Investigating the Conformational Dynamics of HIV-1 Protease through Molecular Dynamics

MISSING IMAGE

Material Information

Title:
Investigating the Conformational Dynamics of HIV-1 Protease through Molecular Dynamics
Physical Description:
1 online resource (146 p.)
Language:
english
Creator:
McGee, Terry Dwight
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemistry
Committee Chair:
ROITBERG,ADRIAN E
Committee Co-Chair:
HORENSTEIN,NICOLE ALANA
Committee Members:
BOWERS,CLIFFORD RUSSELL
FANUCCI,GAIL E
DUNN,BEN M

Subjects

Subjects / Keywords:
dynamics -- hiv -- molecular -- protease
Chemistry -- Dissertations, Academic -- UF
Genre:
Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Human Immunodeficiency Virus (HIV), the virus that causes Acquired Immunodeficiency Syndrome (AIDS), is still one of the most prominent diseases in the world. One of the main targets in anti-retroviral therapy isthe protease because of its role in the life cycle of the virus and inhibition would prevent the maturation and the spread of the virus to neighboring cells. HIV-1 Protease (HIVPR) has two aspartates in active site that are responsible for its catalytic ability. The protonation of the aspartates in the apo and bound forms of the protease has been topic of much debate. Using Constant pH Molecular Dynamics we investigate the protonation of the catalytic aspartates. The mechanism of how drug pressure selected mutations,especially those not located in the active site, confer resistance is not well understood. Through our results we provide an explanation as to how they reduce the efficacy of protease inhibitors. The results of this work could offer valuable insight on ways to improved protease inhibitors.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Terry Dwight McGee.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: ROITBERG,ADRIAN E.
Local:
Co-adviser: HORENSTEIN,NICOLE ALANA.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2014-06-30

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0046101:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 INVESTIGATING THE CONFORMATIONAL DYNAMICS OF HIV 1 PROTEASE THROUGH MOLECULAR DYNAMICS By TERRY DWIGHT MCGEE JR. A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE R EQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Terry Dwight McGee Jr.

PAGE 3

3 To J.P an d Nora Williams

PAGE 4

4 ACKNOWLEDGMENTS It has been through the love and sacrifice of othe rs that I have been able to obtain so much in my life. I would like to commend Dr. Roitberg for his continuous efforts to make sure that I was equipped with the necessary tools to succeed. I appreciate most of all the patience and understanding that Dr. Ro itberg has shown me, which has greatly helped to nurture my development. I would like to acknowledge my committee who has not only been instrumental in my development as a scientist but as a person as well. I appreciate your patience and thoughtful advice I am grateful to Dr. Dunn for his guidance and the opportunity to wor k on a project of such magnitude. I appreciate Dr. Fanucci for your advice and positive reassurance of my work and progress. I would also like to extend my thanks to Dr. Bowers for chal lenging me in and out of the classroom and for your willingness to see me succeed. Dr. Horenstein, I appreciate how your classes strengthen my knowledge of biochemistry, which has had a profound impact on my research. I am indebted to Dr. Edwards for his i nvaluable advice and inspirational messages. You have been a valuable mentor. I would like to acknowledge my parents Terry and Sylvia McGee who have been unbelievably understanding and encouraging in my pursuit of a higher education. Throughout my life, my parents have been a continuous stream of motivation and wisdom. I would like to thank the Roitberg Group for all their support, especially Natali Di Russo Finally, I wou ld like to acknowledge my wife Danielle McGee who has been nothing but supportive thr oughout this entire process. Thank you for being so understanding on the demands, which graduate school places on my time and life.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 1.1 HIV Worldwide Epidemic ................................ ................................ ............... 17 1.2 HIV 1 Genome ................................ ................................ .............................. 17 1.3 Genetic Variability ................................ ................................ ......................... 19 1.4 HIV Life Cycle ................................ ................................ ............................... 20 1.5 HIV 1 Protease ................................ ................................ .............................. 22 1.6 HIV Protease as a Potential Drug Target ................................ ...................... 27 1.7 Protease Inhibitors ................................ ................................ ........................ 28 1.8 Drug Resistance ................................ ................................ ............................ 31 1.9 General Overview ................................ ................................ ......................... 34 2 METHODS ................................ ................................ ................................ .............. 36 2.1 Molecular Dynamics ................................ ................................ ...................... 36 2.1.1 Molecular Dynamics Integrator ................................ ........................... 37 2.1.2 Langevin Dynamics ................................ ................................ ............. 38 2.2 Molecular Mechanics ................................ ................................ ........................ 39 2.3 Solvent Models ................................ ................................ .............................. 40 2.3.1 Explicit Solvent ................................ ................................ .................... 41 2.3.2 Implicit So lvent Models ................................ ................................ ....... 42 2.3.2.1 Poisson Boltzmann solvent model ................................ ............... 42 2.3.2.2 Generalized Born solvent model ................................ .................. 43 2.4 Steered Molecular Dynamics ................................ ................................ ........ 44 2.5 Constant pH Methods ................................ ................................ ................... 45 2.5.1 Constant pH Molecular Dynamic s ................................ ....................... 45 2.5.2 Constant pH Replica Exchange Molecular Dynamics ......................... 46 2.6 Accelerated Molecular Dynamics ................................ ................................ .. 47 2.7 Docking ................................ ................................ ................................ ......... 49 2.7 Molecular Mechanics Poisson Boltzmann Surface Area ............................... 50

PAGE 6

6 3 MOLECULAR DYNAMICS SIM ULATIONS INDICATE THAT THE CATALYTIC DYAD OF HIV 1 PROTEASE IS MONO PROTONATED ................................ ....... 52 3.1 Introduction ................................ ................................ ................................ ... 52 3.2 Methods ................................ ................................ ................................ ........ 54 3.2.1 General Setup ................................ ................................ ..................... 54 3.2.2 Constant pH Molecular Dynamics Protocol ................................ ......... 55 3.2.3 Co nstant pH Replica Exchange Molecular Dynamics Protocol ........... 56 3.2.4 General Analysis ................................ ................................ ................. 57 3.3 Results ................................ ................................ ................................ .......... 57 3.3.1 Constant pH Molecular Dynamics ................................ ....................... 57 3.3.2 Constant pH Replica Exchange Molecular Dynamics ......................... 58 3 .3.4 Location of the proton in the mono protonated state ........................... 59 3.4 Discussion ................................ ................................ ................................ ..... 61 3.4.1 Apo (Free enzyme) ................................ ................................ ............. 61 3.4.2 Cyclic Urea Inhibitors ................................ ................................ .......... 61 3.4.3 Neutral Asymmetric Inhibitors ................................ ............................. 62 3.4.4 Cationic Inhibi tor ................................ ................................ ................. 63 3.4.5 Low Barrier Hydrogen Bond ................................ ................................ 64 3.5 Conclusions ................................ ................................ ................................ ... 64 4 UNDE RSTANDING HOW THE G48T/L89M MUTATIONS REDUCE THE EFFICACY OF THE HIV 1 PROTEASE INHIBITOR SAQUINAVIR ....................... 66 4.1 Introduction ................................ ................................ ................................ ... 66 4.2 Me thods ................................ ................................ ................................ ........ 68 4.2.1 General Setup ................................ ................................ ..................... 68 4.2.2 Steered Molecular Dynamics ................................ .............................. 68 4.2.3 Molecular Mechanics Generalized Born Surface Area Calculations ... 69 4.2.4 General Analysis ................................ ................................ ................. 69 4.3 Results and Discussions ................................ ................................ ............... 69 4.3.1 Inhibitor Simulations ................................ ................................ ............ 69 4.3.1.1 Flexibility ................................ ................................ ...................... 69 4.3.1.2 Hyd rogen bonding analysis ................................ .......................... 73 4.3.1.3 MMGBSA calculations ................................ ................................ 74 4.3.1.4 Steered Molecular Dynamics ................................ ....................... 76 4.3.2 Apo Simulations ................................ ................................ .................. 78 4.3.3 Drug Resistant Mechanism ................................ ................................ 80 4.4 Conclusions ................................ ................................ ................................ ... 82 5 EVALUATING THE EFFECTIVENESS OF SAQUINAVIR AND TIPRANAVIR ON HIV 1 SUBTYPE C WILD TYPE AND THE DRUG RESISTANT N88D/L90M MUTANT PROTEASE ................................ ................................ ............................ 84 5.1 Introduct ion ................................ ................................ ................................ ....... 84 5.2 Methods ................................ ................................ ................................ ........ 86 5.2.1 General Setup ................................ ................................ ..................... 86

PAGE 7

7 5.2.3 Molecular Dynami cs ................................ ................................ ............ 86 5.2.3 Molecular Mechanics Generalized Born Surface Area Calculations ... 87 5.3 Results and Discussion ................................ ................................ ................. 87 5.3.1 Saquinavir bound MD Simulations ................................ ...................... 87 5.3.1.1 Flexibility ................................ ................................ ...................... 8 7 5.3.1.2 Hydrogen bonding analysis ................................ .......................... 90 5.3.1.3 MMGBSA calculations ................................ ................................ 91 5.3.2 Tipranavir bound MD Simulations ................................ ....................... 93 5.3.2.1 Flexibility ................................ ................................ ...................... 93 5.3.2.2 Hydrogen bonding analysis ................................ .......................... 96 5.3.2.3 MMGBSA calculations ................................ ................................ 97 5.3.3 Saquinavir vs. Tipranavir ................................ ................................ ..... 99 5.3.4 Drug Resistant Mechanism ................................ ................................ 99 5.5 Conclusions ................................ ................................ ................................ .... 101 6 INVESTIGATING WHAT EFFECTS THE STABILIZING MUTATIONS HAVE ON THE EPR MEASURED DISTANCE PROFILES ................................ ................... 103 6.1 Introduction ................................ ................................ ................................ 103 6.2 Methods ................................ ................................ ................................ ...... 105 6.2.1 General Setup ................................ ................................ ................... 105 6.2.2 MTSL Parameters ................................ ................................ ............. 106 6.2.3 Molecular Dynamics ................................ ................................ .......... 107 6.3 Results and Discussions ................................ ................................ ............. 107 6.3.1 Inhibit or bound MD simulations ................................ ......................... 107 6.3.2 Apo MD Simulations ................................ ................................ .......... 110 6.3.3 Comparing the N ................................ ......... 113 6.4 Conclusions ................................ ................................ ................................ 114 7 PRELIMINARY STUDIES: EVALUATING THE EFFECTIVENESS OF ANTIRETROVIRAL PROTEASE INHIBITORS ON XMRV ................................ ... 116 7.1 Introduction ................................ ................................ ................................ 116 7.2 Methods ................................ ................................ ................................ ...... 118 7.2.1 Completing XMRV ................................ ................................ ............. 118 7.2.1 General Setup ................................ ................................ ................... 118 7.2.2 Molecular Dynamics ................................ ................................ .......... 119 7.2.3 Molecular Mechanics Poisson Boltzman Surface Area Calculations 119 7.3 Results and Discussion ................................ ................................ ............... 120 7.3.1 Flexibility ................................ ................................ ........................... 120 7.3.2 Hydrogen Bonding Analysis ................................ .............................. 122 7.3.3 Molecular Mechanics Poisson Boltzmann Surface Area Calculations ................................ ................................ ................................ 124 7.3.4 Potent ial Inhibitor Interactions ................................ ........................... 125 7.4 Conclusions and Future Work ................................ ................................ ..... 126 8 CONCLUSIONS AND FUTURE WORK ................................ ............................... 128

PAGE 8

8 APPENDIX: HIV 1 PROTEASE SEQUENCES ................................ ......................... 130 LIST OF REFERENCES ................................ ................................ ............................. 131 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 146

PAGE 9

9 LIST OF TABLES Table page 1 1 The twelve different substrates that HIV 1 protease must cleave in order to produce an infectious virion. ................................ ................................ ............... 27 4 1 Hydrogen bonding analysis of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir. ................................ ................................ ......... 73 4 2 Results of MMGBSA calculations f or the wild type and the mutant (G48T/L89M), subtype B, bound to s aquinavir. ................................ .................. 75 5 1 Hydrogen bonding analysis of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir ................................ ................................ .......... 91 5 2 Results of the MMGBSA calculations for the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. All units are in kcal/mol ............ 92 5 3 Hydrogen bonding analysis of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir. ................................ ................................ ........... 96 5 4 Results of the MMGBSA calculations for the wild type and the mutant (N88D/L90M), subtype C, bound to t ipranavir. ................................ ................... 97 6 1 Distance profiles of inhibitor bound simulations ................................ ................ 109 6 2 Distance profiles of the apo MD simulations ................................ ..................... 112 7 1 Hydrogen bond analysis XMRV (amprenavir) ................................ ................... 123 7 2 Hydrogen bond analysis XMRV (pepstatin A) ................................ .................. 123 7 3 Hydrogen bond analysis XMRV (TL 3) ................................ ............................. 123 7 4 MMPBSA results for the inhibitors: amprenavir, pepstatin A, and TL 3. ........... 125 A 1 HIV 1 Protease Subtype B Sequen ce (Wild type). ................................ ........... 130 A 2 HIV 1 Protease Subtype B S ................................ .................... 130 A 3 HIV 1 Protease Subtype C Sequenc e (Wild type). ................................ ........... 130 A 4 HIV 1 Protease MDR769 Sequence (Drug Resist ant Construct). ..................... 130 A 5 HIV 1 Protease V6 Sequence (Drug Reista nt Construct). ................................ 130

PAGE 10

10 LIST OF FIGURE S Figure page 1 1 Structu re of mature HIV virion. ................................ ................................ .......... 18 1 2 Schematic diagram of the HIV 1 Genome. ................................ ......................... 19 1 3 Genetic classification of HIV. HIV is classified in types, groups, subtypes/cla des and circulating recombinant forms. ................................ .......... 20 1 4 Schematic of t he HIV viral life cycle. ................................ ................................ .. 21 1 5 Schematic representation of HIV 1 pro tea se ................................ ...................... 22 1 6 HIV 1 protease. ................................ ................................ ................................ ........... 23 1 7 Flap confo rmations of HIV 1 protease. ................................ ............................... 24 1 8 General Acid/Base mechanism of HIV 1 protease ................................ ............. 26 1 9 Transition state mimet ics ................................ ................................ .................... 29 1 10 Regions targeted by non active site inhib itors ................................ .................... 31 1 11 Per residue mutation frequency of protease inhibitor nave patients and patients who have been exposed to protease inhibitors (subtypes A, B, C, D, G, AE and AG.) ................................ ................................ ................................ ... 32 1 12 Schematic of HIV 1 protease illustrating the location of the primary and secondary mutations. ................................ ................................ ......................... 33 2 1 Schematic representation of a TIP3P water molecule ................................ ........ 42 2 3 Schematic representation o f REMD simulations. ................................ ............... 47 2 4 Schematic representation of the normal and biased potential of AMD simulations. ................................ ................................ ................................ ........ 49 2 5 Thermodynamic cycle of MMPBS A. ................................ ................................ .. 51 3 1 2D schematic of the protease inhibitors used in the Constant pH Molecular Dynamic Simulations ................................ ................................ ......................... 53 3 2 Plot of Average total molecular protonation versus time for HIVP R complex KNI 272. CpHMD performed at pH 5 ................................ ................................ 58 3 3 Schematic of the four different possible mono pro tonated protonation states .... 60

PAGE 11

11 4 1 HIV 1 protease, the colored spheres indicate locations on the protease the where wild type and mutant differ. Position 48 (G48T) is colored yellow and position 89 (L89M) is colored orange. The catalytic aspartates are also depicted. ................................ ................................ ................................ ............. 66 4 2 Schematic representation of the HIV 1 protease inhibitor saquinavir. ................ 67 4 3 Atomic Fluctuations of the wild type and the mutant (G48T/L89M), s ubtype B, bound t o saquinavir. ................................ ................................ .......................... 70 4 4 Histogram of the RMSD of the C 58) of each monomer MD simulations of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir. ................................ ................................ ......... 71 4 5 Histogram of the RMSD of saquinavir. MD simulations of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir. ................................ 71 4 6 Histograms of the different distances in the active site of the wild type and the mutant (G48T/L89M), subt ype B, bound to saquinavir ................................ 72 4 7 Per residue en ergy decomposition of the binding affinity of the wild type and the mutant (G48T/L89M), subtype B bound to saquinav ir. ................................ 75 4 8 Work Profiles from the SMD simulations for wild type and the mutan t (G48T/L89M), subtype B, bound to saquinavir. ................................ .................. 77 4 9 Free energy profile of removing the protease inhibitor saquinavir from the active site cavity of the wild type and mutant (G48T/L89M), subt ype B. ............ 78 4 10 Plot of the RMSD of wild type subtype B, flap residues, 43 58, versus the closed, semi open and wide open conformations. MD simulations performed in the apo form. ................................ ................................ ................................ ... 78 4 11 Plot of the RMSD of mutant (G48T/L89M) subtype B, flap residues, 43 58, versus the closed, semi open and wide open conformations. MD simulations performed in the apo form. ................................ ................................ ................. 79 4 12 Atomic Fluctuations of the wild type and the mutant (G48T/L89M ), subtype B, ................................ ................................ .......................... 80 4 13 Schematic of the Phe53 pi pi stacking interaction wit h the qui noline ring of saquinavir. ................................ ................................ ................................ ......... 81 5 1 Schematic of HIVPR subtype C ................................ ................................ .......... 84 5 2 Schematic of the protease inhibitors saquinavir an d tipranavir. .......................... 85

PAGE 12

12 5 3 Atomic Fluctuations of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. ................................ ................................ ....................... 88 5 4 Histo gram of the RMSD of saquinavir. MD simulations of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. ................................ 89 5 5 Histogram of the RMSD of the flap residues (43 58) of each mon omer. MD simulations of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. ................................ ................................ ................................ ...... 89 5 6 Histograms of the different distances in the active site of the wild type and the mutant (N88D/L90M), subtype C, bound to s aquinavir. ................................ 90 5 7 Per residue energy decomposition of the binding affinity of the wild type and the mutant (N88D/L90M), subtype C bou nd to saqunavir ................................ ... 92 5 8 Atomic Fluctuations of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavi r. ................................ ................................ ....................... 94 5 9 Histogram of the RMSD of ti pranavir. MD simulations of the wild type and mutant (N88D/L90M), subtype C, bound to tipranavir. ................................ ....... 94 5 10 Histogram of the RMSD of the flap residue (43 58) of each monomer. MD simulations of th e wild type and mutant (N88D/L90M), subtype C, bound to tipranavir. ................................ ................................ ................................ ............ 95 5 11 Histograms of the different distances in the active site of the wild type and mutant (N88D/L90M), subtype C, bound to ti pranavir. ................................ ....... 95 5 12 Per residue energy decomposition of the binding affinity of the wild type and the mutant (N88D/L90M), subtype C bound to tipranavir. ................................ 98 5 13 Schematic diagram of the Asp88/Asn88 interactions with Thr31 and Thr74 A) wild type B) m utant (N88D/L90M). ................................ ................................ .. 100 6 1 Schematic of HIV 1 protease. Highlighted are the spin labels at position 55 of each monom er. ................................ ................................ ................................ 104 6 2 Schematic of the five torsional angles of the spin label MTSL .......................... 106 6 3 Hi stogram s of the RMSD of HIV 1 protease subtype B and the protea se inhibitor bound ................................ ................................ ................................ 108 6 4 Histogram of N N distance distribution profiles of apo form HIVPR subtype B compared to when it is bou nd to protease inhibitors. ................................ .... 110 6 5 Plot of the RMSF of the wild type, subtype B (black), MDR769 (red), V6 (blue). ................................ ................................ ................................ ............. 11 1

PAGE 13

13 6 6 Histog ram of the N N distance profiles. Wild type subtype B, (black), MDR769 (red), V6 (blue) ................................ ................................ .................. 112 6 7 Schematic of the average structures of HIVPR: wild type subtype B (black), MDR769 (red), and V6 ( blue) from spin labeled MD simulations. ..................... 113 6 8 Histogram of the 55 55 C type subtype B (black), MD769 (red), V6 (blue) ................................ ................................ ..................... 114 7 1 Schematic of XMRV protease. Highlighted are the catalytic aspartates ........... 117 7 2 Protease inhibitors: Amprenavir (HIV 1 protease), Pepstatin A (generic inhibitor of aspartic proteases), TL 3 (FIV protease) ................................ ........ 117 7 3 Plot of the RMSF of XMRV bound to amprenavir, pepstatin A ......................... 120 7 4 Histogram of the RMSD of the protease inhibitors: amprenavir, pepstatin A and TL 3. ................................ ................................ ................................ .......... 121 7 5 Histo gram of the RMSD of the flap residues 48 66 of each monomer. ............. 122 7 6 Schematic representation of XMRV protease. Represented by spheres are the residues in the active site that potential inhibit ors can ta rget. ..................... 126

PAGE 14

14 LIST OF ABBREVIATIONS AIDS Acquired Immunodeficiency Syndrome aMD Accelerated Molecular Dynamics APV Amprenavir ASN Asparagine ASP Aspartic Acid ATZ Atazanavir cpHMD Constant pH Molecular Dynamics CR F Circular Recombinant Form DEER Double Electron Electron Resonance DRV Darunavir EPR Electron Paramagnetic Resonance GB Generalized Born GLY Glycine FDA Federal Drug Administration FIV Feline Immunodeficiency Virus HIV Human Immunodefi ci ency Virus HIVPR H uman Immunodeficiency Virus Protease IDV Indinavir ILE Isoleucine LPV Lopinavir MD Molecular Dynamics MET Methionine MM Molecular Mechanics MMPBSA Molecular Mechanics Poisson Boltzmann Surface Area

PAGE 15

15 MMGBSA Molecular Mechanics Genera lized Born Surface Area M TSL (1 Oxyl 2,2,5,5 Tetramethly Pyrroline 3 Methyl) Methanethiosulfonate NFV Nelfinavir NMR Nuclear Magnetic Resonance PDB Protein Data Bank PB Poisson Boltzmann pH REMD pH Replica Exchange Molecular Dynamics RMSD Root Mean Squared Deviation RMSF Root M ean Squared Fluctuations RTV Ritonavir SASA Solvent Accessible Surface Area SDSL Site Directed Spin Label SQV Saquinavir THR Threonine TPV Tipranavir WHO World Health Organization UNAIDS Joint United Nations on HIV/AIDS VAL Valine XMRV Xenotropic Murine Le ukemia Virus Related Virus

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INVESTIGATING THE CONFORMATIONAL DYNAMIC S OF HIV 1 PROTEASE THROUGH MOLECULAR DYNAMICS By Terry Dwight McGee Jr. December 2013 Chair: Adrian E. Roitberg Major: Chemistry Human Immunodeficiency Virus (HIV) the virus that causes Acquired I mmunodeficiency Syndrome (AIDS), is still one of the mo st prominent diseases in the world. One of the main targets in anti retroviral therapy is the protease because of its role in the life cycle of the virus and inhibition would prevent the maturati on and the spread of the virus to neighboring cells. HIV 1 Pr otease (HIVPR) has two a spartates in the active site that are responsible for its catalytic ability. The protonation of the aspartates in the apo and bound forms of the protease has been topic of much debate. Using Constant pH M olecular Dynamics we investi gate the protonation of the catalytic aspartates. The mechanism of how drug pressure selected mutations, especially those not located in the active site, confer resistance is not well understood. Through our results we provide an explanation as to how the y reduce the efficacy of protease inhibitors. The results of this work could offer valuable insight on ways to improved protease inhibitors.

PAGE 17

17 CHAPTER 1 INTRODUCTION 1.1 HIV W orldwide Epidemic Hum an Immunodeficiency Virus (HIV) is the causative agent of Acqu ired Immunodeficiency Disease Syndrome (AIDS) 1 As of 2011, the Joi nt United Nations P rogram on HIV/AIDS (UNAIDS) has projected that there are approximately 34 million peopl e living with HIV. To date there is no cure or vaccine for HIV, however recent advanc es in antiretroviral therapy have improved the life expectancy o f those infected. Consequently, the number of AIDS related deaths has decreased from 2.3 million in 2005 to 1.7 million in 2011 according to the World Health Organ ization (WHO). A majority of those deaths were in Sub Saharan Africa this startling fact can be attributed to the inabilities of these countries to afford antiretroviral t reatment, social stigmatism, etc. There has been a decline in new HIV cases because only 2.5 milli on people were newly infected with HIV in 2011, which is down from 3.2 million i n 2001 according to data provided by t o the UNAIDS 1.2 HIV 1 Genome HIV is a retrovirus and is further classified as a lentivirus due to its long incubation period. HIV is approximately 120 nm in diameter and possesses a spherical shape Figure 1 1. 2 It is comprised of two copies of single stranded RNA that is encapsulated by the viral protein p24 Its genome Figure 1 2, is roughl y 9.4 Kb in length. The group specific antigen ( gag ) gene encodes the stru ctural proteins matrix (MA) capsid (CA), and nucleocapsid (NC) 3 The envelo pe proteins g41 and gp120, are responsible for the viral tropism and are encoded by the envelope ( env ) gene. The polymerase ( pol ) gene encodes for the three enzymes: reverse transcriptase, integrase

PAGE 18

18 and protease. The other genes rev nef vpr vif, vpu and tat encode for the six accessory proteins. Nef has two roles: decreasing the number of CD4 receptors locat ed on the surface of infected lymphocytes and enhancing viral replication. 4 5 Rev is responsible for controlling the expression of HIV proteins by governing t he rate at which mRNAs are exported from the nucleu s to cytoplasm. 6 The Tat enhances transcription of integrated proviral DNA 7 Vpr plays a critical regulatory role during replication in the early stages of the life cycle of the virus. 8 9 The exact function of Vif is not known but has been impli cated to be a factor in the viral replication. 10 11 HIV str ains that have mutations in Vif do not replicate as efficiently and have been linked to low viral loads when compared to the wild type strains. 10 12 Vpu governs the release of the virus from cell surface of the infected host and degradation of CD4 in the endoplasmic reticulum. 10 13 Figure 1 1. Structure of mature HIV virion. Image courtesy of National Institute of Allergy and Infectious Diseases.

PAGE 19

19 Figure 1 2. Schematic diagram of the HIV 1 Genome. 1.3 Genetic Variability HIV has a high genetic variability Figure 1 3 and it can be divided into two distinct types: HIV 1 and HIV 2. HIV 1, which accounts for the majority of infections can M/non the newly disco vered group, P 14 17 Group M is the larges t and is composed of subtypes or clades, A D, F H, J, K and circulating recombinant forms. The subtype relationships and classifications were constructed by performing sequence alignments of the gag, and env regions. 14 18 19 Subtype A is predominantly in West and Central Africa. Subtype B is generally found in North and South Americas, Europe and Australia. Subtype C is essentially in Southern and East Africa and throughout Easte rn Asia. Subtype D is limited mainly to Central Africa. Subtype F is primarily in South America. Subtype H is mostly in Central Africa and subtype J in Central America. Lastly, subtype K is mainly in the countries Congo and Cameroon. Persons infected with HI V are generally infected with subt ypes A D. Moreover, subtype C accounts for the highest percentages of those with HIV. 20 However, Subtype D has been cr edited with being the deadliest and th ose infected with this subtype

PAGE 20

20 develop AIDS at a much faster rate and have a higher mortality rate when compared to the other subtypes. 21 Figure 1 3. Genetic classification of HIV. HIV is classified in types, groups, subtypes/clades and circulating recombinant forms. 1.4 HIV Life Cycle In general, t he surface glycoproteins of enveloped viruses play a critical role in viral invasion of host cells. The HIV envelope glycoprotein (gp) consists of two non covalent subunits gp120 and gp41. First, gp120 is responsible for viral tropism by binding to target cell receptors and gp41 directs fusion betwee n cellular and viral membranes. 22 23 Gp1 20 binds to the CD4 receptor and the CCR5 or CXCR4 co receptors, which are members of the chemokine family. 22 24 Macrophage (M tropic) strains o f HIV 1 use the CD4 and CCR5 receptors to invade the host cell. 25 In contrast, T tropic strains use CD4 and CXCR4 receptors to gain entry into the cell. 25 In addition to CCR5 and CXCR4, other co receptors have been identified that the virus may use to invade a host cell. 26,27 These receptors are present in such cells as T lymphocytes and macrophages. Gp 41 undergoes a conf ormational change to a fusion active state when gp120 binds to the host cell receptor. 28 This conformational change allows the virus to inject its

PAGE 21

21 viral genome into the host through fusion. 28 Soon after entry, the process of reverse transcription occurs the enzyme reverse transcriptase transcribes a compl ementary, double stranded DNA from a single stranded RNA genome. The process of reverse transcription is prone to error. The high frequenc y of mutation s is a direct result of having an average error of 3 x 10 5 per nucleotide base per cycle. 29 Figure 1 4. Schematic of the HIV viral life cycle. Image courtesy of National Institute of Allergy and Infectious Diseases. The newly formed DNA is transported to the nucleus of the cell wh ere it is integrated into the host DNA by the enzyme, integrase. Inclusion of the virus genetic material into the host cell genome results in the viral DNA becoming a provirus. 30 The virus can remain dormant for years in this state. 31 Upon activation 32 the provirus uses the host enzyme RNA polymerase to transcr ibe the viral mRNA. The mRNA must be

PAGE 22

22 exported from the nucleus and remain intact the rev protein binds to the un spliced mRNA and transports it from the nucleus. 30,33 The last and final steps are assembly and budding. The protease enzyme is responsible for cleaving the polypeptide chain into individual proteins, which promotes assembly and is vital for virus maturation. 34,35 The newly formed virus buds from the host cell and can now infect a new cell. 1.5 HIV 1 Protease A protea se is an enzyme that directs proteolysis and cleaves the peptide bonds that link amino acids to form polypeptide chains. Proteases are divided into several groups: serine, threonine, aspartyl, cysteine, metallo and glutamic. 10 HIV protease Figure 1 5 is classified as an aspartyl protease because it has two aspartic residues in the a ctive site and substitution of these aspartic residues leads to a loss of catalytic ability. 36 Fi gure 1 5. Sch ematic representation of HIV 1 p rotease, PDB 1HHP. Flap residues 43 58, flap elbow residues 37 42, active site residues 25 27, cantilever 59 75, fulcrum residues 10 23, dimer interface residues 1 4, and 96 99 The three dimensional structure of HIV protease Protein Data Bank (PDB ID 2HVP 37 3HVP 38 4HVP 39 ) was not determined until 1989. To date, hundreds of crystal

PAGE 23

23 structures of HIVPR in vario us forms have been solved. HIV 1 Protease (HIVPR) is composed of two identical monomers each containing 99 amino acid residues that give stranded d C terminus strands. 38 Figure 1 6. e site of HIV 1 protease. A) Front view B) Side view. Leu24 colored yellow, Asp25 colored gray, Thr26 colored orange, Gly27 colored green. The active site of HIVPR is highly conserved it is located at the interface of the two monomers where each monomer contributes the sequence Asp25 Thr26 Gly27. The three residue motif is referred to as the catalytic triad. The active site is held together, through a hydrogen Figure 1 6. 40 The amide group of Thr 26 donates a hydrogen bond to the Thr26 side chain hydroxyl group located on the opposing monomer Thr26 donates another hydrogen bond to the carbonyl group of Leu24 of the opposite monomer. Previous studies have shown that

PAGE 24

24 Thr26 has an important role in ma intaining the dimer complex the mutation of Thr26 would result in the monomer being favored as opposed to the dimer. 41,42 The active site is co vered by t hairpins referred to as flaps (residues 43 58) The flaps undergo large conformational changes of backbone phi and psi angles in order for a substrate or inhibitor to access the active site. 43 The flaps are in a closed conformation when a n inhibitor/substrate is bound the flaps are stabilized by a water molecule (WAT301) that form s a hydrogen bond with the substrate/inhibitor and amide group of Ile50 The reaction pathways of substrate binding and product released have been investigated using umbrella sampling 44 and coarse grained molecular dynamics 45 Trylska et al 45 sugge sted that the flaps do not need to open in order for the products to be released the products can instead slide out from the binding cleft to the sides of the enzyme. Figure 1 7. Flap conformations of HIV 1 protease A) Closed conformation (orange) B) Se mi open conformation (blue), C) Tucked/Curled conformation (purple) D ) Wide open conformation (red)

PAGE 25

25 In the apo form of HIVPR, the flaps are either in the closed, wide open, tucked/curled, or semi open conformation Figure 1 7 Ishima et al. 43 presented a working model of the flap opening mechanism for the free HIVPR that requires the flaps to transition from semi open conformation in order to reach the wide open conformation. Access to the active site even in the semi open conformation is still blocked by the flaps and does not allow adequate space for a substrate/ligand to bind. Flap openings as large as 32 angstroms, measuring distance between the Ile50 residue of each monomer, have been observed in computational studies performed on HIVPR. 46 NMR as well as MD studie s have concluded that the flaps are predominantly in a semi open conformation in the apo form HIVPR 46,47 when flaps transition between the semi open to closed conformations. NMR studies have indic ated that the entire flaps of the apo protease m ove on the 100 s time scale the flap tips (residues 49 52) undergo many rapid fluctuations on the sub nanosecond time scale. 43 The increase d flexibility of the flap tips when compared to the entire flap is attributed to the flap tips glycine rich sequence. A 10 ns MD study performed by Scott and Schi f fer 48 on HIVPR, suggest ed that the curling of the flap tips et al. 49 claim the conclusions of Scott and Schi f fer 48 could be a result of improper equilibr ation of the flaps Li et al. 50 concluded that the sub nanosecond local motion of the flap tips were essential to the global conformational changes of the flap tips, thus confirming Scott and Schifer 48 previous notion. The HIV protease c leaves its substrate through a general acid/base mechanism. HIVPR must perform twelve proteolytic reactions in order to produce a mature virion. 51

PAGE 26

26 Kinetic studies perform ed on HIVPR with its substrates observed a bell shape curve when the rate as a function of pH was measured, indicating that one aspartate acts as an acid and the other as a base. 52,53 The mechanism proposed by Meeks et al. 52 Figure 1 The n ewly formed nucleophile is very potent and attacks the carbonyl of the substrate. The carbonyl oxygen of the substrate is stabilized by accepting a proton from the Asp25 residue (acting as an acid). The Asp25 accepts proton from the oxygen as it begins to reform the double bond with the carbon. Simultaneously, the carbonyl is reformed as the N C bond begins to elongate and eventually break as the nitrogen accepts a proton Fig ure 1 8. General Acid/Base mechanism of HIV 1 protease

PAGE 27

27 Table 1 1. The t welve different substrate s that HIV 1 protease must cleave in order to produce an infectious virion. The substrate. HXB2 reference Locatio n of Proteolytic Reactions Amino Acid Sequence gag polyprotein cleavage sites MA/CA VSQNY / PIVQN CA/p2 KARVL / AEAMS p2/NC TSAIM / MQRGN NC/p1 ERQAN / FLGKI p1/p6 gag RPGNF / LQSRP gag pol polyprotein cleavage sites NC/TFP ERQAN / FLREN TFP/p6 pol ENLAF / LQGKA p6 pol /PR VSFNF / LQGKA PR/RTp51 CTLNF / PISPI RT/RT/p66 GAETF / YVDGA RTp66/INT IRKVL / FLDGI Nef protein cleavage sites Nef AACAW / LEAQE 1.6 HIV Protease a s a Potential Drug Target First generation antiretroviral drug therapy involved developing inhibitors for reverse transcriptase. The efficacy of r everse transcriptase inhibitors was proven to be limited due to toxicities caused by interference with the hu man cell metabolism. 54 This ignited the need to search for other alternative drug targets. It was not until the m id 1980s that momentum for the development of HIVPR i nhibitors began to take form. Several landmark studies, revealed vital information that helped solidify the protease as a promising drug target. When the complete nucleotide sequence was determined, the results suggest that HIV encoded a protease 55 a

PAGE 28

28 mechanistic relationship of HIVPR with other as partic proteases had been proposed 56 Kramer et al. 57 were the first to show that protease was essential in cleaving the gag and gag pol poly proteins for the HIV virus specifically Sanchez Pescador et al. 58 determined two cleavage sites of HIVPR in the gag protein (Met/Met and Tyr/Pro) Because HIVPR cleaves bonds between Tyr/Phe and Pro residues and no other human enzyme s show this specificity this affords an opportunity to develop protease inhibitors that are very selective and nontoxic to the host cell 1.7 Protease Inhibitors Since HIVPR offered a promising new drug target in the antiretroviral drug therapy, the next issue that had to be addressed was how to design an inhibitor. Only competi tive inhibitors were considered the inhibitor compete s with the substrate for binding to the active site It was important that i nhibitor s be highly selective to HIVPR and have mini mum affinity to the other human aspartic proteases such as pepsin, renin, gastricsin and cathepsins D and E. 59 which had much success in the design of other aspartic proteases such as renin. 54 The transition state mimetic concept 59 when applied to aspartic proteases is where the scissile carbonyl of an oligopeptide substrate is replaced by a nonhydrolyzable analogue with tetrahedral geometry. 60 Many different transition state mimetics were proposed based on the P he Pro and Tyr Pro cleavage site s in order to ensure that the potential inhibitor would be potent as well as selective. Transition state mimetics such as: reduced amide, ketomethylamine, phosphinate, hydroxyethylene and hydroxyethylamine were evaluated as possible inhibitors of HIVPR. 60 62 K inetic studies revealed that the h ydroxyethylamine and hydroxyethylene mimetics were the best candida tes for being potential HIVPR inhibitors 54,62

PAGE 29

29 Figure 1 9. Transition state mimetics. A) Reduced Amide B) Hydroxyethylene C) Hydroxylethylamine D) Phosphinate E) Ketomethylamine In December of 1995, saquinavir, marketed as Invirase, became the first FDA approved HIV protease inhibitor. 54 The structur e of saquinavir is a derivative of the hydroxyethylamine mimetic. T o date, there are ten F DA approved protease inhibitors and the y are a mixture of peptidomimetic and non peptidomimetic drugs The potency of an inhibitor is measured by the free energy of binding, which is composed of both enthalpic and entropic contributions. The earlier generation HIVPR inhibitors were designed to be rigid the dominant driving force for binding was entropically dr iven with unfavorable or small favorable change enthalpy change. 63,64 The large entropi c gains were a result of desolvatio n of the non polar groups, which release s water to the bulk solvent increasing the solvation entropy and ligand being pre shaped

PAGE 30

30 to geometry of the active site, minimizing the loss of conformational entropy. 63 Newer develop ed inhibitors such as darunavir, were found to be instead enthalpically driven and are less susceptible to drug resistance. 65 As mentioned earlier all FDA approved HIVPR inhibitors bind to the active site of HIVPR. These inhibitors are susceptible to drug resistance due to drug pressure selected mut ations on HIVPR In order to rectify this problem, non competitive inhibitors have been proposed as a means to increase the efficacy of the current HIVPR inhibitors. Two different types of non competitive inhibitors have been proposed: inhibitors that prev ent dimerization 66,67 and inhibitors that perturb the dynamics of the protein, more specifically, the dynamics of the flaps 68 70 The protease dimer in a sheet that comprises both N and C termini of each monomer. Residues 1 4 a nd 96 99 of the terminal residues account for over half the hydrogen bonds along the dimer interface 71,72 75% of the dimerization free energy is attributed to the contacts in N and C termini 73 The inhibitors that bind and prevent dimerization target the N and C te rminal strands of each monomer residues 1 4 and 96 99. This region is highly conserved thus making drug resistance minimum. 74 Furthermore, t he mono mer for m the HIVPR is catalytically inactive. 66 The first dimerization inhibitors were derived from the amino ac id sequences of N and C terminus strands 75,76 Dimerization inhibitors have since evolved to N terminal ly cross linked, interfacial protease peptides that were devised to mimic the protease dimerization interface 77 Computational studies have identified two druggable regions and the top/ of the of the protease that could accommodate an allosteric

PAGE 31

31 inhibitor. 69,70 Gly16, Gly17, Gln18, Lys14 and Leu63 Arg57. 69 The non active site allosteric inhibitors can bind to the apo form of H IVPR trapping the flaps in conformation s such as wide open or closed thus rendering the protease catalytically inactive. T hese non active site allosteric inhibitors can also bind the protease active site inhibitor complex and synergistically imp rove the bi nding affinity of an active site protease inhibitor. 69 Figure 1 10. Regions targeted by non site fragment 4DX bound colored yellow 1.8 Drug Resistance Drug resistance has become a dire problem and reduces the efficacy of protease inhibitors. As discussed previously, the replication process of HIV is very error prone, which provides the fou ndation for d rug resistance. Without exposure to any protease inhibitors patients will naturally develop polymorphisms in HIVPR. However, when patients are exposed protease inhibitors drug pressured selected mutations will

PAGE 32

32 develop. When comparing the rat es of mutations versus nave patients, (have not been exposed to any protease inhibitors) and protease inhibitor treated patients (have been exposed to protease inhibitors ) residues in the N and C termini, 1 10, 96 99, active site 25 27 and the flaps tips 49 52 are conserved Figure 1 11 78 80 Figure 1 11. Per residue mutation frequency of protease inhibitor nave patients and patients who have been exposed to protease inhibitors (subtypes A, B, C, D, G, AE and AG.) The drug pressure sel ected mutations are the result of viral replication not being fully suppressed in the presence of an inhibitor Drug resistance can occur not only from the presence of protease inhibitors but from mutations in the cleavage sites of the

PAGE 33

33 protease 81,82 The mutations of HIVPR are classified as either primary or secondary mutations as depicted in Figure 1 12 Primary mutations are often residues in the active site that come into direct contact with the inhibitor These mutations elicit resistance by reducing the number of interact ions with competitive inhibitors by altering the geometry of the active site 83 Consequently primary mutations reduce the catalytic efficiency of the enzyme. 84 Secondary mutations, which are not always a result of protease inhibitor therapy 85 are thou ght to restore the catalytic efficiency of the enzyme while still maintaining resistance to the inhibitor. 86 Furthermore, these mutations often do not directly interact with the inhibitor yet the drug resistance is still maintained. As of now, fif teen sites for major mutations and nineteen for minor mutations have been determined for all the currently approved FDA protease inhbitors. 87 Figure 1 12. Schematic of HIV 1 protease illustrating the location of the primary (yellow) and secondary (purple) mutations.

PAGE 34

34 King and Schiffer et al. 88 proposed a mechanism describing how the primary drug pressured selected mutations evolve in the protease The mechanis m entails that an y inhibitor that protrudes beyon d the substrate envelope causes mutations in the specific residues that it directly contacts. 88 The residues beyond the substrate envelope that the inhibitor interacts with directly are more important for the inhibitor binding than the substrate bind in g and this expl ains why HIVPR can tolerate substitutions at those positions The residues in the active site that directly interact with inhibitor (s) and mutations at these sites often result in drug resistance for one or more protease inhibitors are D30, I47, G48 and I5 0, V82 and I84. Although several mechanisms 83,89,90 have suggested how non active site mutations confer resistance it is still n ot well defined. The hydrophobic sliding mechanism which was developed through MD simulations of the apo form HIVPR, involves nineteen non active site hydrophobic residues Each of these residues in the hydrophobic core slide by each other exchanging, one hydrophobic van der Waa l contact for another, with minimum energy penalty, while maintaining hydrogen bonds that are important to the stability of the protease. 89 Non active site mutations have been implicated in perturbing dynamics of the protease by decreasing the rate of flap closure causing the flaps of the protease to be more open and bind the substrate over the inhibitor. 83 The last mechanism suggests non active mutations alter the fractional occupancy of the equilibrium conformational sampling. 90 1.9 General Overview This body of work involves using MD as a means to stud y HIVPR. The aims of this work are: determine the prot onation of the cata lytic aspartates, explain the mechanism of active/non active site mutations, determine what effects if any the

PAGE 35

35 stabilizing mutations have the on EPR measured distance profiles and make suggestions on developing an inhibitor for XMRV prot ease.

PAGE 36

36 CHAPTER 2 METHODS 2.1 Molecular Dynamics The first MD simulation was published by Alder and Wainwright, in which they reported on the interactions of hard spheres. 91 Bovine pancreatic trypsin inhibitor became the first MD simulation involving a protein, and it was instrumental in paving the way for the application of MD to biological systems. Since then Molecular Dynamics has been beneficial in providing des cription on the atomic level, validating experimental results, structure refinement, drug design, enzymatic reactions, etc. C urrently with the increase of computing power MD simulations can reach time scales of microseconds. MD is s a deterministic method meaning that the state of the system at any future time can be projected from its current state. laws of motion, in particular his second law that states F=ma, where F is the force exerted on the particle, m is its mass and a is its acceleration. (2 1) Equation 2 1 describes the motion of particle of mass m i along a single coordinate x i with the force Fx i acting on the particle in tha t dire ction solving the differential E quation 2 1 yields a trajectory. 92 In order to calculate a trajectory the initial positions, velocities and accelerations must be defined. For example, the initial positions can be from an x ray crystal or NMR structure, and the velocities and the acceleration are calculated by the g radient of the potential energy function. 92,93 MD calculates the dynamics of system as it evolves wit h time, from which time average properties can be calculated. If a system is simulated for infinity then it will eventually cover the entire phase space of system. The ergodic hypothe sis states that

PAGE 37

37 the time average is equal to the ensemble average only if enough phase space has been sampled. The ergodic hypothesis allows MD simulations to calculate macroscopic variables using only a microscopic representation of a system. 2.1.1 Molecu lar Dynamics Integrator The integration of equations of motion can be solved analytically for systems like one dimensional harmonic oscillator. On the contrary, this is not feasible for such complex systems as proteins and these equations must be solved nu merically. 94 Many numerical algorith ms have been developed in order to accomplish this, for example, Verlet 95 leap frog 96 velocity Verlet 97 and Beeman 98 All of these algorithms use a Taylor series expansion to make approximations about the positi ons, velocities and acceleration. No matter what algorithm is employed during a MD simulation it must adhere to the following : conserve energy and momentum, be time reversible and allow 92 The size of the time step is ex tremely important because if it is too small then time is waste d sampling the same conformation O n the other hand if the time step is too big then instabilities such as steric clashes occur and conservation of energy becomes difficult. The leap frog algorithm was used in the MD simulations for this body of work and is defined as: (2 2) (2 3) where q v and a i n Equation s 2 2 and 2 3 c orrespond to the coordinate, velocity an d acceleration vectors of a ll the atoms. 96 In the leap frog algorithm the velocities ( t + are initially calculated from the velocities at time (t and the accelerations at

PAGE 38

38 time t. 92 The positions are calculated from the velocities along with the positions using Equation 2 2. The velocities at time t are determined from: (2 4) Thus, the velocities leap over the positions and vice versa. The advantage of this method is the velocities are explicitly calculated but the drawback of this method is that calculation of the position s and velocities are no t deduced in the same step. 2.1 .2 Langevin Dynamics Langevin Dynamics are often used in conjunction with implicit solvation models or as a thermostat for temperature 92 The Langevin equation of motion is a stochastic cond law of motion in ord er to account for the omitted degrees of freedom. In the description of stochastic dynamics the force exerted on particle is attributed to three different factors. The first force is owed to the interaction of a particles with other particles the magnitud e of the force acting on a particle depends on its position relative to the other particles. 92 The second force is described as the force felt by a particle as it moves through the solvent and is equated to the frictional drag on the particle due to the solvent. 92 The last force exerted on a particle is due to the random fluctuations that are a result of interactions with solvent molec ules. 92 The Langevin Equation can be expressed as : (2 5) where m is the mass, x is the position, F is an interaction force between a particle and orce exerted on a particle due to random fluctuations with solvent molecules.

PAGE 39

39 2.2 Molecular Mechanics Molecular Mechanics is possible because of three main assumptions. First, the Born Oppenheimer approximation which allows the electronic and nuclear moti on in molecules to be separated the nu clear motion is neglected thus simplifying the construction of the Hamiltonian. 92 Second Molecular Mechanics assumes that such processes as bond stretching, bond angles and rotations about the bond can de scribe the interactions within a system 94 Lastly, the parameters calculated for the small molecules can be transl ated to describe larger systems such as proteins. 94 Force fields involve the following parameters: (2 6) where the first three terms describe the covalent interactions and the last two terms describe the non covalent interactions. V denotes the potential energy, which is a function of the positions (r) of N atoms. The first two t erms in E quation 2 6 modeled as a harmonic potential due to the second order truncation of the Taylor series expansion, account for the energy change that occurs when the bonds and angles deviate fr om their reference value k is representative of the forc e constants for bonds and angles The third term in Equation 2 6 defines the torsional angles. The torsional term is expressed in a Fourier series where V n is the dihedral force constant, n is the dihedral 99 There are instances where additional terms called improper torsions are included in the force field to adequately describe such motions as out of plane bending. The incorporation of

PAGE 40

40 imp roper torsions is important when atoms must rem ain in the same plane such as the case for a benzene ring. The fourth and fifth terms of Equation 2 6 are the Lennard Jones and Coulomb potentials. The Lennard Jones potential is used to describe the van der W which the inter particle potential is zero, r 12 is the repulsive term and r 6 is the attractive term No theoretical evidence has been provided to justify to why the rep ulsive term is r 12 many force fields implement the 6 12 Lennard Jones potential because the repulsive term can be calculated by squaring the r 6 term thus reducing the computational cost. 92 The Coulomb potential is used to describe electrostatic interactions where q 1 and q 2 denote the charges of atoms, r represents the in ter ectric constant. In general, most of the force fields employed for describing protein have the same function al form as Equation 2 6 However, the y differ greatly b y what parameters are used in Equation 2 6 for example the value of the forces constants of the bonds, angles and torsions. The AMBER ff99SB 99 CHARM22 100 and OPLS 101 are examples of force fields commonly employed in MD studies of biological systems. The force field AMBER ff99SB was used for the parameters of the MD simulations performed in this study. 2.3 Solvent Models Solvent models are often employed to model the environment that proteins or any other system of interest exist in. The solvent is typical ly represented using two different approaches: explicit and implicit solvent. In an explicit solvent model, the

PAGE 41

41 solvent is modeled with real atoms that interact with the solute and in the implicit solvent model the solvent is represented using a dielectri c continuum. In most cases, t he computational cost of MD simulations is increased when the solvent is explicitly modeled as opposed to implic itly 2.3 .1 Explicit Solvent There are many different explicit water models that are utilized in molecular dynamic s simulations: SP C /E 102 TIP3P, 103 TIP4P 103 and TIP5P 104 The water models have been parameterized by fitting to the bulk properties of water such as: heat of vaporization, density and dipole moment. 103 The water models listed above differ mainly by the number of interaction site s they include. TIP3P and SPC/E have three interaction sites TIP4P has four sites and TIP5P has five sites The computati onal demand increases with the number of sites in the water model the central processing unit (CPU) time is approximately proportional to the number of interatomic distances that must be computed. The TIP3P model has three sites for electrostatic interacti ons a negative charge ( 8.314 e ) is place d on the oxygen atom and a partial positive charge (0.417 e ) is placed on each h ydrogen atom. T he van de r Waals interactions between two waters molecules or atoms of solute are only computed with the oxygen atom. Th e geometry of TIP3P water is identical to the experimental geometry of a water molecule the OH bond length is 0.9572 and the HOH angle of 104.52 Unlike TIP5P, TIP3P does not accurately reproduce the density of water that exhibits a maximum at 4 C. 104 The TIP3P water mod el was used in all of the explicit solvent simulations described in this study.

PAGE 42

42 Figure 2 1. Schematic representation of a TIP3P water molecule 2.3 .2 Implicit Solvent Model s Due to the cheaper computational cost and lower viscosity that enhances conforma tion sampling, the solvent is sometimes modeled implicitly. The methods are particularly helpful when the main focus of a study is the solute and the role of solve nt is miniscule. Two main approaches are used to model the solvent solute electrostatic inter actions: Poisson Boltzmann (PB) and Generalized Born (GB) methods. The GB method is an approximation to the PB implicit solvent model and is less computationally demanding. The free energy of solvation is composed of an electrostatic and a non electrostati c component. The PB and GB equations are used to calculate the electrostatic term while algorithms like LCPO 105 determine the non electrostatic term by calculating the solvent accessib le surface. 2.3 .2.1 Poisson Boltzmann s olvent m odel The PB models have shown that they can replicate conformations and energetic s comparable to explicit solvent simulations as well as experimental measurements for a variety of systems. 106 The Poisson equation is the foundation of all

PAGE 43

43 continuum solvent models. However, The Poisson equation, Equation 2 7 is only valid when the ionic strength of the system is zero. (2 7) The Poisson equation must be amended in order to account for the presence of ions thus the electrostatic energy is computed by solving Poisson Boltzmann equation : (2 8 ) nstant, z i is the charge of the ion type i c i is the bulk number density of ion type i k B is the Boltzmann constant and T is the temperature. 2. 3.2.2 Generalized Born s olvent m odel The Generalized Born model is often used to represent the solvent in protein simulations due to a cheaper computational cost when compared to other implicit solvents. As discussed earlier the GB is an approximation to P B and when given the same parameters such as atomic radii the results of PB and GB are the same for small to medium sized molecules 107 The basis of the GB method is from the equation for the polarization energy of monoatomic ion in atomic units often referred to as so called Born equation Equation 2 9 (2 9 ) An expansion of Equation 2 9 to apply to polyatomic molecules yields the GB equation that can be expressed as: (2 10 )

PAGE 44

44 q is the charge of an atom, r ij is the inter particle distance, and a ij is the Born radii. The f (r ij ,a) which is an arbitrary function developed by Still et al. 108 that can be further simplified to: (2 11 ) where r ij a ij correspond to the inter particle distance and the Born radii. Equation 2 11 has several advantages but the main advantage is that the expression can be differentiated analytically, allowing the solvation term to be incorporated in MD simula tions and gradient based optimization methods. 92 2.4 Steered Molec ular Dynamics Steered Molecular Dynamics involves applying an external force onto a system in order to drive changes in its coordinates within a certain time. SMD can be applied to study such problems as ligand dissociation, ion diffusion, and conformation changes in biological systems on time scales accessible to MD simulations SMD is similar to umbrella sampling except that the center of restraint is time dependent. The second law of thermodynamics mandates that the ensemble average of work performed o n a system by an external force be either larger than or e qual to free energy difference between two states the dissipated work is just the difference between (2 12) The equality in Equation 2 12 is onl y valid for quasistatic processes, which means that the change from A to B must occur infinitely slowly. However, using the so called Jarzynski relationship equality 109 Equation 2 13 allows work calculated from going from

PAGE 45

45 state A to B for example, to be related to equilibrium free energies from non equilibrium processes. (2 13) Because the average work mu st be calculated many MD simulations must be run, each simulation starting from different configurations of the initial state (in this case state A) 2.5 Constant pH Methods The structure of a protein and its function is contingent on the pH of the solve nt. The pK a dictates the protonation of a titrat able residue at a particular pH. T he pK a values of titratable residues are govern by its surrounding electrostatic environment In traditional MD simulations, protonation states of titratable residues are ass igned prior to the simulations. This approach has two issues. First, the assignment of the protonation state is usually taken from the protonation state of the residue in free solution. 110 The pK a of a residue in free solution and it s pKa in the folded protein c an differ significantly Lastly, when the pK a values of titratable residues of a system are near solution pH no single protonation state properly describes the ensemble at tha t pH. 2.5.1 Constant pH Molecular Dynamics Constant pH MD alleviates the problems listed above by allowing proper sampling of coupled conformational and protonation state distributions corresponding to the semi grand canonical ensemble. 111,112 The different CpHMD methods available can be c ategorized in two main approaches: discrete protonation states 112 and continuous protonation states. 113 In discrete protonation state models, Monte Carlo exchanges between different protonation s tates are attempted during the simulation. Cont inuous

PAGE 46

46 protonation state models i to control the process of protonation and deprotonation of a group labeled i The CpHMD algorithm 114 implemented in AMBER treats the solvent implicitly using the Generalized Born method. At each Monte Carlo step, a change in protonation state of either one or two closely interacting titratable residues is attempte d. The transition free energy, Equat ion 2 14 between the present and proposed protonation state is calculated applying the Metropolis criterion 115 to determine if the proposed state is accepted or rejected. (2 14) In Equation 2 14 k B is the Boltzmann constant, T is the temperature, pH is the current pH of the solvent, pK a,ref is the pK a of the appropriate reference compound, G elec is the electrostatic component of the free energy calcul ated for the titratable group in the protein, G elec,ref is the electrostatic component of the transition free energy for the reference compound, a free dipeptide amino acid. 114 2.5.2 Constant pH Replica Exchange Molecular Dynamics Before introducing Constant pH Replica Exchange Dynamics, a brief introduction to Replica Exchange will be given. Replica Exchange Molecular Dynamics is the MD version of parallel tempe ring developed by Sugita and Oka moto. 116 REMD is a method where multiple copies or replicas of the same system are simulated simultaneously with different parameters i.e. tem perature. Periodically, the MD simulation is halted and in formation, i.e. conformations are swapped between adjacent replicas based on the Metropolis criterion. A schematic representation of REMD is shown in Figure 2 2

PAGE 47

47 Figure 2 3. Schematic representation of REMD simulations. Each arrow represents an in dependent MD simulation MD is halted and MC exchange is attempted, which can be accepted or rejected. If the exchange attempt is successful then replicas are swapped between adjacent neighbors if not the MD continues. The concept of coupling CpHMD and Repl ica Exchange Molecular Dynamics, which is referred to as pH Replica Exchange Molecular Dynamics (pH REMD), was first introduced by Itoh et al. 117 and implemented into AMBER by Swails et al. 110 In pH REMD each replica is simulated at a different pH using CpHMD as described previously, and the solution pH is swapped between neighboring replicas. This method has been shown to enhance conformational sampling as well as produce pK a values that converge faster and achieve better precision when compared to just employing CpHMD. 117 110 2.6 Accelerated Molecular Dynamics The potential energy surface of biological systems is compos ed of multiple minima separated by large bar riers. Correct sampling of the PES can be difficult using conventional MD because the time scales of MD simulations are limited to the nanosecond microsecond ranges. Consequently, MD simulations become trapped in

PAGE 48

48 local minima for long periods of time ensemble properties calculated from the MD simulations become erroneous due to incomplete sampling. As stated previously, the ergodic hypothesis states that the time average is equal to the ensemble average only if en ough phase space has been sampled. Hamelberg et al. 118 introduced a bias potential method named Accelerat ed Molecular Dynamics that has been shown to circumvent the problem of kinetic trapping AMD lowers the barrier height therefore decreasing the amount of time it takes to properly sample the conformational space. 119 The boosted energy surface from AMD maintains the integrity of the unde rlying shape of the real one this ensures that original distribution can be recovered by simply reweighing the distribution of structures from the AMD simulation. The criteria for when the potential boost i s added during the simulation are given b y Equation 2 15 and AMD modifies the potential using the following equation: (2 15) (2 16) where V(r) is then normal pot E dictates the energy below which the sy stem is accelerated, the tuning parameter affects the smoothness There has been no theoretical justification as to how to choose the parameters E and parameters used in the AMD must be chosen carefully if the parameters a re too small then little enhancement will be seen in the sampling and if they are too large then s mulation will become unstable.

PAGE 49

49 Figure 2 4. Schematic representation of the normal and biased potential of AMD simulations. (Solid line) 1 D potential energ y landscape, unmodified. (Dashed line) modified potential energy landscape. E is the threshold boost energy 2.7 Docking states only the key (substrate) with the correct size /shape will fit into the lo ck (active site of protein). Docking is a method that predicts the orientation of a ligand when bound to a protein. Docking is often used in conjunction with drug design because large libraries of compounds can be rapidly screened for p otential inhibitors as opposed to experimentally determining the binding affiniti es of the compounds, which would expensive and time consuming Docking calculations proceed in the following fashion: generating a grid around the target receptor, conformati onal sampling of the ligand, and scoring and ranking each docked pose. Each grid point within the grid encompassing the target receptor is used to calculate the interaction potential between the ligand and target receptor. The larger the grid the more expe nsive the docking calculation becomes therefore on ly the region (active site) of the target receptor where the ligand will bind is generally included in the

PAGE 50

50 grid. The idea l scenario would be to generate every possible conformation of ligand and receptor bound together this is only plausible if considering only a few ligands and as the number ligands evaluated increases so does the computational expense. Some docking programs instead only allow certain residues in the protein deemed important to be flexibl e while the rest of protein is rigid. Generally, most docking calculations are performed keeping the receptor rigid and only grant flexibility to the ligand. Different poses of the ligand are generated by an exhaustive search of all angle torsion space for the rotatabl e bonds. Each pose is scored using an empirically derived scori ng function similar in form to Equation 2 6 2.7 M olecular Mechanics Poisson Boltzmann Surface Area Free energy calculations have proven to be great resource in computational biol ogy, especially as it pert ains to drug design. Thermodynamic Integration 120 (TI) and Free Energy Perturbation 121 (FEP) are examples of methods available to calculate free energies. Th ese methods, though theoretically rigorous, become computationally expensive as the size of system increases. The TI and FEP divide the reaction coordinate into intermediate states to calculate the free energy between two states. Molecular Mechanics Poisso n Boltzmann Surface Area 122 is an end state method that eliminates the need for these intermediate states therefore r educing the com putational cost. MMPBSA is a post processing end state method that calculates th e free energies using the th ermodynamic cycle in Figure 2 5 The binding energies in MMPBSA can be calculated in the following fashion. (2 17) The terms on the right hand side of the equation can further simplified to:

PAGE 51

51 (2 18) where H gas is the gas phase energies TS solute is the entropy of the solute, and solvation is the free energy of solvation. H gas is calculated from the MM force field Equation 2 6. The entropy of the solute can be determined using either the normal mode or quasi harmonic approximation. The free energy of solvation is composed of ele ctrostatic and non electrostatic contributions, where the electrostatic contributions are determined from the implicit solvent model i.e. PB/GB and the non polar contribution is calculated from the SASA. Figure 2 5. Thermodynamic cycle of MMPBSA. HIV pr otease and saquinavir PDB 3OXC. The blue background represents the aqueous form of the ligand, receptor and complex. The goal of the MMPBSA is to calculate the free energy difference between two states, which is frequently represented by the bound and unb ound state of two molecules that are solvated (colored blue).

PAGE 52

52 CHAPTER 3 MOLECULAR DYNAMICS SIMU L ATIONS INDICATE THAT THE CATALYTIC DYAD OF HIV 1 PROTEASE IS MONO PROTONATED 3.1 I ntroduction The catalytic dyad has a significant role in protein ligand compl exes as well as on the dynamics of the flaps in the apo form. 43 123 In computational drug design studies it is imperative to assign protonation stat es correctly in o rder to achieve accurate binding affinities and conformational sampling 124 125 The catalytic dyad of HIVPR has been modeled using molecular dynamics in the deprotonated, di protonated, and mono protonated ionization states 50,99,126 Constant pH Molecular Dynamics (CpHMD) is an ideal technique to resolve this problem due to the uncertainty of assigning the correct protonation states of the catalytic dyad. We performed CpHMD and pH REMD simulations on the protease in its apo form and bound to thirteen protease inhibitors, whic h are shown in Figure 3 1 The inhibitors used in this study can be divided into three main groups: cationic, neutral asymmetric and cyclic urea inhibitors. The neutral asymmetric group is chiefly composed of FDA approved inhibitors that are currently bein g used to treat patients infected with HIV. CpHMD was performed at pH 5 (HIVPR is most catalytically active in a pH range of 4 6) 127 and pH 7 (physiological pH). pH REMD simulations were performed at pH 0 11. This study shows evidence that the protonation state of the catalytic dyad is mono protonated for bo th the apo and bound forms of HVIPR over a wide pH range In addition, we determine which of catalytic aspartates sections, should be protonated and which of the oxygen atoms, OD1 or OD2, possesses the proton when the aspartat e is protonated in the bound form of HIVPR.

PAGE 53

53 These results will be compared to previous theoretical and experimental results in which the protonation state of the catalytic dyad have been investigated. Figure 3 1. 2D schematic of the protease inhibitors u sed in the Constant pH Molecular Dynamic Simulations. A) Amprenavir, B ) Atazanavir, C) Darunavir, D) Indinavir, E) Lopinavir, F) Nelfinavir, G) Ritonavir, H) Saquinavir, I) Tipranavir, J) KNI 272, K) DMP 323, L) XV 638, M) MVT 101. Inhibitors A J are the neutral asymmetric inhibitors, K L are the cyclic urea inhibitors, and M belongs to the cationic group.

PAGE 54

54 3.2 Methods 3.2.1 General Setup Fourteen different forms of HIVPR were simulated in which the initial coordinates were X ray crystal structures from the Protein Data Bank, PDB codes: 1HHP apo 128 3NU3 Amprenavir 129 2O4K Atazanavir 130 3TKW Darunavir 131 2BPX Indinavir 132 2O4S Lopinavir 130 1OHR Nelfinavir 133 1HXW Ritonavir 134 2O4P Tipranavir 130 1QBS DMP 323 135 1QBR XV 638 135 4HVP MVT 101 39 and 3FX5 KNI 272. 136 In order to be consistent, all the crystal structure sequences were muta ted to match that of 1HHP T able A 1 Having identical sequences is important because previous studies have shown that active site mutations and non active site mutations, even though not close in proximity to the active site, can have an impact on the dyn amics of the flaps as well as how residues in the active site interact with the ligand. 137 131 Missing hydrogen and heavy atoms were added using the LEaP module in AmberTools 12. 138 The AMBER ff 99SB 99 force field was used for the protein and the parameters for the ligands were obtained using the antechamber module and GAFF force field 139 with AM1 BCC charges. 140 Each H IVPR complex began from three different protonation states: deprotonated, di protonated and mono protonated. In the case of the simulations that began in the mono protonated state, the hydrogen was initially placed on the OD2 atom of D25 and in the simulat ions that started in the di protonated state, a hydrogen was The effects of the solvent were accounted for by using the Generalized Born 108 implicit solvent model, GB OBC model 141 (igb=5 in AMBER) in conjunction with the mbondi2 intrinsic radii. The internal dielectric was set to 1 and the external dielectric to

PAGE 55

55 78.5. The cutoff for the van der Waals and electrostatic interactions was set to 30 and the salt co ncentration, modeled as Debye Hckel screening parameter was set to 0.1 M. Langevin Dynamics was utilized in order to maintain the temperature at 300 K and the collision frequency was set to 50 ps 1 to mimic the viscosity of liquid water. Different initial seeds were applied to each simulation to avoid synchronization artifacts. 142 The SHAKE algorithm 143 was used to constrai n bonds containing hydrogen so that a 2 fs time step could be used. 3.2.2 Constant pH Molecular Dynamics Protocol An equilibration procedure was employed prior to performing CpHMD on the various HIVPR complexes. First, the hydroge n atoms were minimized using 500 steps of steepest descent followed by 7000 steps of conjugate gradient. Next the entire system was minimized using 500 steps of steepest descent and 10000 steps of conjugate gradient. All systems were heated from 100 K to 3 00 K for 1 ns where 100 kcal/mol 2 positional restraints were used on the backbone atoms. Following heating, four 200 ps simulations were performed where the positional restraint was r educed from 100 to 0.1 kcal/mol 2 for the backbone atoms. Finally, in the last step of equilibration all restraints were removed and a 200 ps simulation was performed. No protonation state changes were attempted during the equilibration process The production phase consisted of 12 ns of unrestrained CpHMD of which the last 10 ns were used for analysis. A MC step was performed every 10 fs in order to change protonation state s All minimizations and CpHMD simulations were performed using the s ander module in AMBER 12. 138

PAGE 56

56 3.2.3 Constant pH Replica Exchange Molecular Dynamics Proto col Two dif ferent sets of pH REMD simulations were performed. The first set of pH REMD simulations included all 14 HIVPR complexes where 12 equally spaced replicas spanning from pH 0 11 were used. Each simulation was begun in a mono protonated state as change protonation states. The second set of pH REMD simulations included 12 replicas spanning from pH 0 11. All of the titratable residues of the protein, excluding the argini ne amino acids, N and C termini of the HIVPR complexes were allowed to change protonation states. The arginine amino acid residues, N and C termini, were excluded because parameters are not available. All of the aspartates and glutamates began deprotonated except for one placed in the epsilon position) amino acid residues began each simulation in the neutral form. The lysine amino acid residues started each simulation protonated. The starting structures for the pH REMD simulations were acquired from employing an equilibration procedure for each pH, similar as to what was done for the CpHMD simulations. For the first set of simulations, a protonation state change was at tempted every 10 fs and for the second set of simulations a protonation state change was attempted every 4 fs to determine if a change in protonation should occur. The exchange attempt frequency (EAF) of 100 ps 1 was used for the first set of pH REMD simul ations. The impact of a larger EAF would be negligible since only two residues pro to nation states were probed. The EAF was 10 ps 1 for the second the set of pH REMD simulations. A larger EAF is warranted given that the number of titratable residues to con sider is substantially more, 36 in all, when compared to the first set of

PAGE 57

57 pH REMD simulations. Exchanging more frequently will allow simulations to transverse the replica ladder more rapidly, which improves the conformational sampling. 110 144 The production phase for the first set of pH REMD was 30 ns and the production phase for the second set of pH REMD simulations was 50 ns. 3.2.4 General Analysis The cpptraj 145 module of AmberTools was used to analyze the trajectories from the CpHMD and pH REMD simulations. The program calcpka in AmberTools was utilized to calculate the resulting protonation state distributions from the CpHMD and pH REMD simulations. 3.3 Results This section is orga nized as follows: first, the results from the CpHMD, and secondly, the pH REMD simulations are presented. Lastly, we offer results to dyad is protonated in conjunction with which of the oxygen atoms, OD1 or OD2, sh ould possess the proton. Only the results from the CpHMD and pH REMD simulations in the pH range pH 4 7 will be considered, since this is the most experimentally relevant range. 3.3.1 Constant pH Molecular Dynamics All HIVPR CpHMD simulations executed at pH 5 converge to the mono protonated protonation state regardless of their initial protonation state. Furthermore, the results indicate that 10 ns is more than ample time due mainly in part to each of the different initial protonation states converging to the same pro ton ation state as shown in Figure 3 2 As mentioned in earlier sections, the mono protonated simulations were run with the D25 being initially protonated. In order to ensure that the results of the mono protonated simulations were not biased we ran three HIVPR complexes amprenavir,

PAGE 58

58 DMP 323, and MVT consistent with those obtained from the D25 simulations. Figure 3 2. Plot of Average total molecular protonation versus time for H IVPR complex KNI 272. CpHMD performed at pH 5, where a total molecular protonation of 0.00 indicates that neither aspartate is protonated, 1.00 that only one aspartate is protonated and 2.00 that both are protonated. Because each system converged to the sa me result at pH 5, only the apo structure, one neutral inhibitor, (nelfinavir), one cationic inhibitor, (MVT 101) and one cyclic urea inhibitor, (DMP 323) were each simulated at pH 7. The simulations were initially begun in the same manner as described ear lier. Each system was simulated at three different initial protonation states in order to determine if they will too converge to the same protonation state. All four HIVPR complex simulations converged to the mono protonated state regardless of its initial protonation similar to the CpHMD simulations at pH 5. 3.3.2 Constant pH Replica Exchange Molecular Dynamics In order to ensure that the results obtained from the CpHMD simulations were not a result of conformational trapping, pH REMD was employed. The f irst set of pH REMD simulations included all 14 HIVPR complexes. The pH values of the pH REMD

PAGE 59

59 simulations spanned from 0 11 and only the catalytic aspartates were allowed to change protonation. The results of the pH REMD simulations reveal that the cata lytic aspartates of HIVPR are mono protonated between pH 4 7 The second set of pH REMD simulations consisted of allowing all titratable residues in the protein, excluding those listed in the methods section, to change protonation. The pK a is governed by local electrostatic interactions, which is dictated by conformation of the protein and protonation state of titratable residues because of the tight coupling of protein conformation and protonation state we chose to include all possible titratable residue s. 114 Each replica was initiated in the same fashion as described previously. As was done in the CpHMD simulations, only the apo stru cture, as well as one neutral inhibitor, (KNI 272), one cationic inhibitor, (MVT 101), and one cyclic urea inhibitor (DMP 323) were used in the pH REMD simulations. The catalytic aspartates for all four different HIVPR complexes remained mono protonated be tween pH 4 7. 3.3.4 Location of the proton in the mono protonated state Our results have shown that the catalytic aspartates are mono protonated in the apo and holo forms with the various inhibitors used in the study. The proton could exist on D25 or D25 However, OD1 or OD2 possesses the proton? On the other hand, in the bound d istinguishable. In the bound that is dep rotonated as shown in Figure 3 3

PAGE 60

60 Figure 3 3. Schematic of the fo ur different possible mono protonated protonation OD1 protonated, D) D25 simultan eously deprotonates. This illustrates that either of the catalytic aspartates can be protonated in the neutral asymmetric, cylic urea and cationic inhibitor simulations. In the simulations involving the neutral asymmetric inhibitors, the central hydroxyl group of the inhibitor rota tes to interact with the catalytic aspartate that is deprotonated when a change in The next issue is which of the oxygen atoms possesses the proton when one of the catalytic aspartates is protonated, OD1 (inner) or OD2 (outer) oxygen atoms. The oxygen atoms OD1 and OD2 have been defined based on their orie ntation at t=0 as shown in Figure 2 3 A. The four different mono protonated arrangements are shown in Figure 2 3 It is not clear from the simulations which oxygen atom OD2 or OD1 should

PAGE 61

61 possess the proton However, the proton is never seen in the anti position but only in syn position relative to the oxygen atoms. 3.4 Discussion 3.4.1 Apo (Free enzyme) Our results suggest that the aspartyl dyad of the apo form of HIVPR is mono protonated, which corroborates with previous kinetic and theoretical studies. 53,127,146 148 In experimental kinetic studies done by Ido et al. 127 and Hyland et al. 53 their results show that the aspartyl dyad of free pr otease has two apparent pK a s ~3 and ~6. In theoretical studies done by Trylska et al 146 and Czodrowski et al 147 their results yielded pK a s comparable to those found in the kinetic studies. However, a previous NMR study, conducted by Smith et al. 149 determined that the aspartyl dyad is deprotonated at pH 6.0 in the absence of an inhibitor, which our results contradict. Trylska et al. 146 have already addressed explanations as to why the results of Smith et al. 149 may not agree with the g eneral consensus. 3.4.2 Cyclic Urea Inhibitors When cyclic urea inhibitors, DMP 323 and XV 638 are bound to HIVPR the catalytic aspartates remain mono protonated between pH 4 7. This result is in agreement with the findings of Mardis et al. 150 they sought to explain the observed U shape of binding free energy as function of length for series of cyclic urea inhibitors. The study revealed that the mono protonated model of the catalytic dyad yields the best agreement w ith experimental trends, while the di protonated model was unsuccessful in reproducing the characteristic U shape of measured binding free energy chain length. 150 However, the author did note that the calculated absolute binding free energies suggest that the cyclic urea inhibitors bind to the protein in the di protonated state. Trylska et

PAGE 62

62 al. 146 performed two sets of calculations using the full charge/4 model 151 and the si ngle site/20 model. 152 The results from the single site/20 model yielded pK a s of ~3 and ~6 for the catalytic aspartates that coincide with our results. On the contrary, a previous NMR study, found that when DMP 323 is bound to the protease that the catalytic dyad exists in the di protonated state over the pH range 2 7. 153 Trylska et al 146 full charge/4 model results and Czodrowski et al 147 found that their theoretical results suggest that the aspartyl dyad is fully p rotonated. However, Trylska et al 146 did experimentally determine that the inhibition constant for DMP 323 remained constant from a pH range of 5.5 7. 146 If HIVPR when bound DMP 323 is di protonated then the binding affinity should decrease with increasing pH. Reasons as to why there are some discrepancies between our proposed results and Trylska et al 146 and Czodrowski et al 147 might be that their calculations only incorporated st atic structures, and did not account for the flexibility of the protein ligand complex. Their solvent was modeled implicitly using the Poisson Boltzmann method while we instead used the Generalized Born method. Lastly, different dielectrics were used whic h can have a profound impact on the calculated pK a value. 154 3.4.3 Neutral Asymmetric Inhibitors The catalytic aspartates of HIVPR are mono protonated from pH 4 7 when bound to the neutral asymmetric inhibitors employed in this study. Our results obtained from the CpHMD and pH REMD simulations are found to be in agreement with other previous experimental and theoretical studies suggesting that the protease is mono protonated for neutral asymmetric inhibitors. 146,147,155,156 Unfortunately not all the inhibitors used in this study have experimental and theoretical pK a values that we can compare to but we will compare when possible.

PAGE 63

63 Wang et al. 155 experimentally determined that the aspartyl dyad of HIVPR is mono protonated from a pH range of 2.5 6.2 for the inhibitor KNI 272 and were unable to extend beyond this range. The theoreti cal calculations of Trylska et al 146 and Czodrowski et al 147 does show that the catalytic dyad is mono protonated when KNI 272 is bound, between pH 4 7 with pK a s Our results for amprenavir are found to be in good agreement with those of Czdrowski et al 147 3.4.4 Cationic Inhibitor The catalytic aspartates remained mono protonated between pH 4 to 7 when the inhibitor MVT 101 is bound. The pK a determined for this protein ligand complex. However, our findings agree with previous theoretical studies of Geller et al. 157 and the results from full charge/4 model of Trylska et al. 146 A MD study performed by Harte et al 158 proposed that the protonation state of the aspartyl dyad is deprotonated when cationic protease inhibitor MVT 101 is bound. 158 However, as previously noted, Trylska et al. 146 performed two sets of calculations and the single site/20 calculation suggested that the aspartyl dyad is deprotonated from pH 5 7. 146 Here we offer an explanation as to why Harte et al 158 and our results to do not coincide explan ations as to why the results from Trylska et al. 146 do not possibl y corroborate with ours have been addressed previously. Harte et al 158 performed 100 ps MD simulations using different ionization states of the protein: mono protonated, di protonated, and deprotonated. Their assignment of the protonation state of the D25 and ng the backbone average structure of the ligand from the MD simulations to the orientation of the ligand in the crystal structure and

PAGE 64

64 deprotonated model of the catalytic dyad yielded the lowest deviations of 0.5 The lengths of their simulations might no t be sufficient and longer time scales are needed the correlation between ligand orientation and assigning protonation of a residue is not obvious. 3.4.5 Low Barrier Hydrogen Bond of HIVPR. Furthermore, the proton can exist on either the OD1 or OD2 when the aspartate is protonated. This topic is still under investigation but has been addressed by previous studies. 156 159 160 One explanation as to why the proton can be on either oxygen could be attributed to the low barrier hydrogen bond (LBHB). 148 161 The LBHB is believed to be a compensation for the strong repulsion between the two catalytic aspartates. 148 3.5 Conclusions Assigning the correct protonation state to the catalytic dyad is important in c omputational drug design. We have shown through the use of the CpHMD and pH REMD simulations that the protonation state of the protease is mono protonated in the apo form as well as with the various inhibitors included in this study. This is illustrated by the convergence of the HIVPR complexes to the mono protonated state regardless of the initial protonation state. We conclude that the binding of an inhibitor does not induce a change in the protonation state when transitioning from the apo to bound form o f HIVPR for the inhibitors used in this investigation. There is still some ambiguity concerning where the proton is located in the mono protonated state and further investigation is needed. Possibly the position of the proton

PAGE 65

65 in the mono protonated state can not be defined by a single state, but instead by an ensemble of arrangements in the mono protonated bound form of HIVPR

PAGE 66

66 CHAPTER 4 UNDERSTANDING HOW THE G48T/L89M MUTATIONS REDUCE THE EFFICACY OF THE HIV 1 PROTEASE INHIBITOR SAQUINAVIR 4.1 Introduction The lifetime of HIV patients has been extended due to the success of highly active antiretroviral therapy (HAART). The HIV protease has remained an important for antiretroviral drug therapy target due to its role in the life cycle of the virus. 57 Inhibition of the protease would lead to an immature v irion that is non infectious. The efficacy of protease inhibitors is reduced as a result of drug pressure selected mutations and mutations cleavage sites of the protease 81,82 Figure 4 1. HIV 1 protease, t he c olored spheres indicate locations on the protease the where wild type and mutant differ. Position 48 (G48T) is colored yellow and position 89 (L89M) is colored orange. The catalytic aspartates are also depicted. Mutations that cause resistance to the prote ase inhibitors are not just restricted to residues in the active site but can include residues in other regions of the protein. 86

PAGE 67

67 The mechanism of how an active site mu tation elicits resistance is caused by reduced interactions with the inhibitor and altering the size of the active site cavity 83 Non active site mutations are defined as sites that do not directly i nteract with inhibitors yet confer resistance the mechanism of drug resistance for non active site mutations is not well understood although several mechanisms have been proposed. 83,89,90 Figure 4 2 Schematic representati on of the HIV 1 protease inhibitor saquinavir. Forty two patients failing treatment with saquinavir and ritonavir were discovered to possess the drug resistant mutations G48T and L89M. In this study we describe how the two mutations G48T (active site) loc ated in the flap tips and L89M (non active site) helix, reduce the efficacy of the protease inhibitor saquinavir, shown in Figure 4 1. Saquinavir, Figure 4 2, is a transition state analog inhibitor and was the first FDA approved HIV 1 prot ease inhibitor. Mutations at position 48 frequently occur as a result of treatment with saquinavir. 162 In order to evaluate the mechanism of how t hese mutations diminish the efficacy of saquinavir type and

PAGE 68

68 typ e and mutant in the apo form. 4.2 Methods 4.2.1 General Setup The initial coordinates and sequ e nce for the wild type and the mutant were obtained from the x ray crystal structure PDB ID 3OXC 163 (holo simulations) and PDB ID 1HHP 128 (apo simulations). An in house python script was used to make the mu tations to the protease sequence The stabilizing mutations 164,165 (Q7K, L33I, L63I, C67A C95A) were remov ed from the protease sequence T able A 1 The missing atoms from x ray structure were added using the LEaP module in AmberTools13. The AMBER ff99SB 99 force field was used for the protein and the parameters for saquinavir were generated using the antechamber module and GAFF 139 force field with AM1 B CC 140 charges. Only one of the catalytic aspartates was protonated. The explicit solv ent was modeled using a 15 TIP3P 103 solvent buffer encapsulating the protein in a truncated octahedron box and c hloride ions were added to force neutrality. T he long range electrostatic s were treated with the particle mesh Ewald method (PME) 166 using direct space and van der Waals cutoff of 9 The temperature was maintained at 310 K using Langevin Dynamics with a collision frequency of 1.0 ps 1 The SHAKE 143 algorithm was employed so that a 2 fs time step could be utilized. 4.2.2 Steered Molecular Dynamics The equil ibration procedure for the Steered Molecular Dynamics simulations was begun in the same fashion as previously described in the MD subsection. A 15 ns

PAGE 69

69 explicit solvent MD simulation was performed in order to extract the snapshots necessary for the SMD simu lations. The snapshots were extracted every 100 ps. During the production phase the solvent was modeled implicitly using the Generalized Born (igb=8) 167 solvation model in AMBER in conjunction with the mbondi3 intrinsic radii. No cutoff was used for the non bonded interactions. The temperature was maintained at 310 K using the weak coupling algorithm 168 (ntt=1) in AMBER. The re action coordinate was the distance between the center of mass of the inhibitor and the protease (residues in the flaps 43 58 of each monomer were excluded). The force constant was 400 kcal/mol 2 and the pulling rate was 0.8 /ns. 4.2.3 Molecular Mechani cs Generalized Born Surface Area Calculations The MMGBSA calculations were performed using MMPBSA.py 122 in AmberTools 13. The polar solvation energy was calculated using the Generalized Born Method (igb=8) in conjunction with the mbondi3 intrinsic radii. The non polar solvation energy was calculated using the LCPO 105 method. The quasi harmonic approximation was used to account for the entropic contributions. 4.2.4 General Analysis Analyses of the trajectories were performed using the cpptraj module of AmberTools13 and VMD 169 All MD simulations were performed using pmemd.cuda 119 in AMBER 12 170 4. 3 Results and Discussions 4.3.1 Inhibitor Simulations 4.3.1.1 Flexibility To evaluate the differences in flexibility of the wild type and mutant the root mean ere measured and the difference between the

PAGE 70

70 wild type and mutant (positive value indicates that the wild type residue fluctuates more and a negative value indicates that the mutant residue fluctuates more) Figure 4 3. The largest differences between the wild tip residues (48 ompared to the wild type because of the larger side chain Met89 in the mutant increases the reducing its flexibility. The flap tips of the mutant exhibit more flexibility than the wild type, which will be explained in more detail later. Figure 4 3. Atomic Fluctuations of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir A) Plo t of the RMSF B) Plot of the difference between the wild type and mutant (G48 T/L89M). P ositive value signifies the wild type fluctuates more and a negative value indicates the mutant fluctuates more. Monomer A (1 ).

PAGE 71

71 To probe the conformation of the flaps, we plotted the values of the root mean squared deviatio 58) versus the minimized crystal structure, Figure 4 4. The maximum peak of the wild type is shifted to the left of the maximum peak of the mutant and the width of the distribution is broader for the mutant, indicating that t he flaps of the wild type are less flexible when compared to the mutant. Figure 4 4. Histogram of the RMSD of the C of the flap residues (43 58) of each monomer MD simulations of the wild type and the mutant (G48T/L89M), subtype B, bound to saq uinavir. Figure 4 5. Histogram of the RMSD of saquinavir MD simulations of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir.

PAGE 72

72 To assess the flexibility of saquinavir, the RMSD of saquinavir versus the minimized crystal structures was calculated, and to make the distinction between the two clearer, the data was histogrammed and plotted in Figure 4 5. The RMSD distribution in Figure 4 5 is much broader for saquinavir in the mutant, which is indicative of saquinavir possessing more f lexibility in the active site of the mutant relative to the wild type. Figure 4 6. Histograms of the different distances in the active site of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir A) Ile50 distance B) Val82 V al distance C) Asp25 Ile50 Ile50 distance type and mutant were measured and plotted in order to quantify the flap tip separation, Figure 4 6 A. This distance provides insight as to the openness of the flaps of HIVPR. The wild type and the mutant exhibit a peak at 6 However, the maximum peak of the wild type is 8 and the 7.5 for the mutant, suggesting that the flap tips separation in the wild type is slightly larger. In Figure 4 6 B, the distance between the active site residue Val82 of each monomer was measured, to assess the width of the active site cavity This distance fluctuates less in the wild type because the width of distribution is smaller than the mutant. There is a small tail seen between 17 19 in the mutant that is not p resent

PAGE 73

73 in the wild can become smaller in the mutant than the wild type. Lastly, the distance between the catalytic Asp25 and Ile50 of monomer was measured and plotted Figure 4 6 C. This distance provides insight as to how large the active site is vertically. The distance distribution in the mutant is broader than that of the wild type. The flap tips of the wild type point downward towards the active site thus increasing the flap t ip separation while decreasing the distance between the Asp25 and Ile50. 4.3.1.2 Hydrogen bonding a nalysis Table 4 1. Hydrogen bonding analysis of the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir. Acceptor Donor % Occupancy Wild ty pe O SQV199 O H 97 GLY48 O SQV199 N H 57 SQV199 O ASP25 O H 39 Water Mediated SQV199 18 G48T_L89M O SQV199 O H 90 THR48 O SQV199 N H 23 Water Mediated SQV199 11

PAGE 74

74 The direct and water mediated hydrogen bonds that saquinavir forms with residues in the active site were investigated to determine if the mu tations cause any differences, T able 4 1. The catalytic aspartates of the wild type, Asp25 (protonated) and group of saquinavir more in the wild type when compared to the mutant. Furthermore, the Gly48 backbone carbonyl of the wild type forms a hydrogen bond more frequently with saquinavir than Thr48 of the mutant. Both wild type and mutant form a water mediated hydrogen bond between saquinavir and Asp29. Blackburn et al 171 se the flaps correlates with the number of non water mediated hydrogen bonds between the inhibitor and protease and our data corro borates this story as shown in T able 4 1. The flaps are more closed like in the wild type and the ligand is more rigid, theref ore the saquinavir is forming more interactions with the protein in the wild type than in the mutant. The G48T mutation appears to reduce the number of interactions with the flaps therefore increasing the flexibility of the flaps. 4.3.1.3 MMGBSA c alculatio ns The MMG BSA calculations in T able 4 2 reveal that there is approximately a 2 kcal/mol difference between the binding affinities of the wild type and the mutant. The enthalpic contributions is ~7 kcal/mol higher for the wild type than the mutant. This is indicative of the residues in the active site of the wild type forming more favorable interactions (electrostatic, van der Waals) with the ligand. The solvation free energy is ~4 kcal/mol lower in the mutant. The difference between the entropic contributio ns for the wild type and mutant are negligible.

PAGE 75

75 Table 4 2. Results of MMGBSA calculations for the wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir All units in kcal/mol System H gas G solvation T S quasi harm G bind Wild type 106.79 0 .05 51.94 0 .04 23.05 31.79 0 .02 G48T_L89M 99.22 0 .05 47.72 0 .03 22.38 29.12 0 .02 Figure 4 7. Per residue energy decomposition of the binding affinity of the wild type and the mutant (G48T/L89M) subtype B bound to saquinavir The diff erence between the wil d type and mutant is shown, where a positive binding free energy indicates that the mutant (G48T/L89M) has a more favorable contribution and vice versa for the wild type. Mono mer A (1 99), ) The residues in the plot are highlighted on HIVPR 29 (orange) Gly/Thr48 (yellow), The binding free energy was decomposed into contributions from the individual residues of the protein to gain insight into the role of particular residues in the binding

PAGE 76

76 mechanism of saquinavir. The difference between the wild type and mutant is plotted in Figure 4 7, to make the comparison clearer. One of the main differences between the wild type and mutant is the energetic contributions of the catalytic aspa rtates. Asp25 in the wild type contributes more to the binding free energy because it forms more in the wild type. therefore r educing its desolvation penalty Asp29 of the mutant forms more electrostatic interactions with saquinavir via a hydrogen bond between an oxygen atom of saquinavir and backbone amide group Asp30 in the mutant forms more favorable van der Waals type forms more van der Waals interactions with saquinavir than the mutant. The interactions between saquinavir type contribute more to the binding energy. Phe53 interacts strongly with saquinavir in the wild type, because of pi stacking and this same interaction is not formed in the mutant The difference in the Waals and electrostatic interactions with saquinavir. Thr48 and Ile50 of the mutant have a larger energetic contribution. The larger contribut ions are a result of the orientation of the quinoline ring of saquinavir. 4.3.1.4 Steered Molecular Dynamics Steered Molecular Dynamics simulations were performed to investigate what effect the G48T mutation in the flaps had on the escape pathway of saquin avir. The work profiles of the various SMD simulations are shown in Figure 4 8. The binding free energy profiles of the SMD simulations are plotted in Figure 4 8. The free energy

PAGE 77

77 difference between the mutant and wild type is approximately 2 kcal/mol, whic h is comparable to the difference that the MMPBSA results yielded. Figure 4 9 shows that it takes more energy to remove saquinavir from the active site of the wild type because saquinavir forms more favorable interactions with the residues in the active si te of the wild type than the mutant. These results agree very well from those obtained in MMGBSA that demonstrated that the G48T and L89M reduces the affinity for saquinavir therefore the drug is easier to remove from the active site. Figure 4 8. Work Profiles from the SMD simulations for wild type and the mutant (G48T/L89M), subtype B, bound to saquinavir. All of the ligands escape the active site by the flaps opening; no lateral escapes were seen, in which the flaps do not have to fully open. After t he ligand exits the active site the flaps remain predominantly in the semi open conformation. The SMD simulations revealed that the flaps need to open at least 35 (distance between Ile50

PAGE 78

78 Figure 4 9. Free energy profile of removing the protease inhibitor saquinavir from the active site cavity of the wild type and the mutant (G48T/L89M) subtype B 4.3.2 Apo Simulations Simulations of the apo (unbound) conformations were investigated in order to deter mine what effects the G48T and L89M had on the conformation of the flaps and other regions of the protease. Previously studies have demonstrated that mutations can alter the equilibrium between the open and closed conformations of the flaps. 90,137,172 Figure 4 10 Plot of the RMSD of wild type subtype B, flap residues, 43 58, versus the closed, semi open and wide open conformations MD simulations performed in the apo form. The rmsd of the dynamics of the flaps, residues 43 58 of each monomer of the wild type and mutant versus the closed (PDB ID 3OXC), semi open (PDB ID 1HHP),

PAGE 79

79 and wide open (obtained from a previous MD simulations unpublished) conformations are shown in Figures 4 10 and 4 11. A schematic o f the three reference structures is shown in Figure 1 7. The lower the rmsd signifies that the flaps favor that particular structure as opposed to the others. The dynamics of the flaps for the wild type began in semi open conformation but as the simulation progresses the flaps become closed and no other flap conformation is visited. There are instances where the rmsd of both the closed, semi open, wide open conformations are higher than 4 and in such cases the flaps were in tucked/curled conformation. How ever, the flap dynamics of the mutant favor predominantly the semi open conformation in the early stages of simulation. However as the simulation progresses to ~400 ns the flaps in all three simulations favor the tucked/curled conformation. The wide open c onformation is never seen in either the Figure 4 11 Plot of the RMSD of mutant (G48T/L89M) subtype B, flap residues, 43 58, versus the closed, semi open and wide open conformations MD simulations performed in the apo form. wild type or the mutant sim ulations. The flaps of wild type transition from the semi open to the closed conformation and remain there for the entirety of the simulation, which could be a result of kinetic trapping.

PAGE 80

80 the wild type a n d mutant and the difference between them are plotted in Figure 4 12. The largest deviations are in the residues 43 58 of each monomer, which are the flap tips. This difference is due to the of the flaps of the wild type remaining in the closed conformation for the majority of the simulations. The largest differences between the wild type and mutant are in the flap 12 exhibits the same trend as the saquinavir bound MD simulations. Figure 4 12. Atomic Fluctuations of the wild type and the mutant (G48T/L89M), subtype B, apo. A) Plot of the RMSF. B) Plot of the difference between the wild type and mutant (G48T/L89M). Positive value signifies the wild type fluctuates more and a negative value indicates the mutant fluctuates more. Monomer A (1 4.3.3 Drug Resistant Mechanism The G48T mutation causes a conformational change in the quinoline ring of saquinavir. This conformational change reduces th e ability of the quinoline ring in the

PAGE 81

81 mutant to form pi stacking interactions with Phe53 Figure 4 13 This same observation has been made as a result of the G48V mutation and has been well documented. 173,174 The side chain hydroxyl group of T hr48 possesses the ability to form a water mediated hydrogen bond with saquinavir that Val48 is unable to form. The backbone oxygen of the Gly48 residue in the wild type forms a hydrogen bond with saquinavir approximately 20% more than the Thr48 residue of the mutant. Figure 4 13. Schematic of the Phe53 pi pi stacking interaction with the quinolin e ring of saquinavir. A wild type and the B) mutant Resi due Thr48 of the mutant reduces the interaction between Phe53 and the quinoline ring of saquinavir

PAGE 82

82 The interaction of the central hydroxyl groups of saquinavir with the catalytic aspartates is diminished in the mutant by almost 10%. The diminished interactions between the catalytic aspartates and the hydroxyl group of saquinavir might be attributed to the L 89M mutation although it is not obvious. Sanches et al. 175 proposed that the L89M mutation causes the L90 residue to shift and form more van der Waals interactions with residues close in vicinity to Asp25. No shift of the L90 was observed during the course of o ur MD simulations pair wise energy decomposition calculations revealed that there were no differences in van der Waals interactions of Leu90 in wild type or mutant. The side chain of the larger Met residue does form more van der Waals s strand (Ala71, Ile72, Gly73, Thr74, Val75). The flexibility of the 4.4 Conclusions Through the use molecular dynamics we have described how the G48T and L89M mutations reduce the efficacy of saquinavir on an atomic level. The MMGBSA calculations and SMD simulations demonstrate that the binding affinity of the mutant is lowered and it takes less energy for saquinavir to be removed from the active site of the mutant The major e ffects of the mutations are: the G48T mutation causes the reorientation of the quinoline ring that reduces the number of inte ractions between saquinavir and residues in the flaps, specifically Phe53 and the L89M mutation has been implicated in causing the reduced interactions of the central hydroxyl group of saquinav i r with the catalytic aspart at es though further investigation is needed The flaps are more closed like in the wild type indicating that the saquinavir makes more direct hydrogen bonds with res idues in the active site.

PAGE 83

83 Sa quinavir when bound to the mutant or the wild type forms a water mediated dification such that saquinavir form s a direct bond with Asp29 might it increase it s potency 176 These results can aid in the design of new inhibitors to become more potent against drug resistant proteases

PAGE 84

84 CHAPTER 5 EVALUATING THE EFFECTIVE NESS OF SAQUINAVIR AND TIPRANAVIR ON HIV 1 SUBTYPE C WILD TYPE AND THE DRUG RESISTANT N88D/L90M MUTANT PROTEASE 5.1 I ntroduction Subtype C accounts for the majority of all the HIV infections worldwide. 20 The current anti retroviral drugs used to combat HIV, more specifically the FDA approved protease inhibitors have been developed using subtype B. Fortunately these protease inhibitors have been proven to still inhibit subtype C although the efficacy of these protease inhibitors when compared to subtype B is diminished. 177 179 Figure 5 1. Schematic of HIVPR subtype C Green spheres indicate where the sequences of subtypes B and C differ, (T12S, I15V, L19V, M36I, S37A, H69K, L89M, I93L). Red spheres illustrate the positions of the drug resistant mutations N88D and L90M The first crystal structure of Subtype C protease was not determined until 2008. 177,180 The sequences of subtype C and B differ by only eight residues, T12S,

PAGE 85

85 I15V, L19V, M36I, S 37A, H69K, L89M, I93L. These are non active site mutations that are located in the elbow region of the flaps, and the hydrophobic core, Figure 5 1. Figure 5 2. Schematic of the protease inhibitors saquinavir and tipranavir. As mentioned in previous chapt ers the introduction of protease inhibitors causes drug selected pressure mutations that reduce the efficacy of the protease inhibitors. In this study we investigate how the non active site mutat ions N88D, and L90M, diminish the effectiveness of saquinavir Figure 5 2. The L90M mutation reduces the efficacy of all the protease inhibitors except darunavir and tipranavir. 87 The N88D mutation is a facilitator in the co occurrence of D30N and L90M mutations, causing multidrug resistance. 181 The single N88D mutant increases the dimer stability. 182 The single L90M mutant increases the catalytic ability 129,131,183 of the protease even though it destabilizes the stability of the dimer. 182 The stability of the double mutant N88D/L90M is moderate when compared to the wild type and demonstrated a higher catalytic abil ity than the wild type for certain substrates. 177,182 Coman et al. 177 determined the K I values for the wild type and several d rug resistant mutants for subtype C the N88D/L90M double mutant exhibited 42 fold

PAGE 86

86 increase in K I against saquinavir, while same double mutant displayed only 1.3 fold increase against tipranavir. The aim of this study is to explain how the N88D and L90M mut ations reduce the efficacy of the protease inhibitors saquinavir and tipranavir. 5.2 Methods 5.2.1 General Setup The initial coordinates and sequence for the wild type and mutant were derived from the x ray crystal structure PDB 2R5Q. 177 The stabilizing mutations were removed from the sequences prior to performing any MD simulations Table A 3 The missing atoms from the x ray structure were added using the Leap module in Amb erTools13. The AMBER ff 99SB 99 force field was used f or the protein. The parameters for saquinavir and tipranavir were constructed using the AM1 BCC 140 charges generated from the antechamber module and the GAFF 139 force field. The m ono protonated state was chosen for the catalytic aspartates The explicit solvent was modeled using a 15 TIP3P 103 solvent buffer encapsulating the protein in a truncated octahedron box and chloride ions were added to force neutrality. T he long range electrost atics were treated with the PME method 166 using direct space and van der Waals cut off of 9 Langevin Dynamics was utilized to maintain the temperature at 310 K the collision frequency was 1.0 ps 1 In order to achieve a 2 fs time ste p all atoms bonded to hydrogens were constrained using the SHAKE 143 algorithm. 5.2.3 Molec ular Dynamics The hydrogen atoms were minimized followed by the entire system. Each minimization process consisted of 500 steps of steepest descent followed and 10 000 steps of conjugate gradient. The systems were heated linearly from 100 K to 310 K over

PAGE 87

87 a 1 ns MD simulation, where a 5 kcal/mol 2 positional restraint was placed on the solute. Following heating, the density was equilibrated for 500 ps. Next, five simulations were performed where the positional restraint was reduced from 5 kcal to .1 kcal/mo l 2 In the last step of equilibration all restraints were removed and 5 ns of unrestrained MD were performed. The production phase consisted of running 20 x 51 ns simulations but only the last 50 ns was used for analysis. Snapshots of the MD trajectory were saved every 30 ps. 5.2.3 Molecular Mechanics Generalized Born Surface Area Calculati ons The MMG BSA calculations were performed using MMPBSA.py 122 in AmberTools13. The polar solvation energy was calculated using the Generalized Born Method (igb=8) 167 in conjunction with the mbondi3 intrinsic radii. The non polar solvation energy was calculated using the LCPO 105 method. The quasi harmonic approximation was used to account for the entropic contributions. 5.3 Results and Discussion 5.3.1 Saquinavir bound MD Simulations 5.3.1.1 Flexibility T and the difference bet ween wild type and mutant is plotted in Figure 5 3. The mutant exudes a greater flexibility especially in the flap tips residues 48 53 of monomer A (residues 1 99). There are differences between the two i n 86 92 of monomer B (residues 100 198). The active site residues type. A histogram of the RMSD of saquinavir versus the minimized crystal structure is shown in Fi gure 5 4. The RMSD distribution of saquinavir in the mutant is much broader

PAGE 88

88 than that of the saquinavir in the wild type. This suggests that saquinavir in the mutant is not interacting in a similar fashion as when it is bound to the wild type thus the fle xibility of the ligand is increased. Plotted in Figure 5 5 is a histogram of the RMSD of the flaps (43 58) of each monomer versus the minimized crystal structure (closed). The flaps of the wild type are re is a small tail between 1.5 and 2 that is not present in the wild type. Furthermore, this plot reaffirms the notion that the mutant exhibits more flexibility as was demonstrated in Figure 5 3. Figure 5 3. Atomic Fluctuations of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. A) Plot of the RMSF. B) Plot of the difference between the wild type and mutant ( N88D/L90M ). Positive value signifies the wild type fluctuates more and a negative value indicates the mutant fluctuates mo re. Monomer A (1

PAGE 89

89 Figure 5 4. Histogram of the RMSD of saquinavir. MD simulations of the wild type and mutant (N88D/L90M) subtype C, bound to saquinavir. Figure 5 5. Histogram of the RMSD of the flap r esidues (43 58) of each m onomer. MD simulations of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. measured and plotted to determine the flap tip separation, Figure 5 6 A. The Ile50 distance for the wild type has one large peak at 6 and smaller peak at 8 The Ile50 distance for the mutant has a maximum peak at 6 and second peak that ranges from 7 8 In order to gain further insight on the size of the active site, the distances 6 B, and Asp25 Ile50, of each

PAGE 90

90 monomer, Figure 5 6 C wer e measured. In Figure 5 6 B, the maximum peak of the mutant is shifted to left of the maximum peak of the wild type this is indicative that the tail between 17 20 t hat is located in the mutant but is not as prevalent in the wild type. In Figure 5 6 C, both the wild type and mutant exhibit peaks at 13 and 15 but the wild type peaks are narrower than those of the mutant. The reason why Asp25 Ile50 and istance distributions are not the same is due to the asymmetry of the ligand. Figure 5 6. Histograms of the different distances in the active site of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir. A) Ile50 distance B) Va l82 Ile50 distance 5.3.1.2 Hydrogen bonding a nalysis In order to assess how the mutations N88D and L90M affected the interactions of saquinavir with residues in the active site the direct and the water media ted hydroge n bonds were calculated, T able 5 1. The central hydroxyl group of saquinavir in the wild mutant catalytic aspartates interacts with the central hydroxyl group of saquinavir significantly less.

PAGE 91

91 Table 5 1. Hydrogen bonding analysis of the wild type and the mutant (N88D/L90M), subtype C, bound to saquinavir Acceptor Donor %Occupancy Wild type N88D/L90M O SQV O H 98 86 SQV O Asp25 O H 81 41 Gly48 O SQV N H 52 38 Gly27 O SQV N H 26 21 Water Mediated SQV199 18 11 The most noticeable difference is the hydrogen bond between Asp25 (protonated) and saquinavir, which in the mutant is formed ~50% less. The difference between the frequency of hydrogen bondin the wild type forms this hydrogen approximately 12% more. The backbone carbonyl oxygen of Gly48, located in the flaps, interacts less with saquinavir in the mutant than in the wild type. Lastly, Gly27, located at the base of the active site, interacts only slightly type and mutant forms a hydrogen bond with saquinavir via a water molecule. 5.3.1.3 MMGBSA c alculations Saquinavir binds m ore f avorably to the wild type and there is roughly a 4 kcal/m ol difference between the two, T able 5 2. The H Gas term of the wild type is roughly 10 kcal/mol larger and that is indicative of it forming stronger interactions with saquinavir than the mutant The desolvation penalty is much lower in the mutant this is due to the

PAGE 92

92 mutant not displacing as many waters as wild type in the active site. The difference in the entropy between the two systems is negligible. Table 5 2. Results of the MMGBSA calculation s for the wild type and the mutant (N88D/L90M), subtype C, bound to saquin a v i r All units are in kcal/mol System H gas G solvation T S quasi harm G bind Wild type 125.60 0.08 65.27 0.05 23.88 36.43 0.03 N88D_L90M 114.87 0.08 58.09 0.05 24.52 32.24 0.04 Figure 5 7. Per residue energy decomposition of the binding affinity of the wild type and the mutant (N88D/L90M), subtype C bound to saquinavir. The difference between the wild type and mutant is shown, where a positive binding free energy indicates that the mutant (N88D/L90M) has a more favorable contribution and vice versa for the wild type. Monomer A (1 99), Monomer B (100 198). The residues in the plot are highlighted on HIVPR Asp25/Asp25 le50 (red), Gly27/ Ala28/

PAGE 93

93 In order to gain further insight the binding affinity was decomposed into individual residue contributions and the difference between the wild type and mutant is plotted in Figure 5 7. If the number is negative than the residue of the wild type contributes more to the total binding affinity and vice versa if it is positive. The difference in the contribution of Asp25 between the two is attributed to the Asp25 of the wild type making more favorable electr ostatic interactions (hydrogen bonding) with saquinavir. The contribution to be higher than the mutant. The Gly27 and Ala28 residues of the wild type make for favorable va n der Waals and electrostatic interactions with saquinavir, resulting in a larger contribution to the total binding affinity. A slight difference in the orientation of the quinoline ring causes the energetic contributions of the Gly49 and Ile50 to be diffe rent between the mutant and the wild type. 5.3.2 Tipranavir bound MD Simulations 5.3.2 .1 Flexibility The RMSF o difference between them are plotted in Figure 5 8 The mutant shows a greater flexibility than the wild type. The mo st noticeable differences are Asp35, flap elbow residue, Gln61 and both located in the hydrophobic core, of monomer A (residues 1 helical residue of monomer B (residue 100 198) fluctuates more in the mutant. The difference between the flap residues (43 58 of each monomer) of the wild type and mutant is insignificant <.1 The RMSD of tipranavir versus the minimized crystal structure is plotted in Figure 5 9. The maximum peaks of both the wild type and mutant are centere d at 1 but the

PAGE 94

94 RMSD distribution of the mutant is slightly larger the mutant has a tail between 1.5 and 2 that is not present in wild type. Figure 5 8. Atomic Fluctuations of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir. A ) Plot of the RMSF. B) Plot of the difference between the wild type and mutant (N88D/L90M). Positive value signifies the wild type fluctuates more and a negative value indicates the mutant fluctuates more. Monomer A (1 Figure 5 9 Histogram of the RMSD of tipranavir MD simulations of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir.

PAGE 95

95 The RMSD of the flap residues versus the minimized crystal structure is shown in Figure 5 10 Both systems exhibit maximu m pe aks at approximately .7 There is no obvious difference between the two although the flaps favor the closed structure slightly more for the wild type because its maximum peak is larger. Figure 5 10. Histogram of the RMSD of the flap residue (43 58) of each monomer. MD simulations of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir. Figure 5 11. Histograms of the different distances in the active site of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir. A ) Ile50 distance B) Val82 Ile50 distance The distance between Ile50 of each monomer was determined and plotted in Figure 5 11 A, in order to assess the flap tip separation. A maximum peak at roughly 6 is

PAGE 96

96 exhib ited by both systems. The measured distance between Val82 of each monomer is plotted in Figure 5 11 B. There is no major difference between the two and both wild type and mutant display peaks at 20 The distance between residue Asp25 and Ile50, was compu ted and plotted in Figure 5 11 C. The wild type and mutant have maximum peaks at 13 and 14.5 5.3.2 .2 Hydrogen bonding a nalysis To gauge the impact the mutations had with tipranavir and residues in the active site, a hydroge n bond analysis was performed, T able 5 3. The mutations does not affect However, the protonated Asp25 of both the wild type and mutant do no t interact strongly with the hydroxyl group of tipranavir. Th e hydrogen bond interaction between backbone carbonyl oxygen Gly48 and tipranavir is slightly stronger in the wild type. The wild type with tipranavir via the amide grou p. Table 5 3. Hydrogen bonding analysis of the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir. Acceptor Donor %Occupancy Wild type N88D/L90M O TPV O H 99 99 Gly48 O TPV N H 67 52 TPV O N H 43 43 TPV O Gly48 N H 3 1 33 TPV O Ile50 N H 26 31

PAGE 97

97 5.3.2 .3 MMGBSA c alculations Table 5 4, lists the results of the MMPBSA calculations performed on the wild type and mutant. The difference in binding affinity between the wild type and mutant is approximately 4 kcal/mol. The di fferences in the free energy of solvation and the entropy of th e wild type and mutant are ~1 kcal/mol The wild type makes more van der Waals interactions with tipranavir and explains why the H Gas term is higher the difference between the electrostatic interactions of the wild type and mutant is minute Table 5 4. Results of the MMGBSA calculations for the wild type and the mutant (N88D/L90M), subtype C, bound to tipranavir All units in kcal/ mol System H gas G solvation T S quasi harm G bind Wild type 120.6 0 0.03 53.87 0.03 22.27 44.44 0.02 N88D_L90M 118.13 0.04 54.33 0.03 23.29 40.5 0.02 The binding free energy was decomposed into individual contributions and the diff erences between the wild type and mutant are plotted in Figure 5 12 Th e difference between the wild (protonated) is negligible. On the other hand, the difference of the energetic contribution o is attributed to the wild type having a lower de solvation penalty. The hydrogen bonding analysis T able 5 3, demonstrated d the wild type equally and further investigation is need to explain why the desolvation penalty is lower in the wild type. Ala28 of wild type contributes more to the total binding free energy as a result of forming more favorable van der Waals and electro static interactions. On the other hand, Asp29 of the mutant contributes more by the same token. The only noticeable difference between the wild type and mutant flap residues

PAGE 98

98 interaction with tipranavir is Gly48. The difference between them is a direct resu lt of the electrostatic interactions between tipranavir and Gly48 being more favorable in the wild type. The most interesting difference between the wild type and mutant is Arg87. Arg87 is not located in the active site but does occasionally form hydrogen bonds with the trifluoromethyl group of tipranavir (<3%). The lower desolvation penalty of Arg87 accounts for the majority of the difference between the wild type and mutant. Figure 5 12. Per residue energy decomposition of the binding affinity of the wi ld type the and mutant (N88D/L90M), subtype C bound to tipranavir. The difference between the wild type and mutant is shown, where a positive binding free energy indicates that the mutant has a more favorable contribution and vice versa for the wild type. Monomer A (1 99), Monomer B (100 (blue), Gly48 (yellow), Asp29(green), Ala28 (pink), Arg87 (red)

PAGE 99

99 5.3.3 Saquinavir vs. Tipranavir Tipranavir is not as susceptible to the L90M/N88D mutations as saquinavir which is consistent with kinetic studies of Coman et al. 177 There is approximately a 4 kcal/mol difference in the binding affinity between wild type and mutant when either tipranavir or saquinavir is bound. The reason why binding of tipranavir is decreased in the mutant could be attributed to the fact the subtype C possesse s the polymorphisms 36I, 69K and 89M that are secondary mutations, not located in the active site, associated with tipranavir 87 However the binding affinity of mutant tipranavir is still higher than the binding affinity of wild type saquinavir T ables 5 2 and 5 4 One explanation as to why t ipranavir is more effective than saquinavir is because it fo rms more direct hydrogen bonds with residues in the acti ve site especially in the flaps. The flap hydrogen bonds help to maintain the rigidity of tipranavir allowing it still interact strongly with the catalytic aspartates even though the longer side chain of Met90 disrupts the dimer interface interactions. 5.3.4 Drug Resistant Mechanism The Asn88 residue in the wild type forms direct hydrogen bonds with Thr74, Thr31, and Asp29 through the following interactions: the side chain carbonyl of N88 forms a hydr ogen bond with the side chain hydroxyl group of Thr31, the backbone carbonyl of Thr74 forms a hydrogen bond with the side chain amine of Asn88, the backbone amide of Asn88 forms a hydrogen bond to the backbone carbonyl of Asp29 Figure 5 13 Furthermore As n88 hydrogen bonds with T hr74 via a water molecule. The interactions the Asn88 residue makes within the protease are consistent with previous studies. 178,184 Asp88 does not fo rm any significant interactions Thr74 or Thr31 via

PAGE 100

100 Figure 5 13. Schematic diagram of the A sp88/Asn88 i nteractions with Thr31 and Thr74 A) wild type B) mutant (N88D/L90M) Asp88 (purple) is unable to hydrogen bond to Thr31 and Th r 74 residues but Asn88 (grey) residue can. direct or water mediated hydrogen bonds with Thr74 or Thr31. Asp88 does form a hydrogen bond with Asp29 via backbone, amide and carbo nyl group. Asp88 inability to f o r m a hydrogen bond with Thr74 might affect the dynamics of the Pro79 Val82 loop in

PAGE 101

101 fluctuate s more when compared to the wild type, (saquinavir bound MD simulations). A pairwise energy decomposition calculation, results not shown, reveals that the longer side chain of the Met90 residue does form van der Waals interactions with residues Leu24, Asp2 5 and Thr26 than Leu90. This increased interaction causes a shift the central hydroxyl group of saquinavir. Met90 mutation is responsible for the decreased interaction of Ala28 and Gly27 with both saquinavir and tipranavir due to the weakened interactions at the dimer interface. 173,175 5.5 Conclusions Our MD studies have s hown how the non active site mutations reduce the efficacy of the protease inhibitor saquinavir. The Asp88 mutation affects the dynamics of the Pro79 Val82 loop in the active site by making it more flexible because of the lost hydrogen bond between Thr74 a nd Asp88 This effect is more pronounced when saquinavir is bound. The larger Met90 mutations impacts how the catalytic aspartates interact with the central hydroxyl group of saquinavir the interactions between the central hydroxyl group and the catalytic aspartates are unaffected by the L90M mutation Tipranavir is able to overcome these mutations by forming direct hydrogen bonds with the protein especially in the flap region The binding affinities of saquinavir and tipranavir when bound to the mutant ar e reduced when compared to the wild type. The reduced binding affinity of saquinavir is primarily due to the L90M mutation. The naturally occurring polymorphisms of subtype C 36I, 69K and 89M might be the reason the binding affinity of tipranavir is lower. The binding affinity of the mutant tipranavir is still higher than the wild type saquinavi r

PAGE 102

102 suggesting that tipranavir might still be an effective inhibitor in spite of the lower binding affinity when compared to the wild type. In addition, darunavir and tipranavir are the only inhibitors not susceptible to the L90M mutations. Although these mutations do not form direct contact with the inhibitor they confer resistance by altering the geometry of the active site, which in turn affects how the active site r esidues interact with ligand. These results offer insight as to how new inhibitors can be design ed to overcome drug resistant strains of HIV.

PAGE 103

103 CHAPTER 6 INVESTIGATING WHAT EFFECTS THE STABILIZING MUTATIONS HAVE ON THE EPR MEASURED DISTANCE PROFILES 6.1 Int roduction A variety of biomolecular conformational changes and structures have been studied using double electron electron resonance (DEER) 185 Distance measurements by site directed s pin labeling (SDSL) EPR are based on the degree of the magnetic dipolar coupling of the unpaired nitroxide electrons that scales as 1/r 3 where r denotes the distance between unpaired spins. 186 HIV 1 protease has two flaps that govern access to the active site. The dynamic s of the flaps have been studied using a variety of different methods. 43,46,47,50 In particular, Galiano et al. 186 Kear et al. 137 and Blackburn et al. 172 showed that DEER EPR spectroscopy could discern between the different flap conformations of HIVPR: closed, semi open, wide open and tucked/curled. In the se DEER experiments the distance bet ween K55SL (spin label) of each monomer is measured Figure 6 1 The K55 residue is mutated to a Cys and then the label (1 Oxyl 2,2,5,5 Tetramethly Pyrroline 3 Methyl) Methanethiosulfonate ( MTSL ) is attached. Lys 55 was chosen because it was one the few positions in the flap region that could tolerate an amino acid subst itution while still retaining activity comparable to the wild type 186 Furthermore, the di stances of the flaps span from 2 6 48 186 which is well within the sen sitivity of the methodology 187,188 In order to perform the DEER experiments several mutations w ere made besides the K55SL: the mutations Q7K, L63I, L33I, and D25N prevent auto proteolysis 164 C67 A and C95A are to ensure the spin label attaches exclusively to C55. This sequence is referred to as LAI. MD studies have shown that the EPR measured distances can be

PAGE 104

104 reproduced. 172,189 The question remains, do the mutations (Q7K, L33I, D25N, L63I, C67A, C95A) affect the distance profiles and trends observed in the previous EPR studies 137,171,172,186 performed on HIV PR? Figure 6 1. Schematic of HIV 1 protease. Highlighted are the spin labels at position 55 of each monomer. The dashed line is represents the distance that is measured. A total of t welve MD simulations containing the spin l abel MTSL were performed Nine MD simulations consisted of HIVPR, subtype B, bound to all nine FDA pro tease inhibitors. The others three simulations included the apo form of the wild type (subtype B), MDR769 190 and V6 191 which are drug resistant constructs. The sequence of MDR769 when compared to the wild type differs at positions L10I, M36V, S37N, M46L, I54V, I62V, L63P, A71V, V 82A, I84V, L90M and the sequence of V6 c ompared to the wild type differs at K20R, V32I, L33F M36I, L63P, A71V, V82A, L90M. The sequences of wild given in Appendix A

PAGE 105

105 6.2 Methods 6.2 .1 General Setup Twelve differen t forms of HIVPR were simulated and the initial coordinates were X ray crystal structures fro m the PDB, PDB Codes: 1HHP wild type (apo) 128 1HPV Amprenavir 192 2O4K Atazanavir 130 3TKW Darunavir 131 2BPX Indinavir 132 2O4S Lopinavir 130 1OHR Nelfinavir 133 1HXW Ritonavir 134 2O4P Tipranavir 130 If any of the sequences of the crystal structures posses stabilizing mutations, they were removed prior to running any MD simulations. The initial coordinates for the drug resistant constructs MDR769 an d V6 were from the PDB 1HHP. Due to the large amount of mutations made to 1HHP to match the sequences of MDR769 and V6 a special equilibration procedure was employed prior to solvating the structures and performing any MD simulations. The structures were minimized using 500 steps of steepest descent followed by 15000 steps of conjugate gradient, where only the side chains were allowed to move. Next, the structures were heated linearly from 100 to 300 K for 500 ps and a 50 kcal/mol 2 positional restraint w as used for backbone atoms. The solvent was modeled implicitly using the Generalized Born Method (igb=8) 167 in AMBER in conjunction with mbondi3 intrinsic radii. The missing hydrogen and heavy atoms were added using the LEaP module in AmberTools13. The AMBER ff99SB 99 force field was used for the protein and the parameters for protease inhib itors were generated using the antechamber module and GAFF 139 force field with AM1 BCC 140 charges. Only one of the catalytic aspartates was protonated.

PAGE 106

106 The explicit solvent was model ed using a truncated octahedron box of TIP3P 103 water s extended 15 from the edge of the solute chloride ions were added to force neutrality. Long range electrostatics were treated with PME method 166 using a direct space and a van der Waals interactions cutoff of 9 The temperature was maintained at 300 K using Langevin Dyna mics with a collision frequency of 1.0 ps 1 The SHAKE 143 algorithm was employed so that a 2 fs time step could be utilized. 6.2.2 MTSL Parameters The MTSL label was built using the visualization software Avogadro 193 Gaussian 09 194 was used to pe rform the geometry optimization of the MTSL in conju n ction with functi onal M062X and 6 31+G basis set. The antechamber module was employed to calculate the Restrained Electrostatic Potential (RESP) charges for the label. Figure 6 2. Schematic of the five torsi onal angles of the spin label MTSL The last step in the buildi ng of MTSL was parameterizing the torsions 1 2 3 4 5 Figure 6 4 5 model 195 suggests that the last two dihedrals, 4, and 5 are largely responsible for the internal motion of MTSL and that the other dihedrals, 1 2 and 3 are essentially rigid. Because of this model only the last two torsions have been previously deemed important enough to parameterize. 196 Furthermore, torsional fittings

PAGE 107

107 are generally performed by fitting each dihedral independently of each other because the assumption is that the coupling between torsions is negligible. We have determined that not only are these assumptions incorr ect but that all five torsions of the spin label are coupled and need to be fitted simultaneously. The program Torsion_Fit written by Chad Hopkins (in review), was used to generate the parameters for all five torsions. 6.2.3 Molecular Dynamics The hydrogen atoms were minimized first, followed by minimization of the entire system. Each minimization process consisted of 500 st eps of steepest descent followed by 10000 steps of conjugate gradient. The systems were heated linearly from 100 K to 310 K for 1 ns wh ere a 5 kcal/mol 2 positional restraint was used on the solute. Following heating, the density was equilibrated for 500 ps. Next, five simulations were performed where the positional restraint was reduced from 5 kcal to .1 kcal/mol 2 In the last step of equilibration a ll restraints were removed and 2 ns of unrestrained MD simulation was performed The production phase for the ligand bound simulations consisted of running 4 x 50 .5 ns simulations but only the last 5 0 ns was used for analysis. The apo simul ations consisted of running 10 x 30.5 ns simulations. Snapshots of the MD trajectory were saved every 30 ps. 6.3 Results and Discussions 6.3.1 Inhibitor bound MD simulatio ns A histogram of the RMSD distribution of the backbone atoms of the protein and the heavy atoms of the inhibitor for each of the inhibitor bound MD si mulations was plotted in Figure 6 3 to demonstrate that the simulations were stable. The reference structure for RMSD calculations was the minimized crystal structure of each.

PAGE 108

108 Figure 6 3. Hist ogram s of the RMSD of HIV 1 protease subtype B and the protease inhibitor bound. HIVPR refers to only the protein residues 1 198, and the protease inhibitors The measured 55N 55N distance profiles of each the apo and inhibitor bound forms of HIVPR ar e plotted in Figure 6 4. The interflap distances of the inhibitor bound form of HIVPR are shorter when compared to the apo, and this trend is consistent with previous studies. 171,186,189 The most probable, average, and range span of the interflap distances of all the inhibito rs are approximately the same, T able 6 1. U nlike previous studies 171,186,189,197 the apo distance d istribution profile is narrower than the ligand bound distance distribution in certain instances; possibly as a result of kinetic trapping. The distribution profile of the apo is expected to be broader because the flaps sample

PAGE 109

109 more conformations in the apo form of the enzyme. The apo distance distribution will be discussed in later sections. Blackburn et al. 171 had t wo main conclusions from their experimental DEER study on HIVPR : that of the n ine FDA approved inhibitors nelfinavir, indinavir, and atazanavir were the least effective at closing the flaps and that the percentage of time the flaps are closed correlates to the number of direct hydrogen bonds made between the active site residues and the inhibitor observed in the crystal structure (excluding hydrogen bonds between D25). However our results do not reflect the trend in Blackburn et al. 171 and we find that all of the inhibitors are effective in closing the flaps T he most probable distance for the inhibitors are at ~32 T able 6 1 which is more or less what the results of Blackburn et al. 171 show (excluding atazanavir, indinav ir, nelfinavir). The relatively good agreement for the other inhibitors is no surprise because a similar EPR study performed by Torbeev et al. 197 measured the interflap distances of the active (D25) and inact ive (N25) HIVPR using the protease inhibitors MVT 101, JG 365 and KV201, a nd there were only slight differences between two distance profiles Table 6 1. Distance profiles of inhibitor bound simulations Inhibitor Average Distance Most Probable Range Span Amprenavir 31.9 0.6 32.7 0.5 21.0 36.7 Atazanavir 32.0 0.6 32.7 0.5 21.8 35.7 Daru n a vir 31.7 0.6 32.7 0.5 20.3 35.7 Indinavir 31.5 0.6 32.3 0.5 21.5 36.1 Lopinavir 31.3 0.6 32.4 0.5 21.0 35.7 Nelfinavir 31.7 0.6 32.9 0.5 21.4 35.9 Ritonavir 31.8 0.5 32.6 0.4 21.8 36.5 Saquinavir 31.8 0.7 32.8 0.6 19.7 36.3 Tipranavir 30.8 1.0 32.7 0.8 18.0 35.6

PAGE 110

110 Figure 6 4 Histogram of N N dist ance distri bution profiles of a po form HIVPR subtype B compared to when it is bound to protease inhibitors. One explanation as to why the inhibitors nelfinavir, ata zanavir and indinavir do not close the flaps in the study performed by Blackburn et al. 171 could be that the inhbitors are more susceptible to the D25N mutation than the other inhibitors because it is the only stabi lizing mutation located in the active site. However previous chapters have demonstrated that mutations outside of the active site can alter the interactions of the inhibitor with residues in the active site thus changing the dynamics of the inhibitor. 6.3 .2 Apo MD Simulations In order to investigate the dynamics of the backbone atoms a RSMF calculation was performed, Figure 6 5 The flaps (residues 43 58) of the drug resistant constructs MDR769, and V6 exhibit a greater flexibility than the wild type. The differences between

PAGE 111

111 MDR769 and the wild type residues 1 30 and 60 99 is negligible However this is not the case for the for V6, which fluctuates more than the wild type in th ese regions. Figure 6 5 Plot of the RMSF of the wild type subtype B (black), M DR769 (red), V6 (blue). The RMSF of the two monomers have been averaged together. Figure 6 6 illustrates the interflap distance profiles for the three different constructs. The most probable distance of the wild type and V6 are approximatel y the same, 34. 7 T able 6 2. T he distance distribution profile breadth of V6 is broader than the wild type. The most proba ble distance of MDR769 is 35.7 which is l arger than the wild type Similar to V6, the distance profile breadth of MDR769 is also larger than th e wild type. The order of the distance profile breadths for the three constructs: V6 > MDR769 > wild type, agrees well with the trend in Figure 6 5 where V6 was the most flexible and the wild type was the least flexible. However, there is no direct correl ation between the most probable distance and the range span of the distance profiles. The average structures of the three constructs from MD simulations are depicted in Figure 6 7 The flaps of all three are in a semi open conformation The Ile50 s slightly larger in MDR769 when compared to the other two.

PAGE 112

112 Tab le 6 2 Distance profiles of the apo MD simulations Construct Average Distance Most Probable Range Span Wild type 34.9 0.5 34.7 0.4 24.1 41.5 V6 35.7 0.7 34.7 0.6 23.8 48.8 MDR769 35.7 0.6 35.3 0.5 28.0 50.0 Figure 6 6 Histogram of the N N distance profiles. Wild type subtype B, (black), MDR769 (red), V6 (blue) The results obtained from our MD simulations and those experimentally and computationall y determined by Galiano et al. exhibit different trends for the three constructs. The LAI, which in this case is equivalent to the wild type, has the broadest distance distribution profile, followed by V6 and MDR769 had the narrowest distribution profile. Galiano et al. 172 determined that the most probable distance for V6 is 34.9 0.7 LAI is 3 5.5 1 and MDR769 36.4 0.5 B oth studies find that MDR769 is more open than the wild type/LAI, which agrees with the apo crystal structure of MDR769 PDB ID 1TW7 190 when compared to a po structure of the wild type PDB ID 1HHP 128 There is no apo crystal structure for the construct V6 for comparison o nly inhibitor bound forms. The most probable distance calc ulated for V6 by both studies are fou nd to be in a good agreement ~34

PAGE 113

113 Figure 6 7 Schematic of the average structures of HIVPR: wild type subtype B (black), MDR769 (red), and V6 (blue) from spin labeled MD simulations. 6.3.3 Comparing the N A plot of the distance b etwee n residues 55 is given in Figure 6 8 The plot was to determine how accurately the spin label, could reproduce the trends seen in the backbone dynamics of the protease since that is essentially its aim. The flexibility of the spin label could be altered by its interactions with the neighboring side chains of other residues. The trend in Figure 6 8 is in good agreement with that depicted in Figure 6 6 The wild type has the narrowest distance profi le width and the V6 has the widest distance profil e However, there is a faint peak at 27 exhibited by V6 in Figure 6 8 indicating another conformation of the flaps that is not present in Figure 6 6 N onetheless, this confirms that the EPR measured profiles are dominated by the dynamics of the backbone and not the computational studies done by Galiano et al. 172

PAGE 114

114 Figure 6 8 Histogram of the 55 55 C type subtype B (black), MD769 (red), V6 (blue) Our results indicate that the drug resistance of V6 and MDR769 is due to the flaps showing a greater flexibility than the wi ld type, therefore allowing an escape mechanism for a n inhib itor. In addition a larger degree of flexibility could result in a larger entropic penalty for closing the flaps upon the binding of an inhibitor 6.4 Conclusions Our MD simulations show that the effect of the stabilizing mutations (Q7K, I33L, I63L, D25N, C 67A, C95A) on the EPR measured interflap distances are minimal for the inhibitor bound forms of HIVPR excluding the inhibitors indinavir, nelfinavir, and atazanavir The average most probable distances determined for the inhibitor bound forms of the prot ease are within ~.5 of each other. Maintaining D25, the active form of the HIVPR, causes all the inhibit ors to close the flaps tightly. The biggest difference between was between LAI and the wild type. Our measured spin label distance profile is much mo re narrower than that of the LAI The most probable distances differ by more ~1 of each other. Further investigation is needed to determine if our distance profile is a result of not sampling enough or if it is

PAGE 115

115 caused by the parameters of the spin label etc. However, both studies concluded that MDR769 is more open than the wild type/LAI and the dynamics of t he flaps could explain its drug resistance. Lastly, we showed that the N N distances are a good reflection of the backbone dynamics of the protease and that the difference between them is minimal for the constructs studied in this investigation.

PAGE 116

116 CHAPTER 7 PRELIMINARY STUDIES: EVALUATING THE EFFECTIVENESS OF ANTIRETROVIRAL PROTEASE INHIBITORS ON XMRV 7 .1 Introduction The newly discovered human retrovi rus xenotrophic murine l euk emia virus Related virus (XMRV) has been detected in patients with prostate cancer and chronic fatigue syndrome. 198 203 Furthermore, it is the first gammaretrovirus to be linked with human disea ses. However subsequent studies have failed to detect the virus in patients with prostate cancer or chronic fatigue syndrome despit e earlier reports 204 213 In addition, neither the origin nor the mode of transmission have been established this has lead to much debate and is an on going investigation in the scientific community Because XMRV is a retr ovirus much like HIV, the main targets of inhibition are its main three enzymes: reverse transcriptase, integrase and the protease. Targeting these enzymes has had much success in HIV antiretroviral therapy Various retroviral inhibitors have been tested o n XMRV and r altegravir, an integrase inhib itor designed for HIV, has been shown to be a potent inhibitor of XMRV. 214,215 The first x ray crystal structure of XMRV prot ease (apo) was solved in 2011 216 S ince then there have been several crystal structures of XMRV complexed with various antiretrovir al protease inhibitors. 217,218 The crystal structures revealed that XMRV is a homodimer, composed of 250 residues (125 residues per monomer) Figure 7 1 216 Two aspartates lie at the bottom of the active site that are responsible for its catalytic ability. The N and C termini when compared to other retropepsins are longer. 216 Unlike other retroviral proteases, the dime r of interface of XMRV is not composed of N and C termini that are interdigitated, but is instead composed of hairpins formed by strands close in proximity to the C termini. 216

PAGE 117

117 Figure 7 1. Schematic of XMRV protease. Highlighted are the catalytic aspartates Figure 7 2. Protease inhibitors: Amprenavir (HIV 1 protease), Pepstatin A (generic inhibitor of aspartic proteases), TL 3 (FI V protease) A variety of protease inhibitors have be en tested on XMRV but with limited success. 214, 217 In this study we perform MD with XMRV bound to the protease inhibitor s :

PAGE 118

118 amprenavir, pepstatin A, and TL 3 Figure 7 2 Amprenavir is a FDA approved protease inhibitor for HIV. TL 3 is a protease inhibitor that was developed to inhibit the FIV protease and it has been demonstrated to be effective against other retroviral proteases. 219 Pepstatin A is a nonspecific inhibitor, capable of inhibiting aspartic proteases. 220 The goal of th is study is to analyze the interactions each inhibitor makes in an to effort develop a potent inhibitor of XMRV protease. 7.2 Methods 7.2.1 Completing XMRV The coordinates for the wild type and mutant were derived from the x ray crystal structures PDB ID: 3SLZ TL 3, 3SM1 pepstatin A, 3SM2 ampre navir. In order to model the missing residues from the N and C terminus the crystal structure 3NR6 (apo) was utilized. A description of the equilibration procedure for these residues will be discuss ed in later sections. Pepstatin A was disordered in the cr ystal structure 3SM1. Pepstatin A was constructed using the visualization software Avogadro 193 Gaussian 09 194 was used to perform the geometry optimization of the pepstatin A in conjunction with the density functional M062X and 6 31+G basis set. The pepstatin A inhibitor was then docked into the crystal structure 3SM1 using the GLIDE 221 module of the Schrdinger Suite. 7.2.1 General Setup The missing hydrogen and heavy atoms were added using the Leap module in AmberTools13. The AMBER ff99SB 99 force field was used for the protein. The parameters for am prenavir, TL 3 and pepstatin A were built by using RESP charges generated from the antechamber module and the GAFF 139 force field. The catalytic aspartates were modele d using a mono protonated state

PAGE 119

119 The explicit solvent was modeled using a 15 TIP3P 103 solvent buffer encapsulating the pro tein in a truncated octahedron box and chloride ions were added to force neutrality. L ong range elect rostatics were accounted for using the PME 166 method utilizing a direct space and a van der Waals cutoff of 9 Langevin Dynamics was utilized t o maintain the temperature at 30 0 K the collision frequency was 1.0 ps 1 In or der to achieve a 2 fs ti me all atoms bonded to hydrogen atoms were constrained using the SHAKE 143 algorithm. 7.2.2 Molecular Dynamics The first step of the equilibration procedure consisted of minimizing the missing N and C terminus residues and the solvent followed by a 500 ps MD simulation while placing a positional restraint weight of 10 kca l/mol 2 on the rest of the solute The hydrogen atoms were minimized followed by the entire system. Each minimization process consisted of 500 steps of steepest descent followed and 10000 steps of conjugate gradient. The systems were heated linearly from 100 K to 300 K for 1 ns where a 5 kcal/mol 2 positional restraint was used on the solute. Following heating, the density was equilibrated for 500 ps. Next, five simulations were performed where the positional restraint was reduced from 5 kcal to .1 kcal/m ol 2 In last step of equilibration all restraints were removed and 5 ns of unrestrained MD was performed The production phase consisted of running a 20.5 ns simulation but only the last 20 ns were used for analysis Snapshots of the MD trajec tory were saved every 15 ps. 7.2. 3 Molecular Mechanics Poisson Boltzman Surface Area Calculations The MM PB SA calculations were performed using MMPBSA.py 122 in AmberTools13. The polar solvation energy was calculated using the Poisson Boltzmann Method in conjunction with the mbondi3 intrinsic radii. The non polar solvation energy

PAGE 120

120 was calculated using the LCPO 105 method. The normal mode approximation was used to account for the entropic contributions. Due to the high computational cost of normal mode calculations only 65 frames spaced every 975 ps were used. 7.3 Resu lts and Discussion 7.3.1 Flexibility Figure 7 3 Plot of the RMSF of XMRV bound to amprenavir, pepstatin A, TL 3. Monomer A residues (1 125) Monomer B residues (126 250) The RMSF of XMRV bound to the three different ligands is plotted in Figure 7 3. All t hree systems exhibit high fluctuations at the N and C termini, residues 1 10, 115 125 of each monomer. The flaps residues 48 66 of each monomer remain fairly rigid throughout the MD simulation The fluctuation of residue Thr75 is more pron o unced in the TL 3 simulation. The increas e in flexibility of Thr75 is attributed to the hydrogen bond between its side chain hydroxyl group and the backbone carbonyl of Lys100 being broken due to a small conformational change of residues 73 76 of monomer A in TL 3. A hi stogram of the RMSD of the inhibitors: amprenavir, pepstatin A and TL 3 is plotted in Figure 7 4. The minimized crystal structure was used as the reference. In the case of

PAGE 121

121 pepstatin A the minimized docked structure was the reference. Amprenavir is the mos t rigid of the three inhibitors and has two peaks at 0.8 and 1.2 The t wo peaks are caused by a change in the orientation of the benzyl ring that is solvent exposed and protrudes from the active site. TL 3 has two benzyl rings that protrude from both sid es of the active site and do not form any interactions with residues in the active site, consequently accounting for most of its flexibility. The flexibility of pepstatin A is due to it being a hexa peptide, which contains many more rotatable bonds than th e other two inhibitors. Figure 7 4. Histogram of the RMSD of the protease inhibitors: amprena vir, pepstatin A and TL 3. Figure 7 5 is a histogram of the RMSD distribution of the flap residues 48 66 of each monomer. The minimized crystal structures were used as the reference. The flaps of the XMRV (pepstatin A) are the most flexible of the three inhibitors. Its RMSD distribution is fairly broad and spans from approximately .7 1.6 The flaps of XMRV (amprenavir) and (TL 3) exhibit a unique trend; XMRV ( amprenavir) has the narrowest

PAGE 122

122 RMSD distribution but its maximum peak is shifted to the right of the maximum peak of XMRV (TL3) indicating that XMRV (amprenavir) is farther from its reference. Figure 7 5 Histogram of the RMSD of the flap residues 48 66 o f each monomer. 7.3.2 Hydrogen Bonding Analysis Because the ligands bind non covalently the direct hydrogen bonds and water mediated hydrogen bonds between the ligand and p rotein were calculated, T ables 7 1 7 2 and 7 3 TL 3 makes the most hydrogen bonds with residues in the active site. None of the ligands formed any water mediated hydrogen bonds with the active site residues > 5%. All of the ligands central hydroxyl groups(s) interact strongly with the deprotonated). TL 3 and pepstatin A form hydrogen bonds with the flap residues Gln55 of each monomer. Amprenavir does not form any hydrogen bonds with Gln55 but instead hydrogen bonds to the flap residue Ala57. Pepstatin A is the only inhibitor to make any substantial interactions with residues located at the base of the active site, Gly34. TL 3 hydrogen bonds to Gln36

PAGE 123

123 and Amprenavir hydrogen bonds to His37 which are residues located on the side of the active site Tabl e 7 1 Hydrogen bond analysis XMRV ( amprenavir) Amprenavir Acceptor Donor %Occupancy O APV O H 97 APV O Asp32 O H 79 APV O ALA57 N H 42 O APV N H 34 Table 7 2 Hydrogen bond analysis XMRV (pepstatin A) Pepstatin A Acceptor Donor %Occupancy O Pep O H 99 Pep O Asp32 O H 94 Pep O H 63 O Pep N H 51 Gly34 O Pep N H 45 Pep O Gln55 N H 30 Table 7 3 Hydrogen bond analysis XMRV (TL 3) TL 3 Acceptor Donor %Occupancy O TL3 O H 97 TL3 O H 82

PAGE 124

124 Table 7 3. Continued Acceptor Do nor %Occupancy Gln55 O TL3 N H 76 TL3 O Gln36 N H 74 O TL3 N H 71 TL3 O Asp32 O H 54 TL3 O Gln55 N H 53 TL3 O H 50 7.3.3 Molecular Mechanics Poisson Bol tzmann Surface Area Calculat ions TL 3 had the most favorable binding affinity o f the three inhibit ors, T able 7 4. The inhibitor TL 3 made the most hydrogen bond contacts with XMRV and as a result it has the highest H gas term. Amprenavir has the lowest entropic and desolvation penalty. The computed binding affinities in Table 7 4 do n ot coincide with K I values published in a recent kinetic study 217 Li et al. 217 found that amprenavir had the lowest K I value of the inhibitors used in their study. However our results do agree with Li et al. in that pepstatin A performs the worst of three inhibitors. The reason w hy results do not corroborate those published by Li et al. could be due to the following : inadequate sampling and the MD simulations need to be extended to a longer time scale, MMPBSA is an end state method and assumes that the re is little difference in t he ap o and bound forms of the enzyme or in spite of TL 3 and pepstatin A forming lots of interactions with the XMRV, the flaps of the protein are not flexible enough to accommodate the size of them and explains why the smaller inhibitor amprenavir prevails

PAGE 125

125 Table 7 4 MMPBSA results for the inhibitors: amprenavir, pepstatin A, and TL 3. All units are in kcal/mol System H gas G solvation T S nmode G bind Amprenavir 113.22 0 .22 56.35 0 .14 28.26 1.12 28.61 1.01 Pepstatin A 157.11 0 .18 94.71 0 15 39.78 0 .74 22.62 0 .68 TL 3 195.21 0 .27 108.38 0 .1 6 39.50 2.00 47.33 1.87 7.3.4 Potential Inhibitor Interactions Residues in the active site that a potential protease inhibitor of XMRV could interact with are illustrated in Figure 7 6 Although pepstatin A and TL 3 forms two hydrogen bonds with residues in the flaps, the flaps of XMRV (pepstatin A) when compared to XMRV (amprenavir) are more flexible and the difference between XMRV (amprenavir/TL 3) is minute. An inhibitor that targets ei ther one or both residues Gln55 and Ala57 will be sufficient enough to ensure the closure of the flaps. The positioning of the backbone carbonyl of Gly56 could potentially m ake it a good target but it is solvent exposed The scaffold HEA (amprenavir) and statine (pepstatin A) interacted the best with both catalytic aspartates the scaffold HEA might be a better choice to pursue to increase its selectivity therefore diminishing its toxicity but further investigations are warranted The interactions of t he central hydroxyl groups of TL 3 with Asp32 (deproton ated) are comparable for the two inhibitors but TL 3 interacted the least with Asp32 (protonated). None of the inhibitors formed a hydrogen bond directly or through a water molecule with the residue Cy s 88, which is located in the 86 90 loop of the active site.

PAGE 126

126 The side chain thiol group could act as either a hydrogen bond donor or hydrogen bond acceptor. Gln36 and His37 are resid ues that inhibitors could form interactions with either their backbone or side chains atoms. The backbone atoms of Gln36 and His37 residues might serve as a more attractive target because the side chains of both residues protrude from the active site and are solvent exposed. Although amprenavir and TL 3 form interactions with either Gln36 or His37, none of the inhibitors form interactions with both these residues. The potency of the inhibitor might be increased if it formed interactions with both residues. Figure 7 6 Schematic representation of XMRV protease Represented by spheres are the residues in the active site that potential inhibitors can target. Asp32 (blue), Gly34 (red), Gln36 (orange), His37 (yellow), Gln55 (green), Gly56 (purple), Ala57 (pink) and (Cys88 (silver) 7.4 Conclusions and Future Work Our MD simulation s provide preliminary information as to how to develop a potent inhibitor for XMRV. The HEA or the statine are good scaffolds that interact

PAGE 127

127 strongly with the catalytic aspartates. The flap res idues Gln 55 and Ala57 are good targets for inhibitors, Trp58 and Gly57 might be new considerations. Neither of the inhibitors targeted these residues. Altho ugh pepstatin A formed interactions with the backbone carbonyl of Gly34, located at the base in of the active site, i t was only for approximately 30 % of the time th erefore there is room for improvement. Residues that are located in close proximity to the catalytic aspartates generally have a lower mutational rate and therefore should be targeted. While the inhibitors TL 3 and pepstatin A make more hydrogen bonds with residues in the active site than amprenavir, the K I of the amprenavir was determined to be lower. One possibility might be that flaps of XMRV are flexible enough to accommodate the larger inhibitors TL 3 and pepstatin A and explains why they are not as ef fective inhibiting the virus Future work on XRMV protease will include: extending the time scale of the inhibitor bound simulations, using SMD simulations to gain insight on the dissociation pathway of an inhibitor, and performing long scale simulations o f the apo form of the XMRV protease to in order to study the dynamics of the flaps.

PAGE 128

128 CHAPTER 8 CONCLUSIONS AND FUTURE WORK Through the use of CpHMD simulations we determined that the two catalytic aspartates of the protease are mono protonated in apo as well as the bound form for the inhibitors used in this study. Furthermore, either of the catalytic aspartates can be protonated and the proton is easily transferred due to a low barrier hydrogen bond. Future work will include running these simulations wher e the solvent is instead modeled explicitly. We explained how the active site and non active site mutations G48T/L89M and N88D/L90M confer drug resistance against the protease inhibitor saquinavir in subtype B and C. In both cases the interac tions of saqui navir with other residues in the active site were altered Next simulations running only one mutation will provide insight as how much each mutation contributes to the overall resistance as well as its impact on the dynamics of the protease. More rigorous methods such as TI will be used to calculate the binding affinities. The effects of the stabilizing mutations were investigated on the EPR distance measured profiles. R estoring the HIVPR to i t s active form (N25D) caused all the inhibitors to close the flap s. The most probable distances of the inhibitor bound form of HIVPR were found to be in good agreement with those obtain ed from previous EPR studies where the stabilizing mutations were included. The wild type and LAI distance distributions did not corre la te, and further investigation is warranted. XMRV is a newly discovered re trovirus that has been linked to prostate cancer and chronic fatigue syndrome although it is still controversial. MD simulations with XMRV bound to protease inhibitors amprenavir, TL 3 and pepstatin A provide

PAGE 129

129 information on the how to design a potent inhibitor. With information obtained from the study a pharmacophore can be developed and used to screen possible inhibitors.

PAGE 130

130 APPENDIX HIV 1 PROTEASE SEQUENCES Table A 1. HIV 1 Protease Subtype B Sequence (Wild type) Residues 1 50 (first row) 51 99 (second row) PQITLWQRPL VTIKIGGQLK EALLDTGADD TVLEEMSLPG RWKPKMIGGI GGFIKVRQYD QILIEICGHK AIGTVLVGPT PVNIIGRNLL TQIGCTLNF Table A 2. HIV 1 50 (first row) 51 99 (second row) PQITLWKRPL VTIKIGGQLK EALLNTGADD TVIEEMSLPG RWKPKMIGGI GGFIKVRQYD QIIIEIAGHK AIGTVLVGPT PVNIIGRNLL TQIGATLNF Table A 3. HIV 1 Protease Subtype C Sequence (Wild type). Residues 1 50 (first row) 51 99 (second row) PQITLWQRPL VSIKVGGQIK EALLDTGADD TVLEEIALPG RWKPKMIGGI GGFIKVRQYD QILIEICGKK AIGTVLVGPT PVNIIGRNML TQLGCTLNF Table A 4. HIV 1 Protease MDR769 Sequence (Drug Resistant Construct). Residues 1 50 (first row) 51 99 (second row) PQITLWQRPI VTIKIGGQ LK EALLD TGADD TVLEEVNLPG RWKPKLIGGI GGFVKVRQYD QVPIEICGHK VIGTVLVGPT PANVIGRNLM TQIGCTLNF Table A 5. HIV 1 Protease V6 Sequence (Drug Reistant Construct). Residues 1 50 (first row) 51 99 (second row) PQITLWQRPL VTIKIGGQLR EALLDTGADD TIFEEISLPG RWKP KMIGGI GGFIKVRQYD QIPIEICGHK VIGTVLVGPT PANIIGRNLM TQIGCTLNF

PAGE 131

131 LIST OF REFERENCES (1) Weiss, A.; Hollander, H.; Stobo, J. Annu. Rev. Med. 1985 36 545 562. (2) Kuznetsov, Y. G.; Victoria, J. G.; Rob inson, W. E.; McPherson, A. J. Virol. 2003 77 11896 11909. (3) King, S. R. Ann. Emerg. Med. 1994 24 443 449. (4) Lee, C. H.; Saksela, K.; Mirza, U. A.; Chait, B. T.; Kuriyan, J. Cell 1996 85 931 942. (5) Collins, K. L.; Chen, B. K.; Kalams, S. A .; Walker, B. D.; Baltimore, D. Nature 1998 391 397 401. (6) Zapp, M. L.; Green, M. R. Nature 1989 342 714 716. (7) Ruben, S.; Perkins, A.; Purcell, R.; Joung, K.; Sia, R.; Burghoff, R.; Haseltine, W. A.; Rosen, C. A. J. Virol. 1989 63 1 8. (8) Moon, H. S.; Yang, J. S. Mol. Cells 2006 21 7 20. (9) Rogel, M. E.; Wu, L. I.; Emerman, M. J. Virol. 1995 69 882 888. (10) Whitford, D. Proteins: structure and function ; J. Wiley & Sons, 2005. (11) Mariani, R.; Chen, D.; Schrfelbauer, B.; Navar ro, F.; Knig, R.; Bollman, B.; Mnk, C.; Nymark McMahon, H.; Landau, N. R. Cell 2003 114 21 31. (12) Conticello, S. G.; Harris, R. S.; Neuberger, M. S. Curr. Biol. 2003 13 2009 2013. (13) Fujita, K.; Omura, S.; Silver, J. J. Gen. Virol. 1997 78 619 625. (14) Robertson, D. L.; Anderson, J. P.; Bradac, J. A.; Carr, J. K.; Foley, B.; Funkhouser, R. K.; Gao, F.; Hahn, B. H.; Kalish, M. L.; Kuiken, C.; Learn, G. H.; Leitner, T.; McCutchan, F.; Osmanov, S.; Peeters, M.; Pieniazek, D.; Salminen, M.; S harp, P. M.; Wolinsky, S.; Korber, B. Science 2000 288 55d. (15) Grtler, L. G.; Hauser, P. H.; Eberle, J.; von Brunn, A.; Knapp, S.; Zekeng, L.; Tsague, J. M.; Kaptue, L. J. Virol. 1994 68 1581 1585. (16) Simon, F.; Mauclre, P.; Roques, P.; Louss ert Ajaka, I.; Mller Trutwin, M. C.; Saragosti, S.; Georges Courbot, M. C.; Barr Sinoussi, F.; Brun Vzinet, F. Nat. Med. 1998 4 1032 1037. (17) Plantier, J. C.; Leoz, M.; Dickerson, J. E.; De Oliveira, F.; Cordonnier, F.; Leme, V.; Damond, F.; Robe rtson, D. L.; Simon, F. Nat. Med. 2009 15 871 872.

PAGE 132

132 (18) Penny, M. A.; Thomas, S. J.; Douglas, N. W.; Ranjbar, S.; Holmes, H.; Daniels, R. S. AIDS Res. Hum. Retroviruses 1996 12 741 747. (19) Archer, J.; Robertson, D. L. AIDS 2007 21 1693 1700. ( 20) Hemelaar, J.; Gouws, E.; Ghys, P. D.; Osmanov, S. AIDS Lond. Engl. 2006 20 W13 23. (21) Baeten, J. M.; Chohan, B.; Lavreys, L.; Chohan, V.; McClelland, R. S.; Certain, L.; Mandaliya, K.; Jaoko, W.; Overbaugh, J. J. Infect. Dis. 2007 195 1177 118 0. (22) Chan, D. C.; Fass, D.; Berger, J. M.; Kim, P. S. Cell 1997 89 263 273. (23) Pantophlet, R.; Burton, D. R. Annu. Rev. Immunol. 2006 24 739 769. (24) Dalgleish, A. G.; Beverley, P. C. L.; Clapham, P. R.; Crawford, D. H.; Greaves, M. F.; We iss, R. A. Nature 1984 312 763 767. (25) Coakley, E.; Petropoulos, C. J.; Whitcomb, J. M. Curr. Opin. Infect. Dis. 2005 18 9 15. (26) Deng, H. K.; Unutmaz, D.; KewalRamani, V. N.; Littman, D. R. Nature 1997 388 296 300. (27) Liao, F.; Alkhatib, G.; Peden, K. W. C.; Sharma, G.; Berger, E. A.; Farber, J. M. J. Exp. Med. 1997 185 2015 2023. (28) Chan, D. C.; Kim, P. S. Cell 1998 93 681 684. (29) Robertson, D. L.; Hahn, B. H.; Sharp, P. M. J. Mol. Evol. 1995 40 249 259. (30) Pollard, V. W.; Malim, M. H. Annu. Rev. Microbiol. 1998 52 491 532. (31) Zheng, Y. H.; Lovsin, N.; Peterlin, B. M. Immunol. Lett. 2005 97 225 234. (32) Hiscott, J.; Kwon, H.; Gnin, P. J. Clin. Invest. 2001 107 143 151. (33) Bray, M.; Prasad, S.; Dubay, J. W.; Hunter, E.; Jeang, K. T.; Rekosh, D.; Hammarskjld, M. L. Proc. Natl. Acad. Sci. U. S. A. 1994 91 1256 1260. (34) Witte, O. N.; Baltimore, D. J. Virol. 1978 26 750 761. (35) Crawford, S.; Goff, S. P. J. Virol. 1985 53 899 907.

PAGE 133

133 (36) Kohl, N. E.; Emini, E. A.; Schleif, W. A.; Davis, L. J.; Heimbach, J. C.; Dixon, R. A.; Scolnick, E. M.; Sigal, I. S. Proc. Natl. Acad. Sci. U. S. A. 1988 85 4686 4690. (37) Navia, M. A.; Fitzgerald, P. M. D.; McKeever, B. M.; Leu, C. T.; Heimbach, J. C.; Herb er, W. K.; Sigal, I. S.; Darke, P. L.; Springer, J. P. Nature 1989 337 615 620. (38) Wlodawer, A.; Miller, M.; Jasklski, M.; Sathyanarayana, B. K.; Baldwin, E.; Weber, I. T.; Selk, L. M.; Clawson, L.; Schneider, J.; Kent, S. B. Science 1989 245 616 621. (39) Miller, M.; Schneider, J.; Sathyanarayana, B. K.; Toth, M. V.; Marshall, G. R.; Clawson, L.; Selk, L.; Kent, S. B.; Wlodawer, A. Science 1989 246 1149 1152. (40) Pearl, L.; Blundell, T. FEBS Lett. 1984 174 96 101. (41) Strisovsky, K.; T essmer, U.; Langner, J.; Konvalinka, J.; Krusslich, H. G. Protein Sci. Publ. Protein Soc. 2000 9 1631 1641. (42) Ingr, M.; Uhlkov, T.; Strsovsk, K.; Majerov, E.; Konvalinka, J. Protein Sci. Publ. Protein Soc. 2003 12 2173 2182. (43) Ishima, R .; Freedberg, D. I.; Wang, Y. X.; Louis, J. M.; Torchia, D. A. Struct. Lond. Engl. 1993 1999 7 1047 1055. (44) Rick, S. W.; Erickson, J. W.; Burt, S. K. Proteins 1998 32 7 16. (45) Trylska, J.; Tozzini, V.; Chang, C. A.; McCammon, J. A. Biophys. J. 2007 92 4179 4187. (46) Hornak, V.; Okur, A.; Rizzo, R. C.; Simmerling, C. Proc. Natl. Acad. Sci. U. S. A. 2006 103 915 920. (47) Freedberg, D. I.; Ishima, R.; Jacob, J.; Wang, Y. X.; Kustanovich, I.; Louis, J. M.; Torchia, D. A. Protein Sci. 2002 11 221 232. (48) Scott, W. R.; Schiffer, C. A. Struct. Lond. Engl. 1993 2000 8 1259 1265. (49) Meagher, K. L.; Carlson, H. A. Proteins Struct. Funct. Bioinforma. 2005 58 119 125. (50) Li, D.; Ji, B.; Hwang, K.; Huang, Y. J. Phys. Chem. B 2010 114 3060 3069.

PAGE 134

134 (51) De Oliveira, Tulio, T.; Engelbrecht, S.; Janse van Rensburg, E.; Gordon, M.; Bishop, K.; zur Megede, J.; Barnett, S. W.; Cassol, S. J Virol 2003 77 9422 9430. (52) Meek, T. D.; Rodriguez, E. J.; Angeles, T. S. Methods Enzymol. 1 994 241 127 156. (53) Hyland, L. J.; Tomaszek, T. A.; Meek, T. D. Biochemistry (Mosc.) 1991 30 8454 8463. (54) Ogden, R. C.; Flexner, C. W. Protease Inhibitors in AIDS Therapy ; Taylor & Francis, 2001. (55) Ratner, L.; Haseltine, W.; Patarca, R.; Livak, K. J.; Starcich, B.; Josephs, S. F.; Doran, E. R.; Rafalski, J. A.; Whitehorn, E. A.; Baumeister, K.; Ivanoff, L.; Petteway, S. R.; Pearson, M. L.; Lautenberger, J. A.; Papas, T. S.; Ghrayeb, J.; Chang, N. T.; Gallo, R. C.; Wong Staal, F. Nature 198 5 313 277 284. (56) Toh, H.; Ono, M.; Saigo, K.; Miyata, T. Nature 1985 315 691 691. (57) Kramer, R. A.; Schaber, M. D.; Skalka, A. M.; Ganguly, K.; Wong Staal, F.; Reddy, E. P. Science 1986 231 1580 1584. (58) Sanchez Pescador, R.; Power, M. D .; Barr, P. J.; Steimer, K. S.; Stempien, M. M.; Brown Shimer, S. L.; Gee, W. W.; Renard, A.; Randolph, A.; Levy, J. A.; Et, A. Science 1985 227 484 492. (59) Wolfenden, R. Annu. Rev. Biophys. Bioeng. 1976 5 271 306. (60) Dreyer, G. B.; Metcalf, B. W.; Tomaszek, T. A.; Carr, T. J.; Chandler, A. C.; Hyland, L.; Fakhoury, S. A.; Magaard, V. W.; Moore, M. L.; Strickler, J. E. Proc. Natl. Acad. Sci. 1989 86 9752 9756. (61) Moore, M. L.; Bryan, W. M.; Fakhoury, S. A.; Magaard, V. W.; Huffman, W. F.; Dayton, B. D.; Meek, T. D.; Hyland, L.; Dreyer, G. B.; Metcalf, B. W.; Strickler, J. E.; Gorniak, J. G.; Debouck, C. Biochem. Biophys. Res. Commun. 1989 159 420 425. (62) Tomasselli, A. G.; Olsen, M. K.; Hui, J. O.; Staples, D. J.; Sawyer, T. K.; Heinr ikson, R. L.; Tomich, C. S. C. Biochemistry (Mosc.) 1990 29 264 269. (63) Velazquez Campoy, A.; Todd, M. J.; Freire, E. Biochemistry (Mosc.) 2000 39 2201 2207. (64) Velazquez Campoy A.; Kiso Y.; Freire E. Arch. Biochem. Biophys. 2001 390 169 175.

PAGE 135

135 (65) King, N. M.; Prabu Jeyabalan, M.; Bandaranayake, R. M.; Nalam, M. N. L.; ACS Chem. Biol. 2012 7 1536 1546. (66) Boggetto, N.; Reboud Ravaux, M. Biol. Chem. 2002 383 1321 1324. (67) Breccia, P.; Boggetto, N.; Prez Fernndez, R.; Van Gool, M.; Takahashi, M.; Ren, L.; Prados, P.; Badet, B.; Reboud Ravaux, M.; de Mendoza, J. J. Med. Chem. 2003 46 5196 5207. (68) Perryman, A. L.; Lin, J.; McCammon, J. A. Biopolymers 200 6 82 272 284. (69) Perryman, A. L.; Zhang, Q.; Soutter, H. H.; Rosenfeld, R.; McRee, D. E.; Olson, A. J.; Elder, J. E.; David Stout, C. Chem. Biol. Drug Des. 2010 75 257 268. (70) Genoni, A.; Morra, G.; Merz, K. M.; Colombo, G. Biochemistry (Mosc.) 2010 49 4283 4295. (71) Ishima, R.; Ghirlando, R.; Tzsr, J.; Gronenborn, A. M.; Torchia, D. A.; Louis, J. M. J. Biol. Chem. 2001 276 49110 49116. (72) Weber, I. T. J. Biol. Chem. 1990 265 10492 10496. (73) Todd, M. J.; Semo, N.; Freire, E. J Mol. Biol. 1998 283 475 488. (74) Gustchina, A.; Weber, I. T. Proteins Struct. Funct. Bioinforma. 1991 10 325 339. (75) Zhang, Z. Y.; Poorman, R. A.; Maggiora, L. L.; Heinrikson, R. L.; Kzdy, F. J. J. Biol. Chem. 1991 266 15591 15594. (76) S chramm, H. J.; Nakashima, H.; Schramm, W.; Wakayama, H.; Yamamoto, N. Biochem. Biophys. Res. Commun. 1991 179 847 851. (77) Zutshi, R.; Franciskovich, J.; Shultz, M.; Schweitzer, B.; Bishop, P.; Wilson, M.; Chmielewski, J. J. Am. Chem. Soc. 1997 119 4841 4845. (78) Rhee, S. Y.; Gonzales, M. J.; Kantor, R.; Betts, B. J.; Ravela, J.; Shafer, R. W. Nucleic Acids Res. 2003 31 298 303. (79) Rhee, S. Y.; Kantor, R.; Katzenstein, D. A.; Camacho, R.; Morris, L.; Sirivichayakul, S.; Jorgensen, L.; Brigid o, L. F.; Schapiro, J. M.; Shafer, R. W. AIDS Lond. Engl. 2006 20 643 651. (80) Shafer, R. W. J. Infect. Dis. 2006 194 S51 S58.

PAGE 136

136 (81) Maguire, M. F.; Guinea, R.; Griffin, P.; Macmanus, S.; Elston, R. C.; Wolfram, J.; Richards, N.; Hanlon, M. H.; Por ter, D. J. T.; Wrin, T.; Parkin, N.; Tisdale, M.; Furfine, E.; Petropoulos, C.; Snowden, B. W.; Kleim, J. P. J. Virol. 2002 76 7398 7406. (82) B. M.; Sleasman, J. W.; Go odenow, M. M. Virology 2008 378 272 281. (83) Rose, R. B.; Craik, C. S.; Stroud, R. M. Biochemistry (Mosc.) 1998 37 2607 2621. (84) Wilson, S. I.; Phylip, L. H.; Mills, J. S.; Gulnik, S. V.; Erickson, J. W.; Dunn, B. M.; Kay, J. Biochim. Biophys. A cta BBA Protein Struct. Mol. Enzym. 1997 1339 113 125. (85) Lech, W. J.; Wang, G.; Yang, Y. L.; Chee, Y.; Dorman, K.; McCrae, D.; Lazzeroni, L. C.; Erickson, J. W.; Sinsheimer, J. S.; Kaplan, A. H. J. Virol. 1996 70 2038 2043. (86) Weber, I. T.; Agniswamy, J. Viruses 2009 1 1110 1136. (87) Johnson, V. A.; Calvez, V.; Gunthard, H. F.; Paredes, R.; Pillay, D.; Shafer, R. W.; Wensing, A. M.; Richman, D. D. Top. Antivir. Med. 2013 21 6 14. (88) King, N. M.; Prabu Jeyabalan, M.; Nalivaika, E. A .; Schiffer, C. A. Chem. Biol. 2004 11 1333 1338. (89) Foulkes Murzycki, J. E.; Scott, W. R. P.; Schiffer, C. A. Structure 2007 15 225 233. (90) De Vera, I. M. S.; Smith, A. N.; Dancel, M. C. A.; Huang, X.; Dunn, B. M.; Fanucci, G. E. Biochemistry (Mosc.) 2013 52 3278 3288. (91) Alder, B. J.; Wainwright, T. E. J. Chem. Phys. 1957 27 1208. (92) Leach, A. Molecular Modelling: Principles and Applications ; 2nd ed.; Prentice Hall, 2001. (93) Frenkel, D.; Smit, B. Understanding Molecular Simulat ion, Second Edition: From Algorithms to Applications ; 2nd ed.; Academic Press, 2001. (94) Cramer, C. J. Essentials of Computational Chemistry: Theories and Models ; 2nd ed.; Wiley, 2004. (95) Verlet, L. Phys. Rev. 1967 159 98. (96) Hockney, R. Method s Comput Phys 1970 9 136 211.

PAGE 137

137 (97) Swope, W.; Andersen, H.; Berens, P.; Wilson, K. J. Chem. Phys. 1982 76 649, 637. (98) Beeman, D. J. Comput. Phys. 1976 20 130 139. (99) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Proteins 2006 65 712 725. (100) MacKerell; Bashford, D.; Bellott; Dunbrack; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; Joseph McCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen D. T.; Prodhom, B.; Reiher, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wirkiewicz Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998 102 3586 3616. (101) Jorgensen, W. L.; Tirado Rives, J. J. Am. Chem. Soc. 1988 110 1657 1666. (102) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem. 1987 91 6269 6271. (103) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983 79 926. (104) Mahone y, M. W.; Jorgensen, W. L. J. Chem. Phys. 2000 112 8910 8922. (105) Weiser, J.; Shenkin, P. S.; Still, W. C. J. Comput. Chem. 1999 20 217 230. (106) Honig, B.; Nicholls, A. Science 1995 268 1144 1149. (107) Onufriev, A.; Bashford, D.; Case, D. A. J. Phys. Chem. B 2000 104 3712 3720. (108) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am. Chem. Soc. 1990 112 6127 6129. (109) Jarzynski, C. Phys. Rev. Lett. 1997 78 2690 2693. (110) Swails, J. M.; Roitberg, A. E. J. Chem. Theory Comput. 2012 8 4393 4404. (111) Baptista, A. M.; Martel, P. J.; Petersen, S. B. Proteins Struct. Funct. Bioinforma. 1997 27 523 544. (112) Baptista, A. M.; Teixeira, V. H.; Soares, C. M. J. Chem. Phys. 2002 117 4184 4200. (113) Lee, A. C.; Crippen, G. M. J. Chem. Inf. Model. 2009 49 2013 2033.

PAGE 138

138 (114) Mongan, J.; Case, D. A.; McCammon, J. A. J. Comput. Chem. 2004 25 2038 2048. (115) Metropolis, N.; Rosenbluth, A. W.; Rosenbluth, M. N.; Teller, A. H.; Teller, E. J. Chem. Phys. 1953 21 1087 1092. (116) Sugita, Y.; Okamoto, Y. Chem. Phys. Lett. 1999 314 141 151. (117) Proteins Struct. Funct. Bioinforma. 2011 79 3420 3436. (118) Hamelberg, D.; Mongan, J.; McCammon, J. A. J. Chem. Phy s. 2004 120 11919 11929. (119) Pierce, L. C. T.; Salomon Ferrer, R.; Augusto F. de Oliveira, C.; McCammon, J. A.; Walker, R. C. J. Chem. Theory Comput. 2012 8 2997 3002. (120) Kollman, P. Chem. Rev. 1993 93 2395 2417. (121) Zwanzig, R. W. J. Ch em. Phys. 1954 22 1420. (122) Miller, B. R.; McGee, T. D.; Swails, J. M.; Homeyer, N.; Gohlke, H.; Roitberg, A. E. J. Chem. Theory Comput. 2012 8 3314 3321. (123) Nicholson, L. K.; Yamazaki, T.; Torchia, D. A.; Grzesiek, S.; Bax, A.; Stahl, S. J.; Kaufman, J. D.; Wingfield, P. T.; Lam, P. Y.; Jadhav, P. K. Nat. Struct. Biol. 1995 2 274 280. (124) Chen, X.; Tropsha, A. J. Med. Chem. 1995 38 42 48. (125) Xie, D.; Gulnik, S.; Collins, L.; Gustchina, E.; Suvorov, L.; Erickson, J. W. Biochemistry (Mosc.) 1997 36 16166 16172. (126) Wang, W.; Kollman, P. A. J. Mol. Biol. 2000 303 567 582. (127) Ido, E.; Han, H. P.; Kezdy, F. J.; Tang, J. J. Biol. Chem. 1991 266 24359 24366. (128) Spinelli, S.; Liu, Q. Z.; Alzari, P. M.; Hirel, P. H.; Pol jak, R. J. Biochimie 1991 73 1391 1396. (129) Shen, C. H.; Wang, Y. F.; Kovalevsky, A. Y.; Harrison, R. W.; Weber, I. T. FEBS J. 2010 277 3699 3714. (130) Muzammil, S.; Armstrong, A. A.; Kang, L. W.; Jakalian, A.; Bonneau, P. R.; Schmelmer, V.; Amz el, L. M.; Freire, E. J Virol 2007 81 5144 5154.

PAGE 139

139 (131) Agniswamy, J.; Shen, C. H.; Aniana, A.; Sayer, J. M.; Louis, J. M.; Weber, I. T. Biochemistry (Mosc.) 2012 51 2819 2828. (132) Munshi, S.; Chen, Z.; Li, Y.; Olsen, D. B.; Fraley, M. E.; Hungate, R. W.; Kuo, L. C. Acta Crystallogr. D Biol. Crystallogr. 1998 54 1053 1060. (133) Kaldor, S. W.; Kalish, V. J.; Davies, J. F.; Shetty, B. V.; Fritz, J. E.; Appelt, K.; Burgess, J. A.; Campanale, K. M.; Chirgadze, N. Y.; Clawson, D. K.; Dressman, B. A. ; Hatch, S. D.; Khalil, D. A.; Kosa, M. B.; Lubbehusen, P. P.; Muesing, M. A.; Patick, A. K.; Reich, S. H.; Su, K. S.; Tatlock, J. H. J. Med. Chem. 1997 40 3979 3985. (134) Kempf, D. J.; Marsh, K. C.; Denissen, J. F.; McDonald, E.; Vasavanonda, S.; Fle ntge, C. A.; Green, B. E.; Fino, L.; Park, C. H.; Kong, X. P. Proc. Natl. Acad. Sci. U. S. A. 1995 92 2484 2488. (135) Lam, P. Y. S.; Ru, Y.; Jadhav, P. K.; Aldrich, P. E.; DeLucca, G. V.; Eyermann, C. J.; Chang, C. H.; Emmett, G.; Holler, E. R.; Danek er, W. F.; Li, L.; Confalone, P. N.; McHugh, R. J.; Han, Q.; Li, R.; Markwalder, J. A.; Seitz, S. P.; Sharpe, T. R.; Bacheler, L. T.; Rayner, M. M.; Klabe, R. M.; Shum, L.; Winslow, D. L.; Kornhauser, D. M.; Jackson, D. A.; Erickson Viitanen, S.; Hodge, C. N. J. Med. Chem. 1996 39 3514 3525. (136) Adachi, M.; Ohhara, T.; Kurihara, K.; Tamada, T.; Honjo, E.; Okazaki, N.; Arai, S.; Shoyama, Y.; Kimura, K.; Matsumura, H.; Sugiyama, S.; Adachi, H.; Takano, K.; Mori, Y.; Hidaka, K.; Kimura, T.; Hayashi, Y.; Kiso, Y.; Kuroki, R. Proc. Natl. Acad. Sci. 2009 106 4641 4646. (137) Kear, J. L.; Blackburn, M. E.; Veloro, A. M.; Dunn, B. M.; Fanucci, G. E. J. Am. Chem. Soc. 2009 131 14650 14651. (138) Case, D. A.; Cheatham, T. E.; Darden, T.; Gohlke, H.; Luo, R.; Merz, K. M.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods, R. J. J. Comput. Chem. 2005 26 1668 1688. (139) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004 25 1157 1174. (140) Jakalian, A.; Jack, D. B. ; Bayly, C. I. J. Comput. Chem. 2002 23 1623 1641. (141) Onufriev, A.; Bashford, D.; Case, D. A. Proteins 2004 55 383 394. (142) Sindhikara, D. J.; Kim, S.; Voter, A. F.; Roitberg, A. E. J. Chem. Theory Comput. 2009 5 1624 1631.

PAGE 140

140 (143) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977 23 327 341. (144) Sindhikara, D.; Meng, Y.; Roitberg, A. E. J. Chem. Phys. 2008 128 024103 024103 10. (145) Roe, D. R.; Cheatham, T. E. J. Chem. Theory Comput. 2013 9 3084 3095. (146 ) Trylska, J.; Antosiewicz, J.; Geller, M.; Hodge, C. N.; Klabe, R. M.; Head, M. S.; Gilson, M. K. Protein Sci. Publ. Protein Soc. 1999 8 180 195. (147) Czodrowski, P.; Sotriffer, C. A.; Klebe, G. J. Chem. Inf. Model. 2007 47 1590 1598. (148) Pian a, S.; Carloni, P. Proteins Struct. Funct. Bioinforma. 2000 39 26 36. (149) Smith, R.; Brereton, I. M.; Chai, R. Y.; Kent, S. B. Nat. Struct. Biol. 1996 3 946 950. (150) Mardis, K. L.; Luo, R.; Gilson, M. K. J. Mol. Biol. 2001 309 507 517. (151) Antosiewicz, J.; Briggs, J. M.; Elcock, A. H.; Gilson, M. K.; McCammon, J. A. J. Comput. Chem. 1996 17 1633 1644. (152) Antosiewicz, J.; McCammon, J. A.; Gilson, M. K. J. Mol. Biol. 1994 238 415 436. (153) Yamazaki, T.; Nicholson, Linda; Torchia, Dennis; WIngfield, Paul; Stahl, Stephen; Kaufman, Joshua; Eyermann, Charles; Hodge, C. Nicholas; Lam, Patrick; Ru, Yu; Jadhav, Prabhakar; Chang, Chong Hwan; Weber, Patricia J Am Chem Soc 1994 116 10791 10792. (154) Schutz, C. N.; Warshel, A. Proteins 2001 44 400 417. (155) Wang, Y. X.; Freedberg, D. I.; Yamazaki, T.; Wingfield, P. T.; Stahl, S. J.; Kaufman, J. D.; Kiso, Y.; Torchia, D. A. Biochemistry (Mosc.) 1996 35 9945 9950. (156) Wittayanarakul, K.; Hannongbua, S.; Feig, M. J. Comput. Chem. 2008 29 673 685. (157) Geller, M.; Miller, M.; Swanson, S. M.; Maizel, J. Proteins Struct. Funct. Bioinforma. 1997 27 195 203. (158) Harte, W. E.; Beveridge, D. L. J. Am. Chem. Soc. 1993 115 3883 3886.

PAGE 141

141 (159) Ky Youb Nam; Chang, B. H.; Han, C. K.; Ahn, S. K.; No, K. T. Bull. Korean Chem. Soc. 2003 24 817 823. (160) Das, A.; Mahale, S.; Prashar, V.; Bihani, S.; Ferrer, J. L.; Hosur, M. V. J. Am. Chem. Soc. 2010 132 6366 6373. (161) Northrop, D. B. Accounts Chem. Res. 2001 34 790 797. ( 162) Jacobsen, H.; Hnggi, M.; Ott, M.; Duncan, I. B.; Owen, S.; Andreoni, M.; Vella, S.; Mous, J. J. Infect. Dis. 1996 173 1379 1387. (163) Tie, Y.; Kovalevsky, A. Y.; Boross, P.; Wang, Y. F.; Ghosh, A. K.; Tozser, J.; Harrison, R. W.; Weber, I. T. P roteins Struct. Funct. Bioinforma. 2007 67 232 242. (164) Ros, J. R.; Salto, R.; Craik, C. S. J. Biol. Chem. 1993 268 11939 11945. (165) Mildner, A. M.; Rothrock, D. J.; Leone, J. W.; Bannow, C. A.; Lull, J. M.; Reardon, I. M.; Sarcich, J. L.; How e, W. J.; Tomich, C. S.; Smith, C. W. Biochemistry (Mosc.) 1994 33 9405 9413. (166) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993 98 10089. (167) Nguyen, H.; Roe, D. R.; Simmerling, C. J. Chem. Theory Comput. 2013 9 2020 2034. (168) Be rendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984 81 3684 3690. (169) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996 14 33 38, 27 28. (170) Case, D. A.; Darden, T. A.; Cheatham, T. E. I.; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Walker, R. C.; Zhang, W.; Merz, K. M.; Roberts, B. P.; Wang, B.; Hayik, S.; Roitberg, A. E.; Seabra, G.; Kolossvry, I.; Wong, K. F.; Paesani, F.; Vanicek, J.; Wu, X.; Brozell, S. R.; Steinbrecher, T.; Gohlke, H.; Cai, Q.; Ye, X.; Wang, J.; Hsieh, M. J.; Cui, G.; Roe, D. R.; Mathews, D. H.; Seetin, M. G.; Sagui, C.; Babin, V.; Luchko, T.; Gusarov, S.; Kovalenko, A.; Kollman, P. A. AMBER 12 ; AMBER; University of California, San Francisco, 2012. (171) B lackburn, M. E.; Veloro, A. M.; Fanucci, G. E. Biochemistry (Mosc.) 2009 48 8765 8767. (172) Galiano, L.; Ding, F.; Veloro, A. M.; Blackburn, M. E.; Simmerling, C.; Fanucci, G. E. J. Am. Chem. Soc. 2009 131 430 431.

PAGE 142

142 (173) Hong, L.; Zhang, X. C.; Ha rtsuck, J. A.; Tang, J. Protein Sci. 2000 9 1898 1904. (174) Stoica, I.; Sadiq, S. K.; Coveney, P. V. J. Am. Chem. Soc. 2008 130 2639 2648. (175) Sanches, M.; Krauchenco, S.; Martins, N. H.; Gustchina, A.; Wlodawer, A.; Polikarpov, I. J. Mol. Biol. 2007 369 1029 1040. (176) Hou, T.; McLaughlin, W. A.; Wang, W. Proteins 2008 71 1163 1174. (177) Coman, R. M.; Robbins, A. H.; Fernandez, M. A.; Gilliland, C. T.; Sochet, A. A.; Goodenow, M. M.; McKenna, R.; Dunn, B. M. Biochemistry (Mosc.) 2008 47 731 743. (178) Matsuyama, S.; Aydan, A.; Ode, H.; Hata, M.; Sugiura, W.; Hoshino, T. J. Phys. Chem. B 2010 114 521 530. (179) Liu, X.; Dai, Q.; Li, L.; Xiu, Z. J. Mol. Struct. 2011 986 30 38. (180) Coman, R. M.; Robbins, A. H.; Goodenow, M. M .; Dunn, B. M.; McKenna, R. Acta Crystallogr. D Biol. Crystallogr. 2008 D64 754 763. (181) Mitsuya, Y.; Winters, M. A.; Fessel, W. J.; Rhee, S. Y.; Hurley, L.; Horberg, M.; Schiffer, C. A.; Zolopa, A. R.; Shafer, R. W. AIDS Res. Hum. Retroviruses 2006 22 1300 1305. (182) Mahalingam, B.; Boross, P.; Wang, Y.; Louis, J. M.; Fischer, C. C.; Tozser, J.; Harrison, R. W.; Weber, I. T. Proteins Struct. Funct. Bioinforma. 2002 48 107 116. (183) Kovalevsky, A. Y.; Tie, Y.; Liu, F.; Boross, P. I.; Wang, Y F.; Leshchenko, S.; Ghosh, A. K.; Harrison, R. W.; Weber, I. T. J. Med. Chem. 2006 49 1379 1387. (184) Bihani, S. C.; Das, A.; Prashar, V.; Ferrer, J. L.; Hosur, M. V. Biochem. Biophys. Res. Commun. 2009 389 295 300. (185) Fanucci, G. E.; Cafiso, D. S. Curr. Opin. Struct. Biol. 2006 16 644 653. (186) Galiano, L.; Bonora, M.; Fanucci, G. E. J. Am. Chem. Soc. 2007 129 11004 11005. (187) Pannier, M.; Veit, S.; Godt, A.; Jeschke, G.; Spiess, H. W. J. Magn. Reson. San Diego Calif 1997 2000 142 331 340.

PAGE 143

143 (188) Jeschke, G. Chemphyschem Eur. J. Chem. Phys. Phys. Chem. 2002 3 927 932. (189) Ding, F.; Layten, M.; Simmerling, C. J. Am. Chem. Soc. 2008 130 7184 7185. (190) Martin, P.; Vickrey, J. F.; Proteasa, G.; Jimenez, Y. L.; Wawrzak, Z. ; Winters, M. A.; Merigan, T. C.; Kovari, L. C. Struct. Lond. Engl. 1993 2005 13 1887 1895. (191) Clemente, J. C.; Moose, R. E.; Hemrajani, R.; Whitford, L. R. S.; Govindasamy, L.; Reutzel, R.; McKenna, R.; Agbandje McKenna, M.; Goodenow, M. M.; Dunn, B. M. Biochemistry (Mosc.) 2004 43 12141 12151. (192) Kim, E. E.; Baker, C. T.; Dwyer, M. D.; Murcko, M. A.; Rao, B. G.; Tung, R. D.; Navia, M. A. J. Am. Chem. Soc. 1995 117 1181 1182. (193) Hanwell, M. D.; Curtis, D. E.; Lonie, D. C.; Vandermeersc h, T.; Zurek, E.; Hutchison, G. R. J. Cheminformatics 2012 4 17. (194) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A.F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, Jr., J. A.; Peralta, J. E.; Og liaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J; Dapprich, S.; Daniel s, A. D.; Farkas, .; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian09 (195) Columbus, L.; Hubbell, W. L. Trends Biochem. Sci. 2002 27 288 295. (196) Sezer, D.; Freed, J. H.; Roux, B. J. Am. Chem. Soc. 2009 131 2597 2605. (197 ) Torbeev, V. Y.; Raghuraman, H.; Mandal, K.; Senapati, S.; Perozo, E.; Kent, S. B. H. J. Am. Chem. Soc. 2009 131 884 885. (198) Urisman, A.; Molinaro, R. J.; Fischer, N.; Plummer, S. J.; Casey, G.; Klein, E. A.; Malathi, K.; Magi Galluzzi, C.; Tubbs, R. R.; Ganem, D.; Silverman, R. H.; DeRisi, J. L. PLoS Pathog 2006 2 e25. (199) Schlaberg, R.; Choe, D. J.; Brown, K. R.; Thaker, H. M.; Singh, I. R. Proc Natl Acad Sci U S A 2009 106 16351 16356.

PAGE 144

144 (200) Lombardi, V. C.; Ruscetti, F. W.; Das Gupta, J.; Pfost, M. A.; Hagen, K. S.; Peterson, D. L.; Ruscetti, S. K.; Bagni, R. K.; Petrow Sadowski, C.; Gold, B.; Dean, M.; Silverman, R. H.; Mikovits, J. A. Science 2009 326 585 589. (201) Arnold, R. S.; Makarova, N. V.; Osunkoya, A. O.; Suppiah, S.; Sco tt, T. A.; Johnson, N. A.; Bhosle, S. M.; Liotta, D.; Hunter, E.; Marshall, F. F.; Ly, H.; Molinaro, R. J.; Blackwell, J. L.; Petros, J. A. Urology 2010 75 755 761. (202) Danielson, B. P.; Ayala, G. E.; Kimata, J. T. J. Infect. Dis. 2010 202 1470 147 7. (203) Fischer, N.; Schulz, C.; Stieler, K.; Hohn, O.; Lange, C.; Drosten, C.; Aepfelbacher, M. Emerg. Infect. Dis. 2010 16 1000 1002. (204) Aloia, A. L.; Sfanos, K. S.; Isaacs, W. B.; Zheng, Q.; Maldarelli, F.; Marzo, A. M. D.; Rein, A. Cancer Res 2010 70 10028 10033. (205) Barnes, E.; Flanagan, P.; Brown, A.; Robinson, N.; Brown, H.; McClure, M.; Oxenius, A.; Collier, J.; Weber, J.; Gnthard, H. F.; Hirschel, B.; Fidler, S.; Phillips, R.; Frater, J. J. Infect. Dis. 2010 202 1482 1485. (206 ) Cornelissen, M.; Zorgdrager, F.; Blom, P.; Jurriaans, S.; Repping, S.; van Leeuwen, E.; Bakker, M.; Berkhout, B.; van der Kuyl, A. C. PLoS ONE 2010 5 e12040. (207) Erlwein, O.; Kaye, S.; McClure, M. O.; Weber, J.; Wills, G.; Collier, D.; Wessely, S. ; Cleare, A. PloS One 2010 5 e8519. (208) Groom, H. C. T.; Boucherit, V. C.; Makinson, K.; Randal, E.; Baptista, S.; Hagan, S.; Gow, J. W.; Mattes, F. M.; Breuer, J.; Kerr, J. R.; Stoye, J. P.; Bishop, K. N. Retrovirology 2010 7 10. (209) Hohn, O.; Krause, H.; Barbarotto, P.; Niederstadt, L.; Beimforde, N.; Denner, J.; Miller, K.; Kurth, R.; Bannert, N. Retrovirology 2009 6 92. (210) Hong, P.; Li, J.; Li, Y. Virol. J. 2010 7 224. (211) Jeziorski, E.; Foulongne, V.; Ludwig, C.; Louhaem, D.; C hiocchia, G.; Segondy, M.; Rodire, M.; Sitbon, M.; Courgnaud, V. Retrovirology 2010 7 63. (212) Switzer, W. M.; Jia, H.; Hohn, O.; Zheng, H.; Tang, S.; Shankar, A.; Bannert, N.; Simmons, G.; Hendry, R. M.; Falkenberg, V. R.; Reeves, W. C.; Heneine, W. Retrovirology 2010 7 57. (213) Kim, S.; Kim, N.; Dong, B.; Boren, D.; Lee, S. A.; Das Gupta, J.; Gaughan, C.; Klein, E. A.; Lee, C.; Silverman, R. H.; Chow, S. A. J. Virol. 2008 82 9964 9977.

PAGE 145

145 (214) Smith, R.; Gottlieb, G.; Miller, A. D. Retrovirol ogy 2010 7 70. (215) Singh, I. R.; Gorzynski, J. E.; Drobysheva, D.; Bassit, L.; Schinazi, R. F. PLoS ONE 2010 5 e9948. (216) Li, M.; Dimaio, F.; Zhou, D.; Gustchina, A.; Lubkowski, J.; Dauter, Z.; Baker, D.; Wlodawer, A. Nat. Struct. Mol. Biol. 20 11 18 227 229. (217) Li, M.; Gustchina, A.; Matz, K.; Tzsr, J.; Namwong, S.; Goldfarb, N. E.; Dunn, B. M.; Wlodawer, A. FEBS J. 2011 278 4413 4424. (218) FEBS J. 2012 279 3276 3286. (219) Slee, D. H.; Laslo, K. L.; Elder, J. H.; Ollmann, I. R.; Gustchina, A.; Kervinen, J.; Zdanov, A.; Wlodawer, A.; Wong, C. H. J. Am. Chem. Soc. 1995 117 11867 11878. (220) Umezawa, H.; Aoyagi, T.; Morishima, H.; Matsuzaki, M.; Hamada, M. J. Antibiot. ( Tokyo) 1970 23 259 262. (221) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. J. Med. Chem. 2004 47 1739 1749

PAGE 146

146 BIOGRAPHICAL SKETCH Terry degree in Chemistry from Florida Agricultural and Mechanical University in 2005. In the fall of 2008 he entered the University of Florida chemistry prog ram to study physical chemistry with a specialization in computational chemistry. Upon entering the graduate program, he joined the lab of Prof. Adrian E. Roitberg. During his tenure at the University of Florida, he married Danielle McGee and they had thre e b eautiful children: Brendan, Cameran, and Logan McGee.