Retrocyclin Rc-101 Overcomes Cationic Mutations on the Heptad Repeat 2 of Hiv-1 Gp41

University of Central Florida STARS

Electronic Theses and Dissertations, 2004-2019

2007

Retrocyclin Rc-101 Overcomes Cationic Mutations On The Heptad Repeat 2 Of Hiv-1 Gp41

Christopher Fuhrman University of Central Florida

Part of the Immune System Diseases Commons, Microbiology Commons, and the Molecular Biology Commons Find similar works at: https://stars.library.ucf.edu/etd University of Central Florida Libraries http://library.ucf.edu

This Masters Thesis (Open Access) is brought to you for free and open access by STARS. It has been accepted for inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more information, please contact [email protected].

STARS Citation Fuhrman, Christopher, "Retrocyclin Rc-101 Overcomes Cationic Mutations On The Heptad Repeat 2 Of Hiv-1 Gp41" (2007). Electronic Theses and Dissertations, 2004-2019. 3166. https://stars.library.ucf.edu/etd/3166 RETROCYCLIN RC-101 OVERCOMES CATIONIC MUTATIONS ON THE HEPTAD REPEAT 2 OF HIV-1 GP41

CHRISTOPHER “KIT” ALLEN FUHRMAN B.S. University of Central Florida, 2005

A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Molecular Biology and Microbiology in the Burnett College of Biomedical Sciences at the University of Central Florida Orlando, Florida

Summer Term 2007

Major Professor: Alexander M. Cole

ii ABSTRACT

Retrocyclin RC-101, a θ-defensin with lectin-like properties, potently inhibits infection by many HIV-1 subtypes by binding to the heptad repeat (HR)-2 region of gp41 and preventing six-helix bundle formation. In the present study, we used in silico computational exploration to identify residues of HR2 that interacted with RC-101 and then analyzed the HIV-1 Sequence

Database at LANL for residue variations in the HR1 and HR2 segments that could plausibly impart in vivo resistance. Docking RC-101 to gp41 peptides in silico confirmed its strong preference for HR2 over HR1, and implicated residues crucial for its ability to bind HR2. We mutagenized these residues in pseudotyped HIV-1 JR.FL reporter viruses, and subjected them to single round replication assays in the presence of 1.25-10ug/ml RC-101. Except for one mutant that was partially resistant to RC-101, the other pseudotyped viruses with single-site cationic mutations in HR2 manifested absent or impaired infectivity or retained wild-type susceptibility to RC-101. Overall, these data suggest that most mutations capable of rendering HIV-1 resistant to RC-101 will also exert deleterious effects on the ability of HIV-1 to initiate infections - an interesting and novel property for a potential topical microbicide.

iii ACKNOWLEDGEMENTS

I would like to thank Martine Kline for his excellent technical assistance. I am also very

grateful to Dr. Wilfred Li, Teri Simas and Chris Misleh at the NBCR for the use of the rocks

cluster supercomputer. The NBCR also granted a scholarship to attend a weeklong work in San

Diego on Autodock for which I am very grateful. I also thank my lab-mates: Julie Martellini,

Anna Herasimtschuk, Nick Felts and Nitya Venkataraman. Thanks especially to Andrew Warren whose dedication help push this project along. Thanks to my committee members for the guidance in my work and in my scientific career. And finally, thank you Dr. Alex and Amy Cole for their support and tutelage. I appreciate all that I have learned from both of you throughout the last two years.

iv TABLE OF CONTENTS

LIST OF FIGURES ...... vii

LIST OF ACRONYMS/ABBREVIATIONS...... ix

CHAPTER ONE: INTRODUCTION...... 1

Overview...... 1

Innate Immunity...... 2

Defensins...... 3

Alpha- and Beta-Defensins...... 5

Theta-Defensins ...... 5

HIV-1 Epidemic...... 10

Mechanism of HIV Infection...... 11

The Structure of gp41 ...... 14

HIV-1 and Retrocyclin...... 18

Molecular Docking ...... 20

CHAPTER TWO: MATERIALS AND METHODS ...... 23

Computational Analysis of the Variation in HR1 and HR2 ...... 23

Preparation of HR1, HR2 and RC-Structure Models...... 23

Computational Modeling of RC-101 Binding ...... 24

Defining Hydrogen and Non-Hydrogen Bonds ...... 25

Preparation of Peptide...... 25

Cell Culture...... 26

HIV-1 Plasmid Constructs and Viral Entry Assay ...... 26

v CHAPTER THREE: RESULTS...... 29

Amino Acid Variation of HR1 and HR2 ...... 29

RC-101’s Preference for Anionic Charge...... 32

Anionic-to-Cationic Mutations on HR2...... 35

RC-101 Binds to the HR1-binding regions of HR2...... 39

RC-101 Binds Anionic, Polar, and Hydrophobic Residues of HR2...... 41

Cationic Mutations in the HR1-Binding Domain of HR2 ...... 43

CHAPTER FOUR: DISCUSSION...... 45

CHAPTER FIVE: CONCLUSION...... 47

APPENDIX A: MANUSCRIPT SUBMISSION...... 49

APPENDIX B: COMPUTER PROGRAM ...... 52

Docking Log Parser Command Line Interface (DLGCLI)...... 53

CountBonds...... 54

PrimerFind ...... 55

APPENDIX C: SEQUENCE INFORMATION ...... 59

REFERENCES ...... 63

vi LIST OF FIGURES

Figure 1 Defensins of the Innate Immune System...... 4

Figure 2 Theta-Defensins...... 8

Figure 3 RC-101 and RC-100...... 9

Figure 4 Six-Helix Bundle Formation Drives Viral-Host Membrane Fusion...... 13

Figure 5 Heptamer Designation of Alpha Helix ...... 15

Figure 6 Illustration of the Heptamer Functionality of the 6-Helix Bundle...... 16

Figure 7 Rendering of the 6-Helix Bundle per Heptamer Functionality...... 17

Figure 8 The Lamarckian Genetic Algorithm...... 22

Figure 9 Pseudotype Methodology ...... 28

Figure 10 Amino Acid Diversity of HR1 and HR2 ...... 30

Figure 11 Isoelectric Point of HR1 vs. HR2 of Group M HIV-1 Sequences...... 31

Figure 12 Final Docked Energy of RC-101 Dockings to HR1, HR2, and HR1-HR2 Dimer ...... 34

Figure 13 Anionic-to-Cationic Mutations...... 36

Figure 14 Relative Light Units of JR.FL Pseudotypes with Anionic-to-Cationic Mutations ...... 37

Figure 15 Percent Inhibition of Infectious Anionic-to-Cationic JR.FL Pseudotypes ...... 38

Figure 16 RC-101 Docks to HR2 in a Site Specific Manner...... 40

Figure 17 RC-101 Interaction per HR2 Residues ...... 42

Figure 18 HR1-Binding Region Mutations of HR2...... 44

Figure 19 HIV-1 Remains Susceptible to RC-101 Over 28 Rounds of Infection...... 48

Figure 20 First Page of Submitted Manuscript to Journal of Biological Chemistry...... 50

Figure 21 Confirmation of Primary Editoral Review ...... 51

Figure 22 File Hierarchy Produced by DLGCLI ...... 54

Figure 23 Input for PrimerFind...... 57

vii Figure 24 Sample Output from PrimerFind ...... 58

Figure 25 HR1 Sequence Information ...... 61

Figure 26 HR2 Sequence Information ...... 62

viii LIST OF ACRONYMS/ABBREVIATIONS

6HB Six-Helix Bundle

AIDS Acquired Immune Deficiency Syndrome

CCR5 Chemokine (C-C motif) Receptor 5

CD4 Cluster of Differentiation 4

CXCR4 Chemokine (C-X-C motif) Receptor 4

DC Dendritic Cell

DC-SIGN DC-specific ICAM-2-Grabbing Nonintegrin gp glycoprotein

HAART Highly Active Anti-Retroviral Therapy

HIV Human Immunodeficiency Virus

HOS Human Osteo-Sarcoma

HR Heptad Repeat

LGA Lamarckian Genetic Algorithm

PDB Protein Data Bank

RTD Rhesus Theta Defensins

SEM Standard Error of the Mean

TLR Toll-Like Receptor

ix CHAPTER ONE: INTRODUCTION

Overview

HIV-1 is a worldwide problem with 4.3 million new infections in 2006 (Joint United

Nations Programme on HIV/AIDS. 2006). The theta-defensin RC-101 prevents HIV-1 infection

(Owen, Rudolph et al. 2004). It is similar in structure and function to other theta-defensins of lower primate innate immune systems and has shown a remarkable ability to remain effective as an HIV-1 fusion inhibitor in extended viral culture (Cole, Yang et al. 2006). RC-101’s primary target is gp41, a glycoprotein on the surface of HIV. A trimer of gp41 molecules forms a 6-helix

bundle that catalyzes the fusion of the host and viral membranes. RC-101 prevents the formation

of the six-helix bundle (Gallo, Wang et al. 2006), putatively through its binding to heptad repeat

2 (HR2) region of gp41. However, the exact mechanism of inhibition is still unclear, and is the subject of the current study. One-hundred days of serial passaging of HIV-1BaL in the presence

of RC-101 revealed an escape mutant with three cationic mutations in the envelope gene, of

which two were on the heptad repeat regions that comprise the six-helix bundle (Cole, Yang et

al. 2006). To better understand the mechanism of RC-101 mediated viral inhibition, we used an

in silico computational approach complemented with in vitro HIV-1 inhibition studies to

examine the interaction between RC-101 and the two heptad repeats of gp41. Site-directed

mutagenesis was used to test the sensitivity of the virus to cationic mutations along HR2 while

molecular dockings were performed to uncover site-specific interactions and locate residues of importance for RC-101 binding.

1 Innate Immunity

Innate immunity describes a de facto approach to prevent microbial and viral assaults. It

is the first line of defense for all organisms, and for many it is the only defense (Danilova 2006).

Innate immune systems evolved slowly over millions of years; consequently, effectors largely

respond to molecular patterns whereas the adaptive immune system evolves through somatic cell

rearrangement and can target specific epitopes on a pathogen. The innate immune system

manifests itself in a variety of forms that include physical barriers such as skin, and chemical

mediators such as antimicrobial peptides. Pattern recognition receptors recognize molecular

signatures specific to pathogenic microbes and viruses and trigger an immune system response

when activated. In humans, toll-like receptors (TLR) respond to the molecular patterns of foreign pathogens such as specific sugars, LPS molecules, double-stranded RNA and viral DNA.

TLRs act in reactive fashion to prevent bacterial or viral pathogens from gaining a foothold in

the host cells and the environment (Trinchieri and Sher 2007). The receptors can activate the

release of antimicrobial peptides, activate apoptotic pathways inside the cell and even signal the

adaptive immune system to respond to infection.

Like TLRs, soluble molecular effectors of the innate immune system can also capitalize

on the molecular patterns of bacteria and viruses. For example, lysosyme has been shown to

prevent infections in the ocular region (Bron and Seal 1986) and mucosa of the respiratory system (Cole, Thapa et al. 2005) by attacking the peptidoglycan moieties in many bacterial cell walls. Similarly, lactoferrin sequesters iron, a key ingredient for microbial growth, and binds to the surface of bacteria. It can also modulate the immune response by suppressing inflammation and signaling the cells of the adaptive immune system to proliferate (Legrand, Elass et al. 2006).

2 In response to the innate immune system, viral and bacterial pathogens have evolved a

number of mechanisms to evade or counteract the immune system. Some bacteria maintain

proteases and enzymes that can modify their anionic membranes and therefore reduce the

number of anionic charges on its cell wall or outer membrane. As most antimicrobial peptides

are cationic, this reduced surface charge prevents the peptide’s initial electrostatic interaction

with the microbial membrane. Other bacteria expel the antimicrobial peptides using specialized

protein efflux pumps (Peschel 2002). Yet, many viruses and bacteria remain susceptible to the

effectors of innate immunity because the effectors respond to molecular patterns crucial to the pathogens’ survival.

Defensins

Defensins are the most widely studied family of antimicrobial peptides and are

characterized by their cationic charge at neutral pH, six cysteines which form three disulfide

bonds, and largely beta-sheet conformations (Lehrer 2004). Defensins exert broad spectrum

antimicrobial and antiviral activity against a range of Gram-positive and Gram-negative bacteria,

fungi, and enveloped and non-enveloped viruses. In primates, three defensin classes exist

(alpha-, beta- and theta-defensins), which are characterized by their unique tri-disulfide motif.

The molecular structures of the three classes of defensins are rendered in Figure 1.

Figure 1 Defensins of the Innate Immune System.

The molecules were rendered in the PyMol Molecular Graphics program (DeLano 2002). The top row of pictures illustrates the secondary structure of each defensins. Beta-sheets are rendered in yellow and alpha-helices are rendered in red. The bottom row illustrates the charges on each defensins. Anionic charges are rendered red and cationic charges are rendered in blue.

Alpha- and Beta-Defensins

While the presence of cationic proteins in granules of leukocyte was discovered in 1966, elucidating the structure and molecular evolution of these proteins has occurred only recently

(Lehrer and Ganz 2002). The alpha- and beta-defensin classes are the most widely observed defensins in nature and have been found in a wide range of animals. These two groups of defensins are thought to have arisen through a common precursor protein because of their proximity on human chromosome 8 (Liu, Zhao et al. 1997). Overall, there are six types of alpha- defensins identified in human. Alpha-defensins 1-4 are expressed by human neutrophils and participate in the innate immune system throughout the body. Alpha-defensins 5-6 are secreted by the Paneth cells of the Crypts of Leiberkuhn and play a predominant role in protecting intestinal mucosa from infection (Cunliffe 2003; Bevins 2005). Beta-defensins are produced either constitutively or induced in airway, intestinal, liver, and skeletal tissues (Lehrer and Ganz

2002). Beta-defensins are active against a number of bacterial strains and are relatively salt insensitive (Batoni, Maisetta et al. 2006), thus suggesting a protective role in the physiological salt concentrations of the mucosa. Both alpha- and beta- defensins are potent against many bacterial and viral species, but have marginal to moderate activity against HIV-1 (Wang, Owen et al. 2004; Garzino-Demo 2007). While speculative, this evolutionary oversight may have been a reason for humans’ susceptibility to HIV.

Theta-Defensins

The third defensin classification, theta-defensin, was the first macro-cyclic protein found

in mammals. Theta-defensin precursor proteins share a great deal of homology with alpha-

5 defensins, and it is believed that theta-defensins are the result of a duplicated ancestral alpha-

defensin (Nguyen, Cole et al. 2003). An unknown mechanism splices two precursors together to

create the macrocyclic peptide. Rhesus theta-defensin-1 (RTD-1) was discovered at sub-

micromolar concentrations in the granules of neutrophils and monocytes of rhesus macaque

(Tang, Yuan et al. 1999). Rhesus macaque have two unique theta-defensin genes that can be

spliced together to make three unique peptides, RTD-1, 2, and 3 (Figure 2) (Leonova, Kokryakov

et al. 2001). The heterodimeric combination of both genes, RTD-1, is the predominant form

expressed in rhesus leukocytes, while the homodimeric forms are observed to a lesser extent

(Tran, Tran et al. 2002).

Since their initial discovery, theta-defensins have been found in Old World monkeys,

orangutans and a lesser ape species (Tang, Yuan et al. 1999; Nguyen, Cole et al. 2003).

Humans, gorillas, Bonobos and chimpanzees retain a remarkably intact θ-defensin gene and express θ-defensin mRNA in a variety of cells and tissues, yet their lack of protein expression is

owed to a conserved stop codon in the signal sequence that prevents translation (Nguyen, Cole et

al. 2003). Normally, two theta-defensin precursors would be spliced together by a yet-to-be-

determined mechanism to form a macrocyclic peptide, 18 amino acids long (Lehrer and Ganz

2002). Because human theta-defensins genes only exist as expressed pseudogenes in humans,

the peptide was recreated using solid-phase synthesis and termed Retrocyclin (Cole, Hong et al.

2002) (Figure 3). Retrocyclin showed moderate activity against bacteria, but strong activity

against primary isolates and lab-adapted strains of HIV-1 at low micromolar/ high nanomolar concentrations. Later, an analogue of retrocyclin, RC-101, was engineered to have a single arginine-to-lysine mutation on one end of the beta-turns and has an even more potent inhibitory effect against HIV-1 (Figure 3).

6 Alterations in the primary sequence can have a drastic effect on the ability of the peptide

to properly function (Figure 2). For example, a theta-defensin with arginine-to-lysine substitutions near the disulfide bonds reduced the peptide’s ability to inhibit the HIV-1BaL escape

mutant as compared to RC-101 (Cole, Yang et al. 2006). Additionally, the macrocyclic structure

is important for proper functionality of the peptide. An open form of RTD-1 was considerably

less effective than the circularized RTD-1 even though it formed a similar structure in solution.

It was concluded that the introduction of the N- and C- terminal groups sufficiently altered the

electrostatic landscape to make it less effective against bacteria (Tang, Yuan et al. 1999; Trabi,

Schirra et al. 2001). A 15N solid state NMR study of retrocyclin-2 (RC-100b) revealed its ability

to insert into anionic lipid membranes, while retrocyclin-2 largely remained on the surface of

zwitterionic membranes (Tang, Waring et al. 2006). This result mirrors difference between

bacterial membrane that have high amounts of anionic headgroups and eukaryotic membranes

with zwitterionic lipid headgroups. Obviously, the charge on theta-defensins is important for

functionality, but the role it plays in HIV-1 inhibition is heretofore unknown.

Figure 2 Theta-Defensins

The rhesus theta-defensins (RTD) are expressed in vivo. Deviations in RTD-2 and RTD-3 from

RTD-1 are shown in red. RC-100 represents the retrocyclin (RC) based on the human pseudogene. RC deviations from RC-100 are shown in green. Theta-defensins are cyclic peptides and composed of 18 amino acids.

Figure 3 RC-101 and RC-100

The primary macrocyclic structure of RC-100 and RC-101 is illustrated in A. RC-100 contains and arginine while RC-101 contains a lysine. The biophysical model of RC-101, B, is the same model used in the docking experiments described below. Color in the biophysical structure correlates with those in the cartoon in A.

9 HIV-1 Epidemic

While developed countries have seen a reduction in new HIV-1 infections due to antiviral

therapeutics and sex education (Shattock and Moore 2003), less industrialized countries

especially those in sub-Saharan Africa have seen the rampant spread of HIV. There are currently

over 45 million individuals infected with HIV worldwide and 25 million of those infected are in

Africa (Joint United Nations Programme on HIV/AIDS. 2006). Highly Active Antiretroviral

Therapy (HAART) is the predominant treatment for infection, and this includes the use of combinations of reverse transcriptase inhibitors, protease inhibitors, fusion/entry inhibitors, and integrase inhibitors (Weiss 2003). Only drugs in the first three classifications are routinely used.

At present there is no cure for HIV, only treatments that can delay the onset of AIDS. The crisis

is further exacerbated by the emergence of single- and multi-drug resistant strains. These strains

often occur because of the difficulty of dosage maintenance in a patient, especially when the

drug is in short supply.

There is also need for topically applied microbicides that can prevent the spread of HIV.

In many cultures, promiscuity and the absence of the woman’s control over the sexual practices

of the partner promote the dissemination of HIV. Microbicides, with and without the use of

other “safe sex” practices, are looked to as a way to control the sexual transmission of HIV-1

(Pope and Haase 2003; McGowan 2006). In early studies, the spermicide nonoxynol-9 was

evaluated for its ability to prevent vaginal HIV transmission. Unfortunately, clinical trials were

halted because the damage nonoxynol-9 caused to the vaginal epithelia actually increased the

risk of transmission (Rustomjee and Abdool Karim 2001). Damage to the vaginal epithelia

induced the expression of proinflammatory cytokines and chemokines, which recruit

10 macrophages and dendritic cells, key targets for initial HIV infection. As an alternative to

nonoxynol-9, the theta-defensin RC-101 is currently being considered as a possible microbicide.

RC-101 is a potent anti-HIV peptide that is non-hemolytic for human red blood cells, non-

cytotoxic against several human cell lines at concentrations reaching 500 µg/ml, and is not

proinflammatory (Cole, Hong et al. 2002; Venkataraman, Cole et al. 2005). In addition, RC-101

remained active in a cervicovaginal construct for over 9 days without damaging the tissue and

could inhibit HIV-1BaL infection in the presence of vaginal fluid (Cole, Herasimtschuk et al.

2007). These collective data build a strong case for the development of RC-101 as a topical vaginal microbicide.

Mechanism of HIV Infection

Over the past 25 years, considerable effort has gone into understanding the mechanisms involved in HIV infection. Consequently, much is known of the viral life-cycle and molecular structures. In the developing world, unprotected sex is the principal route of HIV infection. In the vaginal or rectal mucosa, HIV comes in contact with Langerhans cells or dendritic cell (DC).

There are several DCs, each with a unique immunological purpose. In particular, monocyte- derived DCs contain a DC-specific ICAM-2-grabbing nonintegrin (DC-SIGN). The DC-SIGN is a C-type lectin binding protein that binds to gp120, internalizes the virus, and somehow protects the virus from degradation (Knipe, Howley et al. 2001). The DC then transports the virus to the lymph system, where they present the intact virus to the primary host cell, CD4+ T memory cells

(Wu and KewalRamani 2006).

DCs, in general, express more of the chemokine coreceptor CCR5 than CXCR4 – consequently, R5 (M-tropic) specific HIV viruses, which utilize CCR5 as a coreceptor for entry,

11 are the primary transmitted species. Upon sustained infection, a tropism switch to X4 (T-tropic) strains of HIV-1, which utilize CXCR4 as a co-receptor, occurs in most individuals and plays a major role in viral progression (Knipe, Howley et al. 2001).

The envelope proteins play a pivotal role in HIV infection. Trimeric gp120 on the surface of the virion binds to the CD4 receptor and the appropriate co-receptor; gp120 then disengages gp41 and allows the HR1 region form a homotrimer (Maddon, Molineaux et al. 1987;

Berger, Murphy et al. 1999; Gallo, Puri et al. 2001). Next, the HR2 regions fold into the hydrophobic grooves created by the trimerizations of HR1 and create the six-helix bundle (Chan,

Fass et al. 1997; Weissenhorn, Dessen et al. 1997). The formation of the six-helix bundle

(Figure 4) promotes and stabilizes pores in the host membrane and fusion of the host and viral membranes, which in turn, allows viral RNA and proteins to enter the cells (Melikyan,

Markosyan et al. 2000; Markosyan, Cohen et al. 2003). Finally, and most importantly the surface of gp41 is highly variable. The error-prone reverse transcriptase of HIV-1 produces different quasispecies of the virus. This results in epitopes that can readily change in order to evade the adaptive immune system. One amino acid mutation can alter an epitope enough to prevent antibody mediated viral inhibition (Wei, Decker et al. 2003).

The high level of mutation in HIV underscores the inability of the scientific community to develop an effective vaccine against the virus. Conserved epitopes on the envelope proteins are obscured by a glycan shield and other, more variable, residues (Rodriguez-Chavez, Allen et al. 2006). There are only a handful of human cross-neutralizing monoclonal antibodies, but even these can become ineffective if the proper mutation arises (Wei, Decker et al. 2003). It remains to be seen whether any vaccine containing a combination of HIV epitopes and immunogenic reagents can elicit an immune response robust and strong enough to attenuate HIV.

Figure 4 Six-Helix Bundle Formation Drives Viral-Host Membrane Fusion

13 The Structure of gp41

The structure of gp41 is essential for viral infection, but has eluded biochemists who have tried to isolate the full peptide in vitro: when unprotected by gp120, gp41 is very sensitive to proteases. As an alternative, segments have been synthesized that resist degradation and are stable in solution (Weissenhorn, Dessen et al. 1997). These peptides primarily consist of the heptad repeat regions of gp41, the stable, fusogenic core of the protein.

Heptad-repeats are a common element throughout nature and characterize the stacking of amino acids in the alpha-helix confirmation (Figure 5). Every seventh amino acid in an alpha- helix roughly aligns with the seventh amino acid above and below and projects out in the same direction. Each heptamer creates a facet of the alpha-helix, which is important for proper helix functionality (Figure 6 and Figure 7).

In the case of gp41, different heptamers have different roles in the formation of the six- helix bundle and cell-host membrane fusion. HR1 shows a strong propensity to form a trimer in solution (Chan, Fass et al. 1997). Accordingly, heptamers ‘a’ and ‘d’ of HR1 participate in trimerizations of the peptides. In vivo, this step occurs after gp120 binds to the CD4 receptor

(Gallo, Puri et al. 2001). The heptamers ‘a’ and ‘d’ in HR2 then bind to the hydrophobic pocket composed of heptamers ‘g’ and ‘e’ in HR1. Heptamers ‘c’, ‘f’, and ‘b’ of HR1 and ‘g’, ‘c’, ‘f’,

‘b’ and ‘e’ of HR2 are expose to the external environment when gp41 is in its post-fusogenic state, the six-helix bundle (Figure 6).

Figure 5 Heptamer Designation of Alpha Helix

Panel A illustrates the heptamer pattern. Arrows represent the amino acid backbone. The heptamer colors in A correlate to the colors on the cartoon in B. B is a rendering of HR1 and

HR2 of gp41 (PDB-ID: 1AIK) (Chan, Fass et al. 1997). Rendering was done in MacPyMOL

(DeLano 2002).

Figure 6 Illustration of the Heptamer Functionality of the 6-Helix Bundle

Figure 7 Rendering of the 6-Helix Bundle per Heptamer Functionality

The proteins are based off of the PDB file: 1AIK (Chan, Fass et al. 1997). The two extra dimers generated to form the 6-Helix Bundle were generated with crystallographic information in the published file. Color correlate to the heptamers in Figure 6 Illustration of the Heptamer

Functionality of the 6-Helix Bundle. Rendering was done in MacPyMOL (DeLano 2002).

17 HIV-1 and Retrocyclin

Retrocyclin (RC-100) prevents both X4 and R5 HIV-1 replication in CD4+ PBMC cells

and H9 CD4+ T Lymphocytes from infection. It was further determined that RC-100 does not

directly inactive the virus in suspension (Cole, Hong et al. 2002). RC-100 binds gp120 only

when it is glycosylated and binds to the CD4 receptor with lower affinity than gp120-CD4

binding. Retrocyclin is a lectin, and its ability to bind sugar molecules on the surface of proteins

is undoubtedly connected to its ability to prevent HIV infection. In addition, retrocyclin does not

discernibly discriminate between N- and O-linked sugars (Wang, Cole et al. 2003). RC-100 does

not prevent CD4/CCR5/CXCR4-gp120 binding, but seemed to target the virus at a step prior to

entry (Munk, Wei et al. 2003). This is in direct contrast with alpha-defensins that are thought to

hinder HIV at a post entry step (Wang, Owen et al. 2004). The retrocyclin analog, RC-101, is more active than RC-100 against primary R5 and X4 isolates and a variety of HIV-1 subtypes

(Munk, Wei et al. 2003). It is even active against isolates that show a slight resistance to RC-100

(Owen, Rudolph et al. 2004).

Why are retrocyclins and theta-defensins in general so active against HIV-1? Gallo et al

(2006) confirmed that RC-100 does not interfere with CD4-gp120 binding, although it did bind

to the proteins. Retrocyclin did not crosslink CD4 molecules and therefore did not interfere with

the lateral mobility of CD4 molecules. Instead it was determined through cell-cell fusion assays

that RC-100 bound the ectodomain of gp41, a possibility only in the late stages of viral fusion.

In the cell-cell fusion assays, the transfer of dye from CV-1 cells expressing viral envelope

proteins on the cell surface to SupT1 cells with CD4, CCR5 and CXCR4 on the surface

measured the fusion ability of different envelope proteins. Envelopes from HIV-1IIIB (X4) and

18 HIV-1BAL (R5) Env but not HIV-2ROD were inhibited by RC-100. Fusion between CV-1 cells

expressing HIV-2SBL or SIVMAC Env and Sup T1 cells were only slightly inhibited by retrocyclin.

HIV-2 and SIV have a reduced amount of anionic charge on HR2. It is thought that this difference in charge prevented the proper electrostatic interactions for viral fusion. Furthermore,

RC-100 prevented the formation of the six-helix bundle (6HB) of in vitro synthesized heptad

repeat (HR)1 and HR2 regions of gp41 of HIV-1 (Gallo, Wang et al. 2006). Taken together,

these data support the fusogenic core of gp41 as the crucial target for retrocyclins.

To further explore the interaction between the HIV-1 envelope and retrocyclins, a study

was undertaken in this lab to determine exactly how theta-defensins inhibit HIV-1. During 100

days of serial passaging, the HIV-1 strain BaL evolved three cationic mutations in the presence

of sub-lethal concentrations of RC-101 (Cole, Yang et al. 2006). Of the three mutations, one was

found in gp120 and one each in the HR1 and HR2 regions of gp41, and all three mutations

converted a polar or anionic residue to a cationic residue (Cole, Yang et al. 2006). In addition, the cationic mutation in HR2 ablated a commonly glycosylated asparagine residue. Loss of glycosylated residues in gp41 can reduce the fusion ability of the virus, and alter the shape of

discontinuous epitopes (Perrin, Fenouillet et al. 1998; Wang, Song et al. 2005), and thus the

dependence of viral replication on the presence of RC-101 was not surprising. The cationic

mutations and loss of glycosylated residues suggest an attempt by the virus to repel the cationic

lectin, RC-101 (Wang, Cole et al. 2003; Cole, Yang et al. 2006). Our work continues the effort

to determine the exact mechanism of how retrocyclin inhibits HIV fusion, by employing a

combination of life science and computer science approaches.

19 Molecular Docking

AutoDock and a docking methodology called Blind Docking (BD) were used to discriminate between HR1 and HR2 and identify residues of interest on HR2. AutoDock is the most widely used docking software (Sousa, Fernandes et al. 2006). The application samples different torsional conformations and locations using a Lamarckian Genetic Algorithm (LGA) to return the docked molecule that binds best to the macromolecule.

Genetic algorithms borrow the Lamarckian genetic model of genes and phenotype to solve otherwise computational intensive programming problems. In short, a population of variables called the genotype is manipulated by operations similar in concept to those seen in biological genetics. The genotype is then tested by translating it to a phenotype, where the fitness of the genotype is measures. In the LGA the genotype of the ligand molecule is represented as a string of variables: three variables for location (x, y, z) , four for rotation

(quaternion), and one variable for each rotatable bond (-180º to +180º). The phenotype for molecular docking applications is the atomic spatial coordinates. Fitness is measured as the change in free energy upon binding to the macromolecule and is also referred to as the Final

Docked Energy (FDE). To measure electrostatics, AutoDock uses a derivative of the amber electrostatic packages that has been optimized for ligand binding measurements. With an LGA, the fitness of the phenotype can affect the genotype, hence the “Lamarckian” in the name. The genotype undergoes a number of operations (Figure 8; blue boxes). Testing each ligand in a population of spatially unique ligands is computationally intensive, therefore, fitness is initially evaluated using pre-computed electrostatic grids (one grid per atom type) and only a subset of the population is chosen for the local search algorithm (Figure 8; yellow box). In the local

20 search, the spatial and rotational values are manipulated in a stepwise manner according to the

energy gradient. Each move is tested for increased fitness. The new ligand values are then

reincorporated into the genotype and the best ligand gets passed onto the next generation. When

the exit requirements (Figure 8; magenta diamond) are met, the coordinates of the fittest ligand

found to that point are saved. This whole process is then repeated to create a set of docked ligands (Morris, Goodsell et al. 1998).

There have been several blind docking studies done (Hetenyi and van der Spoel 2002;

Hetenyi and van der Spoel 2006), and some even included gp41 (Jun Tan, Kong et al. 2005); however, all have used small molecules as their ligand. Using a highly flexible ligand such as

RC-101 increased the computational complexity; therefore a full exploration of possible ligand conformations was not as feasible. Therefore, a unique approach was taken to analyze the collection of fit ligands. The intermolecular interactions were quantified (McDonald and

Thornton 1994; Wallace, Laskowski et al. 1995) and residues of frequent interaction were targeted for mutagenesis. Furthermore, the fact that theta-defensins contain a large number of rotatable bonds supports the notion that more than one bound ligand conformation is possible.

This flexibility along with their amphipathic nature allows them to recognize patterns over specific epitopes.

Figure 8 The Lamarckian Genetic Algorithm

22 CHAPTER TWO: MATERIALS AND METHODS

Computational Analysis of the Variation in HR1 and HR2

The aligned envelope protein sequences of 913 unique HIV-1 group M viruses were

obtained from the HIV Sequence Database at the Los Alamos National Laboratories (LANL;

www.hiv.lanl.gov), which has been curated by LANL scientific staff for duplicate sequences

from the same source. HIV-1 Group M represents a group of viral isolates that diverged in humans and originated from one chimpanzee-to-human transmission event, and is the most common group found in humans (Robertson, Anderson et al. 2000). The amino acid diversity

2 ⎜⎛ 20 ⎟⎞ 20 2 Daa = ∑ xi − ∑ xi index was calculated as ⎝ i=1 ⎠ i=1 , where x is the proportion of the ith amino

acid of the twenty standard amino acids at that location (Yamaguchi-Kabata and Gojobori 2000).

The value is similar to gene diversity in population genetics (Nei 1987). An amino acid with a

diversity index that is less than .05 is considered monomorphic (Yamaguchi-Kabata and

Gojobori 2000). The residues corresponding to HR1 and HR2 were spliced out of the sequence

file and used for evaluation in the ExPASy pI/Mw tool to determine the isoelectric point

(Bjellqvist, Hughes et al. 1993; Bjellqvist, Basse et al. 1994; Wilkins, Gasteiger et al. 1999).

Preparation of HR1, HR2 and RC-Structure Models

Three separate structural representations were needed for bio-computational experimentation: the HR1 and HR2 regions of JR.FL and the θ-defensin RC-101. In the context of computational data, the HR1/HR2 nomenclature refers only to the N36 and C34 peptides, respectively. The three-dimensional structural models of the HR1 and HR2 regions of JR.FL

23 were generated using Swiss-Model protein homology web server from the HIV-1 gp41 core

structure (PDB-ID: 1AIK) published previously (Chan, Fass et al. 1997; Schwede, Kopp et al.

2003; Arnold, Bordoli et al. 2006). The structure for RC-101 was created with the mutagenesis

function of the PyMOL Molecular Graphics System, based on the structure of retrocyclin-2

(PDB-ID: 2ATG). Two in silico mutations were performed to create RC-101: the second

arginine to a glycine and the fourth arginine to a lysine (Owen, Rudolph et al. 2004). The backbone atoms for both mutated residues remained stationary. There was no need to minimize the rotational bond energy of the mutated bonds, since all carbon-carbon or carbon-nitrogen

bonds were deemed “rotatable” in the docking procedure.

Computational Modeling of RC-101 Binding

Grid and Docking parameter files for all RC-101 dockings to dimer and comparative

monomer macromolecules were prepared with AutoDockTools (ADT) and accompanying

scripts, then run with AutoDock 3.0 and AutoGrid 3.0 (Goodsell, Morris et al. 1996; Morris,

Goodsell et al. 1998). The grid parameters were the same for all three macromolecules. The

number of points in the x, y and z direction was 76, 76, and 126 respectively. The grid spacing

value was 0.452777777778 Å. Finally, the grid center was defined as the xyz-coordinate

(17.449, 13.8, 5.67). All other AutoGrid parameters remained at their default values. The

ligand, RC-101, was prepared with ADT according to the AutoDock Manual (Morris, Goodsell

et al. 1998). For each macromolecule (HR1, HR2, HR1+HR2) and ligand (RC-101), hydrogen

positions were reassigned, non-polar hydrogens were merged, and Kollman United Charges were

assigned to each residue.

24 The genetic algorithm variables of population size (200), maximum number of energy evaluations (2x106), and the maximum number of generations (2x105) were increased from their

default values using Hetenyi et al. as a general guideline (Hetenyi and van der Spoel 2002).

Lower values were used because the protein model contained substantially less solvent-exposed

surface area and contained less than half the average number of residues tested in previous blind- docking studies (Debnath, Radigan et al. 1999; Hetenyi and van der Spoel 2002; Kong, Tan et al.

2006). For each docking simulation the genetic algorithm was run 200 times. Each docking

simulation was executed four times, and the quantitative measures of all four docking

simulations were averaged and the SEM calculated.

Defining Hydrogen and Non-Hydrogen Bonds

Tabulation of hydrogen bonds and non-hydrogen bonds were generated using LIGPLOT

in conjunction with HBPLUS (McDonald and Thornton 1994; Wallace, Laskowski et al. 1995).

The best 25% (50) docked RC-101 molecules, according to the Final Docked Energy, were

tabulated, and the average bonds per residue for four independent docking executions were reported along with the SEM. See appendix B for more information on the computer code used in this work.

Preparation of Peptide

The 18 amino acid peptide RC-101 was synthesized as previously described (Tang, Yuan

et al. 1999; Leonova, Kokryakov et al. 2001; Cole, Hong et al. 2002) with the sequence: cyclic-

GICRC ICGKG ICRCI CGR. After each step the peptide was subjected to MALDI-TOF mass

spectrometry to assess homogeneity (typically ~95%) and to confirm that the observed mass

25 agreed with the theoretical mass. Peptide concentrations were determined by quantitative

peptide analysis.

Cell Culture

HOS-CD4-CCR5 cells (Dr. N. Landau, Salk Institute for Biological Studies, La Jolla,

CA) that allow entry of R5 HIV-1, were acquired from the NIH AIDS Research and Reference

Reagent Program. HOS cells were grown in DMEM supplemented with penicillin, streptomycin, 10% FBS, 1 μg/ml puromycin, and mycophenolic acid selection medium. 293T cells were grown in DMEM with penicillin, streptomycin, and 10% FBS.

HIV-1 Plasmid Constructs and Viral Entry Assay

The expression vectors pNL-LucR–E– and JR.FL env were gifts from Dr. N. R. Landau

(The Salk Institute for Biological Studies, La Jolla, CA). JR.FL is an R5 strain of HIV-1. In

total, 18 JR.FL env mutants were constructed. Nine glutamic acids on the HR2 of HIV-1 gp41

were mutated to lysines. Amino acids 632, 634, 641, 647, 648, 654, 657, and 659 were mutated

from Glu (GAA) to Lys (AAA). Amino acid 636 was mutated from Asp (GAC) to Arg (CGC).

For the second set of mutagenesis studies, polar or hydrophobic residues were mutated to an

arginine or lysine: 638 Tyr (TAC) to Arg (CGC), 640 Ser (AGC) to Arg (AGG), 642 Ile (ATA)

to Lys (AAA), 645 Leu (CTA) to Arg (CGA), 649 Ser (TCG) to Lys (CGC), 651 Asn (AAC) to

Lys (AAA), 653 Gln (CAA) to Lys (AAA), and 656 Asn (AAT) to Lys (AAA).

Each mutation was created from the wildtype JR.FL env plasmid using the QuikChange

Multi Site-Directed Mutagenesis Kit (Stratagene), verified by sequencing (University of Central

Florida Biomolecular Science Center Genomics Core Laboratory, Orlando, FL) and compared to

26 the published JR.FL wild-type sequence (Accession Number: U63632). Subsequently, HIV-1

single-cycle (replication incompetent) luciferase reporter viruses were produced by

cotransfecting 293T cells with 10 µg each of pNL-LucR–E– and one of the JR.FL env clones.

Virus-containing, clarified supernatants were collected after 48 hrs, filtered though a .45-µm

filter and stored at -80°C in aliquots until needed. One aliquot was used to quantify propagated

pseudovirus by p24 ELISA (PerkinElmer). Another aliquot was used to ensure the integrity of the envelope gene. In summary, viral RNA was isolated from the JR.FL pseudotypes (QIAGEN

Viral RNA Mini Kit). A cDNA library was created from the isolated viral RNA with the iScript

Select cDNA Synthesis Kit (BioRad). Then a 666bp envelope region containing HR1 and HR2 was PCR-amplified and detected using a 1.5% agarose gel. The sense primer used was 5’-

CTGTGTTCCTTGGGTTCTTGG-3’ and the antisense primer used was 5’-

CTCCACCTTCTTCTTCGATTCC-3’. To measure infectious ability of JR.FL pseudotypes,

HOS-CD4-CCR5 cells (5 x 103/well; 96-well plate) were infected with 50 ng p24/well of virus in

the presence or the absence of RC-101 (0, 1.25, 2.5, 5, 10 µg/mL), and luciferase activity was

measured 2 days later. The whole process is illustrated in Figure 9.

Figure 9 Pseudotype Methodology

28 CHAPTER THREE: RESULTS

Amino Acid Variation of HR1 and HR2

In order to measure the susceptibility of the envelope gene to mutation and identify viable

escape mutants, we analyzed over 900 HIV-1 Group M envelope protein sequences from the

HIV Sequence Database at LANL. The amino acid diversity of HR1 and HR2 are distinctly

dissimilar (Figure 10A). The majority of sites (28 of 36) on HR1 are monomorphic and do not

readily change, whereas the majority of sites (21 of 34) on HR2 are highly pliable and change readily between viral strains. Of the eight non-monomorphic sites found on HR1, six are

externally exposed to the environment in the 6HB conformation. To visualize the sequence variation as a function of biochemical structure we mapped the amino acid diversity values to the

3-dimensional model of HR2 (Figure 10B). The externally exposed regions of HR2 in the 6HB

show a high amount of amino acid diversity, while the HR1 binding domain on HR2 is

predominantly monomorphic. Yamaguchi-Kabata et al. (Yamaguchi-Kabata and Gojobori 2000)

found that discontinuous epitopes in the α-helices of gp120 were under putative positive selection. By contrast, the monomorphic sites of HR2 suggest a region under very little selection. Alternatively, the regions exposed in the 6HB conformation are under strong putative positive selection from pressure generated by the immune system’s ability to easily access these exposed epitopes. RC-101’s long term potency against HIV-1 BaL could be attributed to interaction with discontinuous epitopes of the monomorphic residues of HR2.

Figure 10 Amino Acid Diversity of HR1 and HR2

A, the amino acid diversity index of HR1 and HR2 was calculated for 913 group M HIV-1 viruses. All points below the dotted line (.05) are considered monomorphic. B, the diversity index was mapped to the 3-dimensional structure of HR2 (N-terminal on top). The monomorphic residues, more red in color, are found in the HR1 binding region of HR2. The highly diverse residues, more white in color, are exposed to the external environment in the 6HB confirmation. The image in B was rendered with MacPyMOL (DeLano 2002).

Figure 11 Isoelectric Point of HR1 vs. HR2 of Group M HIV-1 Sequences

The isoelectric points of the HR1 and HR2 were obtained by inputting the Group M sequences into the pI/MW tool of ExPASy.

31 Because all three known non-synonymous, RC-101-evasive mutations were cationic

residues (Cole, Yang et al. 2006), we chose to measure the isoelectric points of the heptad repeat

regions of all Group M sequences as a marker of charge diversity. The isoelectric points of the heptad repeats illustrate ability of HIV-1 to alter its regional charge in vivo. While the amino acid diversity of HR2 is highly variable, its isoelectric range is acidic and significantly restricted

– 96% of the isoelectric points range between 3.89 and 4.66. Alternatively, HR1 is highly monomorphic but covers a wide range of isoelectric points (Figure 11). In line with having only

eight non-monomorphic sites, the isoelectric points show a strong inclination to cluster around

certain values: 8.49 (n=30), 9.99 (n=52), 10.29 (n=33), 10.83 (n=569), 11 (n=140), 11.71 (n=58),

and 12.01 (n=12). Sequences with higher HR1 isoelectric points have a greater number of

cationic mutations with fewer anionic residues; the converse is true for HR1 sequences with more acidic isoelectric points. The isoelectric range of HR1 is over twice that of HR2, suggesting a greater in vivo variation in electrostatic density.

RC-101’s Preference for Anionic Charge

While charge interaction plays an important role in RC-101 viral inhibition, it is unknown which

residues play an important role in binding. Because RC-101 still binds gp41 in the absence of

linked sugar molecules, we can reasonably exclude the sugar moieties from having a direct

interaction with RC-101 (Cole, Yang et al. 2006; Gallo, Wang et al. 2006). The molecular docking program AutoDock (Goodsell, Morris et al. 1996; Morris, Goodsell et al. 1998) was

used to determine RC-101’s affinity for HR1, HR2 and the dimer (HR1+HR2). Previous docking procedures to the protein models of HR1 and HR2 focused on docking small molecules

to the helices (Jun Tan, Kong et al. 2005). In contrast, RC-101 contains a large number of

32 flexible side chains and flexible side groups. Consequently, our dockings did not elucidate just one docked residue that can be considered the principal docking site of RC-101, but a number of

RC-101 binding conformations. Four in silico docking experiments revealed a significantly lower ΔG for RC-101 upon binding HR2 than HR1 (P = .0005). The docking of RC-101 to HR1 alone did not result in a strong binding energy (Figure 12). Conversely, the minimum energy of binding to HR2 is predominantly lower than values of small molecule inhibitors previously docked to this model (Jun Tan, Kong et al. 2005) (Figure 12).

Figure 12 Final Docked Energy of RC-101 Dockings to HR1, HR2, and HR1-HR2 Dimer

Four in silico docking experiments revealed a significantly lower ΔG for RC-101 upon binding

HR2 than HR1 (P = .0005). The ΔG upon binding is also referred to as the Final Docked Energy

(FDE). Error bars represent the SEM.

34 Anionic-to-Cationic Mutations on HR2

We created HIV-1 env molecular clones to identify mutations that would effectively alter

HIV-1 susceptibility to RC-101. An expression vector containing env from JR.FL, an R5 pseudotype, was subjected to site-directed mutagenesis to create mutant clones (Figure 9).

Because RC-101 viral entry inhibition is glycan-independent and charge alteration is a common mechanism of microbial evasion of antimicrobial peptides (Peschel 2002; Lehrer 2004; Cole,

Yang et al. 2006; Gallo, Wang et al. 2006), we individually mutated each negatively charged amino acid to a positively charged lysine or arginine Figure 13.

After alteration of the env gene, the wild type stock (non-mutated) or mutant JR.FL env clones were then used to create pseudotyped single-cycle HIV-1 luciferase reporter viruses, and

RC-101 activity against each viral clone was measured. Of the ten JR.FL variants, five variants showed scant ability to infect HOS.CD4.CCR5 cells (Figure 14A). The mutation on all five low-

or non-infectious variants was located on the region of HR2 externally exposed in the 6HB

conformation (heptamers b, c, e, f, and g). Of the five pseudotyped variants that effectively

entered HOS.CD4.CCR5 cells, only the pseudotype with a lysine at amino acid position 648

showed a significant amount of partial resistance to RC-101 (P = .05) (Figure 14B, Figure 15).

Residue 648, part of the ‘g’ heptamer, is located in the central region of the helix and through our modeling simulations is a potential binding site for the positive residues on RC-101. These data suggest that the ability of the virus to form the 6HB was significantly decreased and/or the mutants lost the ability to properly form the gp41 pre-fusion complex. Furthermore, certain

anionic charges on HR2 were integral for proper 6HB formation and did not tolerate cationic

mutations.

Figure 13 Anionic-to-Cationic Mutations

The top alpha-helix is HR2 of the gp41 molecule. Only the charged residues side chains are displayed as sticks and labeled. Cationic residues are white, while anionic residues are in black.

Hept. = heptamer location (a-g).

Figure 14 Relative Light Units of JR.FL Pseudotypes with Anionic-to-Cationic Mutations

The JR.FL env molecular clone was mutated using site-directed mutagenesis according to the

HR2 sequences in Figure 13. Pseudotyped viruses were then used to infect HOS-CD4-CCR5

cells. A shows pseudotypes that infected HOS cell very little or not at all with no RC-101. In

contrast, B shows pseudotypes that infected in a manner similar to the wild-type JR.FL

molecular clone. Error bars represent the SEM (n=4).

Figure 15 Percent Inhibition of Infectious Anionic-to-Cationic JR.FL Pseudotypes

Percent inhibition was calculated for normal infectious virus, Figure 14B. All the pseudotypes

were inhibited similar to wild type, except for E648K (P = .05). Error bars represent the SEM

(n=4).

38 RC-101 Binds to the HR1-binding regions of HR2

Figure 16 shows a rendering of the top 5 docked positions from a representative

AutoDock experiment. Ligands binding to HR1 were non-specific as evidenced by the highly dispersed RC-101 molecules. In contrast, RC-101 repeatedly bound to HR2 in the same region.

When examining a helical representation of HR2 (Figure 5), the backbone of RC-101 covers the

‘a’ and ‘d’ heptamers, while the long, flexible sidechains of RC-101 extend out and interact with heptamer locations ‘g’ and ‘e’. Interestingly, these heptamer positions are areas of low amino acid diversity (Figure 10) that coincide with the region that binds HR1 upon 6HB formation.

RC-101’s strong affinity for HR2 would prevent the interaction of HR1 and HR2, the formation of the 6HB, and the subsequent fusion of the host and viral membranes (Figure 7), which is supported by recent in vitro studies (Cole, Yang et al. 2006; Gallo, Wang et al. 2006).

Figure 16 RC-101 Docks to HR2 in a Site Specific Manner.

The top 5 docked RC-101 molecules of a representative docking, as measured by the FDE, are

shown as silver loops near the α-helix that it was docked to. The RC-101 molecules docked to

HR1 are much more dispersed than the RC-101 molecules docked to HR2. The color of each residue of HR1 and HR2 correlates with the heptamer designation shown in Figure 5. The

images were rendered in MacPyMOL (DeLano 2002).

40 RC-101 Binds Anionic, Polar, and Hydrophobic Residues of HR2

The computer programs LIGPLOT and HBPLUS were used to identify specific interactions between the ligand, RC-101, and HR2 based on proximity and atomic angles. We quantified the number of interactions per HR2 residue for the lowest (best) 25% of docked RC-

101 molecules based on final docked energy for each docking experiment of 200 iterations of the

Lamarckian Genetic Algorithms (Figure 17). The applications identified two sets of molecular interactions between RC-101 and HR2: hydrogen bonds at residues Serine-649, Glutamine-653 and Asparagine-656 and hydrophobic or non-hydrogen bonded contacts at residues Tyrosine-

638, Isoleucine-642 and Leucine-645 (Figure 17, asterisks). All six residues are located in the

‘a’ and ‘d’ heptamer regions of HR2, which bind HR1 upon 6HB formation; four of the residues

are monomorphic and the remaining two residues have reasonably low amino acid diversity

values. AutoDock consistently bound RC-101 to a location with low amino acid diversity and

that has an important role in 6HB formation.

Figure 17 RC-101 Interaction per HR2 Residues

Four docking experiments were completed, each with 200 Lamarckian Genetic Algorithm. The best 25% (50) docked RC-101 molecules from each docking experiment were analyzed for intermolecular interactions (hydrogen bonding and hydrophobic contacts) and tabulated per HR2 residue. Asterisks indicate the six residues of HR2 that had the greatest number of interactions with RC-101, and which were subjected to in vitro infection assays (Figure 6). Error bars represent the SEM.

42 Cationic Mutations in the HR1-Binding Domain of HR2

Based on the above mentioned study, we created mutant pseudotyped JR.FL env that contained a cationic mutation at each of the six residues observed to interact with RC-101 in silico. In addition, we mutated two residues on the 6HB-exposed portion of HR2 (heptamers f and c) as negative controls (Figure 18): both of these control pseudotypes infected

HOS.CD4.CCR5 cells and remained sensitive to RC-101. Four mutants were non-infectious even in the absence of RC-101 (Figure 18B). All non-infectious JR.FL mutants were located on

heptamers that indirectly or directly interacted with HR1 (Figure 18C, Figure 18D) (Chan, Fass

et al. 1997; Suntoke and Chan 2005). Of the JR.FL mutants that did infect HOS.CD4.CCR5

cells, none were resistant to RC-101.

Figure 18 HR1-Binding Region Mutations of HR2

The JR.FL env molecular clone was mutated using site-directed mutagenesis according to the

HR2 sequences in A. The top alpha-helix is HR2 of the gp41 molecule. Only the charged

residues side chains are displayed as sticks and labeled as in Figure 13. Hept. = heptamer location (a-g). Pseudotyped viruses were then used to infect HOS-CD4-CCR5 cells. B, cationic mutations of RC-101-interacting residues revealed non-viable mutations, B, or normally inhibited mutant pseudotypes, C. All normal infectious mutants were inhibited similar to the wild-type, C. Error bars represent the SEM (n=4).

44 CHAPTER FOUR: DISCUSSION

The envelope protein of HIV-1 is under many kinetic restraints for proper functionality.

First, the short time between gp120-CD4 interaction and 6HB formations limits the time 6HB-

inhibitors have to act (Gallo, Puri et al. 2001; Dimitrov, Louis et al. 2005). The strong net negative charge of HR2 and net positive charge of RC-101 creates a strong electrostatic attraction that likely promotes binding. This is evident in the marked difference observed between non-specific binding of RC-101 to HR1 and specific binding to HR2 seen in this work.

RC-101 binds reversibly but with high affinity to glycoproteins and associates with the cellular lipids and proteins involved in host-viral fusion (Wang, Cole et al. 2003; Buffy, McCormick et al. 2004). This lectin-like binding places RC-101 in a most opportune location to affect 6HB formation.

As a response to opposing host and environmental factors, HIV-1 employs a number of counter-measures including a “glycan shield” and the expediated rate of evolution due to the error-prone nature of its reverse transcriptase protein. Alterations in the glycan shield affects access to binding sites (Wei, Decker et al. 2003). In addition, the 6HB formed in solution with the synthetic N36 and a glycosylated C34 peptide was less compact than its non-glycosylated counterpart (Wang, Song et al. 2005). This suggests a variation in the inter-helical distance and a possible change in the discontinuous epitopes targeted by site-specific antibodies (Choudhry,

Zhang et al. 2007). In our analysis of HIV-1 protein sequences we observed many non- monomorphic sites with variation primarily within an amino acid chemical grouping (ex. Ile ↔

Leu) further altering possible binding epitopes. Thus the question remains as to whether HIV-1

45 mutations that confer partial resistance against RC-101 change the binding site of RC-101 or alter its access to the binding site. Both scenarios are possible.

In attempting to evade RC-101 inhibition, HIV-1 developed three cationic mutations, one of which removed a glycosylated residue, yet caused the virus to remain dependent on RC-101 for infectivity (Cole, Yang et al. 2006). Anionic-to-cationic mutations along the HR2 region resulted in a normal infectious mutant only 50% of the time, with all mutants susceptible to RC-

101. This suggests that the negative charge on HR2 may be important for maintaining normal replication efficiency of HIV-1, possibly by stabilizing its interaction with HR1 during 6HB formation. Although mutations that alter the negative charge of HR2 may impair RC-101 binding, they may also have the untoward effect (for the virus) of preventing its ability to mediate the fusion process and infect cells.

46 CHAPTER FIVE: CONCLUSION

The virus’s inability to become fully resistant to RC-101 is further illustrated by an

extension of our previous work (Cole, Yang et al. 2006). Passaging the virus from round 20 to

round 28 (days 100 to 140) in the presence of 10-20 ug/ml RC-101 neither induced additional

mutations, nor increased its resistance (Figure 19). Collectively, our data indicate that it is unlikely HIV-1 can mount further resistance to RC-101: aside from one partially resistant virus, mutant viruses either remained infectious yet sensitive to RC-101, or suffered from a significant loss of fusion efficiency.

The predominant problem with current HIV-1 treatments is the eventual emergence of fully resistant mutants that are then transmitted to new hosts. The same problem is theoretically possible for widely used topical microbicides. Our work has shown that the ability of HIV-1 to generate escape mutants against RC-101 is limited, and thus RC-101 holds great potential as an anti-HIV-1 microbicide in part because it remains effective against the virus.

Figure 19 HIV-1 Remains Susceptible to RC-101 Over 28 Rounds of Infection

This figure is an extension of previous work (Cole, Yang et al. 2006). Asterisks represent rounds

where HIV was isolated from the supernatant and sequenced for mutations. Cationic mutations in gp120 and HR1 of gp41 appeared at round 4, whereas the mutation in HR2 appeared at Round

15. The mutations remained stable throughout all successive passages.

48 APPENDIX A: MANUSCRIPT SUBMISSION

Figure 20 First Page of Submitted Manuscript to Journal of Biological Chemistry

Figure 21 Confirmation of Primary Editoral Review

51 APPENDIX B: COMPUTER PROGRAM

52 All the applications developed for this thesis were written in the PHP scripting language

by Zend Technologies (www.php.net). The scripts require php language version 5 (php5). The

first two programs, DLGCLI and countBonds, are command line driven and can take advantage

of the UNIX shell or DOS command interface. DLGCLI can also be used as a webpage. The application PrimerFind requires a webserver with php5 and may not be run as a command line application.

Docking Log Parser Command Line Interface (DLGCLI)

Each docking experiments with AutDock 3.0 returns one computer file, a docking log

file. In summary it contains all the experimental setup variables and most importantly the docked ligand (RC-101) molecules. DLGCLI accepts two files as input: the docking log and the macromolecule (gp41 Heptamers). It creates a unique folder with a summary of the results and sub-folder for each docked ligand (Figure 22). The summary file contains information

(maximum, minimum, average, standard deviation, etc) about all the electrostatic interaction for

all the ligands produced in the experiment. The values file contains the ligands ranks according

to their final docked energy (FDE) i.e. the change in free energy upon binding. The

Experimental Results file also contains the summary information as well as the experimental

values that went into the docking experiment. Finally, the tcsh script (TEE-SHELL script)

run_LIGPLOT will analyze the intermolecular interactions between each docked ligand and the

macromolecule. It uses the LIGPLOT and HBPLUS programs to produces a slew of result files

in each ligand folder (McDonald and Thornton 1994; Wallace, Laskowski et al. 1995). The most

important files produced by the script are the “.nnb” and “.hhb” which contain the non-hydrogen

bonds and hydrogen bonds, respectively, for that particular ligand-macromolecule pairing.

Figure 22 File Hierarchy Produced by DLGCLI

The input files (yellow) produce a slew of data when processed by the DLGCLI application

(blue). All files are placed in a folder with the same name as the docking log. Green boxes represent sub folders with the underlined text as the name. XXX stand for the designation of a particular docked ligand molecule (1 to the total number of LGAs run). The teal boxes represent the output files place in the main folder. The run_LIGPLOT script (purple) further processes the data and produces the files in the green box on the right. It also copies the folders of the best 5 ligands to the summary folder.

CountBonds

The CountBonds application tabulates the number of interactions per macromolecule residue and the specified upper percentile of best fit ligands. It uses the intermolecular bond data

54 produced by LIGPLOT and HBPLUS (“.hhb” and “.nnb” files) and the list of docking values for each ligand produced by DLGCLI. The application can also determine which ligand residues and atoms play the largest role in interacting with the macromolecule. The results are in the form of comma deliminated database that can be imported into any statistics package for further analysis and display.

PrimerFind

The heptad repeat 2 region of gp41 is very GC poor which presents a problem when developing primers for site-directed mutagensis. This program uses a brute force algorithm to generate all the possible primers for a specific mutation and test each one according to the published Stratagene Guidelines for mutagenenic primers (QuikChange Multi Site-Directed

Mutagenesis Kit). The web application, PrimerFind, accepts a DNA sequence with a desired mutation (Figure 23) and outputs a webpage of possible primers (Figure 24). The criteria for primers were as followed:

1. Optimal Length is 25-45 base pairs

a. Primers over 45bp may suffer from 2° structure formation

2. Melting Temperature ≥ 75°C or 78°C

a. Tm = 81.5 + .41(%GC)-675/N-%mismatch

i. N = primer length in bases

ii. %GC and %mismatch are whole numbers

b. In addition the Tm of each side should be above 55°C and as close as possible

3. Optimum GC Content ≥ 40%

4. Terminate with one or more G or C at the 3’ end

55 5. The Tm of each side should be as close as possible

6. The difference in length on each side of the substitution should not exceed 5bp

Figure 23 Input for PrimerFind

The sequence should be input as lowercase with the mutation(s) in uppercase. Multiple mutations should be in close proximity to each other (1-5bp). Values in the boxes represent default values. These values are highly dependant on the reagents used in the mutagenesis reaction and on the kit used for the experiment.

Figure 24 Sample Output from PrimerFind

These are the results from the E to K mutation of the gp41 residue 648. The user is able to refine their primer search without inputting the sequence over again. Also, the application outputs the reverse compliment as a possible primer. Clicking the number on the left of a particular primer opens a new page with summary information about that primer for ordering.

58 APPENDIX C: SEQUENCE INFORMATION

59 A large amount of HIV sequence information has been published at the HIV Sequence

Database at the Los Alamos National Laboratories (LANL; www.hiv.lanl.gov). The following two pages summarize the information garnered from the HR1 and HR2 regions of the Group M viruses in the database. Each row represents a location on the amino acid sequence. The numbering of HR1 and HR2 are based off of the uncleaved gp160 protein of the HXB2 standard strain. Cell in red represent an amino acid that is observed more than 50% of the time. Yellow

cells represent an amino acid observed between 100 and 50% of the time in the database. And

green cells represent amino acids seen between 1 and 99 times in the database. Amino acid

diversity values in red represent values less than .05.

Figure 25 HR1 Sequence Information

Figure 26 HR2 Sequence Information

62 REFERENCES

Arnold, K., L. Bordoli, et al. (2006). "The SWISS-MODEL workspace: a web-based

environment for protein structure homology modelling." Bioinformatics 22(2): 195-201.

Batoni, G., G. Maisetta, et al. (2006). "Human beta-defensin-3: a promising antimicrobial

peptide." Mini Rev Med Chem 6(10): 1063-73.

Berger, E. A., P. M. Murphy, et al. (1999). "Chemokine receptors as HIV-1 coreceptors: roles in

viral entry, tropism, and disease." Annu Rev Immunol 17: 657-700.

Bevins, C. L. (2005). "Events at the host-microbial interface of the gastrointestinal tract. V.

Paneth cell alpha-defensins in intestinal host defense." Am J Physiol Gastrointest Liver

Physiol 289(2): G173-6.

Bjellqvist, B., B. Basse, et al. (1994). "Reference points for comparisons of two-dimensional

maps of proteins from different human cell types defined in a pH scale where isoelectric

points correlate with polypeptide compositions." Electrophoresis 15(3-4): 529-39.

Bjellqvist, B., G. J. Hughes, et al. (1993). "The focusing positions of polypeptides in

immobilized pH gradients can be predicted from their amino acid sequences."

Electrophoresis 14(10): 1023-31.

Bron, A. J. and D. V. Seal (1986). "The defences of the ocular surface." Trans Ophthalmol Soc U

K 105 ( Pt 1): 18-25.

Buffy, J. J., M. J. McCormick, et al. (2004). "Solid-state NMR investigation of the selective

perturbation of lipid bilayers by the cyclic antimicrobial peptide RTD-1." Biochemistry

43(30): 9800-12.

63 Chan, D. C., D. Fass, et al. (1997). "Core structure of gp41 from the HIV envelope

glycoprotein." Cell 89(2): 263-73.

Choudhry, V., M. Y. Zhang, et al. (2007). "Cross-reactive HIV-1 neutralizing monoclonal

antibodies selected by screening of an immune human phage library against an envelope

glycoprotein (gp140) isolated from a patient (R2) with broadly HIV-1 neutralizing

antibodies." Virology.

Cole, A. L., A. Herasimtschuk, et al. (2007). "The retrocyclin analogue RC-101 prevents human

immunodeficiency virus type 1 infection of a model human cervicovaginal tissue

construct." Immunology 121(1): 140-5.

Cole, A. L., O. O. Yang, et al. (2006). "HIV-1 adapts to a retrocyclin with cationic amino acid

substitutions that reduce fusion efficiency of gp41." J Immunol 176(11): 6900-5.

Cole, A. M., T. Hong, et al. (2002). "Retrocyclin: a primate peptide that protects cells from

infection by T- and M-tropic strains of HIV-1." Proc Natl Acad Sci U S A 99(4): 1813-8.

Cole, A. M., D. R. Thapa, et al. (2005). "Decreased clearance of Pseudomonas aeruginosa from

airways of mice deficient in lysozyme M." J Leukoc Biol 78(5): 1081-5.

Cunliffe, R. N. (2003). "Alpha-defensins in the gastrointestinal tract." Mol Immunol 40(7): 463-

Danilova, N. (2006). "The evolution of immune mechanisms." J Exp Zoolog B Mol Dev Evol

306(6): 496-520.

Debnath, A. K., L. Radigan, et al. (1999). "Structure-based identification of small molecule

antiviral compounds targeted to the gp41 core structure of the human immunodeficiency

virus type 1." J Med Chem 42(17): 3203-9.

64 DeLano, W. L. (2002). The PyMOL Molecular Graphics System Palo Alto, CA, USA. , DeLano

Scientific.

Dimitrov, A. S., J. M. Louis, et al. (2005). "Conformational changes in HIV-1 gp41 in the course

of HIV-1 envelope glycoprotein-mediated fusion and inactivation." Biochemistry 44(37):

12471-9.

Gallo, S. A., A. Puri, et al. (2001). "HIV-1 gp41 six-helix bundle formation occurs rapidly after

the engagement of gp120 by CXCR4 in the HIV-1 Env-mediated fusion process."

Biochemistry 40(41): 12231-6.

Gallo, S. A., W. Wang, et al. (2006). "Theta-defensins prevent HIV-1 Env-mediated fusion by

binding gp41 and blocking 6-helix bundle formation." J Biol Chem 281(27): 18787-92.

Garzino-Demo, A. (2007). "Chemokines and defensins as HIV suppressive factors: an evolving

story." Curr Pharm Des 13(2): 163-72.

Goodsell, D. S., G. M. Morris, et al. (1996). "Automated docking of flexible ligands:

applications of AutoDock." J Mol Recognit 9(1): 1-5.

Hetenyi, C. and D. van der Spoel (2002). "Efficient docking of peptides to proteins without prior

knowledge of the binding site." Protein Sci 11(7): 1729-37.

Hetenyi, C. and D. van der Spoel (2006). "Blind docking of drug-sized compounds to proteins

with up to a thousand residues." FEBS Lett 580(5): 1447-50.

Joint United Nations Programme on HIV/AIDS. (2006). AIDS epidemic update. Geneva,

UNAIDS : World Health Organization.

Jun Tan, J., R. Kong, et al. (2005). "Prediction of the binding model of HIV-1 gp41 with small

molecule inhibitors." Conf Proc IEEE Eng Med Biol Soc 5: 4755-8.

65 Knipe, D. M., P. M. Howley, et al. (2001). Fundamental virology. Philadelphia, Lippincott

Williams & Wilkins.

Kong, R., J. J. Tan, et al. (2006). "Prediction of the binding mode between BMS-378806 and

HIV-1 gp120 by docking and molecular dynamics simulation." Biochim Biophys Acta

1764(4): 766-72.

Legrand, D., E. Elass, et al. (2006). "Interactions of lactoferrin with cells involved in immune

function." Biochem Cell Biol 84(3): 282-90.

Lehrer, R. I. (2004). "Primate defensins." Nat Rev Microbiol 2(9): 727-38.

Lehrer, R. I. and T. Ganz (2002). "Defensins of vertebrate animals." Curr Opin Immunol 14(1):

96-102.

Leonova, L., V. N. Kokryakov, et al. (2001). "Circular minidefensins and posttranslational

generation of molecular diversity." J Leukoc Biol 70(3): 461-4.

Liu, L., C. Zhao, et al. (1997). "The human beta-defensin-1 and alpha-defensins are encoded by

adjacent genes: two peptide families with differing disulfide topology share a common

ancestry." Genomics 43(3): 316-20.

Maddon, P. J., S. M. Molineaux, et al. (1987). "Structure and expression of the human and

mouse T4 genes." Proc Natl Acad Sci U S A 84(24): 9155-9.

Markosyan, R. M., F. S. Cohen, et al. (2003). "HIV-1 envelope proteins complete their folding

into six-helix bundles immediately after fusion pore formation." Mol Biol Cell 14(3):

926-38.

McDonald, I. K. and J. M. Thornton (1994). "Satisfying hydrogen bonding potential in proteins."

J Mol Biol 238(5): 777-93.

66 McGowan, I. (2006). "Microbicides: a new frontier in HIV prevention." Biologicals 34(4): 241-

55.

Melikyan, G. B., R. M. Markosyan, et al. (2000). "Evidence that the transition of HIV-1 gp41

into a six-helix bundle, not the bundle configuration, induces membrane fusion." J Cell

Biol 151(2): 413-23.

Morris, G. M., D. S. Goodsell, et al. (1998). "Automated docking using a Lamarckian genetic

algorithm and an empirical binding free energy function." Journal of Computational

Chemistry 19(14): 1639-1662.

Munk, C., G. Wei, et al. (2003). "The theta-defensin, retrocyclin, inhibits HIV-1 entry." AIDS

Res Hum Retroviruses 19(10): 875-81.

Nei, M. (1987). Molecular evolutionary genetics. New York, Columbia University Press.

Nguyen, T. X., A. M. Cole, et al. (2003). "Evolution of primate theta-defensins: a serpentine path

to a sweet tooth." Peptides 24(11): 1647-54.

Owen, S. M., D. Rudolph, et al. (2004). "A theta-defensin composed exclusively of D-amino

acids is active against HIV-1." J Pept Res 63(6): 469-76.

Owen, S. M., D. L. Rudolph, et al. (2004). "RC-101, a retrocyclin-1 analogue with enhanced

activity against primary HIV type 1 isolates." AIDS Res Hum Retroviruses 20(11): 1157-

65.

Perrin, C., E. Fenouillet, et al. (1998). "Role of gp41 glycosylation sites in the biological activity

of human immunodeficiency virus type 1 envelope glycoprotein." Virology 242(2): 338-

45.

Peschel, A. (2002). "How do bacteria resist human antimicrobial peptides?" Trends Microbiol

10(4): 179-86.

67 Pope, M. and A. T. Haase (2003). "Transmission, acute HIV-1 infection and the quest for

strategies to prevent infection." Nat Med 9(7): 847-52.

Robertson, D. L., J. P. Anderson, et al. (2000). "HIV-1 nomenclature proposal." Science

288(5463): 55-6.

Rodriguez-Chavez, I. R., M. Allen, et al. (2006). "Current advances and challenges in HIV-1

vaccines." Curr HIV/AIDS Rep 3(1): 39-47.

Rustomjee, R. and Q. Abdool Karim (2001). "Microbicide research and development--where

to?" HIV Clin Trials 2(3): 185-92.

Schwede, T., J. Kopp, et al. (2003). "SWISS-MODEL: An automated protein homology-

modeling server." Nucleic Acids Res 31(13): 3381-5.

Shattock, R. J. and J. P. Moore (2003). "Inhibiting sexual transmission of HIV-1 infection." Nat

Rev Microbiol 1(1): 25-34.

Sousa, S. F., P. A. Fernandes, et al. (2006). "Protein-ligand docking: current status and future

challenges." Proteins 65(1): 15-26.

Suntoke, T. R. and D. C. Chan (2005). "The fusion activity of HIV-1 gp41 depends on

interhelical interactions." J Biol Chem 280(20): 19852-7.

Tang, M., A. J. Waring, et al. (2006). "Orientation of a beta-hairpin antimicrobial peptide in lipid

bilayers from two-dimensional dipolar chemical-shift correlation NMR." Biophys J

90(10): 3616-24.

Tang, Y. Q., J. Yuan, et al. (1999). "A cyclic antimicrobial peptide produced in primate

leukocytes by the ligation of two truncated alpha-defensins." Science 286(5439): 498-

502.

68 Trabi, M., H. J. Schirra, et al. (2001). "Three-dimensional structure of RTD-1, a cyclic

antimicrobial defensin from Rhesus macaque leukocytes." Biochemistry 40(14): 4211-21.

Tran, D., P. A. Tran, et al. (2002). "Homodimeric theta-defensins from rhesus macaque

leukocytes: isolation, synthesis, antimicrobial activities, and bacterial binding properties

of the cyclic peptides." J Biol Chem 277(5): 3079-84.

Trinchieri, G. and A. Sher (2007). "Cooperation of Toll-like receptor signals in innate immune

defence." Nat Rev Immunol 7(3): 179-90.

Venkataraman, N., A. L. Cole, et al. (2005). "Cationic polypeptides are required for anti-HIV-1

activity of human vaginal fluid." J Immunol 175(11): 7560-7.

Wallace, A. C., R. A. Laskowski, et al. (1995). "LIGPLOT: a program to generate schematic

diagrams of protein-ligand interactions." Protein Eng 8(2): 127-34.

Wang, L. X., H. Song, et al. (2005). "Chemoenzymatic synthesis of HIV-1 gp41 glycopeptides:

effects of glycosylation on the anti-HIV activity and alpha-helix bundle-forming ability

of peptide C34." Chembiochem 6(6): 1068-74.

Wang, W., A. M. Cole, et al. (2003). "Retrocyclin, an antiretroviral theta-defensin, is a lectin." J

Immunol 170(9): 4708-16.

Wang, W., S. M. Owen, et al. (2004). "Activity of alpha- and theta-defensins against primary

isolates of HIV-1." J Immunol 173(1): 515-20.

Wei, X., J. M. Decker, et al. (2003). "Antibody neutralization and escape by HIV-1." Nature

422(6929): 307-12.

Weiss, R. A. (2003). "HIV and AIDS: looking ahead." Nat Med 9(7): 887-91.

Weissenhorn, W., A. Dessen, et al. (1997). "Atomic structure of the ectodomain from HIV-1

gp41." Nature 387(6631): 426-30.

69 Wilkins, M. R., E. Gasteiger, et al. (1999). "Protein identification and analysis tools in the

ExPASy server." Methods Mol Biol 112: 531-52.

Wu, L. and V. N. KewalRamani (2006). "Dendritic-cell interactions with HIV: infection and

viral dissemination." Nat Rev Immunol 6(11): 859-68.

Yamaguchi-Kabata, Y. and T. Gojobori (2000). "Reevaluation of amino acid variability of the

human immunodeficiency virus type 1 gp120 envelope glycoprotein and prediction of

new discontinuous epitopes." J Virol 74(9): 4335-50.