Engineering with GFP: Study of -Protein Interactions In vivo, Protein Expression and Solubility

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Mohosin M. Sarkar, M. Sc.

Graduate Program in Chemistry

The Ohio State University

2009

Dissertation Committee:

Thomas J. Magliery, Advisor

Dennis Bong

Ross E. Dalbey

Christopher M. Hadad

Copyright by

Mohosin M. Sarkar

2009

Abstract

Protein–protein interactions (PPIs) play a key role in most biological processes. Many of these interactions are necessary for cell survival. To understand the molecular mechanisms of biological processes, it is essential to study and characterize protein-protein interactions, identify interacting partners and protein interaction networks. There are a number of methods that have been developed to study protein-protein interactions in vitro and in vivo, such as yeast-2-hybrid, fluorescence resonance energy transfer, co-immunoprecipitation, etc. Split protein reassembly is an in vivo probe of protein interactions that circumvents some of the problems with yeast 2- hybrid (indirect interactions, false positives) and co-immunoprecipitation (loss of weak and transient interactions, decompartmentalization). Split GFP reassembly is especially attractive because the GFP chromophore forms spontaneously on in almost every cell type.

However, existing split systems have limitations of evolving cellular fluorescence slowly (3-4 days), failure to evolve at all for some interactions, and also failure to work at a physiological temperature. Among different variants of GFP tested, we found that split folding-reporter GFP

(frGFP, a hybrid of EGFP and GFPuv) evolves fluorescence much faster (24 - 30 h) with associating peptides and also evolves fluorescence for the RING domain BRCA1/BARD1 wild type pair. Thirty six known cancer-associated BRCA1 RING domain mutants were tested with split-frGFP system for their role in BRCA1/BARD1 interactions. Some of these mutations resulted in significant reduction of complex reassembly and cellular fluorescence.

Split frGFP fragments were further improved by directed evolution (error-prone PCR and

ii DNA shuffling) to obtain fragments for fast and efficient fluorescence reassembly. The evolved fragments were able to generate fluorescence in as little as 12-16 h at 30 °C and in 10-14 h at 37

°C. This system was successfully tested for the detection of interactions of several therapeutically important protein pairs (such as Bcl-xL/Bim, Bcl-2/Bim, p53/hDM2, XIAP/Smac), which have

key roles in apoptosis and cancer. Response to known inhibitors of these interactions was also

tested using this system. These results suggest that the efficient split GFP (esGFP) fragments we

developed will be very useful for in vivo screening of small molecule or cyclic peptide libraries to

develop effective modulators of protein-protein interactions in their native cellular context from

direct fluorescence reassembly.

Human paraoxonase-1 (huPON1) has been known for some time for its broad hydrolytic

specificity against organophosphorus (OP) pesticides and nerve agents, such as, sarin, soman and

tabun, etc. The large-scale expression of the soluble protein and the improvement of the stability

and catalytic activity are the most critical challenges for huPON1 to be used as a drug for

detoxification of OP pesticides and nerve agents. As a human protein, it is considered to be a

potent candidate for the development of a catalytic bioscavenger for effective pre- and post-

exposure treatment of OP intoxication. HuPON1 is very unstable and prone to aggregation when

expressed in E. coli. PON1’s hydrophobic leader sequence, hydrophobic surfaces on the HDL

binding sites and the lack of post-translational modifications in bacteria are considered to be some

of the reasons for its lower stability in E. coli. We applied rational and semirational approaches to

re-engineer huPON1 for higher expression and solubility in E. coli. At the same time, applying

approaches of co-expression and MBP (maltose binding protein) fusion and optimizing

purification conditions, we were able to express active, wild-type human PON1 and the

engineered variants in large-scale with a high degree of purity and solubility.

iii Dedication

To my parents

And

To Ruhnaz and Anousha

for their love and support

iv Acknowledgements

During the course of my graduate school, I have been encouraged, inspired and supported by numerous outstanding individuals whose contribution made it possible to complete my thesis.

This work would never have been possible without many of their support and help. I would like to take an opportunity here to acknowledge each and everyone’s support and contribution and express my sincere gratitude towards them.

I would like to sincerely thank my advisor Professor Thomas J. Magliery for his excellent guidance and support and for giving me the opportunity to be a part of his lab. His depth of knowledge and his thoughtful insights about science and ways of addressing scientific problems always intrigued me. The opportunity to work with him has been a valuable experience through which, I must admit, I have learned a great deal from him. I would like to express my sincere gratitude to him for all the assistance, support and encouragements. I am deeply indebted to him.

During my research work I have been privileged to have worked with a number of talented people from whom I have received invaluable advice and inspiration. I would like express my sincere gratitude to Dr. Sean V. Taylor for giving me the opportunity to learn molecular biology in his lab. It has been a great experience to work with him and learn general molecular biology experiments and enzyme kinetics. I would like to thank and extend my appreciation to Professor Christopher M. Hadad, Professor Terry L. Gustafson, Professor George

P. Wang, Professor Richard Sayre (Danforth Plant Science Center, Missouri), Dr. David E. Lenz

(USAMRICD) for their helpful suggestions and advice on the paraoxonase-1 work. I would also like to thank Professor Christopher P. Jaroniec for valuable advice and suggestions on the v biosynthesis of the Transthyretin peptide. I would like to thank Professor Dustin Maly for providing us with pro- and anti-apoptotic proteins and inhibitor and for valuable suggestions to work with them.

I would like to express my sincere gratitude towards all Magliery Lab members for being very supportive of me and for all the help during the course of my research. I would like to thank

George Matic and Deepti Mathur (Cornell University) for their help with the huPON1 and the

BRCA1/ BARD1 works. In particular, I would like to thank and express my sincere appreciation to Christina Harsch and David Mata for their support and help, for the helpful discussions and feedback and for all the PON1 works that we have done together. I would also, like to thank, Dr.

Vivekanand Shete, Dr. Lihua Nie, Brandon Sullivan, Chau Nguyen, Shila Sen, Sarah Johnston and all other lab members for being supportive and helpful. Especially, I would like to thank Dr.

Jason Lavinder, whom I have spent most of my time in the lab with in my graduate school career.

I have learned a great deal of valuable technicalities about biochemical bench work from him.

Also, I would like to thank him for his valuable suggestions and helpful discussions when,

seemingly, things were not going in the right direction and, of course, for blaming it always on

the luck when nothing was going in his way on the pool table.

I would like to sincerely thank my committee members, Professor Ross E. Dalbey,

Professor Christopher M. Hadad, and Professor Dennis Bong for their time and valuable advice.

I would like to express my gratitude to The Ohio State University, The National Institute

of Health (NIH) Center of Excellence for Catalytic Bioscavenger Medical Defense Research for

their support and generous funding.

Finally, I would like to thank my family members and friends for their unwavering

support and encouragement. I am thankful to my parents Abdus Shahid and Anwara Shahid for

their invaluable love and support. Words are not just enough to express my gratitudes towards

them. Especially, I would like to thank and express my sincere gratitude to my brother Shahadat

vi Hossain for all the support and guidance he extended since my childhood to these days. I would like to thank my wife, Ruhnaz, for her invaluable support and sacrifice she made and for being there for me. My three-year-old daughter, Anousha, has always been a constant source of encouragement, inspiration and strength to get back to work whenever I needed. Without them I could never have made it this far.

vii Vita

1989 ...... Ashek Mahmud College, Jamalpur, Bangladesh

1996 ...... B. Sc, Applied Chemistry, Univeristy of Dhaka, Dhaka, Bangladesh

2001-2003 ...... M. Sc., Chemistry, Kent State University, Ohio

2004-2009 ...... Graduate Associate, Department of Chemistry, The Ohio State University, Ohio

Publications

Philippe S. Nadaud, Mohosin Sarkar, Bo Wu, Cait E. Macphee, Thomas J. Magliery, Christopher P. Jaroniec. Expression and purification of a recombinant amyloidogenic peptide from transthyretin for solid-state NMR spectroscopy. Protein Expr Purif. 2009 Sep 29. [doi: 10.1016/ j.pep.2009.09.017]

Mohosin Sarkar and Thomas J. Magliery. Re-engineering a split-GFP reassembly screen to examine RING-domain interactions between BARD1 and BRCA1 mutants observed in cancer patients. Mol. BioSyst. 2008, 4, 599-605.

Anima B. Bose, Mohosin Sarkar and Rathindra N. Bose. Electrocatalytic reduction of platinum phosphate blue on carbon surfaces: A novel method for preparing fuel cell electrodes. J. Power Sources, 2006, vol 157, 188-192.

M Sarkar, Md Ahad Ali, Rashid. M A. Antibacterial compounds from the roots of Rauwolfia serpentina”, J. Biol. Sci. 1998, vol 3, 205-209.

Fields of Study

Major Field: Chemistry

Specialization: Biological Chemistry

viii Table of Contents

Abstract ...... ii

Dedication ...... iv

Acknowledgements ...... v

Vita...... viii

List of Tables ...... xiii

List of Figures ...... xiv

CHAPTER 1: Introduction ...... 1 1.1 Study of protein-protein interactions in vivo ...... 1 1.1.1 Protein-protein interactions ...... 1 1.1.2 Protein fragment complementation assay ...... 4 1.1.3 Fluorescence complementation assay: Split-GFP technique ...... 6 1.1.3.1 Study of protein-protein interactions and its visualization in living cells ...... 12 1.1.3.2 Identification and characterization of protein-protein interactions and new interacting partners ...... 16 1.1.3.3 Principles of protein fragment reassembly and fluorescence complementation 18 1.1.3.4 Advantages and limitations ...... 19 1.1.4 Need for an efficient and faster split fluorescence reassembly ...... 22 1.2 Protein-protein interactions: small molecule drug discovery for cancer therapy ...... 24 1.2.1 Potential targets for drug discovery for cancer therapy ...... 24 1.2.2 Interaction of Bcl-2, Bcl-xL with proapoptotic proteins Bim, Bid, Bax, Bak ...... 27 1.2.3 Interaction of IAP, XIAP with Smac (DIABLO) ...... 29 1.2.4 Interaction of hDM2 with p53 ...... 33 1.2.5 Fluorescence reassembly: Emerging tool for protein-protein interactions and drug discovery for cancer therapy ...... 36 1.3 BRCA1/BARD1 interactions: Involvement in breast and ovarian cancers ...... 38 1.3.1 BRCA1/BARD1: Background ...... 38 1.3.2 Interaction of BRCA1 with BARD1 and putative physiological role ...... 39 1.3.3 BRCA1/BARD1 RING domain heterodimeric complex: E3 ligase activity and target substrate ...... 42 1.3.4 Known cancer predisposing BRCA1 RING domain mutations: Effect on BRCA1/ BARD1 interactions and E3 ligase activity ...... 47 1.4 Human paraoxonase-1: A potential bioscavenger target ...... 50 ix 1.4.1 Paraoxonase-1: Background ...... 50 1.4.2 Organophosphorous pesticides and nerve agent toxicity ...... 53 1.4.3 OP poisoning and nerve agent detoxification: Treatment of acute OP poisoning . 55 1.4.4 Human paraoxonase-1 (huPON1): A potential bioscavenger ...... 58 1.4.5 Physiological role of huPON1: Prevention of atherosclerosis development and cardiovascular disease ...... 63 1.5 Re-engineering human paraoxonase-1: Rational design and directed evolution ...... 67 1.5.1 Solubility tag for the expression in bacterial systems ...... 67 1.5.2 Rational design ...... 69 1.5.3 Directed evolution ...... 70 1.5.4 Folding reporter GFP (frGFP) as a reporter protein for solubility screen ...... 72

CHAPTER 2: Re-engineering a split-GFP reassembly screen to examine RING-domain interactions between BARD1 and BRCA1 mutants observed in cancer patients ...... 74 2.0 Contributions ...... 74 2.1 Summary ...... 74 2.2 Introduction ...... 75 2.3 Results ...... 78 2.3.1 Split sg100 system ...... 78 2.3.2 Constructing folding-reporter GFP ...... 79 2.3.3 Split folding-reporter GFP and split EGFP ...... 80 2.3.4 BRCA1 cancer-associated mutations ...... 82 2.3.5 Pull-down assay ...... 83 2.4 Discussion ...... 85 2.4.1 Improving the split-GFP system for BRCA1/BARD1 ...... 85 2.4.2 BRCA1/BARD1 interaction ...... 86 2.5 Experimental section ...... 88 2.5.1 Plasmid construction ...... 88 2.5.2 Construction and cloning of GFP variant ...... 88 2.5.3 Engineering split GFP fusion construct ...... 89 2.5.4 Screening ...... 89 2.5.5 BRCA1 mutants ...... 90 2.5.6 Affinity purification of fusion proteins and interacting partners ...... 91 2.5.7 Western blotting using anti-HA-tag antibody ...... 92 2.6 Acknowledgements ...... 92

CHAPTER 3: Re-engineering split-GFP fragments for efficient and faster fluorescence complementation to study protein-protein interactions in vivo ...... 93 3.0 Contributions ...... 93 3.1 Summary ...... 93 3.2 Introduction ...... 94 3.3 Results ...... 99 3.3.1 Re-engineering split folding reporter GFP ...... 99 3.3.2 Re-engineering split frGFP fragments for faster and efficient reassembly ...... 101 3.3.3 Targeting protein-protein interactions for small molecule drug discovery ...... 106 3.3.4 Detection of inhibition with small molecule inhibitors ...... 110 3.4 Discussion ...... 112 3.5 Experimental section ...... 115

x 3.5.1 Plasmid construction ...... 115 3.5.2 Construction and cloning of GFP variants an split GFP fragments ...... 116 3.5.3 Screening ...... 116 3.5.4 Error-prone PCR ...... 117 3.5.5 DNA shuffling ...... 117 3.5.6 In vitro reconstitution experiment ...... 118 3.5.7 Bcl-xL, Bcl-2 and Bim peptide cloning ...... 119 3.5.8 XIAP, Smac peptide, and hDM2 and p53 peptide cloning ...... 119 3.5.9 Nutlin3 and ABT-737 inhibition assay ...... 120 3.6 Acknowledgements ...... 120

CHAPTER 4: Rational design of human paraoxonase-1 (PON1) for higher expression and solubility in E. coli ...... 121 4.0 Contributions ...... 121 4.1 Summary ...... 121 4.2 Introduction ...... 122 4.3 Results ...... 126 4.4 Discussion ...... 136 4.5 Experimental section ...... 142 4.5.1 Cloning huPON1 and frGFP fusion in pET11a vector ...... 142 4.5.2 Rational engineering of huPON1 variants ...... 142 4.5.3 N-terminal deletion mutants ...... 143 4.5.4 Cloning into a pHMT vector ...... 144 4.5.5 Cloning in to a pET11a vector with the C-terminal hexahistidine tag ...... 144 4.5.6 Fusion protein expression and purification ...... 144 4.5.7 Chaperone coexpression ...... 145 4.5.8 TEV cleavage and purification of huPON1 and its variants ...... 145 4.5.9 Enzyme kinetics ...... 146 4.5.10 Thermal inactivation and residual activity determination ...... 147 4.6 Acknowledgements ...... 147

CHAPTER 5: Study of BRCA1 cancer predisposing mutations using split frGFP system and directed evolution of huPON1 ...... 148 5.1 Cancer predisposing BRCA1 mutations: Study of the interactions of BRCA1/BARD1 in vivo ...... 148 5.1.1 Study of BRCA1 cancer associated mutants for their interactions with BARD1 154 5.1.2 Future directions ...... 156 5.2 Directed evolution of human paraoxonase-1 (huPON1) ...... 157 5.2.1 Folding reporter GFP (frGFP) as a reporter protein for solubility screen ...... 158 5.2.2 Construction of screening vector using frGFP as a reporter protein ...... 159 5.2.3 Directed evolution of human paraoxonase-1 ...... 159 5.3 Materials and methods ...... 161 5.3.1 Study of BRCA1 cancer associated mutations...... 161 5.3.1.1 Construction of BRCA1 mutants ...... 161 5.3.1.2 Screening ...... 165 5.3.1.3 Affinity purification of fusion proteins and interacting partners ...... 165 5.3.1.4 Western blotting using anti-HA-tag antibody ...... 166 5.3.2 Directed evolution of huPON1 ...... 167 5.3.2.1 Plasmid construction: Cloning linker and frGFP fusion in pET11a vector ..... 167

xi 5.3.2.2 Cloning T4LTA, yTim, huPON1 and G2E6 into the pET11-link-frGFP vector ...... 167 5.3.2.3 Cellular fluorescence measurement ...... 168 5.3.2.4 Random libraries of huPON1 and the N-terminal deletion mutant of huPON1 ...... 168 5.3.2.5 Library cloning into the pET11-link-frGFP vector and screening ...... 168 5.3.2.6 Paraoxon and phenyl acetate assay ...... 169

CHAPTER 6: Supplemental materials and methods ...... 170 6.1 General biochemical and molecular biology methods ...... 170 6.1.1 General molecular biology reagents ...... 170 6.1.2 Plasmids and strains ...... 171 6.1.3 Molecular biology kits ...... 171 6.2 Instrumentation ...... 172 6.3 General protocols ...... 172 6.3.1 PCR and error-prone PCR...... 173 6.3.2 Phenol-chloroform extraction and ethanol precipitation ...... 174 6.3.3 Enzymatic digestion and ligation ...... 174 6.3.4 Purification of DNA fragments ...... 175 6.3.5 Electroporation and electrocompetent cell preparation ...... 175 6.3.6 Protein purification ...... 176

References ...... 178

xii List of Tables

Table 3.1: Mutations on the N-terminal and C-terminal fragments isolated from C19 clones after erroe-prone PCR and rounds of DNA shuffling...... 105

Table 4.1: Yield of wild type human PON1 and the engineered variants expressed and purified from E. coli. Yields are shown in mg of protein per liter...... 138

Table 4.2: Kinetic measurements for huPON1 and engineered variants including N-terminal deletion mutants for the hydrolysis of phenyl acetate...... 138

Table 4.3: Kinetic measurements for huPON1 and engineered variant for the hydrolysis of paraoxon...... 138

Table 4.4: Volume of the reagents and buffer used for the PON1 enzyme kinetics against paraoxon...... 146

Table 4.5: Volume of threagents and buffer used for the PON1 enzyme kinetics against phenyl acetate ...... 146

Table 5.1: Primers used for creating BRCA1 with point mutations using overlap PCR or Quikchange methods...... 163

Table 5.2: Primers used for huPON1 and ΔhuPON1 cloning in to the pET11a-PON1-frGFP vector...... 167

xiii List of Figures

Figure 1.1: Schematic diagrams of different split protein methods to study protein-protein interactions in vitro and in vivo...... 5

Figure 1.2: (a) X-ray crystal structure of sg100 GFP (PDB:1EMA) with potential dissecting points on loops on both sides of the barrel...... 7

Figure 1.3: Re-engineered Split GFP fragment vectors pET11-link-NGFP and pMRBAD-link- CGFP. Linker sequences are designed with cloning ...... 9

Figure 1.4: Fluorescence reassembly between fragments of different spectral variants fused to bFos and bJun. Fluorescence images were taken from COS-1 cells ...... 10

Figure 1.5: Fluorescence reassembly observed from split GFP fragments expressed from the Pmec- 18 promoter in the six touch receptor neurons ...... 14

Figure 1.6: Principle of fluorescence complementation based FRET assay. The interaction between bJun and bFos facilitates Venus fragments ...... 15

Figure 1.7: Sequences of Antiparallel leucine zipper peptides and EK library used to study the protein-proteininteractions using split GFP assay...... 18

Figure 1.8: Mechanism of fluorophore (p-hydroxybenzylidene-imidazolidone) formation in wtGFP as proposed by Heim et al. and Roger Tsien.75, 76 ...... 20

Figure 1.9: Principles of GFP fragments reassembly and fluorescence complementation governed by the free energy of folding from reporter ...... 20

Figure 1.10: (a) between Bcl-2 family proapoptotic (Bcl-xL, Bcl-2, Bcl-w) and antiapoptotic (Bax, Bak, Bim , Bid, Bad) proteins...... 26

Figure 1.11: Structures of ABT-737 and ABT-263, small molecule inhibitors of Bcl-2 family proteins. Structures colored in red indicate ...... 26

Figure 1.12: Role of Bcl-2 family proteins and IAP/XIAP in extrinsic and intrinsic pathways of apoptosis.105 In response to multiple stimuli ...... 29

Figure 1.13: Figure a shows the NMR structure of XIAP-BIR3, (PDB:1G3F) domain (258-346) and N-terminal Smac peptide as shown ...... 31

xiv Figure 1.14: Chemical structures of potent XIAP inhibitors designed from lead compounds smac001 obtained from the screening of a large library of small ...... 32

Figure 1.15: (a) X-ray crystal structures of (1-109) and N-terminal p53 peptide (17-29), PDB: 1YCQ.140 (c) Cis-Im ...... 34

Figure 1.16: Functional domains and structural homology between BRCA1 and BARD1proteins. Both proteins carry nuclear localization signal s ...... 40

Figure 1.17: Solution NMR structures of RING domains of BRCA1 (1-109 aa) and BARD1 (26- 140 aa) with their Zn2+ (gray spheres) binding ...... 41

Figure1.18: Functional diversity of BRCA1 and its regulatory pathways. BRCA1 is a key component of a number...... 46

Figure 1.19: Yeast two hybrid assay of wild type N-terminal RING domain of BRCA1or mutants for the interaction with N-terminal RING ...... 49

Figure 1.20: Summary of Y2H assay results obtained from the screening of BRCA1 mutation library introduced by error-prone PCR...... 49

Figure 1.21: Chemical structures of organophosphate pesticide metabolites (paraoxon, chlorpyrifos, diazoxon) and nerve agents (sarin, tabun, soman, and V-agents)...... 54

Figure 1.22: Inhibition of acetylcholine esterase by organophosphates. Enzymes form a reversible complex with organophosphates ...... 56

Figure 1.23: Plausible mechanism of reactivation of phosphorylated cholinesterase by oxime. Nitroxyl group of oximate attacks the...... 57

Figure 1.24: Figure a shows the X-ray crystal structures of chimeric recombinant PON1 (G2E6, PDB:1V04).272 Central two calcium ions are shown in red and blue...... 61

Figure 1.25: Role of PON1 in LDL and HDL oxidation, macrophage foam cell formation and atherosclerosis development...... 65

Figure 1.26: Folding reporter GFP as a reporter protein for the screening of soluble proteins from large library. frGFP is expressed as ...... 72

Figure 2.1: Solution structure of BRCA1 and BARD1 RING domain heterodimer complex (1JM7).194 (Left) BARD1 is at left in ...... 77

Figure 2.2: Fluorescence complementation with the split sg100 system. Cells were grown for 24 h at 30 ºC followed by two days at room te ...... 79

Figure 2.3: Schematics of the GFP variants with their corresponding mutations, which are used for split GFP techniques. The frGFP variant was ...... 80

Figure 2.4: Comparative fluorescence complementation for split sg100 and split frGFP systems. Cells were incubated overnight at 30 ºC ...... 81 xv Figure 2.5: Fluorescence complementation with split frGFP system depends on fusion orientation for BRCA1/BARD1. Cells were incubated...... 82

Figure 2.6: Interaction of BARD1 with cancer-associated mutants of BRCA1 observed by split frGFP reassembly. Fluorescence was ...... 83

Figure 2.7: (a) SDS-PAGE of purified, reassembled complexes by Ni-NTA affinity column. Lane 1, MW markers; 2, positive control ...... 84

Figure 3.1: Fluorescence complementation with the split sg100 (left) and frGFP (right) systems. Left: Cells were grown for 24 h at 30 °C ...... 99

Figure 3.2: Directed evolution and bilmolecular selection of split-frGFP fragment libraries. (a) Schematic of directed evolution...... 102

Figure 3.3: Cellular fluorescence observed from selected colonies isolated after 3rd of DNA shuffling and selection at both 30 °C and 37 °C...... 104

Figure 3.4: Cellular fluorescence observed from evolved fragments (C4, C13 and C19) and wild type regular frGFP fragments with leucine ...... 104

Figure 3.5: A model structure of the C19-GFP constructed from the evolved N-terminal and C- terminal fragments isolated from C19 clone...... 105

Figure 3.6: In vitro reconstitution experiment for the evolved split fragments. Purified complex was denatured in 6 M Gu-HCL overnight and dialyzed into a buffer...... 105

Figure 3.7: Cellular fluorescence complementation of protein-protein interactions of Bcl-xL/Bim, Bcl-2/Bim, XIAP/Smac, p53/hDM2 ...... 109

Figure 3.8: Suppression of protein-protein interactions by the small molecule inhibitors ABT-737 and Nutlin3. Cellular fluorescence read out of on plate assay ...... 111

Figure 4.1: Schematic of the designed pET11 vector for using folding reporter GFP (frGFP) as a C-terminal fusion of test proteins...... 127

Figure 4.2: Whole-cell fluorescence measured for the cells expressing proteins with wide range of solubility and as...... 129

Figure 4.3: Whole-cell fluorescence of frGFP fusion proteins of wild type huPON1 and recombinant G2E6 and their variants expressed in E.coli...... 130

Figure 4.4: (a) Schematic of pHMT vector designed for the expression of MBP fusion human PON1 and its variants in E. coli...... 131

Figure 4.5: IMAC (Ni-NTA agarose resin) purified huPON1 variants expressed in E. coli from MBP-fusion pHMT vector...... 133

Figure 4.6: Western blot (right) of cell lysate and purified human PON1 variants from the MBP fusion vector in E. coli. Hexahistidine tag containing...... 133 xvi Figure 4.7: SDS-PAGE gel of IMAC purified MBP fusion wild type human PON1 and it variants. MBP fusion proteins were co-expressed ...... 133

Figure 4.8: Thermal inactivation of huPON1 and the engineered variants. Protein samples were heated for 10 min at different ...... 134

Figure 4.9: Hydrolysis of paraoxon by purified huPON1. Paraoxon activities were assayed in 50 mM Tris-HCl, 10 mM CaCl2 pH 7.4...... 135

Figure 5.1: Table above listed the known cancer associated mutations in the N-terminal RING domain of BRCA1 (Breast Cancer Information Core, NIH)...... 151

Figure 5.2: Plasmid maps, pMRBAD-BRCA1-CfrGFP and pET11-NfrGFP-BARD1, constructed for co-expression and in vivo interaction study...... 153

Figure 5.3: Cellular fluorescence obtained from screening of BRCA1 mutants for their interaction with BARD1 using split-frGFP assay...... 155

Figure 5.4: LB agar plates for the screening of error-prone libraries of the wild type huPON1 and NΔhuPON1. Both libraries were expressed as the C-terminal ...... 160

Figure 5.5: Paraoxonase turnover activity measured for the variants isolated from error-prone libraries of huPON1 and N-terminal deletion mutant ...... 160

Figure 5.6: pMRBAD vector used for constructing cancer predisposing BRCA1 mutants. C- terminal of C-frGFP was tagged with HA (YPYDVPDYAK)...... 162

xvii CHAPTER 1

Introduction

1.1 Study of protein-protein interactions in vivo

1.1.1 Protein-protein interactions

Protein-protein interactions (PPIs) play a central role in most fundamental biological

processes, from signal transduction to and regulation, transcription and

translation, cell growth and differentiation, intercellular communication and programmed cell

death.1-3 Interactions between proteins, and proteins and other macromolecules are necessary for cell survival. It has been estimated that over 80 % of the proteins in do not operate alone but in complexes.2 Proteins frequently interact with themselves or other proteins and function as stable or transient complexes to regulate a wide variety of cellular functions.

Many of these protein-protein interactions are part of larger cellular networks and are tightly regulated by a number of different mechanisms. Results from human and other genome sequencing projects suggest that the biological complexity of organisms is not determined merely by the number of genes but by the number of physiologically relevant protein interactions.4, 5 It is now evident that proteins recognize and bind targets in a highly specific manner and regulate most of the cellular functions and ultimate fate of the cells. Alterations in protein-protein interactions perturb the normal sequence of events in the cell and contribute to diseases.

Therefore, in order to understand the mechanisms of biological processes at the molecular level, identification and characterization of the protein-protein interactions, interacting partners and

1 their networks are essential. Every year enormous number of studies have applied experimental methodologies and in silico analysis for detailed characterization of individual protein-protein interactions in regard to structure, biological role and function. Identifying and understanding the normal pattern of protein-protein interactions can lead to the development of drugs to fight the underlying cause of diseases and cancer.

However, in spite of remarkable advances in biotechnology and genetic engineering, identification and characterization of protein-protein interactions and protein interaction networks

still remains a very challenging job. A large number of genetic and biochemical experimental

methodologies have been developed to facilitate the identification, characterization and analysis

of protein-protein interactions both in vivo and in vitro.1, 6, 7 Co-immunoprecipitation and TAP-

tag, chemical crosslinking, protein arrays, fluorescence resonance energy transfer (FRET),

fluorescence anisotropy and surface plasmon resonance spectroscopy are among the most

prevalent today to study protein-protein interactions in vitro. A more limited number of methods

such as yeast two-hybrid (Y2H) assay, FRET, bioluminescence resonance energy transfer

(BRET), split protein complementation assay (PCA) are available to study protein-protein

interactions in vivo. Coimmunoprecipitation, TAP-tag and Y2H assay are among the most widely

used approaches to screen or confirm PPIs in vivo. Y2H assay, recently developed protein arrays

and phage display have been widely used to screen large libraries of proteins for interacting

partners. Each of these methods has its own advantages and limitations.1, 7 Nonspecific binding

and the loss of transient interactions are potential problems with chemical crosslinking,

coimmunoprecipitation and TAP-tag methods. In addition, these traditional assays are not useful

for high-throughput approaches and require the use of cell lysis, purification of interacting

proteins and creation of experimental conditions that are inconsistent with natural cellular

environment. Since its introduction in 1989 by Stanley Fields,8, 9 Y2H has been one of the most

widely used methods to study protein-protein interactions in vivo.10, 11 It has been very useful for

2 high-throughput screening from a large library and discovery of new protein-protein interactions by means of transcriptional activation of a reporter protein. The classic Y2H method uses a transcription factor (such as yeast Gal4 protein) split into a DNA binding domain (DBD) and transcription activation domain (AD), which are fused to the proteins of interest. It was demonstrated that when the two domains were expressed separately, they were unable to activate transcription. When an interaction between two test proteins takes place, it brings the two domains in close proximity to construct a functional transcription factor and activates the expression of a reporter protein (lacZ, GFP, etc.). Conventional Y2H systems are limited to yeast cells and to the protein interactions that can be localized to the nucleus, and is not useful for membrane protein study. This method does not demand direct interaction of the proteins and is hampered by the proteins initiating transcription by themselves. A variety of improvements have been made over the years to overcome these limitations which has made it possible to study protein interactions in mammalian12 and prokaryotic cells13-15 and to study membrane protein

interactions.16-18 These techniques require the addition of exogenous substrates (fluorogenic,

chromogenic or chemiluminescent reagents) or require a survival selection method to detect

protein-protein interactions. In spite of having many improvements and variations, Y2H still

suffers from very high rates (as high as 50%) of false-positives2 and fails to detect transient or

19 weak interactions (KD 1 µM). Therefore, it is always important to verify the result obtained from

Y2H using other methods. Though FRET and BRET have been very useful for real time detection

and visualization of PPIs from direct read out in vivo, the main draw backs of these methods are

again high rate of false-positives, failure to detect weak interactions, and incompatiblity with

high-throughput screening. To overcome these limitations, the split protein complementation

approach (PCA) has emerged in recent years as a valuable tool to study protein-protein

interactions in vivo in a wide range of cells and subcellular in their natural cellular

environment.

3

1.1.2 Protein fragment complementation assay

Protein fragment complementation assay (PCA) depends on a similar strategy as Y2H in which a functional protein is dissected into two inactive components that are fused in frame to target proteins which upon association can restore the function of the split protein. Protein folding is typically a thermodynamically driven highly cooperative process. Cleavage of a single polypeptide bond in a protein results in protein unfolding and inactivation. In spite of dramatic energetic consequences upon cleavage of a covalent bond, some proteins may remain folded and held together by noncovalent interactions. This idea is supported by work carried out by

Anfinsen20 and Richards,21 in which they have shown that a mixture of polypeptides corresponding to proteolytic cleavage can sometimes result in a reassembly of active protein.22 In some cases protein fragments may not reassemble to intact protein and restore the function spontaneously. Fusion of the split fragments with known interacting proteins will assist reassembly and folding of the fragments to intact protein and will make the processes thermodynamically favorable. Based on this hypothesis, Fields and Song in 1989 split a transcription factor protein into a DNA binding domain and an activation domain, and devised a method widely known as yeast two hybrid assay to study protein-protein interactions in vivo.9

Johnsson et al23 in 1994 first describe the split ubiquitin based method, and later Stagljar et al18, 24-

26 modified this method to make it compatible with Y2H using transcription factor as a reporter protein. Compared to the classic and modified Y2H assays, the PCA techniques have been developed based on the common principle that the reporter protein itself is dissected into two fragments which are genetically fused to the potential interacting partners. Dihydrofolate reductase (DHFR),27 β-lactamase,28, 29 β-galactosidase,30 firefly or Renilla or Gaussia luciferase31-

36 and green fluorescent protein (GFP)37 have been used successfully for this purpose to study protein-protein interactions in a wide variety of cell lines including mammalian cells. A simplified cartoon configuration of different protein fragment complementation assays

4

Figure 1.1: Schematic diagrams of different split protein methods to study protein-protein interactions in vitro and in vivo. Figure was reproduced from Piehler, J. Current Opinion in Structural Biology, 2005.7

developed to study protein-protein interactions in vitro and in vivo is shown in Figure 1.1. These

methods have been developed based on more direct read-out in actual cellular context and applied

for the detection of protein-protein interactions in variety of cell lines and cellular compartments.

In most of these PCA techniques the reporter protein is an enzyme which requires exogenous

substrate for the interaction-signal read-out. Though, the enzyme activity can amplify and

generate signals very fast (minutes to hours), the substrate itself can be a source of background

and the reaction products can accumulate over the course of the assay. Uneven distribution of

exogenous substrates inside cells and cellular compartments is also a potential problem posed by

these methods. To overcome these limitations and for faster detection of interactions with low

background signal, fluorescent protein (green fluorescent and its spectral variants) based PCAs

5 with direct spectroscopic readouts have been developed. The reassembled fragments have strong intrinsic fluorescence that allows direct visualization of the protein interactions in living cells and organisms. Therefore, protein interactions can be detected without any exogenous chromogenic or fluorogenic substrates and ligands avoiding potential perturbation of cells by these agents.

1.1.3 Fluorescence complementation assay: Split-GFP technique

Green fluorescent protein (GFP) and its spectral variants have been the most widely used genetic marker for simple and direct visualization of protein trafficking and localization in wide variety of cells and organisms. Among numerous applications of fluorescent proteins, fluorescence complementation has emerged as a valuable tool to study protein-protein interactions, protein interaction networks and their direct visualization in subcellular compartments. Split GFP fragment reassembly or bimolecular fluorescence complementation

(BiFC) are based on a strategy that a fluorescent protein is split into two nonfluorescent fragments that are fused to two test proteins. Interaction between test proteins facilitates the association of the split fragments to reconstitute intact fluorescent protein which can be detected from direct fluorescence readout. Ghosh et al. in 2000 first introduced the fluorescent protein based protein fragment complementation assay (PCA) using the reassembly of dissected fragments of GFP.37 In their technique they have used a variant of GFP, sg100 GFP (single

excitation maximum at 475 nm and emission maximum at 509 nm), and dissected it at a surface

loop between residues 157 and 158, Figure 1.2. A previous study by Abedi et al. showed that an

insertion of 20 amino acid peptides in this loop position did not significantly perturb the

expression, folding and fluorescence level of GFP.38 Ghosh et al. were able to show that two split

fragments (1-157 residues and 158-238 residues) do not associate to give reassembled GFP in

vitro or in vivo, when purified fragments are mixed in equimolar ratio or overexpressed in E. coli.

However, when two fragments are fused with strongly interacting antiparallel leucine zipper

peptides39-41 reconstitution of intact GFP is achieved in vitro and in vivo, and strong fluorescence 6

Figure 1.2: (a) X-ray crystal structure of GFP (PDB:1EMA) with potential dissecting points on loops on both sides of the barrel. (b) and (c) Schematic of split-GFP fragment reassembly by antiparallel leucine zipper peptides. NGFP (residues 1–157) is green, CGFP (158–238) is green, and the zipper peptides are blue. (d) Cellular fluorescence upon GFP reassembly. Figures b, c and d were reproduced from Magliery et al. J. Am. Chem. Soc., 2005.42

7 signal is observed. In their system, antiparallel leucine zipper peptides were fused to the C- terminus of the N-terminal GFP (residues 1-157, NZGFP) and the N-terminus of the C-terminal

GFP (residues 158-238, CZGFP). This orientation was designed to allow most favorable interaction between the two antiparallel leucine zipper peptides in the reassembled complex.

Magliery et al. later reported an improved split-GFP fragment complementation assay based on original screen. They re-engineered the system and designed two compatible vectors for split fragments that can be co-maintained in E. coli.42, 43 The NGFP fusion vector, pET11-link-NGFP,

contains T7 promoter and terminator sequences, ColE1 origin and ampicillin resistance marker

(Figure 1.3). The CGFP fusion vector, pMRBAD-link-CGFP, is designed to have an arabinose

promoter with araC regulator gene, p15A origin and kanamycin resistant marker. These standard

vectors can readily be used to subclone test proteins to examine their interaction and cellular

localization. Since its introduction in 2000, fluorescence reassembly techniques have been a

valuable tool to study and identify new protein-protein interactions in vivo, to study the patterns

of expression, transcriptional regulation and subcellular localization of protein complexes.

Following Ghosh’s work, potential uses of other variants of GFP and other points of dissection

have been explored by different groups. Hu and Kerppola in 2002 demonstrated that fragments of

enhanced yellow fluorescence protein (EYFP) can reconstitute the fluorophore when fused to the

parallel leucine zipper domains of Jun and Fos, and called their approach bimolecular

fluorescence complementation (BiFC).44 They also examined various dissecting points in YFP for

efficient reassembly. YFP was split into two fragments at several nonconserved amino acid

residues within loops at either end of the β-barrel structure (38-39, 101-102, 144-145, 154-155,

168-169, 172-173 and 192-193) and fused these fragments at the C-terminal ends of bZIP

domains using short linkers between them (5-7 amino acids). From different protein-protein

interactions (combinations among Fos, Jun, ATF2, p50, p65, and IκBα) they have tested, the best

fluorescence reassembly obtained when YFP was split between 154 and 155 residues. They were

8

Figure 1.3: Re-engineered Split GFP fragment vectors pET11-link-NGFP and pMRBAD-link-CGFP. Linker sequences are designed with cloning restriction enzyme sites in such a way that they are in frame with NGFP and CGFP and can be replaced with test protein genes for easy cloning. Figure was reproduced from Wilson et al. Nature Methods, 2004.43

also able to examine the reassembly and detect subcellular localization of bZIP transcription factor in mammalian cells. In 2003, Kerppola and coworkers reported the multicolor fluorescence complementation approach for simultaneous visualization of multiple protein interactions in living cells.45 This approach was based on the complementation between fragments derived from

EYFP (enhanced yellow fluorescence protein), EGFP (enhanced green fluorescence protein),

ECFP (enhanced cyan fluorescence protein) and EBFP (enhanced blue fluorescence protein)

dissecting at 154-155 and 172-173 surface loop positions. Fragments were fused to the C-terminal

of bZIP domains of Fos and Jun. The GFP fragments split at 154-155 and 172-173, and BFP and

CFP fragments split at 172-173 did not exhibit any detectable fluorescence complementation

when expressed in mammalian cells or in E. coli. On the other hand, CFP when split at 154-155

9

Figure 1.4: Fluorescence reassembly between fragments of different spectral variants fused to bFos and bJun. Fluorescence images were taken from COS-1 cells transfected with plasmids expressing the protein fragments indicated in each panel. The C-terminal fragments were fused to bFos and the N-terminal fragments were fused to bJun. Structural models of the reassembled complexes are shown to the right. Figure was reproduced from Hu et al., Nature Biotechnology, 2003.45

and YFP when split at both 154-155 and 172-173 exhibited reassembly and fluorescence.

Surprisingly, it was found that YFP(1-154) with CFP(155-238) and YFP(173-238) with GFP(1-

172) or CFP(1-172) or BFP(1-172) exhibited reassembly and fluorescence with distinct spectral characteristics. No reassembly was observed for GFP(172-238) or CFP(172-238) or BFP(172-

238) with any 1-172 fragments. More surprisingly, they have observed that N-terminal 1-172 fragments of YFP, GFP and CFP were able to reassemble with C-terminal 155-238 fragments of

YFP and CFP despite the overlapping of a β-strand (Figure 1.4). BFP(1-172) fragment was able to reassemble only with CFP(155-238) fragment. This multicolor fluorescence complementation enables the visualization of interactions between different proteins in the same cell and comparison of the efficiencies of complex formation among different interaction partners.45

Following the Ghosh et al. approach, Chalfie and coworkers46 used antiparallel leucine zipper peptides as fusion proteins and showed that CFP and YFP can also reassemble when split at 157- 10

158. They have also showsn that CFP (158-238) can also reassemble with GFP (1-157) and

YFP(1-157) and exhibit fluorescence with distinct spectral properties. More recently, Shyu et al. reported the identification of fluorescent protein fragments derived from Cerulean and Venus47, 48 for improved BiFC and multicolor BiFC with efficient reassembly under physiological culture conditions.49 Among the different combination of fragments examined, ECFP(155-238) with

Cerulean(1-172) and Venus(1-172) have been found to be the best combinations for multicolor complementation assay (Figure 1.4). Spectral variants of red fluorescent proteins (mRFP1-Q66T, mCherry) have been used lately for fluorescence complementation assay by dissecting them at

154-155, 136-137, 159-160 and 174-175 to study the protein interactions between humanized

Renilla GFP (hrGFP) monomers and between LTag (SV40 Large T antigen) and human p53 proteins.50, 51

Umezawa and coworkers developed a modified fluorescence complementation assay, in

which the EGFP fragments dissected at 128-129 were fused with split VMA1 intein from yeast.52

The N-terminal half (1-184) of VDE (modified VMA1, without 185-388 endonuclease domain)

and the C-terminal half (389-454) of VDE are each fused with N-terminal and C-terminal EGFP,

respectively. Interacting proteins, calmodulin and M13, are connected to the split VDE fragments.

When calcium mediated the interaction between calmodulin and M13 in E. coli, GFP fragments

were reassembled, covalently connected and cleaved off by full length intein. Later they have

reported an improved version of their system using bacterial DnaE split-inten instead of split

VDE from yeast for faster and efficient covalent reassembly of the split EGFP fragments.31 In addition to the 128-129, they have also examined the potential of other dissecting points (144-

145, 157-158 and 224-225) and found that split at 157-158 position (with modification of K156Y and Q157C or an insertion of KFAEYC after Q157) produced better reassembly compared to the other fragments with similar modifications. Umezawa, in 2003, also reported the successful use of this intein based method in cultured mammalian cells.53 Cabantous et al. reported a split GFP

11 complementation method and used it as a reporter system for rapid screening of soluble protein from large library in vivo and in vitro.54 In their technique a variant of GFP, kown as super-folder

GFP, was dissected at 214-215 and N-terminal fragment (1-214) was subjected to directed

evolution for improved fluorescence and solubility, and C-terminal fragment (215-230) was

subjected to directed evolution for optimized solubility. Neither of these fragments was

fluorescent alone. Upon mixing or coexpression, both fragments were spontaneously reassembled

and fluorescence was observed. Test protein was fused N-terminally to the small C-terminal GFP

fragment whose proper folding and solubility or aggregation will be depended on the solubility of

the test protein. Though this has been a valuable tool for protein solubility screen, it is not useful

for protein-protein interactions study.

These studies suggested that GFP and its spectral variants and other fluorescence proteins

can be dissected at different loop positions and the fragments will reassemble upon association

with interacting proteins. Dissection at surface loop near 157-158 led to the most successful

reassembly and fluorescence complementation. A few of the peptide bonds in a particular protein

can be cleaved to produce fragments that can associate to produce intact functional protein. It was

also evident that these protein fragment reassemblies can accommodate different fusion

topologies and wide range of linker length and precise alignment of GFP fragments may not be

necessary to initiate the reassembly.

1.1.3.1 Study of protein-protein interactions and its visualization in living cells

Fluorescence complementation based on split GFP and its different spectral variants has

been successfully applied to study many different protein-protein interactions in a variety of cell

lines and organisms such as E. coli, yeast, mammalian cells (COS-I, HEK, HeLa, NIH3T3), C. elegans, etc. Nagai et al. studied the Ca2+ dependent interaction between calmodulin and M13

peptide using full length circularly permutated GFP and dissected YFP.55 Interaction between

calmodulin and M13 peptide has also been studied by Ozawa et al. using modified split EGFP 12 and split intein mediate assay.31, 52 Hu et al. developed a fluorescent complementation assay based

on split YFP and CFP and studied the interaction between bZIP domains from Fos and Jun. Using

split YFP, they have also studied the protein-protein interactions among full length Fos and Jun,

Rel family protein of NFκB (p50 and p65) and IκBa.44 Magliery et al. used the originally

developed split GFP system and examined the interactions of tetratricopeptide repeat (TPR)

domains (TPR I, TPR 2A and TPR 2B) with and .42 Fan et al. developed a PCA

technique based on split mCherry, which in combination with split YFP is used to examine the

interaction between SV40 Large T antigen and human p53 protein, and sp100 and promyelocytic

leukemia protein (PML) in Vero cells.51 Most recently Sarkar et al. developed an improved split

folding reporter GFP based PCA and used their system to study interaction of BARD1 with

different mutants of tumor suppressor protein BRCA1 observed in cancer patients.56 Barnard et

al. used a split EGFP system to study protein-protein interactions between phosphofructokinase

subunits pfk1 and pfk2 in yeast and in different subcellular compartments, such as Idh1p and

Idh2p (mitochondrial matrix), Sdh3p and Sdh4p (mitochondrial membrane) and Pap2p and Mtr4p

(nucleus).57

One of the most remarkable advantages of fluorescence complementation assay is that, in addition to the direct fluorescence readout, it enables simple and direct visualization of protein interactions in their native state in different subcellular locations of almost every cell type and organism. Using split YFP, Kerppola and coworkers demonstrated the subnuclear localization of bZIP domains of Fos and Jun.44 When full length Fos and Jun was fused to the YFP fragments,

localization to the nucleoplasmic region was observed (Figure 1.4). They have also shown that

ATF2/Jun heterodimer predominantly localize in perinuclear region. Interaction between Rel

family NFκB protein p50 and p65 resulted in uniform fluorescence in nucleoplasmic region but

interaction between IkBa and p65 proteins mainly produced cytoplasmic fluorescence.44 Using

split YFP and CFP Kerppola et al. developed a multicolor fluorescence complementation assay

13 and demonstrated simultaneous visualization of multiple protein interaction in COS-1 cells.45

This was based on a principle that the association of different fluorescent protein fragments through their fused interacting partners can lead to distinct emission spectra which will distinguish different associations in the same cell. Hu et al. added further improvement of this system by using two other spectral variants of GFP, Venus and Cerulean.49 Using multicolor fluorescence complementation assay, Grinberg et al.58 examined the competition of dimerization

and precise subcellular locations of Myc/Mad/Max family proteins. Fang et al. reported the

Itch/AIP4 interraction in lysosomes;59 Giese et al. reported Cytokine receptor gp130/LIFR in plasma membrane;60 Niu et al. reported GBF1/Arf1 dimerization in Golgi;61 Ozalp et al. reported

Cyt-p450 2c2 and 2e1 interaction with Cyt-p450 reductase in endoplasmic reticulum;62 Takahashi

et al. reported Bif-1 interaction with Bax/Bak in mitochondria of mammalian cells.63

Figure 1.5: Fluorescence reassembly observed from split GFP fragments expressed from the Pmec-18 promoter in the six touch receptor neurons and from the heat shock promoter Phsp16.2 in C. elegans. Figures were reproduced from Zhang et al. Cell, 2004.46

Using a similar split GFP approach described by Ghosh et al., Chalfie and coworkers in

2004 first reported the visualization of protein interactions in a whole organism, C. elegans.46

They have expressed NZGFP and CZGFP under mec18 promoter (Pmec-18) which is expressed

only in the six touch receptor neurons of C. elegans, and under hsp16.2 heat shock promoter

(Phsp16.2) which is widely expressed in C. elegans (Figure 1.5). Using a number of tissue and cell

specific promoters (Punc-4, Punc-24, Pmec-3, Pegl-44, Psto-6, Pacr-5, etc.), it was shown that the expression

14

Figure 1.6: Principle of fluorescence complementation based FRET assay. The interaction between bJun and bFos facilitates Venus fragments to form reassembled protein which upon folding and maturation serves as a FRET acceptor when Jun/Fos complex interacts with NFAT1 fused with cerulean which acts as a FRET donor. Figure was reproduced from Shyu et al. PNAS, 2008.65

of GFP fragments and reconstitution of GFP were not promoter or tissue specific. From the expression of GFP fragments under different combinations of promoters they were also able to demonstrate and identify the promoter dependent protein expression pattern in C. elegans. When the GFP fragments were expressed under Pmec-24 promoter (cells in ventral cord and six touch receptor neurons specific) and Pmec-2 (six touch receptor neurons specific), fluorescence reassembly was observed only in six touch receptor neurons. In a different set of experiments they have demonstrated that among different combinations of GFP, YFP and CFP fragments tested, NZYFP/CZCFP was able to reassemble with most intense fluorescence. Following

Chalfie’s work, using bZIP domains of Fos and Jun, Hu and coworkers in 2008 demonstrated the use of fluorescence reassemly to identify, visualize and validate temporal and spatial protein interactions in living worms.64

More recently, fluorescent complementation assay has been adapted to plants and used to

study and visualize many protein-protein interactions in numerous plant species.66-71 Waadt et al. 15 demonstrated the use of multicolor fluorescence complementation to study and visualize the interactions between CIPK/CBL complexes in plant cells.69 Using their system they have shown that CIPK1 can form alternative complexes with CBL1 and CBL9 simultaneously in epidermal cells and protoplasts, and CIPK24 can form alternative complex with CBL1 and CBL10 at the plasma membrane and tonoplasts, respectively. In a most recent study, Hu and coworkers described a significant improvement of split GFP assay by combining split Venus approach with a Cerulean-based FRET to study protein-protein interactions in a ternary complex in living cells.65 Recently, a three chromophore-based FRET system has been developed by Galperin et al.72 for the visualization of the ternary complex in living cells. The use of this approach was

limited by the requirement of the expensive and sophisticated optical system. In fact, methods for

studying protein interactions in a complex formed by more than two proteins are extremely

limited. In Hu’s easy-to-operate approach, they have fused bZIP domains with split Venus

fragments and a full length Cerulean with NFAT (Nuclear factor of activated T cells). NFAT is a

transcription factor required for T cell development and is known to interact with Fos-Jun

heterodimer to form a composite NFAT-AP1 element in the regulatory regions of many target

genes. Upon interaction with bZIP domains, split fragments will reconstitute a full length Venus

which will serve as a good acceptor for Cerulean donor when a Fos-Jun-NFAT1 ternary complex

is formed (Figure 1.6). This system has also been successfully used to study the interactions of

transcription factors AP-1 (Fos-Jun) and NF-κB subunit, p65. Since many signaling proteins and transcription factors function as a complex of a number of proteins, split GFP based FRET will be a valuable tool to visualize these complexes and identify the proteins involved in the complex formation.

1.1.3.2 Identification and characterization of protein-protein interactions and new

interacting partners

For deeper understanding of biological function and cellular biological networks, 16 identification and characterization of protein interaction partners are essential. Based on split-

GFP fluorescence assay, Remy et al. first reported a large-scale genome wide screening of a cDNA library to identify and validate new protein-protein interactions and interacting partners.73

A human brain cDNA library (fused to the N-terminal GFP fragment) was screened for the

identification of novel substrate or regulators of the serine/threonine protein kinase PKB/Akt

(fused to the C-terminal GFP fragment). Mammalian COS-1 cells were transiently cotransfected

with the plasmids and positive clones were screened using FACS (Fluorescence activated cell

sorting). After two rounds of selection, plasmids from 100 colonies were sequenced among which

22 sequences corresponding to genes of potential interest (rest of the sequences were either

unreadable, contaminants, or genes with no known function, false positives). Out of the 22

promising hits, one of the proteins they identified was hFt1, a human homolog of mouse Ft1.

Using co-immunoprecipitation further validation of direct interaction of PKB with hFt1 was

carried out. The PKB/hFt1 interaction predominantly takes place at the plasma membrane, and is

induced by PI3K-associated signaling pathways. This study clearly demonstrated the valuable

application of split fragment approach in identification and characterization of novel protein

interactions and interacting partners.

Magliery et al. used their improved split GFP assay for screening of antiparallel leucine

zipper libraries to understand and explore the structural requirements needed for the

interactions.42 It is known that parallel or antiparallel coiled coils are generally stabilized by

hydrophobic interactions in the core and charge-charge interactions of edged residues, but what

controls the orientation is not clearly known. They have constructed a library in which the eight

edge positions in one zipper peptide were randomized between Glu and Lys using the RAA

codon and keeping the buried hydrophobic and charged residues constant. Library variants were

screened based on cellular fluorescence from reassembled GFP. Fluorescence was observed for

the peptide with three or fewer charge-charge mismatches (Figure 1.7). It was noticed that in all

17

Figure 1.7: Sequences of Antiparallel leucine zipper peptides and EK library used to study the protein- proteininteractions using split GFP assay. Helical wheel diagram of antiparallel leucine zipper peptides interactions. Distribution of mutations in the library screened. Figures were adapted from Magliery et al.42

eight position mutations were not equally frequent and the charge mismatches near the ends of the peptide were less significant. Data also suggested that mutations at positions K6, K18 and

K20 are especially disruptive and profoundly affected the interactions of the zipper peptides

(Figure 1.7). From the screening and characterization of the library variants, it was estimated that

split GFP fragments were capable of determining very weak interactions (Kd ~ 1 mM) between

leucine zipper peptides.

1.1.3.3 Principles of protein fragment reassembly and fluorescence complementation

The crucial feature of split fragment complementation assay is that the fragments are

designed not to fold spontaneously without being brought together in close proximity by the

interaction of the fused proteins. The method relies on the proper folding and interaction of

proteins to which fragments are fused. The principle is that the interaction of fusion proteins

increases the effective concentration of the fragments, which favors and facilitates the correct

folding over any other nonproductive processes. It is proven that the interaction between fusion

proteins is required to initiate or nucleate the reassembly reaction.42 The interaction between 18 fusion proteins need not to be strong, as it is reported that weak interactions as low as KD ~1 mM

can lead to the reassembly. The folding of the reporter fluorescent protein depends not only on

close proximity of the fragments, but the requirement of the precise orientation of the fragments

in space to allow for folding of intact protein. Therefore, the amount of cellular fluorescence

depends on both the physical properties of the fused proteins and the affinities of the proteins for

each other. Initial reconstitution of intact protein is considered to be in equilibrium between the

folded and unfolded states. The process is favored to the folded state because of negative free

74 energy (ΔGfolded-unfolded ~ negative) between the folded and unfolded states, (Figure 1.9 a). Once

GFP folds and fluorophore is formed, the process is essentially irreversible. Since the dissected

GFP fragments are nearly insoluble proteins, this irreversible step may shift the equilibrium

towards soluble folded products. The most notable advantage of this irreversible process is that

once the fluorophore is formed, even by weak interaction, it is trapped and can accumulate over

time, which makes the fluorescent complementation especially attractive to detect very weak and

transient interactions.

1.1.3.4 Advantages and limitations

Protein fragment complementation assay (PCA) techniques emerged as a powerful

alternative to the traditional or modified yeast-2-hybrid assays. Split GFP assay has several key

advantages over all other PCA techniques. It does not require any exogenous substrates

(fluorophore, chromophore or other chemical substrate). Thus, it is avoiding any potential cellular

perturbation of the cells caused by these chemical agents. Unlike FRET, Y2H and other PCA

methods, split GFP technique has little or no background or false-positive read out. Many of the

cells do not have significant fluorescence background at the GFP emission/excitation

wavelengths. Therefore, virtually all the signal can be attributed to the reassembled GFP. Fusion

proteins are optimized to expressed at very low level in their native cellular context. The method

requires direct interaction of proteins and can detect very weak and transient interactions. It is 19

Figure 1.8: Mechanism of fluorophore (p-hydroxybenzylidene-imidazolidone) formation in wtGFP as proposed by Heim et al. and Roger Tsien.75, 76

Figure 1.9: Principles of GFP fragments reassembly and fluorescence complementation governed by the free energy of folding from reporter protein interactions (a). Figure (a) was reproduced from Michnick et al. Nature Reviews Drug Discovery, 2007.74 Plausible mechanism of GFP fragments reassembly is shown in (b) as proposed by Magliery et al.42 Interaction between reporter proteins nucleates the folding of GFP fragments and facilitates the chromophore formation. Reassembly of fragments is essentially irreversible process which further favors equilibrium towards folding. Protein-protein interactions are not necessary to maintain the interaction. Figure (b) was adapted from Magliery et al. J Am Chem Soc, 2005.42

20 especially attractive because the GFP chromophore forms spontaneously upon protein folding in virtually every cell types (bacteria, yeast, mammalian cells, etc.) and every subcellular compartment. Therefore, the method can be used to study and visualize protein-protein interactions in their native biological environment in wide variety of cell types and whole organisms,46, 64 and in almost all subcellular locations. The method has been used previously for

high-throughput screening from large library to identify new protein interactions and interacting

partners73 and to study and characterize structural determinants required for protein-protein interactions.42 The most remarkable advantage of split GFP technique is that it not only detects

the interactions between proteins; it enables the direct visualization of protein interactions in

living cells. Multicolor fluorescence complementation based on fragments derived from YFP and

CFP or Venus and Cerulean, to form fluorophores with different spectral properties, makes it very

useful to simultaneously visualize and study protein interactions and expression patterns in the

same cells or organisms.45, 77 A fluorescence complementation based FRET assay has recently

been developed for the visualization and identification of ternary protein complexes.65 There have been an increasing number of studies every year for simultaneous study and visualization of protein-protein interactions in different cell lines and organisms based on fluorescence complementation.

However, the main drawback of this method is that the acquisition of cellular fluorescence from reassembled fragments is quite slow (1-3 days). Because of the slower rate limiting step, oxidation of the chromophore (Figure 1.8), GFP itself requires several hours to mature in the cell. Regardless of cell types and organisms this method requires reasonably low level expression of fusion proteins at 30 °C or at lower temperature and incubation at room temperature for about 4 to 48 h before it evolves any significant cellular fluorescence. The method is also sensitive to the concentration of fusion protein fragments in the cell. Because the dissected fragments are nearly insoluble, expression of the fusion proteins at higher temperature

21

(over 30 °C) may cause formation of inclusion bodies and aggregation, and fail to evolve cellular fluorescence. Fluorescence intensity does not necessarily correlate quantitatively to the interaction affinity of the proteins. Because of the slower fluorescence maturation and irreversibility of the process, fluorescence complementation techniques may not be ideal for real time detection of complex formation. FRET and BRET based techniques are proven to be very useful for the real-time detection of protein-protein interactions, complex formation and dissociation in vivo. However, these techniques are limited by relatively high background of

cellular autofluorescence and as well as by the direct excitation of the fluorescence acceptor.

They also require a higher level of protein expression and are technically and experimentally

demanding.

1.1.4 Need for an efficient and faster split fluorescence reassembly

Though fluorescence complementation assay has been used increasingly for detection and

visualization of protein-protein interactions, the widespread use of this technique has been limited

by a number of factors. Fluorophore formation and maturation driven by cyclization and

oxidation reactions is quite slow. In almost all fluorescence reassembly assay driven by different

spectral variants of GFP reported to date, incubation of the cells or organisms for about 4-48

hours at lower temperature (room temperature or 20 °C) is required. The optimum temperature

for the expression of GFP and most of the spectral variants is 30 °C or lower. In addition to that,

the solubility issue posed by dissected fragments when expressed at higher temperature limits this

assay to be carried out at physiological temperature, 37 °C. Magliery et al. reported that the

optimum cellular fluorescence in E. coli was observed after overnight growth at 30 °C and

followed by 1-2 days at room temperature, or after 3 days all incubation at room temperature.43 It

failed to evolve fluorescence at all for some interactions (such as barnase and barstar; C.G.

Wilson, TJM, and L. Regan, unpublished). Hu et al. reported that the cellular fluorescence was

observed after 36 – 48 h of transfection.44 The same group in 2003 reported the development of 22 multicolor fluorescence complementation and observed cellular florescence after 24 h growth of

COS-1 cells at 37 °C and 0-24 h incubation at 30 °C using inverted fluorescence microscope with charged coupled device (CCD) camera.45 Hu et al. in 2006 reported the use of split Venus based complementation assay at 37 °C. It still required 24 h of incubation to observe fluorescence under

CCD camera, and also, it was reported that split Venus fragments are comparatively more prone to self reassembly.78 Remy et al. demonstrated the large-scale library screen based on cellular fluorescence using FACS after 48 h of transfection.73 Chalfie and coworkers first reported the use

of split-GFP assay in C. elegans and observed fluorescence after 8-24 h growth at 20 °C.46 Hu et

al. in 2008 performed similar experiment to detect protein interactions and expression patterns in

C. elegans and reported that it is possible to observe fluorescence using fluorescence microscope

about 3 h after young worms were heat shocked at 33 °C and incubated at 20 °C.64 Authors also reported that it may require the induction of worms at 20-26 °C for 24-48 h before it evolves detectable fluorescence. Most of these split GFP based screenings work only at 30 °C or lower partly because of the fact that individual fragments are not stable at higher temperature.

Therefore, the slower maturation of fluorophore and the inability to perform this assay at physiological temperature limits the widespread use of fluorescence complementation for studying protein-protein interactions. It is highly desirable that the fluorescence complementation method needs to be improved for faster and efficient reassembly with robust fluorescence which will enable it to detect without the use of a CCD camera. Detection of fluorescence under UV illumination will make it faster for high-throughput processing. With increasing demand of studying and detection of protein interactions in vivo in a more native like environment and at

physiological condition, the search for a pair of GFP fragments that will allow fluorescence

complementation at 37 °C with the shortest possible time for reassembly is required. It is also

required that the fragments will not reassemble spontaneously to eliminate the background

fluorescence. The total assay time should be limited to the time required for sufficient growth of

23 the cells, not by reassembly or fluorophore maturation time. These possible improvements will make fluorescence complementation a powerful alternative technique to study and simultaneously visualize the protein-protein interactions in vivo in their native cellular milieu.

1.2 Protein-protein interactions: small molecule drug discovery for cancer therapy

1.2.1 Potential targets for drug discovery for cancer therapy

Proteins frequently interact with themselves or other proteins and function as stable or transient complexes to regulate a wide variety of cellular processes. Many of these protein-protein interactions are part of larger cellular networks and are tightly regulated by a number of different mechanisms. Alterations in protein-protein interactions or protein interaction networks from mutation or deregulation of one of the partners or both perturb the normal sequence of events in the cell and contribute to diseases. Experimental methodologies and in silico analysis has been using extensively for detail characterization of individual protein-protein interactions in regard to its structure, biological role and function. It is evident that almost every pathway in cell biology is critically mediated and regulated by protein-protein interactions.

Therefore, small molecule modulators of protein-protein interactions offer novel possibilities for the development of potential anticancer drugs. It has been a very challenging job to develop small molecule modulators of protein-protein interactions with high affinity and specificity. The development of inhibitors of protein–protein interactions is a far more complicated process due to a number of specific challenges, including fairly large protein interaction surfaces (approximately 750 – 1500 Å2),79 lack of adequate information in molecular level about interaction surface, structure of the interacting partners, types of interaction taking place, etc.79, 80 It is evident that, despite large surface area involved in protein interactions, only a

small subset of residues in the interface contribute to the affinity and binding. These regions of

“hot spots” tend to be found on both sides of the protein interaction interface and are considered

to be very crucial.79 Therapeutic antibodies, peptidomimetics and small molecule inhibitors are 24 considered to be most effective emerging class of drugs for cancer therapy.

With huge development of genetic engineering and molecular biology, an increasing number of protein–protein interactions have been identified as potential therapeutic targets for the development of small molecule drugs for cancer therapy. The most notable ones are the protein- protein interactions that are critically involved in the regulation of cell proliferation and apoptosis or programmed cell death.81, 82 Apoptosis plays a critical role in the development and homeostasis

of multicellular organisms. Dysregulation of apoptotic pathway by inappropriate suppression or

activation of apoptosis causes the disruption of balance between cell proliferation and cell death

and leads to many diseases in humans, including cancer and tumorgenesis, autoimmunity, and

neurodegenerative disorders.83 In fact, in many human cancers, proapoptotic proteins are found mutated or the expression of antiapoptotic proteins are upregulated, which lead to unchecked proliferation of cells to tumorgenesis. This is considered to be the most common mechanism by which cancer cells avoid cell death. In most of the cases these cancer cells are resistant to chemotherapy. Restoring drug-induced apoptosis signaling pathway to induce the cell death has been proven to be very effective in different animal models to destroy the cancer cells.81, 82

Apoptosis takes place by both extrinsic and intrinsic pathways through a strictly regulated

cascade of signals mediated by protein-protein interactions. Therefore, targeting protein-protein

interactions in the apoptosis pathway is a promising approach for drug discovery. There are a

number of potential protein-protein interaction targets in the intrinsic apoptotic pathway that have

been studied extensively and explored for small molecule drug discovery, including interaction

between Bcl-2/Bcl-xL and proapoptotic Bcl-2 family proteins (Bim, Bak, Bid, Bax, Bak), X-

linked inhibitor of apoptosis proteins (XIAP) and their modulator Smac (DIABLO), p53 and its

negative regulator hDM2 (human homolog of maurine double minute-2, MDM2), BRCA1 and its

stabilizing partner BARD1. These proteins play direct role in cell proliferation and apoptosis and

are considered to be the most effective targets for cancer therapy.

25

Figure 1.10: (a) Sequence homology between Bcl-2 family proapoptotic (Bcl-xL, Bcl-2, Bcl-w) and antiapoptotic (Bax, Bak, Bim , Bid, Bad) proteins. (b) High resolution NMR structure of Bcl- xL (1-197) and Bak peptide (72-87) (PDB:1Bxl). BH1, BH2 and BH3 domains are shown in yellow, red and cyan, respectively. Bak peptide is shown in orange. (c) Crystal structure of Bcl-xL (1-197) and Bim peptide (83- 115) (PDB:1Pq1). Bim peptide is shown in red.

Figure 1.11: Structures of ABT-737 and ABT-263, small molecule inhibitors of Bcl-2 family proteins. Structures colored in red indicate the places of chemical modification on ABT-737 to obtain ABT-263. The 84 crystal structure of the Bcl- xL bound to ABT-737 (PDB:2YXJ) is shown on right.

26

1.2.2 Interaction of Bcl-2, Bcl-xL with proapoptotic proteins Bim, Bid, Bax, Bak

Members of the Bcl-2 (B-cell Lymphoma-2) family of proteins are critical regulators of apoptosis by either inhibiting or promoting cell death and are involved in a number of cancers such as melanoma, breast, prostate, and lung carcinomas. Bcl-2 family proteins are composed of antiapoptotic proteins (Bcl-2, Bcl-xL, Bcl-w, Bcl-2L1, Mcl-1 and A1) and proapoptotic proteins

(Bax, Bim, Bid, Bak, etc.).85-87 These proteins share conserved regions known as BH (Bcl-2

Homology) domains and cooperate through protein-protein interactions to regulate the intrinsic apoptotic pathway, Figure 1.10.87-89 Protein-protein interactions and balance between these opposing factors determine whether cells live or die. The pro-apoptotic family members are subdivided into two structurally distinct classes; proteins (Bax and Bak) with multiple BH domains (BH1, BH2 and BH3) and the BH3-only proteins (Bim, Puma, Bid, Bad, Bik, Noxa).

Biological function of antiapoptotic proteins are to interact with proapoptotic proteins and inhibit them from their functions and protect cells from apoptosis.90, 91 For both classes of proteins, BH3 domains are necessary for high affinity binding and heterodimerization with antiapoptotic proteins and promote mitochondria-initiated apoptosis when overexpressed. The molecular mechanism by which the pro- and antiapoptotic proteins initiate apoptotic pathway is unclear. But it is believed that upon a multitude of stimuli, including DNA damage, BH3 only proteins initiate

apoptosis either by activating proapoptotic Bcl-2 proteins or by binding and inhibiting

92 antiapoptotic family members. The current model suggests that Bcl-xL/Bcl-2 or BH3 only proteins activate the Bax/Bak homooligomerization on mitochondria, which in turn activates the release of apoptogenic factors, cytochrome c or Smac/DIABLO. Cytochrome c activates the apoptosome and caspase mediated pathway to cell death, Figure 1.12.93

It is reported that in variety of human cancer cells Bcl-2 family proteins are found to be overexpressed which contribute to tumor initiation, progression and resistance to therapy.94

Apoptosis is initiated by the interaction of BH3 only proteins through a BH3 α-helix into the

27 large hydrophobic pocket on antiapoptotic proteins. Therefore, inhibiting pro-survival proteins

Bcl-xL, Bcl-2 and Bcl-w has been a potential target for anticancer therapy. Molecules that mimic

the BH3 domains and bind to the hydrophobic groove of the antiapoptotic proteins can be used as

potential anticancer drugs. Applying structure-based and fragment-based approaches, and using

various high-throughput or semi-high-throughput screening, several groups have reported the

discovery of a number of potent small molecule inhibitors of Bcl-2/Bcl-xL/Bcl-w with low micromolar to nanomolar affinity (obatoclax mesylate, 95 terphenyl scaffold, antimycin A,96

chelerythrine,97 YC-137,98 HA14-1,99 gossypol100). In a different approach, Klasa et al.101

designed an antisense oligonucleotide (Oblimersen); Holinger et al.102 and Wang et al.90 designed

BH3 peptide mimic; and Walensky et al.103 designed a hydrocarbon “stapled” BH3 helix to inhibit antiapoptotic proteins to induce apoptosis. Recently, using structure and fragment- based design, and high-throughput NMR screening, Rosenberg and coworkers from Abbott

Laboratories, discovered ABT-737, which is by far the most potent small molecule inhibitor of

104 the anti-apoptotic proteins Bcl-xL, Bcl-2 and Bcl-w with high affinity (Ki ≤ 1 nM). ABT-737

alone exhibits cytotoxicity of lymphoma and small-cell lung cancer (SCLC) cell lines and

primary patient-derived chronic lymphatic B-cell leukaemia cells. In animal models it was shown

to cause regression of solid tumors. ABT-737 antagonized Bcl-2 protection at concentrations

about 10 nM, but did not induce cytochrome c release at concentrations up to 1 µM. Using

mammalian two-hybrid system they have also demonstrated that it was able to disrupt Bcl-2

family protein-protein interactions in vivo by 45% and 55 % at about 0.10 µM and and 1 µM

concentrations, respectively. However, ABT-737 is not orally bioavailable and has very low

84 aqueous solubility. Based on crystal structure of ABT-737/Bcl-xL (PDB:2YXJ), the same group

from Abbott Laboratories chemically modified ABT-737 (exchange of nitro group for

trifluoromethylsulfonyl group) and redesigned the second generation Bcl-2 proteins inhibitor,

ABT-263, with better pharmacokinetics, Figure 1.11. ABT-263 disrupts Bcl-xL/Bcl-2

28

Figure 1.12: Role of Bcl-2 family proteins and IAP/XIAP in extrinsic and intrinsic pathways of apoptosis.105 In response to multiple stimuli intrinsic pathway triggers the activaton of Bax/Bak, which in turn regulates the release of apoptogenic factors, cytochrome c and smac, from intermembrane space of mitochondria. Smac and cytochrome c activate the procaspase 9 and effector caspase 3 and 7 and initiate the caspase mediated apoptosis pathway. BH3 only proteins Bim, Bid or Bad etc bind with proapoptotic proteins Bcl-2, Bcl-xL etc. and inhibit the activation of Bax/Bak. Downstream of signaling pathway IAP/XIAP binds with smac and inhibits the activation of caspases. Cell proliferation or apoptosis is strictly regulated by the balance between these proapoptotic and antiapoptotic proteins. Figure was reproduced from Makin et al. from Trends in Molecular Medicine, 2003.105

interactions with proapoptotic proteins (e.g., Bim) and exhibits initiation of apoptosis to Bcl-

2/Bcl-xL–dependent cells in vitro. It elicits complete tumor regressions in xenograft models of

SCLC and acute lymphoblastic leukemia. ABT-263 is currently undergoing phase II clinical trials for the oral treatment of variety of tumors.106 Though these results are very promising with some negative in vivo effects, there is still a huge demand for discovery of new Bcl-2 proteins inhibitors with better druggability and pharmacokinetic properties.

1.2.3 Interaction of IAP, XIAP with Smac (DIABLO)

IAPs (inhibitor of apoptosis proteins) are another class of regulatory proteins that are critically involved in cell proliferation or apoptosis and cell death through caspase (cysteine- 29 aspartate proteases) mediated mitochondrial apoptotic pathway. Like Bcl-2 family antiapoptotic proteins, inhibitor of apoptosis proteins (IAP) are found to be upregulated in many cancer cell lines and have been implicated in tumor growth, pathogenesis, and resistance to chemo- or radiotherapy.107-109 Overexpressed IAP avoids programmed cell death through their ability to

directly inhibit members of the caspase family of apoptotic enzymes. Human XIAP (X-linked

Inhibitor of Apoptosis Protein), the best characterized IAP family protein, selectively binds

caspases 3, 7 and 9, and functions as a negative regulator of apoptosis, Figure 1.12. It contains

three BIR (Baculovirus IAP Repeat) domains (BIR1, BIR2 and BIR3) and a C-terminal RING

finger domain. BIR2 (163-234) and linker (124-162) between BIR1 and BIR2 domains of XIAP

bind to, and inhibit the effector caspase 3 and 7, which prevents normal substrate processing and

leads to the suppression of apoptosis.110-113 On the other hand, BIR3 (256-349 aa) domain of

XIAP binds directly and specifically to the small subunit of initiator caspase 9 and prevents the

formation of catalytically active of caspase 9 homodimer.114 Because of its potent caspase inhibitory functions, XIAP is considered to be a potential target for cancer therapy. Studies reported that overexpression XIAP or XIAP knockout by siRNA or antisense oligonucleotides makes cells more vulnerable to proliferate uncontrollably or to apoptosis and cell death, respectively.115, 116 Smac (second mitochondria-derived activator of caspases) also known as

DIABLO (Direct IAP-Binding protein with low pI) is a potent proapoptotic protein which functions as a natural endogenous antagonist of IAPs.117, 118 In response to apoptotic stimuli,

Smac is released from mitochondria, directly binds to IAPs and prevent them from inhibiting

caspases which induce apoptosis. Previous studies demonstrated that Smac interacts with BIR3

domain of XIAP in the same binding groove where caspase 9 binds via its N-terminal AVPI motif

which removes the caspase 9 by direct competition with BIR3. High-resolution crystal and NMR

structures of Smac/DIABLO complexed with the third BIR3 of XIAP showed that the N-terminal

four residues (Ala-Val-Pro-Ile) in Smac recognize a surface groove on BIR3, with the first

30 residue Ala binding into a hydrophobic pocket and making five hydrogen bonds to neighboring residues on BIR3 (Figure 1.13).110, 115, 116 Modeling studies suggest that full length Smac extends

its binding to XIAP BIR2 domain and prevents the binding of XIAP to caspases 3 and 7.110, 111, 119

Figure 1.13: Figure a shows the NMR structure of XIAP-BIR3, (PDB:1G3F) domain (258-346) and N- terminal Smac peptide as shown as blue and red color atom type (AVPI).115 Figure b shows the crystal structure of XIAP-BIR3 domain and smac (1G73).116 N-terminal regions of smac proteins (colored blue and green) are bound to the BIR3 domains (colored orange and yellow) of XIAP.

Recent studies reported that Smac forms a homodimer and inactivates XIAP by binding to both the BIR2 and BIR3 domains and promotes the activity of caspase 9, and also caspase 3 and 7.120 Therefore, inhibition of XIAP and other IAPs by Smac mimetic peptides has been

considered as a potential target for cancer therapy. Several Smac peptide mimetics and small

molecule inhibitors have been reported to induce apoptosis and treat cancers in vitro and in vivo.113, 121-124 These Smac mimetic XIAP inhibitors act by binding more strongly to the BIR3

domain of XIAP which release and activate the caspases and induce apoptosis. Oost et al. in 2004

reported the designing of capped tripeptides with low nanomolar affinity to the BIR3 domain of

XIAP.125 They have shown that these inhibitors potently induce caspase activation and apoptosis

in several human cancer cell lines and inhibit the growth of the tumors in xenograft mouse

models. In a more recent study, Cossu et al.127 reported the discovery of potent small molecule

inhibitors by screening from a library of 4-substituted azabicylo alkanes (library was designed 31

Figure 1.14: Chemical structures of potent XIAP inhibitors designed from lead compounds smac001 obtained from the screening of a large library of small compounds.126, 127

based on a known inhibitor smac001) and designed a more potent inhibitor (smac0037) based on the structures of the lead compounds, Figure1.14. Wu et al. discovered a novel small molecule

(TWX024) from high-throughput biochemical screen of a combinatorial chemical library to disrupt the XIAP/caspase-3 interaction.128 The activity of TWX024 has been characterized both in

vitro and in cells and shown that in 50% of the cells it can induce apoptosis. Using structure-

based design, Fesik and coworkers designed and synthesized a number of Smac mimetic small

molecules (Figure 1.14) with micromolar to nanomolar binding affinity.129 Wang et al., Zobel et

al. and several other groups reported the discovery of small molecule inhibitors for XIAP130-132

and examined their potential in inducing apoptosis in several cancer cell lines. In a recent study

Sun et al.133 reported a bivalent compound (SM-164126 Figure 1.14) and demonstrated that it can

bind with BIR2 and BIR3 domains simultaneously with high affinity in vitro (IC50 ~ 1.4 nM). It

was also shown that it inhibited cellular proliferation and induced apoptosis of human leukemia

cells at a concentration as low as 1 nM. These stuides clearly validate the use of small molecule

XIAP antagonists against cancers that overexpress XIAP and provide insights into the 32 development of therapeutic agents that act by modulating apoptotic pathways. For therapeutic applications, however, it is highly desirable to improve their stability, cell permeability, and pharmacokinetics.

1.2.4 Interaction of hDM2 with p53

Because of its tumor suppressor activity, p53 has long been considered as an attractive therapeutic target for cancer treatment. p53 functions as a transcription factor and is found to be implicated with wide variety of cellular functions such as DNA damage repair, cell cycle regulation and apoptosis to maintain genomic integrity.134 In response to stress, p53 enables the

activation and expression of an array of genes which play critical role in major pathways

protecting cells from malignant transformation. Upon signal from oncogene overexpression or

failing to repair DNA damage, p53 induces cell cycle arrest and cell death by apoptosis. It has

been shown that cells that lack p53 (Tp53-) or functional p53 (mutated or deleted) are unable to

respond properly to stress and are prone to tumors.135, 136 The cellular level of p53 is tightly regulated by its natural binding partner DM2 (double minute 2, human homolog is hDM2) through its E3 activity.137, 138 DM2 binds to the N-terminal transcriptional activation domain of tumor suppressor p53 and negatively regulates its activity and stability.139

MDM2 and p53 are involved in an auto-regulatory feedback loop in which both proteins mutually

control their endogenous cellular concentrations. The expression of mdm2 gene is controlled by

p53 under both normal and stressful conditions. Therefore, levels of MDM2 rise as p53 is

activated and stabilized. The elevated level of MDM2 regulates p53 protein activity by physically

binding to p53 and blocking its transcriptional activity, inducing nuclear export and stimulating

the degradation of p53 by the E3 ubiquitin ligase and proteosome degradation pathway. Cell

cycle progression, and genomic integrity and stability are tightly regulated by both MDM2 and

p53 and an overall balance between them. Over expression of MDM2 protein has negative

consequences for the cell by inactivating the p53 which results unchecked cell proliferation. In 33

Figure 1.15: (a) X-ray crystal structures of MDM2 (1-109) and N-terminal p53 peptide (17-29), PDB: 1YCQ.140 (c) Cis-Imidazoline analog inhibitors (Nutli1, Nutlin2 and nutlin3) of MDM2, which are designed from the lead compounds identified from the screening of a large library of synthetic molecules. (b) X-ray crystal structure of MDM2 (1-109) complexed with Nutlin2, PDB: 1RV1.141

over 50% of human cancer cells, p53 was found o be mutated or deleted.137, 138 In all other cancer

cells functional wild type p53 was inactivated by binding with the hDM2 (human isoform). The

increased level of hDM2 found in many human tumor cell lines suggests its involvement in

cancer.139 In tumors, gene amplifications and other alterations can result in elevated hDM2 and

lead to the constitutive inhibition of p53. Therefore, strategies to block the interaction between

hDM2 and p53 to restore and activate the p53 mediated apoptosis pathway could be the important

target for cancer drug discovery to induce apoptosis and growth inhibition of human tumor cells.

Kussie et al. reported the X-ray crystal structure of MDM2-p53 complex (Figure 1.15)

and showed that N-terminal region of p53 forms an α-helix from which the side chains from three

amino acids F19, W23 and L26 project into the hydrophobic pocket of MDM2, which promotes

the interaction of p53-hDM2.140 Based on the crystal structure, several groups reported the designing of peptide mimetics which bind tightly to hDM2 and inhibit its interaction with p53, 34 and showed that they exhibit cytotoxicity to human cancer cells.142, 143 However, these compounds lacked the ability to penetrate cells and did not possess drug-like properties. Verdine and coworkers144 in 2007 reported a “stapled” α-helical peptide of p53 that binds hDM2 with subnanomolar affinity and showed the activation of p53 mediated apoptosis pathway in Jurkat cells. Using structure-based design and chemical modifications, several groups have also reported the discovery of small molecule inhibitors hDM2/p53 interaction. Stoll et al.145 reported the chalcone derivatives and most recently Shangary et al.146 from Ascent Therapeutics reported more potent spiro-oxindole with excellent selectivity with nanomolar affinity for binding with hDM2.

Spiro-oxindole exhibited activation of p53, which led to cell cycle arrest and apoptosis in human tumor cells and mouse xenograft models. Duncan et al.147 reported the discovery of fungal metabolites chlorofusin which inhibits p53-MDM2 interaction with an IC50 of 5 µM. Based on

computational approaches Holak and Weber, and Carlson and Wang screened out, designed and

tested small molecules with low micomolar to low nanomolar affinity.125, 129, 148 Other small molecules identified to date are sulfonamides and benzodiazepines that bind weakly to MDM2 and displace p53 peptides.149-151 Vassilev and coworkers more recently discovered the most potent

small molecule inhibitors, cis-imidazoline analogs (called nutlins, Nutlin1, Nutlin2 and Nutlin3),

141 of p53-MDM2 interaction with IC50 of 100 nM. They have also showed that nutlin caused the

accumulation and activation p53 and subsequent cell cycle arrest and apoptosis in tumor cells.

The X-ray crystal structures of MDM2 and p53 peptide, and Nutlin2 and MDM2 are shown in

Figure 1.15. Among cis-imidazole analogs, Nutlin3 was found to be the most potent, which

exhibited inhibition of tumor growth in xenograft mouse model without causing significant

toxicities. These results indicated that the peptidomimetics or small molecule inhibitors of hDM2

could be useful for the treatment of cancer. However, because of lower druggability and poor

pharmacokinetic properties, the potential of these molecules as therapeutic drug needs to be

thoroughly examined. Therefore, there is still an immense need to discover potent small molecule

35 inhibitors with more drug like properties in the p53-mediated pathway cancer therapy.

1.2.5 Fluorescence reassembly: Emerging tool for protein-protein interactions and drug

discovery for cancer therapy

The success of target-based approaches to drug discovery relies on identification of a potential biological target and discovery of a therapeutic agent that specifically modulates the activity of this target with limited or no adverse effects in humans. There are a limited number of methods based on high-throughput or semi-high-throughput screening approaches available for small molecule drug discovery, including structure-based design,141 in silico screening

(QSAR),152, 153 and fragment based drug design.154 These methods have been successfully used to mainly identify initial leads for modulation of several therapeutically important protein-protein interactions. Extensive chemical modifications on these compounds have led to the discovery of several anticancer drugs, some of which are in clinical trials. Since its introduction by Bohm and coworkers, fragment-based screening has been a very popular method to develop small molecule drugs. In this method, the low affinity binders were screened from a library using high-throughput

NMR approach. Low affinity binders from two complementary sites were then combined through tethering based on structure-activity relationship to develop a hybrid high affinity inhibitor.155

Using this approach, Fesik and coworkers104 reported discovery of ABT-737, a small molecule inhibitor of anti-apoptotic Bcl-2 family proteins (Bcl2, Bcl-xL, Bcl-w) with very high affinity (Ki

~1 nM). Vassilev and coworkers141 discovered a small molecule called Nutlin3 from a diverse

library of synthetic chemicals based on crystal structure of p53 peptide and MDM2 with IC50 value about 100 nm. A number of computation-based approaches are available to predict the PPIs and types of PPIs from genome wide sequencing.156-159 These experimental approaches and

computer aided modeling of interaction surfaces have been used to design small molecule drugs

with limited success. The success rate was limited principally because of the screening of initial

36 leads using high-throughput NMR or high-throughput crystallography is time consuming, very expensive and requirs highly specialized labs. These methods are in vitro and need large quantities of purified proteins, which, in most of the cases, is a major challenge. Since they have been designed based on in vitro screening methods, critical issues with small molecule inhibitor

identified to date have been the lower druggability and poor pharmacokinetic properties. These

difficulties suggest that large, functionally-diverse libraries might be essential for finding unique

molecules that are capable of modulating specific protein–protein interactions with higher affinity

and with more drug-like properties. This can be achieved by screening large libraries of small

molecules or peptides or cyclic peptides, against specific target in natural cellular environment.

The major challenge is the lack of functional high-throughput in vivo screening methods for

detection of small molecule modulators. Because of the absence of simple and easy to operate

high-throughput screening assays, this field of novel drug discovery targeting protein-protein

interactions has been very slow. Therefore, efficient and reliable high-throughput screening

methods are necessary to study protein-protein interactions in vivo to discover therapeutic drugs.

In recent years, protein fragment complementation assay has drawn interest for its

simplicity, convenience and ease to use for therapeutic drug discovery targeting protein-protein

interactions. Split GFP or fluorescence complementation assay would be especially useful for

high-throughput screening of small molecule or peptide libraries in vivo from direct fluorescence

read out. Macdonald et al. used this technique to screen out four different drugs which were

proven to have antiproliferative effect on five different human tumor cell lines.160 Sarkar et al.

(manuscripts in preparation, included in Chapter -3) re-engineered and developed an improved split GFP system which possesses faster (overnight growth of the cells) and efficient fluorescence reassembly with robust fluorescence signal. Improved split GFP fragments successfully detect the interactions and evolved cellular fluorescence for several therapeutically important protein- protein interactions such as Bcl-xL/Bim, Bcl-2/Bim, p53/MDM2, XIAP/Smac and

37

BRCA1/BARD1. Two known small molecule inhibitors, ABT-737 (for Bcl-xL/Bim, Bcl-2/Bim

interactions) and Nutlin3 (for p53/hDM2 interactions) were tested using the new system. The

new efficient split GFP fragments were able to respond to the inhibitors in a concentration

dependent manner. These results validate the potential of split GFP technique as an efficient

screening method for small molecule drug discovery. There are a number of advantages which

make the fluorescent complementation assay very attractive. The development of the split GFP

system is relatively simple and does not require the production of recombinant proteins and

protein purification, which is often a very difficult problem for the development of large-scale in

vitro-based assays of drug targets. It does not require the use of expensive equipment like NMR

or crystallography. It enables the use of a combination of drugs which may affect multiple targets

simultaneously. Primary screens are carried out in living cells in their native state. The most

remarkable advantage of this approach is that it can target any protein-protein interactions in a

wide variety of cell types and living organisms and in almost any subcellular localization. Chalfie

and coworkers, and several other groups have studied PCA in whole living organisms (C.

elegans, S. cerevisiae and D. melanogaster). These studies demonstrate the broad potential of this

approach, as well as its applicability to target discovery and drug screening. In order to carry out

fluorescence based screens, small molecules have to be membrane permeable and therefore more

likely to be bioavailable. Fluorescence reassembly methods can be used to aid our understanding

of molecular mechanisms and action of drug in human cells and has the potential to enhance the

productivity of therapeutic drug discovery targeting protein-protein interactions.

1.3 BRCA1/BARD1 interactions: Involvement in breast and ovarian cancers

1.3.1 BRCA1/BARD1: Background

By gene mapping and family linkage analysis, Mary-Claire King in 1990 first identified a

gene (17q21) related to breast and ovarian cancer, which was later termed as BRCA1 (BReast

CAncer associated gene 1).161 In 1994 Mark Skolnick and coworkers at Myriad Genetics 38 reported the cloning of BRCA1 cDNA.162 In the same year Stratton and coworkers identified the

BRCA2 gene located at 13q12-13.163, 164 About 10% of women diagnosed with breast cancer have

inherited mutations in BRCA1162 or BRCA2164 tumor suppressor proteins, that have been

associated with predisposition to breast and ovarian cancer.165 Mutations of BRCA1 gene are associated with more than half the cases of hereditary breast cancer. Several studies reported that women who inherit mutations in one of the BRCA gene (i.e., heterozygous) have a significantly increased (~85%) lifetime risk of breast and/or ovarian cancer before 50 years of age.166, 167

Usually the inheritance of a mutant copy of the gene and the cancer occurs when the second copy

of the gene is mutated or lost.168 This theory is supported by mouse model experiment carried out

by Ludwig and coworkers169 and Deng and coworkers.170 In their experiment they have shown that the tissue-specific inactivation of either BRCA1 or BRCA2 leads to mammary carcinogenesis.171-175

1.3.2 Interaction of BRCA1 with BARD1 and putative physiological role

BRCA1 tumor suppressor protein has been implicated with a wide array of fundamental cellular processes involving homologous recombination repair of DNA damage, chromatin remodeling, cell cycle check points and regulation, transcriptional regulation, regulation of centrosome duplication and apoptosis.176-182 To date, BRCA1 has been found to be implicated in

interactions with more than 100 different proteins and identified as part of three different

macromolecular assemblies.180, 183-185 Many of these interactions were found to be regulated by both spatially and temporally in response to DNA damage183, 186-188 and also in a cell cycle-

dependent manner.168, 184, 188-190 In most of these cellular processes BRCA1 was found to be

associated with a protein, called BARD1 (BRCA1 associated RING domain 1), discovered in

1994 in a yeast two-hybrid screen as a binding partner of BRCA1 to form a stable heterodimeric

complex.191 Although BRCA1 has been reported to form a complex with a number of other

proteins, its interaction with BARD1 is remarkable in terms of its stoichiometry and stability. 39

Figure 1.16: Functional domains and structural homology between BRCA1 and BARD1proteins. Both proteins carry nuclear localization signal sequences and nuclear export signal sequences, and two BRCT (BRCA1 carboxyterminal) domains at their C-terminus. The RING finger domain of BRCA1binds to RING finger domain of BARD1 and this binding enhances the ubiquitin-ligase function of BRCA1.

BRCA1 and BARD1 complex has been shown to colocalize within nuclear foci during S-phase of cell cycle.169, 189 Using co-immunoprecipitation experiments, Wu et al. confirmed that more than

75% of the endogenous cellular BRCA1 exists in vivo in the form of a BRCA1–BARD1 heterodimeric complex.191 It is now evident that BARD1 is required for the stability of BRCA1

protein in vivo and most of its diverse cellular functions.192 Ludwig et al. showed that BARD1-

deficient mice exhibited similar phenotypes to that of BRCA1 and causes early embryonic

lethality and chromosomal instability.193 Both of these proteins have striking structural similarity

as shown in Figure 1.16. BRCA1 is an 1,863 amino acid multidomain protein and consists of N-

terminal RING finger domain, nuclear localization signal (NLS), nuclear export signal (NES) and

two tandem C-terminal BRCT (BRCA1 Carboxy Terminal) domains. BARD1, on the other hand,

is a 777 amino acid protein and contains NES, NLS and N-terminal RING finger domain and two

C-terminal BRCT domains. BRCA1 interacts with BARD1 through their N-terminal RING

domains which includes their RING motifs and the antiparallel α-helices formed by the sequences

encompassing the RING motifs.191 These antiparallel α-helices (immediately followed by RING

domains) form a four-helix bundle to form BRCA1/BARD1 heterodimeric complex. Interaction

chiefly takes place through RING motifs by forming a four-helix bundle. In both proteins, RING

motifs are characterize by a conserved pattern of eight residues, C3HC4, arranged in pairs in

40 primary sequences and form two Zn2+ binding sites, Figure 1.17. Klevit et al. in 2001 reported the NMR solution structure of N-terminal RING-RING complex of BRCA1/BARD1 heterodimer.194 The BRCA1 RING motif is characterized by a short antiparallel three-stranded β- sheet, two large Zn2+ binding sites and a central α-helix (Figure 1.17). The BARD1 RING motif is structurally homologous but lacks a central helix between the third and fourth pairs of Zn2+

ligands. Regions adjoining the central RING motifs are critical for the proper association of the

two proteins.195, 196 Residues 8–22 and 81–96 of BRCA1 form antiparallel α-helices that flank the

Figure 1.17: Solution NMR structures of RING domains of BRCA1 (1-109 aa) and BARD1 (26-140 aa) with their Zn2+ (gray spheres) binding sites are shown at the top. BRCA1/BARD1 dimerize by forming an antiparallel four-helix bundle through the sequences encompassing RING motifs (PDB:1BXL).194 The Figure at the bottom shows the CXXC----CXXC motif for first Zin2+ (Site I) and CXH---- CXXC motif for the second Zn2+ binding site, and their sequence alignment and conserved residues for BRCA1 and BARD1.

41 central RING motif (BRCA1 residues 23–76). Likewise, helices formed by residues 36–48 and

101–116 in the BARD1 subunit also bracket its central RING motif (BARD1 residues 49–100).

Klevit et al. also reported that heterodimerization is primarily governed by extensive interactions between BRCA1 and BARD1 N-terminal subunits within the hydrophobic core of the four-helix bundle.194 Side chains of residues L3, L6, V8, V11, V14, I15, A17, M18, I21, L22, L82, L86, I89,

A92 and L95 of BRCA1, and residues A33, W34, H36, A40, L44, L47, L48, L101, M104, L107,

L111, L114 and L115 of BARD1 are either partially or mostly buried at the subunit interface.

These interactions may be augmented by several other polar interactions as it is observed that side chains of R7, E10, E85 and D96 in BRCA1 are close to the side chains of D117, K110, R43 and

H36 of BARD1, respectively.

1.3.3 BRCA1/BARD1 RING domain heterodimeric complex: E3 ubiquitin ligase activity

and target substrate

Despite huge effort, the molecular mechanism by which the loss of BRCA1 or its function promote tumor formation is still unclear. Significant advances in understanding the mechanism of BRCA1–BARD1 function came from a number of recent studies in which it is reported that this heterodimer has an enzymatic role and acts as E3 ubiquitin ligase.169, 192, 197-199

Hashizume et al. investigated that the individually BRCA1 and BARD1 have very low ubiquitin ligase activities in vitro, whereas BRCA1/BARD1 complex has dramatically much higher ligase

activity. Later this group, Klevit et al. and Solomon et al. showed that the N-terminal BRCA1 (1-

109) and BARD1 (26-140) only are capable of forming a functional RING domains to form four-

helix heteroduplex and maintaining its E3 ligase activity.169, 194, 200-202 Ubiquitination occurs through a sequential process involving E1 ubiquitin-activating enzyme, E2 ubiquitin-conjugating enzyme, and an E3 ubiquitin ligase. In the final step of this process, an ubiquitin monomer is transfered covalently to a lysine residue of the ultimate substrate by isopeptide bond formation. In many cases, the monoubiquitinated substrate will then serve as a nucleus for assembly of a 42 polyubiquitin chain. Generally, polyubiquitination is restricted to certain lysine residues of ubiquitin, such as K48 (in K48-linked chains), K63 and K29. Polyubiquitination using K48 of Ub commonly serves as a signal for rapid degradation of protein by the 26S proteasome. Although definitive substrates of BRCA1/BARD1 have not yet been identified, autoubiquitination of the

BRCA1 subunit itself is observed during in vitro and in vivo reactions catalyzed by

BRCA1/BARD1.203, 204 In addition to the ubiquitination on K29, K48 and K63 residues, BRCA1–

BARD1 heterodimer mediates an unconventional linkage of polymerization of ubiquitin primarily through K6 residue.203, 204 Ubiquitin ligase activity of the heterodimer in auto-ubiquitylation at

non-K48 ubiquitin residues is 20-fold higher than that of its substrate. It is believed that while

ubiquitin chains on K29 and K48 can target their conjugated substrates for proteasomal

degradation, K63-linked chains serve diverse non-proteolytic functions.204 However, the exact cellular functions of K6 polyubiquitin chains are not known. The autoubiquitination of BRCA1 and BARD1 stabilizes each other in vivo and enhances its E3 ligase activity. The K6 linked

chains are recognized by the 26S proteasome for deubiquitination in an ubiquitin-aldehyde-

sensitive manner as opposed to degradation.203 These observations suggest that the ubiquitination

mediated by BRCA1/ BARD1 could signal a process other than degradation. The

auotoubiquinated BRCA1/BARD1 complex efficiently monoubiquitinates nucleosome core

histones in vitro.205 Several studies reported the observation of autoubiquinated BRCA1/BARD1

complex at the sites of DNA replication and repair.165 This indicates that the complex might regulate BRCA1– BARD1 ubiquitin ligase activity and induce DNA repair pathways.

Although the ubiquitin ligase activity of BRCA1/BARD1 complex is very important for its role as a tumor suppressor, the mechanism by which this activity contributes to its biological function remains to be determined. The involvement of BRCA1 with wide array of cellular processes indicates that the BRCA1/BARD ligase may have variety of substrate candidates.

BRCA1 is required for maintaining genomic integrity, partially through its contributions to

43 efficient DNA double-strand break repair.181 Upon DNA damage, BRCA1/BARD1 complex was found to be colocalized with RAD51 and PCNA (proliferating cell nuclear antigen) in different nuclear foci, DNA damaged, replicating regions. Rad51 is a key component involved in the repair of DNA double-strand breaks in the homologous recombination.187 BRCA1’s association with

RAD51 in subnuclear clusters indicates the role for BRCA1 in DNA repair. BRCA1 was also found to be an important part of BASC (BRCA1 Associated genome-Surveillance Complex) which includes RAD50-MRE11-NBS (functions as exonuclease at the double-strand breaks) and

ATM (ataxia telangiectasia mutated, functions upstream of BRCA1 in the double-strand-break repair pathway).206-208 Upon irradiation or treatment with hydroxyurea, BRCA1 colocalizes with

RAD50-MRE11-NBS1 complexes at large nuclear foci that also contain PCNA.209 Studies by

Chene et al. and Mallery et al. reported that the histone H2AX as a possible target for

BRCA1/BARD1 E3 ligase.205, 210 They have shown that purified BRCA1/BARD1 complex

mediates the UbcH5α-mediated monoubiquitination of histone H2AX and other core histones in

vitro and in vivo. These results indicate the possible role for the BRCA1/BARD1 ubiquitin ligase

activity in DNA repair by chromatin remodeling. BRCA1 was also found to copurify with

SWI/SNF-related chromatin remodeling complex.211 In vivo and in vitro studies have shown that

BRCA1 can regulate gene transcription through its C-terminal BRCT domains. It was shown that

BRCA1 transcriptionally activates the expression of p21WAF1/CIP1 in a p53 dependent manner through SWI/SNF complex. Harkin et al. demonstrated the BRCA1’s role in transcriptional control in apoptosis signaling through induction of a proapoptotic protein GADD45.212 Under

normal conditions, BRCA1 interacts with ZBRK1 in an obligate manner and the complex

represses the transcription of GADD45.213 Upon DNA damage, ZBRK1 is ubiquitinated and

degraded, which releases the repressor, activates the expression of GADD45, and results in

apoptosis.214

BRCA1 has also been shown to regulate transcriptional activity negatively during

44 transcription-coupled DNA repair. In response to DNA damage, it has been reported that, RNA polymerase II is phosphorylated, ubiquitinated and degraded.185, 215 In vitro experiments by

Kleiman et al.216 and Starita et al.185 suggested that RNA Pol II holoenzyme and phosphorylated

largest subunit of RNA Pol II (POR2A) are ubiquitinated by BRCA1/BARD1 upon stimulation of

DNA damage. Kleiman et al. have also shown that, in both cases, after DNA damage

ubiquitination was significantly reduced by depletion of BRCA1 and BARD1 by siRNA

treatment. It is hypothesized that BRCA1–BARD1 E3 ligase activity controls cell-cycle

progression by targeting proteins of the stalled RNA Pol II complex. The BRCA1 BRCT motif

directly binds the BACH1 helicase and CtIP (carboxy-terminal-binding-protein interacting

protein) proteins and is required to inhibit cell cycle progression in the G2 and S cell cycle phases

after DNA damage.167, 217 Since BACH1 and CtIP are mutually exclusive partners of the BRCA1

BRCT domain, siRNA for each protein made it possible to achieve specific perturbation of a particular BRCA1/BARD1 super complex. BRCA1-regulated gene products have been implicated directly or indirectly in cell cycle regulation and DNA repair. Previous studies established that the expression of BRCA1 is regulated in a cell cycle-dependent manner. BRCA1 maintains low expression and steady-state level in resting (G0) cells and G1 cycling cells.218 But the BRCA1 expression level increases considerably as cells enter S phase. It is shown that in late

G1 stage there is a sharp increase in the level of BRCA1 mRNA. Therefore, it is believed that the higher expression level of BRCA1 protein G1/S in interphase and S phase is probably driven by the transcriptional activation of BRCA1.

Parvin et al. in 2006 reported the detection of centrosome amplification and hypertrophy in ductal carcinoma in situ (DCIS), which is reportedly the first stage of breast cancer. Later centrosome hyperactivity is found to be associated with more advanced breast tumors.185, 219 The

previous studies reported that BRCA1 is localized into centrosomes during all phases of cell cycle

and mitosis and has a very active role in regulating the centrosomes.174, 220 It is demonstrated that

45

Figure1.18: Functional diversity of BRCA1 and its regulatory pathways. BRCA1 is a key component of a number of important pathways that regulate DNA repair, cell-cycle progression, ubiquitylation and transcriptional regulation.168 BRCA1 interacts with BARD1 through N-terminal RING domains and form a stable heteroduplex which functions as E3 ubiquitin ligase whose physiological targets are not known clearly.

the transient inhibition of BRCA1 function caused a rapid amplification and fragmentation of centrosomes in cell lines derived from mammary tissue. Using BRCA1 knockout mice, Xu et al. demonstrated that loss of BRCA1 function caused abnormal centrosome amplification, defective

G2–M checkpoint control and genetic instability.174 It is hypothesized that BRCA1/BARD1 may

target a protein in the centrosome for ubiquitination during the mitosis, which may protect the

centrosome from aberrant, additional cycles of duplication. In support of these studies, from in

vitro experiments, Starita et al. showed that among a number of centrosome proteins, γ-tubulin,

which is an essential microtubules nucleation protein, was found to be ubiquitylated by BRCA1–

46

BARD1 using both K48 and K34 residues.221 They have also shown that the expression of γ- tubulin K48R mutant protein caused a marked amplification of the centrosomes that lead to unequal segregation, abnormal nuclear division, and cellular aneuploidy.221

Nucleophosmin1 (Npm1) is another centrosome protein that is targeted by the BRCA1/BARD1 ubiquitin ligase.222 Npm1, an important regulator of chromosomal stability, interacts and

colocalizes with BRCA1/BARD1 during mitosis. Sato et al also showed that exogenous co-

expression of BARD1 and BRCA1 with NPM leads to the stabilization of NPM. It is possible that

this stabilization of NPM is due to the ubiquitylation of the protein using K6 polyubiquitin links

instead of K48. Results from these studies clearly suggest that the ubiquitin ligase activity of

BRCA1/BARD1 complex is linked to the cell-cycle checkpoint functions and tightly regulate the

cell cycle during cell proliferation. It is very intriguing to see that how vast of a role BRCA1

plays physiologically. Figure 1.18 shows a schematic of genome wide interaction pattern of

BRCA1/BARD1 in numerous biological functions.

1.3.4 Known cancer predisposing BRCA1 RING domain mutations: Effect on BRCA1/

BARD1 interactions and E3 ligase activity

It is already evident that BRCA1-BARD1 interaction is required for several of the cellular and tumor-suppressor functions of BRCA1. Although several reports have demonstrated the BRCA1-independent functions of BARD1 might have on its own crucial role in apoptosis and

mitosis, few cancer associated mutations have been found in BARD1, compared with more than

650 (Human Gene Mutation Database) for BRCA1. It is now clearly evident that the ubiquitin E3

ligase activity of the BRCA1/BARD1 complex is responsible for many of the diverse biological

functions attributed to BRCA1, including its ability to suppress tissue-specific tumor generation

in normal cells. Several in vitro and in vivo studies reported that the missense mutations in the

RING domain, especially C61G and C64G, abolish the E3 ligase activity of BRCA1/BARD1

47 complex. Twenty percent of the clinically relevant mutations of BRCA1 occur within the N- terminal 100 residues, which contain the RING motif.188 A total of 44 different nonpolymorphic

missense mutations have been reported to date in cancer patient’s DNA encoding this region

(Breast Cancer Information Core, BIC, NIH). These missense mutations observed in cancer

patients are divided into two classes. The Zn2+-ligating residue (C24, C39, C44, C47, C61 and

C64; C27 and H41sites have not been identified in cancer patients) mutations clearly predispose individuals to breast and ovarian cancers. A second class of mutations is distributed throughout

N-terminal domain of BRCA1 and affects the BRCA1/BARD1 interaction interface and presumably BRCA1/substrate (ubiquitin conjugating enzymes) binding interface. Several in vitro

and in vivo studies reported that the missense mutations in the RING domain, especially C61G

and C64G, abolish the E3 ligase activity of BRCA1/BARD1 complex. Klevit et al.194, 200 extensively studied the mutations in Zn2+ binding sites and showed that mutations on site I (C24 and C44) cause RING motif not to fold properly200 and mutations on site II (C39, C61 and C64)

cause the local perturbation of the RING motif structure (structure of second Zn2+ binding loop)

and decrease the affinity for second Zn2+ ion binding.201, 202 These mutations do not affect the

binding interface with BARD1; instead they have hypothesized that it is likely that these

mutations directly affect the BRCA1 RING (E3)-E2 binding interface. Using Yeast Two-Hybrid

(Y2H) assay, Solomon et al. reported a comprehensive study on cancer predisposing missense

mutations on BRCA1 and their affect on BARD1 binding or E2 conjugating enzyme binding and

clinical role in tumor suppression.202 Their data revealed that some of these mutations disrupt the

binding of BRCA1 with BARD1 or an E2 enzyme (UbCh5α) and which in turn caused the loss of

its E3 ubiquitin ligase activity. They generated a random library of BRCA1 mutants using

mutagenic PCR and screened the mutants for the interaction with BARD1 or E2 enzyme

(UbCh5α) activity using Y2H assay. These screens identified residues within BRCA1 that disrupt

interaction with BARD1, and 19 of the 35 BRCA1 variants co-purified with

48

Figure 1.19: Yeast two hybrid assay of wild type N-terminal RING domain of BRCA1or mutants for the interaction with N-terminal RING domain of BARD1 and full length E2 enzyme, UbcH5α. Growth on media lacking histidine (-HIS) indicates protein-protein interactions leading to HIS3 transcription. The BRCA1- protein interactions and ubiquitin ligase data are summarized in bottom panel together with known (K) or predicted pathogenic status. Substitutions predicted to be deleterious (D), likely to be neutral or of little significance (N) or because of splice junction defect (S) are shown. Figure was reproduced from Solomon and coworkers, Human Molecular Genetics 2006.202

Figure 1.20: Summary of Y2H assay results obtained from the screening of BRCA1 mutation library introduced by error-prone PCR. Single amino acid substitutions in BRCA1 identified by split-hybrid selection against E2 (UbcH5α) and BARD1 are listed in line1 and line2, respectively. Known cancer predisposing BRCA1 mutants recorded in BIC database are listed in line3. Green-shaded boxes illustrate residues substituted in variants identified from the UbcH5α and BARD1 screens that are located in residues also substituted in patienent DNA. Figure was reproduced from Solomon and coworkers, Human Molecular Genetics 2006.202

49

BARD1 showed reduced E3 ligase activity. All 7 mutations in Zn2+-ligand binding residues inhibited ubiquitin ligase activity. A summary of the screening results and activity for E3 ligase is

shown in Figures 1.19 and 1.20. These results are in consistent with RING motif structural

analysis and Zn2+ ligating mutation studies reported by Klevit et al. But, these results are in contrast with the hypothesis that any charged or polar residue mutations in the hydrophobic core may perturb the BRCA1/BARD1 interaction.200, 201 And also because Y2H assay is prone to high rates of false positives, further studies of the role of mutations on BRCA1 in BRCA1/BARD1 interaction were warranted. BRCA/BARD1 interaction studies were also carried out by Sarkar et al. in 2008 on a small subset of BRCA1/BARD1 interface mutants using improved split folding reporter GFP.56 We reported that helix-helix interface mutants V11A and M18K inhibit the interaction of BRCA1 and BARD1, and a mutation on the RING motif away from four-helix bundle, L52F, also disrupts the interaction. Therefore, further investigation of all cancer predisposing mutations observed in N-terminal RING domain of BRCA1 and their role in

BRCA1/BARD1 binding and E3 ligase activity is needed to be carried out using a more reliable approach to validate the partial results obtained by Y2H assay. The analysis of BRCA1/BARD1 interactions and ubiquitin ligase activities of RING-domain mutations is important not only to investigate the biological function of BRCA1 but also to predict a person’s predisposition to cancer.

1.4 Human paraoxonase-1: A potential bioscavenger target

1.4.1 Paraoxonase-1: Background

Paraoxonase-1 is a calcium dependent serum protein belonging to a family of arylesterase enzymes. Serum paraoxonase received its name from the ability to hydrolyze paraoxon, the toxic metabolite of the insecticide parathion.223 In 1991, Furlong and coworkers first reported the

purification of human and rabbit paraoxonase 1 (PON1) and later the isolation and

characterization of cDNA of these enzymes.224, 225 Human paraoxonase 1 (huPON1, EC 3.1.8.1) is

50 a 355 amino acid glycosylated protein which completely retains its hydrophobic signal sequence uncleaved.225, 226 Retention of the signal sequence is resulting from the presence of the large polar

amino acids (H, Q, K and R) in the -3 position from the theoretical cleavage site.

PON1 has three conserved cysteine residues (C42, C284 and C353) with a free cysteine at C284

and a disulfide bond between C42 and C353. PON1 possesses two independent calcium binding

sites with Kd values of 0.36 and 6.6 μM; the higher-affinity site is essential for the enzyme stability and is considered to be structural; the other one is considered to be essential for enzymatic hydrolytic activity.227, 228 There are a number of potential N-glycosylation sites on

huPON1. Deglycosylation of serum PON1 with PNGase F led to a 20% decrease of the apparent

molecular mass suggesting that huPON1 is exclusively N-glycosylated.229 Primo-parmo et al. in

1996 identified two other PON family proteins, PON2 and PON3.230 Both show a remarkable sequence similarity with the PON1 gene. The degree of identity between PON1 and PON2 (73%) is slightly higher than that between PON1 and PON3 (65%). Human PON2 and PON3 have very limited paraoxonase and arylesterase activities, but are similar to PON1 in that both hydrolyze aromatic and aliphatic lactones and thiolactones.231 It was observed that PON1 and PON3 are expressed mostly in the liver and at low levels in the kidney.232 La Du and Draganov et al. observed that PON3 is capable of hydrolyzing lovastatin, simvastatin and spironolactone.231 On

the other hand, it is reported that PON2 is widely expressed in a number of tissues, such as in the

brain, liver, kidney, and testis.233, 234 The only substrate that has been reported to date for PON2 is

dihydrocoumarin.235 Studies from Reddy et al. and several other groups reported that like PON1, both PON2 and PON3 have been shown to prevent HDL and LDL from oxidative damage.232, 234

PON3 appears to exhibit enhanced ability to protect low-density lipoprotein against oxidation relative to PON1. The three PON family genes are aligned in chromosome7 (7q21-22) in the order of PON1, PON3 and PON2. On the basis of the structural homology and predicted evolutionary distance between them, it appears that PON2 is the oldest member of the family;

51

PON3 arose from it next, and, more recently, PON1 appeared.230 PON1 is the most widely

studied member of this family.

It was found that the concentration of PON1 in human plasma (60 mg/L) varies between

individuals by as much as 13-fold.236 The activity level of PON1 in plasma is determined by a combination of complex genetic interactions and environmental/dietary factors. In the human population, the PON1 gene exists in several polymorphisms in the promoter and coding regions.

There are two coding region polymorphisms in PON1, Q192R and L55M, among which substrate dependent polymorphism Q192R has been studied extensively. The R isoform hydrolyzes substrates, such as paraoxon more rapidly but diazoxon, soman and especially sarin more slowly than the Q isoform. Whereas, the Q isoform hydrolyzes the other substrates, such as diazoxon, soman and sarin, more rapidly than the R192 isoform.237, 238 The lactonase activity of the PON1

R192 isoform was shown to be little higher than that of Q192 isoform.239 Both isoforms, on the other hand, are found to hydrolyze phenyl acetate at approximately the same rate. The L55M polymorphism does not affect catalytic activity but has been associated with the level of PON1 in plasma and increased risk of cardiovascular disease and diabetes.240, 241 Studies from Leviev et al.

and other groups demonstrated that PON1 has a promoter region polymorphism at position -107

which affects the promoter activity and has a major impact on the gene expression, and thus on

the PON1 concentration in blood.242, 243 Certain drugs were found to affect the expression of

PON1 through promoter stimulation.244 It was reported that the L55M polymorphism can

influence the expression of plasma PON1 through interaction with the C-107T promoter

polymorphism.241 Several studies have shown that the L55 and R192 alleles are in strong

disequilibrium, with >98% of the R192 alleles having L at position 55.229 Therefore, it is evident

that the polymorphism of the PON1 gene affects the blood-level PON1 and its catalytic

efficiency, and both factors strongly affect an individual’s susceptibility to arteriosclerosis and

toxicity from OP poisoning.226, 245

52

1.4.2 Organophosphorous pesticides and nerve agent toxicity

Organophosphates (OPs) are a major class of insecticides which have been widely used around the world since late 1940s. Chemicals in this class kill insects by disrupting their brain and nervous system. Unfortunately, the toxic metabolites of these chemicals have significant adverse effects on non-target animals and humans. Exposure to organophosphates causes a significant number of poisonings and deaths each year. Organophosphate pesticide poisoning is a major clinical and public health problem in rural areas across the world. The number of intoxications (accidental, suicidal, homicidal) with organophosphorus pesticides (OPs) is estimated at 3000,000 per year, and the number of deaths and casualties are 300, 000 per year.246

In the 1930s, Gerhard Schrader in Germany was the first to experiment with these compounds as insecticides and synthesized a large number of organophosphorus compounds including Tabun.

Shortly thereafter, Schrader and his group accidentally discovered the potential of tabun as a nerve agent and named it ‘GA’. Schrader’s work also led to the development of even more potent nerve agents, such as sarin (GB), soman (GD) and cyclosarin (GF), Figure 1.21. Second generation nerve agents, produced in UK, Russia and USA, are known as V-series (VE, VG, VM,

VX and VR) nerve agents, which are at least 10 times more potent than G-agents. Of more than

100 OP pesticides used worldwide, the majority of them are dimethyl or diethyl phosphoryl compounds; parathion, malathion, chlorpyrifos, diazinon, dimethoate are among the most prominent ones, Figure 1.21.

Sulfur-containing pesticides undergo oxidative desulfuration to their toxic oxon metabolites by cytochrome P450 and glutathione-S-oxidase enzymes in the liver.247 These oxon

metabolites of organophosphate pesticides and nerve agents are the potent inhibitors of carboxyl

ester hydrolases, particularly acetylcholinesterase (AChE) and butyrylcholinesterase (BChE),

which is localized primarily in the central and peripheral nervous system, in the neuromuscular

junctions, and in red blood cells (RBCs). AChE is an enzyme that degrades the neurotransmitter

53

Figure 1.21: Chemical structures of organophosphate pesticide metabolites (paraoxon, chlorpyrifos, diazoxon) and nerve agents (sarin, tabun, soman, and V-agents).

acetylcholine (ACh) into choline and acetate. Exposure to organophosphate pesticides and nerve agents inactivates AChE by phosphorylating the serine hydroxyl group located at the active site

of the enzyme. The phosphorylation occurs by loss of an organophosphate leaving group and

establishment of a covalent bond with AChE. Inactivation of AChE causes an excessive

accumulation of acetylcholine throughout the central and peripheral nervous system, resulting in

the overstimulation of muscarinic and nicotinic receptors. This leads to the development of

cholinergic syndrome (which includes increased sweating and salivation, profound bronchial

secretion, bronchoconstriction, miosis, increased gastrointestinal motility, diarrhea, tremors,

muscular twitching, etc) which can in turn progress to paralysis and respiratory arrest. In addition

to respiratory failure, OP exposure can result in temporary or permanent damage to neurological,

cognitive, and motor functions.248-250 The mechanism of acetylcholinesterase inhibition and aging is shown in Figure 1.22.251, 252 54

Inhibition of cholinesterases by oxons is preceded by reversible formation of a complex followed by pseudo-unimolecular phosphorylation reaction.253 The phosphorylated enzymes then

may undergo endogenous hydrolysis and reactivation by strong nucleophile such 2-PAM

(pyridine aldoxime methiodide or Pralidoxime chloride-Cl), a powerful reactivator of

alkylphosphate-inhibited acetylcholinesterase, or irreversible binding and permanent enzyme

inactivation known as ‘aging’.253, 254 The aging process takes place through the monodealkylation of the phosphorylated enzyme complex that results in the formation of very stable enzyme-OP complex, which is then resistant to spontaneous hydrolysis and reactivation by oximes. It is shown that the spontaneous recovery of the dimethyl phosphorylated enzymes is much faster than that of the diethyl phosphorylated enzymes and that BChE recovers by an order of magnitude slower than AChE.251, 255 The rate of spontaneous reactivation of AChE is variable, which partly accounts for differences in acute toxicity between the nerve agents. Monodealkylation occurs to some extent with all dialkylphosphonylated AChE complexes, but in particular with soman, aging takes place very fast, which determines its acute toxicity. In case of soman, aging occurs so fast that no clinically relevant spontaneous reactivation of AChE occurs before it forms an irreversible complex.

1.4.3 OP poisoning and nerve agent detoxification: Treatment of acute OP poisoning

Although the acute toxic effects of organophosphate insecticides and nerve agents on humans and animals have been known for long time, surprisingly, there are no effective detoxification measures available to date. Despite the increasing threat posed by nerve agents and organophosphate pesticides across the world, the work to develop an efficient pre- and post- exposure treatment is in a very early stage. The conventional pharmacological approach for the treatment of OP intoxication involves multiple drug administration. Current medical measures consist of pretreatment with spontaneously reactivating acetylcholinesterase inhibitors, such as pyridostigmine bromide and cholinergic drugs, such as atropine (a competitive antagonist 55

Figure 1.22: Inhibition of acetylcholine esterase by organophosphates. Enzymes form a reversible complex with organophosphates and get phosphorylated at the hydroxyl group of serine releasing leaving group, HX. Phosphorylated enzyme undergoes monodealkylation to form an irreversible enzyme-phosphate complex (known as aging) or hydrolysis of phosphate group for spontaneous reactivation of enzyme which liberate inactive acidic organophosphate.251

of the muscarinic acetylcholine receptor) to antagonize the effects of elevated level of acetylcholine resulted from the inhibition of AChE.256, 257 An anticonvulsant drug, such as

diazepam, is administered to control OP-induced tremors and convulsions. In case of acute

toxicity, an oxime (parlidoxime or 2-PAM or bispralidoxime)251, 258 is also used in combination

for the reactivation of phosphorylated cholinesterase to prevent or slow down the enzyme’s aging

process. The proposed mechanism of cholinesterase reactivation is shown in Figure 1.23. The

rate of aging is dependent on the alkyl group on OPs. In case of sarin and soman, the

phosphorylated enzyme undergoes dealkylation very fast so that the spontaneous and oxime

mediated reactivation is ineffective. Although this antidotal treatment is effective in preventing

lethality from OP poisoning, it does not prevent post-exposure incapacitation, convulsions and

permanent brain damage. Moreover, the antidotes cause multiple long-term side effects. These

problems led to the development of alternative strategies using bioscavengers as a pretreatment to

sequester and neutralize highly toxic OPs before they reach their physiological targets.

The idea of a bioscavenger using proteins or enzymes to detoxify OP agents has been

very popular because of its potential for pre- and post-exposure treatment with no side effects.

Mainly, there are three classes of potential bioscavengers that have been explored to date for the 56 detoxification of OP poisoning. The first class is categorized as stoichiometric bioscavengers, such as acetylcholinesterase (ChE) and butyrylcholinesterase (BChE). These enzymes bind to OP agents in a 1:1 mole ratio. The second class is categorized as “pseudo-catalytic” bioscavengers, for example a combination of ChE and an oxime pre-treatment. The most recent class is categorized as catalytic bioscavengers that can catalytically hydrolyze OPs and thus render them non-toxic, such as paraoxonase and phosphotriesterase, etc.

Figure 1.23: Plausible mechanism of reactivation of phosphorylated cholinesterase by oxime. Nitroxyl group of oximate attacks the phosphoryl moiety and forms a pentacoordinate transition state which liberates the reactivated enzyme and phosphoryloxime.251, 259

In “pseudo-catalytic” bioscavenger, oximes rapidly and continuously react with the OP- inhibited phosphorylated enzyme and restore its catalytic activity. Among the first generation stoichiometric bioscavengers, butyrylcholinesterase has been proven to be relatively more effective in treating OP intoxication.260 ProtexiaTM, a PEGylated recombinant human butyrylcholinesterase (rHuBChE), produced in the milk of transgenic goats (Nexiabiotech and

PharmAthene), is the first operational stoichiometric scavenger. The capability of rHuBChE as a medical countermeasure has been demonstrated in vivo to protect animals from multiple lethal doses of a broad spectrum of nerve agents, including sarin, soman, tabun and VX.261 Protexia was in phase I human safety clinical trial in 2008 and the results of which will be available soon.

However, the usefulness of these stoichiometric bioscavengers as therapeutic was limited by the 57 fact that a large amount of these drugs is necessary for higher levels of OP exposure. Because of this disadvantage, it has long been sought to develop and obtain a catalytic enzyme for effective pre- and post-exposure treatment of OP intoxication. In an effort to do that, based on molecular modeling, a G117H human BuChE mutant was designed, but it failed to show sufficient OP hydrolase activity.236, 262

As a human protein and for its known hydrolyzing activity against OP pesticides and nerve agents, human paraoxonase is considered to be an excellent candidate for the development of therapeutic drug as a catalytic bioscavenger.223, 237, 263-265

1.4.4 Human paraoxonase-1 (huPON1): A potential bioscavenger

Paraoxonase-1 has long been known for its hydrolyzing activity against organophosphate insecticides and nerve agents. Though the exact physiological role of PON1 has not been unambiguously established yet, several studies reported that PON1 has a broad substrate specificity and is capable of hydrolyzing a wide variety of organophosphate substrates and a number of aromatic esters, thioesters, phosphotriesters, carbonates, lactones and thiolactones. As

PON family proteins (PON1, PON2, PON3) have a common functionality of hydrolyzing lactones, it is believed that native function of PON1 may be as lactonase. 239 Mazur in 1946 first reported the enzymatic hydrolysis of organophosphates by animal tissues.266 In 1950s Aldridge et al.248 reported the hydrolysis of paraoxon and phenyl acetate in human blood serum and other

mammalian species. Recently La Du et al.267 in 1992 and several groups reported that human serum paraoxonase is capable of hydrolyzing the toxic metabolites of a number of insecticides such as parathion, diazinon and chlorpyrifos, and nerve agents, such as sarin and soman.237, 268

Later a series of in vivo studies on different animal models also demonstrated the role of serum

PON1 in detoxification of organophosphate (OP) pesticide metabolites such as paraoxon,

chlorpyrifos oxon, diazoxon.223, 245, 263-265 Some of these studies investigated the exogenous administration of paraoxonase-1 into rats and mice to find out its role in acute OP intoxication. 58

Costa and coworkers223 in 1990 demonstrated that injection of purified paraoxonase-1from rabbit serum in rats increases the peak hydrolytic activity of rat serum by 9-fold toward paraoxon and by

50-fold toward chlorpyrifos-oxon. The increase of serum paraoxonase activity was long-lasting, with a 2- and 10-fold increase, respectively, still present after 24 hr. Later, Li et al.265 purified

PON1 to homogeneity from rabbit serum and administered in mice. Serum PON activity was

increased by 7-fold against paraoxon and 30- to 40- fold against chlorpyrifos oxon. The mice

were then subjected to an acute chlorpyrifos (CPS) intoxication and observed for its influence.

The effect of OP poisoning was decreased in PON1-treated mice and the complete protection

against AChE inhibition was observed in the brain and diaphragm. The administration of

paraoxonase after 30 min of exposure to an acute level of CPS abolished cholinergic signs and a

significant protection against CPS toxicity was still observed. These results indicate that

increased levels of serum paraoxonase activity can affect the toxicity of paraoxon and

chlorpyrifos-oxon (CPO) and protect the animals from these OP poisoning.

The most convincing results about the role of PON1 in OP poisoning came from in

vivo experiments by Shih et al. in 1998 and Li et al. in 2000.245, 264 They investigated the toxicity of OPs (paraoxon, diazoxon and chlorpyrifos oxon) in PON1 knockout mice generated by targeted disruption of exon 1 of the PON1 gene. On average wildtype (PON1+/+) mice plasma had

367 units per liter of paraoxonase activity, whereas PON1 knockout (PON1-/-) mice had no

detectable paraoxonase or diazoxon activity and only 2% of chlorpyrifos oxon activity in their

plasma. PON1 deficient mice were found to be extremely sensitive to the toxic effects of

chlorpyrifos oxon245, 269and diazoxon264 compared to the wild type mice. At a low dose (1-1.5 mg

kg-1), CPO did not affect AChE activity in wild type mice but reduced it by 80% and 74% in the brain and diaphragm, respectively, in PON1-null mice. At a dose of 3 mg kg-1, CPO inhibited brain AChE activity in PON1+/+ mice by 31%, whereas it killed PON1-/- mice within 4 h of exposure. The AChE activity in both the brain and diaphragm of PON1-/- mice was significantly

59 lower than that of PON1+/+ mice. All of the PON1-/- mice that received 6 mg kg-1 of CPO died

within 2 hours of exposure, whereas the PON1+/+ mice that received the same dose of CPO survived, their AChE activity was mildly inhibited and showed no symptoms of CPO poisoning.

Similar results were observed for diazoxon.264 Surprisingly, these knockout mice did not show increased sensitivity to paraoxon exposure. Exogenous administration of human PON1 isoforms

(192R and 192Q) restored the PON1 activity for CPO (192R afforded 50% better protection than

192Q isoform) and diazoxon (equal protection from both isoforms).245, 264 The administration of

rabbit or human PON1 isoforms failed to protect the knockout mice from paraoxon exposure.

Although the catalytic efficiency of PON1 against some OPs is low, such as paraoxon, its

capacity to hydrolyze OPs and nerve agents, such as sarin and soman in vivo and in vitro makes

this enzyme a promising candidate as catalytic bioscavenger.236 There are several key advantages of huPON1 for using as a bioscavenger. It is a human protein and may not have any immunological response which will give it long half-life in the bloodstream. Its catalytic efficiency against sarin and soman is similar to that of bacterial phosphotriesterase (PTE).270

Several in vivo studies in rat and mice have already proven its role against OP and nerve agent

poisoning.270, 271 Though there have been a large number of studies because of its involvement of huPON1 in cardiovascular disease, diabetes and antiatherogenicity, there is little known about its structure and catalytic activity and mechanism. HuPON1 is highly unstable protein when expressed in E. coli. The very low solubility and expressibility in E. coli and lower thermal stability in non-native environment make it very difficult to obtain in reasonably large quantities to study the structure-function relationship.

To address these issues, Tawfik and coworkers engineered a recombinant chimeric (mix of rabbit mouse, rat and human) forms of PON1 by directed evolution.273 DNA shuffling of genes of a close homologous family has been proven to be a very successful technique for the directed evolution of enzymes with specific functions.274, 275 In their approach, they have carried

60

Figure 1.24: Figure a shows the X-ray crystal structures of chimeric recombinant PON1 (G2E6, PDB:1V04).272 Two central calcium ions are shown in red and blue. The N-terminal residues 1–15 and a surface loop (residues 72–79) connecting two central strand strands are not visible in the structure. Figures b and c show the side view with three α-helices at the top. Hydrophobic residues proposed to be involved in HDL anchoring (side chains yellow) are shown in c. The line defined by the side chains of Tyr185, Phe 186, Tyr190, Trp194, Trp202 (helix H2 and the adjacent loops) and Lys21 (helix H1) models the putative interface between HDL’s hydrophobic interior and the exterior aqueous phase. Figure c was reproduced from Harel et al., Nat. Struct. Mol. Biol 2004.272

61 out family shuffling of PON1 genes from human, mouse, rat and rabbit (79 - 93% DNA sequence identity). The PON1 library was screened on agar plates for arylesterase activity using 2NA (2- naphthylacetate) and DepCyC (7-O-diethylphosphoryl-3-cyano-7-hydroxycoumarin) dye. The 20 best clones were subcloned and used as a template for further rounds of DNA shuffling. The isolated best variants after the second (G2E6) and third (G3C9) round of DNA shuffling have better solubility, stability and expressibility in E. coli (~20 mg/L of culture). Among these variants, G2E6 has very high hydrolysis activity against paraoxon. G2E6 has 70% sequence homology with rabbit and 84% with human PON1 and a total of 59 mutated amino acids compared to human PON1. The crystal structure of this recombinant protein, reported in 2004

(Figure 1.24 a), showed that chimeric G2E6 is a six-bladded β-propeller protein in which each blade contains four strands.272 The two central calcium ions are shown in red and blue. A side

view of the structure with the proposed HDL (high density lipoprotein) binding sites formed by

the hydrophobic side chains from H1, H2 and H3 and the loops connecting H2 is shown in

Figure 1.24 b, c. The absence of substrates or analogs in the crystal structure failed to shed light

on the structure-activity relationship and on the catalytic mechanism of PON1. On the other hand,

recombinant G3C9 has even better thermostability than G2E6 and has 50 amino acids mutated

compared to the huPON1.272 Though both of these chimeric PON1 variants have better solubility and thermodynamic stability, they failed to show catalytic activity against OPs and nerve agents other than paraoxon.

HuPON1 is highly unstable and prone to aggregation when expressed in E. coli. All previous attempts to express and purify huPON1 in a large-scale have not been successful.

Furlong and coworkers,276 in 2008, reported the first successful expression and purification of

human PON1 from E. coli. Although the yield was very low (5 mg from 12 L of fermentor

preparation or 450 µg per fermented L of culture), wild type huPON1 was found to be active

against both paraoxon and phenyl acetate. When purified huPON1 was injected into PON1

62 knockout mice, it increased the response against CPO and DZO but not paraoxon, which is consistent with previous results shown by Shi et al. and Li et al.245, 264, 269

An enzyme must fulfill a number of critical requirements before it acquires the status of an effective OP bioscavenger in vivo.277 Among these, the most difficult challenges are; high specificity and catalytic activity against OP molecules; long half-life (or mean residence time,

MRT) in vivo to be effective over a prolonged time; no unexpected immunogenic response; no adverse effects on physiological processes, in particular, no behavioral effects; and availability in large quantities. There are also several technological constraints the enzyme faces. It must display both thermal and storage stability to be kept and used under various climate conditions without the loss of activity. For using it as a therapeutic drug, huPON1 is facing a number of critical challenges, such as large-scale expression and purification of fully-humanized protein; thermal stability of the protein when expressed in E. coli or other nonnative environment; and improvement of activity towards OP pesticides and nerve agents. The rational design and or directed evolution of the huPON1 and the huPON1 variants are the obvious choices to address these issues.

1.4.5 Physiological role of huPON1: Prevention of atherosclerosis development and

cardiovascular disease

It is worth noting here that human PON1 plays a significant role in the prevention of atherosclerosis development and cardiovascular disease. Paraoxonase-1 is expressed in the liver and secreted into the blood where it is exclusively associated with high density lipoprotein (HDL) particles.278 The close association of PON1 with HDL particles led Mackness et al. in 1989 to propose that the enzyme might have some physiological role in the prevention of atherosclerosis development.279, 280 A number of studies, thereafter, attempted to determine the relationship

between genetic polymorphism of PON1 and the development of atherosclerosis and

cardiovascular disease. Some of these case-control studies reported by Zama et al., Pati et al., 63

Sanghera et al., Ruiz et al. and Odawara et al. showed that PON1 Q192R polymorph is significantly associated with coronary heart disease (CHD).241, 281-284 However, some other studies did not find any relationship between aetherosclerosis change and PON1 alleles.285-289 The

discrepancies among these findings could be explained by the fact that each of these studies used

a different type of population with different dietary habits and environment. In recent studies it

has been found that serum PON1 concentration and activity were found to be reduced in several

groups of patients with coronary heart disease (CHD), non-insulin-dependent diabetes, vascular

disease, independent of the PON1 polymorphism.290 Oxidation of low density lipoprotein (LDL) is directly related to the initiation and progression of atherosclerosis.237, 281, 290-294 Although it was

known for quite some time that HDL-associated enzymes inhibit the LDL oxidation, the

mechanism of this antioxidation was unknown. It was believed that human serum paraoxonase 1,

which is exclusively associated with HDL, might have a partial role in inhibiting LDL oxidation.

In fact, Mackness et al. in 1991 first showed that the HDL-associated purified human

paraoxonase retards the in vitro copper-catalyzed oxidation of LDL by preventing the generation

of lipid hydroperoxides.295, 296 This provided the first molecular hypothesis that HDL might affect

and prevent LDL oxidation. From the inbred strains of mice (which have dramatically different

susceptibility to atherosclerosis), the experiments by Shih et al. revealed that there was a direct

relationship between the level of PON1 expression and the risk of atherosclerosis development.297

Navab et al. generated a number of transgenic and knockout mouse models [Apolipoprotein E knockout and LDL receptor (LDLR) knockout] to study the atherosclerosis.298, 299 Both models

exhibited hypercholesterolemia and developed advanced atherosclerotic lesions after feeding of

the high-fat diet. In both models the plasma PON1 levels were greatly reduced as compared to the

wild type mice. HDL isolated from ApoE KO mice also exhibited significantly lower PON1

activity and failed to prevent LDL oxidation in cultured artery wall cells. The supplementation of

transgenic HDL with purified PON1 restored the protective ability of HDL and paraoxon activity.

64

Figure 1.25: Role of PON1 in LDL and HDL oxidation, macrophage foam cell formation and atherosclerosis development. Under oxidative stress, plasma LDL binds to the extracellular matrix proteoglycans and is oxidized by arterial wall macrophages to yield oxidized LDL (Ox-LDL). The extent of LDL oxidation is determined by the balance between LDL lipids and antioxidants, including vitamin E, β- carotene, lycopene and certain flavonoids. Oxidized LDL is selectively taken up by the macrophage scavenger receptors, which contributes to foam cell formation. HDL associated PON1 can hydrolyzes lipid peroxides in LDL and convert this atherogenic lipoprotein to a non-atherogenic lipoprotein (LDL shown in pink). HDL-associated paraoxonase also hydrolyze these lipid peroxides in macrophage cells and thus reduces macrophage atherogenicity. HDL-associated PON1 also protects HDL from oxidation and promotes HDL-mediated cholesterol efflux from macrophage foam cells. The figure was reproduced from Aviram et al. from Molecular Medicine Today, 1999.281

65

These findings suggested that the PON1 is directly involved in protecting LDL from oxidation and inhibiting atherosclerosis development. Watson et al.300 later showed that PON1 was also capable of hydrolyzing specific lipid peroxide on LDL. The most convincing results of PON1’s involvement in preventing atherosclerosis development came from studies by Lusis and coworkers245 in 1998 from the use of PON1 knockout (KO) mice. In order to study the role of

PON1 in vivo, they created the PON1 KO mice by gene targeting. It was observed that the KO

mice were more susceptible to atherosclerosis compared to their wild type littermates when fed

on a high fat, high cholesterol diet. HDL particles isolated from PON1-knockout mice were

unable to prevent LDL oxidation in a co-cultured cell model of the artery wall. They have also

shown that both the HDL and LDL isolated from PON1 knockout mice were more susceptible to

oxidation by co-cultured cells than the lipoproteins of wild type littermates. These studies and the

human population case studies suggested the possible role of paraoxonase-1 in the prevention of

atherosclerosis development (Figure 1.25).

The above findings led to an enormous number of studies to investigate the in-depth role

and mechanism of PON1 in the prevention of atherosclerosis. La Du and coworkers293 in 1998 demonstrated that, during oxidative stress PON1 was also capable of preventing HDL from oxidation and preserve its function. They have also demonstrated that this effect was mainly due to the PON1’s ability to hydrolyze lipid hydroperoxides. These results were supported by transgenic mice experiments by Oda et al.301 in 2002, in which they have reported that overexpression of PON1 in transgenic mice inhibited lipid hydroperoxides formation on HDL and protected its integrity and function. One of the major functions of HDL is that of reverse cholesterol transport, a process whereby HDL removes excess cholesterol from foam cells.

Macrophage foam cell formation is the hallmark of early atherogenesis.302 Using PON1 knockout mice and transgenic mice, Rosenblat et al. demonstrated that paraoxonase-1 enhances the HDL- mediated macrophage cholesterol efflux via ABC A1 (ATP binding cassette transporter A1)

66 transporter and attenuates the atherosclerosis development.303

Most of these studies determined the physiological roles of PON1 in atherosclerosis development and cardiovascular disease. However, some other recent studies found PON1 to be implicated with several physiologically significant functions. Low serum PON1 activity has been found in diabetes mellitus (both type I, insulin-dependent, and type II, noninsulin-dependent) and renal disease,304 in patients with vascular complications.305 Diabetic complications in the

Japanese population, such as retinopathy306, 307 or central retinal vein occlusion,307 may be

associated with the PON1(55L) and PON1(192R) alloenzymes, respectively. A higher frequency

of the PON1(192R) allele has been reported in type-II diabetic patients with macrovascular

disease,308 cerebrovascular disease,309 and coronary artery disease.310

1.5 Re-engineering human paraoxonase-1: Rational design and directed evolution

For catalytic bioscavengers, large-scale expression, and in vitro and in vivo

thermostability are crucial properties for huPON1. Bacterial expression of recombinant human

paraoxonase-1 typically yields misfolded or unfolded proteins in the form of inclusion bodies or

aggregates. Very low solubility and expressibility in E. coli and lower thermodynamic stability in

non-native environment make it very difficult to obtain in a reasonably large quantity and to

study the structure-function relationships. A number of structural features contribute to this.

Mature huPON1 retains its hydrophobic leader sequence, three α-helices and the loops before and

after the second α-helix form a hydrophobic patch for HDL binding sites, and it has two putative

glycosylation sites on the surface and three cysteine residues. When expressed in bacteria, these

unmodified hydrophobic surfaces on human PON1 make it less stable and prone to misfolding

and aggregation.

1.5.1 Solubility tag for the expression in bacterial systems

E. coli offers a number of advantages that make it the most widely used host organism for

67 recombinant protein expression in molecular biology. The bacteria can grow rapidly to high cell density in a media which is inexpensive. The plasmid construction of the recombinant genes with affinity tags is easy, the protein can be expressed at very high levels, and the affinity purification of the target protein is simple. Despite these many advantages, the insolubility of the target protein remains a major limitation of the system. Especially the expression of human proteins in bacteria has been a very difficult challenge. Expression at lower levels and at lower temperatures sometimes helps overcome these limitations. In another approach, fusion of the recombinant protein with more soluble bacterial proteins helps increase the solubility and the expressibility of the test proteins. Glutathione S-transferase (GST), thioredoxin (Trx) and maltose-binding protein

(MBP) have been successfully applied as a fusion protein to increase the solubility and expressibility of a number of difficult-to-express proteins or human proteins.311, 312 MBP has proven to be a better fusion partner to express protein with better solubility and with a higher yield.311 Chaperone coexpression has proven to be another approach to obtain folded functional

difficult-to-express proteins with higher yield from the bacterial system.313 Overexpression of the recombinant proteins in bacteria sometimes leads to misfolding and aggregation. Molecular chaperones are a group of specialized proteins that control the protein folding process in the cytosol and help refold the misfolded or aggregated proteins in an ATP-dependent manner.313-315

They also assist in the folding of newly synthesized proteins to their native states, prevent aggregation and promote refolding of preexisting proteins which are thermodynamically trapped in a low energy conformation. In E. coli, the folding of newly synthesized proteins is assisted by mainly three different chaperone sets, ribosome-associated trigger factor, DnaK with its co- chaperones DnaJ and GrpE (KJE), and GroEL with its co-chaperone GroES (ELS).314 Several

studies reported that the co-expression of these chaperone sets increases the expressibility and

solubility of human proteins such as ORP150, p50CSK protein tyrosine kinase, and β-

galactosidase etc.313 Bukau and coworkers tried to explore the full potential of different bacterial

68 chaperone sets and demonstrated the chaperone mediated procedure to increase the yield and solubility of 70% of 64 different heterologous proteins expressed in E. coli.313

Although these methods have been occasionally successful, they fail to modify intrinsic

folding stability and the solubility of the protein. It is possible to obtain proteins in soluble form

during initial expression and chaperone-mediated refolding, the protein may still aggregate

irreversibly during the subsequent workup. Directed evolution method or rational design offer

alternative approaches to obtain soluble proteins.

1.5.2 Rational design

It may also be possible to improve the expression and the solubility through various

amino acid substitutions that promote the innate folding ability and solubility of a protein. In

recent years, rational design or directed evolution approaches have been successful to achieve the

efficient heterologous expression of proteins with increasing solubility and stability in bacteria

and yeast.316 In rational design, site-specific changes are made on the target enzyme based on the

detailed information and knowledge about the protein structure, function and catalytic

mechanism. There are several examples of proteins that have been stabilized by the introduction

of point mutations with cumulative stabilizing effects.316-319 The best known example is the rational design of subtilisin E by introducing nine point mutations to obtain a calcium- independent hyperstable variant whose inactivation rate at higher temperature is 105 times slower than that of the wild type.320 Engelman and coworkers in 2001 reported the re-engineering of phospholamban, a protein that forms a stable helical homopentamer within the sarcoplasmic reticulum membrane, into a soluble pentameric helical bundle by replacing its lipid-exposed hydrophobic residues with charged and polar residues.321 Based on computational design,

DeGrado and coworkers rationally engineered a water soluble analog of phospholamban by

changing membrane-exposed positions to polar or charged amino acids, while the putative core

was left unaltered.322 This was based on the hypothesis that membrane proteins and water-soluble 69 proteins share a similar core and it should be possible to water-solubilize membrane proteins by mutating only their lipid-exposed residues. The redesigned phospholamban mimics all of the reported properties of PLB including oligomerization state, helical structure, and stabilization upon phosphorylation. Based on the same approach, DeGrado and coworkers redesigned a water soluble variant of a membrane protein, potassium channel KcsA, by mutating the lipid-contacting side chains to more polar groups.323 It is still unknown how the stability of a protein is encoded in its primary sequence and how the changes in individual amino acids contribute to stability.

Recent developments in the field include increasing awareness of the importance of the protein surface for stability, as well as the notion that normally a very limited number of mutations can yield a large increase in stability.324 Unfortunately, the forces that govern protein folding and structure are not fully understood, and much research in this direction is still needed. General application of this approach is limited by the lack of sufficient structural and functional information about the target protein. Successful rational design using site-directed mutagenesis requires a high level of understanding of structure and mechanism. Despite many successful efforts to understand the structural basis of protein stability, there is still no universal strategy to stabilize any protein by a limited number of rationally designed mutations.

1.5.3 Directed evolution

Directed evolution has emerged as a powerful alternative approach for engineering enzymes with new or improved functions, such as activity, substrate specificity, stability, and solubility. This approach has proven particularly advantageous in cases in which prior knowledge of the protein’s structure and function were not available. Directed evolution involves the iterative cycles of random mutagenesis by error-prone PCR or recombination by DNA shuffling followed by functional screening or selection under evolutionary pressure to obtain the desired properties. Directed evolution can alter or generate a remarkable range of new enzyme properties and is now well established and highly effective method for protein engineering and 70 optimization.325 Following the pioneering works by Willem Stemmer and Frances Arnold, several successful applications of directed evolution have been documented to re-engineer proteins with new or altered properties and functionalities. One notable example is the pioneering work by

Arnold and coworkers in which they have used laboratory evolution methods to enhance the thermostability and activity of psychrophilic protease subtilisin S41.326 Using a combined strategy of random mutagenesis, saturation mutagenesis and in vitro recombination (DNA shuffling), they were able to screen out variants with 500 times the stability and 3 times the activity (10 to 60 °C) of wild type. The cytochrome P450 monooxygenase BM-3 (P450-BM3) is a fatty acid (chain length between C12 to C18) hydroxylase which also oxidizes corresponding alcohols and amides but does not oxidize alkanes. The directed evolution on P450-BM3, screening and selection converted it into an efficient alkane hydroxylase and oxidizes octane.327 Sun et al. reported the

directed evolution of a fungal enzyme, galactose oxidase (GAO), and selected variants which

were more thermostable, expressed much higher level in E. coli with 60 fold increasing of GAO

activity.328

A successful example of directed evolution is the generation of an orthogonal aminoacyl

tRNA sythetase which enables the incorporation of unnatural aminoacids into proteins in

response to the amber stop codon in vivo.329, 330 Tyrosyl-tRNA synthetase from M. jannaschii was

evolved to an efficient O-methyl-tyrosyl-tRNA synthetase by using a combination of positive and

negative selection. Other examples of evolvable enzymes for thermostability include glutathione

transferase,331 lipase B from Candida Antarctica,332 β-glucuronidase,333 glucose dehydrogenase,334 etc. As mentioned earlier, a chimeric recombinant variant of paraoxonase-1 (G2E6) with increased solubility and expressibility in E. coli was generated by Tawfik and coworkers by directed evolution from rabbit, mouse, rat and human isoforms.273 Application of computational design and rational engineering in addition to directed evolution has recently very successful to evolve proteins with altered or new properties. Numerous widely used methods such as error-

71 prone PCR, DNA shuffling, family shuffling, etc., are available to produce a large library with huge diversity. Efficient high-throughput methods are necessary for screening the large library.

Of particular interest, in 1999 Waldo and coworkers reported a variant of GFP (known as folding reporter GFP, frGFP) and used it for rapid screening of soluble proteins from large library.

Figure 1.26: Folding reporter GFP as a reporter protein for the screening of soluble proteins from a large library. frGFP is expressed as C-terminal fusion of test proteins. If the test protein is folded properly, frGFP is folded and fluorescence is observed. If the upstream test protein is not folded correctly the downstream frGFP fails to fold and no fluorescence is observed.

1.5.4 Folding reporter GFP (frGFP) as a reporter protein for solubility screen

Waldo and coworkers demonstrated that if the upstream protein folds, frGFP can fold and catalyze the formation of a chromophore.335 But if the upstream protein is misfolded or aggregates, GFP fails to fold properly which in turn will cause little or no fluorescence (Figure

1.26). Waldo and coworkers have also examined potential of frGFP as a reporter protein fusing with a panel of 20 different proteins from Pyrobaculum aerophilum with variable solubility and demonstrated that the observed cellular fluorescence had a direct correlation with solubility of test proteins.335 They have used this reporter protein for rapid screening of soluble protein

variants from random libraries of mutant C33T of gene V protein and bullfrog H-subunit of

ferritin to evolve these proteins that are normally prone to aggregation during expression in E. coli.335 Methyl transferase (MT), tartrate dehydratase β-subunit (TD-β), and nucleoside

72 diphosphate kinase (NDP-K) from the hyperthermophilic Pyrobaculum aerophilum were also subjected to directed evolution and screened for the variants with improved folding and solubility using folding reporter GFP as a reporter protein.335

HuPON1 is highly unstable and prone to aggregation when expressed in E. coli. Re-

engineering wild type huPON1 using rational and semi-rational design approaches or directed

evolution methods will be very useful to obtain variants with desired properties. Large libraries

can be screened for variants with better solubility, thermostability and higher expressibility in E.

coli.

73

CHAPTER 2

Re-engineering a split-GFP reassembly screen to examine RING-domain interactions

between BARD1 and BRCA1 mutants observed in cancer patients

2.0 Contributions

This chapter was published as a full paper in Molecular BiosyStems under the authorship of Mohosin Sarkar and Thomas J. Magliery. The work presented in this chapter was produced by the primary author. Analysis of the results and writing of the paper was brought about by the primary and corresponding author.

2.1 Summary

Identification of protein-protein interactions is critical for understanding protein function and regulation. Split protein reassembly is an in vivo probe of protein interactions that circumvents some of the problems with yeast 2-hybrid (indirect interactions, false positives) and co-immunoprecipitation (loss of weak and transient interactions, decompartmentalization). Split

GFP reassembly, also called Bimolecular Fluorescence Complementation (BiFC), is especially attractive because the GFP chromophore forms spontaneously on protein folding in virtually every cell type tested. However, cellular fluorescence evolves slowly in bacteria and fails to evolve at all for some interactions. We aimed to use split-GFP reassembly to examine the determinants of association for a heterodimeric four-helix bundle, and we chose the N-terminal

RING domains of BARD1 and the tumor suppressor BRCA1 as our test system. The wild-type interaction failed to give fluorescence with the split sg100 GFP variant. We found that split

74 folding-repo rter GFP (a hybrid of EGFP and GFPuv) evolves fluorescence much faste

(overnight) with associating peptides and also evolves fluorescence for the BRCA1/BARD1 wild- type pair. Six cancer-associated BRCA1 interface mutants were examined with the system, and only two resulted in a significant reduction in complex reassembly. These results are generally in accord with Y2H studies, but the differences highlight the utility of complementary approaches.

The split frGFP system may also be generally useful for other proteins and cell types, as the split-

Venus system has proven to be in mammalian cells.

2.2 Introduction

Protein-protein interactions are involved in every level of cellular function, from signal

transduction to gene expression and regulation. To understand protein function, identifying

interacting partners is of prime importance. A number of methods are available to study protein-

protein interactions in vitro and in vivo. Co-immunoprecipitation and TAP-tag, protein arrays,

mass spectrometry, yeast two-hybrid analysis, and split protein complementation assays are

among the most prevalent today.6, 7 The split protein complementation approach has become a valuable tool to study protein-protein interactions in a range of cells. In this approach, a protein is split into two fragments which are then fused to potential interacting partners. If the fused proteins interact, the split fragments associate and/or fold, and the function of the split protein is restored. Dihydrofolate reductase (DHFR),27 β-lactamse,29 β-galactosidase,30 and green fluorescent protein (GFP)37 have been used successfully for this purpose.

Ghosh et al. first reported the split GFP fluorescence complementation method in 2000.37

In their system, GFP variant sg100 was dissected in a loop between residues 157 and 158. When

expressed in trans in bacteria, the fragments do not associate to give reassembled GFP. When the

fragments were fused with strongly-interacting leucine zipper peptides and coexpressed, the GFP

reassembled and green fluorescence was observed (Figure 1.2). In 2005, Magliery et al.

75 engineered two compatible plasmid vectors for expression of the fragment fusions that can be co- maintained in E. coli.42, 43 The two plasmids (pET11a-Z-NGFP and pMRBAD-Z-CGFP) have

different origins of replication (ColE1 and p15A, respectively), antibiotic resistances (AmpR and

KanR) and inducible promoters (T7 and araBAD). Upon co-transformation and induction with

IPTG and arabinose for 16 h at 37 °C and 24-48 h at room temperature, or overnight at 30 °C and

3 days at room temperature, cellular fluorescence was observed. Using a library of antiparallel leucine zippers, the method was shown to be very sensitive, detecting interactions as weak as 1 mM in KD due to the irreversibility of the GFP reassembly. However, some known protein

interactions cannot be trapped using this system (such as barnase and barstar; C.G. Wilson, TJM,

and L. Regan, unpublished), and the slow evolution of cellular fluorescence is problematic.

Kerppola and colleagues have also applied this method to a variety of proteins in mammalian

cells, using EGFP and several spectral variants.44, 45 (They call the method Bimolecular

Fluorescence Complementation, or BiFC). Several studies reported that this method has also

been applied in plants and C. elegans.46, 68, 336, 337

We wished to use this system to examine the association of heterodimeric four-helix

bundles. As a test case, we chose the RING domains of the tumor suppressor BRCA1 and its

binding partner BARD1, since they are known to associate in bacteria and their binding has been

explored by yeast 2-hybrid (Y2H) analysis. Mutations of BRCA1 are associated with familial

breast and ovarian cancer, and the protein is involved in various cellular functions such as double-

stranded DNA break repair,209 cell cycle regulation, chromatin remodeling, transcriptional regulation187, 188, 338 and protein ubiquitylation.205, 339 BRCA1 (1,863 aa) principally interacts with

BARD1 (777 aa) through N-terminal RING domains that together constitute an E3 ubiquitin

ligase, the targets of which are largely unknown (although the estrogen receptor appears to be

one340). Klevit and coworkers reported the solution NMR structure of the heterodimer complex, which is essentially composed of four Zn2+ binding sites associated through a four-helix bundle

76

Figure 2.1: Solution structure of BRCA1 and BARD1 RING domain heterodimer complex (1JM7).194 (Left) BARD1 is at left in blue and BRCA1 at right in grey. The bound Zn2+ ions are rendered as pink spheres. The positions of cancer-associated interface mutations, as well as the cancer-associated L52F mutation, are rendered in sticks. (Right) The BRCA1 RING domain is rendered alone for clarity. Indicated mutations are examined in this study. Rendered with PyMOL (Delano Scientific).

interface (Figure 2.1).194 The truncated BRCA1/BARD1 RING/RING complex has E3 ubiquitin ligase activity in vitro.192, 210, 339

About 20 % of clinically-relevant mutations of BRCA1 occur within the N-terminal 100

residues of BRCA1.188 Two classes of known N-terminal cancer-associated mutations have been

reported: RING motif mutations in residues involved in Zn2+ binding sites, and helical packing

residues involved in the RING domain binding interface.195, 200, 341 Altering Zn2+ ligating residues prevents BRCA1 from folding properly and impairs its function. How the interface mutations cause loss of function of BRCA1, what effects they have on interaction of BRCA1 with BARD1, and how they contribute to tumorogenesis is not fully clear yet. One curious observation is that virtually no cancer-associated BARD1 mutants are known, although some truncation isoforms have been observed.342-344 However, a Y2H library approach detected BARD1 interface mutants that reduce association.201 On the other hand, none of the cancer-associated interface mutations of BRCA1 detectably reduce association by the same Y2H screen, although some of them reduce the association with an E2 ubiquitin ligase, and some Zn2+ binding site mutations do appear to abrogate complex formation.202 It is not entirely clear how distal mutations at the interface ablate 77 E2 binding but do not perturb complex formation, although NMR shifts are seen into the helical subdomain upon E2 binding (e.g., K20).195 Mutations at Zn2+-ligating residues (C39A, H41A,

C61G, C61A and C64A) result in a stable complex in vitro,200 although there is little detectable complex in vivo by immunoprecipitation for C61G or C64G.195 By yeast 2-hybrid, C39R, H41R, and C64G appear to bind, but C61G does not.202

We attempted to observe split-GFP reassembly driven by the BRCA1 and BARD1 RING domains in E. coli using the pET11a/pMRBAD plasmid system with the original sg100 GFP variant. Regardless of which protein was fused to which GFP fragment, no fluorescence was observed on plates even after growing cells overnight at 30 °C and 3 days at room temperature.

We therefore set out to optimize the split GFP system for more reliable, robust, faster fluorescence complementation, at least with BRCA1 and BARD1. In 1999, Waldo et al. reported an engineered GFP variant known as folding reporter GFP (frGFP) for use in a screen for soluble proteins.335 The frGFP is a hybrid of EGFP345 (F64L S65T) and the “Cycle 3” mutations (F99S

M153T V163A) of GFPuv reported by Stemmer and coworkers.346 We engineered the same variant from EGFP and GFPuv and used it to make a split frGFP system using the pET11a/pMRBAD plasmids. In this study, we report that the split frGFP system results in a very robust and bright fluorescence complementation within 24 h with the leucine zipper peptides.

Moreover, we observed fluorescence for the pET11a-BARD1-NfrGFP/pMRBAD-BRCA1-

CfrGFP combination. Several BRCA1 RING domain mutants were assayed with the screen, and some but not all appear to abrogate binding.

2.3 Results

2.3.1 Split sg100 system

Magliery et al. developed42, 43 an improved plasmid system for GFP reassembly in E. coli,

dissecting the sg100 GFP between residues 157-158, and resulting in C-terminal fusion to the

78

Figure 2.2: Fluorescence complementation with the split sg100 system. Cells were grown for 24 h at 30 ºC followed by two days at room temperature. (a) pET11-link-NGFP/pMRBAD-link-CGFP, negative control; (b) pET11-BARD1-NGFP/pMRBAD-BRCA1-CGFP; (c) pET11-BRCA1-NGFP/pMRBAD-BARD1- CGFP; (d) pET11a-Z-NGFP/pMRBAD-Z-CGFP, positive control.

NGFP fragment and N-terminal fusion to the CGFP fragment. We replaced the leucine zipper peptides in the original constructs with N-terminal RING domains of BRCA1 (1-109) and

BARD1 (26-140) in both possible orientations (i.e., pET11a-BRCA1-NGFP/pMRBAD-BARD1-

CGFP, pET11a-BARD1-NGFP/pMRBAD-BRCA1-CGFP). Upon co-transformation, plates were incubated at 30 °C for 1 day and at room temperature for 3 days. Green fluorescence was visible for the leucine zipper positive control after 3 days, but no cellular fluorescence was observed for

BRCA1/BARD1 in either orientation, or for a non-cognate negative control (Figure 2.2).

2.3.2 Constructing folding-reporter GFP gene

We generated the frGFP335 from two other variants of GFP (EGFP and GFPuv) available in our lab. Waldo et al. engineered this soluble variant of GFP by introducing F64L (folding mutation) and S65T (red-shift mutation) into a variant (Cycle 3 mutant reported by Stemmer et al. in 1996346, which contains F99S, M153T and V163A) that folds well in E. coli. The frGFP

distinguishes proteins that fold robustly and are highly soluble when expressed in E. coli from

those that tend to misfold and aggregate. The schematic of creating frGFP gene is shown in

Figure 2.3. The 5' fragment (residues 1-84 containing mutations F64L and S65Tmutations) from

EGFP and the 3' fragement (residues 85-238 containing F99S, M153T and V163A mutations)

from GFPuv were PCR amplified, and the full-length folding reporter GFP (frGFP) gene with all 79 5 mutations was assembled using overlap PCR. Cells expressing frGFP were visibly brighter than those with sg100 itself and were about as bright as with EGFP or GFPuv under a hand-held UV lamp (not shown).

Figure 2.3: Schematics of the GFP variants with their corresponding mutations, which are used for split GFP techniques. The frGFP variant was generated from EGFP and GFPuv. The 5' fragment (residues 1-84 containing mutations F64L and S65Tmutations) from EGFP and the 3' fragement (residues 85-238 containing F99S, M153T and V163A mutations) from GFPuv were PCR amplified and the full length frGFP gene was assembled and amplified using overlap PCR.

2.3.3 Split folding-reporter GFP and split EGFP

The frGFP and EGFP were split between 157 and 18 residues using PCR to create the N- terminal fragments from 1-157 residues and the C-terminal fragments from 158-238 residues.

Using the pET11a/pMRBAD plasmid system, the sg100 fragments were replaced with the corresponding fragments of EFGP and, separately, of frGFP, using the positive-control leucine zipper peptides as fusions. The length of time to cellular fluorescence was not improved by

EGFP over sg100. However, overnight incubation (~20 h) at 30 °C and 4-12 h at room temperature was enough to observe bright fluorescence with frGFP (Figure 2.4). Incubation for longer times from 24-72 h showed little or no change in the intensity of fluorescence for split frGFP with zipper peptides or with BARD1/ BRCA1 (Figure 2.4). 80

Figure 2.4: Comparative fluorescence complementation for split sg100 and split frGFP systems. Cells were incubated overnight at 30 ºC (left) followed by additional 48 h at room temperature (right). (a) pET11a-BARD1-NfrGFP/pMRBAD-BRCA1-CfrGFP, new system; (b) pET11a-Z-NfrGFP/pMRBAD-Z- CfrGFP, positive control, new system; (c) pET11a-link-NGFP/pMRBAD-link-CGFP, negative control; (d) pET11a-BARD1-NGFP/pMRBAD-BRCA1-CGFP, old system; (e) pET11a-Z-NGFP/pMRBAD-Z-CGFP, positive control, old system.

Leucine zipper peptides of pET11a-Z-NfrGFP and pMRBAD-Z-CfrGFP vectors were replaced with the N-terminal RING domains of BRCA1 (1-109) and BARD1 (26-140) to obtain four different constructs (pET11-BRCA1-NfrGFP/pMRBAD-BARD1-CfrGFP, pET11-BARD1-

NfrGFP/pMRBAD-BRCA1-CfrGFP). In contrast to the experiment with sg100, 24 h incubation at 30 °C resulted in cellular fluorescence for pET11-BARD1-NfrGFP/pMRBAD-BRCA1-

CfrGFP and the zipper positive control (Figures 2.4 and 2.5). No fluorescence was observed for negative controls (non-cognate pET11-BARD1-NfrGFP/pMRBAD-Z-CfrGFP and pET11-Z-

NfrGFP/pMRBAD-BRCA1-CfrGFP) or for the other cognate fusion orientation (pET11-BRCA1-

NfrGFP/pMRBAD-BARD1-CfrGFP). Fluorescence was observed only when BARD1 is fused to the C-terminus of NfrGFP and BRCA1 is fused to the N-terminus of CfrGFP, possibly due to differences in expression or to the conformational impossibility of binding or GFP reassembly in the other orientation.

81

Figure 2.5: Fluorescence complementation with split frGFP system depends on fusion orientation for BRCA1/BARD1. Cells were incubated overnight for 12-16 h at 30 ºC and for 4-8 h at room temperature befor taking the picture under UV-illumination. The plasmid set pET11a-Z-NfrGFP/pMRBAD-Z-CfrGFP used as positive control (a) and the plasmid sets pET11-Z-NfrGFP/pMRBAD-BRCA1-CfrGFP (b) and pET11-Z-NfrGFP/pMRBAD-link-CGFP (e) used as negative control (b) pET11a-BARD1-NfrGFP/ pMRBAD-Z-CfrGFP, negative control. The two orientations used for BRCA1/BARD1 interactions are pET11a-BARD1-NfrGFP/pMRBAD-BRCA1-CfrGFP (c) and pET11a-BRCA1-NfrGFP/pMRBAD- BARD1-CfrGFP (d ).

2.3.4 BRCA1 cancer-associated mutations

Our goal is to use the split-GFP system to study the determinants of heterodimeric four- helix bundle association. We aim to completely map the determinants of BRCA1/BARD1 interaction using a library approach. As a test, we constructed a small number of cancer- associated mutants of BRCA1, which have been examined by Y2H methods, in vitro methods, or

both. We selected five mutations (see Figure 2.1) which are mostly buried in the helical interface

(V11A, I15T, M18T, M18K and I21V) and one mutation in the RING motif away from the interface (L52F). These mutations were introduced into the BRCA1 N-terminal RING domain gene sequence of pMRBAD-BRCA1-CfrGFP vector by site directed mutagenesis using overlap

PCR. All five mutants were co-transformed with the pET11a-BARD1-NfrGFP plasmid expressing wild-type BARD. Bright fluorescence was observed for both positive controls

(pET11-Z-NfrGFP/pMRBAD-Z-CfrGFP and wild-type pET11a-BARD1-NfrGFP/pMRBAD-

BRCA1-CfrGFP), and no fluorescence was observed for a non-cognate negative control (Figure

2.6). The V11A and M18K mutations showed little or no fluorescence relative to the 82

Figure 2.6: Interaction of BARD1 with cancer-associated mutants of BRCA1 observed by split frGFP reassembly. Fluorescence was observed after 24 h of incubation at 30 ºC. Mutants are labeled as mutations in N-terminal RING domain of BRCA1 which were fused with CfrGFP in the pMRBAD vector. NfrGFP was fused with the N-terminal RING domain of wild-type BARD1 in the pET11a vector.

negative control. The I15T, I21V, and M18T mutations did not markedly reduce the fluorescence, although the L52F mutation away from the interface did reduce fluorescence.

2.3.5 Pull-down assay

The N-terminus of the NGFP fragments in our systems is fused to a hexahistidine tag.

The re-assembled complex can be purified using IMAC methods from the soluble fraction of the

cellular lysate. If the fragments do not assemble, a small amount of the His6-NGFP-analyte is

found in the soluble fraction but most is found in the insoluble fraction, because the individual

GFP fragments are unfolded and poorly soluble. IMAC purification of reassembled complex

after several days of growth is a very sensitive test of interaction, because the reassembly of GFP

is irreversible, and the soluble, refolded complex builds up over time even for weak or transient

interactions. After Ni-NTA purification of the cleared lysate, strong bands corresponding to the

complex components are observed from the positive zipper control by SDS-PAGE (with

Coomassie staining, (Figure 2.7 a). The complex dissociates on the SDS gel due to denaturation

83

Figure 2.7: (a) SDS-PAGE of purified, reassembled complexes by Ni-NTA affinity column. Lane 1, MW markers; 2, positive control (pET11a-Z-NfrGFP/pMRBAD-Z-CfrGFP); 3, wild-type BRCA1; 4: V11A; 5, I15T; 6, M18T; 7, M18K; 8, I21V; 9, L52F; 10, negative control (pET11a-BARD1-NfrGFP/pMRBAD-Z- CfrGFP). The calculated molecular weights of the constructs are H6-NfrGFP-Z, 23 kD; Z-CfrGFP, 13 kD; H6-NfrGFP-BARD1, 33 kD; BRCA1-CfrGFP, 22 kD. (b) Western blot of HA tag-CFrGFP using HRP conjugated anti-HA-Goat poyclonal antibody to confirm the expression level of C-terminal fragments.

of the proteins. Virtually nothing is observed for the non-cognate negative control, although weak Ni-NTA binders from the E. coli proteome purify more strongly in the absence of a His6- tagged protein. For wild-type BRCA1/BARD1 and all of the mutants except V11A and M18K, strong bands corresponding to the complex components are observed. A much smaller amount of complex is purified for V11A and M18K. Any mutation on BRCA1 may affect the expression level of CfrGFP-BRCA1 fusion protein and thereby it may affect the fluorescence level. To confirm the expression levels of BRCA1-mutant fusion proteins we performed a Western blot against an HA tag (fused to the C-terminus of the fusion protein) using HRP conjugated Goat- anti-HA-tag polyclonal antibody (from GenScript). Figure 2.7 b shows the expression levels of all six BRCA1 mutant-fusion proteins and wild type BRCA1. This result suggested that the 84 lowered cellular fluorescence levels we observed for V11A, M18K and L52F, were not due to the lack of expression. In whole-cell lysate, there is no obvious difference in the band patterns for any of the BRCA1 variants, suggesting that there is not a marked difference in the expression of the unassembled fragments (not shown).

2.4 Discussion

Our lab takes combinatorial approaches to understand the folding and association of

proteins. For example, we use a cell-based screen developed by Magliery & Regan to study the

association of the homodimeric four-helix bundle, Rop.347 We wished to use the spilt-GFP

approach introduced by Ghosh & Regan and improved by Magliery et al. to examine the

determinants of heterodimeric four-helix bundle association, and we selected the

BRCA1/BARD1 RNG-domain complex as a test case. BRCA1/BARD1 has the same helical

topology as the well-studied Rop protein, it has been expressed in E. coli as a complex and

characterized structurally, and the association of the RING domains and mutants thereof has been

examined by Y2H and in vitro methods. Moreover, mutations in the RING domains and at the

helical interface between them are associated with familial breast and ovarian cancer, although

the effects of those mutations are not entirely clear. Most notably, none of the interface mutants

appear to abrogate complex formation by Y2H, although some do seem to perturb binding of the

complex to an E2 ubiquitin ligase away from the helical interface.

2.4.1 Improving the split-GFP system for BRCA1/BARD1

Split-GFP reassembly is a complementary approach to Y2H and in vitro methods. By its

mechanism, it demands a direct interaction and is not prone to false positives caused by

autoactivation of gene expression by the “bait” protein sometimes observed in Y2H. It can detect

weak and transient interactions that might be lost by in vitro methods. It is amenable to use in

virtually any type of cell, making it especially useful for library approaches in E. coli, where

85 transformation efficiency is excellent. However, the time to reassembly and chromophore formation is long in the original implementation of the screen, and some known interactions have not successfully led to reassembly. In particular, we found that the BRCA1/BARD1 interaction could not be detected in bacteria using the original fragments of the sg100 variant of GFP.

We set out to improve the screen in bacteria specifically for the BRCA1/BARD1 complex, starting with the EGFP variant and improving it with mutation. We found that EGFP did not perform better than sg100, but that the three additional mutations in GFPuv dramatically improve its performance in reassembly. Reassembly can be observed for the BRCA1/BARD1 complex; for that complex and for strongly-associating leucine zippers, fluorescence is observed in a much shorter time. The improved reassembly of frGFP is likely a result of improved protein folding. Rational mutation to other known fast-folding GFP variants may yield further improvements. For example, a “superfolder” variant of frGFP was engineered by Pedelacq et al.,348 and it shows excellent reassembly at a different dissection point even in the absence of fused, interacting proteins.349 (It is unlikely that it would spontaneously associate if dissected at

157-158, since no other GFP variants self-assemble when dissected at this position.) Shyu and

colleagues have recently demonstrated that split Venus, a fast-folding variant of YFP, shows

improved reassembly in mammalian cells with certain proteins.49, 65 However, none of these proteins were specifically engineered for fast bimolecular folding, which we are now selecting for using directed evolution of the frGFP fragments (MS & TJM, manuscript in preparation).

2.4.2 BRCA1/BARD1 interaction

Our results indicate that three cancer-associated interface mutants of BRCA1 bind to wild-type BARD1 about as well as wild-type BRCA1. However, two interface mutants, V11A and M18K, do not promote reassembly, and one mutation in the RING domain away from the helical interface reduces reassembly somewhat. These data together show that the interface is

86 fairly insensitive to mutation but that underpacking and charge burial can be sufficiently deleterious to prevent binding. Our results are generally consistent with those of Solomon and

coworkers, who examined cancer-associated mutations of BRCA1 RING domain using Y2H.

They found that none of the interface mutants has reduced binding, although some Zn2+-ligating positions resulted in loss of binding. They also found (by Y2H) that some of the interface mutations abrogated binding of the complex to and E2 ligase, even though that interface is far from the helical interface. Of course, it is not known if all of the cancer-associated mutations are causative, and it is conceivable that the interface mutations result in a non-productive association that perturbs the E2 binding site at a distance. It is curious that several similar interface mutations of BARD1 were engineered that abrogate binding by Y2H, but that no cancer-associated BARD1 interface mutations are known.

Our results suggest that more investigation is warranted, including split-GFP assay of other cancer-associated mutations and in vitro biophysical characterization (for example, by gel filtration chromatography). The apparent inconsistency between Y2H, co-IP and gel filtration chromatography for the Zn2+-ligating positions also highlights the necessity of using

complementary techniques for detecting protein interactions to fully understand a given system.

Y2H, split-protein reassembly and in vitro methods have different read-outs, operate in different

concentration regimes and are not equally useful at all interaction strengths. For example, Y2H

has an indirect read-out (gene expression) and the bait and prey fusions are typically

overexpressed. In co-IP, proteins are typically at physiological concentrations, but weak

interactions can be lost. Split protein reassembly has a very direct read-out of association and is

applicable to very weak interactions. One potential limitation of in vivo methods (Y2H and spilt-

GFP) is that weak expression of the fusion can be mistaken for lack of interaction, and we cannot

rule that out for the V11A or M18K cases without in vitro characterization of binding of the

purified, unfused complex. However, our lysate and IMAC purification data is consistent with

87 the fluorescence results. The confluence of data from in vivo and in vitro methods gives the most comprehensive picture of the actual association event. Due to the amenability of the split-GFP

screen to bacteria, our improved system is especially robust for the examination of libraries of

BRCA1 and BARD1 variants to comprehensively probe the determinants of interaction. It is also

easily compatible with high-throughput screening for small molecules that restore complex

formation in disabled mutants, such as V11A, M18K or possibly at the Zn2+-ligating sites,

potentially as leads to novel therapeutics.

2.5 Experimental section

2.5.1 Plasmid construction

The plasmids pET11a-Z-NGFP, pMRBAD-Z-CGFP, pET11a-link-NGFP and pMRBAD-

link-CGFP have unique restriction sites for convenient subcloning of bait and prey proteins. They

are useful in any E. coli strain that expresses T7 polymerase.42 The N-terminus of the NGFP fusion contains a hexahistidine (His6) tag for direct purification of reassembled complex. Cloning into these plasmids was carried out as described.43 Plasmids encoding the N-terminal RING domains of BARD1 (amino acids 26-140) and BRCA1 (1-109) were kindly provided by Rachel

Klevit (University of Washington). BRCA1 and BARD1 were PCR amplified with primers containing XhoI and BamHI sites to subclone into pET11-Z-NGFP and AatII and BsrGI sites to

subclone in to pMRBAD-Z-CGFP.

2.5.2 Construction and cloning of GFP variant

The frGFP gene was constructed from EGFP and GFPuv. N-terminal EGFP (1-84) was

PCR amplified with primers 5'-AATAATAAT CATATG ATGGTGAGCA AGGGCGAGGA G and 3'-

GTA CATAACCTTC GGGCATGGC G GACTTGAAGA AGTCGTGCTG C. C-terminal GFPuv (85-

238 aa) was PCR amplified with primers 5'-GCCATGCC CGAAGGTTATGTA C and 3'-AATATA

GGATCCTTATTTGTAGAGCTCATCCATG CC. Full length frGFP is constructed from these

88 fragments by overlap PCR and amplified with two terminal primers containing NdeI and BamHI

cloning sites to subcloned into a pET11a vector for further use.

2.5.3 Engineering split GFP fusion construct

The frGFP was dissected and created two fragments, residues 1-157 as the N-terminal

fragment and residues 158-238 as the C-terminal fragment. The N-terminal fragment was PCR

amplified with primers, 5'-AAT AAT AAT CATATG GCTAGT CATCAC CACCAT CACCAC GGC

GTGAGC AAGGGC GAGGAG CTG and 3'-AATAAT CTCGAG CCAG AGCCAGAG CCACC

TTGTTTGTCTGCC GTGATG, and cloned into pET11a-Z-NGFP, replacing NGFP (sg100) with

NfrGFP. The C-terminal fragment was PCR amplified with 5'-AATAAT GACGTC GGGTGGAAG

CGGT A AGAATGGAAT CAAAGCTAAC TTC and 3'-AATATA GCGGCCGC TTA

TTTGTAGAGCTCATCCATGC and cloned into pMRBAD-Z-CGFP, replacing CGFP with

CfrGFP. Similar approaches were used to create plasmid sets with EGFP split fragments. These

new plasmid vectors were used to subclone the N-terminal RING domains of BRCA1 and

BARD1 in place of leucine zipper peptides. Similar protocols, as described for sg100 system,

were used to construct pET11a-BRCA1-NfrGFP, pMRBAD-BARD1-CfrGFP, pET11a-BARD1-

NfrGFP and pMRBAD-BRCA1-CfrGFP.

2.5.4 Screening

All the screening experiments were carried out following the protocols described by

Regan and coworkers.42, 43 Compatible pairs of plasmids (e.g., pET11a-BARD1-NfrGFP and pMRBAD-BRCA1-CfrGFP) were cotransformed into BL21(DE3) E. coli electrocompetent cells by electroporation. Cells were grown overnight to a saturation at 37 ºC in LB supplemented with

100 µg mL-1 ampicillin and 35 µg mL-1 kanamycin. Five to 10 µL of 1:1000 dilutions of saturated culture were plated on LB agar media supplemented with 20 µM IPTG (Isopropyl β-D-

Thiogalactopyranoside), 0.2% arabinose and necessary antibiotics. Plates were incubated at 30 °C

89 for 18 to 24 h. For the sg100 system, cells were grown at 30 °C for 24 h and 48-72 h at room

temperature. In each case green fluorescence was observed on a Transilluminator (UVP Inc.)

using long wavelength (365 nm) UV irradiation.

2.5.5 BRCA1 mutants

All BRCA1 mutants were made by the overlap site directed mutagenesis method. V11A,

I15T, M18T, M18K and I21V all four of these are within NcoI and BclI restriction sites which are

117 bases apart from each other. Two synthetic oligonucleotides with the required point mutations and with 18 bases overlap were used to generate a gene sequence between NcoI and

BclI sites. Four sets of oligonucleotides were used for this purpose: VA5'-fw: GCTGATT

TATCTGCTCT ACGCGTTGAA GAAGCCCAAA ATGTCATTAA TGCTATGCAG, VA3'-re:

CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGCATAGCA TTAATGACAT;

IT5'-fw: GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCACTAA

TGCTATGCAG, IT3'-re: CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT

CTGCATAGCA TTAGTGACAT; MT5'-fw: GCTGATT TATCTGCTCT ACGCGTTGAA

GAAGTACAAA ATGTCATTAA TGCTACGCAG, MT3'-re: CCTTGA TCAA CTCTAGACAG

ATGGGACACT CTAAGATTTT CTGCGTAGCA TTAATGACAT; MK5'-fw: GCTGATT

TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCATTAA TGCTAAACAG, MK3'-re:

CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGTTTAGCA TTAATGACAT;

IV5'-fw: GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCATTAA

TGCTATGCAG, IV3'-re: CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGACTTT

CTGCATAGCA TTAATGACAT (bold letter shows the mutation). Two terminal primers

(AATAATTAA CCATG GCTGATT TATCTGCTCTACGC and AATAATAATCCTTGATCAACTCT

AGACAG ATG ) were used to amplify the extended sequence with NcoI and BclI sites. The

plasmid pMRBAD-BRCA1-CfrGFP is used as the vector to clone in these DNA fragments by

replacing wild type BRCA1 for mutant one.

90 For L52F mutation, a modified overlap method is used to insert the point mutation.

Primers used for this purpose are: (5'-fw) CTAACCGGTTCCTT AGCTCGACTCGGC ACGCGT

AACAAAAGTGTCTA T; (5'-re) CTTTCTT CTGGTTGAAT AGTTTCAGCA TG; (3'fw) CATGCTG

AAACTATTCA ACCAGAAGAA AG; (3're) CATAGT CACACG TACGAC GCGAGA GC

AGAATTCTTATGTACA TTATTTGTAGAGCTC. Primers GATTGGCCAA GGAATCGAGCT

GAGCCG and GTATCAGTGTGCATGCTGCGCT CTCG were used for final amplification of the whole sequence between MluI and BsrGI restriction enzyme sites. Inserted mutations were

confirmed by DNA sequencing from Genewiz Inc (South Plainfield, NJ).

2.5.6 Affinity purification of fusion proteins and interacting partners

BL21(DE3) E.coli cells containing compatible plasmids were grown overnight to

saturated culture. Two mL of LB broth containing kanamycin and ampicillin was inoculated with

40 µL of saturated culture for each sample and grown at 37 ºC to OD600 ~0.60. Cells were diluted

1:1000 and 100 µL was plated on screening plates supplemented with 10 µM IPTG, 0.2% arabinose, 100 µg mL-1 ampicillin and 35 µg mL-1 kanamycin. After growing for 24 h at 30 °C and 48 h at room temperature, cells were resuspended in 1X phosphate buffered saline (PBS).

The OD600 of 100-fold diluted cells were measured to normalize the cell densities. Cells were

then harvested by centrifugation and each pellet was resuspended in 2.5 mL of lysis buffer (50

mM Tris-HCl, 200 mM NaCl, 100 µM ZnCl2, 0.1% Tween 20, 5 mM β-mercaptoethanol, 10 mM imidazole, pH 8.0) containing 0.5 mg mL-1 HEW lysozyme, DNase, RNase, PMSF

(polymethyl sulphonyl fluoride), and 0.5 mM MgCl2. After sonication and centrifugation, cleared

lysate was collected, mixed with 100 µL of Ni-NTA agarose (Qiagen) equilibrated with lysis

buffer and left at 4 °C for 2 h with gentle shaking. The resin was washed twice with 5 volumes of

wash buffer (lysis buffer containing 20 mM imidazole) and purified proteins were eluted with 300

µL of elution buffer (lysis buffer containing 250 mM imidazole).

91 2.5.7 Western blotting using anti-HA-tag antibody

Split fragments with fused proteins or peptides were expressed following the same procedure mentioned above. An appropriate amount of cell pellet was resuspended in 200 µL of

lysis buffer (50 mM Tris-HCl, 200 mM NaCl, 100 µM ZnCl2, 0.1% Tween 20, 5 mM β-

mercaptoethanol, 10 mM imidazole, pH 8.0) and mixed with 100 µL of glass beads (Biospec,

Sigma), vortex very hard to get the complete lysis. Following centrifugation for 30 min at

maximum speed, cleared lysate was collected and mixed with an equal volume of SDS buffer.

Samples were electrophoresed on 12.5 % SDS-PAGE gel. Protein bands were transferred on a

PVDF membrane (from Pierce) using TE22 Mighty Small Transphor Unit (80-6205-59) from

Amersham Biosciences following manufacturer’s protocol. Following Pierce-OneStep TMB Blot

protocol, membrane was treated with 40 mL of 1-4,000 dilution anti-HA-Goat-HRP in TBST

buffer (25 mM Tris-HCl, 150 mM NaCl, 0.05% Tween-20, pH 7.6). Membrane was then treated

with OneStep TMB substrate (from Pierce) following the supplier’s protocol. Image of the blot

was taken under White Light UV illumination using GelLogic 100 imaging system from Kodak.

2.6 Acknowledgements

We like to thank Professor Rachel Klevit for the plasmids containing BRCA1 and

BARD1 N-terminal RING domain genes. This work was supported by The Ohio State University.

92 CHAPTER 3

Re-engineering split-GFP fragments for efficient and faster fluorescence complementation

to study protein-protein interactions in vivo

3.0 Contributions

This is a manuscript that will be submitted shortly with the authorship being, in order:

Mohosin Sarkar, Dustin Maly and Thomas J. Magliery. The work reported in this chapter was produced and written up by the primary author. Part of the work was designed in collaboration with second author from University of Washington, Seattle. The experimental design and data analysis was accomplished by the primary and corresponding author.

3.1 Summary

Protein–protein interactions play a critical role in most biological processes, and offer attractive opportunities for drug design. With remarkable progress in molecular biology, an increasing number of protein–protein interactions have been identified as potential targets for the development of anticancer drugs. Developing small molecules that modulate protein-protein interactions has been very difficult because of the lack of adequate knowledge about protein interaction surfaces and the lack of high-throughput in vivo screening methods in their native cellular context. Split protein reassembly is an in vivo probe of protein interactions that has some significant advantages over other widely used methods, such as yeast 2-hybrid assay, TAP-tag and coimmunoprecipitation. Unlike the split GFP assay, these techniques suffer from high background noise, false positive read out and can not detect weak or transient interactions.

93 Among all protein fragment complementation techniques, split GFP fragment reassembly is especially attractive because the GFP chromophore forms spontaneously on protein folding in virtually every cell type and subcellular localization tested. However, with the existing split sg100 GFP variant, cellular fluorescence evolves slowly in bacteria (2- 4 days) and fails to evolve at all for some interactions. It also failed to give fluorescence for the interaction of BARD1 with tumor suppressor protein BRCA1. Re-engineered split folding-reporter GFP (frGFP, hybrid of

EGFP and GFPuv) evolved faster and produced efficient fluorescence in 20-30 h at 30 °C with associating peptides and also evolved fluorescence for the BRCA1/BARD1 wild type pair. Split frGFP fragments were re-engineered and further improved by directed evolution (error-prone

PCR followed by several rounds of DNA shuffling) and bimolecular selection to obtain very

efficient, fast and robust fluorescence complementation. The evolved N-terminal and C-terminal

fragments were able to reassemble and generate fluorescence in as little as 12-16 h at 30 °C and in 10-14 h at 37 °C. As a proof of principle experiments, in addition to the BRCA1/BARD1, the new system was successfully tested for interactions of several therapeutically important protein pairs (Bcl-xL/Bim, Bcl-2/Bim, p53/hDM2 and XIAP/Smac) which have key roles in apoptosis and abnormal cell proliferation. The evolved split fragments were able detect the interactions and generate cellular fluorescence in 10 to 14 h for all protein pairs. The new system was also tested for inhibition of these interactions using known inhibitors. Reduced or no fluorescence was detected with increasing concentration of inhibitors, such as ABT-737 against Bcl-xL/Bim or Bcl-

2/Bim and Nutlin3 against p53/hDM2. These results suggest that the efficient split GFP method

we developed will be a valuable tool for in vivo screening of small molecule or cyclic peptide

libraries to develop effective modulators of therapeutically important protein-protein interactions

in their native cellular context from direct fluorescence reassembly readout.

3.2 Introduction

Protein-protein interactions (PPIs) play key roles in every fundamental biological 94 process, from signal transduction to gene expression and regulation, transcription and translation, cell growth and differentiation, intercellular communication and programmed cell death. Proteins frequently interact with themselves or other proteins and function as stable or transient complexes to regulate a wide variety of cellular functions. Many of these protein-protein interactions are part of the larger cellular networks and are tightly regulated by a number of different mechanisms.

Alterations in these interactions perturb the normal sequence of events in the cell and contribute to diseases. Therefore, for a deeper understanding of the biological functions and the mechanisms, the identification and characterization of protein interaction networks, new interaction partners are essential. Understanding the pattern of protein-protein interactions, interaction networks and involved mechanisms can lead to the development of drugs to fight the underlying cause of diseases.

A large number of methods have been developed to study protein-protein interactions in vitro. Co-immunoprecipitation and TAP (tandem affinity purification)-tag, protein arrays, mass spectrometry and surface plasmon resonance spectroscopy are among the most prevalent today.6, 7

Only a limited number of methods, such as Yeast two-hybrid (Y2H) assay, fluorescence resonance energy transfer (FRET), bioluminescence resonance energy transfer (BRET), and split protein complementation assays are available to study protein-protein interactions in vivo. Each of these methods has its own advantages and limitations.1, 7 Y2H is one of the most widely used

systems to screen or confirm PPIs. Classic Y2H systems are limited to protein interactions in the

yeast nucleus. In spite of having many improvements and variations, this technique still suffers

from high background noise, high rates (~50%) of false-positives2 and failure to detect weak or

transient interactions (Kd > 1 µM). Though FRET has been very useful for real-time detection of

PPIs from direct read-out in vivo, it also suffers from similar drawbacks and is not useful for

high-throughput screening. To overcome these limitations, the split protein complementation

assay (PCA) has become a valuable tool to study protein-protein interactions in a wide range of

95 cells. In this approach, a functional protein is split into two inactive fragments which are then fused to potential interacting partners, ‘prey’ and ‘bait’ proteins. When expressed separately, the two fragments remain nonfunctional and do not reassemble spontaneously. Interaction between fused prey and bait proteins facilitates the association of split fragments to reconstitute the intact protein and restore its original function. Examples other proteins that have been used successfully for this purpose are dihydrofolate reductase (DHFR),27 β-lactamse,29 β-galactosidase,30

luciferase29, and green fluorescent protein (GFP).37 Among all PCA techniques, the split GFP method is more attractive because the GFP chromophore forms upon protein folding in almost every cell type and subcellular . Unlike FRET and other PCA methods, the split GFP technique has little or no false-positive read-out. It does not require any exogenous substrate and detects protein-protein interactions in their native cellular milieu. It can detect very weak (Kd ~1

mM ) and transient interactions because of the irreversibility of the process. The method has been

used for high-throughput screening from large libraries to identify new protein interactions and

interacting partners,73 and also to study and characterize structural determinants required for

interactions between antiparallel leucine zipper peptides.42

One of the most remarkable advantages of fluorescence complementation is that in

addition to the direct fluorescence readout, it enables simple and direct visualization of protein

interactions in their native state in different subcellular locations of almost every cell type and

organism.44 45 Using split GFP reassembly, Chalfie and coworkers in 2004 reported the

visualization of protein interactions in a whole organism, C. elegans, Figure 1.5.46 Using a

number of tissue and cell specific promoters they have shown that the expression of GFP

fragments and reconstitution of GFP were not promoter or tissue specific. Following Chalfie’s

work, using bZIP domains of Fos and Jun, Hu and coworkers in 2008 demonstrated the use of

fluorescence complementation assay to identify, visualize and validate temporal and spatial

protein interactions in living worms.64 In a most recent study, they have also described a

96 significant improvement of split GFP assay by combining split Venus approach with a Cerulean-

based FRET to study protein-protein interactions in a ternary complex in living cells.65

Though fluorescence complementation assay has been used increasingly for detection and

visualization of protein-protein interactions, the widespread use of this technique has been limited

by a number of factors. The main drawback of this method is that cellular fluorescence evolves

slowly (3-4 days) in bacteria and fails to evolve at all for some interactions tested [such as,

barnase and barstar (CG Wilson, TJ Magliery, and L Regan, unpublished); and BRCA1/BARD1

(M Sarkar and TJ magliery, Chapter 2)]. In almost all fluorescence reassembly assay driven by

different spectral variants of GFP reported to date, incubation of the cells or organisms for about

4-48 hours at lower temperature (room temperature or 20 °C) is required. Also, in almost all of

these cases screening failed to work at 37 °C partly because individual fragments are not stable at

higher temperature. Several other groups, including Kerppola, Hu, Umezawa and Michnick,

reported the development of split GFP assay using different spectral variants of GFP and splitting

at various positions on the loops; but still all these systems suffer from similar kind of limitations.

With increasing demand of studying and detecting protein interactions in vivo in more native-like

environment and at physiological condition, it necessitates the search for a pair of GFP fragments

that will allow fluorescence complementation to perform at both 30 and 37 °C with shortest

possible reassembly time.

In this work we report a faster and more efficient split folding reporter GFP (frGFP)

system which results bright cellular fluorescence in less than 30 h (compared to the existing

systems which take about 2-4 days42). We further improved the system by applying directed evolution of split fragments (error-prone PCR and DNA shuffling) and selection to obtain highly efficient reassembly. The evolved fragments with leucine zipper peptides showed efficient reassembly and robust cellular fluorescence in 12-16 h at 30 °C, and 10-14 h at 37 °C, which was limited by the time required for sufficient growth of the cells. To our knowledge, this is the

97 fastest split GFP method to observe any direct phenotype under UV lamp for studying protein- protein interactions in vivo. The re-engineered efficient split fragments will be very usefull for

studying protein-protein interactions in the native cellular context in vivo.

Alterations in protein-protein interactions or protein interaction networks from mutation

or deregulation of one of the partners (or both) perturb the normal sequence of events in the cell

and contribute to diseases. With remarkable progress in molecular biology, biotechnology and genomics, an increasing number of protein–protein interactions have been identified as potential therapeutic targets for the development of small molecule drugs for cancer therapy.79, 91 Currently, it has been of a great therapeutic interest to inhibit specific protein-protein interactions in order to induce apoptosis to treat a variety of disease states. A number of protein-protein interactions have

been targated that are critically involved in the regulation of cell proliferation and apoptosis or

programmed cell death. The most widely studied potential protein-protein interaction targets in

the intrinsic apoptotic pathway that have been studied extensively and explored for small

molecule drug discovery are interaction between Bcl-2/Bcl-xL and proapoptotic Bcl-2 family proteins (Bim, Bak, Bid, Bax, Bak); X-linked inhibitor of apoptosis proteins (XIAP) and their modulator Smac (DIABLO); p53 and its negative regulator hDM2 (human double minute-2) and

BRCA1 and its stabilizing partner BARD1.106, 113, 115, 125, 136, 141, 350-352

We tested evolved esGFP fragments for the detection of several therapeutically important

protein-protein interactions, such as Bcl-xL/Bim, Bcl-2/Bim, p53/hDM2, XIAP/Smac. All of

these interactions are considered to be an important target for small molecule drug discovery for

cancer therapy. Bcl-xL/Bim, Bcl-2/Bim and p53/hDM2 protein/peptide pairs were studied in vivo

with known inhibitors to determine if the evolved fragments were able to respond to the

suppression of interactions. Preliminary results suggest that this technique has great potential for

small molecule drug discovery to find potent modulators of therapeutically important protein-

protein interactions in vivo in their native cellular environment.

98 3.3 Results

3.3.1 Re-engineering split folding reporter GFP

Ghosh et al. first reported the split GFP fluorescence complementation method in 2000 to study protein-protein interactions in vivo.37 In their system, the GFP variant sg100 was dissected in a loop between residues 157 and 158. When expressed in trans in bacteria, the fragments do not associate to give reassembled GFP. When the fragments were fused with strongly-interacting leucine zipper peptides and coexpressed, the GFP fragments reassembled and green fluorescence was observed (Figure 1.2). In 2005, Magliery and colleagues reported the improvement of the system by engineering two compatible orthogonal plasmids for expression of the fusion fragments that can be co-maintained in E. coli.42, 43 The two plasmids (pET11a-Z-NGFP and

pMRBAD-Z-CGFP) have different origins of replication (ColE1 and p15A, respectively),

Figure 3.1: Fluorescence complementation with the split sg100 (left) and frGFP (right) systems. Left: Cells were grown for 24 h at 30 °C followed by two days at room temperature. (a) pET11-link- NGFP/pMRBAD-link-CGFP, negative control; (b) pET11-BARD1-NGFP/pMRBAD-BRCA1-CGFP; (c) pET11-BRCA1-NGFP/pMRBAD-BARD1-CGFP; (d) pET11a-ZNGFP/pMRBAD-Z-CGFP, positive control. Right - Fluorescence complementation with split frGFP system depends on fusion orientation for BRCA1/BARD1. Cells were incubated overnight for 12-16 h at 30 ºC and for 4-8 h at room temperature befor taking the picture under UV-illumination.

antibiotic resistance markers (AmpR and KanR) and inducible promoters (T7 and araBAD). Upon co-transformation and induction with IPTG and arabinose for 16 h at 37 °C and 24-48 h at room temperature, or overnight at 30 °C and 3 days at room temperature, cellular fluorescence was 99 observed, Figure 1.2. Using a library of antiparallel leucine zipper peptides, they were also reported that the method was very sensitive and able detect interactions as weak as 1 mM in KD

due to the irreversibility of the GFP reassembly. Kerppola and colleagues have also applied this

method to a variety of proteins in mammalian cells using EGFP and several spectral variants.44, 45

They have also used this system for simultaneous visualization of multiple protein interactions in

the same cell (they call this Bimolecular Fluorescence Complementation, or BiFC).45

We wished to use this split GFP system to examine the association of heterodimeric four-

helix bundles. As a test case, we chose the RING domains of the tumor suppressor BRCA1 and

its binding partner BARD1 (BRCA1 associated RING domain 1), since they are known to

associate in bacteria and their binding has been explored by yeast 2-hybrid (Y2H) analysis.194, 202

BRCA1interacts with BARD1 through their N-terminal RING domains and forms a stable antiparallel four-helix bundle heterodimeric complex.194 The complex is involved in various cellular functions such as double-stranded DNA break repair, cell cycle regulation, chromatin remodeling, transcriptional regulation and protein ubiquitylation.179, 183, 184, 192, 194, 342 A number of

studies reported that mutations in BRCA1 occur in 50 % of women with inherited breast cancer

and up to 90 % of women with combined breast and ovarian cancer. We found that the

BRCA1/BARD1 interaction could not be detected in bacteria using the original fragments of the

sg100 variant of GFP (Figure 1.2). We set out to improve the screen in bacteria specifically for

the BRCA1/BARD1 complex, starting with the EGFP variant and improving it with mutation.

We found that EGFP did not perform better than sg100 system. In 1999, Waldo et al. reported an

engineered GFP variant known as folding reporter GFP (frGFP) for use in a screening of soluble

proteins from a large library.335 The frGFP is a hybrid of EGFP345 (F64L S65T) and the “Cycle 3”

mutations (F99S M153T V163A) of GFPuv reported by Stemmer and coworkers.346 We engineered the same variant in our lab from EGFP and GFPuv (as described in Chapter 2) and

used it to make a split frGFP fragments splitting between 157 and 158 residues (N-frGFP residues

100 1-157 and C-frGFP residues 158-238). Using the standard pET11a/pMRBAD plasmids (Figure

1.3), N-frGFP and C-frGFP fragments were cloned in by replacing sg100 fragments. We observed faster fluorescence complementation and improved fluorescence with this new split frGFP system Figure 3.1. Significant cellular fluorescence complementation was observed after

12-16 h at 30 °C and 8-12 h at room temperature. No fluorescence reassembly was observed at 37

°C. We have used this new system to examine the interaction of BRCA1/BARD1. Leucine zipper

peptides of pET11a-Z-NfrGFP and pMRBAD-Z-CfrGFP vectors were replaced with the N-

terminal RING domains of BRCA1 (1-109) and BARD1 (26-140) to obtain four different

constructs (pET11-BRCA1-NfrGFP/pMRBAD-BARD1-CfrGFP or pET11-BARD1-

NfrGFP/pMRBAD-BRCA1-CfrGFP sets). In contrast to the experiment with sg100 fragments,

20-30 h incubation at 30 °C resulted in cellular fluorescence for BRCA1/BARD1interactions,

Figure 3.1. No fluorescence was observed for negative controls (non-cognate pET11-BARD1-

NfrGFP/pMRBAD-Z-CfrGFP and pET11-Z-NfrGFP/pMRBAD-BRCA1-CfrGFP sets).

3.3.2 Re-engineering split frGFP fragments for faster and efficient reassembly

It was evident from our results that split frGFP evolves fluorescence complementation much faster (20 - 30 h) than most of the split GFP system including original split sg100 (3-4 days), and also worked with noncognate protein pairs tested such as BRCA1/BARD1. Although this is a huge improvement for the split GFP system, it still takes around 20 to 30 h to give any significant visible fluorescence reassembly and it does not work at 37 °C.

We attempted to further improve the split frGFP system using directed evolution (error- prone PCR followed by rounds of DNA shuffling) of split fragments and bimolecular selection.

We chose to use BRCA1/BARD1 as a model system, since it was already known that it worked with the frGFP system and had intermediate fluorescence (Figure 3.1). The N-terminal and the

C-terminal frGFP fragments were subjected to error-prone PCR separately to introduce random

101

Figure 3.2: Directed evolution and bilmolecular selection of split-frGFP fragment libraries. (a) Schematic of directed evolution (Error-prone PCR and DNA Shuffling) of split-GFP fragments and selection, (b) shows the screening plate and hit (brighter colony inside red circle) from error-prone PCR libraries, (c) and (d) show hits at 30 °C and 37 °C after several rounds of DNA shuffling and bimolecular selection for faster reassembly and brighter fluorescence.

102 mutations and cloned both libraries with BARD1 and BRCA1 fusion into pET11-BARD1 and pMRBAD-BRCA1 vectors, respectively, designed by Magliery et al.43 Compatible plasmids

containing N-terminal and C-terminal libraries were cotransformed into BL21(DE3)

electrocompetent cells and screened on agar media plates at 30 °C and 37 °C for brighter and

faster cellular fluorescence reassembly. By screening hundreds of thousands of colonies, we were

able to isolate a total of 24 positive colonies with bright fluorescence. Schematic of 1st round of

directed evolution and screen results are shown in Figures 3.2 a and b.

The N-terminal and the C-terminal GFP fragments were PCR amplified separately from

24 hits obtained from screening of error-prone libraries and subjected them to a round of DNA shuffling following the protocol described by Joyce et al.353 Following reassembly and amplification, both library fragments were subcloned back into pET11-Null-BARD1 and pMRBAD-BRCA1-Null vectors with N-terminal RING domain of BARD1 and BRCA1, respectively. Compatible plasmids were cotransformed into BL21(DE3) cells and bimolecular selection was carried out at both 30 and 37 °C on agar plates. No hits were obtained at 37 °C, whereas the hit rate at 30 °C was much higher than that obtained from error-prone libraries. The

200 brightest colonies were collected and pooled; from which N-terminal and C-terminal libraries of GFP fragments were PCR amplified separately. The library fragments were subjected to multiple rounds of DNA shuffling and selection. In each round of shuffling, screening was carried out at both 30 and 37 °C on agar plate for brighter and faster fluorescence reassembly (10 to 18 h). After three rounds of DNA shuffling and bimolecular selection, we isolated a number of colonies which showed significantly improved split fragment complementation and bright fluorescence in 14-18 h at 30 °C and in 10 -14 h at 37 °C as shown in Figures 3.2 c and d.

The eight brightest colonies were isolated from the selection at each 30 and 37 °C screening as shown in Figure 3.3. The N-terminal and the C-terminal fragments were further characterized and their genes were sequenced. Out of sixteen hits, the three single brightest

103

Figure 3.3: Cellular fluorescence observed from selected colonies isolated after 3rd of DNA shuffling and selection at both 30 °C and 37 °C. NZ/CZ represents the split frGFP fragments fused with leucine zipper peptides. Celles were grown for 14 h at 30 °C and 10 h at 37 °C following 2-6 h at room temperature.

Figure 3.4: Cellular fluorescence observed from evolved fragments (C4, C13 and C19) and wild type regular frGFP fragments with leucine zipper peptides after 14 h at 30 °C and 10 h at 37 °C following 2 h at room temperature. Cells were cotransformed with evolved NfrGFP fused with leucine zipper peptide and CrrGFP fused with linker for negative controls.

colonies (C4, C13 and C19) were isolated which have significantly lower temperature sensitivity, robust fluorescence reassembly and work at both 30 °C and 37 °C. The N-terminal and C- terminal fragments of those were PCR amplified and subcloned back with leucine zipper peptides into standard plasmid sets (pET11-Z-NGFP and pMRBAD-CGFP-Z vectors, Figure 1.3).

Cellular fluorescence reassembly (for clones C4, C13 and C19) were tested at 30 and 37 °C and

compared with that produced by wild type split frGFP fragments as shown in Figure 3.3. These 104 fragments were able to evolve very efficient fluorescence reassembly in as little as 10-16 h at 30 and 37 °C. The evolved fragments were tested for any spontaneous reassembly by cotransforming the compatible plasmid sets containing N-terminal fragments fused to leucine zipper peptide and

C-terminal fragment fused to a linker. All three evolved split fragment pairs generated no detectable cellular fluorescence and showed no spontaneous (non-cognate) reassembly as shown in Figure 3.4 labeled as (-). Among these three selected clones, C19 showed very robust and

efficient fluorescence reassembly at both 30 and 37 °C compared to other two evolved pairs.

Mutations accumulated on C19 frgamnets (we call this efficient split GFP or esGFP)

from error-prone PCR and rounds of DNA shuffling are listed in Table 4.1 and shown on split

fragments in Figure 3.5.

Table 3.1: Mutations on the N-terminal and C-terminal fragments isolated from C19 clones after erroe- prone PCR and rounds of DNA shuffling.

C19GFP Frgamnents Mutations accumulated during evolution N-C19GFP E17K, K26R, T62S, A87T, T97I C-C19GFP R168C, I171V, E172K

Figure 3.5: A model structure of the GFP constructed from the evolved N-terminal and C-terminal fragments isolated from C19 clone. Mutations on the evolved fragments accumulated after error-prone PCR and rounds of DNA shuffling are shown cyan sticks. The N-terminal fragment is in green and the C- terminal fragment is in red. The model structure was created by SWISS Homology Modeling (http://swissmodel.expasy.org/) and by PYMOL (DeLano Scientific). 105 To investigate the viability of the split fragments for protein reassembly an in vitro reconstitution experiment was carried out. Purified complex (with leucine zipper peptides) was denatured in 6

M guanidine hydrochloride and slowly dialyzed into 10 mM phosphate buffer containing 2 mM

DTT at pH 7.2 over 14 h at 4 °C for the reconstitution of complex and recovery of fluorescence complementation following an experiment originally reported by Ghosh et al.8 The split fragments showed about 30% recovery of the reassembly and excitation and emission spectra match with that of wild type frGFP ones as shown in Figure 3.6 a and b. Following similar experiments, a comparison of in vitro reassembly of sg100 fragments, frGFP fragments and evolved C19 fragments is shown in Figure 3.6 c. Lower recovery and much slower reassembly were observed for sg100 fragments; whereas, recovery and reassembly were much higher for frGFP and C19 fragments. Between the last two, the evolved fragments showed a higher percentage of recovery of fluorescence reassembly. This may be because the evolved fragments are comparatively more soluble and stable at 30 and 37 °C, and can reassemble faster.

3.3.3 Targeting protein-protein interactions for small molecule drug discovery

Protein interactions have critical roles in most biological processes, and provide attractive opportunities for drug design. It is now evident that proteins recognize and bind targets in a highly specific manner and regulate most of the cellular functions and ultimate fate of the cells.

Interaction of tumor suppressor protein p53 with its natural antagonist E3 ubiquitin ligase hDM2 is the most widely studied system. DM2 binds with p53 through its transactivation domain with high affinity and negatively regulates its stability and transcriptional activity. It is found that in many human tumors hDM2 was over expressed, which effectively impairs p53 function.

Inhibition of hDM2-p53 interaction can stabilize p53 and restore its function. This may lead cancerous cells to apoptosis and offers a novel strategy for cancer therapy. Anti-apoptotic Bcl-2 family proteins (Bcl-2, Bcl-xL, Mcl1, etc.) protect cells from apoptosis by inhibiting the action of pro-apoptotic proteins (Bim, Bax, Bak, Bid, etc). Inhibition of the interaction of Bcl-2 family 106

Figure 3.6: In vitro reconstitution experiment for the evolved split fragments. Purified complex was denatured in 6 M Gu-HCL overnight and dialyzed into a buffer containing 2 mM DTT, 10 mM phosphate buffer at pH 7.2 over 14 h at 4 °C.for the recovery of fluorescence reassembly. Samples were taken out every 1 h, quenched and absorption and emission spectra were recorded (a). About 30 % of recovery was observed after 14 h of dialysis. (b) In vitro reconstitution of fluorophore observed for evolved c19, frGFP and sg100 split fragments. (c) comparison of excited and emission wavelength scan for three complexes. sg100 complex: ex 475, em 507; frGFP complex: ex 488, em 510; C19 complex: ex 488, em 512. 107 proteins with pro-apoptotic proteins will help these proteins to function positively, which will activate Bax/Bak mediated cytochrome c release, and initiate mitochondria mediated apoptotic pathway and lead cancerous cells to apoptosis. Another important target for cancer therapy is the interaction of XIAP (X-linked inhibitor of apoptosis protein) with its pro-apoptotic partners Smac

(second mitochondria- mediated activator protein)/DIABLO (direct IAP-binding protein with low

pI).354 The BIR3 domain of XIAP specifically binds with and inactivates caspase 9, while BIR1

and 2 bind more specifically with caspases 3 and 7 and inactivate them. DIABLO or Smac, an

endogenous inhibitor of XIAP, is released from mitochondria in response to multiple apoptotic

stimuli112, 113 and bind to the BIR3 domain of XIAP, which in turn inhibits binding of caspase to

XIAP. These initiate the caspase-activated pathways and induce apoptosis and cell death.

Target based rational design of small molecule drugs has not been very successful because of lack of adequate knowledge and information about the PPI surface and its structure.

Library based approaches have been very useful but are hampered by the lack of available efficient and high-throughput methods for large library screening in vivo. ELISA (enzyme linked

immunosorbant assay) and surface plasmon resonance have been used successfully for in vitro

characterization of small library. T7 phage display library method has only been applied recently

to p53/hDM2 system with limited success.355 For large proteins, the phage display technique may

not be useful. We speculated that the evolved esGFP fragments might be very useful for this

purpose to study PPIs in vivo and screen small molecule libraries in the context of cellular

environment. As a test case, we aimed to study several therapeutically important protein pairs,

p53/hDM2, Bcl-xL/Bim, Bcl-2/Bim, XIAP/Smac and BRCA1/BARD1. The N-terminal DNA binding domain of hDM2 (25-109 aa), the N-terminal Bcl-xL (1-197 aa) and Bcl-2 (1-200 aa), and

the N-terminal RING domain of BRCA1 were fused with evolved C-esGFP. Bim peptide (59-72

aa, MRPEIWIAQELRRI GDEFNA), p53 peptide (15-29 aa, SQETFSDLWKL LPEN) and

BARD1 were expressed as a fusion protein with evolved N-esGFP. On the other hand the BIR3

108

Figure 3.7: Cellular fluorescence complementation of protein-protein interactions of Bcl-xL/Bim, Bcl- 2/Bim, XIAP/Smac, p53/hDM2 and BRCA1/BARD1, obtained using evolved efficient split GFP fragments. (10 h at 37 °C and 4-6 h at room temperature). Smac1: AVPI, Smac : AVPFYAAK.

domain (242-353 aa) of XIAP was expressed as a fusion of N-esGFP and Smac peptides (AVPI or AVPFYAAK) were expressed as fusion of C-esGFP. Since all of these proteins and peptides are of similar sizes, first we optimized the linker length between GFP fragments and protein or peptide of interest for Bcl-xL and Bim interactions.

With optimized linker length for both fragments, Bcl-xL/Bim pairs generated efficient fluorescence complementation in 12-18 h at both 30 and 37 °C (Figure 3.7). Re-engineered split fragments were also able to detect protein-protein interactions and fluorescence complementation for all other protein pairs we tested (Bcl-2/Bim, p53/hDM2, XIAP/Smac and BRCA1/BARD1) as shown in Figure 3.7. Efficient and robust fluorescence reassembly was observed for all the protein pairs in 12 -16 h at 37 °C and 14-18 h at 30 °C. Relatively higher fluorescence was observed for Bcl-2/Bim compared to Bcl-xL/Bim. This is may be because Bcl-2 binds more

strongly and the protein is more stable. Several studies reported that binding between Bcl-2 and

Bim (KD ~450 nM) is much stronger than Bcl-xL and Bim (KD ~800 nM). The cellular

fluorescence level we observed was in consistent with these reported values. The Smac peptide

109 with BIR3 domain of XIAP and p53 peptide with hDM2 binds with dissociation constant values in the nanomolar range. We also observed that both Smac peptides (AVPI, AVPFAYYK) with

BIR3 domain of XIAP and p53 peptide with hDM2 bind significantly tightly and evolved very strong cellular fluorescence (Figure 3.7).

3.3.4 Detection of inhibition with small molecule inhibitors

Using target-based design, Vassilev et al.141 from Hoffmann-La Roche discovered and developed a modified small molecule, Nutlin3, which binds with hDM2 with low nanomolar affinity and inhibit the interaction with p53. Suppression of the hDM2 interaction led to the activation of p53 pathway, cell cycle arrest and apoptosis. Using a fragment-based drug discovery and high- throughput NMR screening (SAR by NMR), Rosenberg, Fesik et al.104 discovered a small

molecule, ABT-737, targeting the interaction of Bcl-2 family protein with pro-apoptotic proteins

Bak and Bim. ABT-737 was further modified and derivatized to have significantly higher potency

against Bcl-2 family protein interactions. These compounds bind to Bcl-2 family proteins and

impair their function to inhibit pro-apoptotic proteins, which in turn activate apoptosis pathways.

Oost et al. reported a series of tripeptides that bind to the XIAP BIR-3 domain with nanomolar

affinities against the human breast cancer cell lines BT-549 and MDA-MB-231; the melanoma

cell line (SK-MEL-5), and the acute promyelocytic leukaemia cell line HL-60.125 As a proof of principle experiment, we aimed to test the re-engineered split fragments against both of these compounds for the inhibition of protein-protein interactions and suppression of fluorescence complementation. ABT-737 was tested against both Bcl-xL/Bim and Bcl-2/Bim, and Nutlin3

against p53/hDM2 interactions. Cells were grown on plates with different concentrations of

inhibitor (100 µM, 10 µM, 1 µM and 0 µM) with appropriate resistant markers and observed for

fluorescence reassembly. It is evident from Figure 3.8 that in the presence of inhibitors, protein pairs were not able to interact with each other as strongly as the positive control and generated lower cellular fluorescence reassembly. Suppression of fluorescence reassembly of Bcl-xL/Bim 110

Figure 3.8: Suppression of protein-protein interactions by the small molecule inhibitors ABT-737 and Nutlin3. Cellular fluorescence read out of on plate assay obtained after 10 h at 37 °C and 2-6 h at room temperature using a filter that allows only blue light to pass through for the excitation of GFP. (a) Bcl- 2/Bim interaction inhibition with ABT-737; (b) Bcl-xL/Bim interaction inhibition with ABT-737; and (c) The inhibition of hDM2/p53 interaction with Nutlin3. 1 - 100 µM, 2 - 10 µM, 3 - 1 µM and 4 - 0 µM of total concentration of inhibitor on the agar plates.

111 was higher with increasing concentration of ABT-737. Because of the stronger binding between

Bcl-2 and Bim, and p53 and hDM2, it was difficult to obtain increasing suppression with higher concentration of inhibitors. But, it was evident from screening on agar plates (Figure 3.8) that at higher concentration of inhibitors (100 µM) reassembly was suppressed to a great extent for Bcl- xL/Bim, Bcl-2/Bim and p53/hDM2.

3.4 Discussion

Split-GFP reassembly is a complementary approach to Y2H and in vitro methods. By its mechanism, it demands a direct interaction and is not prone to false positives caused by auto- activation of gene expression by the ‘‘bait’’ protein sometimes observed in Y2H. This method is limited to protein interactions in the nucleus of yeast. Though there have been significant improvements and different variations of this system for studying PPIs in mammalian cells, in bacteria and in yeast cytoplasm, this method still suffers from high rate of false positives. The other widely used method of studying protein-protein interactions, fluorescence resonance energy transfer (FRET), also has limited use because of its false positives read out and higher background noise. On the contrary, split GFP methods have very little or no false positives and can detect weak and transient interactions that might be lost by in vitro methods. It is amenable to use in virtually any type of cell including human cells. It is especially useful for library approaches in E. coli, where transformation efficiency is excellent. However, the use of this method has been limited because of slower fluorescence maturation and low overall fluorescence.

It requires overnight incubation of plates at 30 °C and 2-3 days at room temperature before it evolves any significant fluorescence. Screening does not work at all at physiological temperature and also for some known interactions (such as, Barnase/barstar and BRCA1/BARD1) the sg100 system did not successfully lead to reassembly. Split frGFP system we developed has been proven to be very useful and is able to give efficient and bright fluorescence reassembly in less than 30 h. It has been very useful for studying protein-protein interactions such as RING domains 112 of BRCA1 and BARD1. We tested for the interaction of BARD1 with several cancer predisposing mutants of BRCA1 using this system. Preliminary studies showed that the interface of BRCA1 and BARD1 is fairly insensitive to mutation but underpacking and charge burial can be sufficiently deleterious to prevent binding.56 Our results are generally in consistent with those

of Solomon and coworkers, who examined cancer-associated mutations of BRCA1 RING domain

using Y2H, but highlight some important differences. We are in progress of making all known

cancer predisposing mutants of BRCA1 [44 point mutations reported to date in Breast Cancer

Infromation Core (BIC), National Institute of Health (NIH)] and examine their interaction in vivo

with BARD1.

Using error-prone PCR and rounds of DNA shuffling of individual frGFP fragments, we

have re-engineered the split GFP fragments and further improved the system for very efficient

and fast fluorescence reassembly. In each round the evolved split fragments were selected for

faster reassembly at both 30 and 37 °C. We were able to select and isolate the fragments whose

reassembly time was limited by the growth of the cells from 10 to 14 h at 37 °C and 12- 16 h at

30 °C. Fluorescence complementation was efficient, robust and fast compared to sg100, wild type

frGFP and other spectral variants of GFP systems. The mutations that accumulated in the N-

terminal and C-terminl GFP during directed evolution were listed in Table 3.1. Though we were

not able to determine the effect of individual mutations, it was evident that their combined effect

helps fragments reassemble faster and efficiently. Since, most of these mutations were on the

surface of both fragments (except T62S), Figure 3.5, their additive effect may have improved the

overall solubility and stability of the fragments in E. coli. This may, in turn, helps fragments

reassemble faster. Since, the T62S mutation is close to the GFP chromophore, this may have

affected and improved the fluorophore maturation and its brightness.

Re-engineered efficient split GFP (esGFP) fragments will provide an efficient and

reliable approach for faster complementation with bright a cellular fluorescence for the study of

113 protein-protein interactions in vivo in their natural cellular environment. This method will also be very useful for high-throughput library approaches to study and identify protein-protein interactions and new interacting partners. With this new development, the split GFP technique

eliminates most of the drawbacks of Y2H, FRET and the existing split GFP system.

Protein–protein interactions have critical roles in most biological processes, and provide

attractive opportunities for drug design. The development of inhibitors of protein–protein

interactions is a complicated process due to a number of specific challenges; such as fairly large

interaction surfaces (approximately 750 – 1500 Å2);79 lack of adequate information about

interaction surface, structure of the interacting partners, and types of interaction taking place; and

lack of efficient high-throughput screening methods.79, 80 A number of computation-based

approaches are available to predict the PPIs and their types from genome-wide sequencing.156-159

Scientists have used these approaches with modeling of interaction surface to design small molecule drugs with limited success. Because of these limitations, fragment -based and library- based methods have been the best choices for small molecule drug discovery for cancer therapy.356 For this purpose an efficient and reliable high-throughput method is necessary to study

protein-protein interactions in vivo. As a proof of principle experiment, we tested the new split

system with known inhibitors of BCL-xL/Bim and Bcl-2/Bim, and p53/hDM2 for the inhibition of interactions and suppression of fluorescence complementation. Reassembly was suppressed to a great extent for Bcl-xL/Bim, Bcl-2/Bim and p53/hDM2, as detected by the esGFP fragment

screening, shown in Figure 3.8. These results were very promising in a sense that the re-

engineered esGFP fragments were not only able to detect very efficient and robust fluorescence

reassembly with these non-native protein pairs, they were also sensitive enough and capable of

detecting the suppression of protein-protein interactions by small molecule inhibitors like ABT-

737 and Nutlin3. These preliminary results suggest that the evolved esGFP system will be a

valuable tool for high-throughput screening of small molecule library or cyclic peptide library for

114 lead compounds for drug discovery and cancer treatment.

In recent years, protein fragment complementation assay has drawn interest for its simplicity, convenience and ease to use for therapeutic drug discovery targeting protein-protein interactions. Split GFP or fluorescence complementation assay would be especially useful for high-throughput screening of small molecule or peptide libraries in vivo from direct fluorescence

read out. There are a number of advantages which make the fluorescence complementation assay

very attractive. It is very simple to set up, and does not require the production of recombinant

protein and protein purification. It does not require the use of expensive equipment like NMR or

crystallography. It enables the use of combination of drugs which may affect multiple targets

simultaneously. Primary screens are carried out in living cells in their native cellular

environment. In order to carry out the screen, small molecules have to be membrane permeable

and therefore more likely to be bioavailable. Fluorescence complementation assay could also be

useful for studying two further classes of therapeutic agents such as antibodies or other protein

therapeutics, and proteins that activate their targets by an allosteric mechanism. This method can

be used to aid our understanding in molecular mechanisms and action of drug in human cells and

has the potential to enhance the productivity of therapeutic drug discovery targeting protein-

protein interactions.

3.5 Experimental section

3.5.1 Plasmid construction

The plasmids pET11a-Z-NGFP, pMRBAD-Z-CGFP, pET11a-link-NGFP and pMRBAD-

link-CGFP have unique restriction sites for convenient subcloning of bait and prey proteins. They

are useful in any E. coli strain that expresses T7 polymerase.42 The N-terminus of the NGFP fusion contains a hexahistidine (His6) tag for direct purification of reassembled complex. Cloning into these plasmids was carried out as described.43 Plasmids encoding the N-terminal RING domains of BARD1 (amino acids 26-140) and BRCA1 (1-109) were kindly provided by Rachel 115 Klevit (University of Washington). BRCA1 and BARD1 were PCR amplified with primers containing XhoI and BamHI sites to subclone into pET11-Z-NGFP and AatII and BsrGI sites to

subclone in to pMRBAD-Z-CGFP. Oligonucleotide primers were obtained from IDT (Integrated

DNA Technology, Coralville, IA).

3.5.2 Construction and cloning of GFP variants an split GFP fragments

The frGFP (F64L, S65T, F99S, M153T and V163A) gene was constructed from EGFP

containing F64L and S65T mutations at the N-terminus and GFPuv containing F99S, M153T and

V163A mutations at the C-terminus. The schematic of constructing of frGFP gene was described

in details in Chapter 2.

The frGFP was dissected as residues 1-157 and 158-238. The N-terminal fragment (1-

157 residues) was PCR amplified with primers, AAT AAT AAT CATATG GCTAGT CATCAC

CACCAT CACCAC GGC GTGAGC AAGGGC GAGGAG CTG, AATAAT CTCGAG

CCAGAGCCAGAGCCACC TTGTTTGTCTGCC GTGATG, containing a sequence encoding His6 tag, and cloned into pET11a-Z-NGFP, replacing NGFP (sg100) with NfrGFP. The C-terminal fragment was PCR amplified with AATAAT GACGTC GGGTGGAAG CGGT A AGAATGGAAT

CAAAGCTAAC TTC and AATATA GCGGCCGC TTA TTTGTAGAGCTCATCCATGC and cloned

into pMRBAD-Z-CGFP, replacing CGFP with CfrGFP. These new plasmid vectors were used to

subclone the N-terminal RING domains of BRCA1 and BARD1 in place of leucine zipper

peptides. Similar protocols, as described for sg100 system, were used to construct pET11a-

BRCA1-NfrGFP, pMRBAD-BARD1-CfrGFP, pET11a-BARD1-NfrGFP and pMRBAD-

BRCA1-CfrGFP.

3.5.3 Screening

All the screening experiments were carried out following the protocols described by

Regan and coworkers.42, 43 Compatible pairs of plasmids (e.g., pET11a-BARD1-NfrGFP and

116 pMRBAD-BRCA1-CfrGFP) were cotransformed into BL21(DE3) E. coli electrocompetent cells

by electroporation. Cells were grown overnight to a saturation at 37 ºC in LB supplemented with

100 µg mL-1 ampicillin and 35 µg mL-1 kanamycin. A 5 to 10 µL of 1:1000 dilution of saturated culture were plated on LB agar media supplemented with 20 µM IPTG (Isopropyl-β-D-

Thiogalactopyranoside), 0.2% arabinose and antibiotics. Plates were incubated at 30 °C for 18 to

24 h. For the sg100 system, cells were grown at 30 °C for 24 h and at room temperature for 48-72

h. In each case, green fluorescence was observed directly on a transilluminator (UVP Inc.) or

using a Visi Blue Converter Plate (P/N 38-0200-01 fromUVP Inc) that passes blue light only (for

the excitation of GFP chromophore) using a long wavelength (365 nm) UV irradiation. The

Image of the plates was captured using GelLogic 100 imaging system from Kodak or by using a

digital camera.

3.5.4 Error-prone PCR

Error-prone PCR amplification of N-terminal and C-terminal frGFP were carried out

357 following the protocol described by Joyce et al. For each 50 µL reaction 7 mM MgCl2, 0.5 mM

MnCl2, 1mM d(C/T)TP, 0.2mM d(A/G)TP, 2.5 Units of Taq polymerase and 500 nM each

oligonucleotides were used. Primers used for the amplification of NfrGFP are 5': AATAAT

CCCGGG ATGGCTAG TCATCATCAT C and 3': AATAAT CTCGAG CCAGAGCC AGAGCCACC

TTGTTTGTCTGCCGTGATG; and CfrGFP are fw: AATAAT GACGTC GGGTGGAAG CGGT A

AGAATGGAAT CAAAGCTAAC TTC and re: AATATA GCGGCCGC TTA TTTGTAGAGCTC

ATCCATGC.

3.5.5 DNA shuffling

The initial PCR of split fragments has carried out using a standard PCR protocol using primers for: NfrGFP, 5': AATAAA CGATCCCGCG AAATTAATACG and 3': ATGCTAG

TTATTGCT CAGCGG; and CfrGFP, 5': AATCGAGATTT AGTCAACTTG TTGAAGAGC and 3':

117 GATAT GGGGC AAATG GTGG. PCR fragments were gel purified using the Qiagen gel extraction

kit or membrane-GF/F filter paper capture method (described in Chapter 6). For DNA digestion

reactions, DNaseI (10 U/µL from Roche) was prepared to 0.01 U/ µL by a dilution of 1:1000

using 500 µL glycerol, 100 µL 10X DNaseI buffer (200 mM Tris HCl pH 8), 100 µL 10X CaCl2

(50 mM) and 300 µL dH2O. The template DNA (1-10 µg) was dissolved to a concentration of

100 ng/µL in 1X DNaseI buffer containing 10 mM MnCl2. Each 50 µL of this diluted DNA

sample was equilibrated to 15 °C for 10 min before treating with 1-2 µL of diluted DNaseI. The

digestion reaction was carried out for 10 min at 15 °C and the reaction was immediately heated to

90 °C for 10 min to inactivate the DNaseI. The reaction was electrophoresed on 2 % agarose gel

to check the desired fragments (typically 25-200 bp). The reassembly reaction included 10-30

ng/µL of each fragments, with 1X Taq buffer, 2.2 mM MgCl2, 0.25 mM dNTPs (each), 0.5-1 µL

Taq (2.5-5 U) polymerase per 100 µL. The reaction was carried out using the following program:

94 ºC for 2:30 sec, 94 ºC for 0:30 sec, 47.5 ºC for 0:45 sec, 72 ºC for 0:10 sec, 39 more cycles

(steps 2-4 with extension time increasing by 5 sec per cycle), final extension at 72 ºC for 10 min, and 4 ºC for storage. Final amplification was carried out using the terminal primers used for error- prone PCR. A 1-10 or 1-40 dilution of the reassembly reaction was used as template in 100 µL of reaction which was carried out using the standard PCR protocol.

3.5.6 In vitro reconstitution experiment

The C19 split fragments with leucine zipper peptides were cotransformed into

BL21(DE3) E. coli cells and overexpressed on agar media for 2 days for complete reassembly.

Cells were resuspended in 1X PBS and the complex was purified using Ni-NTA agarose resin

following the standard protocol. Purified complex was denatured in 6 M guanidine hydrochloride

overnight. Denatured fragments were dialyzed into a buffer containing 2 mM DTT and 10 mM

phosphate buffer at pH 7.2 over 14 h at 4 °C for the reconstitution of complex and recovery of

fluorescence complementation. 100 uL of reactions were taken out every hour and tested for 118 fluorescence using wavelenghts 488 nm for excitation and 509 nm for emission.

3.5.7 Bcl-xL, Bcl-2 and Bim peptide cloning

A pET11a vector containing Bcl-xL gene and a pCDNA vector containing Bcl-2 gene

were generous gift from professor Dustin Maly, Chemistry Department, University of

Washington, Seattle. Bcl-xL gene sequence coding 1-197 aa was PCR amplified using 5'-fw:

AATAATAATCCATGGCG TCTCAGA GTAATCGGGAGCTGGTGGTTGACTTTCTC, and 3'-re:

AATAATAATGACGTC CCGCTACCGCTGCTACCGCCGCTAC CGCTGCTA CCGCT ACCGC

TACCCTTG TTCCCATAGAGTTCCACAAA AGTATCC, and subcloned into the pMRBAD vector

as an N-terminal fusion of C19-CGFP between Nco1 and AatII sites. Similarly, the N-terminal

Bcl-2 gene (coding for 1-200 aa) was cloned in using following primers, 5'-fw: AATAATAAT

CCATGGCCGCGCACGCTGGGAGAACAGGGTACG and 3'-re: AATAATAAT GACGTC

CCGCTACCGCTGCTA CCGCCGCTACCGCTGCTACCGC TACCGC TACCCTT TTCCACAAAT

GCATCCCAGCC TCCGTTATCCTGG. The Bim peptide sequence used for Bcl-xL or Bcl-2

interaction was ‘MRPEIWIAQELRRIGDEFNA’. A synthetic DNA sequence (from IDT) coding

for this peptide with linker SSGSSGSG was cloned in to pET11a vector as a C-terminal fusion of

C19-NGFP between XhoI and BamHI sites.

3.5.8 XIAP, Smac peptide, and hDM2 and p53 peptide cloning

A pCDNA3.1 vector containing XIAP gene was a generous gift from Professor Dustin

Maly, Department of Chemsitry, University of Washington, Seattle. A pCMV vector containing

hDM2 gene was a genereous gift from Professor Jiandong Chen, MOFFITT Cancer Center and

Research Institute, Tampa, Florida. The BIR3 doamin of XIAP (242-355 aa) was PCR amplified

with linker using primers AATAATAATCTCGAGCGGTAGCAGTGGTTCAGGTGGTAGCAG

TGGTTCAGGT TTCCCAAATTCAACAAATCTTCCAA GAAATCCATC (5') and AATAATATA

GGATCC TTAAGTAGTTCTTACCAGACACTCCTC AAGTG (3') and subcloned as a C-terminal

119 fusion of C19-NGFP between XhoI and BamHI sites. The Smac peptides AVPI and AVPFYAAK were used for hDM2/Smac interaction study. Synthetic DNA sequences (CCATGGCGGTGCCTA

TAAGC GGAG GT AGCGGTAGCGGTAGCAGCGGTAGCGGCGGTAGCAG CGGTAGCGGGAC

GTC and CCATGGCGGTGCCTTTTTATGCAGCGAAGAGCGGAGGTAGCGGTAGCG GTAGC

AGCG GTAG C GGC GGTAGCAGCGGTAGCGGGACGTC, from IDT) coding for these peptide

with appropriate linker were cloned into the pMRBAD vector as an N-terminal fusion of C19-

CGFP between Nco1 and AatII sites. Similarly, hDM2 (25-109 aa) and p53 peptide (17-29 aa)

coding sequences were PCR amplified with appropriate linker and subcloned as a fusion of C19-

CGFP between XhoI and BamHI sites, and C19-NGFP between Nco1 and AatII sites.

3.5.9 Nutlin3 and ABT-737 inhibition assay

ABT-737 was obtained from Dustin Maly, Chemistry Department, University of

Washington, Seattle, and Nutlin3 was bought from Cayman Chemical, Ann Arbor, Michigan.

Inhibition experiments were carried out on agar plates supplemented with appropriate antibiotics,

10 µM IPTG, 0.2% arabinose. Agar media on plates was also supplemented with 100 µM, 10

µM, 1 µM or 0.0 µM inhibitors (ABT-737 or Nutlin3). Overnight saturated culture of cells was

diluted to 1 to 10,000 and 20 µL of which was grown on appropriate plates. Cells were grown for

9-12 h at 37 °C or 12-14 h at 30 °C following 2-6 h at room temperature to obtain significant

fluorescence complementation or inhibition.

3.6 Acknowledgements

We are grateful to Professor Rachel Klevit, University of Washington, Seattle, for

providing us with BRCA1 and BARD1 RING domain containing plasmids. We like to thank

Professor Jiandong Chen, MOFFITT Cancer Center and Research Institute, Tampa, Florida for

providing us with hDM2 gene containing plasmid.

120 CHAPTER 4

Rational design of human paraoxonase-1 (PON1) for higher expression and solubility in E.

coli

4.0 Contributions

This is a manuscript that will be submitted shortly with the authorship being, in order:

Mohosin Sarkar, George Matic, Braden Competty and Thomas J. Magliery. The vast majority of the work reported in this chapter was produced and written up by the primary author. Cloning and expression of one of the engineered huPON1 variants, huPON1H, were carried out by the second author. Optimization of chaperone co-expression was carried out by the third author. The experimental design and data analysis was accomplished by the primary and corresponding authors.

4.1 Summary

High density lipoprotein (HDL) associated serum paraoxonase-1 (PON1) has been implicated in diabetes and cardiovascular diseases. It is shown that the overexpression of this protein can prevent atherosclerosis development in mice. Paraoxonase-1 also exhibits broad substrate specificity against organophophorous (OP) pesticides and nerve agents. Human paraoxonase-1 is considered to be an excellent candidate for the development of a drug as a catalytic bioscavenger for effective pre- and post-exposure treatment of OP intoxication. Low expressibility and poor stability from expression in a non-native environment have made it very difficult to study the structure-activity relationship of PON1 and its involvement in OP

121 intoxication and antiatherogenicity. To acquire the status of a catalytic bioscavenger, solubility,

stability and large-scale expression and purification of active human PON1 are among the most

difficult challenges. Expression and purification of active human PON1 from bacterial systems

have not been very successful. PON1’s hydrophobic leader sequence, hydrophobic surfaces on

the HDL binding sites and the lack of post-translational modifications in bacteria are considered

to be some of the reasons for its low stability and solubility in E. coli. In addition to this, to

increase the level of soluble folded protein in E. coli, we have applied the approaches of

chaperone co-expression and an MBP (maltose binding protein) fusion at the N-terminus of

huPON1. By optimizing purification conditions, we were able to express active, wild-type human

PON1 and the engineered variants in large-scale with a high degree of purification and solubility.

With a method to obtain sufficient quantities of active protein in hand, it is now possible to apply

site-directed mutagenesis or directed evolution to reengineer the human paraoxonase-1 to obtain a

variant with higher catalytic activity against OP pesticides and nerve agents.

4.2 Introduction

The importance of human PON1 in OP pesticide and nerve agent detoxification has been

known for long time.236, 276, 358, 359 Human PON1 is a calcium-dependent protein which is

synthesized in the liver and secreted into the blood where it is exclusively associated with

HDL.225, 280, 290, 360 Kuo and La Du in 1998 reported that the enzyme is associated with two

227 calcium ions with Kd values of 0.36 and 6.6 µM. The higher affinity site is considered to be

essential for enzyme stability, and the other one is presumed to be for PON1’s hydrolytic catalytic

activity. The full length protein in serum retains its hydrophobic leader sequence. Studies from

Sorenson et al. and Tawfik et al. suggested that the retained N-terminal signal sequence is

necessary for its binding with HDL.226, 361 Their studies reported that PON1 is able to bind with

HDL by directly binding to HDL-associated phospholipids through its retained N-terminal

122

hydrophobic peptide. The crystal structure of recombinant human PON1 (G2E6) showed that this

signal peptide is flexible and unstructured.272 In a non-native environment and with no HDL

particles or possibly no human phosphate binding protein around, huPON1 is highly unstable.

More recent studies have implicated PON1 in the pathogenesis of cardiovascular disease and

reported that serum PON1 plays a significant role in the prevention of aetherosclerosis

development.362, 363 Human PON1 is synthesized mostly in the liver (and at a low level in kidney)

and secreted into the blood where it is exclusively associated with high-density lipoproteins

(HDL).225, 280, 290, 364-367 It was found that the activity of PON1 is inversely related the risk of cardiovascular diseases and diabetes.281, 368, 369 Shih et al. demonstrated that the transgenic mice

lacking serum paraxonase are susceptible to atherosclerosis development and organophosphate

poisoning.245, 269 Later Oda et al. reported that the overexpressing of PON1 in mice is capable of preventing the development of atherosclerosis.301 The antiatherosclerosis role of PON1 is believed to be due in part to its antioxidative properties. PON1 has been shown to protect against oxidative stress by hydrolyzing the hydroperoxide lipids in LDL and HDL.293, 370

More importantly here, human PON1 is known to catalyze the hydrolysis of neurotoxic

organophosphate (OP) pesticides and nerve agents.371 PON1, named for its ability to hydrolyze the organophosphate paraoxon, has been studied in the field of toxicology for a long time. In

1992 La Du and coworkers reported that human PON1 can catalyze the hydrolysis of a number of

OP pesticides such as parathion, diazinon, and chlorpyrifos.267 Later, several studies from other groups found that it can also hydrolyze OP nerve agents such as sarin, tabun and soman.237, 372

However, the catalytic efficiency of PON1 with most of these OP compounds is poor.

Organophosphate pesticides (oxon metabolites of pesticides) and nerve agents are potent inhibitors of acetylcholinesterase. Exposure to these compounds leads to the inhibition of acetylcholinesterase, thereby leading to an accumulation of excessive acetylcholine in the central and peripheral nervous system. This causes cholinergic syndrome, which can progress to 123

paralysis and respiratory arrest. OP exposure can also cause temporary or permanent damage to

the neurological and motor functions. The current multi-drug treatments of OP exposure using

anticholinergic drugs, oximes and the anticonvulsant atropine, has not been that effective because

it does not prevent from post-exposure incapacitation, seizures and permanent neurological

damage.258, 373 Despite tremendous need in fighting against biological threat posed by OP nerve agents and exposure to OP pesticides, scientists have not been very successful at finding efficient drugs or bioscavengers.

Lately, the idea of an enzymatic bioscavenger to detoxify OP agents has been very popular because of its potential to pre- and post-exposure treatment with no side effects. Among first generation stoichiometric bioscavengers, acetylcholinesterase and butyrylcholinesterase have been proven to be very effective in treating OP intoxication.260, 374 However, usefulness of these scavengers as therapeutics is limited by the fact that large stoichiometric amounts are necessary for higher level of OP exposure treatment. Because of these disadvantages, it has long been sought to develop and obtain catalytic bioscavengers for effective pre- and post-exposure treatment of OP intoxication. Because of its known hydrolysis activity against OP pesticides and nerve agents, human paraoxonase is considered to be an excellent candidate for the development of therapeutic drug as a catalytic bioscavenger.223, 263-265

Though there have been many ongoing studies due to its involvement in cardiovascular disease, diabetes and antiatherogenicity, there is little known about its structure, catalytic activity and mechanism. Very low solubility and thermal stability in a non-native environment make it very difficult to obtain PON1 in a reasonable quantity and study the structure-function relationship. Tawfik and coworkers engineered a recombinant chimeric (mix of rabbit mouse, rat and human) form of huPON1 (known as G2E6) with better stability and expressibility in E. coli by directed evolution and screening against 2-napthylacetate, DepCyC and paraoxon hydrolysis activity.273 It has 70% sequence homology with rabbit PON1 and a total of 59 amino acid 124

mutations compared to human PON1. The crystal structure of this chimeric PON1 was reported

in 2004 by the same group (Figure 1.24).272 But, it failed to shed light on the structure-activity

relationship and on the catalytic mechanism of human PON1. From further DNA shuffling and

screening, Tawfik et al. have isolated another variant of recombinant chimeric PONI, G3C9, with

better biophysical characteristics.273 Both of these chimeric PON1 variants have better solubility and thermal stability, but they failed to show improved catalytic activity against OP pesticides and nerve agents other than paraoxon (Otto TC, Harsch CK, Cerasoli DM and Magliery TJ, unpublished results).

The previous attempts to express and purify huPON1 in large-scale from non-native hosts have not been very successful. More recently Furlong and coworkers reported the first successful expression and purification of wild type human PON1 and variants from E. coli with limited yield.276 For its use as a therapeutic agent, huPON1 faces a number of critical challenges. Large-

scale expression and purification of fully-humanized protein, stability of the protein when

expressed in E. coli or other non-native hosts, and improvement of activity towards OP pesticides

and nerve agents have been among the most difficult ones. Human PON1 is highly unstable and it

aggregates when expressed in E. coli. A number of structural features contribute to this. Mature

huPON1 retains its hydrophobic leader sequence, hydrophobic patches for HDL binding sites

formed by three α-helices and the loops before and after the second helices and two N-linked

glycosylation sites on the surface. When expressed in bacteria, these unmodified hydrophobic

surfaces on human PON1 make it less stable and prone to aggregation. We attempted to address

these issues by rational design and/or directed evolution of wild type huPON1 and recombinant

variants.

In recent years rational design or directed evolution approaches have been applied to

achieve the efficient heterologous expression of proteins with increasing solubility and stability in

bacteria and yeast but with limited success.316 There have not been many strategies and studies 125

available about where and exactly what kind of mutations need to be applied to increase the

solubility of a protein. In rational design, site-specific changes are made on the target enzyme

based on the detailed information and knowledge about the protein structure, function and

catalytic mechanism. Bryan and coworkers demonstrated the rational designing of subtilisin E by

introducing nine point mutations to obtain a calcium-independent hyperstable variant.320

Engelman and coworkers in 2001 reported the re-engineering of phospholamban to a soluble pentameric helical bundle by replacing its lipid-exposed hydrophobic residues with charged and polar residues.321 Based on computational design, DeGrado and coworkers rationally engineered a water soluble analog of phospholamban by changing membrane-exposed positions to polar or charged amino acids while the putative core was left unaltered,322 and also a water soluble analog

of potassium channel KcsA, by mutating the lipid-contacting side chains to more polar groups.323

Here we report the re-engineering of wild type huPON1 and its variants using rational

and semi-rational design to obtain variants with higher solubility and expressibility in E. coli. By

redesigning the expression vector and optimizing the expression and purification conditions, we

were able to express and purify active wild type huPON1 and its variants from E. coli in large-

scale with high degree of purity.

4.3 Results

We tried to address the poor solubility and folding of huPON1 using rational and semi-

rational approaches for large-scale expression and purification of active enzyme from E. coli. The

crystal structure of recombinant chimeric PON1 indicated that H1 and H2 helices form

hydrophobic patches in the proximity of the active site provide a potential surface for interaction with the lipid layer of HDL.272 Hydrophobic residues proposed to be involved in HDL anchoring are Y185, F186, Y190, W194, W202 (from helix H2 and the adjacent loops) and K21 (from helix

H1).272 We mutated the putative HDL binding surface with polar or charged residues

126

hypothesizing that it might help PON1 to be more soluble and stable in the absence of HDL in E.

coli. Residues mutated for these purposes are F24E, Y185E, F186Q, I187K, Y190K, L191Q,

W194K, L198E, L200Q, W202K, M289Q, and F293E (we call this variant huPON1H). These

surface mutations may stabilize the protein and also may help E. coli expressibility and solubility.

Figure 4.1: Schematic of the designed pET11 vector for using folding reporter GFP (frGFP) as a C- terminal fusion of test proteins. A linker was inserted between NdeI and KpnI site for standard cloning of test proteins by replacing a TEV protease site (ENLYFQSG) and a linker (GSSG) in frame between test proteins and frGFP. The frGFP contains a C-terminal hexahistidine (His6) tag for purification purpose.

In a different approach, we engineered another variant of huPON1 by applying solubilizing mutations over the whole surface of huPON1. Chimeric recombinant PON1, G2E6, is comparatively more stable and soluble when expressed in E. coli. Compared to the wild

type huPON1, there are a total of 59 residues in G2E6 that are different, most of which are on the surface of the protein. We hypothesized that mutations that are on the surface and more polar or charged compared to huPON1 might contribute most to its solubility. We selected 15 of such mutations (I5T, N19R, Q21K, L31H, N78D, N80D, S81K, P82S, L96S, G101E, A137S, N166S,

127

Q192K, Y197H, N265D and N309D) from G2E6 and put them on the huPON1 surface. In

addition to these 15 residues, a buried N166S mutation which is in very close proximity of K192

was also introduced into huPON1. This mutation is considered to be a possible compensatory

mutation of K192 in G2E6. The second variant of huPON1 was engineered by introducing all 16

mutations in huPON1 (we call this variant huPON1G). We rationalized that this variant would be

more human like with better stability and solubility when expressed in E. coli.

A third variant of huPON1 was generated by deleting the unstructured hydrophobic part

(residues 4 to 17) of the N-terminal leader sequence. Two additional variants of huPON1 were

also generated by deleting the N-term 4 to 17 residues from huPON1H and huPON1G.

All five of these variants and wild type huPON1 were expressed as an N-terminal fusion

of a reporter protein, folding reporter GFP. Waldo and coworkers reported this variant of GFP in

1999 and demonstrated that if the upstream protein folds, frGFP can fold and catalyze the

formation of the chromophore.335 But if the upstream protein is misfolded or aggregates, GFP fails to fold properly which in turn will cause little or no fluorescence (Figure 1.26). They have also demonstrated that the observed cellular fluorescence has a direct correlation with solubility of protein and reported the use of it as a reporter protein for rapid screening of soluble protein from a random library. The frGFP distinguishes proteins that fold robustly and are highly soluble when expressed in E. coli from those that tend to misfold and aggregate. We designed a pET expression vector using frGFP as a C-terminal fusion of PON1with a TEV protease site in between, as shown in Figure 4.1. Several proteins (Cys free T4 lysozyme, T4LTA; yeast triose phosphate isomerase, yTim; G2E6 and huPON1) with varying degree of solubility were cloned into this pET11a vector as an N-terminal fusion of frGFP to test the correlation of cellular fluorescence with the solubility of expressed proteins. T4LTA lysozyme and yTIM are known as highly soluble and well-expressed proteins in E. coli, chimeric G2E6 has intermediate solubility and huPON1 has very low solubility when expressed 128

Figure 4.2: Whole-cell fluorescence measured for the cells expressing proteins with wide range of solubility and as C-terminal fusion of reporter protein frGFP. Fusion proteins were expressed at 30 °C and cells were normalized to same cell density before measuring the fluorescence using 475 nm excitation and 490 to 550 nm emission scan.

in E. coli. We observed that the solubility of the proteins expressed in E. coli was directly

correlated to the fluorescence of the cells expressing the corresponding frGFP fusion (Figure

4.2). This observation further validated the potential of frGFP as a reporter protein for the rapid

screening of folded soluble proteins.

From the cellular fluorescence obtained from the expression of wild type huPON1 and its

variants, it was apparent that the mutations we introduced into wild type huPON1 were in fact

contributing to solubility of the protein in E. coli (Figure 4.3) since all engineered variants have

higher fluorescence level compared to wild type huPON1. Interestingly, we have noticed that the

engineered variant huPON1G and chimeric G2E6 have similar solubilities in E. coli. Only 16

charged or polar residue mutations on the huPON1 surface made the protein as soluble as the

chimeric G2E6. The polar or charged residue mutations on putative HDL binding sites made

huPON1 most soluble compared to the wild type and huPON1G. It was also evident from the

cellular fluorescence data that the deletion of only the N-terminal hydrophobic leader sequence 129

made the huPON1 and engineered variants more soluble compared to their full-length

counterparts. The data also suggested that when the two approaches (mutations on surface polar

residues and deletion of N-terminal hydrophobic leader sequence) were combined, their effects

were additive and increased the solubility of huPON1 significantly. Among all the variants we

engineered, ΔhuPON1H (HDL binding site mutations and N-terminal leader sequence deletion)

acquired the highest level of soluble protein expression in E. coli. Attempts to purify these

variants as frGFP fusions from E. coli were not successful because of the low expression level of

the frGFP fusions.

Figure 4.3: Whole-cell fluorescence of frGFP fusion proteins of wild type huPON1 and recombinant G2E6 and their variants expressed in E.coli. Fusion proteins were expressed in E. coli for 4 h and cell density (OD600) was adjusted to 0.1 before measuring the fluorescence spectra at 510 nm.

There are a number of studies available that reported the successful expression of human proteins in E. coli as a fusion to a more soluble and expressible partner, such as maltose binding protein (MBP), thioredoxin (Trx), glutathione S-transferase (GST), etc.311 As a matter of fact,

even the recombinant PON1 (G2E6) was expressed as a fusion wit a Trx-tag in E. coli.273

Because of its high expressibility and solubility, we used MBP.311 We designed a vector with 130

Figure 4.4: (a) Schematic of pHMT vector designed for the expression of MBP fusion human PON1 and its variants in E. coli. The MBP and test proteins were fused with a TEV protease site in between. (b) Schematic of pET11a vector designed for the expression of MBP fusion human PON1 and its variants with C-terminal His6 tag in E. coli.

huPON1 as a C-terminal fusion of MBP with a TEV protease cleavage site in the middle. A schematic of the designed vector is shown in Figure 4.4 a. We were able to express all these variants in large-scale using MBP as a fusion partner as. As seen in Figure 4.5 all these variants co-purified with several other bands on the gel. After several attempts to optimize the purification conditions and remove unwanted proteins, we were still unable to purify reasonable amount of homogeneous protein. We reasoned that the co-eluted proteins may be the products of the truncation or proteolytic fragments with the N-terminal hexa-histidine tag. To figure out whether the truncated bands were proteolytic products, we carried out a Western blot experiment using

131

HRP (Horse Radish Peroxidase) conjugated Goat-anti-His antibody (against the hexahistidine

tag) using ECL (enhanced chemiluminescence) substrate from Pierce Biotechnology. The

HisProbe experiment indeed confirmed that the co-eluted bands were fragments with an N-

terminal His6 tag (Figure 4.6). To avoid co-purification of these truncated fragments, we have re-

engineered all the vectors by moving the His6 tag from the N-terminus to the C-terminus of the fusion proteins. The MBP fused huPON1 and engineered variants with the C-terminal His6-tag

were cloned into the pET11a vector under the T7 promoter, as shown in Figure 4.4 b.

At the same time, our lab was also exploring the possibility of using different chaperone

sets to obtain more folded soluble huPON1 from E. coli. Among different chaperone sets we

tested, the dnaKJE7 (DnaK, DnaJ, and GrpE from Takara Bioscience) set was found to be more

effective and was able to express more folded active huPON1. Plasmids expressing wild type

huPON1 and its variants were co-transformed with dnaKJE7 vector into Origami B (DE3) cells

for large-scale expression. Expressed proteins were first purified by using an IMAC column (Ni-

NTA resin column). To obtain untagged protein, IMAC-purified protein was treated with TEV

protease and repurified in an Ni-NTA chromatography column to remove the cleaved MBP. The

elution from the second Ni-NTA column was purified by anion exchange chromatography to

obtain untagged purified protein. By combining these two approaches, we were able to express

active huPON1 and its variants in large-scale (~4 mg/L) from E. coli with reasonable purity, as shown in Figure 4.7. The yield of the protein we obtained is shown in Table 4.1. Rationally

designed variants were produced higher levels than wild type. It was observed that the N-terminal

deleted variants resulted in more soluble protein compared with the full-length wild type human

PON1 and its variants. These results are in consistent with the results obtained from fluorescence

screening of the frGFP fusion proteins. Crude cell lysate and purified proteins of all the variants

were found to be active against both paraoxon and phenyl acetate.

132

Figure 4.5: IMAC (Ni-NTA agarose resin) purified huPON1 variants expressed in E. coli from MBP- fusion pHMT vector. Lanes 1- MBP-delhuPON1, 2- MBP - huPON1H, 3 - MBP-delhuPON1H and 4 - MBP-huPON1G.

Figure 4.6: Western blot (right) of cell lysate and purified human PON1 variants from the MBP fusion vector in E. coli. Hexahistidine tag containing proteins were visualized by HRP-conjugated-Goat-antiHis antibody using chemiluminescent substrate from Pierce Biotechnology. Comassieblue-stained PAGE gel after transferring to PVDF membrane is on the left. Lanes 1-huPON1H cell lysate; 2, 3-huPON1H; 4- cleaved MBP; 5-huPON1G, and 6-delhuPON1.

Figure 4.7: SDS-PAGE gel of IMAC purified MBP fusion wild type human PON1 and it variants. MBP fusion proteins were co-expressed with DNA KJE7 (Chaperone set from Takara Bioscience) and reengineered pET11-MBP-PON1-His6 vector. Lanes 1- huPON1, 2- delhuPON1, 3- huPON1H, 4- delhuPON1H, 5- huPON1G and 6- delhuPON1G.

133

Figure 4.8: Thermal inactivation of huPON1 and the engineered variants. Protein samples were heated for 10 min at different temperatures ranging from 25 to 80 °C and their activity was determined against phenyl acetate hydrolysis. The residual hydrolysis activity at 20 °C was taken as 100 %.

134

The kinetics of substrate hydrolysis of paraoxon and phenyl acetate of wild type huPON1

and its variants and the N-terminal deletion mutant of all the variants were determined following

the protocol described by Yeung et al.375 A representative Michaelis-Menten curve for paraoxon

hydrolysis by purified huPON1 is shown in Figure 4.9. Kinetic data are summarized in Tables

4.2 and 4.3. Wild type human PON1 was found to be more active amongst all the variants we

tested. The N-terminal deletion variants were found to be less active than their full length

counterparts. To measure the thermal stability of wild type huPON1, huPON1H and huPON1G

Figure 4.9: Hydrolysis of paraoxon by purified huPON1. Paraoxon activities were assayed in 50 mM Tris- -1 -1 HCl, 10 mM CaCl2 pH 7.4. Formation of hydrolysis product, p-nitrophenolate (ε405 17,000 cm M ), was monitored at 405 nm using Agilent 8453 UV-Vis Spectrophotometer.

using CD failed to produce any useful data as proteins started precipitating very quickly with increased temperature. Instead, we applied the thermal inactivation and residual activity 135

experiment reported by Masson et al. in 2007.277 Purified proteins were heated at different temperatures for 10 min before their residual activity was determined against phenyl acetate. The

HuPON1H was less stable and lost activity at lower temperatures compared to the wild type huPON1. On the other hand, huPON1G showed slower thermal inactivation than the wild type huPON1. For wild type huPON1 and huPON1H, the N-terminal deletion mutants showed slower inactivation than their full length counterparts as shown in Figure 4.8. Deletion of the

unstructured hydrophobic leader sequence may have affected the cooperative unfolding of the

proteins and contributed to stability to some extent.

4.4 Discussion

For large-scale expression of recombinant proteins, a bacterial system is the most widely

used host. Expression and purification of human PON1 from a bacterial system has been a very

difficult challenge. Furlong and coworkers in late 2008 first reported the successful expression

and purification of huPON1 from E. coli.276 HuPON1 was expressed at low temperatures in

enriched media in a fermentor. After purification through five to seven diethylaminoethyl

(DEAE) and hydrophobic interaction (HIC) columns, they were able to isolate a small quantity of

purified protein. Though the specific activity of the purified huPON1 was comparable to huPON1

from serum, the final yield was only about 450 µg per liter of fermented culture. In a different

approach, Arslan et al. in 2008 described an improved method for the purification of huPON1

from serum in several hundred micrograms.376 They have used a two-step method of ammonium sulfate precipitation and a modified HIC column.

The large-scale expression of soluble protein and, the improvement of stability and catalytic activity are the foremost challenges for huPON1 to be used as a drug for OP pesticides and nerve agent detoxification. Here we describe a method of large-scale expression and purification of active human PON1 and its solubilized variants from a bacterial system. Rational

136

design or directed evolution approaches have been applied to achieve the efficient heterologous

expression of proteins with increasing solubility and stability in bacteria and yeast with limited

success. It is still unknown how the changes in individual amino acids contribute to stability.

There are a limited number studies available which give an idea about where and exactly what

kind of mutations need to be applied to increase the solubility of a protein. Despite many

successful efforts to understand the structural basis of protein stability and solubility, there is still

no universal strategy to stabilize any protein by a limited number of rationally designed

mutations. There are few examples of proteins that have been stabilized by inserting the point

mutations with cumulative stabilizing effects.316-319 The most notable ones are rational design of

subtilisin E by introducing nine point mutations to obtain a calcium-independent hyperstable

variant,320 re-engineering of phospholamban to a soluble pentameric helical bundle by replacing

its lipid-exposed hydrophobic residues with charged and polar residues,321 rationally engineering

of a water soluble analog of phospholamban by changing membrane-exposed positions to polar or

charged amino acids,322 and rationally engineering of a water soluble potassium channel KcsA by mutating the lipid-contacting side chains to more polar groups.323 In these designs, site-specific changes are made on the target enzyme based on the detailed information and knowledge about

the protein structure and function. In our effort, to mainly improve the solubility, we re-

engineered the huPON1 surface to be more polar by applying the polar or charged residue

mutations in putative HDL binding sites, or by polar or charged residue mutations from chimeric

G2E6 surface or by deleting the N-terminal hydrophobic signal sequence. The relative solubilities

of the engineered variants, wild type human PON1 and chimeric PON1 (G2E6) were tested using

foding reporter GFP as a reporter protein.

From solubility screen experiments, we were able to successfully demonstrate that the

engineered huPON1H and huPON1G had better solubility than the wild type huPON1. We also

noticed that the deletion of the N-terminal leader sequence of wild type and engineered variants 137

Table 4.1: Yield of wild type human PON1 and the engineered variants expressed and purified from E. coli. Yields are shown in mg of protein per liter of regular shake-flask culture. Protein concentration was determined using Bradford assay and lysozyme standard on SDS-PAGE gel.

huPON1 delhuPON1 huPON1H delhuPON1H huPON1G delhuPON1G ~3.0 ~3.0 ~4.0 ~5.0 ~4.0 ~4.0

Table 4.2: Kinetic measurements for huPON1 and engineered variants including N-terminal deletion mutants for the hydrolysis of phenyl acetate.

K k k /K Variants M cat cat M mM min-1 mM-1 min-1

huPON1 1.6 2870 1790 delhuPON1 1.9 802 411 huPON1H 0.60 123 183 delhuPON1H 1.4 115 80 huPON1G 2.0 2620 1310 delhuPON1G 1.8 509 270

Table 4.3: Kinetic measurements for huPON1 and engineered variant for the hydrolysis of paraoxon.

K k k /K variants M cat cat M mM min-1 mM-1 min-1 huPON1 1.0 3.2 3.2 huPON1G 0.61 2.5 4.1

138

improved the solubility in E. coli compared to their full length counterparts. The charged or polar

residue mutations on putative HDL binding sites had the most pronounced effect on solubility of

the protein. The ΔhuPON1H (HDL binding site mutations and N-terminal leader sequence

deletion) variant achieved the highest solubility and expresssbility as was evident from Figures

4.3 and 4.5. Interestingly, we have noticed that the 16 polar mutations from the G2E6 protein

surface to the wild type huPON1 made huPON1G as soluble as G2E6 itself. Activity assay results

demonstrated that all engineered variants still possessed turnover activity against paraoxon and

phenyl acetate. These results suggested that the surface mutations we introduced on huPON1

were contributing to the solubility of the protein. Although the expression level of all the variants

were relatively low as frGFP fusions.

To increase the expression level of all the variants including wild type human PON1, we

have used a higher expression pHMT vector311 with the MBP-fusion instead of the frGFP-fusion.

Using the MBP-fusion, we were able to achieve a very high level of expression for all the variants

(Figure 4.5). However, we were unable to purify huPON1 and its variants in a reasonably higher degree since all the variants were co-purified with either truncated or proteolytic proteins. To get around these problems, a His6-tag was used at the C-terminus of the PON1 variants so that only full length protein would be purified on a Ni-NTA resin column.

A number of studies reported that the use of chaperones improves the expression and solubility of some difficult-to-express protein in bacterial systems by preventing aggregation and repacking proteins to its native structure.377-379 In this study, all the variants including wild type

human PON1 were coexpressed with chaperone set DnaKJE7 (from Takara Bioscience) in

Origami B (DE3) cells. Using the chaperone coexpression of MBP-fused PON1 with the C-

terminal His6-tag, we were able to achieve high levels of expression and a higher degree of purification of wild type huPON1 and its variants. From regular shake-flask expression, we were able to obtain around 4-6 mg of purified wild type huPON1. As listed in Table 4.1, engineered 139

huPON1H and huPON1G were produced at higher levels than the wild type protein. The soluble

expression level was further improved by deleting the N-terminal hydrophobic leader sequence

from all these variants. The reported kcat/ KM and KM values of huPON1 purified from serum are

47 mM-1min-1 and 0.9 mM-1 against paraoxon, and 1×104 mM-1min-1 and 1.1 mM-1 against phenyl acetate.273 The catalytic activities we determined against both substrates (Table 4.2) were about an order of magnitude lower than the serum huPON1. This may be because the huPON1 purified from E. coli was not in its native environment (absence of glycosylation and HDL particles).

Several studies reported that PON1 is very stable when it is associated with HDL particles and

shows the highest catalytic activities against its substrates such as phenyl acetate and paraoxon.226,

361, 380

Though the mutations on huPON1G and huPON1H are mostly on the surface, these mutations might have affected the structure and catalytic activity of the engineered variants. The

N-terminal deletion mutants were also less active compared to their full length counterparts.

Although these surface mutations may not have anything to do with catalytic activity, these may have some effect on structure, dynamics and substrate recognition.

Though huPON1H has resulted in the most soluble protein, it is evident from thermal inactivation experiments that the mutations on the putative HDL binding sites made it thermally less stable. This variant showed faster thermal inactivation compared to the wild type huPON1.

On the other hand huPON1G showed slower inactivation on thermal denaturation compared to that of the wild type huPON1. In addition to the solubility, the surface polar mutations might have contributed to thermal stability of the protein. The N-terminal deletion mutants of wild type huPON1 and the engineered variants showed slower inactivation than their full length counterparts as shown in Figure 4.8. Though the N-terminal deletion of the leader sequence resulted in a loss of activity by an order of magnitude, it might have helped the gaining of stability to some extent for wild type huPON1 and the engineered variants by increasing the 140 solubility. The studies carried out by La Du et al. reported that a similar N-terminal signal sequence deleted huPON1 (expressed in HEK cells) had about 20-fold reduced turnover activity against phenyl acetate and paraoxon.226 Contrary to huPON1 and huPON1H, deletion of the signal sequence did not make any significant change on the thermal inactivation of huPON1G variant. Although the deletion of the signal sequence from huPON1G (containing surface polar mutations from G2E6) increased the solubility, it may not have any (or may have very little) contribution to stability of the protein, since G2E6 (chimeric PON1) is already a stable protein. A recent study by Tawfik et al. demonstrated the effect of the N-terminal leader sequence on the

PON1’s stability in presence of recombinant HDL particle, Tergitol and PC/FC

(phosphatidylcholine and free cholesterol).361 Their results showed that the truncated (1-20 residues) recombinant PON1 has lower stability than full length protein because the protein could not bind with the exogenous substrates without the N-terminal leader sequence. However, these studies have not addressed the effect of the signal sequence on the protein’s stability in absence of any exogenous substrates.

We have successfully optimized a bacterial system and have been able to express and purify wild type huPON1 and the engineered variants in high yields (about 4-10 mg/L of regular shake-culture). We believe that further optimization of this method and the engineering of huPON1 is possible to obtain an even higher yield. It will also be possible to apply directed evolution of huPON1 and the engineered variants and select against OP pesticides and nerve agents or their analogs for developing a huPON1 bioscavenger with a higher catalytic activity and stability. One of the major problems of studying huPON1 was availability of this protein in reasonable amounts. We hope that this new approach for bacterial expression system and the optimized purification process will make it easier to express active huPON1 variants in reasonable quantities and screen the active variants from large libraries in a high-throughput

141

manner. This will also make it easier to study the structure-activity relationship of human PON1

and to study its role in anti-atherogenicity, cardiovascular disease and diabetes.

4.5 Experimental section

4.5.1 Cloning huPON1 and frGFP fusion in pET11a vector

A frGFP (Waldo et al.) gene was generated in our lab from the GFPuv and the EGFP

gene by PCR amplification and overlap PCR (Chapter 2). The frGFP gene was PCR amplified with two primers containing the sequence encoding His6-tag and AatII site at the 5'-end, and

EcoRI site at the 3'-end. Wild type human PON1 containing pCDNA3.1 plasmid was obtained from David E. Lenz from US Army Millitary Rsearch Institute for Chemical Defense

(USAMRICD). HuPON1 gene was PCR amplified with a 5'-primer (AATAATTATCATAT

GGCTAAGCTGATTGCGC TCACCC) containing an NdeI site and with a 3'-primer (ATAAT

GAATTC GCCGCTG CTTCCGCTCT GAAAATACAGATTCTCACCGCC GGTACCGAGTTCGCAG

TAA AGAG CTTTGTGA AACAC) containing KpnI site, TEV protease (ENLYFQG) site, linker

(GSSG) and EcoRI site at the C-terminus. A fusion of huPON1-KpnI-TEV-Linker-EcoRI-frGFP

was cloned into pET11a vector between NdeI and AatII sites using a three-piece ligation. The

sequence of the fusion proteins were confirmed by DNA sequencing (GeneWiz Inc, New Jersey).

4.5.2 Rational engineering of huPON1 variants

To generate a more soluble and expressible variant of huPON1, twelve point mutations

(F24E, Y185E, F186Q, I187K, Y190K, L191Q, W194K, L198E, L200Q, W202K, M289Q,

F293E) on the putative HDL particle binding sites were introduced in wild type huPON1. Wild

type residues were mutated to polar or charged residues using E, Q and K. All of these mutations

were introduced at once by a 3-piece overlap PCR method. Three pairs of primers were used to

PCR amplify three fragments of a whole gene with all 12 mutations. The three fragments were

then subjected to reassembly and amplification using two terminal primers (5'-fw: ATAGATATAC

142

ATATGGCGAA GCTGATTGCA CTCACGCTCT TGGGGATGGG ACTGGC ACTC TTCAGGAACC

ACC, 3'-re: CTCACCGCCGGTACCGAGTTCGCAGTAAAGAGCTTTG) containing restriction enzyme sites NdeI and KpnI. The amplified huPON1H gene was cloned into pET11a-huPON1-

TEV-frGFP vector replacing huPON1 for huPON1H. The sequence of the cloned gene was

confirmed by DNA sequencing.

A second variant of huPON1 was engineered by introducing surface hydrophobic to polar

or charged residue mutations from chimeric recombinant PON1 (G2E6) to wild type huPON1.

Fifteen of such mutations (I5T, N19R, Q21K, L31H, N78D, N80D, S81K, P82S, L96S, G101E,

A137S, Q192K, Y197H, N265D and N309D) and an N166S compensatory mutation of Q192K

were introduced into huPON1 by total gene synthesis using the TBIO method.381 Thirty oligos of

60 bases long containing all 16 point mutations were designed using DNAWORKS

(http://helixweb.nih.gov/dnaworks). Reassembled full length gene (huPON1G) was amplified using two terminal primers (5'-fw: GTTTAACTTTAAGAAGGAGATATACATATG GCAAAGC

TGACCGC and 3'-re: TGAAAATACAG ATTCTCACCGCCGGTACCTAATTC ACAG) with NdeI

and KpnI cloning sites. The huPON1G gene was cloned into pET11a-huPON1-TEV-frGFP vector

by replacing huPON1 for huPON1G. The sequence of the cloned gene was confirmed by DNA

sequencing.

4.5.3 N-terminal deletion mutants

The N-terminal 16 hydrophobic residues (residues 4 to 17, LIALTLLGMGLALF) from

the leader sequence of huPON1 and engineered variants were deleted to obtain better solubility

and expressibility in E. coli. Deletion mutants were PCR amplified with 5'-primers

(huPON1/huPON1G: AATAATAAT CATATGGCA AAG AGG AACCAC CAGTCTTCT TAC, huPON1H: AATAATAAT CATATGGCGA AAAGGAAC CACCAGTCTT CAGAAC) and 3'-primers

(huPON1/huPON1H: AATAAT GAATTC GCCGCCGCTTCCG CTCTGAA AATACAGATTCTC and

143

huPON1G: AATAATAATGGTACCTAATTCACAGTAT AATGCT TTATGGAAAACCG) and cloned

into the pET11a-huPON1-TEV-frGFP vector between NdeI and KpnI sites. The sequence of the

cloned gene was confirmed by DNA sequencing.

4.5.4 Cloning into a pHMT vector

All three full-length and N-terminal deletion mutants were cloned into a high level

expression vector, pHMT (a generous gift from Professor Mark P. Foster Lab, The Ohio State

University) between NcoI and PstI sites as a C-terminal fusion of MBP (Maltose Binding

Protein). The vector was designed to have a His6-tag at the N-terminus of MBP, and a linker

(EFGSSRVD) and TEV protease site (ENLYFQG) between MBP and fusion protein. For the

convenience of cloning and to make the vector more useful, the SalI site in original vector was

replaced with an NcoI site and an 1400 bp stuffer fragment was inserted between the NcoI and

PstI sites.

4.5.5 Cloning in to a pET11a vector with the C-terminal hexahistidine tag

The MBP (without the N-terminal His6-tag) fusion of full length and N-terminal deletion

mutants of huPON1 were PCR amplified directly from the pHMT-TEV-PON1 vector with a 5'-

primer containing NdeI site, and a 3'-primer containing His6-tag and XhoI site. Fusion genes with the TEV protease site and linker in between and a His6-tag at the C-terminus of huPON1 variants

were cloned into pET11a vector between NdeI and XhoI sites.

4.5.6 Fusion protein expression and purification

The Origami B (DE3) E. coli cells (from Novagen) was transfected with frGFP fusion or

MBP fusion constructs and grown overnight to saturation. LB media (1 L) supplemented with

appropriate antibiotics and 1 mM CaCl2 was inoculated with 20 mL of saturated culture and

grown at 37 °C until OD600~ 0.80. Cells were induced with 0.1 mM Isopropyl β-D-

144

Thiogalactopyranoside (IPTG) and grown at 30 °C for four hours for overexpression of fusion

proteins. Cells were harvested by centrifugation and stored at -80 °C. All purification was carried

out at 4 °C unless otherwise stated. Cell pellets were resuspended in buffer A (50 mM Tris-HCl

pH 8.0, 50 mM NaCl, 1 mM CaCl2 and 10% glycerol) containing 1 mM DTT, extruded through a syringe needle (22G1 from B-D) and lysed by sonication. Tergitol (NP10, 0.1%) was added to the crude cell lysate and protein was recovered for 2 h at 4 °C with gentle shaking. After centrifugation the cleared lysate was mixed with Ni-NTA resin and incubated with gentle shaking for 4 h for binding. The resin slurry was washed with buffer A containing 40 mM imidazole and eluted with buffer A containing 150 mM imidazole. Using a PD-10 desalting column (GE

HealthCare) the buffer was exchanged to PON Assay buffer (50 mM Tris-HCl, 10 mM CaCl2, pH

7.4) containing 10% glycerol. The protein samples were dialyzed against assay buffer containing

50% glycerol and stored at -20 °C before kinetics measurements.

4.5.7 Chaperone coexpression

The chaperone plasmid DnaKJE7 (DnaK, DnaJ and GrpE from Takara Bioscience) construct and pET11a fusion protein constructs were cotransformed into Origami B (DE3) cells for the overproduction of fusion protein. LB media containing appropriate antibiotics, 1 mM

CaCl2 and 0.1% L(+)-Arabinose was inoculated with an overnight saturated culture and grown at

37 °C for chaperone expression until OD600 ~ 0.8. Cells were then induced with 0.1 mM IPTG and grown at 30 °C for 4 h before harvesting. Proteins were purified following the protocols described above.

4.5.8 TEV cleavage and purification of huPON1 and its variants

Proteins were purified on a Ni-NTA agarose column and exchanged to Assay Buffer containing 2 mM DTT using a PD-10 desalting column. Samples were then treated with TEV protease for 4-6 h at room temperature before subjecting them to the equilibrated Ni-NTA resin.

145

After 4 h of binding at 4 °C, the resin slurry was washed with buffer A containing 20 mM

imidazole and eluted with buffer A containing 150 mM imidazole. Protein samples were

exchanged to storage buffer (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 1 mM CaCl2) using PD-10

desalting column. Samples were subjected to anion exchange chromatography (Resource Q, from

Amersham, GE Healthcare) to separate any coeluted proteins. Pure human PON1 and its variants

were eluted using a gradient of 50% buffer B (A: 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 1 mM

CaCl2, B: 50 mM Tris-HCl pH 8.0, 1 M NaCl, 1 mM CaCl2) over 50 mL elution and collected on

fraction collector. Pooled fractions were tested for aryl esterase activity and dialyzed overnight to

storage buffer (50 mM Tris-HCl at pH 7.4, 1.0 mM CaCl2, 50 % glycerol) before kinetic measurements.

4.5.9 Enzyme kinetics

Kinetic parameters for the hydrolysis of arylesterase (phenyl acetate, from Sigma-

Aldrich) and paraoxon (diethyl p-nitrophenyl phosphate, from Sigma-Aldrich), were determined

Table 4.4: Volume of the reagents and buffer used for PON1 enzyme kinetics against paraoxon.

Vol. of Assay Vol. of Paraoxon Vol. of PON1 Total concentration Buffer, µL (0.13 M Stock), µL Enzyme, µL of Paraoxon, µL 189.6 0.4 10 0.26 188.8 1.2 10 0.78 188.0 2.0 10 1.30 187.0 3.0 10 1.90 186.4 3.6 10 2.34 186.0 4.0 10 2.60

Table 4.5: Volume of the reagents and buffer used for PON1 enzyme kinetics against phenyl acetate. Vol. of Assay Vol. of Phenyl acetate Vol. of PON1 Total concentration Buffer, µL (0.13 M Stock), µL Enzyme, µL of Phenyl acetate, µL 189.6 0.4 10 0.26 188.8 1.2 10 0.80 188.0 2.0 10 1.30 187.0 3.0 10 1.90 186.0 4.0 10 2.60 186.0 5.0 10 3.3 146

following a protocol described by Yeung et al.375 using the Assay buffer containing 50 mM Tris-

HCl, 10 mM CaCl2 at pH 7.4. Typically, paraoxon was used from 0.26 to 2.6 mM concentrations

and phenyl acetate was used from 0.26 to 3 mM concentration as shown in Tables 4.4 and 4.5. A

130 mM stock concentration of the substrates was prepared in MeOH. Initial rate of formation of

hydrolysis product was monitored in a quartz cuvette by following the absorbance at A405 for p-

nitrophenolate for paraoxon, and at A270 for phenol for phenyl acetate using Agilent 8453 UV-Vis

Spectrophotometer. The amount of hydrolysis product was determined from the extinction

-1 -1 -1 -1 coefficient (ε405 17,000 cm M for p-nitrophenoxide, and ε270 1,310 cm M for phenoxide).

Kinetic parameters (KM and kcat) were determined using the Michaelis-Menten equation.

4.5.10 Thermal inactivation and residual activity determination

The wild type human PON1 and the engineered variants were heated for 10 min at a

different ranges of temperatures from 20 to 80 °C. The samples were incubated on ice before their

residual hydrolysis activity was measured against phenyl acetate at 25 °C. Phenyl acetate (3.6

mM) was used in assay buffer (50 mM Tris-HCl, 10 mM CaCl2, pH 7.4) for activity

measurement. Formation of the hydrolysis product, p-nitrophenoxide, was monitored as above

and the residual activity was measured from the slope of the line from the rate of the p-

nitrophenoxide formation. The data sets were then normalized to 100 % for activity at 20 °C.

4.6 Acknowledgements

We are grateful to David E. Lenz, US Army MRICD for the pCDNA3.1 plasmid and

Professor Mark P. Foster, The Ohio State University, for the pHMT vector. This work was

supported by The National Institute of Health (NIH) Center of Excellence for Catalytic

Bioscavenger Medical Defense Research (NIH U54 NS058183, Center PI David Lenz, PIs on

two projects, Christopher M. Hadad and Thomas J. Magliery) and United States Army Medical

Research Institute for Chemical Defense (USAMRICD). 147

CHAPTER 5

Study of BRCA1 cancer predisposing mutations using split frGFP system and directed

evolution of huPON1

5.1 Cancer predisposing BRCA1 mutations: Study of the interactions of

BRCA1/BARD1 in vivo

BRCA1 forms a stable heterodimeric complex191 with a structurally homologous protein

BARD1 (BRCA1 Associated RING Domain 1), and the BRCA1/BARD1 complex has been implicated in a wide variety of key cellular processes, such as homologous recombination repair of DNA damage, chromatin remodeling, cell cycle check points and regulation, transcriptional regulation, regulation of centrosome duplication and apoptosis, etc.176-182 It is evident that the

BARD1 is required for the stability of the BRCA1 protein in vivo and for most of its diverse

cellular functions.192 Although BRCA1 has been reported to form complexes with a number of other proteins, its interactions with BARD1 are remarkable in terms of its stoichiometry and stability. Both of these proteins have striking structural similarity as shown in Figure 1.16.

BRCA1 interacts with BARD1 through their N-terminal RING domains which includes their

RING motifs and the antiparallel α-helices formed by the sequences encompassing the RING

motifs.191 These antiparallel α-helices (immediately followed by RING domains) form a stable four-helix bundle. Klevit et al. reported that the heterodimerization primarily takes place by extensive interactions between the BRCA1 and BARD1 N-terminal subunits within the hydrophobic core of the four-helix bundle.194 Side chains from residues L3, L6, V8, V11, V14,

148

I15, A17, M18, I21, L22, L82, L86, I89, A92 and L95 of BRCA1, and residues A33, W34, H36,

A40, L44, L47, L48, L101, M104, L107, L111, L114 and L115 of BARD1, are either partially or

mostly buried at the binding interface. These interactions may be enhanced by several other polar

interactions as it is observed that side chains of R7, E10, E85 and D96 in BRCA1 are close to the

side chains of D117, K110, R43 and H36 of BARD1, respectively. Several recent studies

revealed that the BRCA1/BARD1 heterodimer has an enzymatic role and acts as an E3 ubiquitin

ligase.169, 192, 197-199 Hashizume et al. demonstrated that individually the BRCA1 and BARD1 have very low ubiquitin ligase activities in vitro, whereas the BRCA1/BARD1 complex has dramatically higher ligase activity. Later this group, Klevit et al. and Solomon et al. revealed that the N-terminal BRCA1 (1-109) and BARD1 (26-140) only are capable of forming a functional

RING domains to form four-helix heteroduplex and maintaining its E3 ligase activity.165 169, 194,

200-202 Although the definitive substrates of BRCA1/BARD1 have not yet been identified,

autoubiquitination of the BRCA1 subunit itself is observed during in vitro and in vivo reactions

catalyzed by BRCA1/BARD1.203-205

Although the ubiquitin ligase activity of BRCA1/BARD1 complex is very important for

its role as a tumor suppressor, the mechanism by which this activity contributes to biological

function is still not clear. BRCA1’s implication in wide array of cellular processes indicates that

the BRCA1/BARD1 ligase may have a variety of substrate candidates. In response to DNA

damage, the BRCA/BARD1 complex was found to co-interact with RAD51 and PCNA

(proliferating cell nuclear antigen) in different nuclear foci and in DNA damaged and replicating

regions; with BASC (BRCA1 Associated genome-Surveillance Complex) which includes

RAD50-MRE11-NBS (functions as an exonuclease at the double-strand breaks) and ATM (ataxia

telangiectasia mutated, functions upstream of BRCA1 in the double-strand-break repair

pathway);206-208 with histone H2AX and other histones through UbcH5α-mediated

monoubiquitination; and with RNA polymerase II (POL2A) and proteins on RNA Pol II 149

complex185, 215 for DNA repair and chromatin remodeling.205, 210 In transcriptional control and

apoptosis the complex interacts with proapoptotic protein GADD45.212 The loss of BRCA1

function caused abnormal centrosome amplification, defective G2–M checkpoint control and

genetic instability,221 abnormal nuclear division, and cellular aneuploidy.174 Among the number of centrosome proteins, γ-tubulin which is an essential microtubule nucleation protein was found to be ubiquitylated by BRCA1/BARD1 using both K48 and K344 residues.221 Results from these

studies clearly suggest that the ubiquitin ligase activity of BRCA1/BARD1 complex is linked to

the cell-cycle checkpoint functions and tightly regulate the cell cycle during cell proliferation.

Figure 1.18 shows a schematic of the genome wide interaction pattern of BRCA1/BARD1 in

numerous biological functions.

It is already evident that the BRCA1/BARD1 interaction is required for several of the

cellular and tumor-suppressor functions of BRCA1. Although several reports have demonstrated

that the BRCA1-independent functions of BARD1 might have its own crucial role in apoptosis

and mitosis, few cancer associated mutations have been found in BARD1, compared with more

than 650 (Human Gene Mutation Database, Institute of Medical Genetics in Cardiff,

www.hgmd.cf.ac.uk) for BRCA1 gene. The ubiquitin E3 ligase activity of the BRCA1/BARD1

complex is responsible for most of the diverse biological functions attributed to BRCA1,

including its ability to suppress tissue specific tumor generation in normal cells. It is reported that

20 % of the clinically relevant mutations of BRCA1 occur within the N-terminal 100 residues,

which contain the RING motif.188 Several in vitro and in vivo studies reported that the missense

mutations in the RING domain, especially C61G and C64G, abolish the E3 ligase activity of

BRCA1/BARD1 complex.192, 194, 200

To date a total of 44 different nonpolymorphic missense mutations have been reported to date in cancer patients DNA encoding the N-terminal RING domain of BRCA1 [Breast Cancer

Information Core (BIC), National Human Genome Research Institute, National Institute of 150

Figure 5.1: Table (above) lists the known cancer-associated mutations in the N-terminal RING domain of BRCA1 (Breast Cancer Information Core, NIH). Mutations highlighted yellow were found in the BRCA1/BARD1 binding interface. Mutations highlighted cyan and red (Zn2+ ions binding sites) were found on the RING motif. b shows the solution NMR structures of RING domains of BRCA1 (1-109 aa, cyan) and BARD1 (26-140 aa, gray) with associated Zn2+ (spheres) ions (PDB:1BXL).194 Sticks represent the known cancer predisposing mutations on BRCA1 RING domain. c shows the BRCA1 sequence with conserved residues (in red) binding to the Zn2+ ions. Gray letters represent the reported known cancer predisposing mutations on N-terminal RING domain of BRCA1.

151

Health, http://research.nhgri.nih.gov/bic/], listed in Figure 5.1. These missense mutations

observed in cancer patients are divided into two classes. The first is Zn2+-ligating residues (C24,

C39, C44, C47, C61 and C64), mutations which clearly predispose individuals to breast and ovarian cancers. The mutations in C27 and H41 sites have not been identified in cancer patients.

Several in vitro and in vivo studies reported that the missense mutations in the RING domain, especially C61G and C64G, abolish the E3 ligase activity of the BRCA1/BARD1 complex.

Klevit et al.194, 200 extensively studied the mutations in Zn2+ binding sites and showed that the

mutations on site I (C24 and C44) cause the RING motif not to fold properly (Klevit et al.

unpublished results) and mutations on site II (C39, C61 and C64) cause the local perturbation of

the RING motif structure (structure of second Zn2+ binding loop) and decrease the affinity for the

second Zn2+ ion binding.201, 202 These mutations do not affect the binding interface with BARD1;

instead they have hypothesized that it is likely that these mutations directly affect the BRCA1

RING (E3)-E2 binding interface. The second class is the mutations that are found in the

BRCA1/BARD1 interface.

Using the yeast two-hybrid (Y2H) assay Solomon et al. reported a comprehensive study

on cancer predisposing missense mutations on BRCA1 and their effect on BARD1 binding or E2-

conjugating enzyme binding.202 Their data revealed that some of these mutations disrupt the binding of BRCA1 with BARD1 or E2-enzyme (for example, UbCh5α) and which in turn caused the loss of its E3-ubiquitin ligase activity. They generated a random library of BRCA1 mutants using mutagenic PCR and screened the mutants for the interaction with BARD1 or E2-enzyme

(UbCh5α) using the Y2H assay. These screens identified residues within BRCA1 that are required for interaction with each protein. Among the mutants selected in the positive screen, 22 missense mutations in 19 residues inhibit interactions with E2-enzyme, 8 missense mutants in 7 residues disrupt interaction with BARD1, and 19 of the 35 BRCA1 variants co-purified with

BARD1 showed reduced E3 ligase activity. All 7 mutations in Zn2+-ligand binding residues 152

Figure 5.2: Plasmid maps, pMRBAD-BRCA1-CfrGFP and pET11-NfrGFP-BARD1, constructed for co- expression and the in vivo interaction study of N-terminal RING domain of BARD1 with that of wild type BRCA1 and its mutants. Cancer-associated mutations of BRCA1 were introduced into the BRCA1 gene of the pMRBAD plasmid by either QuikChange or overlap PCR methods.

inhibited binding to E2 substrate. A summary of the screening results and the activity for E3- ligase are shown in Figures 1.19 and 1.20. These results are in consistent with the RING motif structural analysis and Zn2+-ligating mutation studies reported by Klevit et al. However, these results are in contrast with the hypothesis that a charged or polar residue mutation in the hydrophobic core may perturb the BRCA1/BARD1 interaction.200, 201 Also, since the Y2H assay is prone to false positives, further studies of the role of mutations on BRCA1 in the

BRCA1/BARD1 interactions are warranted. The BRCA1/BARD1 interaction studies carried out by Sarkar et al.56 in 2008 (Chapter 2) on a small subset of BRCA1/BARD1 interface mutants

using improved split folding reporter GFP reported that helix-helix interface mutants V11A and

M18K disrupt the interaction of BRCA1 and BARD1, and a mutation, L52F, on the RING motif,

away from the four-helix bundle also disrupts the interaction. Therefore, further investigation of

all cancer-predisposing mutations in the N-terminal RING domain of BRCA1 for their role in

BRCA1/BARD1 binding and E3 ligase activity is needed to validate the partial results obtained

by the Y2H assay. The analysis of the BRCA1/BARD1 interaction and the ubiquitin ligase

153

activity of RING-domain mutations are important not only to investigate the biological function

of BRCA1 but also to predict a person’s predisposition to cancer.

5.1.1 Study of BRCA1 cancer associated mutants for their interactions with BARD1

The split frGFP we re-engineered was able to detect the BRCA1/BARD1 N-terminal

RING domain interactions efficiently and evolve fluorescence in vivo.56 In our previous study, using this improved split GFP assay, we examined a small number of cancer-associated mutants of BRCA1, which had been examined before by either Y2H methods, or in vitro methods, or both. We selected five mutations which are mostly buried in the helical interface (V11A, I15T,

M18T, M18K and I21V) and one mutation in the RING motif away from the interface (L52F),

Figure 2.1. The V11A and M18K mutations showed a reduced fluorescence level compared to the wild type BRCA1/BARD1 Figure 2.6. The I15T, I21V, and M18T mutations did not markedly reduce the fluorescence, although the L52F mutation away from the interface did reduce the fluorescence level. These results show that the interface is fairly insensitive to mutation but that under packing and charge burial can be sufficiently deleterious to prevent binding.

In this study, we aimed to examine the whole spectrum of mutations on the N-terminal

RING domain of BRCA1 observed in cancer patients for their role in the BRCA1/BARD1 interactions using the same approach. The cancer-associated mutations that we tested were shown in Figure 5.1. Mutations were introduced into the BRCA1 N-terminal RING domain gene sequence of the pMRBAD-BRCA1-CfrGFP vector (Figure 5.2) by site directed mutagenesis using either QuikChange (Stratagene) or overlap PCR methods. All thirty six mutants were co- transformed with the pET11a-BARD1-NfrGFP plasmid (Figure 5.2) which was expressing the

wild-type BARD1. In addition to these, the pET11a-BARD1-NfrGFP plasmid was co-

transformed with wild type BRCA1 (positive control) and a plasmid containing linker,

154

Figure 5.3: Cellular fluorescence obtained from screening of BRCA1 mutants for their interaction with BARD1 using split-frGFP assay. BRCA1 mutants were cotrnasformed with BARD1 and cells were grown on plate at 30 C for 12 h and at room temperature for 12-16 h before taking the picture. Pictures were taken under UV-Illuminator using long wave UV lamp using a filter that passes blue light only to excite the fluorophore. (+) indicates wild type BRCA1 with BARD1and (-) indicates pET11-BARD1-NfrGFP cotransformed with CGFP fused with a linker (pMRBAD-Link-CfrGFP).

pMRBAD-linker-CfrGFP (negative control). Both fusion proteins were co-expressed and screened for cellular fluorescence on LB agar plates supplemented with 10 µM IPTG and 0.2%

L(+)-arabinose with required antibiotics. Cells were grown for 9-12 h at 30 ºC and 12-16 h at room temperature and observed for fluorescence under long-wave UV illumination. Pictures of cellular fluorescence were taken under UV illumination using a filter (VisiBlue converter) that passes blue light only for excitation of the cells. Any perturbation of the BRCA1/BARD1 interaction reduces the reassembly efficiency and thereby evolves reduced or no cellular fluorescence, which is shown in Figure 5.3. Bright fluorescence was observed for the positive 155

control and no fluorescence was observed for a non-cognate negative control. The cellular

fluorescence levels of mutants were compared with the fluorescence levels obtained from the

wild-type BRCA1/BARD1 and with the negative control, as shown in Figure 5.3. In addition to

V11A, M18K and L52F (observed as well in our previous studies) mutations, we observed that

Zn2+-binding site mutations C39R, C39Y, C61G perturb the interaction of BRCA1 and BARD1.

Mutations in the four-helix bundle interface, G98R and D96N, and in the RING motif, T37R,

K38N and R71G also showed reduced or no cellular fluorescence. The reduced levels of

fluorescence or no fluorescence we observed may have resulted from the perturbation of the

interactions due to the mutations. Any mutation on BRCA1 may affect the expression levels of

CfrGFP-BRCA1 fusion protein and thereby affecting the fluorescence level. To confirm the

expression levels of the BRCA1-mutant fusion protein, we performed a Western blot against an

HA tag (fused to the C-terminus of the C-terminal frGFP fragment) using HRP-conjugated Goat-

anti-HA-tag polyclonal antibody (from GenScript). Figures 2.7 a and b show the expression levels of the wild type BRCA1 and the mutants. These results suggested that the lowered cellular fluorescence levels we observed for V11A, M18K and L52F were not due to the lack of expression of the fusion proteins. To confirm the expression levels of the fusion proteins, we are working on to carried out a Western blot for other BRCA1 mutants (T37R, C39R, C39Y, C61G,

R71G, D96N and G98R) that showed the perturbation of interactions.

5.1.2 Future directions

It is evident that in cells BRCA1 is stabilized by BARD1 and the interaction is essential for BRCA1 for its regular cellular functions, such as DNA break repair, chromatin remodeling etc. Most of its cellular functions come from the E2 ubiquitin ligase activity of the

BRCA/BARD1 complexes through autoubiquitylation of itself or various other mostly unknown substrates. Though the clinical relevance of the most of the cancer predisposing BRCA1

156

mutations is not clearly known, it is believed that the loss or perturbation of the interactions and

consequently the loss of the E2 ligase activity that perturbs the regular events in the cells and

leads to uncontrolled cell proliferation. The restoration of the biological function of the

BRCA1/BARD1 complex in cells by restoring their interactions to its native state using small

molecule drugs has the potential to treat the cancer caused by BRCA1 mutations. We are working

on developing a system for screening of small molecule library or cyclic peptide library

(SICLOPPS, split intein-mediated circular ligation of peptides and proteins382) using the split

GFP screening method.

The BRCA1/BARD1 complex for the BRCA1 mutants which caused reduced or no fluorescence (V11A, M18K, T37R, K38N, C39R, C39Y, L52F, C61G, R71G, D96N and G98R) can also be studied and characterized by 1H,15N-HSQC NMR to confirm if there is any structural

perturbation of interaction due to the mutation.

5.2 Directed evolution of human paraoxonase-1 (huPON1)

Directed evolution has emerged as a powerful alternative approach for engineering

enzymes with new or improved functions such as activity, substrate specificity, stability, and

solubility. This approach has proven particularly advantageous in cases in which prior knowledge

of the protein’s structure and function was not available. Tawfik and coworkers generated

chimeric recombinant variants of PON1 (G2E6 and G3C9 are the most notable one) by family

shuffling using PON1 genes from rabbit, mouse, rat and human.273 Though they have successfully

isolated a number of PON1 variants with better biophysical characteristics, these variants failed to

show improved catalytic activity against OP pesticides and nerve agents other than paraoxon.

Because of their higher sequence homology with rabbit isoform (G2E6, 70% and G3C9, 93%),

these variants may have very high immunological response in human body. Therefore, these

variants are not considered to be as a therapeutic agent for OP detoxification. In this study we

157

wanted to apply the directed evolution approach to huPON1 only to obtain a variant with higher

expressibilty and solubility in E. coli. To screen for the soluble variants efficiently from a large

library, we have used a variant of GFP as a reporter protein for solubility following an

approach described by Waldo and coworkers in 1999.335

5.2.1 Folding reporter GFP (frGFP) as a reporter protein for solubility screen

An efficient high-throughput method is necessary for screening the large library. In 1999

Waldo and coworkers reported a variant of GFP (known as folding reporter GFP or frGFP) and

used it for the rapid screening of soluble proteins from a large library. They demonstrated that if

the upstream protein folds, frGFP can fold and catalyze the formation of the chromophore.335 But if the upstream protein is misfolded or aggregates, GFP fails to fold properly which in turn will cause little or no fluorescence (Figure 1.26). Waldo and coworkers have also examined the potential of frGFP as a reporter protein fusing with a panel of 20 different proteins from

Pyrobaculum aerophilum with variable solubility and demonstrated that the observed cellular fluorescence had a direct correlation with the solubility of the test proteins.335 They have used this reporter protein for the rapid screening of soluble protein variants from random libraries of mutant C33T of the gene V protein and of bullfrog H-subunit of ferritin to evolve these proteins that are normally prone to aggregation during expression in E. coli.335 Methyl transferase (MT),

tartrate dehydratase β-subunit (TD-β), and nucleoside diphosphate kinase (NDP-K) from the

hyperthermophilic Pyrobaculum aerophilum were also subjected to directed evolution and

screened for the variants with improved folding and solubility using folding reporter GFP as a

reporter protein.335 We constructed the same frGFP gene in our lab from the GFPuv and EGFP variants (as described in Chapter 2) and cloned it into a pET11a vector. The HuPON1 and

ΔhuPON1 libraries, generated by error-prone PCR, were cloned into the pET11-linker-frGFP vector (Figure 4.1) as an N-terminal fusion of frGFP for the screening of soluble folded proteins.

158

5.2.2 Construction of screening vector using frGFP as a reporter protein

A standard screening vector was constructed based on the pET11a vector for the cloning

of the wild type huPON1, error-prone PCR libraries of huPON1 and its variants, as a C-terminal

fusion of frGFP. The NdeI and AatII digested pET11a vector, the NdeI and EcoRI digested stuffer

fragment, and the EcoRI and AatII digested frGFP gene were ligated to create the cloning vector.

The 3'-end of the stuffer fragment was designed to have a unique KpnI site followed by a

sequence encoding the GSSG linker and the TEV protease site (ENLYFQG). A schematic of the

designed fusion construct is shown in Figure 4.1. To examine the correlation between observed

fluorescence of the frGFP and the solubility of the test proteins, the designed vector was tested

using several different proteins with a wide range of solubility. Cysteine free T4 lysozyme

(T4LTA) and yeast triosephosphate isomerase (yTIM) are known as highly soluble and well-

expressed proteins in E. coli, the chimeric G2E6 has an intermediate solubility and the huPON1

has very low solubility when expressed in E. coli. All these proteins were cloned into the frGFP

fusion vector. In addition to the wild type frGFP, these fusion proteins were expressed in E. coli at 30 °C for 6 h. Cells were normalized for the same density at OD600 before measuring the

whole-cell fluorescence. After excitation at 475 nm, cellular fluorescence was measured at 509

nm. We observed that the solubility of the proteins expressed in E. coli was directly correlated to

the fluorescence of the cells expressing the corresponding frGFP fusion (Figure 4.2). This

observation further validated the potential of frGFP as a reporter protein for the screening of

folded soluble proteins from a random library.

5.2.3 Directed evolution of human paraoxonase-1

The human paraoxonase-1 (huPON1) gene was amplified by error-prone PCR to

introduce random mutations. The PCR reaction was optimized to obtain random mutations at a

range of about 1-5 per gene. A random library was also generated for ΔhuPON1(a soluble variant

159

Figure 5.4: LB agar plates for the screening of error-prone libraries of the wild type huPON1 and NΔhuPON1. Both libraries were expressed as the C-terminal fusion of reporter protein frGFP. Cells were grown at 30 °C for 12-16 h and and at room temperature for 4-12 h.

0.5

0.4

0.3

0.2

Change in absorbance , AU in absorbanceChange , 0.1

0 0246810 time, min

Figure 5.5: Paraoxonase turnover activity measured for the variants isolated from the error-prone libraries of huPON1 and the N-terminal deletion mutant of huPON1 fused with frGFP. Fusion proteins were expressed at 30 °C and cells were lysed using glass beads. Cleared lysates (10 µL) were used in 200 µL reactions containing 2.3 mM of paraoxon. Paraoxonase turnover was measured for the formation of the hydrolysis product p-nitrophenoxide at 405 nm.

160 of huPON1) in which the N-terminal 14 hydrophobic residues (4-17) were deleted. The random libraries were subcloned into the redesigned pET11a vector as an N-terminal fusion of the frGFP.

Fusion protein libraries were expressed at both 30 and 37 °C overnight for screening on LB agar plates supplemented with ampicillin and 0.1 mM IPTG. Plates were incubated at room temperature for 4-6 h before observing under the UV lamp for any fluorescent colonies. Even after screening of almost 10 % of the library no fluorescent colonies were observed at 37 °C for the full length protein or for ΔhuPON1. Several fluorescent colonies with a dynamic range of brightness were observed at 30 °C with a hit rate of about 1 in about 10,000 colonies for

ΔhuPON1. For full length huPON1 the hit rate was very low. Figure 5.4 shows the agar plates that were screened at 30 °C and 37 °C. A total of 20 fluorescent colonies were isolated from the

ΔhuPON1 library and 4 fluorescent colonies were isolated from the huPON1 library.

Isolated variants have yet to be fully characterized. From the sequencing results of the eight variants, it was observed that the mutation rate was a little higher than expected (1-8 mutations per gene). All 24 variants were expressed separately and tested for their activity against paraoxon. After 3 h of expressing huPON1 at 30 °C cells were lysed with glass beads and the cleared lysates were used for paraoxonase assay following the protocol described in Chapter 4.

Though some of the isolated full length and the the N-terminal deleted variants of evolved huPON1 lost paraoxonase activity during evolution, most of the variants retained the paraoxonase activity comparable to wild type huPON1. Paraoxonase hydrolyzing activity of all the isolated variants is shown in Figure 5.5.

5.3 Materials and methods

5.3.1 Study of BRCA1 cancer associated mutations

5.3.1.1 Construction of BRCA1 mutants

An HA (YPYDVPDYAK) tag was introduced at the C-terminus of the CfrGFP gene in

161

the pMRBAD-BRCA1-CfrGFP vector using a 5' forward primer (AATAATAAT CCATG

GATTTATCTGCTCTCCG CGTTGAAGAAG) and a 3' reverse primer (AATAATAAT TGTACA

TTACTTAGCGTAATCTGGAACATCGTATGGGTAAGAGGAGCCACTCGAACCTTTGTAGAG

CTCATCC ATGCCATG) encoding the SGSS linker and an HA-tag. A total of 36 BRCA1 point

mutations (listed in Figure 5.1) were introduced in the BRCA1 gene in pMRBAD-BRCA1-

CfrGFP-HA (Figure 5.6) vector between the NcoI and AatII sites. Mutations in the BRCA1 gene

were introduced by site-directed mutagenesis method either by using overlap PCR or the

QuikChange (Stratagene) mutagenesis method. Mutations S4P, R7C, V11A, I15T, M18T, M18K

Figure 5.6: pMRBAD vector used for constructing cancer predisposing BRCA1 mutants. C-terminal of C- frGFP was tagged with HA (YPYDVPDYAK).

and I21V were introduced by overlap PCR method using mutant 5'-primers and a 3' primers within the NcoI and BclI restriction sites which are 117 bases apart from each other. Since the

BclI site is sensitive to methylation on bases C or G, the vector pMRBA-BRCA1-CfrGFP was

transformed into a dcm-/dam- strain for plasmid preparation. Two synthetic oligonucleotides with the required point mutations and with an 18 base-overlap were used to generate a gene sequence between the NcoI and BclI sites. For the L52F mutation, a modified overlap method was used to 162

Table 5.1: Primers used for creating the BRCA1 genes with point mutations using overlap PCR or QuikChange methods.

S4P-5'fw AATAATAAT CCATG GAT TTA CCT GCT CTCCGCGTTGAAGAAGTACAAAATGTC R7C-5'fw AATAATAAT CCATGGATTTATCTGCTCTCTGC GTTGAAGAAGTACAAAAT GTCATTAATG V11A-5'fw GCTGATT TATCTGCTCT ACGCGTTGAA GAAGCCCAAA ATGTCATTAA TGCTATGCAG V11A-3're CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGCATAGCA TTAATGACAT I15T-5'fw GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCACTAA TGCTATGCAG I15T-3're CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGCATAGCA TTAGTGACAT M18T-5'fw GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCATTAA TGCTACGCAG M18T-3're CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGCGTAGCA TTAATGACAT M18K-5'fw GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCATTAA TGCTAAACAG M18K-3're CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGTTTAGCA TTAATGACAT I21V-5'fw GCTGATT TATCTGCTCT ACGCGTTGAA GAAGTACAAA ATGTCATTAA TGCTATGCAG 3're BclI CCTTGATCAA CTCTAGACAG ATGGGACACT CTAAGATTTT CTGCATAGCA TTAATGACAT L52F-5'fw MluI CTAACC GGTTCCTT AGCTCG ACTCGGCACGCGTAACAAAAGTGTCTAT L52F-5're CTTTCTT CTGGTTGAAT AGTTTCAGCA TG L52F-3'fw CATGCTG AAACTATTCA ACCAGAAGAA AG L52F-3're BsrGI CATAGT CACACG TACGAC GCGAGA GC AGAATTCTTATGTACA TTATTTGTAGAGCTC BRCA1_NcoI AATAATAAT CCATGG CTGATT TATCTGCTCT aCGCGTTGAA GAAGTACAAA ATGTCATTAA 1fwWT S TGCTATGCAG BRCA1 NcoI AATAATtAT CCATGG CTGATT TATCTGCTCT aCGCGTTGAA GAAGTACAAA AT GTC ATT AAT fw WT L GCTATGCAG AAA aTC TTA G AGTGTCCC ATC TGT CTG GGTT CCTTGATCAACTCTAGACAGATGGGACGCTCTAAGATTTTCTGCATAGCATTAATGACAT BRCA1C24R re TTTG GGTT CCTTGATCAACTCTGGACAGATGGGACACTCTAAGATTTTCTGCATAGCATTAATGACAT BRCA1L28P re TTTG AGTTTCAGCATGCAAAATTTACAAAATATGTGGTCACACTTTGTGGAGAC I31M-3're AGGTTCCTTCATCAACTCCAGACAGATGGGACACTCTAAG AGTTTCAGCATGCAAAATTTACAAAATATGTGGTCACACTTCCTTGAGAC T37R-3're AGGTTCCTTGATCAACTCCAGACAGATGGGACACTCTAAG AGTTTCAGCATGCAAAATTTACAAAATATGTGGTCCCTCTTTGTGGAGACAGGTTCCTTGATCAAC C39R-3're TCCAGACAGATGGGACACTCTAAG AGTTTCAGCATGCAAAATTTACAAAATATGTGGTCATACTTTGTGGAGAC C39Y-3're AGGTTCCTTGATCAACTCCAGACAGATGGGACACTCTAAG AGTTTCAGCATGCAAAATTTACAAAATACGTGGTCACACTTTGTGGAGAC AGGTTCCTTGATC I42V-3're AACTCCAGACAGATGGGACACTCTAAG TGAGAAGTTTCAGCATGCAGAATTTAAAAAATATGTGGTCACACTTCGTG GATACAGGTTC C44F-3're CTTGATCAACTCCAGAC TGAGAAGTTTCAGCATGCAGAAGTTACAAAATATGTGGTCACACTTCGTG K45N-3're GATACAGGTTCCTTGATCAACTCCAGAC TGAGAAGTTTCAGCATGCCGAATTTACAAAATATGTGGTCACACTTCGTG C47G-3're GATACAGGTTCCTTGATCAACTCCAGAC K38N-5'fw GATCAAGG AACCTGTCTC CACCAAcTGT GACCACATAT TTTGCAAATT TTG K38N-3're CAAAATTTGCAAAATATGTGGTCACAGTTGGTGGAGACAGGTTCCTTGATC C61G-5'fw GGGCCTTCA CAGGGTCCTT TATGTAAGAA TG C61G-3're CATT CTTACATAA AGGACCCTG TGAAGGCCCT L63F-5'fw GAA AGGGCCTTCA CAGTGTCCgT TCTGTAAGAA TGATATAACC AAAAGGAGCC L63F-3're GGCTCCTTTTGGTTATATCATTCTTACAGAACGGACACTGTGAAGGCCCTTTC GCATGCTG AAACTTCTCA ACCAGAAaAA AGGGCCTTCA CAGTGTCCgT TAGGTAAGAA C64G-5'fw TGATATAACCAAAAGGAGCC TACAAGAAAG GCATGCTG AAACTTCTCA ACCAGAAaAA AGGGCCTTCA CAGTGTCCgT TAcGTAAGAA C64R-5'fw TGATATAACC AAAAGGAGCC TACAAGAAAG GCATGCTG AAACTTCTCA ACCAGAAaAA AGGGCCTTCA CAGTGTCCaT TATacAAGAA C64Y-5'fw cGATATAACC AAAAGGAGCC TACAAGAAAG D67Y-5'fw GGGCCTTCA CAGTGTCCaT TATGcAAGAAT TAcATcACC AAAAGGAGCC TACAAGAAAG TACG I68K-5'fw GGGCCTTCA CAGTGTCCgT TATGcAAGAAT gat AaaACC AAAAGGAGCC TACAAGAAAG TACG I68R-5'fw GGGCCTTCA CAGTGTCCgT TATGcAAGAAT gat AaaACC AAAAGGAGCC TACAAGAAAG TACG GCTCTTCAACAAGTTGACTAAATCTCGTACTTTCTTGTAGGCTCCCTTTG GTTATATCATTC R71G-3're TTACATAACGGAC

(continued..) 163

Table 5.1 Continued..

GCTCTTCAACAAGTTGACTAAA TCTCGTACTTTCTTGTAGGCTCCCTTTG GTTATATCATT S72R-5'fw CTTACATAACGGAC GCTCTTCAACAAGTTGACTAAATCTCGTACTTTCTTGTAGCCTCCTTTTG S72R-3're GTTATATCATTCTTACATAACGGAC CCTGTGTCAAGCTGAAAAGCACAAATGATCTTCAATAGCTCTTCAACAAG T77M-3're TTGACTAAATCTCATACTTTCTTGTAGGCTCCTTTTGG CCTGTGTCAAGCTGAAAAGCACAAATGATCTTCACTAGCTCTTCAACAAGTTGACTAAATCTCGTA L87V-3're CTTTCTTGTAGGCTCCTTTTGG CCAAAAGGAGCCTACAAGAAAGTAC GAGATTTAGTCAACTTGTTGAAGAG I89M-3're CTATTGAAGAtgATTT GTGCTTTTCAGCTTGACACAGG CCTGTGTCAAGCTGAAAAGCACAAATGGTCTTCAATAGCTCTTCAACAAG I89T-3're TTGACTAAATCTCGTACTTTCTTGTAGGCTCCTTTTGG CCTGTGTCAAGCTGAAAAGCACAAGTGATCTTCAATAGCTCTTCAACAAG I90T-3're TTGACTAAATCTCGTACTTTCTTGTAGGCTCCTTTTGG CATTT GTGCTT TTCAGCTTGA CACAgGTTTG GAGTATGCGA ACAGCTATAA TTTTGCAAAG D96N-5'fw AAGG GTAAAG GGACGTCGGG TGGA A CATTT GTGCTT TTCAGCTTGA CACACGTTTG GAGTATGCGA ACAGCTATAA TTTTGCAAAG G98R-5'fw AAGG GTAAAG GGACGTCGGG TGGA A BRCA1 SphI GGTTGAGAAGTTTCAGCATGCAAAATTTACAAAAT ATGTGGTCACACTTT GTGGAGACAGGTT WTre2L CCTTGAT CAACTCCAGACAGATGGGACACTCTAAG BRCA1 SphI GCATGCTG AAACTTCTCA ACCAGAAaAA AGGGCCTTCA CAGTGTCCgT TATGTAAGAA WTfw3L TGATATAACC AAAAGGAGCC TACAAGAAAG BRCA1 GCATGCTG AAACTTCTCA ACCAGAAaAA AGGGCCTTCA CAGTGTCCgT TATGTAAGAA SphIfwWTL3 S TGATATAACC BRCA1 AatII re CCGCTTCCACCCGACGTCCCTTTACCC BRCA1 AATAATAAT CCATGGATTTATCTGCTCTCCG CGTTGAAGAA G pMRBAD fw HAtagBRCA1 AATAATAAT TGTACA TTACTTAGCGTAATCTGGAACATCGTATGGGTA AGAgGAGCCaCTCGA pMRBAD re aCC TTT GTA GAG CTC ATC CAT GCCATG BRCA1 AatII TTCCACCCGACGTCCCTTTACCC WTre

insert the point mutation using MluI and BsrG1 sites for cloning. Mutations C24R, L28P, I31M,

T37R, C39R, C39Y, I42V, C44F, K45N and C47G were introduced by overlap PCR methods

using a 5'-primer and mutant 3'-primer within the NcoI and SphI sites. Mutations K38N, C61G and L63F were introduced by QuikChange PCR methods (Stratagene). Mutations C64G, C64R,

C64Y, D67Y, I68K, I68R, R71G, S72R, T77M, L87V, I89M, I89T, I90T, D96N AND G98R were introduced by synthesizing the whole gene using six synthetic oligonucleotides (5'WT-

NcoIfwL, 3'WT-SphIreL, 5'WT-SphIfwL). All inserted mutations were confirmed by DNA sequencing from GeneWiz, Inc. The oligonucleotides used for creating all BRCA1 mutants are listed in Table 5.1.

164

5.3.1.2 Screening

All the screening experiments were carried out following the protocols described by

Regan and coworkers.42, 43 Compatible pairs of plasmids (e.g., pET11a-BARD1-NfrGFP and pMRBAD-BRCA1-CfrGFP) were cotransformed into BL21(DE3) E. coli electrocompetent cells by electroporation. Cells were grown overnight to a saturation at 37 ºC in LB supplemented with

100 µg mL-1 ampicillin and 35 µg mL-1 kanamycin. Five to 10 µL of 1:1000 dilutions of saturated culture were plated on LB agar media supplemented with 20 µM IPTG, 0.2% arabinose and required antibiotics. Plates were incubated at 30 °C for 10-12 h and 12-16 h at room temperature

before taking pictures. In each case, green fluorescence was observed on a transilluminator (UVP

Inc.) using long wavelength (365 nm) UV irradiation. Pictures were taken using a digital camera

under UV-transillumination using a Visi Blue Converter Plate (P/N 38-0200-01 fromUVP Inc)

that passes blue light only to excite the chromophore for GFP fluorescence.

5.3.1.3 Affinity purification of fusion proteins and interacting partners

BL21 (DE3) cells were cotransformed with compatible plasmids (pET11-BARD1-NGFP

and pMRBAD-BRCA1-CGFP) and grown overnight to a saturated culture. LB broth (5 mL)

containing kanamycin and ampicillin was inoculated with 100 µL of saturated culture for each

sample and grown to an OD600 ~0.60 at 37 ºC. The saturated culture was diluted to1:1000 and 100

µL of which wasgrown on LB agar plates supplemented with 10 µM IPTG, 0.2% L(+)-arabinose,

100 µg mL-1 of ampicillin and 35 µg mL-1 of kanamycin. After growing for 14 h at 30 °C and 36

72 h at room temperature, the cells were collected from plates and resuspended in 1X phosphate

buffered saline (PBS). The OD600 of 100-fold diluted cells were measured to normalize the cell

densities. The cells were then harvested by centrifugation and the pelleted cells were resuspended

completely in 2.5 mL of lysis buffer (50 mM Tris-HCl, 100 mM NaCl, 100 µM ZnCl2, pH 8.0) containing 0.1% Tween 20, 5 mM β-mercaptoethanol, 10 mM imidazole, DNase (5-10 U),

165

RNase (10 µg/mL) PMSF (Phenylmethylsulphonyl fluoride from Pierce, 1 mM), and 0.05 mM

MgCl2. After sonication and centrifugation, the cleared lysate was collected, mixed with 100 µL of equilibrated (with lysis buffer) Ni-NTA Agarose Resin (Qiagen) and incubated at 4 °C for 2 h with gentle shaking for binding. The bound resins were loaded on to a disposable chromatography column, washed twice with 5 volumes of wash buffer (lysis buffer containing 20 mM imidazole). The purified proteins were eluted with 300 µL of elution buffer (lysis buffer containing 250 mM imidazole).

5.3.1.4 Western blotting using anti-HA-tag antibody

The split GFP fragments with the fused proteins or peptides were expressed following the similar procedure mentioned above. An appropriate amount of cell pellet was resuspended in 200

µL of lysis buffer (50 mM Tris-HCl, 200 mM NaCl, 100 µM ZnCl2, 0.1% Tween 20, 5 mM β- mercaptoethanol, 10 mM imidazole, pH 8.0), mixed with 100 µL of glass beads (Biospec,

Sigma), and vortexed vigorously to get the complete lysis. Following centrifugation for 30 min at maximum speed, the cleared lysate was collected and mixed with an equal volume of SDS gel loading buffer. Samples were electrophoresed on 12.5 % SDS-PAGE gels. Protein bands were transferred on to PVDF membrane (Pierce) using TE22 Mighty Small Transphor Unit (80-6205-

59) from Amersham Biosciences following the manufacturer’s protocol. PAGE gel was washed with transfer buffer (25 mM Tris-HCl, 192 mM Glycine, 0.1% SDS, 20 % methanol, pH 8.6) and blocked with TBST buffer (25 mM Tris-HCl, 150 mM NaCl, 0.05% Tween-20, pH 7.6) containing 1% BSA (Bovine Serum Albumine) for overnight at 4 °C. Following Pierce-OneStep

TMB Blot protocol, the membrane was treated with 40 mL of 1-4,000 dilution of anti-HA-Goat-

HRP in TBST buffer. Membrane was then treated with OneStep TMB substrate (from Pierce) following the supplier’s protocol. The Image of the blot was taken under White Light UV

Transillumination using GelLogic 100 imaging system from Kodak.

166

5.3.2 Directed evolution of huPON1

5.3.2.1 Plasmid construction: Cloning linker and frGFP fusion in pET11a vector

A frGFP335 gene was generated in our lab from the GFPuv and EGFP gene by PCR amplification and overlap PCR methods. The frGFP gene was PCR amplified with two primers containing a His6 tag and the AatII site at the C-terminus, and the EcoRI site at the N-terminus. A

stuffer fragment ( a random DNA sequence from pBR322 vector) was PCR amplified with 5'-

primer (AATAATTATCATATGGCT AAG CTGATTGCG) containing NdeI site; and 3'-primer

(ATAAT GAATTC GCCGCTG CTTCCG CTCTGAAAATACAGATTCTCACCGCC GGTACCGAGTT

CGCAGTAA AGAG CTTTGTGA AACAC) containing KpnI site, EcoRI site at the 3'-end and a

Table 5.2: Primers used for huPON1 and ΔhuPON1 cloning into the pET11a-PON1-frGFP vector.

delhuPON1 -5'fw AATAATAAT CATATG GCa AAG AGG AACCAC CAGTCTTCT TAC delhuPON1 -3're AATAAT GAATTC GCCGCCGCTTCCGCTCTGAAAATACAGATTCTCACCGCC GGTACCGAGTTCGCAGTAAAGAGCTTTGTGAAACAC huPON1 -5'fw AATAATTATCATATGGCTAAGCTGATTGCGCTCACCC huPON1 -3're AATAAT GAATTC GCCGCCGCTTCCGCTCTGAAAATACAGATTCTCACCGCC GGTACCGAGTTCGCAGTAAAGAGCTTTGTGAAACAC

sequence encoding the TEV (Tobacco Etch Virus) protease site (ENLYFQG) and the linker

(GSSG) in between. A fusion of NdeI-Linker-KpnI-TEV-EcoRI and EcoRI-frGFP-His6-AatII genes were cloned into pET11a vector between NdeI and AatII sites using three-piece ligation reaction. The sequences of the fusion protein genes were confirmed by DNA sequencing

(GeneWiz Inc, New Jersey).

5.3.2.2 Cloning T4LTA, yTim, huPON1 and G2E6 into the pET11-link-frGFP vector

Genes encoding for the T4LTA, yTim, huPON1 and G2E6 proteins were PCR amplified with 5'-primers containing NdeI site and 3'-primers containing KpnI site. The Vector and the PCR amplified genes were treated with NdeI and KpnI enzymes and purified from 1 % agarose gel using QIAquick Gel Extraction Kit (Qiagen) before setting up the ligation reactions. Sequences of 167

the genes with the frGFP fusion were confirmed by DNA sequencing (GeneWiz Inc, New

Jersey).

5.3.2.3 Cellular fluorescence measurement

The fusion proteins were expressed in BL21(DE3) cells for 4 h at 30 °C. Cells were

incubated at 4 °C for 4-6 h before harvesting by centrifugation. An empty vector and the frGFP

vector were also expressed in a similar fashion to use as a negative and positive control,

respectively. Cell pellets were resuspended in 1X PBS and normalized by adjusting the OD600 ~

0.1. Whole-cell fluorescence was measured in a Perkin Elmer LS-50B Fluoremeter using wavelengths 480 nm for excitation and 509 nm for emission maxima.

5.3.2.4 Random libraries of huPON1 and the N-terminal deletion mutant of huPON1

Error-prone PCR amplification of the N-terminal and the C-terminal frGFP were carried

357 out following the protocol described by Joyce et al. MgCl2 (7 mM), 0.5 mM MnCl2, 1 mM d(C/T)TP, 0.2 mM d(A/G)TP, 2-5 Units of Taq polymerase and 500 nM of each oligonucleotide were used for a 50 µL reaction to introduce random mutations. Primers listed in Table 5.2 were used for the amplification of huPON1 and ΔhuPON1libraries and were designed to have NdeI and

EcoRI sites for cloning into the pET11a-link-frGFP vector.

5.3.2.5 Library cloning into the pET11-link-frGFP vector and screening

Error-prone PCR libraries of huPON1 and N-delhuPON1 were digested with NdeI and

EcoRI restriction enzymes and purified from an agarose gel either by using QIAquick Gel

Extraction Kit from Qiagen or by a filter paper-membrane capture method (described in Chapter

6). The cloning vector was also treated with same enzymes and purified from an agarose gel

using similar approaches. For each library, approximately, 1 µg of vector DNA was used with

300 ng of error-prone library DNA in a 50 µL ligaton reaction. Following an overnight incubation

168 at 16 °C, the reaction was treated with BlpI restriction enzyme to reduce the background.

Reaction was purified using phenol-chloroform extraction and ethanol precipitation before transforming into BL21(DE3) electrocompetent cells. Cells were recovered in 500 mL of LB

(Lauria Bertani) media for an hour at 37 °C. The recovered cells (200 µL ) were grown on LB agar plates supplemented with 100 µg/mL ampicillin and 0.02 mM of isopropyl-β-D- thiogalactopyranoside (IPTG) for 12-16 h at 30 °C. Plates were incubated at room temperature for

6 to 24 h for fluorophore maturation and observed for fluorescent colonies under long wavelength

UV-Transillumination.

5.3.2.6 Paraoxon and phenyl acetate assay

The kinetic parameters for the hydrolysis of arylesterase (phenyl acetate, from Sigma-

Aldrich) and paraoxon (diethyl p-nitrophenyl phosphate, from Sigma-Aldrich), were determined following a modified protocol described by Yeung et al.375 Details of the protocols were described in Chapter 4.

169

CHAPTER 6

Supplemental materials and methods

6.1 General biochemical and molecular biology methods

6.1.1 General molecular biology reagents

Unless otherwise mentioned, all chemical or biochemical reagents were purchased mainly from Sigma-Aldrich, Fisher Scientific, American Bioanalytical Inc and Research Products

International. Molecular biology reagents such as DNA polymerases (Taq and Phusion polemerases), Restriction enzymes, T4 DNA ligase, electrophoretic molecular weight standards for protein and DNA were purchased from New England Biolabs Inc (NEB). Pfu Ultra II was purchased from Stratagen. Kanamycin, ampicillin, dithiothreiotol (DTT), IPTG (Isopropyl-β-D-

Thiogalactopyranoside) and L(+)-arabinose were purchased from Research Products

International. Deoxyribonucleotide mixtures (dNTPs) or individual deoxyribonucleotides (dNTP) were purchased from American Bioanalytical at 100 mM concentrations and mixed equimolar to produce 10 mM stocks of dNTPs. Stock solutions were usually made in 1000X (ampicillin at 100 mg mL-1 in ddwater, kanamycin at 35 mg mL-1 in ddwater, chloramphenicol at 30 mg mL-1 in ethanol, DTT (Dithiothreiotol) at 1 M in ddwater, IPTG at 500 mM in ddwater, arabinose at 20% in ddwater) and filter sterilized using 0.2 μm syringe filters from Millipore. Oligonucleotides for

PCR were purchased mainly from either Sigma Genosys (Sigma-Aldrich Life Science) or

Integrated DNA Technology (IDT) and resuspended typically at 200 μM stock with 1X TE (10 mM Tris-Cl, 1 mM EDTA, pH 8.0) buffer. Bacto Tryptone, Bacto Yeast extract and Bacto Agar

170

were purchased from BD (Becton Dickinson & Company). Disodium monohydrogen phosphate,

monosodium dihydrogen phosphate, Tris-Base and Tris-HCl, sodium chloride and calcium

chloride were purchased from Fisher Scientific. Phenol-chloroform-Isoamyl alcohol (25:24:1)

and Phenol-chloroform (24:1) reagents were purchased from USB Corporation. Tergitol (NP-10)

and Triton-X100 were purchased from Sigma-Aldrich. Water for all molecular biology

experiments was purified using a Barnstead NANOpure Diamond system to 18 MΩ•cm.

Solutions and buffers were typically filter sterilized with 0.22 µM Millex-GV PVDF syringe

driven filter unit (from Millipore) or autoclaved.

6.1.2 Plasmids and strains

The pET11-Z-NGFP and pMRBAD-Z-CGFP plasmids were obtained from Professor

Lynne Regan’s Lab, Yale University. pET11-BRCA1(1-109 aa), the pET28-BARD1 (26-140 aa)

vectors were generous gift from Professor Rachel Klevit’s Lab, University of Washington,

Seattle. The pSFFV-neo-Bcl-xL, pCDNA3.1-Bcl-2, pCMVSport6-XIAP plasmids and ABT-737

inhibitor were obtained from Professor Dustin Maly’s Lab, University of Washington, Seattle.

pCDN3.1-hDM2 vector was a generous gift from Professor Jiandong Chen, MOFFITT Cancer

Center and Research Institute, Tampa, Florida. High expression pHMT vector for MBP fusion

was a generous gift from Professor Mark Foster, The Ohio State University. The pRK793 plasmid

containing TEV (Tobacco Etch Virus) protease gene was purchased from ATCC (American Type

Culture Collection). The dam-/dcm- strain was a generous gift from Professor Michael Poirier’s

Lab, The Ohio State University. E. coli strains, BL21(DE3), DH10B, Origami B (DE3) cells were purchased from Stratagene.

6.1.3 Molecular biology kits

Typically plasmid DNA was isolated using QIAprep Spin Miniprep Kit from Qiagen following the supplier’s protocol. DNA fragments were purified from agarose gel (1-2 %) using 171

QIAquick Gel Extraction Kit from Qiagen. The PCR reactions and the ligation (if necessary)

reactions were purified using QIAquick PCR Purification Kit from Qiagen following supplier’s

protocols. Western blot experiments were carried out using OneStep Western Blot or ECL

western Blot Kit using either TMB substrate or ECL (enhanced chemiluminescence) substrate

form Pierce. Bio-Rad Protein Assay Kit was purchased from Bio-Rad Laboratories Inc.

6.2 Instrumentation

PCR was typically performed in a PTC-200 DNA Engine from MJ Research or

Flexigene Thermal Cycler from Techne Inc. Concentration of DNA solutions was typically done

in a Savant SC110 SpeedVac. Electrophoresis was carried out in a 10 cm horizontal gel

apparatus (for DNA) and 6.5 cm vertical gel apparatus (for protein) using either PowerPac Basic

or HC power supply from Bio-Rad. Electroporation was done with a MicroPulser Eectroporator

(Micropulser) from Bio-Rad. Sonication for cell lysis was carried out in a Misonix Sonicator

3000 (S3000) from Misonix Incorporated using 12.7 mm standard probe from the same company.

Centrifugation was carried out in an Eppendorf Centrifuge 5415 (for small volumes), Eppendorf

Centrifuge 5810R (for 5-50 mL volumes), or Sorvall RC-6 (for high speed and large volumes).

Enzymatic assays were carried out either by using Agilent 8453 UV-Vis Spectrophotometer in a

quartz cuvette or by using the SpectraMax M5 plate reader from Molecular Devices.

6.3 General protocols

Unless otherwise stated general molecular biology and biochemical methods including

PCR amplification, cloning and subcloning, protein expression and purification, etc. were

followed from Current Protocols in Molecular Biology (2002) by John Wiley and Sons, Inc; and

from Molecular Cloning: A Laboratory Manual by Joseph Sambrook & David Russel, 2001, 3rd

Edition, from Cold Spring Harbor Laboratory, New York.

172

6.3.1 PCR and error-prone PCR

PCR was typically performed in a PTC-200 DNA Engine (MJ Research) or a Flexigene

thermal cycler (Techne Inc.) with heated lid on. Most of the PCR amplification reactions were

carried out following standard protocol using either Taq DNA polymerase (from NEB), or Pfu

polymerase (Magliery Lab) or Phusion High-Fidelity polymerase (from NEB) in corresponding

1X buffer. A typical reaction volume was 50 μl (or 25 μl) with 200 μM each dNTP, 500 nM of

each primer, and approximately 10-100 ng of template and 1-5 units of polymerase. The

amplification reaction was typically 30 cycles with an initial denaturation step for 2 minutes at 95

°C for Taq or Pfu polymerase (98 °C for phusion polymerase) followed by thermal cycling at 95

°C for 30 seconds, 54 °C for 30 seconds, and 72 °C for approximately 10 – 120 seconds

(approximately 60 seconds per kb DNA for Taq, approximately 120 seconds per kb DNA for Pfu

polymerase and about 30 seconds per kb DNA for Phusion polymerase) with a final extension

step at 72 °C for 6 minutes. Annealing temperatures, extension time and template concentratios

were optimized for each PCR reaction to obtain best results. Reassembly and amplification

reactions were carried out following identical protocols for first 5-7 thermal cycles without

terminal amplification primers and additional 20-25 cycles with amplification primers.

Error-prone PCR reactions were carried out following the protocol described by Joyce et

353, 357 al. to introduce random mutations. Typical reactions contain 7 mM MgCl2, 0.5 mM MnCl2,

1mM d(C/T)TP, 0.2mM d(A/G)TP, 2.5 Units of Taq polymerase and 500 nM of each oligonucleotide in a 50 µL reaction. The amplification reaction was typically 30 cycles with an initial denaturation step for 2 minutes at 96 °C followed by thermal cycling at 95 °C for 30 seconds, 54 °C for 30 seconds, and 72 °C for approximately 60 seconds per kb DNA with a final extension step at 72 °C for 6 minutes.

All PCR reactions and DNA fragments were purified by either agarose gel electrophoresis (QIAquick Gel Extraction kit) or by phenol-chloroform extraction and ethanol 173 precipitation. Electrophoresis was carried out in a 10 cm horizontal gel apparatus in 1-2% agarose gels based on the size of the DNA products. Concentration of DNA solutions was typically done in a Savant SC110 Speed-Vac with low heating.

6.3.2 Phenol-chloroform extraction and ethanol precipitation

Reactions were mixed with equal volume of Phenol-Chloroform- Isoamylalcohol

(25:24:1), mixed well by vortexing and the mixtures were centrifuged down at maximum spedd for 30 seconds. The top aqueous layer was collected and mixed with Chloroform-Isoamylalcohol

(24:1), mixed well by vortexing and the mixtures were centrifuged down at maximum spedd for

30 seconds. The top aqueous layer was collected and mixed well with 1/10 th volume of 3 M sodium acetate at pH 5.2 and 5 volumes of 95% ethanol by vortexing. The reaction was incubated at -80 ºC for 1 h (to overnight) before centrifuging down and washing the DNA pellet with 75% ethanol. The pellet DNA was air dried and rehydrated typically in 20 – 50 µL of TE buffer (10 mM Tris-Cl, 1 mM EDTA, pH 8.0).

6.3.3 Enzymatic digestion and ligation

The enzymatic digestion reactions were typically done under standard conditions following supplier’s recommendations. Reactions were run typically at 30-50 μl volumes for two to six hours in 1X NEBuffer with approximately 10-20 units of each restriction enzyme for preparative scale. All digested plasmid or PCR reactions were purified on 1-2 % agarose gel

(based on the size) using the QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer’s protocol or by GF/F filter paper/membrane capture method following the protocol described below.

The ligation reactions for regular cloning were typically done in 10 μl volume in 1X T4 ligase buffer (NEB) with total DNA concentration of about 20-30 ng μL-1 typically at about 3:1 molar ratio of insert to vector DNA using 200-400 units of T4 DNA ligase (NEB). For the 174 ligation of error-prone PCR and DNA shuffling libraries, approximately 1:1 molar ratio of vector to insert DNA was maintained. All ligation reactions were typically carried out at 16 °C overnight or for 10 to 16 h. For library cloning, the ligation reactions were usually purified by ethanol precipitation or by using a QIAquick PCR Purification Kit (Qiagen) following the manufacturer’s protocol.

6.3.4 Purification of DNA fragments

The DNA fragments (restriction enzyme digested plasmid DNA or PCR amplified DNA) were routinely purified from agarose gel using either by QIAquick Gel Extraction Kit (Qiagen) following recommended protocols or by using membrane capture methods. In the later approach, the DNA fragments were first sufficiently resolved on 1-2 % agarose gel by electrophoresis.

Downstream to the DNA band, an incision was made in the gel parallel to the band using a scalpel or a razor blade. A sandwich of dialysis membrane and Whatman GF/F filter paper corresponding to the size of the DNA bands was placed into the cut with GF/F filter paper facing towards the DNA bands. The DNA fragments were electrophoresed through the GF/F filter paper on to the dialysis membrane. The membrane-filter paper sandwich was collected and placed in a

Ultrafree-MC Centrifuge Filter unit (UFC30HV00 from Fishe Scientific) with filter paper side down. The DNA fragments were eluted by centrifugation and by an additional washing with 100

µL TE buffer. The DNA fragments were further purified by phenol-chloroform extraction and ethanol precipitation as described above and finally rehydrated in 30-50 µL of TE buffer.

6.3.5 Electroporation and electrocompetent cell preparation

The plasmid DNA was routinely transformed into competent cells using MicroPulser

(MicroPulser from Bio-Rad) electroporator using 0.2 cm cuvettes from USA Scientific.

Electrocompetent cells were prepared following the standard protocol. Super competent DH10B

E. coli cells were used typically for library cloning and prepared to a competency of 107-109 cfu 175

per µg of supercoiled DNA. Electrocompetency was tested by electroporating known

concentration of pBR322 plasmid DNA (4361 bp from NEB) and counting cfu (colony forming

unit) on the plate. Typically, 40 µL of DH10B competent cells were electroporated with 1 ng of

pBR322 plasmid DNA, the cells were quenched in 1 mL of LB media and recovered for an hour

at 37 °C. Out of 1 mL recovered culture 1µL was spreaded on agar plate and grown over night at

37 °C. Total number of cfu from 1 µL of culture was counted from the agar plate and the

transformation efficiency was calculated using the following equation-

(𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑜𝑙𝑖𝑛𝑖𝑒𝑠 𝑜𝑛 𝑝𝑙𝑎𝑡𝑒)(𝑣𝑜𝑙𝑢𝑚𝑒 𝑖𝑛𝑞𝑢𝑒𝑛𝑐ℎ𝜇𝐿) TE[cfu μg]= (𝑣𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑞𝑢𝑒𝑛𝑐ℎ 𝑝𝑙𝑎𝑡𝑒𝑑𝑖𝑛𝜇𝐿)(𝜇𝑔 𝑜𝑓 𝑝𝑙𝑎𝑠𝑚𝑖𝑑)

6.3.6 Protein purification

The huPON1 variants were Co-erexpressed with chaperone set dnaKJE7 (Takara

Bioscience) in Origami B (DE3) E. coli cells typically in 1 L 2YT media. Cells were grown at 37

°C to an OD600 ~ 0.5 before induction with 0.1% L(+)-arabinose for the expression of chaperones

and to an OD600 ~ 0.8 before induction with 0.1 mM IPTG for the expression of huPON1 variants.

Protein expression was typically carried out at 30 °C for 4 h. The cells were harvested by

centrifugation and and stored at -80 °C before use. Cell pellet was resuspended completely in 30

mL lysis buffer (50 mM Tris-HCl, 100 mM NaCl, 1 mM CaCl2, pH 8.0) containing 10 mM imidazole, 2 mM DTT, and the cell suspension was extruded through a syringe using 22 G1 needle (from B-D) to aid cell wall breaking. The cell suspension was subjected to sonication for

30- second burst with 2 minutes interval (repeated 6 times) with power level setting at 7.0. A nonionic detergent Tergitol (NP-10, 0.10 %) was added to the cell lysate and mixed usually for 2 h at 4 °C on a Nutator with gentle shaking. The cleared cell lysate was collected after centrifugation at 32,000 g for 30 min at 4 °C, mixed with 1.5 mL Ni-NTA Agarose Resin 176

(Qiagen) and incubated for 2-4 h at 4 °C for binding. The bound agarose resin was loaded on to a

disposable chromatography column and washed with 80 mL of lysis buffer (containing 40 mM

imidazole). The protein was eluted with 10 mL of elution buffer (lysis buffer with 150 mM

imidazole). The eluted protein samples were desalted with PD10 column (from GE Healthcare)

and dialyzed overnight at 4 °C against activity buffer (50 mM Tris-HCl, 1 mM CaCl2, pH 8.0) containing 10% glycerol (or 50% glycerol for storage at extended period of time).

A modified protocol was used for the purification BRCA1/BARD1 complex. BL21

(DE3) cells were cotransformed with compatible plasmids (pET11-BARD1-NGFP and pMRBAD-BRCA1-CGFP) and grown overnight to a saturated culture. LB broth (5 mL) containing kanamycin and ampicillin was inoculated with 100 µL of saturated culture for each sample and grown to an OD600 ~0.60 at 37 ºC. The saturated culture was diluted to1:1000 and 100

µL of which wasgrown on LB agar plates supplemented with 10 µM IPTG, 0.2% L(+)-arabinose,

100 µg mL-1 of ampicillin and 35 µg mL-1 of kanamycin. After growing for 14 h at 30 °C and 36

72 h at room temperature, the cells were collected from plates and resuspended in 1X phosphate

buffered saline (PBS). The OD600 of 100-fold diluted cells were measured to normalize the cell

densities. The cells were then harvested by centrifugation and the pelleted cells were resuspended

completely in 2.5 mL of lysis buffer (50 mM Tris-HCl, 100 mM NaCl, 100 µM ZnCl2, pH 8.0) containing 0.1% Tween 20, 5 mM β-mercaptoethanol, 10 mM imidazole, DNase (5-10 U),

RNase (10 µg/mL) PMSF (Phenylmethylsulphonyl fluoride from Pierce, 1 mM), and 0.05 mM

MgCl2. After sonication and centrifugation, the cleared lysate was collected, mixed with 100 µL

of equilibrated (with lysis buffer) Ni-NTA Agarose Resin (Qiagen) and incubated at 4 °C for 2 h

with gentle shaking for binding. The bound resins were loaded on to a disposable

chromatography column, washed twice with 5 volumes of wash buffer (lysis buffer containing 20

mM imidazole). The purified proteins were eluted with 300 µL of elution buffer (lysis buffer

containing 250 mM imidazole). 177

References

1. Phizicky, E.M. & Fields, S. Protein-protein interactions: methods for detection and analysis. Microbiol Rev 59, 94-123 (1995).

2. Berggard, T., Linse, S. & James, P. Methods for the detection and analysis of protein-protein interactions. Proteomics 7, 2833-2842 (2007).

3. Pedrazzi, G. & Stagljar, I. Protein-protein interactions. Methods in molecular biology (Clifton, N.J 241, 269-283 (2004).

4. Stumpf, M.P. et al. Estimating the size of the human interactome. Proc Natl Acad Sci U S A 105, 6959-6964 (2008).

5. Lander, E.S. et al. Initial sequencing and analysis of the human genome. Nature 409, 860-921 (2001).

6. Zhu, H., Bilgin, M. & Snyder, M. Proteomics. Annu Rev Biochem 72, 783-812 (2003).

7. Piehler, J. New methodologies for measuring protein interactions in vivo and in vitro. Current Opinion in Structural Biology 15, 4-14 (2005).

8. Chien, C.T., Bartel, P.L., Sternglanz, R. & Fields, S. The two-hybrid system: a method to identify and clone genes for proteins that interact with a protein of interest. Proc Natl Acad Sci U S A 88, 9578-9582 (1991).

9. Fields, S. & Song, O. A novel genetic system to detect protein-protein interactions. Nature 340, 245-246 (1989).

10. Uetz, P. et al. A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 403, 623-627 (2000).

11. Uetz, P. & Hughes, R.E. Systematic and large-scale two-hybrid screens. Current opinion in microbiology 3, 303-308 (2000).

12. Lee, J.W. & Lee, S.K. Mammalian two-hybrid assay for detecting protein-protein interactions in vivo. Methods in molecular biology (Clifton, N.J 261, 327-336 (2004).

13. Karimova, G., Pidoux, J., Ullmann, A. & Ladant, D. A bacterial two-hybrid system based on a reconstituted signal transduction pathway. Proc Natl Acad Sci U S A 95, 5752-5756 (1998).

14. Joung, J.K., Ramm, E.I. & Pabo, C.O. A bacterial two-hybrid selection system for studying protein-DNA and protein-protein interactions. Proc Natl Acad Sci U S A 97, 7382-7387 (2000).

178

15. Hays, L.B., Chen, Y.S. & Hu, J.C. Two-hybrid system for characterization of protein-protein interactions in E. coli. BioTechniques 29, 288-290, 292, 294 passim (2000).

16. Aronheim, A. Protein recruitment systems for the analysis of protein-protein interactions. Biochemical pharmacology 60, 1009-1013 (2000).

17. Broder, Y.C., Katz, S. & Aronheim, A. The ras recruitment system, a novel approach to the study of protein-protein interactions. Curr Biol 8, 1121-1124 (1998).

18. Stagljar, I., Korostensky, C., Johnsson, N. & te Heesen, S. A genetic system based on split- ubiquitin for the analysis of interactions between membrane proteins in vivo. Proc Natl Acad Sci U S A 95, 5187-5192 (1998).

19. Deane, C.M., Salwinski, L., Xenarios, I. & Eisenberg, D. Protein interactions: two methods for assessment of the reliability of high throughput observations. Mol Cell Proteomics 1, 349-356 (2002).

20. Anfinsen, C.B. Principles that govern the folding of protein chains. Science (New York, N.Y 181, 223-230 (1973).

21. Richards, F.M. On the Enzymic Activity of Subtilisin-Modified Ribonuclease. Proc Natl Acad Sci U S A 44, 162-166 (1958).

22. Ullmann, A., Jacob, F. & Monod, J. Characterization by in vitro complementation of a peptide corresponding to an operator-proximal segment of the beta-galactosidase structural gene of . Journal of molecular biology 24, 339-343 (1967).

23. Johnsson, N. & Varshavsky, A. Split ubiquitin as a sensor of protein interactions in vivo. Proceedings of the National Academy of Sciences of the United States of America 91, 10340- 10344 (1994).

24. Thaminy, S., Miller, J. & Stagljar, I. The split-ubiquitin membrane-based yeast two-hybrid system. Methods in molecular biology (Clifton, N.J 261, 297-312 (2004).

25. Miller, J. & Stagljar, I. Using the yeast two-hybrid system to identify interacting proteins. Methods in molecular biology (Clifton, N.J 261, 247-262 (2004).

26. Fetchko, M. & Stagljar, I. Application of the split-ubiquitin membrane yeast two-hybrid system to investigate membrane protein interactions. Methods (San Diego, Calif 32, 349-362 (2004).

27. Pelletier, J.N., Campbell-Valois, F.X. & Michnick, S.W. Oligomerization domain-directed reassembly of active dihydrofolate reductase from rationally designed fragments. Proceedings of the National Academy of Sciences of the United States of America 95, 12141-12146 (1998).

28. Wehrman, T., Kleaveland, B., Her, J.H., Balint, R.F. & Blau, H.M. Protein-protein interactions monitored in mammalian cells via complementation of beta -lactamase enzyme fragments. Proceedings of the National Academy of Sciences of the United States of America 99, 3469-3474 (2002).

29. Galarneau, A., Primeau, M., Trudeau, L.E. & Michnick, S.W. beta-Lactamase protein fragment complementation assays as in vivo and in vitro sensors of protein-protein interactions. Nature Biotechnology 20, 619-622 (2002).

179

30. Rossi, F., Charlton, C.A. & Blau, H.M. Monitoring protein-protein interactions in intact eukaryotic cells by beta -galactosidase complementation. Proceedings of the National Academy of Sciences of the United States of America 94, 8405-8410 (1997).

31. Ozawa, T., Kaihara, A., Sato, M., Tachihara, K. & Umezawa, Y. Split luciferase as an optical probe for detecting protein-protein interactions in mammalian cells based on protein splicing. Analytical chemistry 73, 2516-2521 (2001).

32. Kim, S.B., Sato, M. & Tao, H. Split Gaussia luciferase-based bioluminescence template for tracing protein dynamics in living cells. Analytical chemistry 81, 67-74 (2009).

33. Villalobos, V., Naik, S. & Piwnica-Worms, D. Detection of protein-protein interactions in live cells and animals with split firefly luciferase protein fragment complementation. Methods in molecular biology (Clifton, N.J 439, 339-352 (2008).

34. Kaihara, A., Kawai, Y., Sato, M., Ozawa, T. & Umezawa, Y. Locating a protein-protein interaction in living cells via split Renilla luciferase complementation. Analytical chemistry 75, 4176-4181 (2003).

35. Paulmurugan, R. & Gambhir, S.S. Monitoring protein-protein interactions using split synthetic renilla luciferase protein-fragment-assisted complementation. Analytical chemistry 75, 1584-1589 (2003).

36. Chen, H. et al. Firefly luciferase complementation imaging assay for protein-protein interactions in plants. Plant physiology 146, 368-376 (2008).

37. Indraneel Ghosh, A.D.H., and Lynne Regan Antiparallel Leucine Zipper-Directed Protein Reassembly: Application to the Green Fluorescent Protein. Journal of American Chemical Society. 122, 5658-5659 (2000).

38. Abedi, M.R., Caponigro, G. & Kamb, A. Green fluorescent protein as a scaffold for intracellular presentation of peptides. Nucleic acids research 26, 623-630 (1998).

39. Monera, O.D., Zhou, N.E., Kay, C.M. & Hodges, R.S. Comparison of antiparallel and parallel two-stranded alpha-helical coiled-coils. Design, synthesis, and characterization. The Journal of biological chemistry 268, 19218-19227 (1993).

40. O'Shea, E.K., Lumb, K.J. & Kim, P.S. Peptide 'Velcro': design of a heterodimeric coiled coil. Curr Biol 3, 658-667 (1993).

41. Harbury, P.B., Kim, P.S. & Alber, T. Crystal structure of an isoleucine-zipper trimer. Nature 371, 80-83 (1994).

42. Magliery, T.J. et al. Detecting protein-protein interactions with a green fluorescent protein fragment reassembly trap: scope and mechanism. Journal of the American Chemical Society 127, 146-157 (2005).

43. Wilson, C.G., Magliery, T.J. & Regan, L. Detecting protein-protein interactions with GFP- fragment reassembly. Nature methods 1, 255-262 (2004).

44. Hu, C.D., Chinenov, Y. & Kerppola, T.K. Visualization of interactions among bZIP and Rel family proteins in living cells using bimolecular fluorescence complementation. Molecular cell 9, 789-798 (2002). 180

45. Hu, C.D. & Kerppola, T.K. Simultaneous visualization of multiple protein interactions in living cells using multicolor fluorescence complementation analysis. Nat Biotechnol 21, 539-545 (2003).

46. Zhang, S., Ma, C. & Chalfie, M. Combinatorial marking of cells and organelles with reconstituted fluorescent proteins. Cell 119, 137-144 (2004).

47. Nagai, T. et al. A variant of yellow fluorescent protein with fast and efficient maturation for cell- biological applications. Nat Biotechnol 20, 87-90 (2002).

48. Rizzo, M.A., Springer, G.H., Granada, B. & Piston, D.W. An improved cyan fluorescent protein variant useful for FRET. Nat Biotechnol 22, 445-449 (2004).

49. Shyu, Y.J., Liu, H., Deng, X. & Hu, C.D. Identification of new fluorescent protein fragments for bimolecular fluorescence complementation analysis under physiological conditions. BioTechniques 40, 61-66 (2006).

50. Jach, G., Pesch, M., Richter, K., Frings, S. & Uhrig, J.F. An improved mRFP1 adds red to bimolecular fluorescence complementation. Nature methods 3, 597-600 (2006).

51. Fan, J.Y. et al. Split mCherry as a new red bimolecular fluorescence complementation system for visualizing protein-protein interactions in living cells. Biochemical and biophysical research communications 367, 47-53 (2008).

52. Ozawa, T., Nogami, S., Sato, M., Ohya, Y. & Umezawa, Y. A fluorescent indicator for detecting protein-protein interactions in vivo based on protein splicing. Analytical chemistry 72, 5151-5157 (2000).

53. Ozawa, T., Sako, Y., Sato, M., Kitamura, T. & Umezawa, Y. A genetic approach to identifying mitochondrial proteins. Nature biotechnology 21, 287-293 (2003).

54. Cabantous, S. & Waldo, G.S. In vivo and in vitro protein solubility assays using split GFP. Nature methods 3, 845-854 (2006).

55. Nagai, T., Yamada, S., Tominaga, T., Ichikawa, M. & Miyawaki, A. Expanded dynamic range of fluorescent indicators for Ca(2+) by circularly permuted yellow fluorescent proteins. Proceedings of the National Academy of Sciences of the United States of America 101, 10554-10559 (2004).

56. Sarkar, M. & Magliery, T.J. Re-engineering a split-GFP reassembly screen to examine RING- domain interactions between BARD1 and BRCA1 mutants observed in cancer patients. Molecular bioSystems 4, 599-605 (2008).

57. Barnard, E., McFerran, N.V., Trudgett, A., Nelson, J. & Timson, D.J. Detection and localisation of protein-protein interactions in Saccharomyces cerevisiae using a split-GFP method. Fungal Genet Biol 45, 597-604 (2008).

58. Grinberg, A.V., Hu, C.D. & Kerppola, T.K. Visualization of Myc/Max/Mad family dimers and the competition for dimerization in living cells. Molecular and cellular biology 24, 4294-4308 (2004).

59. Fang, D. & Kerppola, T.K. Ubiquitin-mediated fluorescence complementation reveals that Jun ubiquitinated by Itch/AIP4 is localized to lysosomes. Proc Natl Acad Sci U S A 101, 14782-14787 (2004).

181

60. Giese, B. et al. Dimerization of the cytokine receptors gp130 and LIFR analysed in single cells. Journal of cell science 118, 5129-5140 (2005).

61. Niu, T.K., Pfeifer, A.C., Lippincott-Schwartz, J. & Jackson, C.L. Dynamics of GBF1, a Brefeldin A-sensitive Arf1 exchange factor at the Golgi. Molecular biology of the cell 16, 1213-1222 (2005).

62. Ozalp, C., Szczesna-Skorupa, E. & Kemper, B. Bimolecular fluorescence complementation analysis of cytochrome p450 2c2, 2e1, and NADPH-cytochrome p450 reductase molecular interactions in living cells. Drug metabolism and disposition: the biological fate of chemicals 33, 1382-1390 (2005).

63. Takahashi, Y. et al. Loss of Bif-1 suppresses Bax/Bak conformational change and mitochondrial apoptosis. Molecular and cellular biology 25, 9369-9382 (2005).

64. Hiatt, S.M., Shyu, Y.J., Duren, H.M. & Hu, C.D. Bimolecular fluorescence complementation (BiFC) analysis of protein interactions in Caenorhabditis elegans. Methods (San Diego, Calif 45, 185-191 (2008).

65. Shyu, Y.J., Suarez, C.D. & Hu, C.D. Visualization of AP-1 NF-kappaB ternary complexes in living cells by using a BiFC-based FRET. Proc Natl Acad Sci U S A 105, 151-156 (2008).

66. Citovsky, V. et al. Subcellular localization of interacting proteins by bimolecular fluorescence complementation in planta. Journal of molecular biology 362, 1120-1131 (2006).

67. Tzfira, D.W.a.T. Imaging protein–protein interactions in plant cells by bimolecular fluorescence complementation assay. Trends in Plant Science 14, 58-63 (2008).

68. Walter, M. et al. Visualization of protein interactions in living plant cells using bimolecular fluorescence complementation. Plant J 40, 428-438 (2004).

69. Waadt, R. et al. Multicolor bimolecular fluorescence complementation reveals simultaneous formation of alternative CBL/CIPK complexes in planta. Plant J 56, 505-516 (2008).

70. Marrocco, K. et al. Functional analysis of EID1, an F-box protein involved in phytochrome A- dependent light signal transduction. Plant J 45, 423-438 (2006).

71. Chen SB, T.L., Zeng LR, Vega-Sanchez ME, Umemura K,Wang GL. A highly efficient transient protoplast system for analyzing defence gene expression and proteinprotein interactions in rice. Molecular Plant Pathology 7, 417-427 (2006).

72. Galperin, E., Verkhusha, V.V. & Sorkin, A. Three-chromophore FRET microscopy to analyze multiprotein interactions in living cells. Nature methods 1, 209-217 (2004).

73. Remy, I. & Michnick, S.W. A cDNA library functional screening strategy based on fluorescent protein complementation assays to identify novel components of signaling pathways. Methods (San Diego, Calif 32, 381-388 (2004).

74. Michnick, S.W., Ear, P.H., Manderson, E.N., Remy, I. & Stefan, E. Universal strategies in research and drug discovery based on protein-fragment complementation assays. Nature reviews 6, 569-582 (2007).

182

75. Heim, R., Prasher, D.C. & Tsien, R.Y. Wavelength mutations and posttranslational autoxidation of green fluorescent protein. Proc Natl Acad Sci U S A 91, 12501-12504 (1994).

76. Tsien, R.Y. The green fluorescent protein. Annual review of biochemistry 67, 509-544 (1998).

77. Kerppola, T.K. Bimolecular fluorescence complementation (BiFC) analysis as a probe of protein interactions in living cells. Annual review of biophysics 37, 465-487 (2008).

78. Y. John Shyu, C.-D.H. Fluorescence complementation: an emerging tool for biological research. Trends in Biotechnology 26, 622-630 (2008).

79. Arkin, M.R. & Wells, J.A. Small-molecule inhibitors of protein-protein interactions: progressing towards the dream. Nature reviews 3, 301-317 (2004).

80. Murray, J.K. & Gellman, S.H. Targeting protein-protein interactions: lessons from p53/MDM2. Biopolymers 88, 657-686 (2007).

81. Fesik, S.W. Promoting apoptosis as a strategy for cancer drug discovery. Nat Rev Cancer 5, 876- 885 (2005).

82. Hunter, A.M., LaCasse, E.C. & Korneluk, R.G. The inhibitors of apoptosis (IAPs) as cancer targets. Apoptosis 12, 1543-1568 (2007).

83. Thompson, C.B. Apoptosis in the pathogenesis and treatment of disease. Science (New York, N.Y 267, 1456-1462 (1995).

84. Lee, E.F. et al. Crystal structure of ABT-737 complexed with Bcl-xL: implications for selectivity of antagonists of the Bcl-2 family. Cell death and differentiation 14, 1711-1713 (2007).

85. Adams, J.M. & Cory, S. The Bcl-2 protein family: arbiters of cell survival. Science (New York, N.Y 281, 1322-1326 (1998).

86. Huang, D.C., Adams, J.M. & Cory, S. The conserved N-terminal BH4 domain of Bcl-2 homologues is essential for inhibition of apoptosis and interaction with CED-4. The EMBO journal 17, 1029-1039 (1998).

87. Borner, C. The Bcl-2 protein family: sensors and checkpoints for life-or-death decisions. Molecular immunology 39, 615-647 (2003).

88. van Delft, M.F. & Huang, D.C. How the Bcl-2 family of proteins interact to regulate apoptosis. Cell research 16, 203-213 (2006).

89. Yin, X.M., Oltvai, Z.N., Veis-Novack, D.J., Linette, G.P. & Korsmeyer, S.J. Bcl-2 gene family and the regulation of programmed cell death. Cold Spring Harbor symposia on quantitative biology 59, 387-393 (1994).

90. Wang, J.L. et al. Cell permeable Bcl-2 binding peptides: a chemical approach to apoptosis induction in tumor cells. Cancer research 60, 1498-1502 (2000).

91. White, A.W., Westwell, A.D. & Brahemi, G. Protein-protein interactions as targets for small- molecule therapeutics in cancer. Expert reviews in molecular medicine 10, e8 (2008).

183

92. Huang, D.C. & Strasser, A. BH3-Only proteins-essential initiators of apoptotic cell death. Cell 103, 839-842 (2000).

93. Wei, M.C. et al. Proapoptotic BAX and BAK: a requisite gateway to mitochondrial dysfunction and death. Science (New York, N.Y 292, 727-730 (2001).

94. Reed, J.C. Apoptosis-based therapies. Nature reviews 1, 111-121 (2002).

95. Nguyen, M. et al. Small molecule obatoclax (GX15-070) antagonizes MCL-1 and overcomes MCL-1-mediated resistance to apoptosis. Proceedings of the National Academy of Sciences of the United States of America 104, 19512-19517 (2007).

96. Tzung, S.P. et al. Antimycin A mimics a cell-death-inducing Bcl-2 homology domain 3. Nature cell biology 3, 183-191 (2001).

97. Chan, S.L. et al. Identification of chelerythrine as an inhibitor of BclXL function. The Journal of biological chemistry 278, 20453-20456 (2003).

98. Real, P.J. et al. Breast cancer cells can evade apoptosis-mediated selective killing by a novel small molecule inhibitor of Bcl-2. Cancer research 64, 7947-7953 (2004).

99. Wang, J.L. et al. Structure-based discovery of an organic compound that binds Bcl-2 protein and induces apoptosis of tumor cells. Proceedings of the National Academy of Sciences of the United States of America 97, 7124-7129 (2000).

100. Kitada, S. et al. Discovery, characterization, and structure-activity relationships studies of proapoptotic polyphenols targeting B-cell lymphocyte/leukemia-2 proteins. Journal of medicinal chemistry 46, 4259-4264 (2003).

101. Klasa, R.J., Gillum, A.M., Klem, R.E. & Frankel, S.R. Oblimersen Bcl-2 antisense: facilitating apoptosis in anticancer treatment. Antisense & nucleic acid drug development 12, 193-213 (2002).

102. Holinger, E.P., Chittenden, T. & Lutz, R.J. Bak BH3 peptides antagonize Bcl-xL function and induce apoptosis through cytochrome c-independent activation of caspases. The Journal of biological chemistry 274, 13298-13304 (1999).

103. Walensky, L.D. et al. Activation of apoptosis in vivo by a hydrocarbon-stapled BH3 helix. Science (New York, N.Y 305, 1466-1470 (2004).

104. Oltersdorf, T. et al. An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 435, 677-681 (2005).

105. Makin, G. & Dive, C. Recent advances in understanding apoptosis: new therapeutic opportunities in cancer chemotherapy. Trends in molecular medicine 9, 251-255 (2003).

106. Tse, C. et al. ABT-263: a potent and orally bioavailable Bcl-2 family inhibitor. Cancer research 68, 3421-3428 (2008).

107. Deveraux, Q.L. & Reed, J.C. IAP family proteins--suppressors of apoptosis. Genes & development 13, 239-252 (1999).

108. Krajewska, M. et al. Elevated expression of inhibitor of apoptosis proteins in prostate cancer. Clin Cancer Res 9, 4914-4925 (2003). 184

109. Tamm, I. et al. Expression and prognostic significance of IAP-family genes in human cancers and myeloid leukemias. Clin Cancer Res 6, 1796-1803 (2000).

110. Chai, J. et al. Structural basis of caspase-7 inhibition by XIAP. Cell 104, 769-780 (2001).

111. Srinivasula, S.M. et al. A conserved XIAP-interaction motif in caspase-9 and Smac/DIABLO regulates caspase activity and apoptosis. Nature 410, 112-116 (2001).

112. Huang, Y. et al. Structural basis of caspase inhibition by XIAP: differential roles of the linker versus the BIR domain. Cell 104, 781-790 (2001).

113. Riedl, S.J. et al. Structural basis for the inhibition of caspase-3 by XIAP. Cell 104, 791-800 (2001).

114. Shiozaki, E.N. et al. Mechanism of XIAP-mediated inhibition of caspase-9. Molecular cell 11, 519-527 (2003).

115. Liu, Z. et al. Structural basis for binding of Smac/DIABLO to the XIAP BIR3 domain. Nature 408, 1004-1008 (2000).

116. Wu, G. et al. Structural basis of IAP recognition by Smac/DIABLO. Nature 408, 1008-1012 (2000).

117. Du, C., Fang, M., Li, Y., Li, L. & Wang, X. Smac, a mitochondrial protein that promotes cytochrome c-dependent caspase activation by eliminating IAP inhibition. Cell 102, 33-42 (2000).

118. Verhagen, A.M. et al. Identification of DIABLO, a mammalian protein that promotes apoptosis by binding to and antagonizing IAP proteins. Cell 102, 43-53 (2000).

119. Chai, J. et al. Structural and biochemical basis of apoptotic activation by Smac/DIABLO. Nature 406, 855-862 (2000).

120. Shiozaki, E.N. & Shi, Y. Caspases, IAPs and Smac/DIABLO: mechanisms from structural biology. Trends in biochemical sciences 29, 486-494 (2004).

121. Arnt, C.R., Chiorean, M.V., Heldebrant, M.P., Gores, G.J. & Kaufmann, S.H. Synthetic Smac/DIABLO peptides enhance the effects of chemotherapeutic agents by binding XIAP and cIAP1 in situ. The Journal of biological chemistry 277, 44236-44243 (2002).

122. Yang, L. et al. Predominant suppression of apoptosome by inhibitor of apoptosis protein in non- small cell lung cancer H460 cells: therapeutic effect of a novel polyarginine-conjugated Smac peptide. Cancer research 63, 831-837 (2003).

123. Sun, H. et al. Structure-based design of potent, conformationally constrained Smac mimetics. Journal of the American Chemical Society 126, 16686-16687 (2004).

124. Sun, H. et al. Structure-based design, synthesis, and evaluation of conformationally constrained mimetics of the second mitochondria-derived activator of caspase that target the X-linked inhibitor of apoptosis protein/caspase-9 interaction site. Journal of medicinal chemistry 47, 4147-4150 (2004).

125. Oost, T.K. et al. Discovery of potent antagonists of the antiapoptotic protein XIAP for the treatment of cancer. Journal of medicinal chemistry 47, 4417-4426 (2004). 185

126. Berg, T. Small-molecule inhibitors of protein-protein interactions. Current opinion in drug discovery & development 11, 666-674 (2008).

127. Cossu, F. et al. Designing Smac-mimetics as antagonists of XIAP, cIAP1, and cIAP2. Biochemical and biophysical research communications 378, 162-167 (2009).

128. Wu, T.Y., Wagner, K.W., Bursulaya, B., Schultz, P.G. & Deveraux, Q.L. Development and characterization of nonpeptidic small molecule inhibitors of the XIAP/caspase-3 interaction. Chemistry & biology 10, 759-767 (2003).

129. Park, C.M. et al. Non-peptidic small molecule inhibitors of XIAP. Bioorganic & medicinal chemistry letters 15, 771-775 (2005).

130. Nikolovska-Coleska, Z. et al. Discovery of embelin as a cell-permeable, small-molecular weight inhibitor of XIAP through structure-based computational screening of a traditional herbal medicine three-dimensional structure database. Journal of medicinal chemistry 47, 2430-2440 (2004).

131. Chen, J., Nikolovska-Coleska, Z., Wang, G., Qiu, S. & Wang, S. Design, synthesis, and characterization of new embelin derivatives as potent inhibitors of X-linked inhibitor of apoptosis protein. Bioorganic & medicinal chemistry letters 16, 5805-5808 (2006).

132. Li, L. et al. A small molecule Smac mimic potentiates TRAIL- and TNFalpha-mediated cell death. Science (New York, N.Y 305, 1471-1474 (2004).

133. Sun, H. et al. Design, synthesis, and characterization of a potent, nonpeptide, cell-permeable, bivalent Smac mimetic that concurrently targets both the BIR2 and BIR3 domains in XIAP. Journal of the American Chemical Society 129, 15279-15294 (2007).

134. Vogelstein, B., Lane, D. & Levine, A.J. Surfing the p53 network. Nature 408, 307-310 (2000).

135. Donehower, L.A. et al. Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumours. Nature 356, 215-221 (1992).

136. Chene, P. Inhibiting the p53–MDM2 Interaction: An Important Target For Cancer Therapy. Nature Review 3, 102 (2003).

137. Levine, A.J. p53, the cellular gatekeeper for growth and division. Cell 88, 323-331 (1997).

138. Hollstein, M. et al. Database of p53 gene somatic mutations in human tumors and cell lines. Nucleic acids research 22, 3551-3555 (1994).

139. Kubbutat, M.H., Jones, S.N. & Vousden, K.H. Regulation of p53 stability by Mdm2. Nature 387, 299-303 (1997).

140. Kussie, P.H. et al. Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science (New York, N.Y 274, 948-953 (1996).

141. Vassilev, L.T. et al. In vivo activation of the p53 pathway by small-molecule antagonists of MDM2. Science (New York, N.Y 303, 844-848 (2004).

142. Bottger, A. et al. Molecular characterization of the hdm2-p53 interaction. Journal of molecular biology 269, 744-756 (1997). 186

143. Garcia-Echeverria, C., Chene, P., Blommers, M.J. & Furet, P. Discovery of potent antagonists of the interaction between human double minute 2 and tumor suppressor p53. Journal of medicinal chemistry 43, 3205-3208 (2000).

144. Bernal, F., Tyler, A.F., Korsmeyer, S.J., Walensky, L.D. & Verdine, G.L. Reactivation of the p53 tumor suppressor pathway by a stapled p53 peptide. Journal of the American Chemical Society 129, 2456-2457 (2007).

145. Stoll, R. et al. Chalcone derivatives antagonize interactions between the human oncoprotein MDM2 and p53. Biochemistry 40, 336-344 (2001).

146. Shangary, S. et al. Temporal activation of p53 by a specific MDM2 inhibitor is selectively toxic to tumors and leads to complete tumor growth inhibition. Proceedings of the National Academy of Sciences of the United States of America 105, 3933-3938 (2008).

147. Duncan, S.J. et al. Isolation and structure elucidation of Chlorofusin, a novel p53-MDM2 antagonist from a Fusarium sp. Journal of the American Chemical Society 123, 554-560 (2001).

148. Bowman, A.L., Nikolovska-Coleska, Z., Zhong, H., Wang, S. & Carlson, H.A. Small molecule inhibitors of the MDM2-p53 interaction discovered by ensemble-based receptor models. Journal of the American Chemical Society 129, 12809-12814 (2007).

149. Grasberger, B.L. et al. Discovery and cocrystal structure of benzodiazepinedione HDM2 antagonists that activate p53 in cells. Journal of medicinal chemistry 48, 909-912 (2005).

150. Galatin, P.S. & Abraham, D.J. A nonpeptidic sulfonamide inhibits the p53-mdm2 interaction and activates p53-dependent transcription in mdm2-overexpressing cells. Journal of medicinal chemistry 47, 4163-4165 (2004).

151. Parks, D.J. et al. 1,4-Benzodiazepine-2,5-diones as small molecule antagonists of the HDM2-p53 interaction: discovery and SAR. Bioorganic & medicinal chemistry letters 15, 765-770 (2005).

152. Shoichet, B.K. Virtual screening of chemical libraries. Nature 432, 862-865 (2004).

153. Scapin, G. Structural biology and drug discovery. Current pharmaceutical design 12, 2087-2097 (2006).

154. Hajduk, P.J. & Greer, J. A decade of fragment-based drug design: strategic advances and lessons learned. Nature reviews 6, 211-219 (2007).

155. Shuker, S.B., Hajduk, P.J., Meadows, R.P. & Fesik, S.W. Discovering high-affinity ligands for proteins: SAR by NMR. Science (New York, N.Y 274, 1531-1534 (1996).

156. Martin, S., Brown, W.M. & Faulon, J.L. Using product kernels to predict protein interactions. Advances in biochemical engineering/biotechnology 110, 215-245 (2008).

157. Guo, J., Wu, X., Zhang, D.Y. & Lin, K. Genome-wide inference of protein interaction sites: lessons from the yeast high-quality negative protein-protein interaction dataset. Nucleic acids research 36, 2002-2011 (2008).

158. Ben-Hur, A. & Noble, W.S. Kernel methods for predicting protein-protein interactions. Bioinformatics (Oxford, England) 21 Suppl 1, i38-46 (2005).

187

159. Shen, J. et al. Predicting protein-protein interactions based only on sequences information. Proceedings of the National Academy of Sciences of the United States of America 104, 4337-4341 (2007).

160. MacDonald, M.L. et al. Identifying off-target effects and hidden phenotypes of drugs in human cells. Nature chemical biology 2, 329-337 (2006).

161. Hall, J.M. et al. Linkage of early-onset familial breast cancer to chromosome 17q21. Science (New York, N.Y 250, 1684-1689 (1990).

162. Miki, Y. et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science (New York, N.Y 266, 66-71 (1994).

163. Stratton, M.R. et al. Familial male breast cancer is not linked to the BRCA1 locus on chromosome 17q. Nature genetics 7, 103-107 (1994).

164. Wooster, R. et al. Identification of the breast cancer susceptibility gene BRCA2. Nature 378, 789- 792 (1995).

165. Irminger-Finger, I. & Jefford, C.E. Is there more to BARD1 than BRCA1? Nat Rev Cancer 6, 382-391 (2006).

166. King, M.C., Marks, J.H. & Mandell, J.B. Breast and ovarian cancer risks due to inherited mutations in BRCA1 and BRCA2. Science (New York, N.Y 302, 643-646 (2003).

167. Greenberg, R.A. et al. Multifactorial contributions to an acute DNA damage response by BRCA1/BARD1-containing complexes. Genes & development 20, 34-46 (2006).

168. Narod, S.A. & Foulkes, W.D. BRCA1 and BRCA2: 1994 and beyond. Nat Rev Cancer 4, 665-676 (2004).

169. Baer, R. & Ludwig, T. The BRCA1/BARD1 heterodimer, a tumor suppressor complex with ubiquitin E3 ligase activity. Current opinion in genetics & development 12, 86-91 (2002).

170. Deng, C.X. & Scott, F. Role of the tumor suppressor gene Brca1 in genetic stability and mammary gland tumor formation. Oncogene 19, 1059-1064 (2000).

171. Ludwig, T., Fisher, P., Ganesan, S. & Efstratiadis, A. Tumorigenesis in mice carrying a truncating Brca1 mutation. Genes & development 15, 1188-1193 (2001).

172. Ludwig, T., Fisher, P., Murty, V. & Efstratiadis, A. Development of mammary adenocarcinomas by tissue-specific knockout of Brca2 in mice. Oncogene 20, 3937-3948 (2001).

173. Xu, X. et al. Conditional mutation of Brca1 in mammary epithelial cells results in blunted ductal morphogenesis and tumour formation. Nature genetics 22, 37-43 (1999).

174. Xu, X. et al. Centrosome amplification and a defective G2-M cell cycle checkpoint induce genetic instability in BRCA1 exon 11 isoform-deficient cells. Molecular cell 3, 389-395 (1999).

175. Somasundaram, K. Breast cancer gene 1 (BRCA1): role in cell cycle regulation and DNA repair-- perhaps through transcription. Journal of cellular biochemistry 88, 1084-1091 (2003).

188

176. Scully, R. Role of BRCA gene dysfunction in breast and ovarian cancer predisposition. Breast Cancer Res 2, 324-330 (2000).

177. Johnson, R.D. & Jasin, M. Double-strand-break-induced homologous recombination in mammalian cells. Biochemical Society transactions 29, 196-201 (2001).

178. Scully, R., Puget, N. & Vlasakova, K. DNA polymerase stalling, sister chromatid recombination and the BRCA genes. Oncogene 19, 6176-6183 (2000).

179. Deng, C.X. & Brodie, S.G. Roles of BRCA1 and its interacting proteins. Bioessays 22, 728-737 (2000).

180. Scully, R. & Livingston, D.M. In search of the tumour-suppressor functions of BRCA1 and BRCA2. Nature 408, 429-432 (2000).

181. Venkitaraman, A.R. Chromosome stability, DNA recombination and the BRCA2 tumour suppressor. Current opinion in cell biology 13, 338-343 (2001).

182. Monteiro, A.N. BRCA1: exploring the links to transcription. Trends in biochemical sciences 25, 469-474 (2000).

183. Welcsh, P.L. & King, M.C. BRCA1 and BRCA2 and the genetics of breast and ovarian cancer. Hum Mol Genet 10, 705-713 (2001).

184. Welcsh, P.L., Owens, K.N. & King, M.C. Insights into the functions of BRCA1 and BRCA2. Trends Genet 16, 69-74 (2000).

185. Starita, L.M. & Parvin, J.D. Substrates of the BRCA1-dependent ubiquitin ligase. Cancer biology & therapy 5, 137-141 (2006).

186. Scully, R. Interactions between BRCA proteins and DNA structure. Exp Cell Res 264, 67-73 (2001).

187. Scully, R. et al. BRCA1 is a component of the RNA polymerase II holoenzyme. Proceedings of the National Academy of Sciences of the United States of America 94, 5605-5610 (1997).

188. Scully, R. et al. Association of BRCA1 with Rad51 in mitotic and meiotic cells. Cell 88, 265-275 (1997).

189. Jin, Y. et al. Cell cycle-dependent colocalization of BARD1 and BRCA1 proteins in discrete nuclear domains. Proceedings of the National Academy of Sciences of the United States of America 94, 12075-12080 (1997).

190. Deng, C.X. Roles of BRCA1 in centrosome duplication. Oncogene 21, 6222-6227 (2002).

191. Wu, L.C. et al. Identification of a RING protein that can interact in vivo with the BRCA1 gene product. Nature genetics 14, 430-440 (1996).

192. Hashizume, R. et al. The RING heterodimer BRCA1-BARD1 is a ubiquitin ligase inactivated by a breast cancer-derived mutation. J Biol Chem 276, 14537-14540 (2001).

189

193. McCarthy, E.E., Celebi, J.T., Baer, R. & Ludwig, T. Loss of Bard1, the heterodimeric partner of the Brca1 tumor suppressor, results in early embryonic lethality and chromosomal instability. Molecular and cellular biology 23, 5056-5063 (2003).

194. Brzovic, P.S., Rajagopal, P., Hoyt, D.W., King, M.C. & Klevit, R.E. Structure of a BRCA1- BARD1 heterodimeric RING-RING complex. Nature structural biology 8, 833-837 (2001).

195. Brzovic, P.S. et al. Binding and recognition in the assembly of an active BRCA1/BARD1 ubiquitin-ligase complex. Proceedings of the National Academy of Sciences of the United States of America 100, 5646-5651 (2003).

196. Meza, J.E., Brzovic, P.S., King, M.C. & Klevit, R.E. Mapping the functional domains of BRCA1. Interaction of the ring finger domains of BRCA1 and BARD1. The Journal of biological chemistry 274, 5659-5665 (1999).

197. Oyake, D., Nishikawa, H., Koizuka, I., Fukuda, M. & Ohta, T. Targeted substrate degradation by an engineered double RING ubiquitin ligase. Biochemical and biophysical research communications 295, 370-375 (2002).

198. Morris, J.R. & Solomon, E. BRCA1 : BARD1 induces the formation of conjugated ubiquitin structures, dependent on K6 of ubiquitin, in cells during DNA replication and repair. Human molecular genetics 13, 807-817 (2004).

199. Reid, L.J. et al. E3 ligase activity of BRCA1 is not essential for mammalian cell viability or homology-directed repair of double-strand DNA breaks. Proceedings of the National Academy of Sciences of the United States of America 105, 20876-20881 (2008).

200. Brzovic, P.S., Meza, J.E., King, M.C. & Klevit, R.E. BRCA1 RING domain cancer-predisposing mutations. Structural consequences and effects on protein-protein interactions. The Journal of biological chemistry 276, 41399-41406 (2001).

201. Morris, J.R., Keep, N.H. & Solomon, E. Identification of residues required for the interaction of BARD1 with BRCA1. The Journal of biological chemistry 277, 9382-9386 (2002).

202. Morris, J.R. et al. Genetic analysis of BRCA1 ubiquitin ligase activity and its relationship to breast cancer susceptibility. Human molecular genetics 15, 599-606 (2006).

203. Nishikawa, H. et al. Mass spectrometric and mutational analyses reveal Lys-6-linked polyubiquitin chains catalyzed by BRCA1-BARD1 ubiquitin ligase. The Journal of biological chemistry 279, 3916-3924 (2004).

204. Wu-Baer, F., Lagrazon, K., Yuan, W. & Baer, R. The BRCA1/BARD1 heterodimer assembles polyubiquitin chains through an unconventional linkage involving lysine residue K6 of ubiquitin. The Journal of biological chemistry 278, 34743-34746 (2003).

205. Mallery, D.L., Vandenberg, C.J. & Hiom, K. Activation of the E3 ligase function of the BRCA1/BARD1 complex by polyubiquitin chains. Embo J 21, 6755-6762 (2002).

206. Wang, Y. et al. BASC, a super complex of BRCA1-associated proteins involved in the recognition and repair of aberrant DNA structures. Genes & development 14, 927-939 (2000).

190

207. Trujillo, K.M., Yuan, S.S., Lee, E.Y. & Sung, P. Nuclease activities in a complex of human recombination and DNA repair factors Rad50, Mre11, and p95. The Journal of biological chemistry 273, 21447-21450 (1998).

208. Paull, T.T. & Gellert, M. Nbs1 potentiates ATP-driven DNA unwinding and endonuclease cleavage by the Mre11/Rad50 complex. Genes & development 13, 1276-1288 (1999).

209. Zhong, Q. et al. Association of BRCA1 with the hRad50-hMre11-p95 complex and the DNA damage response. Science 285, 747-750 (1999).

210. Chen, A., Kleiman, F.E., Manley, J.L., Ouchi, T. & Pan, Z.Q. Autoubiquitination of the BRCA1*BARD1 RING ubiquitin ligase. The Journal of biological chemistry 277, 22085-22092 (2002).

211. Bochar, D.A. et al. BRCA1 is associated with a human SWI/SNF-related complex: linking chromatin remodeling to breast cancer. Cell 102, 257-265 (2000).

212. Harkin, D.P. et al. Induction of GADD45 and JNK/SAPK-dependent apoptosis following inducible expression of BRCA1. Cell 97, 575-586 (1999).

213. Zheng, L. et al. Sequence-specific transcriptional corepressor function for BRCA1 through a novel zinc finger protein, ZBRK1. Molecular cell 6, 757-768 (2000).

214. Yun, J. & Lee, W.H. Degradation of transcription repressor ZBRK1 through the ubiquitin- proteasome pathway relieves repression of Gadd45a upon DNA damage. Molecular and cellular biology 23, 7305-7314 (2003).

215. Lee, K.B., Wang, D., Lippard, S.J. & Sharp, P.A. Transcription-coupled and DNA damage- dependent ubiquitination of RNA polymerase II in vitro. Proceedings of the National Academy of Sciences of the United States of America 99, 4239-4244 (2002).

216. Kleiman, F.E. et al. BRCA1/BARD1 inhibition of mRNA 3' processing involves targeted degradation of RNA polymerase II. Genes & development 19, 1227-1237 (2005).

217. Xu, B., Kim, S. & Kastan, M.B. Involvement of Brca1 in S-phase and G(2)-phase checkpoints after ionizing irradiation. Molecular and cellular biology 21, 3445-3450 (2001).

218. Choudhury, A.D., Xu, H. & Baer, R. Ubiquitination and proteasomal degradation of the BRCA1 tumor suppressor is regulated during cell cycle progression. The Journal of biological chemistry 279, 33909-33918 (2004).

219. Lingle, W.L. et al. Centrosome amplification drives chromosomal instability in breast tumor development. Proceedings of the National Academy of Sciences of the United States of America 99, 1978-1983 (2002).

220. Hsu, L.C., Doan, T.P. & White, R.L. Identification of a gamma-tubulin-binding domain in BRCA1. Cancer research 61, 7713-7718 (2001).

221. Starita, L.M. et al. BRCA1-dependent ubiquitination of gamma-tubulin regulates centrosome number. Molecular and cellular biology 24, 8457-8466 (2004).

222. Sato, K. et al. Nucleophosmin/B23 is a candidate substrate for the BRCA1-BARD1 ubiquitin ligase. The Journal of biological chemistry 279, 30919-30922 (2004). 191

223. Costa, L.G. et al. Serum paraoxonase and its influence on paraoxon and chlorpyrifos-oxon toxicity in rats. Toxicology and applied pharmacology 103, 66-76 (1990).

224. Furlong, C.E., Richter, R.J., Chapline, C. & Crabb, J.W. Purification of rabbit and human serum paraoxonase. Biochemistry 30, 10133-10140 (1991).

225. Hassett, C. et al. Characterization of cDNA clones encoding rabbit and human serum paraoxonase: the mature protein retains its signal sequence. Biochemistry 30, 10141-10149 (1991).

226. Sorenson, R.C. et al. Human serum Paraoxonase/Arylesterase's retained hydrophobic N-terminal leader sequence associates with HDLs by binding phospholipids : apolipoprotein A-I stabilizes activity. Arteriosclerosis, thrombosis, and vascular biology 19, 2214-2225 (1999).

227. Kuo, C.L. & La Du, B.N. Calcium binding by human and rabbit serum paraoxonases. Structural stability and enzymatic activity. Drug metabolism and disposition: the biological fate of chemicals 26, 653-660 (1998).

228. Kuo, C.L.a.L.D.B.N. Comparison of human and rabbit serum paraoxonase. Drug metabolism and disposition: the biological fate of chemicals 23, 935-944 (1995).

229. Costa, L.G.F., Clement E Paraoxonase (PON1) in Health and Disease: Basic and Clinical Aspects (Kluwer Academic Publishers, 2002).

230. Primo-Parmo, S.L., Sorenson, R.C., Teiber, J. & La Du, B.N. The human serum paraoxonase/arylesterase gene (PON1) is one member of a multigene family. Genomics 33, 498- 507 (1996).

231. Draganov, D.I., Stetson, P.L., Watson, C.E., Billecke, S.S. & La Du, B.N. Rabbit serum paraoxonase 3 (PON3) is a high density lipoprotein-associated lactonase and protects low density lipoprotein against oxidation. The Journal of biological chemistry 275, 33435-33442 (2000).

232. Reddy, S.T. et al. Human paraoxonase-3 is an HDL-associated enzyme with biological activity similar to paraoxonase-1 protein but is not regulated by oxidized lipids. Arteriosclerosis, thrombosis, and vascular biology 21, 542-547 (2001).

233. Mochizuki, H. et al. Human PON2 gene at 7q21.3: cloning, multiple mRNA forms, and missense polymorphisms in the coding sequence. Gene 213, 149-157 (1998).

234. Ng, C.J. et al. Paraoxonase-2 is a ubiquitously expressed protein with antioxidant properties and is capable of preventing cell-mediated oxidative modification of low density lipoprotein. The Journal of biological chemistry 276, 44444-44449 (2001).

235. Rosenblat, M. et al. Mouse macrophage paraoxonase 2 activity is increased whereas cellular paraoxonase 3 activity is decreased under oxidative stress. Arteriosclerosis, thrombosis, and vascular biology 23, 468-474 (2003).

236. Rochu, D., Chabriere, E. & Masson, P. Human paraoxonase: a promising approach for pre- treatment and therapy of organophosphorus poisoning. Toxicology 233, 47-59 (2007).

237. Davies, H.G. et al. The effect of the human serum paraoxonase polymorphism is reversed with diazoxon, soman and sarin. Nature genetics 14, 334-336 (1996).

192

238. Durrington, P.N., Mackness, B. & Mackness, M.I. Paraoxonase and atherosclerosis. Arteriosclerosis, thrombosis, and vascular biology 21, 473-480 (2001).

239. Khersonsky, O. & Tawfik, D.S. The histidine 115-histidine 134 dyad mediates the lactonase activity of mammalian serum paraoxonases. The Journal of biological chemistry 281, 7649-7656 (2006).

240. Garin, M.C. et al. Paraoxonase polymorphism Met-Leu54 is associated with modified serum concentrations of the enzyme. A possible link between the paraoxonase gene and increased risk of cardiovascular disease in diabetes. The Journal of clinical investigation 99, 62-66 (1997).

241. Ruiz, J. et al. Gln-Arg192 polymorphism of paraoxonase and coronary heart disease in type 2 diabetes. Lancet 346, 869-872 (1995).

242. James, R.W. et al. Promoter polymorphism T(-107)C of the paraoxonase PON1 gene is a risk factor for coronary heart disease in type 2 diabetic patients. Diabetes 49, 1390-1393 (2000).

243. Leviev, I. & James, R.W. Promoter polymorphisms of human paraoxonase PON1 gene and serum paraoxonase activities and concentrations. Arteriosclerosis, thrombosis, and vascular biology 20, 516-521 (2000).

244. Gouedard, C., Koum-Besson, N., Barouki, R. & Morel, Y. Opposite regulation of the human paraoxonase-1 gene PON-1 by fenofibrate and statins. Molecular pharmacology 63, 945-956 (2003).

245. Shih, D.M. et al. Mice lacking serum paraoxonase are susceptible to organophosphate toxicity and atherosclerosis. Nature 394, 284-287 (1998).

246. Eddleston, M., Buckley, N.A., Eyer, P. & Dawson, A.H. Management of acute organophosphorus pesticide poisoning. Lancet 371, 597-607 (2008).

247. Tang, J. et al. Metabolism of chlorpyrifos by human cytochrome P450 isoforms and human, mouse, and rat liver microsomes. Drug metabolism and disposition: the biological fate of chemicals 29, 1201-1204 (2001).

248. Aldridge, W.N. Serum esterases. II. An enzyme hydrolysing diethyl p-nitrophenyl phosphate (E600) and its identity with the A-esterase of mammalian sera. The Biochemical journal 53, 117- 124 (1953).

249. Davison, A.N. Inhibition of the cholinesterase of the central nervous system by organophosphorus compounds. The Biochemical journal 54, xix-xx (1953).

250. Aldridge, W.N. & Davison, A.N. The mechanism of inhibition of cholinesterases by organophosphorus compounds. The Biochemical journal 55, 763-766 (1953).

251. Eyer, P. The role of oximes in the management of organophosphorus pesticide poisoning. Toxicological reviews 22, 165-190 (2003).

252. Kovarik, Z. et al. Acetylcholinesterase active centre and gorge conformations analysed by combinatorial mutations and enantiomeric phosphonates. The Biochemical journal 373, 33-40 (2003).

193

253. Marrs, T.C., Rice, P. & Vale, J.A. The role of oximes in the treatment of nerve agent poisoning in civilian casualties. Toxicological reviews 25, 297-323 (2006).

254. Wilson, I.B. & Ginsburg, B. A powerful reactivator of alkylphosphate-inhibited acetylcholinesterase. Biochimica et biophysica acta 18, 168-170 (1955).

255. Amitai, G., Moorad, D., Adani, R. & Doctor, B.P. Inhibition of acetylcholinesterase and butyrylcholinesterase by chlorpyrifos-oxon. Biochemical pharmacology 56, 293-299 (1998).

256. Gordon, J.J., Leadbeater, L. & Maidment, M.P. The protection of animals against organophosphate poisoning by pretreatment with a carbamate. Toxicology and applied pharmacology 43, 207-216 (1978).

257. Vale, J.A., Meredith, T.J. & Heath, A. High dose atropine in organophosphorus poisoning. Postgraduate medical journal 66, 878 (1990).

258. Gray, A.P. Design and structure-activity relationships of antidotes to organophosphorus anticholinesterase agents. Drug metabolism reviews 15, 557-589 (1984).

259. Ashani, Y. et al. Amino acid residues controlling reactivation of organophosphonyl conjugates of acetylcholinesterase by mono- and bisquaternary oximes. The Journal of biological chemistry 270, 6370-6380 (1995).

260. Saxena, A. et al. Bioscavenger for protection from toxicity of organophosphorus compounds. J Mol Neurosci 30, 145-148 (2006).

261. Cerasoli, D.M. et al. In vitro and in vivo characterization of recombinant human butyrylcholinesterase (Protexia) as a potential nerve agent bioscavenger. Chemico-biological interactions 157-158, 363-365 (2005).

262. Wang, Y. et al. Resistance to organophosphorus agent toxicity in transgenic mice expressing the G117H mutant of human butyrylcholinesterase. Toxicology and applied pharmacology 196, 356- 366 (2004).

263. Li, W.F., Costa, L.G. & Furlong, C.E. Serum paraoxonase status: a major factor in determining resistance to organophosphates. Journal of toxicology and environmental health 40, 337-346 (1993).

264. Li, W.F. et al. Catalytic efficiency determines the in-vivo efficacy of PON1 for detoxifying organophosphorus compounds. Pharmacogenetics 10, 767-779 (2000).

265. Li, W.F., Furlong, C.E. & Costa, L.G. Paraoxonase protects against chlorpyrifos toxicity in mice. Toxicology letters 76, 219-226 (1995).

266. Mazur, A. An enzyme in animal tissue capable of hydrolyzing the phosphorus-fluorine bond of alkyl fluorophosphates. . The Journal of biological chemistry 164, 271 - 289 (1946).

267. La Du, B.N. Human serum paraoxonase/arylesterase. Genetic factors influencing the metabolism of foreign compounds. International encyclopedia of pharmacology and therapeutics. Pergamon Press, New York, 51-91 (1992).

268. Broomfield, C.A., Ford, K. W. Hydrolysis of nerve gases by plasma enzymes. . Proceedings of the 3rd International Meeting on Cholinesterases, La Grande-Motte, 161 (1991). 194

269. Shih, D.M. et al. Combined serum paraoxonase knockout/apolipoprotein E knockout mice exhibit increased lipoprotein oxidation and atherosclerosis. The Journal of biological chemistry 275, 17527-17535 (2000).

270. Ghanem, E. & Raushel, F.M. Detoxification of organophosphate nerve agents by bacterial phosphotriesterase. Toxicology and applied pharmacology 207, 459-470 (2005).

271. Dumas, D.P., Durst, H.D., Landis, W.G., Raushel, F.M. & Wild, J.R. Inactivation of organophosphorus nerve agents by the phosphotriesterase from Pseudomonas diminuta. Archives of biochemistry and biophysics 277, 155-159 (1990).

272. Harel, M. et al. Structure and evolution of the serum paraoxonase family of detoxifying and anti- atherosclerotic enzymes. Nature structural & molecular biology 11, 412-419 (2004).

273. Aharoni, A. et al. Directed evolution of mammalian paraoxonases PON1 and PON3 for bacterial expression and catalytic specialization. Proceedings of the National Academy of Sciences of the United States of America 101, 482-487 (2004).

274. Abecassis, V., Pompon, D. & Truan, G. High efficiency family shuffling based on multi-step PCR and in vivo DNA recombination in yeast: statistical and functional analysis of a combinatorial library between human cytochrome P450 1A1 and 1A2. Nucleic acids research 28, E88 (2000).

275. Crameri, A., Raillard, S.A., Bermudez, E. & Stemmer, W.P. DNA shuffling of a family of genes from diverse species accelerates directed evolution. Nature 391, 288-291 (1998).

276. Stevens, R.C. et al. Engineered recombinant human paraoxonase 1 (rHuPON1) purified from Escherichia coli protects against organophosphate poisoning. Proceedings of the National Academy of Sciences of the United States of America 105, 12780-12784 (2008).

277. Rochu, D., Renault, F., Clery-Barraud, C., Chabriere, E. & Masson, P. Stability of highly purified human paraoxonase (PON1): association with human phosphate binding protein (HPBP) is essential for preserving its active conformation(s). Biochimica et biophysica acta 1774, 874-883 (2007).

278. Bog-Hansen, T.C., Krog, H.H. & Back, U. Plasma lipoprotein-associated arylesterase is induced by bacterial lipopolysaccharide. FEBS letters 93, 86-90 (1978).

279. Mackness, M.I. 'A'-esterases. Enzymes looking for a role? Biochemical pharmacology 38, 385- 390 (1989).

280. Mackness, B., Durrington, P.N. & Mackness, M.I. Human serum paraoxonase. General pharmacology 31, 329-336 (1998).

281. Aviram, M. Does paraoxonase play a role in susceptibility to cardiovascular disease? Molecular medicine today 5, 381-386 (1999).

282. Sanghera, D.K., Saha, N., Aston, C.E. & Kamboh, M.I. Genetic polymorphism of paraoxonase and the risk of coronary heart disease. Arteriosclerosis, thrombosis, and vascular biology 17, 1067-1073 (1997).

283. Zama, T. et al. A 192Arg variant of the human paraoxonase (HUMPONA) gene polymorphism is associated with an increased risk for coronary artery disease in the Japanese. Arteriosclerosis, thrombosis, and vascular biology 17, 3565-3569 (1997). 195

284. Odawara, M., Tachi, Y. & Yamashita, K. Paraoxonase polymorphism (Gln192-Arg) is associated with coronary heart disease in Japanese noninsulin-dependent diabetes mellitus. The Journal of clinical endocrinology and metabolism 82, 2257-2260 (1997).

285. Antikainen, M. et al. The Gln-Arg191 polymorphism of the human paraoxonase gene (HUMPONA) is not associated with the risk of coronary artery disease in Finns. The Journal of clinical investigation 98, 883-885 (1996).

286. Cao, H. et al. Lack of association between carotid intima-media thickness and paraoxonase gene polymorphism in non-insulin dependent diabetes mellitus. Atherosclerosis 138, 361-366 (1998).

287. Herrmann, S.M. et al. The Gln/Arg polymorphism of human paraoxonase (PON 192) is not related to myocardial infarction in the ECTIM Study. Atherosclerosis 126, 299-303 (1996).

288. Ko, Y.L. et al. The Gln-Arg 191 polymorphism of the human paraoxonase gene is not associated with the risk of coronary artery disease among Chinese in Taiwan. Atherosclerosis 141, 259-264 (1998).

289. Rice, G.I., Ossei-Gerning, N., Stickland, M.H. & Grant, P.J. The paraoxonase Gln-Arg 192 polymorphism in subjects with ischaemic heart disease. Coronary artery disease 8, 677-682 (1997).

290. Mackness, M.I. et al. Paraoxonase and coronary heart disease. Current opinion in lipidology 9, 319-324 (1998).

291. Steinberg, D., Parthasarathy, S., Carew, T.E., Khoo, J.C. & Witztum, J.L. Beyond cholesterol. Modifications of low-density lipoprotein that increase its atherogenicity. The New England journal of medicine 320, 915-924 (1989).

292. La Du, B.N. Structural and functional diversity of paraoxonases. Nature medicine 2, 1186-1187 (1996).

293. Aviram, M. et al. Paraoxonase inhibits high-density lipoprotein oxidation and preserves its functions. A possible peroxidative role for paraoxonase. The Journal of clinical investigation 101, 1581-1590 (1998).

294. Heinecke, J.W. & Lusis, A.J. Paraoxonase-gene polymorphisms associated with coronary heart disease: support for the oxidative damage hypothesis? American journal of human genetics 62, 20- 24 (1998).

295. Mackness, M.I., Arrol, S., Abbott, C. & Durrington, P.N. Protection of low-density lipoprotein against oxidative modification by high-density lipoprotein associated paraoxonase. Atherosclerosis 104, 129-135 (1993).

296. Mackness, M.I., Arrol, S. & Durrington, P.N. Paraoxonase prevents accumulation of lipoperoxides in low-density lipoprotein. FEBS letters 286, 152-154 (1991).

297. Shih, D.M. et al. Genetic-dietary regulation of serum paraoxonase expression and its role in atherogenesis in a mouse model. The Journal of clinical investigation 97, 1630-1639 (1996).

298. Navab, M. et al. High density associated enzymes: their role in vascular biology. Current opinion in lipidology 9, 449-456 (1998).

196

299. Navab, M. et al. Mildly oxidized LDL induces an increased apolipoprotein J/paraoxonase ratio. The Journal of clinical investigation 99, 2005-2019 (1997).

300. Watson, A.D. et al. Protective effect of high density lipoprotein associated paraoxonase. Inhibition of the biological activity of minimally oxidized low density lipoprotein. The Journal of clinical investigation 96, 2882-2891 (1995).

301. Oda, M.N. et al. Paraoxonase 1 overexpression in mice and its effect on high-density lipoproteins. Biochemical and biophysical research communications 290, 921-927 (2002).

302. Aviram, M. Macrophage foam cell formation during early atherogenesis is determined by the balance between pro-oxidants and anti-oxidants in arterial cells and blood lipoproteins. Antioxidants & redox signaling 1, 585-594 (1999).

303. Rosenblat, M., Vaya, J., Shih, D. & Aviram, M. Paraoxonase 1 (PON1) enhances HDL-mediated macrophage cholesterol efflux via the ABCA1 transporter in association with increased HDL binding to the cells: a possible role for lysophosphatidylcholine. Atherosclerosis 179, 69-77 (2005).

304. Mackness, B., Durrington, P.N., Boulton, A.J., Hine, D. & Mackness, M.I. Serum paraoxonase activity in patients with type 1 diabetes compared to healthy controls. European journal of clinical investigation 32, 259-264 (2002).

305. Boemi, M. et al. Serum paraoxonase is reduced in type 1 diabetic patients compared to non- diabetic, first degree relatives; influence on the ability of HDL to protect LDL from oxidation. Atherosclerosis 155, 229-235 (2001).

306. Kao, Y.L., Donaghue, K., Chan, A., Knight, J. & Silink, M. A variant of paraoxonase (PON1) gene is associated with diabetic retinopathy in IDDM. The Journal of clinical endocrinology and metabolism 83, 2589-2592 (1998).

307. Murata, M., Nakagawa, M. & Takahashi, S. Molecular variant of the human paraoxonase/arylesterase gene is associated with central retinal vein occlusion in the Japanese population. Ophthalmologica. Journal international d'ophtalmologie. International journal of ophthalmology 212, 257-259 (1998).

308. Hu, Y., Tian, H. & Liu, R. Gln-Arg192 polymorphism of paraoxonase 1 is associated with carotid intima-media thickness in patients of type 2 diabetes mellitus of Chinese. Diabetes research and clinical practice 61, 21-27 (2003).

309. Koch, M. et al. Paraoxonase 1 192 Gln/Arg gene polymorphism and cerebrovascular disease: interaction with type 2 diabetes. Exp Clin Endocrinol Diabetes 109, 141-145 (2001).

310. Osei-Hyiaman, D. et al. Coronary artery disease risk in Chinese type 2 diabetics: is there a role for paraxonase 1 gene (Q192R) polymorphism? European journal of endocrinology / European Federation of Endocrine Societies 144, 639-644 (2001).

311. Kapust, R.B. & Waugh, D.S. Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused. Protein Sci 8, 1668-1674 (1999).

312. Rabhi-Essafi, I., Sadok, A., Khalaf, N. & Fathallah, D.M. A strategy for high-level expression of soluble and functional human interferon alpha as a GST-fusion protein in E. coli. Protein Eng Des Sel 20, 201-209 (2007). 197

313. Bleimling, N., Alexandrov, K., Goody, R. & Itzen, A. Chaperone-assisted production of active human Rab8A GTPase in Escherichia coli. Protein expression and purification 65, 190-195 (2009).

314. de Marco, A., Deuerling, E., Mogk, A., Tomoyasu, T. & Bukau, B. Chaperone-based procedure to increase yields of soluble recombinant proteins produced in E. coli. BMC biotechnology 7, 32 (2007).

315. Horwich, A.L., Farr, G.W. & Fenton, W.A. GroEL-GroES-mediated protein folding. Chemical reviews 106, 1917-1930 (2006).

316. Martin, A., Sieber, V. & Schmid, F.X. In-vitro selection of highly stabilized protein variants with optimized surface. Journal of molecular biology 309, 717-726 (2001).

317. Wintrode, P.L. & Arnold, F.H. Temperature adaptation of enzymes: lessons from laboratory evolution. Advances in protein chemistry 55, 161-225 (2000).

318. Wintrode, P.L., Miyazaki, K. & Arnold, F.H. Cold adaptation of a mesophilic subtilisin-like protease by laboratory evolution. The Journal of biological chemistry 275, 31635-31640 (2000).

319. Lehmann, M. et al. The consensus concept for thermostability engineering of proteins: further proof of concept. Protein engineering 15, 403-411 (2002).

320. Alexander, P.A., Ruan, B., Strausberg, S.L. & Bryan, P.N. Stabilizing mutations and calcium- dependent stability of subtilisin. Biochemistry 40, 10640-10644 (2001).

321. Li, H., Cocco, M.J., Steitz, T.A. & Engelman, D.M. Conversion of phospholamban into a soluble pentameric helical bundle. Biochemistry 40, 6636-6645 (2001).

322. Slovic, A.M., Summa, C.M., Lear, J.D. & DeGrado, W.F. Computational design of a water- soluble analog of phospholamban. Protein Sci 12, 337-348 (2003).

323. Slovic, A.M., Kono, H., Lear, J.D., Saven, J.G. & DeGrado, W.F. Computational design of water- soluble analogues of the potassium channel KcsA. Proceedings of the National Academy of Sciences of the United States of America 101, 1828-1833 (2004).

324. Eijsink, V.G. et al. Rational engineering of enzyme stability. Journal of biotechnology 113, 105- 120 (2004).

325. Tracewell, C.A. & Arnold, F.H. Directed enzyme evolution: climbing fitness peaks one amino acid at a time. Current opinion in chemical biology 13, 3-9 (2009).

326. Miyazaki, K., Wintrode, P.L., Grayling, R.A., Rubingh, D.N. & Arnold, F.H. Directed evolution study of temperature adaptation in a psychrophilic enzyme. Journal of molecular biology 297, 1015-1026 (2000).

327. Farinas, E.T.S., Ulrich; Glieder, Anton; Arnold, Frances H Directed Evolution of a Cytochrome P450 Monooxygenase for Alkane Oxidation. Advanced Synthesis Catalysis 343, 601-606 (2001).

328. Sun, L., Petrounia, I.P., Yagasaki, M., Bandara, G. & Arnold, F.H. Expression and stabilization of galactose oxidase in Escherichia coli by directed evolution. Protein engineering 14, 699-704 (2001).

198

329. Wang, L., Brock, A., Herberich, B. & Schultz, P.G. Expanding the genetic code of Escherichia coli. Science (New York, N.Y 292, 498-500 (2001).

330. Liu, D.R., Magliery, T.J., Pastrnak, M. & Schultz, P.G. Engineering a tRNA and aminoacyl-tRNA synthetase for the site-specific incorporation of unnatural amino acids into proteins in vivo. Proceedings of the National Academy of Sciences of the United States of America 94, 10092- 10097 (1997).

331. Jackel, C., Kast, P. & Hilvert, D. Protein design by directed evolution. Annual review of biophysics 37, 153-173 (2008).

332. Fox, R.J. et al. Improving catalytic function by ProSAR-driven enzyme evolution. Nature biotechnology 25, 338-344 (2007).

333. Flores, H. & Ellington, A.D. Increasing the thermal stability of an oligomeric protein, beta- glucuronidase. Journal of molecular biology 315, 325-337 (2002).

334. Baik, S.H., Ide, T., Yoshida, H., Kagami, O. & Harayama, S. Significantly enhanced stability of glucose dehydrogenase by directed evolution. Applied microbiology and biotechnology 61, 329- 335 (2003).

335. Waldo, G.S., Standish, B.M., Berendzen, J. & Terwilliger, T.C. Rapid protein-folding assay using green fluorescent protein. Nat Biotechnol 17, 691-695 (1999).

336. Bracha-Drori, K. et al. Detection of protein-protein interactions in plants using bimolecular fluorescence complementation. Plant J 40, 419-427 (2004).

337. Stolpe, T. et al. In planta analysis of protein-protein interactions related to light signaling by bimolecular fluorescence complementation. Protoplasma 226, 137-146 (2005).

338. Zhang, H. et al. BRCA1 physically associates with p53 and stimulates its transcriptional activity. Oncogene 16, 1713-1721 (1998).

339. Lorick, K.L. et al. RING fingers mediate ubiquitin-conjugating enzyme (E2)-dependent ubiquitination. Proceedings of the National Academy of Sciences of the United States of America 96, 11364-11369 (1999).

340. Eakin, C.M., Maccoss, M.J., Finney, G.L. & Klevit, R.E. Estrogen receptor alpha is a putative substrate for the BRCA1 ubiquitin ligase. Proceedings of the National Academy of Sciences of the United States of America 104, 5794-5799 (2007).

341. Brzovic, P.S., Meza, J., King, M.C. & Klevit, R.E. The cancer-predisposing mutation C61G disrupts homodimer formation in the NH2-terminal BRCA1 RING finger domain. The Journal of biological chemistry 273, 7795-7799 (1998).

342. Thai, T.H. et al. Mutations in the BRCA1-associated RING domain (BARD1) gene in primary breast, ovarian and uterine cancers. Human molecular genetics 7, 195-202 (1998).

343. Sauer, M.K. & Andrulis, I.L. Identification and characterization of missense alterations in the BRCA1 associated RING domain (BARD1) gene in breast and ovarian cancer. Journal of medical genetics 42, 633-638 (2005).

199

344. Li, L. et al. Oncogenic BARD1 isoforms expressed in gynecological cancers. Cancer research 67, 11876-11885 (2007).

345. Zhang, G., Gurtu, V. & Kain, S.R. An enhanced green fluorescent protein allows sensitive detection of gene transfer in mammalian cells. Biochem Biophys Res Commun 227, 707-711 (1996).

346. Crameri, A., Whitehorn, E.A., Tate, E. & Stemmer, W.P. Improved green fluorescent protein by molecular evolution using DNA shuffling. Nature biotechnology 14, 315-319 (1996).

347. Magliery, T.J. & Regan, L. A cell-based screen for function of the four-helix bundle protein Rop: a new tool for combinatorial experiments in biophysics. Protein Eng Des Sel 17, 77-83 (2004).

348. Pedelacq, J.D., Cabantous, S., Tran, T., Terwilliger, T.C. & Waldo, G.S. Engineering and characterization of a superfolder green fluorescent protein. Nature biotechnology 24, 79-88 (2006).

349. Cabantous, S., Terwilliger, T.C. & Waldo, G.S. Protein tagging and detection with engineered self-assembling fragments of green fluorescent protein. Nature biotechnology 23, 102-107 (2005).

350. Fletcher, S. & Hamilton, A.D. Protein-protein interaction inhibitors: small molecules from screening techniques. Current topics in medicinal chemistry 7, 922-927 (2007).

351. Fry, D.C. Protein-protein interactions as targets for small molecule drug discovery. Biopolymers 84, 535-552 (2006).

352. Sattler, M. et al. Structure of Bcl-xL-Bak peptide complex: recognition between regulators of apoptosis. Science (New York, N.Y 275, 983-986 (1997).

353. Joyce, G.F. Directed molecular evolution. Scientific American 267, 90-97 (1992).

354. Schimmer, A.D. Inhibitor of apoptosis proteins: translating basic knowledge into clinical practice. Cancer research 64, 7183-7190 (2004).

355. Ishi, K. & Sugawara, F. A facile method to screen inhibitors of protein-protein interactions including MDM2-p53 displayed on T7 phage. Biochemical pharmacology 75, 1743-1750 (2008).

356. Cai, Z., Greene, M.I. & Berezov, A. Modulation of biomolecular interactions with complex- binding small molecules. Methods (San Diego, Calif 46, 39-46 (2008).

357. Cadwell, R.C. & Joyce, G.F. Randomization of genes by PCR mutagenesis. PCR methods and applications 2, 28-33 (1992).

358. Costa, L.G. Current issues in organophosphate toxicology. Clinica chimica acta; international journal of clinical chemistry 366, 1-13 (2006).

359. Wills, A.M. et al. Paraoxonase 1 (PON1) organophosphate hydrolysis is not reduced in ALS. Neurology 70, 929-934 (2008).

360. Mackness, M.I., Durrington, P.N., Ayub, A. & Mackness, B. Low serum paraoxonase: a risk factor for atherosclerotic disease? Chemico-biological interactions 119-120, 389-397 (1999).

200

361. Gaidukov, L. & Tawfik, D.S. High affinity, stability, and lactonase activity of serum paraoxonase PON1 anchored on HDL with ApoA-I. Biochemistry 44, 11843-11854 (2005).

362. Bradshaw, G. et al. Facilitated replacement of Kupffer cells expressing a paraoxonase-1 transgene is essential for ameliorating atherosclerosis in mice. Proceedings of the National Academy of Sciences of the United States of America 102, 11029-11034 (2005).

363. Tward, A. et al. Decreased atherosclerotic lesion formation in human serum paraoxonase transgenic mice. Circulation 106, 484-490 (2002).

364. Jakubowski, H. Homocysteine thiolactone: metabolic origin and protein homocysteinylation in humans. The Journal of nutrition 130, 377S-381S (2000).

365. Jakubowski, H. Protein N-homocysteinylation: implications for atherosclerosis. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie 55, 443-447 (2001).

366. Lacinski, M. et al. Determinants of homocysteine-thiolactonase activity of the paraoxonase-1 (PON1) protein in humans. Cellular and molecular biology (Noisy-le-Grand, France) 50, 885-893 (2004).

367. Domagala, T.B. et al. The correlation of homocysteine-thiolactonase activity of the paraoxonase (PON1) protein with coronary heart disease status. Cellular and molecular biology (Noisy-le- Grand, France) 52, 4-10 (2006).

368. Mackness, B. et al. Low paraoxonase activity predicts coronary events in the Caerphilly Prospective Study. Circulation 107, 2775-2779 (2003).

369. Rosenblat, M. et al. The catalytic histidine dyad of high density lipoprotein-associated serum paraoxonase-1 (PON1) is essential for PON1-mediated inhibition of low density lipoprotein oxidation and stimulation of macrophage cholesterol efflux. The Journal of biological chemistry 281, 7657-7665 (2006).

370. La Du, B.N., Jr. Are we ready to try to cure alkaptonuria? American journal of human genetics 62, 765-767 (1998).

371. Furlong, C.E. et al. Human and rabbit paraoxonases: purification, cloning, sequencing, mapping and role of polymorphism in organophosphate detoxification. Chemico-biological interactions 87, 35-48 (1993).

372. Broomfield, C.A., Ford, K. W in Proceedings of the 3rd International Meeting on Cholinesterases, La Grande-Motte 1611991).

373. Dunn, M.A. & Sidell, F.R. Progress in medical defense against nerve agents. Jama 262, 649-652 (1989).

374. Doctor BP, R.L., Wolfe AD, Maxwell DM, Ashani Y Enzymes as pretreatment drugs for organophosphate toxicity. Neuroscience and biobehavioral reviews 15, 123-128 (1991).

375. Yeung, D.T. et al. Structure/function analyses of human serum paraoxonase (HuPON1) mutants designed from a DFPase-like homology model. Biochimica et biophysica acta 1702, 67-77 (2004).

201

376. Gencer, N. & Arslan, O. Purification human PON1Q192 and PON1R192 isoenzymes by hydrophobic interaction chromatography and investigation of the inhibition by metals. Journal of chromatography 877, 134-140 (2009).

377. Nishihara, K., Kanemori, M., Yanagi, H. & Yura, T. Overexpression of trigger factor prevents aggregation of recombinant proteins in Escherichia coli. Applied and environmental microbiology 66, 884-889 (2000).

378. Thomas, J.G., Ayling, A. & Baneyx, F. Molecular chaperones, folding catalysts, and the recovery of active recombinant proteins from E. coli. To fold or to refold. Applied biochemistry and biotechnology 66, 197-238 (1997).

379. Xu, H.M. et al. Expression of soluble, biologically active recombinant human endostatin in Escherichia coli. Protein expression and purification 41, 252-258 (2005).

380. Rochu, D. et al. Stabilization of the active form(s) of human paraoxonase by human phosphate- binding protein. Biochemical Society transactions 35, 1616-1620 (2007).

381. Herman, A. & Tawfik, D.S. Incorporating Synthetic Oligonucleotides via Gene Reassembly (ISOR): a versatile tool for generating targeted libraries. Protein Eng Des Sel 20, 219-226 (2007).

382. Tavassoli, A. & Benkovic, S.J. Split-intein mediated circular ligation used in the synthesis of cyclic peptide libraries in E. coli. Nature protocols 2, 1126-1133 (2007).

202