mqbptoirpf

Analytical Biochemistry 277, 247–253 (2000)doi:10.1006/abio.1999.4388, available online at http://www.idealibrary.com on

Yeast Two-Hybrid Assay for Examining HumanImmunodeficiency Virus Protease HeterodimerFormation with Dominant-Negative Inhibitorsand Multidrug-Resistant Variants

Stephen Todd,* Martha C. A. Laboissiere,† and Charles S. Craik*,‡,§,1

*Department of Pharmaceutical Chemistry, ‡Department of Molecular and Cellular Pharmacology, and §Departmentof Biochemistry and Biophysics, University of California, San Francisco, California 94143-0446; and†Departamento de Ciencias Farmaceuticas, Faculdade de Ciencias da Saude, Universidadede Brasılia, Campus Universitario Darcy Ribeiro, Asa Norte, 70910-900 Brasılia, DF, Brazil

Received August 17, 1999

as

The yeast two-hybrid assay was used to study the

dimerization of engineered and naturally occurringvariants of human immunodeficiency virus (HIV) pro-tease (PR) monomers. Defective monomers that werepreviously shown to exhibit a dominant-negative(D-N) effect in cultured mammalian cells were testedfor their ability to interact in the two-hybrid assay.Similarly, monomers with dimer-interface substitu-tions and monomers harboring in vivo selected muta-tions that confer multidrug resistance (mdr) in anAIDS patient were tested for interaction in yeast.Dimer formation between wt monomers with catalyticaspartates was not detected in yeast, whereas thedimerization of PR monomers harboring the acid ac-tive site substitution D25N was readily demonstrated.The use of inactive monomers harboring the D25Nsubstitution as a genetic background for studying ad-ditional HIV PR mutations allowed for the probing ofinteractions between monomers with mdr-associated

utations with those based on the HIV-1 HXB2R se-uence. The HTLVIII/HIV-1 HXB2R clone has been theasis for a large number of HIV-related plasmids,rimers, antibodies, and other specific reagentshroughout the HIV research community. The resultsf our assay suggest that HXB2R-based D-N PR inhib-tors associate with variant monomers based on theecently obtained nucleotide sequence from an AIDSatient with a multidrug-resistant virus. These resultsurther encourage the use of D-N PR inhibitors as

1 To whom correspondence should be addressed. Fax: (415) 502-8298. E-mail: craik@cgl.ucsf.edu.

0003-2697/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

ntiviral agents which may complement existingmall-molecule combination therapies. © 2000 Academic

Press

Key Words: HIV; protease; monomers; dominant-neg-ative; yeast two-hybrid.

The yeast two-hybrid approach has been used toidentify protein–protein interactions in numerous bio-logical systems [for reviews see (1, 2)]. In most cases, aknown protein (or protein domain) is tested against alibrary of proteins (or domains) to identify novel inter-molecular interactions. More recently, the two-hybridapproach has been used in the study of known protein–protein interactions to define structural determinantsresponsible for these interactions (3, 4, and others).Predictably, the yeast two-hybrid assay has been usedto investigate protein-protein interactions involved inthe replication of human immunodeficiency virus(HIV)2 type I, the causative agent of acquired immu-nodeficiency syndrome (AIDS). Interactions betweenhost-cell factors and HIV regulatory proteins [e.g., Rev(5, 6), Nef (7–11), and Tat (12, 13)] have been investi-gated, as well as interactions among gag polypeptideswhich allow the formation of virus core particles (14,15). The two-hybrid assay has also been used to studythe requirement for proximal gag sequences in the

2 Abbreviations used: HIV, human immunodeficiency virus; AIDS,acquired immunodeficiency syndrome; KWW, three amino acidchanges of D25K, G49W, and I50W; D-N, dominant-negative; PR,

protease; AD, activation domain; 3-AT, 3-amino-1,2,4-triazole; DBD,DNA-binding domain; UAS, upstream activating sequence.

247

evshmsat

HcaticPobiram

C

cs[GPiaG

sY

248 TODD, LABOISSIERE, AND CRAIK

efficient dimerization of gag–protease fusion products(16). Herein, we describe the use of the yeast two-hybrid assay to investigate the relative efficiency ofdimerization between engineered and naturally occur-ring variants of HIV PR in the absence of additionalgag or pol domains.

Expression of HIV PR monomers harboring activesite mutations has been shown to interfere with gagprocessing in cultured cells, presumably due to theformation of catalytically defective protease het-erodimers (17). A PR variant containing three aminoacid changes [D25K, G49W, and I50W (henceforth ab-breviated KWW, refer to Fig. 1A)] was the most effec-tive dominant-negative (D-N) inhibitor assayed in cul-tured cells (18, 19). The large, bulky side chainsubstitutions presumably promote heterodimer forma-tion by filling the substrate binding pocket of thedimeric protease while minimizing defective ho-modimer formation due to steric hindrance (18). Dem-onstrating HIV PR dimer formation in mammaliancells has proven arduous, if for no reason other thanthe difficulty in detecting HIV PR by immunoblottingor metabolic labeling. Formation of wt-KWW het-erodimers has recently been shown in vitro, using bac-terially expressed and purified monomers, and in Esch-richia coli, using plasmids which coexpress wt andariant monomers (Rozzelle et al., unpublished re-ults); nonetheless, substantiating data demonstratingeterodimer formation in eukaryotic cells have re-ained elusive. The yeast two-hybrid assay allows the

tudy of PR dimer formation in eukaryotic cells byssociating a growth phenotype with PR monomer in-eractions.

Our findings show that dimer formation betweenIV PR monomers harboring active site substitutions

an be detected in yeast. We have used the two-hybridssay to determine whether monomers harboring mu-ations at or near the dimer interface region of HIV PRnteract with “wt” monomers based on the HXB2Rlone (20) of HIV-1. A highly substituted variant of HIVR [cloned from a multidrug-resistant clinical isolatef HIV (21)], interacted efficiently with the HXB2R-ased inhibitor. This finding suggests that the D-Nnhibitor may represent a clinically important antivi-al agent which is effective against virus isolates thatre resistant as well as sensitive to current small-olecule inhibitor combinations.

MATERIALS AND METHODS

Yeast two-hybrid strains and constructs. pACTIIand pAS1-CYH2 (Clontech) encode the GAL4 activa-tion domain (AD or ACT) and DNA binding domain(DBD), respectively. Oligonucleotides Y2H12(1)[59-

ATATGGCCATGGGAGGCGGTGGAGGGCCCCA-GGTC-ACTCTTTGGAAACGACCCCTCGTCACA-39

(NcoI site, underlined; Gly5 linker, italics; first codon ofHIV PR, bold; Q7K, bold and underlined)] andY2H5(2) [59-TCGGATCCAGTACTAAAAATTTAAAG-TGCAACCAATCTG-39 (BamHI site, underlined; stopodon, bold)] were used to amplify the HIV PR codingequences from pHIV-gpt vectors (22) harboring the wtdefined as HIV-1 HXB2 (20)], D25N (23), or D25K/49W/I50W [referred to as KWW (18)] mutations. TheCR products were digested with NcoI and BamHI and

nserted into pACTII and pAS1-CYH2 using the NcoInd BamHI sites in the polylinkers to produce pACT-5-Q7K-wt, pACT-G5-Q7K-D25N, pACT-G5-Q7K-KWW,

pDBD-G5-Q7K-wt, pDBD-G5-Q7K-D25N, pDBD-G5-Q7K-KWW which express wt or variant HIV PR as aC-terminal fusion protein following a five-glycine (G5)linker after the respective GAL4 domain. The combi-nation of oligonucleotides Y2H12(1) with Y2H19(2)[59-TCGGATCCAGTACTAAAAATTTACAGTGCAA-CCAATCTG-39(BamHI site, underlined; stop codon,bold; L97V mutation, bold and underlined)] orY2H12(1) with Y2H20(2) [59-TCGGATCCAGTACTA-AAAATTCGAAG-TGCAACCAATCTG-39 (BamHI site,underlined; stop codon, bold; L97S mutation, bold andunderlined)]) were used to amplify a PCR product con-taining an additional L97V or L97S mutations whichwere subsequently cloned into the same pACTII andpAS1-CYH2 vectors. Oligonucleotide pairs Y2H12(1)/NWW-ant (59-CATCTGCTCCTGTGTTTAATAGAG-CTTC-39) and NWW-sen (59-GAAGCTCTATTAAACA-CAGGAGCA-GATG-39)/Y2H5(2) (Asn-25 codons inbold) were used to introduce the D25N mutation into amultidrug-resistant (mdr) HIV PR gene obtainedfrom N. Bandrapalli and R. Grant at San FranciscoGeneral Hospital [(21), see Results]. OligonucleotidesY2H12(1)/Y2H5(2) were used to amplify the completeD25N/mdr PR-coding sequence for insertion into pAC-TII and pAS1-CYH2 yielding pACT-G5-Q7K-D25N-mdr and pDBD-G5-Q7K-D25N-mdr. All cloning wasperformed in E. coli strain XL-1 Blue or HB101 andplasmids were sequenced by the UCSF BiomolecularResource Center.

Yeast transformations, media, and b-galactosidaseassays. Transformations of Saccharomyces cerevisiaestrain Y190 (Clontech) were performed using the lith-ium acetate method (24). Transformed Y190 cells wereplated on SD/Glu/Leu2/Trp2/His1 plates and incubatedfor 3–4 days at 30°C. pACTII and pAS1-CYH2 (orderivatives) contain the LEU2 and TRP1 genes, re-pectively. Resulting colonies were grown overnight inPD and plated on fresh Leu2/Trp2/His1 plates for

further analysis. Colony filter-lift assays using 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside (X-gal)as a substrate were performed as described (25), except

that filters were incubated at 37°C. 3-Amino-1,2,4-tria-zole [3-AT (Sigma)] was added to SD/Glu/Leu2/Trp2/

fp

ol

fDFbs

T

pcp

taiim3tpssPratMrcdsn(foldsaD

249HETERODIMER FORMATION IN THE TWO-HYBRID ASSAY

His2 drop-out media at various concentrations (rang-ing from 20 to 60 mM, final) to inhibit yeast growth dueto leakage of the GAL4-responsive HIS3 promoter.

RESULTS

Yeast two-hybrid assay to characterize HIV PR mono-mer–monomer interactions. Wt, D25N, and KWWmonomers were expressed as C-terminal fusion pro-teins with the GAL4 activation domain (AD) or DNA-binding domain (DBD) (Fig. 1B). Interactions betweenmonomers resulted in the expression of two reportergenes (HIS3 and lacZ) under control of a GAL4-respon-sive upstream activating sequence (UAS). Four seriesof constructs were tested in the two-hybrid assay (Fig.1C) which featured the HIV PR-coding sequence (a)inserted into the polylinker of the two-hybrid vectors,(b) inserted into the polylinker and immediately fol-lowed by a stop codon, (c) inserted into the polylinkerwith a Gly5-linker between the GAL4 and PR domains,ollowed by a stop codon, and (d) inserted into theolylinker with the Gly5-linker, stop codon, and the

Q7K mutation, which has been shown to inhibit HIVPR autoprocessing (26). Simply cloning the HIV PR-coding sequence into the polylinker of the yeast two-hybrid constructs yielded very poor signals that wereindistinguishable from background noise. The pres-ence of a stop codon immediately after the PR-codingsequence improved the interactions. The Gly5-linkerfurther improved the signals, whereas the Q7K muta-tion, which reduces autoproteolysis in vitro (26), had nobvious effect (data not shown). Only data from theatter set of constructs [Fig. 1C(d)] will be discussed.

All combinations of wt, D25N, and KWW monomersused to the AD or DBD (as well as the free AD andBD) were transformed into S. cerevisiae strain Y190.our colonies from each transformation were assayedy two different methods: (i) filter-lift assays to mea-ure b-galactosidase activity using the substrate, X-

gal, and (ii) the ability of the yeast to grow on Leu2/rp2/His2 plates supplemented with 20 (data not

shown), 40, or 60 mM 3-AT. (Fig. 2). The D25N mono-mers (i.e., the plasmid combination D25N-AD:D25N-DBD) interacted to give robust b-galactosidase signals,and colonies grew on Leu2/Trp2/His2 supplementedwith 60 mM 3-AT. The free AD produced a high level ofbackground with the D25N monomer in b-galactosi-dase assays, consistent with previous results [(16), seeDiscussion]. Nonetheless, background signal wasreadily distinguished from specific monomer interac-tions (e.g., D25N–D25N) using media containing40–60 mM 3-AT. The combination of two wt monomersdid not result in a significant b-galactosidase signal,

resumably as a consequence of creating a proteolyti-

ally active HIV PR dimer which likely cleaves aolypeptide involved in the production of b-galactosi-

FIG. 1. Diagram of HIV-1 protease and schematic representationof plasmids used for yeast two-hybrid assays. (A) a carbon trace ofhe HIV PR homodimer, including the carbon backbones of the cat-lytic aspartates (residues 25). Residues important for this study arendicated on the right-hand monomer. The dimer interface regionncludes residues 1–4 and 96–99 of each monomer with additional

onomer contacts located in the flap region (residues 48–52) (36,7). The variant KWW consists of D25K/G49W/G50W, resulting inhe positioning of tryptophan side chains in the substrate-bindingocket which presumably mimics the presence of substrate andtabilizes the heterodimer (18). Substitutions at position 82 (in theubstrate-binding cavity) confer resistance to most small-moleculeR inhibitors [(29) and others]. Many substitutions associated withesistance to PR inhibitors (e.g., M36I, L63P, and L90M) are locatedt the periphery of the PR dimer. Position 97 (in combination withhe V82A mutation) is associated with resistance to the DuPont–erck compound XM323 (27), residing within the dimer interface

egion. (B) The modular GAL4 transcription factor contains a dis-rete DNA-binding domain and activation domain. Fusion of theseomains to exogenous proteins (e.g., HIV PR monomers) can recon-titute GAL4 function if the exogenous domains interact in theuclease of the yeast [for reviews of the yeast two-hybrid assay see1, 2) and others]. (C) The constructs tested in the two-hybrid assayeatured the HIV PR-coding sequence (a) inserted into the polylinkerf the two-hybrid vectors, (b) inserted into the polylinker and fol-owed by a stop codon, (c) following a Gly5 linker after the GAL4omains, and followed by a stop codon, and (d) with the Gly5 linker,top codon, and the Q7K mutation to minimize autoprocessing. ADH,lcohol dehydrogenase promoter; ACT or DBD, activation domain orNA-binding domain; G5, five-glycine linker; PR, HIV protease;

small black rectangle, engineered stop codon; term, translation ter-mination sequence in the vector.

le

wCQo

250 TODD, LABOISSIERE, AND CRAIK

dase (e.g., the AD or DBD fusion proteins). The obser-vation that at least one transformant containing theplasmid combination wt-AD/D25N-DBD interactedweakly (even on plates containing 60 mM 3-AT) pro-vides evidence for this model (Fig. 2).

Previously reported transient transfection data hasindicated that wt–KWW heterodimers form efficientlyin mammalian cells (18), in the context of the gag-polpolyprotein. D25N–KWW heterodimers should formwith similar efficiency while avoiding the problemsassociated with the expression of wt PR monomers inthe yeast two-hybrid assay. Curiously, KWW mono-mers fused to the AD interacted only weakly withD25N monomers fused to the DBD and no interactionwas detected when the monomers were expressed inthe reverse arrangement. The relative insolubility ofKWW monomers in bacteria (Rozzelle et al., unpub-ished results) and unexpectedly low solubility of teth-red wt–KWW heterodimers in vitro (Todd and Craik,

unpublished observations) may have been portent forthe failure to efficiently detect wt–KWW interactionsin the yeast nucleus. Results from the two-hybrid assayshow in Fig. 2 are summarized in Fig. 4.

Characterization of variant PR monomers harboring

FIG. 2. Results of two-hybrid assay with wt, D25N, and KWWmonomers. Yeast were transformed with pairwise combinations ofpACTII (encoding the free GAL4 activation domain), pACT-G5-Q7K-

t, pACT-G5-Q7K-D25N, or pACT-G5-Q7K-KWW along with pAS1-YH2 (encoding the free GAL4 DNA-binding domain), pACT-G5-7K-wt, pACT-G5-Q7K-D25N, or pACT-G5-Q7K-KWW and plated

n SD/Glu/Leu2/Trp2/His1 plates. Four colonies from each transfor-mation were grown overnight in YPD and dropped onto fresh SD/Glu/Leu2/Trp2/His1 plates. Two-hybrid interactions were assayed byperforming b-galactosidase filter-lift assays and by replica-platingyeast onto SD/Glu/Leu2/Trp2/His2 plates supplemented with 40 or60 mM 3-aminotriazole (3-AT) to suppress growth due to leaky HIS3expression.

mutations in the dimer interface region. The L97Vmutation (combined with V82A) has been associated

with resistance to XM323 (27), a DuPont–Merck, non-peptidyl, symmetrically substituted, cyclic urea com-pound (28). L97V localizes to the highly conserveddimer interface region consisting of residues 96–99(refer to Fig. 1). L97V monomers (lacking the V82Amutation) form active protease capable of processingthe gag polypeptide in cultured COS cells (data notshown), indicating that the V82A mutation is not re-quired for the function of L97V substitution-containingmonomers. The yeast two-hybrid assay was used todetermine whether L97V monomers preferentially (orobligatorily) form homodimers or instead form a mix-ture of homo- and heterodimers when mixed with wtmonomers. The yeast two-hybrid combination D25N/L97V-AD:D25N-DBD interacted based on b-galactosi-dase colony lift assays. D25N/L97V monomers fused tothe DBD did not interacted with either D25N norD25N/L97V monomers. The combination D25N/L97V-AD:D25N/L97V-DBD produced less b-galactosidasesignal than AD:DBD (data not shown). Figure 3 showsgrowth of these yeast transformants on Leu2/Trp2/His2 plates supplemented with 50 mM 3-AT. Despiteshowing similar levels of b-galactosidase activity infilter-lift assays, D25N-AD:D25N-DBD transformantsgrew on 3-AT plates, whereas D25N/L97V-AD:D25N-DBD transformant did not. This result suggested theL97V mutation perturbs dimerization in the two-hy-brid assay. Monomers harboring a less conservativedimer interface region mutation, L97S (designed spe-cifically to interfere with dimerization), also failed to

FIG. 3. Growth of yeast expressing two-hybrid fusion proteins withD25N/L97V, D25N/L97S, and D25N/multidrug-resistant (mdr) vari-ants of HIV PR on SD/Glu/Leu2/Trp2/His2 media containing 50 mM3-AT. The mdr protease contains seven substitutions (M36I/M46I/L63P/A71T/V82A/N88D/L90M) associated with resistance to small-molecule PR inhibitors (21, 29) and five substitutions (T12K/L19V/E35D/S37D/I93V) which may or may not be involved in drugresistance. The controls for the D25N/mdr-AD:D25N-DBD interac-

tion were D25N/L97S-AD:D25N-DBD (monomers that should notinteract) and D25N-AD:D25N-DBD (monomers that do interact).

tfmtmafeor

cret

fCpvHyyftissommn

le

tg

251HETERODIMER FORMATION IN THE TWO-HYBRID ASSAY

interact with D25N monomers fused to the DBD (Fig.3).

Characterization of a multidrug-resistant variantmonomers in the two-hybrid assay. The PR gene froma clinical isolate of HIV-1, resistant to all FDA-ap-proved inhibitors of HIV replication, including small-molecule PR inhibitors, has recently been cloned (21).This multidrug-resistant (mdr) protease contains theM36I/M46I/L63P/A71T/V82A/N88D/L90M substitu-tions, associated with resistance to small-molecule PRinhibitors [(21, 29), and others], in addition to othervariations (T12K/L19V/E35D/S37D/I93V) which mayor may not be involved in drug resistance. The activesite D25N mutation was cloned into the mdr PR-codingsequence to allow the study of D25N/mdr monomers inhe yeast two-hybrid assay. D25N/mdr PR monomersormed homodimers and heterodimers with D25Nonomers regardless of the orientations with respect

o the GAL4 AD or DBD domains (Fig. 3). A D25N/mdronomer harboring an additional R41G substitution

lso interacted with D25N and D25N/mdr monomersused to the DBD, a further testimony to the well-stablished structural plasticity of HIV PR. The resultsf the two-hybrid assays described above are summa-ized in Fig. 4.

DISCUSSION

Previous data from this laboratory has examined theefficacy of dominant-negative HIV PR monomers asmacromolecular inhibitors of HIV replication (17–19).Obtaining biochemical data from these studies hasbeen hampered by the difficulty in detecting PR incultured mammalian cells. Recent data have demon-strated the formation of heterodimers in vitro and in E.oli (Rozzelle et al., unpublished results). The dataeported herein demonstrate the formation of het-rodimers in eukaryotic cells using the versatile yeastwo-hybrid assay.

The expression of HIV PR monomers as C-terminalusion proteins is consistent with the expression of PR-terminal with respect to gag during HIV gene ex-ression. With the added flexibility and isolation pro-ided by the Gly5-linker, it is not surprising that theIV PR dimerization results in robust signals in the

east two-hybrid assay. Nonetheless, the use of theeast two-hybrid assay to study dimer formation suf-ers from at least three significant complications (refero Fig. 4). First, the formation of active protease dimersn the yeast nucleus appears to interfere with the as-ay, mandating the use of monomers harboring activeite substitutions. This necessity limits the applicationf the assay by precluding the evaluation of monomer–onomer interactions between PR variants harboring

utations near the active site that may depend on theegative charge of the catalytic aspartates. While it is

possible that the inhibition of protease activity in yeastcells, through dominant-negative interactions, could beuseful as an assay, it must be noted that active pro-tease may be formed between fusion proteins with thesame GAL4 domain. For example, the two-hybrid com-bination mdr-DBD:D25N-AD may result in the forma-tion of active mdr-DBD homodimers (which adverselyaffect cell viability) as well as inactive mdr-DBD:D25N-AD heterodimers (which activate the reportergenes). The ultimate level of reporter gene activity insuch an assay would depend upon (i) the kinetics ofHIV protease heterodimer formation resulting inGAL4 activity, (ii) the kinetics of active protease ho-modimer formation not resulting in GAL4 activity, and(iii) the relative activity (and perhaps specificity, ifchanged) of the resulting active protease and its effecton the yeast cell. Since the presence of any wt proteasemonomers in the yeast cells makes interpretation ofthe assay problematic, we chose to use only inactivemonomers to screen additional protease monomer in-teractions.

Second, interactions between the most-substitutedD-N monomers (KWW) and wt or D25N monomers[recently demonstrated in vitro and in E. coli (Rozzelleet al., unpublished results)] are poorly detected in thetwo-hybrid assay, likely due to misfolding. This hy-pothesis is supported by the relative insolubility ofKWW monomers in bacteria (Rozzelle et al., unpub-ished results) and unexpected low solubility of teth-red wt–KWW heterodimers in vitro [,1.5 mg/ml

FIG. 4. Summary of observed yeast two-hybrid interactions. Rel-ative efficiency of interaction is based upon growth on SD/Glu/Leu2/Trp2/His2 media supplemented with 3-AT. The scoring of interac-ions [2, no growth; 1/2, growth in some experiments; 1, consistentrowth; 11, robust growth] is consistent with that indicated in Figs.

2 and 3. The absence of an entry signifies that the interaction was nottested.

(Todd and Craik, unpublished observations) under con-ditions where wt PR is soluble to 6 mg/ml]. Based on

Ryia

cGfm

abfi

c

252 TODD, LABOISSIERE, AND CRAIK

microcalorimetry experiment following denaturationand renaturation of KWW monomers in the presence ofwt monomers, wt–KWW heterodimers are indeed morestable than wt–wt homodimers, as predicted by molec-ular modeling (Rozzelle et al., unpublished results).

elying upon the yeast two-hybrid assay for the anal-sis of more subtle characteristics of protein–proteinnteractions is problematic when known interactionsre undetectable in the assay.Finally, high background is observed due to nonspe-

ific interactions when HIV PR monomers fused to theAL4 DNA-binding domain are coexpressed with the

ree GAL4 activation domain. This observation wasade previously and it was postulated that the b-sheet

conformations responsible for both PR dimerization(refer to Fig. 1A) and GAL4 transcriptional activationalso caused nonspecific interactions between the freeGAL4 activation domain and HIV PR monomers (ordimers) (16). The use of 40–60 mM 3-AT in His2 mediallows the distinction between specific interaction andackground signal, while more qualitative assays (e.g.,lter-lift assays for b-galactosidase activity) were inca-

pable of making this distinction (Fig. 2). Even liquidculture-based b-galactosidase assays were unreliablefor measuring HIV protease monomer–monomer inter-actions.

While the yeast two-hybrid assay has been a power-ful tool for the examination of protein–protein interac-tions in many systems, there is clearly an importantdistinction between interaction in the nucleus of S.erevisiae and biological relevance, in vivo. This apho-

rism is immediately evident upon examination of re-cent literature reporting the use of the yeast two-hy-brid assay to examine putative interactions betweenHIV regulatory proteins and host factors. Reportedly,Nef interacts with (i) beta-COP, a component of non-clathrin-coated vesicles essential for membrane traffic(7), (ii) CD4 (8), (iii) the HsN3 proteasomal subunit (9),and (iv) novel thioesterase(s) (10, 11). The multifunc-tional nature of Nef is exceeded only by that of Vprwhich apparently interacts with (i) the uracil DNAglycosylase DNA repair enzyme (30), (ii) virus matrixprotein p17 (31), (iii) a human 34-kDa mov34 homo-logue linked to the G2/M phase transition of the mam-malian cell cycle (32), (iv) Lys-tRNA synthetase (33)and (v) HHR23A, a cellular protein implicated in nu-cleotide excision DNA repair (34, 35). HIV regulatoryproteins Rev and Tat have also been studied for theability to interact with host-cell proteins using yeasttwo-hybrid screens. Taken together, it is likely thatsome of these interactions are specific to the nucleus ofS. cerevisiae. These interactions, and those that havebeen described herein, must ultimately be demon-strated by more direct methods. Nonetheless, by opti-

mizing our yeast two-hybrid fusion proteins to allowthe detection of HIV PR monomer interactions, and

then using this genetic background to assay the effectsof additional PR mutation on PR dimerization, we haveattempted to minimize the contribution of nonspecificinteractions in our assay.

In this study, we have examined the interactionsbetween wt (HXB2R) HIV PR, two dominant-negative(D-N) variants (D25N, KWW), two dimer-interfacevariants (L97V, L97S), and a multidrug-resistant vari-ant (mdr) in which greater than 10% of the amino acidresidues differ from those of the HXB2R sequence. Insome cases, monomer interactions were not detected(e.g., KWW and L97V), while robust signals were gen-erated with other monomer combinations (e.g., D25Nand mdr) (Fig. 4). This latter interaction is of particu-lar significance because it suggests that a D-N mono-mer based on the widely studied HXB2R amino acidsequence can interact with a PR variant which hasbeen selected in the presence of small-molecule PRinhibitor cocktails. This finding provides encourage-ment to pursue D-N PR inhibitors as antiviral agentswhich can potentially be delivered to CD41 lympho-cytes (or CD341 progenitor cells) to control the replica-tion of HIV.

ACKNOWLEDGMENTS

Yeast two-hybrid reagents and advice were generously provided byRob Brazas and Don Ganem. Additional assistance in yeast geneticswas kindly provided by Sean O’Rourke and Mary Maxon. The mul-tidrug-resistant HIV PR gene was obtained from Nirmala Ban-drapalli and Robert Grant at San Francisco General Hospital. Oli-gonucleotides NWW-ant and NWW-sen were provided by DeborahDauber. We thank D. Dauber and Todd Pray for critical readings ofthe manuscript. This work was funded by University-wide AIDSResearch Program Postdoctoral Fellowship Award F97-SF-155 (PI:S. Todd) and NIH Grants GM56531 (PI: C. S. Craik) and AI35707(PI: C. S. Craik).

REFERENCES

1. Bartel, P. L., and Fields, S. (1995) Methods Enzymol. 254, 241–263.

2. Fields, S., and Sternglanz, R. (1994) Trends Genet. 10, 286–292.3. Nolandt, O., Kern, V., Muller, H., Pfaff, E., Theilmann, L.,

Welker, R., and Krausslich, H. G. (1997) J. Gen. Virol. 78, 1331–1340.

4. Yasugi, T., Benson, J. D., Sakai, H., Vidal, M., and Howley, P. M.(1997) J. Virol. 71, 891–899.

5. Fritz, C. C., Zapp, M. L., and Green, M. R. (1995) Nature 376,530–533.

6. Luo, Y., Yu, H., and Peterlin, B. M. (1994) J. Virol. 68, 3850–3856.

7. Benichou, S., Bomsel, M., Bodeus, M., Durand, H., Doute, M.,Letourneur, F., Camonis, J., and Benarous, R. (1994) J. Biol.Chem. 269, 30073–30076.

8. Rossi, F., Gallina, A., and Milanesi, G. (1996) Virology 217,397–403.

9. Rossi, F., Evstafieva, A., Pedrali-Noy, G., Gallina, A., andMilanesi, G. (1997) Virology 237, 33–45.

1

1

253HETERODIMER FORMATION IN THE TWO-HYBRID ASSAY

10. Liu, L. X., Margottin, F., Le Gall, S., Schwartz, O., Selig, L.,Benarous, R., and Benichou, S. (1997) J. Biol. Chem. 272,13779–13785.

1. Watanabe, H., Shiratori, T., Shoji, H., Miyatake, S., Okazaki, Y.,Ikuta, K., Sato, T., and Saito, T. (1997) Biochem. Biophys. Res.Commun. 238, 234–239.

12. Fridell, R. A., Harding, L. S., Bogerd, H. P., and Cullen, B. R.(1995) Virology 209, 347–357.

13. Kamine, J., Elangovan, B., Subramanian, T., Coleman, D., andChinnadurai, G. (1996) Virology 216, 357–366.

14. Luban, J., Alin, K. B., Bossolt, K. L., Humaran, T., and Goff, S. P.(1992) J. Virol. 66, 5157–5160.

15. Li, X., Yuan, B., and Goff, S. P. (1997) J. Virol. 71, 5624–5630.16. Zybarth, G., and Carter, C. (1995) J. Virol 69, 3878–3884.7. Babe, L. M., Rose, J., and Craik, C. S. (1995) Proc. Natl. Acad.

Sci. USA 92, 10069–10073.18. McPhee, F., Good, A. C., Kuntz, I. D., and Craik, C. S. (1996)

Proc. Natl. Acad. Sci. USA 93, 11477–11481.19. Junker, U., Escaich, S., Plavec, I., Baker, J., McPhee, F., Rose,

J. R., Craik, C. S., and Bohnlein, E. (1996) J. Virol. 70, 7765–7772.

20. Ratner, L., Haseltine, W., Patarca, R., Livak, K. J., Starcich, B.,Josephs, S. F., Doran, E. R., Rafalski, J. A., Whitehorn, E. A.,Baumeister, K., et al. (1985) Nature 313, 277–284.

21. Hecht, F. M., Grant, R. M., Petropoulos, C. J., Dillon, B.,Chesney, M. A., Tian, H., Hellmann, N. S., Bandrapalli, N. I.,Digilio, L., Branson, B., and Kahn, J. O. (1998) N. Engl. J. Med.339, 307–311.

22. Page, K. A., Landau, N. R., and Littman, D. R. (1990) J. Virol.64, 5270–5276.

23. Babe, L. M., Pichuantes, S., and Craik, C. S. (1991) Biochemistry30, 106–111.

24. Ausubel, F. M. (1995) Short Protocols in Molecular Biology: A

Compendium of Methods from Current Protocols in MolecularBiology, Wiley, New York.

25. Schneider, S., Buchert, M., and Hovens, C. M. (1996) BioTech-niques 20, 960–962.

26. Rose, J. R., Salto, R., and Craik, C. S. (1993) J. Biol. Chem. 268,11939–11945.

27. King, R. W., Garber, S., Winslow, D. L., Reid, C., Bacheler, L. T.,Anton, E., and Otto, M. J. (1995) Antiviral Chem. Chemother. 6,80–88.

28. Otto, M. J., Reid, C. D., Garber, S., Lam, P. Y., Scarnati, H.,Bacheler, L. T., Rayner, M. M., and Winslow, D. L. (1993) Anti-microb. Agents Chemother. 37, 2606–2611.

29. Mellors, J. W., Larder, B. A., Schinazi, R., and Deverson, C.(1995) J. Acquired Immune Defic. Syndromes Hum. Retrovirol.10, S45.

30. Bouhamdan, M., Benichou, S., Rey, F., Navarro, J. M., Agostini,I., Spire, B., Camonis, J., Slupphaug, G., Vigne, R., Benarous, R.,and Sire, J. (1996) J. Virol. 70, 697–704.

31. Sato, A., Yoshimoto, J., Isaka, Y., Miki, S., Suyama, A., Adachi,A., Hayami, M., Fujiwara, T., and Yoshie, O. (1996) Virology 220,208–212.

32. Mahalingam, S., Ayyavoo, V., Patel, M., Kieber-Emmons, T.,Kao, G. D., Muschel, R. J., and Weiner, D. B. (1998) Proc. Natl.Acad. Sci. USA 95, 3419–3424.

33. Stark, L. A., and Hay, R. T. (1998) J. Virol. 72, 3037–3044.34. Withers-Ward, E. S., Jowett, J. B., Stewart, S. A., Xie, Y. M.,

Garfinkel, A., Shibagaki, Y., Chow, S. A., Shah, N., Hanaoka, F.,Sawitz, D. G., Armstrong, R. W., Souza, L. M., and Chen, I. S.(1997) J. Virol. 71, 9732–9742.

35. Gragerov, A., Kino, T., Ilyina-Gragerova, G., Chrousos, G. P.,and Pavlakis, G. N. (1998) Virology 245, 323–330.

36. Navia, M. A., Fitzgerald, P. M., McKeever, B. M., Leu, C. T.,Heimbach, J. C., Herber, W. K., Sigal, I. S., Darke, P. L., andSpringer, J. P. (1989) Nature 337, 615–620.

37. Wlodawer, A., Miller, M., Jaskolski, M., Sathyanarayana, B. K.,

Baldwin, E., Weber, I. T., Selk, L. M., Clawson, L., Schneider, J.,and Kent, S. B. (1989) Science 245, 616–621.