Crystal Structures of 8Cl and 9Cl TIBO Complexed with Wild-type HIV1 RT and 8Cl TIBO Complexed with the Tyr181Cys HIV1 RT Drug-resistant Mutant

J. Mol. Biol. (1996) 264, 1085–1100

Crystal Structures of 8-Cl and 9-Cl TIBO Complexedwith Wild-type HIV-1 RT and 8-Cl TIBO Complexedwith the Tyr181Cys HIV-1 RT Drug-resistant Mutant

Kalyan Das 1, Jianping Ding 1, Yu Hsiou 1, Arthur D. Clark Jr 1

Henri Moereels 2, Luc Koymans 2, Koen Andries 3, Rudi Pauwels 4

Paul A. J. Janssen 2, Paul L. Boyer 5, Patrick Clark 5

Richard H. Smith Jr 5,6, Marilyn B. Kroeger Smith 5

Christopher J. Michejda 5, Stephen H. Hughes 5 and Edward Arnold 1*

Human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT)1Center for Advancedis an important target for chemotherapeutic agents used in the treatmentBiotechnology and Medicine

and Department of of AIDS; the TIBO compounds are potent non-nucleoside inhibitors ofHIV-1 RT (NNRTIs). Crystal structures of HIV-1 RT complexed with 8-ClChemistry, RutgersTIBO (R86183, IC50 = 4.6 nM) and 9-Cl TIBO (R82913, IC50 = 33 nM) haveUniversity, 679 Hoes Lanebeen determined at 3.0 Å resolution. Mutant HIV-1 RT, containing Cys inPiscataway, NJ 08854-5638place of Tyr at position 181 (Tyr181Cys), is highly resistant to manyUSANNRTIs and HIV-1 variants containing this mutation have been selected2Center for Molecular Design in both cell culture and clinical trials. We also report the crystal structure

Janssen Research Foundation of Tyr181Cys HIV-1 RT in complex with 8-Cl TIBO (IC50 = 130 nM)Antwerpsesteenweg 37 determined at 3.2 Å resolution. Averaging of the electron density mapsB-2350 Voselaar, Belgium computed for different HIV-1 RT/NNRTI complexes and from diffraction

datasets obtained using a synchrotron source from frozen (−165 °C) and3Janssen Researchcooled (−10 °C) crystals of the same complex was employed to improveFoundation, Turnhoutsewegthe quality of electron density maps and to reduce model bias.30, B-2340 Beerse, BelgiumThe overall locations and conformations of the bound inhibitors in the4TIBOTEC, Institute for complexes containing wild-type HIV-1 RT and the two TIBO inhibitors are

Antiviral Research, Drie very similar, as are the overall shapes and volumes of the non-nucleosideEikenstraat 661, B-2650 inhibitor-binding pocket (NNIBP). The major differences between the twoEdegem, Belgium wild-type HIV-1 RT/TIBO complexes occur in the vicinity of the TIBO

chlorine substituents and involve the polypeptide segments around the5ABL-Basic Researchb5-b6 connecting loop (residues 95 to 105) and the b13-b14 hairpinProgram, NCI-Frederick(residues 235 and 236). In all known structures of HIV-1 RT/NNRTICancer Research andcomplexes, including these two, the position of the b12-b13 hairpin or theDevelopment Center, P.O.‘‘primer grip’’ is significantly displaced relative to the position in theBox B, Frederick, MDstructure of HIV-1 RT complexed with a double-stranded DNA and in21701-1013, USAunliganded HIV-1 RT structures. Since the primer grip helps to position

6Department of Chemistry the template-primer, this displacement suggests that binding of NNRTIsWestern Maryland College would affect the relative positions of the primer terminus and theWestminster, MD 21157 polymerase active site. This could explain biochemical data showing thatUSA NNRTI binding to HIV-1 RT reduces efficiency of the chemical step of

DNA polymerization, but does not prevent binding of either dNTPs orDNA.

*Corresponding author When the structure of the Tyr181Cys mutant HIV-1 RT in complex with

Abbreviations used: HIV-1, human immunodeficiency virus type 1; NNIBP, non-nucleoside inhibitor-bindingpocket; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor;RT, reverse transcriptase; Tyr181Cys HIV-1 RT, mutant HIV-1 RT containing cysteine in place of tyrosine atposition 181; a-APA, a-anilinophenylacetamide; nevirapine (BI-RG-587), 11-cyclopropyl-5,11-dihydro-4-methyl-6H-dipyrido{3,2-b:2',3'-e}(1,4)diazepin-6-one; TIBO, tetrahydroimidazo(4,5,1-jk)(1,4)-benzodiazepin-2(1H)-one and-thione; NNIBP, NNRTI binding pocket.

0022–2836/96/501085–16 $25.00/0 7 1996 Academic Press Limited

Structural Biology of HIV-1 RT/TIBO Interaction1086

8-Cl TIBO is compared with the corresponding structure containingwild-type HIV-1 RT, the overall conformations of Tyr181Cys andwild-type HIV-1 RT and of the 8-Cl TIBO inhibitors are very similar. Somepositional changes in the polypeptide backbone of the b6-b10-b9 sheetcontaining residue 181 are observed when the Tyr181Cys and wild-typecomplexes are compared, particularly near residue Val179 of b9. In the p51subunit, the Cys181 side-chain is oriented in a similar direction to theTyr181 side-chain in the wild-type complex. However, the electron densitycorresponding to the sulfur of the Cys181 side-chain in the p66 subunit isvery weak, indicating that the thiol group is disordered, presumablybecause there is no significant interaction with either 8-Cl TIBO or nearbyamino acid residues. In the mutant complex, there are slightrearrangements of the side-chains of other amino acid residues in theNNIBP and of the flexible dimethylallyl group of 8-Cl TIBO; theseconformational changes could potentially compensate for the interactionsthat were lost when the relatively large tyrosine at position 181 wasreplaced by a less bulky cysteine residue. In the corresponding wild-typecomplex, Tyr181 in the p66 subunit has significant interactions with thebound inhibitor and the position of the Tyr181 side-chain is well definedin both subunits. Apparently the Tyr181:Cys mutation eliminatesfavorable contacts of the aromatic ring of the tyrosine and the boundinhibitor, reducing the stability of NNRTI binding. This is consistent withthe observation that the Tyr181Cys mutant HIV-1 RT is more resistant toNNRTIs that have extensive interactions with the Tyr181 side-chain. Thisinterpretation is supported by a recent biochemical study which found thatan NNRTI dissociates from its complex with Tyr181Cys HIV-1 RT fasterthan it does from a complex with wild-type HIV-1 RT.

7 1996 Academic Press Limited

Keywords: antiviral inhibitors; reverse transcriptase; drug resistance;protein structure; drug design

Introduction

The rate of replication of human immunodefi-ciency virus type 1 (HIV-1) in infected patients ishigh; billions of new virus particles are producedevery day (Ho et al., 1995; Wei et al., 1995). Reversetranscriptase (RT) is responsible for copying thesingle-stranded viral RNA genome into double-stranded DNA, which is subsequently integratedinto host cell chromosomes by the viral enzymeintegrase. RT is an important target for drugtherapy, not only because it is essential for viralreplication but it also contains multiple sites wheredrugs can bind.

Two major classes of HIV-1 RT inhibitors havebeen developed. The nucleoside RT inhibitors(NRTIs) have been widely used to treat AIDSpatients (for reviews, see Larder, 1993; Schinazi,1993; De Clercq, 1995a,b). However, many of thesedrugs are relatively toxic, which, coupled with theemergence of drug-resistant viral variants, haslimited the therapeutic efficacy of the NRTIs (forreviews, see Schinazi, 1993; Tantillo et al., 1994;De Clercq, 1994, 1995a). The other class of HIV-1 RTinhibitors are the non-nucleosides (NNRTIs) whichare highly specific for HIV-1 RT (for reviews, seeYoung, 1993; De Clercq, 1995a,b; Arnold et al.,1996). The NNRTIs are much less toxic than theNRTIs; nonetheless, the emergence of drug-resist-

ant viral strains has limited the therapeutic efficacyof NNRTIs. The available structural and biochemi-cal data make it clear that, although the NNRTIs arechemically diverse, they all bind at a common sitein HIV-1 RT. However, the binding affinities andinhibitory activities of NNRTIs vary significantly,even among different derivatives in a single family.Mutations that engender resistance to NNRTIscluster around the NNRTI binding pocket (NNIBP).Although many HIV-1 RT NNRTI-resistant vari-ants show some degree of cross-resistance toseveral NNRTIs, individual inhibitors engender aparticular and characteristic spectrum of resistancemutations. These variations could be due todifferences in the specific interactions between aparticular inhibitor and the amino acid residuesthat make up the NNIBP.

R86183, an 8-chloro TIBO derivative (8-Cl TIBO,tivirapine), and R82913, a 9-chloro TIBO derivative(9-Cl TIBO), are two members of the TIBO familyof NNRTIs (Figure 1). 8-Cl TIBO has a selectivityindex (ratio of cytotoxic to inhibitory concen-trations) of about 30,000 and inhibits HIV-1 RT inMT-4 cells with an IC50 value of 4.6 nM (Pauwelset al., 1994). The selectivity index for 8-Cl TIBO isabout ten times higher than that for the unsubsti-tuted prototype TIBO derivative R82150 and iscomparable to that for AZT. With a selectivity indexof 1000 and an IC50 value of 33 nM, 9-Cl TIBO is

Structural Biology of HIV-1 RT/TIBO Interaction 1087

relatively less potent and somewhat more toxicthan 8-Cl TIBO. The cytotoxicity CC50 value for 8-ClTIBO is 140 mM and the CC50 value for 9-Cl TIBOis 34 mM. Various biochemical and clinical studies(Larder, 1992; Balzarini et al., 1993a,b; Byrnes et al.,1993; Boyer et al., 1994; Pauwels et al., 1994;Vandamme et al., 1994) have shown that mutationsconferring resistance to TIBO compounds includeLeu100 : Ile, Lys103 : Asn, Val106 : Ala,Glu138 : Lys, Val179 : Asp, Tyr181 : Cys/Ile,and Tyr188 : His/Leu. Changes of amino acidresidues in the NNIBP region of HIV-1 RT havedifferential effects on inhibition by 8-Cl TIBO and9-Cl TIBO. For example, HIV-1 RT mutantscontaining Tyr181 : Cys and Val179 : Asp mu-tations in p66, and Glu138 : Lys in p51 aredifferentially sensitive to these two TIBO deriva-tives (Boyer et al., 1994). The Tyr181 : Cys mutant(Tyr181Cys) of HIV-1 RT is frequently selected inthe presence of NNRTIs, and is quite resistant tomost NNRTIs including 9-Cl TIBO, but is relativelysusceptible to inhibition by 8-Cl TIBO (theIC50 value for this mutant is 130 nM; Pauwelset al., 1994). For nevirapine (Byrnes et al.,1993; Richman et al., 1994) and a-APA (R89439;Pauwels et al., 1993) the IC50 values for theTyr181Cys HIV-1 RT mutant are 22.7 mM (0113-fold higher than for wild-type HIV-1 RT) and5.9 mM (01000-fold higher than for a wild-typeHIV-1 RT), respectively.

It has been proposed that losing favorableinteractions between the aromatic ring of Tyr181and bound NNRTIs is responsible for NNRTIresistance in the Tyr181Cys HIV-1 RT variant(Sardana et al., 1992; Nanni et al., 1993; Tantilloet al., 1994). Recent kinetic studies of the Tyr181CysHIV-1 RT enzyme (Spence et al., 1996), alone and incomplex with nevirapine, have suggested that themutation primarily affects inhibitor binding. Nevi-rapine was found to dissociate faster from theTyr181Cys mutant HIV-1 RT complex than from acomplex with wild-type HIV-1 RT. Both biochemi-cal studies and the ease with which the variant isselected suggest that the polymerization efficiencyof the Tyr181Cys mutant HIV-1 RT is notsignificantly impaired compared to wild-typeHIV-1 RT (Spence et al., 1996).

Knowledge of the structures of a complex ofTyr181Cys HIV-1 RT with 8-Cl TIBO and ofcomplexes of 8-Cl TIBO and 9-Cl TIBO withwild-type HIV-1 RT, which have been crystallizedusing similar conditions and in the same crystalform, should permit a direct comparison of theseclosely related structures. This paper reports thecrystal structures of wild-type HIV-1 RT incomplexes with both 8-Cl TIBO and 9-Cl TIBO at3.0 A resolution, and the crystal structure ofTyr181Cys mutant HIV-1 RT in complex with 8-ClTIBO at 3.2 A resolution. Prior to these structuredeterminations, the X-ray crystal structures of anumber of HIV-1 RT/NNRTI complexes(Kohlstaedt et al., 1992; Ding et al., 1995b; Ren et al.,1995a) and a complex of HIV-1 RT/DNA/Fab

(Jacobo-Molina et al., 1993) were reported. Ren et al.(1995b) reported an independent structure determi-nation of 9-Cl TIBO complexed with HIV-1 RT in adifferent crystal form (this structure is discussed ina later section). Structures of unliganded HIV-1 RThave also been determined in a number of distinctcrystal forms (Esnouf et al., 1995; Rodgers et al.,1995; Hsiou et al., 1996; E. Arnold et al.,unpublished results).

Results and Discussion

Structure of the NNIBP in complexes of HIV-1RT with TIBO and other NNRTIs

HIV-1 RT is a heterodimer consisting of twopolypeptide chains, p66 and p51. The p51 subunithas the same sequence as the 440 N-terminal aminoacid residues of p66. This constitutes the poly-merase domain. The p66 subunit also containsRNase H, which p51 lacks. In the HIV-1 RT/TIBOcomplex crystals, an asymmetric unit contains onep66/p51 heterodimer and one inhibitor molecule.The folding of the polymerase domains of p66 andp51 into fingers, palm, thumb, and connectionsubdomains is similar to that reported for the otherHIV-1 RT structures.

As in other HIV-1 RT/NNRTI structures,the TIBO inhibitors are located in the NNIBP andadopt a ‘‘butterfly-like’’ conformation (Ding et al.,1995a; Ren et al., 1995a). In this analogy, thedimethylallyl group of TIBO, or wing I of thebutterfly (Ding et al., 1995a), interacts with aminoacid residues Pro95, Tyr181, Tyr188, Gly190,and Trp229 of p66, and the benzodiazepinonegroup of TIBO (wing II) interacts with Lys101,Lys103, Val106, Phe227, His235, Pro236, and Tyr318of p66 (Figures 1 and 2). Residues Leu100 andLeu234 of p66 interact with both wings I and II fromthe top, Val179 from the front, and Tyr188 from theback (or the tail) of the butterfly (Figure 2(a)).The amino acid residues surrounding thebound inhibitor are mostly hydrophobic; fivecontain aromatic side-chains. The only aminoacid residues around the NNIBP that havecharged side-chains (Lys101 and Lys103 of p66 andGlu138 of p51) are located near the putativeentrance to the binding pocket (Ding et al., 1995b).In the following discussion, residues that arementioned without any chain identification are inthe p66 subunit.

In contrast to TIBO derivatives, the otherNNRTIs whose structures have been determined incomplex with HIV-1 RT (Kohlstaedt et al., 1992;Ding et al., 1995b; Ren et al., 1995a) contain twodistinct aromatic rings corresponding to the twowings in the butterfly analogy (Ding et al., 1995a).In those structures, the wing I aromatic ring hasextensive hydrophobic contacts with the surround-ing amino acid residues; wing II has relativelyfewer interactions with the surrounding aminoacids. In the HIV-1 RT complexes with a-APA(Ding et al., 1995b; Ren et al., 1995a), nevirapine


Figure 1. Chemical structureswith the numbering scheme usedand distances (E3.6 A) betweenatoms of the TIBO inhibitor and ofthe amino acid residues of theNNIBP for: (a) 8-Cl TIBO (R86183,tivirapine) complexed with wild-type HIV-1 RT; (b) 8-Cl TIBOcomplexed with Tyr181Cys mutantHIV-1 RT; and (c) 9-Cl TIBO(R82913) complexed with wild-typeHIV-1 RT. An NNIBP residue isshown only if atoms of that residueare E3.6 A from an inhibitor atomwith the exception of Cys181 in (b).The wings I and II portions of theinhibitors in the butterfly-like anal-ogy for NNRTIs (Ding et al., 1995a)are indicated here and in sub-sequent Figures by Roman nu-merals I and II. The dotted line in (a)indicates the subdivision of atomsbetween wings I and II.


Figure 2. Stereoviews of superpositions of HIV-1 RT/TIBO complex structures drawn using program RIBBONS(Carson, 1987). (a) Structures of the NNIBP regions of the HIV-1 RT/8-Cl TIBO complex (backbone colored dark red)and the HIV-1 RT/9-Cl TIBO complex (backbone colored light purple) which have been superposed based on all 985Ca atoms. The 8-Cl TIBO inhibitor is colored green and 9-Cl TIBO is gold. The chlorine atoms are shown in magenta.Side-chains of amino acid residues with close contacts to the bound inhibitors are displayed for the HIV-1 RT/8-ClTIBO complex (cyan) and for the HIV-1 RT/9-Cl TIBO complex (light purple). (b) Structures of the NNIBP regionsin the wild-type HIV-1 RT/8-Cl TIBO complex (backbone in dark red) and in the Tyr181Cys HIV-1 RT/8-Cl TIBOcomplex (gray) which were superposed based on all Ca atoms. The 8-Cl TIBO inhibitor is colored green in the wild-typecomplex and is gold in the mutant complex. Side-chains of amino acid residues with close contacts to the boundinhibitors are displayed for the wild-type HIV-1 RT/8-Cl TIBO complex (cyan) and for the Tyr181Cys HIV-1 RT/8-ClTIBO complex (gray). The perimeter of the base of the cone depicts the possible loci of the sulfur atom (Sg) of thedisordered Cys181 side-chain in the mutant complex.

(Smerdon et al., 1994; Ren et al., 1995a), and HEPT(Ren et al., 1995a; Hopkins et al., 1996), Tyr181 andTyr188 have aromatic-aromatic ring interactionswith wing I of the bound inhibitors, albeit theextent of interaction varies considerably amongthese complexes. Substitutions of non-aromaticamino acids at positions 181 and 188 lead toresistance to most NNRTIs both in cell culture andin clinical trials (for example, see Nunberg et al.,

1991; Richman et al., 1991; Balzarini et al., 1993a;Byrnes et al., 1993; Richman, 1993). Site-directedmutagenesis indicated that HIV-1 RT variantslacking aromatic amino acids at positions 181 and188 display a significantly reduced sensitivity to anumber of NNRTIs, including TIBO derivatives(Sardana et al., 1992). It was surprising to find, inthe structures of HIV-1 RT/TIBO complexes, thatthe aromatic benzodiazepine ring of the bound


TIBO inhibitors occupies the wing II position,whereas the dimethylallyl group is located at thewing I position. Consequently, TIBO compoundshave relatively few close contacts with the aromaticrings of Tyr181 and Tyr188 (Figure 1(a) and (c)). Ascan be seen in Figure 2(a), the dimethylallyl groupof TIBO is nearly perpendicular to the indole planeof Trp229 and is flanked by the Tyr181 and Tyr188rings. The methyl groups of the dimethylallylmoiety of TIBO have a number of interactions withthe side-chain of Trp229 (Figure 1). So far, none ofthe mutations reported to cause NNRTI resistanceinvolve Trp229.

Binding of individual TIBO inhibitors towild-type HIV-1 RT

Both 8-Cl TIBO and 9-Cl TIBO bind to HIV-1 RTin a similar, butterfly-like, fashion (Figure 2), butthe specific interactions of the bound TIBOinhibitors with the protein and the precisegeometry of the NNIBP are different. Superpositionof all 985 Ca atoms in the p66/p51 heterodimers inthe respective structures showed a root-mean-squared (RMS) deviation of 0.86 A. On the basis ofthis superposition, the two bound TIBO inhibitorsalso superimpose very well, with the maximumseparation between equivalent pairs of atoms beingabout 0.4 A. The chlorine substituents, which are atdifferent positions on the two drugs, are separatedby 3.1 A in the superposed complexes (Figure 2(a)).Most of the interactions between the protein andwing I of the TIBO inhibitors are conserved in both

structures (Figure 1(a) and (c)). The conformationsof the amino acid residues surrounding wing I arequite similar in the two complexes. The majordifferences in the protein structure are around wingII of TIBO, in particular near the chlorine atoms. Inthe HIV-1 RT/8-Cl TIBO complex structure thechlorine atom is positioned near the tail of thebutterfly. The closest contacts with the 8-Cl atomare 3.8 A with the Cd2 atom of Phe227 and 4.1 Awith the Cb atom of Leu234. The chlorine atom inthe 9-Cl TIBO/HIV-1 RT complex is positionedtoward the outer edge of wing II and pointstowards the b13-b14 turn with nearest-neighboratoms Cd2 of Phe227 (at a distance of 4.0 A), Cb ofLeu234 (3.3 A), N and O of His235 (3.7 A and 3.4 A),and Oh of Tyr318 (3.2 A).

There is an expansion of the NNIBP near theb13-b14 turn (residues 235 to 238) in the 9-Cl TIBOcomplex relative to the 8-Cl complex. The Ca atomof Pro236 is displaced 1.5 A relative to the positionof the corresponding atom in the HIV-1 RT/8-ClTIBO structure (Figure 3). Hopkins et al. (1996) haveshown that the position of this structural segmentvaries considerably in different HIV-1 RT/NNRTIcomplex structures. If the 9-Cl TIBO inhibitor ismodeled into the NNIBP of the HIV-1 RT/8-ClTIBO structure, the distances between the 9-chlor-ine atom of the inhibitor and the N and O atomsof His235 would be 3.0 A and 2.8 A, respectively.These distances are slightly shorter than arenormally observed for non-bonded interactionsinvolving these atoms. The converse experimentdid not show any unfavorable short contacts for

Figure 3. Residue-by-residue displacement of equivalent Ca atoms between 8-Cl and 9-Cl TIBO complexes withwild-type HIV-1 RT (open bars) and between 8-Cl TIBO complexes with wild-type and Tyr181Cys mutant HIV-1 RT(filled bars). The Ca atoms of the amino acid residues of structural elements aE, aF, b6, and b10 (which form the coreof the palm subdomain) were used as the basis for superposition. The RMS deviation in this region between thewild-type and the Tyr181Cys mutant HIV-1 RT/8-Cl TIBO complexes is 0.3 A and the RMS deviation between thewild-type HIV-1 RT complexes with 8-Cl TIBO and 9-Cl TIBO is 0.5 A.


8-Cl TIBO in the binding pocket of the HIV-1RT/9-Cl TIBO complex. The solvent-accessiblevolumes of the NNIBP, calculated using VOIDOO(Kleywegt & Jones, 1994a) and a solvent proberadius of 1.4 A, were quite similar (9-Cl TIBO,133 A3 and 8-Cl TIBO, 129 A3). However, visualiz-ation of the cavity volumes using the program O(Jones et al., 1991) indicated that the NNIBP cavityin the 9-Cl TIBO complex is extended towardsSer105, Pro225, Phe227, His235, and Pro236. Thedistance between the b5-b6 connecting loop andthe b13-b14 hairpin, which are in the vicinity ofthe chlorine substituent of 9-Cl TIBO, is increasedrelative to the 8-Cl structure. This may be due torepositioning of the nearby NNIBP residues toavoid steric conflicts with the chlorine atom. AnHIV-1 RT variant containing the mutationPro236 : Leu was found to have increasedsensitivity to 9-Cl TIBO (Dueweke et al., 1993),suggesting that this substitution might favorablyenhance interactions between the protein and theinhibitor, which is consistent with the proximity ofthis residue to the TIBO compound. In the HIV-1RT/8-Cl TIBO structure, the interactions of Tyr318with the surrounding protein atoms are differentfrom those in the HIV-1 RT/9-Cl TIBO structure.In the 8-Cl TIBO complex, the Oh atom of Tyr318is 2.5 A from the carbonyl oxygen of His235(C-O . . . O angle of 104°) and may form ahydrogen bond. The distance between the Tyr318Oh atom and the chlorine atom of 8-Cl TIBO is5.6 A. In the HIV-1 RT/9-Cl TIBO structure, thedisplacement of the b13-b14 hairpin causes thecarbonyl oxygen atom of His235 to be 3.3 A awayfrom the Tyr318 Oh atom (C-O . . . O angle of 87°).The Tyr318 Oh atom is relatively close (3.2 A) tothe chlorine atom of 9-Cl TIBO. Superposition ofthe two structures also reveals that the b5-b6connecting loop in the region near Lys103 hasdifferent positions in the two complexes, with theseparation of Lys103 Ca atoms being 1.0 A (Figure3). The potential hydrogen bond interactionbetween N1 of TIBO and the carbonyl oxygenatom of Lys101 is apparently weaker in the 9-ClTIBO complex structure (with an N . . . O distanceof 3.0 A) than in the 8-Cl TIBO complex structure(with an N . . . O distance of 2.6 A). The b12 strandnear Phe227 and Leu228 in the 9-Cl TIBO complexis also moved outward from the pocket by about1.5 A relative to its position in the 8-Cl TIBOcomplex (Figures 2(a) and 3). These confor-mational differences are indicative of the flexibilityof the NNIBP, which can adopt shapes that arecomplementary to different bound inhibitors(Smith et al., 1995).

Binding of 8-Cl TIBO to Tyr181Cysmutant HIV-1 RT

The folding of the polypeptide chains in theTyr181Cys mutant HIV-1 RT/8-Cl TIBO complex issimilar to that of the wild-type HIV-1 RT/8-ClTIBO complex (Figure 2(b)). The RMS deviation is

0.54 A for all 985 Ca atoms. This high degree ofsimilarity indicates that the Tyr181 : Cys mutationcauses only minor changes in the global confor-mation of HIV-1 RT. This is consistent withbiochemical data (Spence et al., 1996) showingalmost identical polymerization activities for wild-type HIV-1 RT and the Tyr181Cys mutant. In themutant complex the position and the side-chainorientation of Cys181 in the p51 subunit is welldefined. The Sg atom of this residue points in asimilar direction (x1 = −154°) to the side-chain ofTyr181 of p51 (x1 = −175°) in the correspondingcomplex with wild-type HIV-1 RT. However, thereplacement of tyrosine at position 181 by cysteinein the p66 subunit has both structural andfunctional consequences. In the p66 subunit of themutant HIV-1 RT/8-Cl TIBO complex, the main-chain conformation of amino acid residue 181appears to be relatively unperturbed by theTyr181 : Cys mutation. However, there is anotable difference in the disposition of theside-chains of amino acid residue 181 in p66between the two complexes. In the 8-Cl TIBOcomplex with wild-type HIV-1 RT and those ofother wild-type HIV-1 RT/NNRTI complexes, theTyr181 side-chain is well defined and is positionedsuch that it points away from the NNIBP andtowards the polymerase active site. Substitution ofTyr181 by cysteine replaces the large aromaticside-chain with a thiol (SH) group. No clearelectron density was observed (in maps calculatedat various stages of structure determination andusing various weighting schemes; see Materials andMethods) that could be assigned reliably to the Sg

atom of Cys181 of p66. This is in contrast to thewell-ordered electron density observed for Cys181of p51. Calculations which assume that theside-chain of Cys181 in p66 adopts one of the threeideal rotamers of a cysteine side-chain predict thatthe Sg atom of Cys181 would have a short contactwith an atom of 8-Cl TIBO with a distance of 2.4 Afor one rotamer (x1 = 60°), but no short contacts forthe other two rotamers (x1 = 180° and x1 = 300°).However, a minor adjustment of x1 away from 60°would permit adoption of this rotamer withoutcausing steric interference with the bound inhibitor.None of the three preferred rotamers of the Cys181side-chain would lead to steric interference withother protein atoms. The electron density maps(even at lower contour levels) do not indicate apreference for any of the rotamers of the Cys181side-chain in p66, indicating that the thiol group ofthe Cys181 side-chain in p66 is flexible and mayadopt multiple positions in the complex with 8-ClTIBO. The Sg of Cys181 in p51 makes contacts withthe nearby Pro97 residue (with an Sg-Cd distance of3.4 A; and an Sg-Cg distance of 3.5 A); theseinteractions potentially account for the relativelystable position of Sg in Cys181 in p51.

The position of the inhibitor, 8-Cl TIBO, isessentially the same in the mutant and wild-typeHIV-1 RT complexes. For example, the hydrogenbond between the N1 atom of 8-Cl TIBO and the


Figure 4. Solvent-accessible sur-face areas for the inhibitor-bindingpocket in the 8-Cl TIBO complexwith wild-type HIV-1 RT (green)and with Tyr181Cys mutant HIV-1RT (red). The 8-Cl TIBO inhibitorsin the wild-type and mutant com-plexes are colored light blue andlight purple, respectively. Selectedside-chains of the Tyr181Cys mu-tant are shown. The orientation ofthe side-chain of Cys181 has beenassumed to be similar to that ofTyr181 in the p66 subunit in thewild-type complex (x1 = − 176°; seethe text). The Figure was generatedusing programs O (Jones et al., 1991)and VOIDOO (Kleywegt & Jones,1994a) with a solvent probe radiusof 1.4 A.

carbonyl O atom of Lys101 is almost unaffected bythe Tyr181 : Cys mutation (Figure 1). However,the torsion angles around the C5-N6-C12-C13 andN6-C12-C13-C14 bonds of the dimethylallyl groupof 8-Cl TIBO are different in the wild-type (111° and153°) and mutant HIV-1 RT (121° and 164°)structures. The side-chain of Tyr188 of p66 in themutant HIV-1 RT/8-Cl TIBO complex also hasslightly different torsion angles (x1 = −79°, x2 = 66°)from those of Tyr188 in the wild-type HIV-1RT/8-Cl TIBO complex (x1 = −77°, x2 = 55°). Smallchanges in the torsion angles of both thedimethylallyl group of 8-Cl TIBO and the side-chain of Tyr188 combine to yield increasedinteractions between the dimethylallyl group of theinhibitor and the Tyr188 side-chain in the mutantcomplex (Figure 1(a) and (b)). Additionally, thepositions of Val179 are somewhat different in themutant and wild-type HIV-1 RT/8-Cl TIBOcomplexes. Superposition of the two structuresindicates that the Ca atom of Val179 in the mutantHIV-1 RT/8-Cl TIBO complex is further away fromthe bound inhibitor by 0.8 A compared to itsposition in the wild-type HIV-1 RT/8-Cl TIBOcomplex (Figures 1(a), (b), 2(b), and 3). As aconsequence, the interaction between Val179 andthe bound TIBO inhibitor is weaker in theTyr181Cys mutant HIV-1 RT/8-Cl TIBO structure(Figure 1).

Pocket volume calculations (Figure 4) show thatthe NNIBP in the Tyr181Cys mutant HIV-1RT/8-Cl TIBO complex is substantially larger thanthat in the wild-type HIV-1 RT/8-Cl TIBO complex(160 A3 versus 129 A3), assuming the disorderedCys181 side-chain of p66 in the mutant structurehas the same x1 side-chain torsion angle as that ofTyr181 of p66 in the wild-type HIV-1 RT/8-Cl TIBOstructure (x1 = −176°). The expansion of the NNIBPin the mutant HIV-1 RT/8-Cl TIBO complex occursin the region around amino acid residues 179 and181 (Figures 2(b) and 4).

Possible mechanism of drug resistancecaused by the Tyr181 : Cys mutation

Residue Tyr181 forms an integral part of theNNIBP in complexes of wild-type HIV-1 RT andNNRTIs; the extent of its interaction with theinhibitor depends on which inhibitor is bound.Since the Tyr181Cys HIV-1 RT variant is quiteresistant to many NNRTIs, it is reasonable to expectsignificant differences in the shape, volume, andchemical character of the NNIBP in the Tyr181Cysmutant HIV-1 RT. The disorder of the side-chain ofCys181 in p66 in the Tyr181Cys HIV-1 RT/8-ClTIBO structure suggests that the hydrophobicinteractions of the side-chain of amino acid residue181 with the bound NNRTI and side-chains ofsurrounding amino acid residues are lost when thetyrosine present at this position in wild-type HIV-1RT is replaced by cysteine. This significant changein the NNIBP may be responsible for reducedaffinity of Tyr181Cys HIV-1 RT for 8-Cl TIBO andother NNRTIs. On the other hand, in the Tyr181CysHIV-1 RT complex with 8-Cl TIBO, Tyr188 showsrelatively more interactions with the boundinhibitor compared to the corresponding wild-typecomplex. The loss of interactions of 8-Cl TIBO withamino acid residue 181 in the Tyr181Cys mutantcould be compensated to some extent by increasedinteractions with the Tyr188 side-chain. This mightpartly explain why 8-Cl TIBO is still a relativelypotent inhibitor of Tyr181Cys HIV-1 RT.

Numerous studies have indicated that theTyr181 : Cys HIV-1 RT mutant, in general, is moreresistant to nevirapine and a-APA derivatives thanto TIBO derivatives (Byrnes et al., 1993; Pauwelset al., 1993, 1994; Richman et al., 1994). Structures ofHIV-1 RT complexes with nevirapine (Smerdonet al., 1994; Ren et al., 1995a) and a-APA (Ding et al.,1995b; Ren et al., 1995a) have shown that thesecompounds have more extensive interactions withTyr181 than do TIBO compounds (Ding et al.,


1995a; Ren et al., 1995b). This could explain whythese and numerous other NNRTI compounds arerelatively inefficient in inhibiting HIV-1 RTcarrying the Tyr181 : Cys mutation. It seems likelythat the Tyr181 : Cys mutation has more profoundeffects on the binding of NNRTIs which interactextensively with Tyr181 in the wild-type HIV-1RT/NNRTI complex. However, considering thatTyr181 is also positioned at the putative entranceto the NNIBP, as seen in HIV-1 RT structures inthe absence of an NNRTI, we cannot rule outthe possibility that the Tyr181 : Cys mutationcould also affect entry into or egress from thepocket.

The flexibility of the torsion angles orienting thedimethylallyl moiety of 8-Cl TIBO permits theinhibitor to have more interactions with Tyr188 inthe Tyr181Cys mutant complex; this confor-mational flexibility could also explain why 8-ClTIBO is still relatively potent in inhibiting thismutant. One suggestion for inhibitor design thatemerges from these considerations is that flexibilityof the wing I portion of NNRTIs may be useful forincreasing potency against several key NNRTI-re-sistant HIV-1 RT variants, including the Tyr181Cysmutant.

Displacement of the primer grip by NNRTIscould explain the kinetic data

Comparison of the HIV-1 RT/TIBO complexstructures with other NNRTI-bound HIV-1 RTstructures (Smerdon et al., 1994; Ding et al., 1995b),unliganded HIV-1 RT (Rodgers et al., 1995; Hsiouet al., 1996), and the HIV-1 RT/DNA/Fab complex

structure (Jacobo-Molina et al., 1993) indicates thatthe b6-b10-b9 sheet is a relatively stable structuralelement whose gross conformation and positionchanges little upon NNRTI binding. However, theb12-b13-b14 sheet has relatively greater flexibilityand undergoes large conformational and positionalchanges (>4 A movements) upon inhibitor binding.

Various mechanisms of inhibition by NNRTIshave been proposed on the basis of structural andbiochemical data (Kohlstaedt et al., 1992; Tantilloet al., 1994; Ding et al., 1995b; Esnouf et al., 1995;Rittinger et al., 1995; Rodgers et al., 1995; Smithet al., 1995; Spence et al., 1995; Hsiou et al., 1996. Fora review, see Sarafianos et al., 1996). Detailedbiochemical and kinetic studies of HIV-1 RTinhibition by TIBO and other NNRTIs haveindicated that NNRTIs do not prevent dNTPbinding but instead reduce the rate of the chemicalstep of DNA polymerization (Rittinger et al., 1995;Spence et al., 1995). Comparison of the HIV-1RT/TIBO structures with the HIV-1 RT/DNA/Fabstructure (Jacobo-Molina et al., 1993) reveals a largeshift in the positions of the b12-b13-b14 sheet(Figure 5). In the unliganded HIV-1 RT structures(Rodgers et al., 1995; Hsiou et al., 1996) this b-sheetoccupies a position similar to that found in theHIV-1 RT/DNA/Fab structure. Hence, the dis-placement of the b12-b13-b14 sheet relative to therest of the palm subdomain seems to be a consistentconsequence of NNRTI binding. Since the b12-b13-b14 sheet (which contains the primer grip) interactswith the 3'-terminal phosphate of the primer strand,movement of this sheet is likely to affect theposition and/or mobility of the primer terminusand consequently the position of the 3'-OH group

Figure 5. A stereoview of the superposition (based on the Ca atoms of the b6-b10-b9 sheet) of the HIV-1 RT/DNA/Fabcomplex structure (in gray) (Jacobo-Molina et al., 1993) on the HIV-1 RT/9-Cl TIBO complex structure (in cyan) in theregions near the NNIBP and the polymerase active site showing the disposition of the b12-b13-b14 sheet containingthe primer grip. Bound 9-Cl TIBO in the HIV-1 RT/9-Cl TIBO complex is shown in gold and the two 3'-terminalnucleotides 17 and 18 of the primer strand in the HIV-1 RT/DNA/Fab complex are shown with a yellow ball-and-stickmodel. The broken line represents interactions between the primer grip and the primer terminal phosphate in the HIV-1RT/DNA/Fab complex and the arrow indicates the movement of the primer grip that accompanies NNRTI binding.


Table 1. Statistics of the structure determination of TIBO complexes with wild-type and Tyr181Cys mutant HIV-1 RTA. Summary of diffraction data used HIV-1 RT/ Cys181 HIV-1 RT

HIV-1 RT/8-Cl TIBO HIV-1 RT/9-Cl TIBO 8-I TIBO 8-Cl TIBODatasets: Frozen Cooled Frozen Cooled Frozen Frozen

Data collection system F1a A1b & F1 F1 F1 R-axis IIc F1Temperature at crystal (°C) −165 −10 −165 −10 −165 −165Number of crystals used 1 7 1 13 1 1Number of images 109 83 69 78 94 93Reflections measured 105,552 91,062 56,695 56,296 18,729 61,059Unique reflections 25,846 28,355 26,209 24,057 8,422 23,319Rmerge

d 0.074 0.076 0.064 0.068 0.082 0.105Intensity cutoff Ie1s(I) Ie1s(I) Ie1s(I) Ie1s(I) Ie1s(I) I > 0Completeness (%) 84 87 84 76 63 90Resolution range (A) 40.0–3.0 40.0–3.0 40.0–3.0 40.0–3.0 40.0–4.0 30.0–3.2Space group C2 C2 C2 C2 C2 C2Unit cell parametersa (A) 225.4 227.2 226.0 227.8 225.4 225.7b (A) 69.4 70.2 69.3 70.1 69.4 69.2c (A) 104.4 105.7 104.1 105.8 104.4 104.9b (°) 106.4 105.6 107.0 105.7 106.4 106.6B. Refinement statisticsNumber of reflections 24,961 21,653 20,766Resolution limit (A) 10.0–3.0 10.0–3.0 15.0–3.2Sigma cutoff Fobse2s(Fobs) Fobse2s(Fobs) Fobse1s(Fobs)Data completeness (%) 80 72 85Number of atoms refined 7833 7845 7828R-factore 0.249 0.259 0.274Free R-factor 0.356 Not calculated 0.360Number of reflections used 1288 N/A 1059

for free R calculationRMS deviation for bond lengths (A) 0.014 0.015 0.011RMS deviation for bond angles (deg.) 1.9 2.2 1.8

a F1 beam line at CHESS, l = 0.908 A.b A1 beam line at the Cornell High Energy Synchrotron Source (CHESS), l = 0.903 A.c R-axis II mounted on a Rigaku RU-200 rotating anode X-ray generator, l = 1.5418 A.d Rmerge = S=Iobs − �I�=/S�I�.e R = S>Fobs= − =Fcalc>/S=Fobs=.

which is the site of attachment of the incomingnucleotide (Tantillo et al., 1994; Ding et al., 1995b;Rodgers et al., 1995; Smith et al., 1995; Hsiou et al.,1996). It is reasonable to expect that modification ofthe position or mobility of the 3'-OH by NNRTIbinding would interfere with the chemical step ofDNA polymerization, which is consistent withbiochemical observations (Rittinger et al., 1995;Spence et al., 1995). Alternatively, comparison ofthe structures of the HIV-1 RT/DNA/Fab andHIV-1 RT/NNRTI complexes reveals significantconformational differences of the YMDD motif(part of the b6-b10-b9 sheet) at the catalytic site (asreflected by displacements of Ca positions of 2 A ormore; (Ding et al., 1995b). Even though thesepositional changes are smaller in magnitude thanthose of the primer grip, they involve thecatalytically essential Asp185 and Asp186 atthe active site, and thus could potentially impair thechemical step of polymerization. A serious limi-tation of the available information is that none ofthe NNRTI-bound HIV-1 RT structures containeither template-primer of dNTP. The variousmechanisms of NNRTI inhibition that have beenproposed need not be mutually exclusive. Furtherstructural studies of HIV-1 RT complexed with bothNNRTIs and nucleic acid substrates should help toresolve some of the remaining questions.

Materials and Methods

Crystallization and diffraction data collection

Samples of HIV-1 RT were prepared and purified usingmethods described (Arnold et al., 1992; Jacobo-Molinaet al., 1993; Clark et al., 1995). Samples of the Tyr181Cysmutant HIV-1 RT were prepared and purified in amanner similar to wild-type HIV-1 RT, except that allbuffers contained 5 mM DTT. However, the finalTyr181Cys HIV-1 RT preparations were aliquoted andstored at −70°C at a concentration of 16 mg/ml (due toextensive losses of the mutant enzyme during thepurification and concentration procedures) rather than at40 mg/ml as for wild-type HIV-1 RT preparations.Wild-type HIV-1 RT complexes with 8-Cl TIBO and 9-ClTIBO were crystallized using conditions (Clark et al.,1995) modified from those described earlier byKohlstaedt et al. (1992) for crystallization of HIV-1 RTcomplexed with nevirapine. Initial attempts at co-crystal-lization of HIV-1 RT with 8-Cl TIBO yielded smallcrystals. Larger crystals of wild-type HIV-1 RT/8-ClTIBO were grown by streak seeding (Stura & Wilson,1990) of three-day-old equilibrated hanging drops, usingmicro seeds obtained from crushing small HIV-1 RT/8-ClTIBO crystals. Crystals of wild-type HIV-1 RT/9-Cl TIBOwere grown without seeding. The typical size for thesecrystals was about 1.0 mm × 0.5 mm × 0.2 mm. Crystalsof the Tyr181Cys mutant HIV-1 RT/8-Cl TIBO complexgrew much more slowly than wild-type HIV-1 RT/TIBOcrystals; the maximum size of the mutant complex


crystals was only about 0.2 mm × 0.2 mm × 0.05 mm,sufficient for data collection using flash freezingtechniques (Teng, 1990; Rodgers, 1994). All crystals weregrown at 4°C.

Crystals of wild-type HIV-1 RT with 8-Cl TIBO and9-Cl TIBO and the Tyr181Cys mutant HIV-1 RT complexwith 8-Cl TIBO all crystallized in space group C2 withsimilar unit cell parameters (Table 1), which were withina few percent of the values for the structures of the HIV-1RT/nevirapine (Kohlstaedt et al., 1992) and HIV-1RT/a-APA (Ding et al., 1995b) complexes. Shrinkage ofthe unit cell dimensions by approximately 1% wasconsistently observed when comparing the unit cells infrozen (−165°C) versus cooled (−10°C) crystals. Changesin structure between the frozen and cooled crystal formsdo not correspond to a uniform shrinkage, but rather tosubtle rearrangements of the structure. Thus, even theserelatively minor differences in cell parameters permittedeffective application of the multiple crystal formaveraging procedure for phase improvement (see below).

In the crystal structure determination of the HIV-1RT/TIBO complexes, six different X-ray diffractiondatasets were used (Table 1). To freeze crystals, thewild-type HIV-1 RT/TIBO crystals were soaked in 20 mlsolutions of synthetic mother liquor (50 mM bis-Tris-propane (pH 6.8), 100 mM (NH4)2SO4, and 12% (w/v)polyethylene glycol 8000) containing 1 ml of TIBO stocksolution (10 mM TIBO in dimethyl sulfoxide and 5%(v/v) b-octylglucoside) and 17%, 23%, and 30% (v/v)glycerol, respectively, in time steps of approximately 30minutes each. The soaked crystals were directly frozen inthe liquid N2-cooled gaseous stream (−165°C). However,this stepwise soaking protocol did not work for crystalsof the Tyr181Cys mutant HIV-1 RT/8-Cl TIBO complex.These crystals were found to be very sensitive to thecryoprotectant solutions described above. Crystals of theTyr181Cys HIV-1 RT/8-Cl TIBO complex were success-fully frozen by quickly dipping a crystal into 20 ml ofsynthetic mother liquor (see above) containing 1 ml ofTIBO stock solution and 30% (v/v) glycerol and thentransferring directly to the liquid N2-cooled gaseousstream. All diffraction datasets were processed withDENZO (Minor, 1993; Otwinowski, 1993) and scaled witheither SCALEPACK (Otwinowski, 1993) or SCALA(Evans, 1993) of the CCP4 suite (Collaborative Compu-tational Project, 1994). Datasets from the cooled wild-typeHIV-1 RT/8-Cl TIBO complex crystals and from a frozenwild-type HIV-1 RT/9-Cl TIBO complex crystal (Table 1)were relatively more complete, and the structures ofwild-type HIV-1 RT/TIBO complexes discussed in thispaper have been refined using these datasets.

Structure determination and refinement

The structure of the wild-type HIV-1 RT/8-Cl TIBOcomplex was determined using the molecular replace-ment technique as implemented in X-PLOR 3.1 (Brunger,1993) and coordinates of the HIV-1 RT/a-APA complexas a starting search model against the cooled diffractiondata. In order to improve the phase quality and reducemodel bias, at the earlier stages of structure determi-nation the electron density maps from the HIV-1 RT/8-ClTIBO, HIV-1 RT/a-APA, and HIV-1 RT/BHAP com-plexes were averaged at 3.0 A resolution (Ding et al.,1995b) using the multiple crystal form averaging routinesin RAVE (Kleywegt & Jones, 1994b) combined withprograms from the CCP4 package (Collaborative Compu-tational Project, 1994). Model building based on theaveraged maps using the program O (Jones et al., 1991)

and positional refinement were carried out. The proteinmodel from the cooled crystal structure was placed in thefrozen crystal unit cell by applying translations of 2.8 Aand 1.4 A along the x and z directions, respectively, thatwere determined from a translation search in Pattersonspace. In the later stages of structure determination,electron density maps calculated with the cooled (−10°C)and frozen crystal (−165°C) datasets were averaged usingmultiple crystal form averaging.

In order to further reduce model bias, omit maps werecalculated and used in the multiple electron density mapaveraging. In omit map calculations, the asymmetric unitwas divided into ten sections with each sectioncontaining an approximately equal number of atoms.The omit map for an individual section was calculatedusing the phase information from the partial modelconsisting of the remaining nine sections. In an attemptto reduce the ‘‘memory’’ of the omitted atoms, the partialmodel was refined for three cycles using PROLSQ(Hendrickson, 1985) and the phases computed from theresulting model were used to calculate the omit map forthe omitted section. Then the ten omit maps calculatedfor all ten individual sections were merged together tocomplete one asymmetric unit which was then extendedto cover the entire molecule.

In general, the averaged electron density maps were ofsuperior quality (Figure 6(a) and showed improvedelectron density in many regions of the protein. In thefinal stages of model building and refinement, theelectron density map averaging was performed usingprogram Dmmulti (Cowtan, 1994). The model building ofthe protein, prior to inclusion of the inhibitor, was guidedusing both the individual and averaged omit maps.

Both the difference Fourier map (Fobs − Fcalc) (Fig-ure 6(b)) and the averaged electron density map (Fig-ure 6(a)) computed for the HIV-1 RT/8-Cl TIBO complexclearly indicated the position of the bound inhibitor. Bothelectron density maps were calculated using phasesderived from only the protein model, before anystructural information about the bound inhibitor wasincluded either in structure refinements or in the mapcalculations. Initial coordinates for the TIBO model wereobtained from small-molecule crystal structures (Liawet al., 1991). The small-molecule crystal structure of 9-ClTIBO showed two possible conformations (designated asA and B) that vary in the configuration of theseven-membered diazepine ring. In both conformationsthe benzimidazole ring is nearly planar. Conformation Ahas its seven-membered ring in the TS configuration(Boessenkool & Boeyens, 1980). The dimethylallyl (ormethylbutenyl) side group is above the mean plane withC12 (C15 in the small molecule crystal structurenumbering scheme) in an axial position. However, in theB conformation the seven-membered ring has the TC/BSconfiguration with C12 in an equatorial position (Liawet al., 1991). Both of these conformations could beaccommodated in the electron density with minormodifications of a few torsion angles. A major differencein fitting the two possible conformations of TIBO in theNNIBP corresponds to the orientation of the aromaticring system. In order to determine the position of thechlorine atom of TIBO in the NNIBP, a difference Fouriermap (Figure 6(b)) was calculated between the datasets forthe HIV-1 RT/8-I TIBO (R89076) and the HIV-1 RT/8-ClTIBO complexes at 4.5 A resolution. This map unambigu-ously identified the correct position for the halogen atomand consequently resolved the potential ambiguity inthe overall orientation of TIBO in the NNIBP. Theinformation from the difference Fourier map computed


between the iodinated and chlorinated TIBO derivatives,together with the asymmetric shape of the electrondensity for the inhibitor, permitted us to select the Bconformation of TIBO as a starting model. The fit of theB conformation of TIBO into the omit electron densitywas optimized by a small change in the torsion anglearound the N6–C12 bond that permitted the methylcarbon atoms of the dimethylallyl group to fit the electrondensity well (Figure 6(b)).

Structure refinement of the wild-type HIV-1 RT/8-ClTIBO complex yielded an R value of 24.9% using 24,961reflections and a free R value (Brunger, 1992) of 35.6%using 1288 reflections with Fobse2s(Fobs) in the resolutionrange between 10.0 A and 3.0 A for the cooled dataset(Table 1). Individual isotropic thermal parameters wererefined (Bar = 39 A2). The final model contains 7833non-hydrogen protein atoms and 21 TIBO atoms. Due toweak electron density for the side-chains, amino acid

Figure 6. Electron density mapsused in the structure determinationof wild-type HIV-1 RT complexeswith 8-Cl and 9-Cl TIBO (drawnusing program O (Jones et al., 1991).(a) Stereoview of a portion of anaveraged (2wFobs − Fcalc) omit map(calculated using SIGMAA weights(Read, 1986); contour level 1.2s,magenta) at 3.2 A resolution in theNNIBP region of the HIV-1 RT/8-ClTIBO complex structure. The mapwas averaged between diffractiondatasets measured from frozen andcooled crystals of the HIV-1 RT/8-Cl TIBO complex using programsRAVE (Kleywegt & Jones, 1994) andthe CCP4 package (CollaborativeComputational Project, 1994). (b) A3.0 A resolution difference Fourier(Fobs − Fcalc) electron density map(contour level 2s, cyan) for 8-ClTIBO in the HIV-1 RT/8-Cl TIBOcomplex calculated before the in-hibitor was included in structurerefinement and map calculation,and the difference Fourier map(contour level 4s, magenta) com-puted between the HIV-1 RT/8-ITIBO complex and the HIV-1 RT/8-Cl TIBO complex at 4.5 A resol-ution, showing the location of thehalogen atom. (c) A 3.0 A resolutiondifference Fourier (Fobs − Fcalc) elec-tron density map (contour level 2s,cyan) for 9-Cl TIBO in the HIV-1RT/9-Cl TIBO complex. Again, themap was calculated before anyinformation about the bound 9-ClTIBO inhibitor was incorporatedin structure refinement and mapcalculation.


Figure 7. Electron density (calcu-lated using 2Fobs − Fcalc coefficients;contour level 1.2s, magenta) forportions of the Tyr181Cys HIV-1 RTmutant complex with HIV-1 RT at3.2 A resolution. (a) In the p66subunit, there is no ordered electrondensity for the side-chain sulfuratom of Cys181 (residue highlightedin green with an arbitrary positionof Sg shown), indicating disorder ofthe thiol group. In contrast, the 8-ClTIBO inhibitor has well-orderedelectron density. (b) In the p51subunit, the electron density for theCys181 side-chain is well defined.

residues 36, 66, 71, 72, 134 to 139, 218 to 223, 249, 281 to293, 297 to 302, 311, 312, 323, 356 to 358, 366, 451, and 556to 558 in p66 and 13, 121, 197, 199, 220, 222 to 230, 270,278, 281, 357, 358, and 361 to 363 in p51 were modeledas alanine residues. The portions corresponding toresidues 65 to 72 and 134 to 139 in p66 and 222 to 230 inp51 have poorly defined electron density that did notpermit reliable modeling in those regions.

The crystal structure of the wild-type HIV-1 RT/9-ClTIBO complex was determined starting with the proteinmodel of the HIV-1 RT/8-Cl TIBO structure. Theaveraged omit maps corresponding to the HIV-1 RT/9-ClTIBO structures in the frozen and cooled crystals werecalculated in a similar fashion to that described for theHIV-1 RT/8-Cl TIBO complex. The electron density forthe inhibitor (Figure 6(c)), obtained before including theinhibitor in the structure refinement and map calcu-lations, was adequate to define the position andorientation of the TIBO inhibitor in the binding pocket.The statistics of the structure refinement and the finalmodel are listed in Table 1. In the final model, amino acidresidues 36, 66, 71, 72, 134, 135, 139, 218 to 223, 249, 281to 293, 297 to 302, 311, 312, 323, 356 to 358, 366, 451, and556 to 558 in p66 and 13, 121, 197, 199, 220, 222 to 230,270, 278, 281, 357, 358, and 361 to 363 in p51 weremodeled as alanine residues.

The crystal structure of the Tyr181Cys mutant HIV-1RT/8-Cl TIBO complex was determined starting with themodel of the wild-type HIV-1 RT/8-Cl TIBO complexwith the exclusion of the inhibitor and the side-chains ofthe amino acid residue 181 in both p66 and p51 subunits.The difference Fourier maps and the omit maps

calculated using the scheme described earlier showedclear electron density for the bound inhibitor thatallowed us to define the position and orientation of 8-ClTIBO without any ambiguity (Figure 7(a)). The electrondensity was well defined for the side-chain of Cys181 inthe p51 subunit (Figure 7(b)). However, no clear electrondensity was observed for the side-chain of Cys181 in thep66 subunit (Figure 7(a)), suggesting that the side-chainof Cys181 in p66 is disordered. The statistics of thestructure refinement and the final model are listed inTable 1. In the final model, the following amino acidresidues were modeled as alanines: 36, 66, 71, 72, 134 to139, 218 to 223, 249, 281 to 293, 297 to 302, 311, 312, 323,356 to 358, 366, 451, and 556 to 558 in p66 and 13, 121,197, 199, 220, 222 to 230, 270, 278, 281, 357, 358, and 361to 363 in p51.

Comments on the conformation of TIBO boundto HIV-1 RT

A structure determination of the HIV-1 RT/9-Cl TIBOcomplex at 2.6 A resolution in a different crystal form(P212121) has been reported (Ren et al., 1995b). Thoughthis paper confirmed that 9-Cl TIBO bound to HIV-1 RTin the same overall mode as we had described for theHIV-1 RT/8-Cl TIBO complex (Ding et al., 1995a), itproposed that the conformation of the bound 9-Cl TIBOdiffered slightly from that which we reported for 8-ClTIBO. In the structure of the HIV-1 RT/9-Cl TIBOcomplex reported by Ren et al. (1995b), the conformationof the seven-membered ring and the chirality of thesp3-hybridized N6 atom appear to be different from what


was reported for the small-molecule crystal structures(Liaw et al., 1991) and from what we have found forstructures of 8-Cl TIBO (Ding et al., 1995a) and 9-Cl TIBOin complexes with HIV-1 RT (this paper). In the structurereported by Ren et al. (1995b) the seven-membereddiazepine ring is nearly coplanar with the benzimidazolering and the methyl group at the C5 position and thedimethylallyl group at the N6 position lie on either sideof the mean plane of the three fused ring system.

The coordinates for the HIV-1 RT/9-Cl TIBO complexstructure reported by Ren et al. (1995b) are unavailable sowe are unable to compare their structure directly withour structures. We built a model of 9-Cl TIBO withgeometry similar to that shown in Figures 3 and 4 of Renet al. (1995b) (hereafter referred to as conformation C todistinguish from the A and B conformations (Liaw et al.,1991)). Omit difference Fourier maps (using either(Fobs − Fcalc) or (2Fobs − Fcalc) coefficients) were calculatedprior to inclusion of the TIBO inhibitors in both thewild-type HIV-1 RT/ TIBO complexes and the fit of eachof the three conformations (A, B, and C) was evaluated.The difference Fourier maps showed electron density thatcould be fit most consistently with the B conformation ofTIBO. Experiments including the modeled C confor-mation of TIBO in electron density map calculations,followed by structure refinement, showed poor densityfor the dimethylallyl side group, suggesting that the Cconformation of TIBO is not consistent with ourexperimental data. In our structure determination of theHIV-1 RT/8-Cl TIBO complex, the electron density forthe methyl group attached at the C5 position wasreasonable in the averaged omit maps (Figure 6a).Although the electron density for this methyl group wasnot clear in the HIV-1 RT/9-Cl TIBO complex (Figure 6c),all electron density maps showed clear density for thelarge dimethylallyl side group in both the HIV-1 RT/8-ClTIBO and HIV-1 RT/9-Cl TIBO complexes. Sinceconformational changes of the seven-membered ring arecoupled with the position of the dimethylallyl group ofTIBO, the conformation of the dimethylallyl group ofTIBO in the C conformation is not consistent with theelectron density maps.

In the C conformation of TIBO, the dimethylallyl groupand the methyl group, attached to N6 and C5,respectively, are in trans configuration and the dihedralangle of C11-C5-N6-C12 is approximately 140°. Thecorresponding torsion angles in the A and B forms of TIBOare 71° and −54°, respectively. In order to assess theobserved distribution of this dihedral angle for availablebenzodiazepine derivative structures, we searched theCambridge Crystallographic Data Bank for small-molecule structures containing a benzodiazepine group(fused benzene and seven-membered rings). Out of 23closely related benzodiazepine structures that have achiral center at position 5 and a tetrahedral carbon atposition 6, none has the two chemical groups/atomsattached at positions 5 and 6 in the trans configuration. Thedihedral angle for C11-C5-N6-C12 or its equivalent in anyof these structures is within 275° and the two chemicalgroups attached at positions 5 and 6 tend to lie on the sameside of the mean plane of the benzodiazepine ring.

Taking all of these considerations into account, webelieve that the orientation and conformation of TIBOinhibitors presented provide the best explanation of ourexperimental observations. With moderate resolutiondiffraction data, we have taken precautions to improvethe quality of electron density and reduce model biasthrough multiple crystal form map averaging and omitmaps. However, we do not rule out the possibility that

the TIBO inhibitor might adopt other conformations inthe HIV-1 RT/TIBO complex crystallized in differentcrystal forms such as that seen by Ren et al. (1995b).

Data deposition

The full coordinates of the HIV-1 RT/8-Cl TIBO(R86183) complex (PDB entry 1HNV), the HIV-1 RT/9-ClTIBO (R82913) complex (PDB entry 1TVR), and theTyr181Cys mutant HIV-1 RT/8-Cl TIBO (R86183)complex (PDB entry 1UWB) have been deposited withthe Brookhaven Protein Data Bank for immediate release.

AcknowledgementsWe thank our collaborators and the other members of

our laboratories, including Gail Ferstandig Arnold, MikeKukla, Karen Lentz, Stefan Sarafianos, Chris Tantillo, andWanyi Zhang, for helpful discussions and assistance. Wealso thank members of Stephen Harrison’s group(Harvard) for advice on multiple crystal form averaging;Dan Thiele, Rick Walter, and the staff at the Cornell HighEnergy Synchrotron Source and Richard Leidich andGreg Listner at our local X-ray diffraction facility forinvaluable help in data collection; Robert Hooper at theKeck Structural Biology Computing Center at CABM forhelp in computing; Helen Berman and Tom Emge forhelp with the small molecule database search; and StafVan Reet and Mike Zelesko for support. The work inEdward Arnold’s laboratory is supported by JanssenResearch Foundation and NIH grants (AI 27690 and AI36144). Research in Stephen Hughes’ laboratory issponsored by the National Cancer Institute, DHHS,under contract with ABL and by NIGMS.

ReferencesArnold, E., Jacobo-Molina, A., Nanni, R. G., Williams,

R. L., Lu, X., Ding, J., Clark, A. D., Jr, Zhang, A.,Ferris, A. L., Clark, P., Hizi, A. & Hughes, S. H.(1992). Structure of HIV-1 reverse transcriptase/DNA complex at 7 A resolution showing active sitelocations. Nature, 357, 85–89.

Arnold, E., Das, K., Ding, J., Yadav, P. N. S., Hsiou, Y.,Boyer, P. L. & Hughes, S. H. (1996). Targeting HIVreverse transcriptase for anti-AIDS drug design.Drug Design Discov. 13, 29–47.

Balzarini, J., Karlsson, A., Perez-Perez, M.-J., Vrang, L.,Walbers, J., Zhang, H., Oberg, B., Vandamme, A.-M.,Camarasa, M.-J. & De Clercq, E. (1993a). HIV-1-specific reverse transcriptase inhibitors show differ-ential activity against mutant strains containingdifferent amino acid substitutions in the reversetranscriptase. Virology, 192, 246–253.

Balzarini, J., Karlsson, A., Perez-Perez, M. J., Camarasa,M. J., Tarpley, W. G. & De Clercq, E. (1993b).Treatment of human immunodeficiency virus type 1(HIV-1)-infected cells with combinations of HIV-1-specific inhibitors results in a different resistancepattern than does treatment with single-drugtherapy. J. Virol. 67, 5353–5359.

Boessenkool, I. K. & Boeyens, J. A. (1980). Identificationof the configurational type of seven-membered rings.J. Cryst. Mol. Struct. 10, 11–18.

Boyer, P. L., Ding, J., Arnold, E. & Hughes, S. H. (1994).Drug resistance of human immunodeficiency virus


type 1 reverse transcriptase: subunit specificity ofmutations that confer resistance to nonnucleoside in-hibitors. Antimicrob. Agents Chemother. 38, 1909–1914.

Brunger, A. T. (1992). The free R value: a novel statisticalquantity for assessing the accuracy of crystalstructures. Nature, 355, 472–474.

Brunger, A. T. (1993). X-PLOR Manual version 3.1: asystem for X-ray crystallography and NMR. YaleUniversity Press, New Haven and London.

Byrnes, V. W., Sardana, V. V., Schleif, W. A., Condra,J. H., Waterbury, J. A., Wolfgang, J. A., Long, W. J.,Schneider, C. L., Schlabach, A. J., Wolanski, B. S.,Graham, D. A., Gotlib, L., Rhodes, A., Titus, D. L.,Roth, E., Blahy, O. M., Quintero, J. C., Staszewski, S.& Emini, E. A. (1993). Comprehensive mutantenzyme and viral variant assessment of humanimmunodeficiency virus type 1 reverse transcriptaseresistance to nonnucleoside inhibitors. Antimicrob.Agents Chemother. 37, 1576–1579.

Carson, M. (1987). Ribbon models of macromolecules.J. Mol. Graph. 5, 103–106.

Clark, A. D., Jr, Jacobo-Molina, A., Clark, P., Hughes, S.H. & Arnold, E. (1995). Crystallization of humanimmunodeficiency virus type 1 reverse transcriptasewith and without nucleic acid substrates, inhibitors,and an antibody Fab fragment. Methods Enzymol. 262,171–185.

Collaborative Computational Project No. 4 (1994). TheCCP4 suite: programs for protein crystallography.Acta Crystallog. sect. D, 50, 760–763.

Cowtan, K. (1994). Dmmulti. Joint CCP4 and ESF-EACBMNewsletter on Protein Crystallography, 31, 34–38.

De Clercq, E. (1994). HIV resistance to reverse trans-criptase inhibitors. Biochem. Pharmacol. 47, 155–169.

De Clercq, E. (1995a). Antiviral therapy of humanimmunodeficiency virus infections. Clin. Microbiol.Rev. 8, 200–239.

De Clercq, E. (1995b). Toward improved anti-HIVchemotherapy: therapeutic strategies for inter-vention with HIV infections. J. Med. Chem. 38,2491–2517.

Ding, J., Das, K., Moereels, H., Koymans, L., Andries, K.,Janssen, P. A. J., Hughes, S. H. & Arnold, E. (1995a).Structure of HIV-1 RT/TIBO R 86183 revealssimilarity in the binding of diverse non-nucleosideinhibitors. Nature Struct. Biol. 2, 407–415.

Ding, J., Das, K., Tantillo, C., Zhang, W., Clark, A. D., Jr,Jessen, S., Lu, X., Hsiou, Y., Jacobo-Molina, A.,Andries, K., Pauwels, R., Moereels, H., Koymans, L.,Janssen, P. A. J., Smith, R., Koepke, M. K., Michejda,C., Hughes, S. H. & Arnold, E. (1995b). Structure ofHIV-1 reverse transcriptase in a complex with thenon-nucleoside inhibitor a-APA R 95845 at 2.8 Aresolution. Structure, 3, 365–379.

Dueweke, T. J., Pushkarskaya, T., Poppe, S. M., Swaney,S. M., Zhao, J. Q., Chen, I. S. Y., Stevenson, M. &Tarpley, W. G. (1993). A mutation in reversetranscriptase of bis(heteroaryl)piperazine-resistanthuman immunodeficiency virus type 1 that confersincreased sensitivity to other non-nucleoside inhibi-tors. Proc. Natl Acad. Sci. USA, 90, 4713–4717.

Esnouf, R., Ren, J., Ross, R., Jones, Y., Stammers, D. &Stuart, D. (1995). Mechanism of inhibition of HIV-1reverse transcriptase by non-nucleoside inhibitors.Nature Struct. Biol. 2, 303–308.

Evans, P. R. (1993). Data reduction. In Data Collection andProcessing (Sawyer, L., Issacs, N. & Bailey, N., eds),pp. 114–122, SERC Daresbury Laboratory, Dares-bury, UK.

Hendrickson, W. A. (1985). Stereochemically restrainedrefinement of macromolecular structures. MethodsEnzymol. 115, 252–270.

Ho, D. D., Neumann, A. U., Perelson, A. S., Chen, W.,Leonard, J. M. & Markowitz, M. (1995). Rapidturnover of plasma virions and CD4 lymphocytes inHIV-1 infection. Nature, 373, 123–126.

Hopkins, A. L., Ren, J., Esnouf, R. M., Willcox, B. E.,Jones, E. Y., Ross, C., Miyasaka, T., Walker, R. T.,Tanaka, H., Stammers, D. K. & Stuart, D. I. (1996).Complexes of HIV-1 reverse transcriptase withinhibitors of the HEPT series reveal conformationalchanges relevant to the design of potent non-nu-cleoside inhibitors. J. Med. Chem. 39, 1589–1600.

Hsiou, Y., Ding, J., Das, K., Clark, A. D., Jr, Hughes, S. H.& Arnold, E. (1996). Structure of unliganded humanimmunodeficiency virus type 1 reverse transcriptaseat 2.7 A resolution: implications of conformationalchanges for polymerization and inhibition mechan-isms. Structure, 4, 853–860.

Jacobo-Molina, A., Ding, J., Nanni, R. G., Clark, A. D., Jr,Lu, X., Tantillo, C., Williams, R. L., Kamer, G., Ferris,A. L., Clark, P., Hizi, A., Hughes, S. H. & Arnold, E.(1993). Crystal structure of human immunodefi-ciency virus type 1 reverse transcriptase complexedwith double-stranded DNA at 3.0 A resolutionshows bent DNA. Proc. Natl Acad. Sci. USA, 90,6320–6324.

Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M.(1991). Improved methods for building proteinmodels in electron density maps and the location oferrors in these models. Acta Crystallog. sect. A, 47,110–119.

Kleywegt, G. J. & Jones, T. A. (1994a). Detection,delineation, measurement and display of cavities inmacromolecular structures. Acta Crystallog. sect. D.50, 178–185.

Kleywegt, G. J. & Jones, T. A. (1994b). Halloween, masksand bones. In From First Map to Final Model (Bailey,S., Hubbard, R. & Waller, D., eds), pp. 59–66, SERCDaresbury Laboratory, Daresbury, U.K.

Kohlstaedt, L. A., Wang, J., Friedman, J. M., Rice, P. A.& Steitz, T. A. (1992). Crystal structure at 3.5 Aresolution of HIV-1 reverse transcriptase complexedwith an inhibitor. Science, 256, 1783–1790.

Larder, B. A. (1992). 3'-Azido-3'-deoxythymidine resist-ance suppressed by a mutation conferring humanimmunodeficiency virus type 1 resistance to non-nu-cleoside reverse transcriptase inhibitors. Antimicrob.Agents Chemother. 36, 2664–2669.

Larder, B. A. (1993). Inhibitors of HIV reversetranscriptase as antiviral agents and drug resistance.(1993) In Skalka, A. M., Goff, S. P., eds, ReverseTranscriptase. pp. 205–222. Cold Spring HarborLaboratory Press, Cold Spring Harbor, N.Y.

Liaw, Y. C., Gao, Y. G., Robinson, H. & Wang, A. H. J.(1991). Molecular structure of a potent HIV-1inhibitor belonging to TIBO family. J. Am. Chem. Soc.113, 1857–1859.

Minor, W. (1993). XDISPLAYF program. Purdue Univer-sity, West Lafayette, IN, USA.

Nanni, R. G., Ding, J., Jacobo-Molina, A., Hughes, S. H.& Arnold, E. (1993). Review of HIV-1 reversetranscriptase three-dimensional structure: impli-cations for drug design. Perspect. Drug Discov.Design, 1, 129–150.

Nunberg, J. H., Schleif, W. A., Boots, E. J., O’Brien, J. A.,Quintero, J. C., Hoffman, J. M., Emini, E. A. &Goldman, M. E. (1991). Viral resistance to human


immunodeficiency virus type 1-specific pyridinonereverse transcriptase inhibitors. J. Virol. 65, 4887–4892.

Otwinowski, Z. (1993). Oscillation data reductionprogram. In Data Collection and Processing (Sawyer,L., Issacs, N. & Bailey, S., eds), pp. 55–62, SERCDaresbury Laboratory, Daresbury, U.K.

Pauwels, R., Andries, K., Debyser, Z., Van Daele, P.,Schols, D., Stoffels, P., De Vreese, K., Woestenborghs,R., Vandamme, A.-M., Janssen, C. G. M., Anne, J.,Cauwenbergh, G., Desmyter, J., Heykants, J.,Janssen, M. A. C., De Clercq, E. & Janssen, P. A. J.(1993). Potent and highly selective human im-munodeficiency virus type 1 (HIV-1) inhibition by aseries of a-anilinophenylacetamide derivativestargeted at HIV-1 reverse transcriptase. Proc. NatlAcad. Sci. USA, 90, 1711–1715.

Pauwels, R., Andries, K., Debyser, Z., Kukla, M. J., Schols,D., Breslin, H. J., Woestenborghs, R., Desmyter, J.,Janssen, M. A. C., De Clercq, E. & Janssen, P. A. J.(1994). New TIBO derivatives are potent inhibitors ofHIV-1 replication and are synergistic with 2',3'-dideoxynucleoside analogues. Antimicrob. AgentsChemother. 38, 2863–2870.

Read, R. J. (1986). Improved Fourier coefficients or mapsusing phases from partial structures with errors. ActaCrystallog. sect. A, 42, 140–149.

Ren, J., Esnouf, R., Garman, E., Somers, D., Ross, C.,Kirby, I., Keeling, J., Darby, G., Jones, Y., Stuart, D.& Stammers, D. (1995a). High resolution structuresof HIV-1 RT from four RT-inhibitor complexes.Nature Struct. Biol. 2, 293–302.

Ren, J., Esnouf, R., Hopkins, A., Ross, C., Jones, Y.,Stammers, D. & Stuart, D. (1995b). The structure ofHIV-1 reverse transcriptase complexed with 9-chloro-TIBO: lessons for inhibitor design. Structure,3, 915–926.

Richman, D. D. (1993). Resistance of clinical isolates ofhuman immunodeficiency virus to antiretroviralagents. Antimicrob. Agents Chemother. 37, 1207–1213.

Richman, D., Shih, C. K., Lowy, I., Rose, J., Prodanovich, P.,Goff, S. & Griffin, J. (1991). Human immunodeficiencyvirus type 1 mutants resistant to non-nucleosideinhibitors of reverse transcriptase arise in tissueculture. Proc. Natl Acad. Sci. USA, 88, 11241–11245.

Richman, D. D., Havlir, D., Corbeil, J., Looney, D.,Ignacio, C., Spector, S. A., Sullivan, J., Cheeseman, S.,Barringer, K., Pauletti, D., Shih, C.-K., Myers, M. &Griffin, J. (1994). Nevirapine resistance mutations ofHIV-1 selected during therapy. J. Virol. 68, 1660–1666.

Rittinger, K., Divita, G. & Goody, R. (1995). Humanimmunodeficiency virus reverse transcriptase sub-strate-induced conformational changes and themechanism of inhibition by non-nucleoside inhibi-tors. Proc. Natl Acad. Sci. USA, 92, 8046–8049.

Rodgers, D. W. (1994). Cryocrystallography. Structure, 2,1135–1140.

Rodgers, D. W., Gamblin, S. J., Harris, B. A., Ray, S., Culp,J. S., Hellmig, B., Woolf, D. J., Debouck, C. &Harrison, S. C. (1995). The structure of unligandedreverse transcriptase from the human immunodefi-ciency virus type 1. Proc. Natl Acad. Sci. USA, 92,1222–1226.

Sarafianos, S. G., Das, K., Ding, J., Hsiou, Y., Hughes,

S. H. & Arnold, E. (1996). Current perspectives onmechanisms of HIV-1 reverse transcriptase inhi-bition by nonnucleoside inhibitors. In Anti-Infec-tives: Recent Advances in Chemistry and StructureActivity Relationships (Bentley, P. H. & O’Hanlon,P. J., eds), The Royal Society of Chemistry,Cambridge, England. In the Press.

Sardana, V. V., Emini, E. A., Gotlib, L., Graham, D. J.,Lineberger, D. W., Long, W. J., Schlabach, A. J.,Wolfgang, J. A. & Condra, J. H. (1992). Functionalanalysis of HIV-1 reverse transcriptase amino acidsinvolved in resistance to multiple non-nucleosideinhibitors. J. Biol. Chem. 267, 17526–17530.

Schinazi, R. F. (1993). Competitive inhibitors of humanimmunodeficiency virus reverse transcriptase. Per-spec. Drug Discov. Design, 1, 151–180.

Smerdon, S. J., Jager, J., Wang, J., Kohlstaedt, L. A.,Chirino, A. J., Friedman, J. M., Rice, P. A. & Steitz,T. A. (1994). Structure of the binding site fornon-nucleoside inhibitors of the reverse transcrip-tase of human immunodeficiency virus type 1. Proc.Natl Acad. Sci. USA, 91, 3911–3915.

Smith, M. B. K., Rouzer, C. A., Taneyhill, L. A., Smith, N.A., Hughes, S. H., Boyer, P. L., Janssen, P. A. J.,Moereels, H., Koymans, L., Arnold, E., Ding, J., Das,K., Zhang, W., Michejda, C. J. & Smith, R. H., Jr(1995). Molecular modeling studies of HIV-1 reversetranscriptase non-nucleoside inhibitors: total energyof complexation as a predictor of drug placementand activity. Protein Sci. 4, 2203–2222.

Spence, R. A., Kati, W. M., Anderson, K. S. & Johnson,K. A. (1995). Mechanism of inhibition of reversetranscriptase by non-nucleoside inhibitors. Science,267, 988–993.

Spence, R. A., Anderson, K. S. & Johnson, K. A. (1996).HIV-1 reverse transcriptase resistance to non-nu-cleoside inhibitors. Biochemistry, 35, 1054–1063.

Stura, E. A. & Wilson, I. A. (1990). Analytical andproduction seeding techniques. Methods, 1, 38–49.

Tantillo, C., Ding, J., Jacobo-Molina, A., Nanni, R. G.,Boyer, P. L., Hughes, S. H., Pauwels, R., Andries, K.,Janssen, P. A. J. & Arnold, E. (1994). Locations ofanti-AIDS drug binding sites and resistance mu-tations in the three-dimensional structure of HIV-1reverse transcriptase: implications for mechanismsof drug inhibition and resistance. J. Mol. Biol. 243,369–387.

Teng, T. Y. (1990). Mounting of crystals for macromolecu-lar crystallography in a free-standing thin film.J. Appl. Crystallog. 24, 387–391.

Vandamme, A.-M., Debyser, Z., Pauwels, R., DeVreese,K., Goubau, P., Youle, M., Gazzard, B., Stoffels, P. A.,Cauwenbergh, C. F., Anne, J., Andries, K., Janssen,P. A. J., Desmyter, J. & De Clerq, E. (1994).Characterization of HIV-1 strains isolated frompatients treated with TIBO R82913. AIDS Res. HumanRetro. 10, 39–46.

Wei, X., Ghosh, S. K., Taylor, M. E., Johnson, V. A., Emini,E. A., Deutsch, P., Lifson, J. D., Bonhoeffer, S.,Nowak, M. A., Hahn, B. H., Saag, M. S. & Shaw,G. M. (1995). Viral dynamics in human immunodefi-ciency virus type 1 infection. Nature, 373, 117–122.

Young, S. D. (1993). Non-nucleoside inhibitors of HIV-1reverse transcriptase. Perspect. Drug Discov. Design, 1,181–192.

Edited by J. Karn

(Received 26 September 1996; accepted 29 October 1996)

Crystal Structures of 8Cl and 9Cl TIBO Complexed with Wild-type HIV1 RT and 8Cl TIBO Complexed with the Tyr181Cys HIV1 RT Drug-resistant Mutant

Documents