Structure, Vol. 13, 1887–1895, December, 2005, ª2005 Elsevier Ltd All rights reserved. DOI 10.1016/j.str.2005.11.005

‘‘Wide-Open’’ 1.3 A Structure of aMultidrug-Resistant HIV-1 Proteaseas a Drug Target

Philip Martin,1,5 John F. Vickrey,4,5

Gheorghe Proteasa,1 Yurytzy L. Jimenez,1

Zdzislaw Wawrzak,2 Mark A. Winters,3

Thomas C. Merigan,3 and Ladislau C. Kovari1,*1Department of Biochemistry and Molecular BiologyWayne State University School of Medicine540 East Canfield AvenueDetroit, Michigan 482012DND-CAT Synchrotron Research CenterNorthwestern UniversityArgonne, Illinois 604393Center for AIDS ResearchStanford UniversityStanford, California 943044Department of ChemistryUniversity of Texas—Pan American1201 West University DriveEdinburg, Texas 78541

Summary

This report examines structural changes in a highly

mutated, clinical multidrug-resistant HIV-1 protease,and the crystal structure has been solved to 1.3 A res-

olution in the absence of any inhibitor. This proteasevariant contains codon mutations at positions 10, 36,

46, 54, 62, 63, 71, 82, 84, and 90 that confer resistanceto protease inhibitors. Major differences between the

wild-type and the variant include a structural changeinitiated by the M36V mutation and amplified by addi-

tional mutations in the flaps of the protease, resulting

in a ‘‘wide-open’’ structure that represents an openingthat is 8 A wider than the ‘‘open’’ structure of the wild-

type protease. A second structural change is triggeredby the L90M mutation that results in reshaping the 23–

32 segment. A third key structural change of the prote-ase is due to the mutations from longer to shorter

amino acid side chains at positions 82 and 84.

Introduction

The causative agent for AIDS, the human immunodefi-ciency virus (HIV), has become one of the most challeng-ing infectious entities known to humanity. About 42 mil-lion people throughout the world are infected with HIV,and an estimated 14,000 are infected per day worldwide(UNAIDS, 2002).

Combination antiretroviral therapy, or HAART, is keyto the medical management of HIV infection, and thegoal of combination antiviral therapy is to suppressHIV replication in the patient for as long as possible.Eradication of the HIV infection is not possible with cur-rently available HAART regimens.

Protease inhibitors are a potent and selective classof antiviral drugs, and they target the HIV-1 protease

*Correspondence: kovari@med.wayne.edu5 These authors contributed equally to this work.

enzyme required for the cleavage of the Gag-Pol viralpolyprotein. Drug resistance is probably the most im-portant factor influencing failure of present treatmentapproaches to HIV infection. Numerous inhibitors ofHIV-1 protease have been described in an attempt toblock viral proteolytic maturation. Clinical studies ofseveral HIV-1 protease inhibitors have established thatthese compounds can profoundly suppress the levelsof virus in the blood (Holodniy et al., 1993; Katzensteinand Holodniy, 1995). However, treatment with thesedrugs, as with the reverse transcriptase inhibitors, se-lects for resistant viral variants (Deutsch et al., 1994),and the protease gene mutations associated with resis-tance to inhibitors are reviewed periodically (Johnsonet al., 2003).

The rapidly evolving HIV-1 protease requires contin-ued attention as a drug target for rational drug designdue to its essential role in the proteolytic processingof the viral Gag and Gag-Pol polyproteins (Tomasselliet al., 1990). Viral progeny that do not have a func-tional protease are noninfectious (Kohl et al., 1988),and small-molecule inhibitors of the protease efficientlyblock replication of HIV in vitro (Huff, 1991). With the de-velopment and distribution of protease inhibitors for thetreatment of HIV, there has been an increase in the num-ber of resistance mutations within proteases, which di-minishes the success of protease inhibitors (Baldwinet al., 1995; Coffin, 1995; Gatanaga et al., 2002; Jacob-sen et al., 1995; King et al., 1995; Patick et al., 1995;Ridky and Leis, 1995; Rose et al., 1996; Weber et al.,2002).

According to the active site expansion model, HIV-1protease multi-drug resistance is associated with a se-ries of conformational changes of the HIV-1 proteaseleading to an expanded active site cavity and, as a result,a diminished binding of the protease inhibitor (Logsdonet al., 2004). It is important to note that the active siteexpansion model is consistent with both the substrateenvelope hypothesis (Prabu-Jeyabalan et al., 2002a,2002b, 2003; Wu et al., 2003) and thermodynamic mea-surements of ligand binding (Kurt et al., 2003). In thesubstrate envelope model, Dr. Schiffer and coworkershave determined a series of crystal structures of HIVprotease-substrate complexes, and the authors pro-pose a hypothesis in which inhibitors that fit within thesubstrate envelope of HIV-1 protease may be more ef-fective and less susceptible to drug resistance muta-tions. Microcalorimetric measurements by Freire et al.indicate that drug-resistant mutants lower the affinityof the licensed inhibitors by two or three orders of mag-nitude (Leavitt and Freire, 2001; Ohtaka et al., 2002;Velazquez-Campoy et al., 2000, 2001).

The focus of this work is to investigate the structuralchanges in the HIV-1 viral protease that render proteaseinhibitors unable to inhibit viral replication. Protease in-hibitors currently in use were designed to bind to pro-teases with closed flaps. In this study, we used X-raycrystallographic analysis and report a 1.3 A resolution‘‘wide-open’’ structure of a multidrug-resistant HIV-1protease clinical isolate.

Structure1888

Results and Discussion

Overall Differences in the Multidrug-ResistantHIV-1 Protease

The HIV-1 protease variant, MDR 769, contains codonmutations at positions 10, 36, 46, 54, 62, 63, 71, 82, 84,and 90 that are known to confer drug resistance toU.S. FDA-approved protease inhibitors. One of the strik-ing differences between the wild-type and the MDR 769HIV-1 protease is a very large conformational change inthe flap region with a movement of more than 8 A relativeto the native structure. The wild-type structure (3PHV),known as the ‘‘open’’ form, crystallizes in space groupP41212 with cell constants very similar to those of theMDR 769 mutant, which crystallizes in space groupP41 (a = 52.24 A versus 45.04 A and c = 107.12 A versus105.77 A, respectively). Figure 1 illustrates the relativemovement in the amino acid Ca atoms along with thecrystallographic temperature factor B values of themultidrug-resistant HIV-1 protease relative to the wild-type crystal structure. Figure 2 shows an overlappingribbon diagram of MDR 769 and the wild-type structure(3PHV). Since the crystal structures have very closeenvironments—virtually identical in places—one cannotinvoke crystal packing as the reason for these confor-mational differences. It is clear from Figure 2 why werefer to this MDR structure as the ‘‘wide-open’’ form ofHIV protease.

The crystallographic 2-folds in the native structure,which orient two of the monomers in the asymmetricunit into a functional dimer, have become local 2-foldsin our structure. In performing the superposition (LSQABfrom CCP4 V4.2.2 [Fitzgerald, 1994]), we used only thehinge region of the structure, and only residues 3–6,24–28, and 67–69 were used to calculate the leastsquares superposition. The rms deviation of the Casatoms for the hinge region residues is 0.52 A, versus1.86 A if all of the Cas atoms of the protease are used.Therefore, we consider the deviations greater than 1 A,or roughly twice the rms deviation Ca, as significant.

Reshaping the HIV-1 Protease, Resultingin the ‘‘Wide-Open’’ Structure

The high-resolution structure of a MDR HIV-1 protease(isolate 769) permits the examination of structuralchanges relative to the wild-type, and it is correlatedwith the drug-resistance properties of this protease var-iant (Palmer et al., 1999). The largest observed structuralchanges affect the amino acid sequence from residue 36through 64 (Figure 2), resulting in a dramatic ‘‘opening’’of the flaps.

Figure 3 shows the symmetric movement of the flapsaway from the ‘‘open’’ form in the wild-type to the ‘‘wide-open’’ form in the MDR protease. The ‘‘open’’ form isstabilized at the tips of the flaps at residues I50–G51by four van der Waals (vdW) contacts and two hydrogenbonds. The hydrogen bonds are between pairs of amidenitrogens (G51A-N donor, G51B-N acceptor, or viceversa, and I50A-N donor, I50B-N acceptor). The hydro-gen bonding pairs are separated by 3.1 A (the G51pair) and 2.8 A (the I50 pair). The only way for this to hap-pen is for one of the peptides to be in the form of an enoltautomer (Ca-(OH)C = N-Ca) and the other to be in theform of a keto tautomer (Ca-(O = )C-N(H)-Ca). Only under

these circumstances can there be a donor nitrogen atomwith a hydrogen and acceptor nitrogen atom having apair of electrons to share.

The new position for each flap produces a single vdWinteraction, albeit a very good one, at a distance of 3.6 Afor each chain. This contact occurs between I50 Cd1 ofchain A and P81 Cg of chain B, and vice versa. The con-sequence of this is that P81 is displaced 2.6 A, pullingA82, itself a mutation, 1.2 A from its position in the nativestructure.

Figure 4 illustrates a fragment in the chain of eventsleading to the rearrangement in the flaps of the HIV-1protease. The first mutation in this region, M46L, dis-places the backbone over 2 A simply because a vdWcontact between the side chains of M36 and I15 (3.1 A)is not present in the mutated species. The effect of theM36V mutation has an impact on amino acid residues lo-cated at a distance. It extends three residues toward theN terminus and is extensively propagated in the forwarddirection (toward the C terminus). The Cg2 of V36 (Fig-ure 4) makes a new vdW contact with the Cd2 of L38(3.7 A). These two side chain atoms form two new vdWcontacts with G16 Ca (3.9 and 3.8 A, respectively), mov-ing the G16 loop over 2 A. In this case, the forces actingon the two sections of chain are reciprocal in nature. Notonly is G16 displaced by 2 A toward L38, but the entireG40 loop (Figure 4), from M36V to the next mutation,M46L, is shifted severely by 2.9 A in the vicinity of resi-due 46. The polymorphic M36V mutation contributesto resistance when present in combination with one ormore resistance mutations. The result is drug resistanceto amprenavir, atazanavir, indinavir, lopinavir, nelfinavir,ritonavir, and saquinavir (Stanford HIV Drug ResistanceDatabase, http://hivdb.stanford.edu; Shafer et al., 1999).

The G40 loop is not only acted upon by the M36V sub-stitution (Figure 4). The side chain of V36 (Cg1) makes anadditional vdW contact with V77 Cg1 (3.7 A), which, ac-companied by the I50-P80 interaction, shifts the main

Cα

(Å)

B v

alue

(Å

2 )

Residue number

00

5

10

20

15

25

30

35

40

0

5

10

20

15

25

30

35

40

20 40 60 80 100

Figure 1. Changes in Ca and B Values for HIV-1 MDR 769 Protease

Relative to the Wild-Type Structure

The protease amino acid residues are represented on the x axis, and

Ca (A) and B (A2) are shown on the double y axis. The differences in

atomic positions for the Ca atoms of MDR 769 HIV-1 protease and

the wild-type protease are shown (solid histogram). The temperature

factors (B values) for the corresponding residues are shown by the

striped histogram.

MDR HIV-1 Protease Represents a Drug Target1889

Figure 2. Structural Differences between

MDR 769 HIV-1 Protease and the Wild-Type

The dark-red ribbon indicates the MDR 769

protease variant, and the red ribbon repre-

sents the wild-type protease. The mutated

MDR amino acid side chains are light blue.

Orange, highlighted mutated residues are hy-

drophilic, and the sulfur-containing residue,

M90, is highlighted in yellow. A single sodium

ion is indicated as a yellow sphere.

chain containing R57 by over 2 A. Continuing with thedomino effect, R57 is in a stretch of b sheet that formshydrogen bonds to the G40 loop at K45. These multiplecontacts all contribute to the overall conformationalchange in the mutant, and they were all started by a sim-ple methionine-to-valine substitution at position 36 ina particularly vulnerable section of the chain.

The MDR patient isolate also contains the M46L muta-tion located in the flap region of the structure that hasvirtually no interaction with the rest of the protein, ex-cept for the very tips of the flap. From about residues46–55, the flap region is composed of an antiparallelb sheet, with residue I50 at the tip in a Type II b turn. Con-sequently, all of the side chains are pointing either aboveor below the plane of the b sheet into the solvent.

The analysis of the 46–55 segment reveals the impactof deleted side chain atoms. The wild-type Sd atom of

methionine at M46 forms two excellent vdW contactswith K55 Ca and Cg (3.5 and 3.3 A, respectively). Whenthese atoms disappear due to a mutation in MDR 769,completely new interactions are formed in the b sheet.In particular, the interactions twist the b sheet in sucha way that M46L Cb can make a vdW contact with F53Cb (4.0 A). The side chain of F56 then reorients itself sothat F53 Cz can form a new vdW contact with G49 Ca

(3.6 A).The outcome of this chain of events is that once a sin-

gle mutation is in place, an additional mutation at a favor-able location could then further contribute to any devia-tions that might be advantageous to the organism. Werefer to this as successive mutational reinforcement.

The next mutation present in the MDR isolate is theI54V substitution. In the native structure, the side chainof I54 is pointed out into the solvent and makes no

Figure 3. Stereodiagrams of Water Mole-

cules Forming a Network Bridge between

the Flaps of the HIV-1 MDR 769 Protease

The gray ball-and-stick diagram represents

the MDR protease, and the green ball-and-

stick diagram is the wild-type protease. Up-

per panel: blue spheres represent the water

molecules, and the hydrogen bond scaffold-

ing is represented in red. Lower panel: com-

parison between the MDR 769 variant and

the wild-type protease in green. Long, black

arrows represent movement between the

variant and the wild-type flaps. The short ar-

row (magenta) represents new van der Waals

contacts present in MDR 769 protease.

Structure1890

Figure 4. The Impact of the M36V Substi-

tution

A severe chain shift produced by a single mu-

tation is illustrated (native in solid green,

769MDR in yellow). The substitution of valine

for methionine at position 36 has lost a single

van der Waals (vdW) contact between Met-36

Sd and Ile-15 Cg1 (green spheres). The entire

chain moves so that a new vdW contact can

form between Leu-38 Cd2 and a repositioned

Ile-15 Cg1 (yellow spheres).

contacts whatsoever. The valine substitution leads toimproved vdW contacts with K55 O and V56 Cg1 (3.8 Aand 4.0 A, respectively). These added contacts areamong those responsible for getting the flaps to shiftby more than 8 A.

By now, the chain trace has essentially shifted relativeto the wild-type. Carbonyl groups are falling where Casshould be, and where the Cas amide nitrogen atomsare in the native structure, and so on. However, fromthe I62V position, the deviations are starting to diminish(Figure 1).

Figure 5 shows a close-up of the interactions. Wecomputed the surface area covered by intermolecularsymmetry contacts (Richards, 1985) for both the wild-type and MDR 769. The wild-type contact area has pri-

Figure 5. Crystal Packing Close-Up at the I50 Loop Region

A crystal packing close-up shows the symmetry-related residues

(green for one monomer, yellow for the other) that are within contact

distance (4 A) of the I50 loop (standard colors) for MDR 769 and the

wild-type HIV protease. The I50 loop forms new contacts with resi-

dues in the mutated I40 loop (M36V, S37N, M46L), Q61 (next to

I62V and L63P), and I72 (adjacent to A71V). Residues T91–I93, which

are in the region of the L90M mutation, complete the contact envi-

ronment.

mary interactions with symmetry-related residues atT91–G94 and I72, with a total contact area of 1,222 A2.In MDR 769, that area increases to 2,621 A2, primarilydue to new interactions with residues P39–R41 andQ61. Of particular interest is that the new contacts areall in regions that are within, or adjacent to, mutationsthemselves. In other words, the packing scheme forthe mutated structure has been adjusted to accommo-date the mutations. Also, as we have shown throughout,the rearrangement primarily deals with changes in atomvolumes and vdW contacts. This appears to be drivingthe conformational rearrangements.

Reshaping the HIV-1 Protease Active Site Cavity

around Residue 25The next major difference is at residue D30 with a Ca de-viation of 1.2 A. The residues from 23 through 32 formthe ‘‘floor’’ of the inhibitor binding cavity. The D30N mu-tation confers resistance to nelfinavir, the L24I mutationconfers resistance to indinavir and lopinavir, and theV32I mutation confers resistance to amprenavir, indina-vir, lopinavir, and ritonavir (Johnson et al., 2003).

By ‘‘tracing the shifts’’ in the amino acid residues, wecan follow the effect all the way back to the mutation atresidue L90M. The insertion of the C3 of M90 into a smallpocket formed by residue 25 (which moves only slightly)and G86-R87 forms several new vdW contacts that pre-viously did not exist.

The new vdW interactions pull the main chain G86,R87, and N88 closer in the direction of the M90 C3

atom. These residues, in turn, form backbone hydrogenbonds from N88 N to D29 O, and from R87 N to A28 O,drawing them in to include residue D30. An additionalside chain hydrogen bond between N88 OD1 and T31N also helps ‘‘pull in’’ the D30 main chain. The impactof the L90 mutation is cogent, since the leucine-to-methionine mutation affects the efficacy of eight proteaseinhibitor drugs, namely, amprenavir, fos-amprenavir,atazanavir, indinavir, lopinavir, nelfinavir, ritonavir, andsaquinavir (Johnson et al., 2003). Thus, there is a dominoeffect from the mutation of L90 that is 9.7 A away from

MDR HIV-1 Protease Represents a Drug Target1891

D30 (Figure 6). The L90M mutation results in successiveperturbations via residues 86–88, leading to the reshap-ing of the floor of the inhibitor binding cavity.

The Ca deviations were not only deviations at or nearthe mutation points, but also in places that were atsome distance from the mutations. For example, residueG16 moved more than 2 A (2.2 A) relative to the wild-typestructure. Although the residue has good electron den-sity, it has one of the highest B values of the structure(Figure 1). The wild-type coordinate set used for com-parison did not have refined B values. This could bedue to the low resolution of 3PHV and the flexibility ofthe G16 region (Figure 1).

We observed that positions 16 and 17, the Gly-Gly res-idues, deviate significantly between the two structures.Upon closer inspection, we discovered that the largemovement at residue G16 is connected by a chain ofevents because of the replacement of methionine 36by valine and is a result of the propagation of mutationalchanges. While the deviation is less for residues C67 andG68, it is close enough to the 1 A cutoff (C67 = 0.88 Ain chain A, C67 = 0.99 A in chain B) that we believe thatthe successive mutational reinforcement occurs hereas well.

Reshaping of the HIV-1 Protease Active Site Cavityaround Positions 82 and 84

Two key drug-resistance mutations are represented byamino acid substitutions at positions 82 and 84, andthe patient isolate MDR 769 contains both of these mu-tations (V82A and I84V). One important feature of MDR

Figure 6. L90M Influence on Chain Movement

Semitransparent spheres (yellow) represent the new van der Waals

contacts that develop between the L90M-CE atom and the G86-C

(3.6 A) and R87-CA (4.0 A) atoms. The resulting shifts in the back-

bone and side chains of MDR 769 (yellow) versus the wild-type

(green), as well as the new polar contacts that formed as a result

of this mutation, illustrate the changes that manifest themselves

via the domino effect.

769 is that the V82A and I84V mutations cause an in-crease in the volume of the active site cavity (Logsdonet al., 2004), and the loss of a sigma carbon-carbonbond results in an approximate change of 1.5 A in eachof the 82, 182, 84 and 184 amino acid residues. Sincethese four residues are in opposite corners of the activesite cavity, there is an approximate 3.0 A expansion in-side the active site.

A Water Scaffold Stabilizes the ‘‘Wide-Open’’

Active Site Cavity of the Multidrug-ResistantHIV-1 Protease

The quality and resolution of this crystal structure al-lowed us to determine the positions of 380 water mole-cules surrounding the MDR 769 HIV-1 protease. Ratherthan losing the stabilizing interchain hydrogen bondsfound in the wild-type ‘‘open’’ form, separated by morethan 12 A, the ordered water molecules provide stabilityto the ‘‘wide-open’’ structure of the MDR protease. Ascaffold of hydrogen bonds has formed between theends of the flaps by intervening water molecules, addingto the stabilization of the ‘‘wide-open’’ form (Figure 3).About 100 water molecules from the total of 380 are lo-cated in the active site cavity (Figure 7, upper panel),and they form a scaffold in the active site cavity, pre-venting the MDR HIV-1 protease from collapsing in theabsence of a ligand. The hydrogen bonding networkformed by about three layers of crystallographic watersin the active site cavity is illustrated in the lower panel ofFigure 7. The replacement of the waters by an inhibitormight be a logical step in designing protease inhibitorsagainst the MDR HIV.

Biologically Relevant Conformationsof the Crystal Form

While no one can either prove or disprove that any crys-tal structure is the same as the solution structure, weoffer the following relevant points for consideration. Fig-ure 8 illustrates the 769 MDR and wild-type HIV proteasedimers (standard CHNO colors) as space-filling modelsto indicate the volume and contact area of these struc-tures. The molecules that are in contact with the I50loop by crystallographic symmetry are also shown(gray). It is evident that a large cavity exists in the wild-type structure (space group P41212) that very neatly ac-cepts the MDR 769 I50-loop flaps in their new conforma-tion (space group P41). With the unit cells between thetwo structures very close to one another, basicallya crystallographic 2-fold along the diagonal in the wild-type crystal has been replaced by a local 2-fold in themutant. Indeed, our latest mutants of MDR 769 (A82F,A82S, and A82T) have all reverted back to space groupP41212 while still maintaining the ‘‘wide-open’’ formatand similar unit cell.

There is a lack of intermolecular hydrogen bonds inthis area in both structures. Both have a K55 N to Q92OE1 hydrogen bond at 2.9 A. The wild-type has one ad-ditional Q61 NE2 to G49 O hydrogen bond at 3.2 A. Note,however, that these are all through flexible side chains.MDR 769 does have additional flexible side chain hydro-gen bonds: K55 NZ to T91 O (3.0 A), K55 NZ to G92 O(3.3 A), Q92 NE2 to K55 O (2.9 A), and R41 NH2 (twoside chain conformations) to G48 O (2.5 A). Only one,R41 N to G49 O (2.9 A), is a main chain hydrogen bond.

Structure1892

Figure 7. HIV-1 MDR 769 Protease Stereodiagrams

Upper panel: Stereodiagram of water molecules shown in light blue in the active site cavity of the MDR HIV-1 protease. Lower panel: Stereo

diagram of the hydrogen bonding network (red color) formed by the water molecules in the active site cavity.

We observe that intermolecular symmetry contactsare made along stretches of chain that do not directly in-volve mutated residues. They are, rather, adjacent to, orin between (in the case of the I50 loop), these mutations.We would propose that, since all of the surface residuesat which the symmetry contacts are made are present inboth structures, MDR 769 packs the way it packs be-cause it is folded differently in the first place.

In general, we suggest that the controversy between‘‘crystal structure’’ and ‘‘solution structure’’ cannot beproved, except where physical evidence exists. Werethe solution structure significantly different from thecrystal structure, one would have to devise some asyet unknown mechanism whereby a molecule wouldhave to somehow adopt the crystal conformation beforenucleation could occur. Once a nucleation event did oc-cur, an additional unknown mechanism would need tobe devised to get the molecule being added to the grow-

ing crystal surface, again, into the ‘‘crystal confor-mation.’’ Or, we can simply say that the molecules in so-lution are in the ‘‘crystal conformation’’ already, theynucleate spontaneously when their solubility is de-creased sufficiently by the precipitant, and they thengrow in the normal way.

However, while we favor the crystal structure as beingone of a possible collection of solution structures, thereis ample evidence that there can be conformationalstates in crystals (DePristo et al., 2004). We cannot ruleout that the crystal structure is selected for from a num-ber of solution conformations (Straub, 1964) that are inrapid equilibrium with one another. We also distinguishthese from a ‘‘triggered’’ conformational change suchas occurs when a ligand (substrate or inhibitor) encoun-ters the active site of an enzyme. These latter conforma-tional states describe the transition from one equilibriumconformation to another (Ringe and Petsko, 1986).

Figure 8. Crystal Contact Comparisons be-

tween MDR 769 and the Wild-Type HIV-1 Pro-

tease

MDR 769 and the wild-type HIV protease

(standard colors) along with the molecules

that contact the I50 loop, which are related

by crystallographic-related symmetry (gray),

are shown. These mobile loops are seen to

fill a cavity in the wild-type structure (space

group P41212) and form different crystallo-

graphic contacts in the MDR 769 unit cell

(space group P41). The change in packing is

due to the multitude of small and large con-

formational differences between the two

structures, but it requires no significant

changes between the unit cells.

MDR HIV-1 Protease Represents a Drug Target1893

The structures of both the wild-type and MDR 769when overlapped with a structure of a HIV protease con-taining a peptide or drug (for example, wild-type HIVwith a peptidomimetic inhibitor, 1IIQ, or the mutantV82A-L90M plus indinavir, 1SDT) undergo a consider-able conformational change (not shown). Most of thisis at the I50 loop region and is an example of a ‘‘trig-gered’’ conformational change.

It is accepted that proteins must ‘‘breathe’’ in order tobind ligands such as substrates or inhibitors. Our ‘‘wide-open’’ form of HIV protease is the only one that is openenough for a substrate to enter the active site. Boththe wild-type structures and the drug bound structureswould require substances to enter the protease as ifthreading through the eye of a needle.

The S2, S1, S10, and S20 sites of the HIV protease havebeen mapped, and the peptides and drugs all span themonomers of the dimer to which they bind. We wouldlike to propose the following mechanism. In order fora drug to bind tightly, it might bind first to one monomer(or the other) while it is in a ‘‘wide-open’’ conformation,for example, in the P sites. The close proximity of thedrug extending into the P0 sites via electrostatic, hydro-phobic, and hydrogen bonding interactions (Ringe andPetsko, 1986) would then act as the ‘‘trigger’’ to closethe other monomer around the drug and bind it tightly.In the MDR 769 structure, the geometry of the bindingcavity has been so drastically altered that the closuredoes not occur. Also, the local geometry of each mono-mer has also been altered, and the initial binding eventis probably not nearly as strong as with the wild-typeenzyme.

Experimental Procedures

Protease Expression and Purification

The MDR 769 protease was overexpressed by using a T7 promoter

expression vector in conjunction with the E. coli host, BL21(DE3). In

brief, a fresh transformant of BL21(DE3) with the MDR 769 plasmid

was cultured in 5 ml YT medium containing 100 mg/ml ampicillin to

an A600 of 0.5. At this point, the 5 ml culture was used to inoculate

a liter of YT medium with 100 mg/ml ampicillin. After 20 hr of incuba-

tion with shaking at 37ºC, these cells were harvested at 10,000 3 g

for 5 min at 4ºC by using a Sovall RC5B Plus centrifuge (Vickrey

et al., 2003).

The MDR protease was isolated from inclusion bodies by using

a series of buffered washes, followed by denaturing in 6 M urea.

For purification of the unfolded protease, an anion exchange resin

(Q Sepharose, Amersham Biosciences) was used that allowed the

protease to pass through and the contaminants to bind. The pro-

tease was determined to be purified greater than 95% through

Coomassie blue-stained SDS-PAGE (Laemmli, 1970).

The 6 M urea was removed to refold the protease by using a series

of dialysis exchanges that were carried out at 4ºC. The first four

buffer changes consisted of 40 mM sodium phosphate (pH 7),

0.2% bME, and 10% glycerol, and the last two buffer changes

were 10 mM sodium acetate, 1 mM DTT (pH 5). The protease was

concentrated to between 5 and 13 mg/ml for storage at 4ºC.

Protease Crystallization and Data Collection

The hanging drop vapor diffusion method was used to form the bi-

pyrimidal crystals of the MDR 769 protease. Using a matrix screen

consisting of pH (5.5 to 7.5) verses sodium chloride (0.3 to 0.9 M),

the HIV-1 protease crystals formed overnight at 22ºC. Routinely,

0.2 mm crystals in the longest dimension were obtained after

14 days of incubation.

Protease crystals were placed in a cryoprotectant consisting of

30% 400 PEG with the well solution and were frozen in liquid nitro-

gen, and data were collected at 1.00 A wavelength at the Advanced

Photon Source (APS) (IMCA-CAT beamline), Argonne National Lab-

oratory (Argonne, IL). Crystallographic data were collected at the

APS with an Oxford cryostream.

Data were reduced to structure amplitudes with HKL2000 (Otwi-

nowski and Minor, 1997). The data statistics are shown in Table 1.

Crystallographic refinement was initiated by the addition of solvent

to a model previously refined at a lower (1.8 A) resolution (Logsdon

et al., 2004) with 20 cycles of ARP/wARP (version 6.0, [Perrakis et al.,

1997]), as implemented in the CCP4 suite of programs (version 4.2.2,

[Fitzgerald, 1994]). The initial R and Rfree went from 0.313 and 0.335,

respectively, to 0.183 and 0.228, respectively. When done, the

Table 1. X-Ray Diffraction Data for MDR 769

Experimental Conditions

X-ray source APS (IMCA-CAT)

Wavelength (A) 1.0 A

Sample temperature 100 K

Crystal Parameters

Resolution range (A) 20–1.3

Unit cell (A) a = b = 45.04; c = 105.77

Space group P41

Mosaicity 0.3º

Percent solventa 42.7%

Data Processing

Number of unique reflections 42,456

Redundancy 4.50 (1.7)

I/s(I) 30.5 (1.5)

Completeness (%)b 82 (26)

Rsymc (%) 4.3 (41.5)

a Computed based on SHELXL.b Values in parentheses represent the numbers in the highest-reso-

lution shell.c Rsym = S jI 2 <I>j/S I.

Table 2. Refinement Statistics for the 1.3 A MDR 769 Structure

Refinement Parameters

Number of reflections used 38,356

Resolution range (A) 20–1.3

Number of protein atoms 1,562a

Number of water molecules 382

R Factors

Rcrystb 0.14

Rfreec 0.21

s cutoff none

<B> Average Atomic Temperature Factors (A2)

<B> Protein 20.50

<B> Side chains 23.46

<B> Main chains 17.62

<B> Solvent 35.83

Root-Mean-Square Deviations from Ideal Geometry

Bonds (A) 0.01

Atom1 to Atom3 distances (A) 0.03

Ramachandran Plot

Favorable (%) 94.2

Additional (%) 5.8

Generous (%) 0

Forbidden (%) 0

a The refinement statistics include residues in alternate conforma-

tions. The true number of protein atoms is 1508.b Rcryst = S kFobsj 2 jFcalck/SjFobsj.c Rfree = S kFobsj2 jFcalck/SjFobsj, where Fobs are test set amplitudes

(851 reflections) not used in refinement.

Structure1894

procedure had added 382 waters. At this point, the structure was ex-

amined graphically with XtalView (McRee, 1999), and alternate con-

formation side chains were added to the model.

At this time, we also noticed that the first two waters added to the

structure had unusual coordination properties. Rather than the four

(maximum for water) roughly tetrahedral ligands associated with

a typical hydrogen bonded water, these two had five ligands (a sixth

ligand position was occupied by a symmetry related carbon atom)

arranged in a roughly octahedral pattern (Figure 1). This is typical

of a metal ion rather than a water. Since sodium was the only metal

in the media (from the sodium acetate buffer), we concluded that this

was, indeed, a sodium ion, and it was included in the high-resolution

refinement.

Due to the high resolution (1.3 A), the data and coordinates were

ported into SHELXL (Sheldrick and Schneider, 1997), and the refine-

ment continued. A total of 15 cycles of isotropic refinement gave

R and Rfree values a bit higher (0.192 and 0.247, respectively) than

ARP/wARP, presumably from tighter restraints (38356 data [2071

data were used for Rfree], 7779 parameters). Full anisotropic refine-

ment for an additional 15 cycles (38356 data, 17,510 parameters)

lowered the R and Rfree to the final values of 0.142 and 0.211, respec-

tively. The lowering of Rfree by 3.6% showed that we were justified

in using anisotropic refinement at a data:parameter ratio of 2.2:1

(Table 2; 38,356 data and 17,519 parameters).

Ribbon diagrams were generated with PYMOL (DeLano, 2002)

and were polished with Adobe Photoshop, V7.0. Molecular cartoons

and electron density (CNS maps) were generated with SPOCK

(Christopher, 1998).

Acknowledgments

This work was supported by National Institute of Health grants

(GM62990 and A1065294) and an American Foundation for AIDS re-

search grant (AmFAR: 106457-34-RGGN) to L.C.K. and a National

Foundation for Cancer Research grant to T.C.M. The Michigan Life

Science Corridor provided funding to enhance the structural biology

facility at Wayne State University. Portions of this work were per-

formed at the DuPont-Northwestern-Dow Collaborative Access

Team (DND-CAT) Synchrotron Research Center located at Sector

5 of the Advanced Photon Source (APS). DND-CAT is supported

by E.I. DuPont de Nemours & Co., The Dow Chemical Company,

the U.S. National Science Foundation through Grant DMR-9304725

and the State of Illinois through the Department of Commerce and

the Board of Higher Education Grant IBHE HECA NWU 96. The use

of the APS was supported by the U.S. Department of Energy, Basic

Energy Sciences, Office of Energy Research under Contract No.

W-31-102-Eng-38.

Received: April 26, 2004

Revised: August 24, 2005

Accepted: August 25, 2005

Published: December 13, 2005

References

Baldwin, E.T., Bhat, T.N., Liu, B., Pattabiraman, N., and Erickson,

J.W. (1995). Structural basis of drug resistance for the V82A mutant

of HIV-1 proteinase. Nat. Struct. Biol. 2, 244–249.

Christopher, J.A. (1998). SPOCK: The Structural Properties Obser-

vation and Calculation Kit (Program Manual) (College Station, TX:

The Center for Macromolecular Design, Texas A&M University).

Coffin, J.M. (1995). HIV population dynamics in vivo: implications for

genetic variation, pathogenesis, and therapy. Science 267, 483–489.

DeLano, W.L. (2002). The PyMOL Molecular Graphics System (San

Carlos, CA: Delano Scientific).

DePristo, M.A., de Bakker, P.I., and Blundell, T.L. (2004). Heteroge-

neity and inaccuracy in protein structures solved by x-ray crystallog-

raphy. Structure (Camb) 12, 831–838.

Deutsch, P., Teppler, H., Squires, K., Saag, M., Pomerantz, T.,

Bjornsson, T., Waldman, S., Osborne, B., Justice, S., Deloach, J.,

et al. (1994). Antiviral activity of L-735,524, an HIV protease inhibitor,

in infected patients. (Abstr.). Intersci. Conf. Antimicrob. Agents Che-

mother. Proc. 34, 159.

Fitzgerald, P.M. (1994). A generalized approach to the fitting of non-

peptide electron density. From first map to final model. Proceedings

of the CCP4 study weekend 35, 125–132.

Gatanaga, H., Suzuki, Y., Tsang, H., Yoshimura, K., Kavlick, M.F.,

Nagashima, K., Gorelick, R.J., Mardy, S., Tang, C., Summers, M.F.,

and Mitsuya, H. (2002). Amino acid substitutions in Gag protein at

non-cleavage sites are indispensable for the development of

a high multitude of HIV-1 resistance against protease inhibitors.

J. Biol. Chem. 277, 5952–5961.

Holodniy, M., Katzenstein, D., Winters, M., Montoya, J., Shafer, R.,

Kozal, M., Ragni, M., and Merigan, T.C. (1993). Measurement of

HIV virus load and genotypic resistance by gene amplification in

asymptomatic subjects treated with combination therapy. J. Acquir.

Immune Defic. Syndr. 6, 366–369.

Huff, J.R. (1991). HIV protease: a novel chemotherapeutic target for

AIDS. J. Med. Chem. 34, 2305–2314.

Jacobsen, H., Yasargil, K., Winslow, D.L., Craig, J.C., Krohn, A., Dun-

can, I.B., and Mous, J. (1995). Characterization of human immunode-

ficiency virus type 1 mutants with decreased sensitivity to protein-

ase inhibitor Ro 31–8959. Virology 206, 527–534.

Johnson, V.A., Brun-Vezinet, F., Clotet, B., Conway, B., D’Aquila,

R.T., Demeter, L.M., Kuritzkes, D.R., Pillay, D., Schapiro, J.M., Tele-

nti, A., and Richman, D.D. (2003). Drug resistance mutations in

HIV-1. Top. HIV Med. 11, 215–221.

Katzenstein, D.A., and Holodniy, M. (1995). HIV viral load quantifica-

tion, HIV resistance, and antiretroviral therapy. AIDS Clin. Rev. 277–

303.

King, R.W., Winslow, D.L., Garber, S., Scarnati, H.T., Bachelor, L.,

Stack, S., and Otto, M.J. (1995). Identification of a clinical isolate

of HIV-1 with an isoleucine at position 82 of the protease which re-

tains susceptibility to protease inhibitors. Antiviral Res. 28, 13–24.

Kohl, N.E., Emini, E.A., Schleif, W.A., Davis, L.J., Heimbach, J.C.,

Dixon, R.A., Scolnick, E.M., and Sigal, I.S. (1988). Active human im-

munodeficiency virus protease is required for viral infectivity. Proc.

Natl. Acad. Sci. USA 85, 4686–4690.

Kurt, N., Scott, W.R., Schiffer, C.A., and Haliloglu, T. (2003). Cooper-

ative fluctuations of unliganded and substrate-bound HIV-1 prote-

ase: a structure-based analysis on a variety of conformations from

crystallography and molecular dynamics simulations. Proteins 51,

409–422.

Laemmli, U.K. (1970). Cleavage of structural proteins during the as-

sembly of the head of bacteriophage T4. Nature 227, 680–685.

Leavitt, S., and Freire, E. (2001). Direct measurement of protein bind-

ing energetics by isothermal titration calorimetry. Curr. Opin. Struct.

Biol. 11, 560–566.

Logsdon, B.C., Vickrey, J.F., Martin, P.D., Proteasa, G., Koepke, J.I.,

Terlecky, S.R., Wawrzak, Z., Winters, M.A., Merigan, T.C., and

Kovari, L.C. (2004). Crystal structure of a multidrug-resistant human

immunodeficiency virus type 1 protease reveals an expanded

active-site cavity. J. Virol. 78, 3123–3132.

McRee, D.E. (1999). XtalView/Xfit–a versatile program for manipulat-

ing atomic coordinates and electron density. J. Struct. Biol. 125,

156–165.

Ohtaka, H., Velazquez-Campoy, A., Xie, D., and Freire, E. (2002).

Overcoming drug resistance in HIV-1 chemotherapy: the binding

thermodynamics of Amprenavir and TMC-126 to wild-type and

drug-resistant mutants of the HIV-1 protease. Protein Sci. 11,

1908–1916.

Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction

data collected in oscillation mode. Methods Enzymol. 276, 307–326.

Palmer, S., Shafer, R.W., and Merigan, T.C. (1999). Highly drug-

resistant HIV-1 clinical isolates are cross-resistant to many antiretro-

viral compounds in current clinical development. AIDS 13, 661–667.

Patick, A.K., Rose, R., Greytok, J., Bechtold, C.M., Hermsmeier,

M.A., Chen, P.T., Barrish, J.C., Zahler, R., Colonno, R.J., and Lin,

P.F. (1995). Characterization of a human immunodeficiency virus

type 1 variant with reduced sensitivity to an aminodiol protease in-

hibitor. J. Virol. 69, 2148–2152.

Perrakis, A., Sixma, T.K., Wilson, K.S., and Lamzin, V.S. (1997).

wARP: improvement and extension of crystallographic phases by

MDR HIV-1 Protease Represents a Drug Target1895

weighted averaging of multiple refined dummy atomic models. Acta

Crystallogr. D Biol. Crystallogr. 53, 448–455.

Prabu-Jeyabalan, M., King, N.M., Nalivaika, E., Scott, W.R.P., and

Schiffer, C.A. (2002a). Drug resistance and substrate recognition in

HIV-1 protease. Antivir. Ther. 7, S36.

Prabu-Jeyabalan, M., Nalivaika, E., and Schiffer, C.A. (2002b). Sub-

strate shape determines specificity of recognition for HIV-1 prote-

ase: analysis of crystal structures of six substrate complexes. Struc-

ture (Camb) 10, 369–381.

Prabu-Jeyabalan, M., Nalivaika, E.A., King, N.M., and Schiffer, C.A.

(2003). Viability of a drug-resistant human immunodeficiency virus

type 1 protease variant: structural insights for better antiviral ther-

apy. J. Virol. 77, 1306–1315.

Richards, F.M. (1985). Calculation of molecular volumes and areas

for structures of known geometry. Methods Enzymol. 115, 440–464.

Ridky, T., and Leis, J. (1995). Development of drug resistance to

HIV-1 protease inhibitors. J. Biol. Chem. 270, 29621–29623.

Ringe, D., and Petsko, G.A. (1986). Study of protein dynamics by

X-ray diffraction. Methods Enzymol. 131, 389–433.

Rose, R.E., Gong, Y.F., Greytok, J.A., Bechtold, C.M., Terry, B.J.,

Robinson, B.S., Alam, M., Colonno, R.J., and Lin, P.F. (1996). Human

immunodeficiency virus type 1 viral background plays a major role in

development of resistance to protease inhibitors. Proc. Natl. Acad.

Sci. USA 93, 1648–1653.

Shafer, R.W., Stevenson, D., and Chan, B. (1999). Human immuno-

deficiency virus reverse transcriptase and protease sequence data-

base. Nucleic Acids Res. 27, 348–352.

Sheldrick, G.M., and Schneider, T.R. (1997). SHELXTL: high resolu-

tion refinement. Methods Enzymol. 277, 319–343.

Straub, F.B. (1964). Formation of the secondary and tertiary struc-

ture of enzymes. Adv. Enzymol. Relat. Areas Mol. Biol. 26, 89–114.

Tomasselli, A.G., Hui, J.O., Sawyer, T.K., Staples, D.J., Bannow, C.,

Reardon, I.M., Howe, W.J., DeCamp, D.L., Craik, C.S., and Heinrik-

son, R.L. (1990). Specificity and inhibition of proteases from human

immunodeficiency viruses 1 and 2. J. Biol. Chem. 265, 14675–14683.

UNAIDS (2002). AIDS Epidemic Update (www.unaids.org/html/pub/

Publications/IRC-pub03/epiupdate2002_en_pdf.pdf).

Velazquez-Campoy, A., Todd, M.J., and Freire, E. (2000). HIV-1 pro-

tease inhibitors: enthalpic versus entropic optimization of the bind-

ing affinity. Biochemistry 39, 2201–2207.

Velazquez-Campoy, A., Kiso, Y., and Freire, E. (2001). The binding

energetics of first- and second-generation HIV-1 protease inhibitors:

implications for drug design. Arch. Biochem. Biophys. 390, 169–175.

Vickrey, J.F., Logsdon, B.C., Proteasa, G., Palmer, S., Winters, M.A.,

Merigan, T.C., and Kovari, L.C. (2003). HIV-1 protease variants from

100-fold drug resistant clinical isolates: expression, purification and

crystallization. Protein Expr. Purif. 28, 165–172.

Weber, J., Mesters, J.R., Lepsik, M., Prejdova, J., Svec, M., Sponar-

ova, J., Mlcochova, P., Skalicka, K., Strisovsky, K., Uhlikova, T., et al.

(2002). Unusual binding mode of an HIV-1 protease inhibitor explains

its potency against multi-drug-resistant virus strains. J. Mol. Biol.

324, 739–754.

Wu, T.D., Schiffer, C.A., Gonzales, M.J., Taylor, J., Kantor, R., Chou,

S., Israelski, D., Zolopa, A.R., Fessel, W.J., and Shafer, R.W. (2003).

Mutation patterns and structural correlates in human immunodefi-

ciency virus type 1 protease following different protease inhibitor

treatments. J. Virol. 77, 4836–4847.

Accession Numbers

Coordinates have been deposited in the Protein Data Bank with ac-

cession code 1TW7.