Vol. 66, No. 12

Novel Mutation in the Human Immunodeficiency VirusType 1 Reverse Transcriptase Gene That Encodes

Cross-Resistance to 2',3'-Dideoxyinosineand 2',3'-Dideoxycytidine

ZHENGXIAN GU, QING GAO, XUGUANG LI, MICHAEL A. PARNIAK,AND MARK A. WAINBERG*

Lady Davis Institute-Jewish General Hospital and McGill University AIDS Centre,3755 Chemin C6te Ste-Catherine, Montreal,

Quebec, Canada H3T 1E2

Received 8 July 1992/Accepted 22 September 1992

We have used the technique of in vitro selection to generate variants of human immunodeficiency virus type1 (H1V-i) that are resistant to 2',3'-dideoxyinosine (ddl) and cross-resistant to 2',3'-dideoxycytidine (ddC).The complete reverse transcriptase (RT)-coding regions, plus portions of flanking sequences, of virusespossessing a ddI-resistant phenotype were cloned and sequenced by polymerase chain reaction (PCR)-basedmethods. We observed that several of these viruses possessed mutations at amino acid sites 184 (Met -- Val;ATG -- GTG) and 294 (Pro -- Ser; CCA -* TCA). These mutations were introduced in the pol gene of

infectious, cloned HXB2-D DNA by site-directed mutagenesis. Viral replication assays confirmed theimportance of site 184 with regard to resistance to ddl. The recombinant viruses thus generated displayed morethan fivefold-greater resistance to ddl than parental HXB2-D did. Moreover, more than fivefold-greaterresistance to ddC was also documented; however, the recombinant viruses continued to be inhibited byzidovudine (AZT). No resistance to ddI, ddC, or AZT was introduced by inclusion of mutation site 294 in thepol gene of HXB2-D. PCR analysis performed on viral samples obtained from patients receiving long-term ddltherapy confirmed the presence of mutation site 184 in five of seven cases tested. In three of these five positivecases, the wild-type codon was also detected, indicating that mixtures of viral quasispecies were apparentlypresent. Viruses possessing a ddI resistance phenotype were isolated from both subjects whose virusescontained only the mutated rather than wild-type codon at position 184 as well as from a third individual,whose viruses appeared to be mostly of the mutated variety.

Resistance to nucleoside antiretroviral compounds hasbeen reported to occur among isolates of human immunode-ficiency virus type 1 (HIV-1) from patients receiving pro-longed therapy with these drugs (10, 17, 27, 31). In addition,some investigators have reported that drug-resistant HIVvariants can be amplified in vitro by gradually increasing theconcentrations of these drugs in tissue culture medium (12,16). Generation of HIV resistance to nonnucleoside antago-nists of viral reverse transcriptase (RT) has also been dem-onstrated through in vitro selection procedures (23, 25).3'-Azido-3'-deoxythymidine (zidovudine; AZT) and otherantiretroviral drugs have been shown to impact positively onboth quality of life and survival of HIV-1-infected individu-als (8, 34-36). Although many patients become intolerant ofthese drugs after prolonged therapy and suffer clinical dete-rioration (7, 21, 22), the relationship between the appearanceof drug-resistant viruses and clinical status is still unclear.The RT of retroviruses is widely considered to be the

target of antiretroviral agents which can act as chain termi-nators of proviral DNA synthesis (11). The RT of HIV-1displays considerable infidelity during the replication of viralRNA (24, 26, 32). Unsurprisingly, a number of mutationshave been found in the pol genes of HIV-1 variants thatdisplay resistance to AZT (18), 2',3'-dideoxyinosine (ddI)(31), and 2',3'-dideoxycytidine (ddC) (10). It has been shownthat a mutation at site 74 in the RT-coding region was

* Corresponding author.

responsible for cross-resistance between ddl and ddC; stud-ies using site-directed mutagenesis showed that this muta-tion could also increase susceptibility to AZT against abackground of preexisting mutations that conferred initialresistance to this drug (31).A number of nucleoside analogs, including ddl, have been

studied in clinical trials to treat HIV-1-infected individualswho are intolerant of AZT, or in combination with AZT, todelay or prevent the emergence of drug resistance (5, 35).We now report on the cloning and sequencing of the com-plete RT-coding regions of several ddI-resistant variants ofHIV-IIIB generated in vitro and describe a novel mutation atsite 184 of the HIV-1 RT that is associated with resistance toddI and ddC but not AZT.

(This work was largely performed by Z. Gu in partialfulfillment of the requirements for a Ph.D. degree, Faculty ofGraduate Studies and Research, McGill University, Mon-treal, Quebec, Canada.)

MATERIALS AND METHODSCells and viruses. MT-4 cells were used to propagate both

wild-type and resistant variants of HIV-1. Briefly, the cellswere maintained in suspension culture in RPMI 1640 me-dium (GIBCO Laboratories, Mississauga, Ontario, Canada),supplemented with 10% fetal bovine serum (Flow Laborato-ries, Toronto, Ontario, Canada), 2mM L-glutamine, 100 U ofpenicillin per ml, and 100 ,ug of streptomycin per ml. Boththe HIV-IIIB laboratory strain of HIV-1 (kindly supplied by

7128

JOURNAL OF VIROLOGY, Dec. 1992, p. 7128-71350022-538X/92/127128-08$02.00/0Copyright C 1992, American Society for Microbiology

HIV RESISTANCE TO ddI 7129

prRTQO B

nt. 4256 B

pro I RT l intnt. 2515 prne

RT02

184U nt. 3117 nt. 3239-~~~~~~II

RT

nt. 2984 prirer pr184G or

184W1 84D

HXB2-D1 84FIG. 1. Construction of a molecular clone of HIV-1 carrying the

Val-184 mutation. As described in Materials and Methods,mpHIVRT was constructed by cloning a HincII-KpnI 1.3-kb frag-ment of the RT-coding region into M13mpl8. pHlVpol was con-structed by cloning the HincII-EcoRI 2.1-kb fragment of the HIV-1pol gene into pGEM-3Z. mpHIVRT184 was produced by site-directed mutagenesis to introduce the RT codon 184 mutation (Met-- Val) into mpHIVRT. After complete digestion with HincII and

partial digestion with KpnI, a 1.3-kb fragment of pHIVpol wasreplaced with the appropriate mutated fragment of mpHIVRT184 toyield pHIVpoll84. HXB2-D184 was constructed by replacing theBall 1.9-kb fragment of HXB2-D with the mutated Ball fragment ofpHIVpoll84. The boxes represent HIV-1 sequences. The dashedlines refer to vectors. Abbreviations: H, HincII; B, Ball; K, KpnI;E, EcoRI.

R. C. Gallo, National Institutes of Health, Bethesda, Md.)and a ddI-resistant variant that had been selected underconditions of tissue culture passage (12) were studied exten-sively. In addition, we employed the HXB2-D clone offull-length infectious DNA (9). A concentration of 40 ,uM ddI(Bristol-Myers Squibb, Wallingford, Conn.) was utilized forroutine propagation of viruses possessing a ddI resistancephenotype. This is approximately 10-fold the usual 50%inhibitory concentration (IC50) for ddI-sensitive viral strains.

In some experiments, viruses that had initially been grownon MT-4 cells were passaged onto phytohemagglutinin-prestimulated cord blood lymphocytes (CBL) obtained fromthe Department of Obstetrics of our hospital, as describedpreviously (28). For subsequent analysis, samples of CBL (5x 105 cells per ml) were pretreated with various concentra-tions of ddl ranging between 0 p,M and 200 ,uM for 4 h andwere then inoculated with CBL-grown HIV-1 at a multiplic-ity of infection of 1.0 (as determined by plaque assay onMT-4 cells) (12) in the concentration of ddl used for pretreat-ment. Fresh medium, including the appropriate concentra-tion of ddI, was added three times weekly, and freshphytohemagglutinin-prestimulated CBL (5 x 105 cells perml) were added at 2-day intervals. Similar studies wereperformed with ddC or AZT at concentrations of between 0and 10 p,M.

Oligonucleotides. The construction of HXB2-D used inthese studies is shown in Fig. 1. All of the oligonucleotidesused in this work were mapped as shown in Fig. 2 and listedin Table 1. The RT01-RT02 primer pair was used to amplifythe complete RT-coding region and to determine the direc-tion of a mutated BalI fragment which was inserted intowild-type HXB2-D. 184G, 184W, 184U, and 184D were

FIG. 2. Diagrammatic representation of PCR primer pairs. Idenotes the RT01-RT02 primer pair used to amplify the completeRT-coding region of HIV-1 and to determine the direction of themutated BalI fragment which replaced the appropriate fragment ofHXB2-D. B indicates the Ball fragment. II denotes the 184U-184W,184U-184G, and 184U-184D primer pairs used to discriminate wild-type from mutated codon Val-184 in the RT-coding region of HIV-1.Nucleotide (nt.) positions are indicated.

primers used to discriminate wild-type from mutated se-quences at the Val-184 codon in the RT-coding region ofHIV-1. 184mG and 294mT were used only in site-directedmutagenesis experiments (see below).

Cloning and sequencing of the RT region. Total cellularDNA was extracted from about 2 x 106 MT-4 cells that hadbeen infected with a ddI-resistant isolate of HIV-1 initiallyderived from HIV-IIIB by tissue culture passage selection(12). The cells were harvested and washed twice withphosphate-buffered saline. They were then lysed with 0.5 mlof 0.5% sodium dodecyl sulfate-Tris-EDTA (pH 8.0) buffer,digested with 20 ,ug of pronase per ml for 5 h at 37°C, andthen phenol extracted. A 1,742-bp segment containing thecomplete RT-coding sequence plus 34 bases of the 3' end ofthe protease-coding sequence and 28 bases of the 5' end ofthe integrase-coding sequence was amplified by polymerasechain reaction (PCR) (29) by using the RT01 and RT02primers for the 5' and 3' ends, respectively. About 0.5 ,ug (5pJ of extract) of extracted cellular DNAwas initially used foreach PCR mix (total of 100 pAl), which contained 50 mM KCl,10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 0.01% (wt/vol)gelatin, 2.5 U of Taq polymerase, a 0.2 mM concentration ofeach deoxynucleoside triphosphate (Pharmacia Fine Chem-icals, Montreal, Quebec, Canada), and 0.4 ,uM each RT01and RT02. Samples were overlaid with 100 ,ul of light mineraloil and heated to 94°C for 5 min before being subjected to 35thermal cycles of 1 min at 94°C for denaturing, 2 min at 55°C

TABLE 1. Oligonucleotides used in this study

Oligonucleotide Sequence (5' n 3') coordinates

RT01 GTAGAATTCTGTTGACTCAGATTGG 2515-2532aRTO2 GATAAGCTTGGGCCTTATCTATTCCAT 4256-4236a184U TACAATGTGCTTCCACAGGG 2984-3003184D CCATCCAAAGGAATGGAGG 3239-3221184G CCTACATACAAATCATCCAC 3117-3098184W TCCTACATACAAATCATCCAT 3118-3098184mG TCTATCAATACGTGGATGATTTG 3087-3109294mT CAGAAGTAATATCACTAACAGAAG 3417-3440

a Coordinates for RT01 and RT02 do not include 7 and 6 bases added to the5' ends of each of these two constructs to serve as recognition sites for EcoRIand HindIII, respectively.

VOL. 66, 1992

7130 GU ET AL.

for annealing, and 3 min at 72°C for extension. Amplificationwas completed by a final incubation at 72°C for 10 min in aPerkin-Elmer Cetus thermal cycler. After purification fromagarose gels by electroelution, the amplified segments weredigested with HindIlI and EcoRI (Pharmacia Fine Chemi-cals), whose recognition sites were built into the 5' and 3'ends of the PCR primers, and ligated with digested M13mp19(30). Eschenchia coli TG1 cells were transfected with therecombinants as described previously (30) and screened bydigesting double-stranded DNA with restriction endonu-cleases. Single-stranded DNA prepared from recombinantM13 clones was used for nucleotide sequencing (30). Theentire length of the RT was sequenced by using a set of fiveoligomers and a Taq Track Sequencing kit (Promega Inc.,Madison, Wis.).

In some experiments, cloning and sequencing studies weresimilarly performed on 2 x 10 CBL that had been infectedwith wild-type or drug-resistant variants of HIV-1.PCR detection of mutation sites. As stated above, Fig. 2

shows a map of primer pairs used in PCR analyses in thisstudy. We used a mutant primer (184G), a wild-type primer(184W), an upstream primer (184U), and a downstreamprimer (184D) to discriminate wild-type from mutatedcodons for RT site Val-184 (primer pair consisting of 184Uand either 184G or 184W) and to detect HIV-1 DNA as apositive control in our PCR (184U-184D primer pair). The184G-184U or 184W-184U primer pair produces a 134-basefragment, and the 184U-184D primer pair yields a 256-basefragment. We selected seven clinical isolates obtained frompatients who had received at least 6 months of ddI therapyfor further analysis. Some of these individuals were alsoshown to possess ddI-resistant variants of HIV-1 in theircirculation. Wild-type HXB2-D was used as a control toindicate specific reactivity to the 184G primer, and a blankcontrol was used to ensure noncontamination of samples inevery experiment. We also employed DNA from uninfectedMT-4 cells to confirm that our primers did not nonspecifi-cally amplify cellular DNA.

Thus, MT-4 cells were infected with clinical isolates, andtotal cellular DNA was extracted. Concentrations of PCRprimers utilized in each reaction mix were as follows: 184U,1.2 ,uM; 184G and 184W, 0.65 ,uM; and 184D, 0.5 ,uM. Thesamples were subjected to 35 thermal cycles of 40 s at 94°Cfor denaturing, 20 s at 53°C for annealing, and 30 s at 72°C forextension. The PCR products were separated through 2%agarose. Other procedures were performed as describedabove.

In addition, we performed PCR analysis for detection ofthe previously described site 41 mutation (Met -- Leu)conferring resistance to AZT (14). This was accomplishedthrough the use of two primers, RT01 (described above) and41C (which possessed the sequence 5'GAAATT7T'CCCTTCCT'TTCCAG3' [HXB2-D coordinates 2691 to 2669]) toyield a 177-base fragment. As a control for detection ofHIV-1 DNA, we used a primer pair consisting of RT01 and41D (5'CTCTGAAATCTACTAAT1TTTCTCC3', coordi-nates 2783 to 2760) to yield a 269-base segment.

Site-directed mutagenesis. The construction of mutatedHXB2-D is diagrammed in Fig. 1. The codon 184 (Met >Val; ATG -- GTG) mutation of RT was introduced intowild-type HXB2-D by site-directed mutagenesis as de-scribed elsewhere (15, 30). Briefly, a 1.3-kb fragment ofHXB2-D, derived by HincII and KpnI digestion, was clonedinto bacteriophage M13mpl8 (mpHIVRT). E. coli TG1 cellswere transfected with this recombinant. E. coli CJ236 cellswere infected in medium containing 0.25 pLg of uridine

(Boehringer Mannheim Biochemica) per ml with 108 recom-binant phage particles to prepare uracil-containing single-stranded DNA (U-DNA). The synthetic mutated oligonucle-otide 184mG was phosphorylated with T4 polynucleotidekinase and then annealed to U-DNA template at a 184mG/U-DNA ratio of 22:1 by heating the mixture to 80°C andthen slowly cooling it to room temperature. After a secondstrand was synthesized and ligated with T4 DNA polymerase(1 U) and T4 DNA ligase (1 U), the double-stranded DNAwas digested with uracil-DNA glycosylase (1 U) and thentreated with alkali to remove any remaining U-DNA strands.The remaining single-stranded DNA was used for transfect-ing E. coli TG1 cells. Single-stranded DNA was isolatedfrom plaques, and potentially relevant mutations (termedmpHIVRT184) were analyzed by dideoxynucleotide se-quencing.pHIVpol was constructed by cloning the 2.1-kb fragment

generated by restriction digestions (HincIl and EcoRI) of theHXB2-D pol gene into pGEM-3Z. After complete digestionwith HincIl and partial digestion with Ijpnl, the 1.3-kbfragment of pHIVpol thus generated was replaced by themutated fragment from mpHIVRT184 to produce pHIVpoll84.A 1.9-kb BalI fragment from pHIVpoll84 was substituted

for the appropriate fragment of HXB2-D to produce HXB2-D184. E. coli TG1 cells were transformed with HXB2-D184and screened by endonuclease digestion. The orientation ofthe cloned fragment was determined by PCR with the RT01and RT02 primers shown in Fig. 2. The Val-184 (ATAGTA) mutation was confirmed by DNA sequencing.A recombinant virus, HXB2D-294, containing a CCA -

TCA substitution at codon 294 (resulting in replacement ofproline by serine) was constructed in the same manner asdescribed above for HXB2D-184, except that oligonucleo-tide 294mT was used instead of 184mG. In some experi-ments, a recombinant that contained the substitutions men-tioned above at positions 184 and 294 (HXB2-D184+294)was constructed. For purposes of such construction, anoligonucleotide/template ratio of 40:1 was used instead of the22:1 ratio that was used for single-codon substitutions.

Transfection and preparation of virus stock. MT-4 cells (5x 106) were harvested, washed twice with cold RPMI 1640medium containing 10% fetal bovine serum (FBS), penicillin(100 U/ml), and streptomycin (100 ,ug/ml), resuspended in0.8 ml of cold supplemented RPMI 1640 medium, and mixedwith 20 ,ug of DNA. After standing on ice for 10 min, thecells were transfected by electroporation at 250 V and 960,uF (6) with a Bio-Rad Gene Pulser. The electroporated cellswere kept on ice for 10 min and then washed with coldsupplemented RPMI 1640 medium. The transfected cellswere resuspended in 10 ml of RPMI 1640 medium containing20% FBS plus antibiotics and incubated at 37°C under 5%CO2. To generate virus stock, fresh MT-4 cells were addedto expand the culture after cytopathic effects were observed,and the cultures were maintained for 2 to 3 days. Culturesupernatants were harvested and frozen in aliquots at -700Cprior to use.Assays of viral replication and drug sensitivity. Assays of

HIV-1 susceptibility to drugs, RT assays, and indirect im-munofluorescence assays for detection of p24 were per-formed as previously described (3, 28). Antigen capture testsfor occasional and confirmatory determination of p24 levelsin culture fluids were carried out with kits purchased for thispurpose from Abbott Laboratories (North Chicago, Ill.).Isolation of virus from patients was performed with CBL inthe absence of drug, as described elsewhere (12). After

J. VIROL.

HIV RESISTANCE TO ddI 7131

amplification of virus by further replication in CBL in theabsence of drug, studies examined the ability of such isolatesto replicate when the drug was present at a variety ofconcentrations. This permitted the calculation of IC50 deter-minations on the basis of RT levels in culture fluids, aspreviously described (12).

RESULTS

To determine the oligonucleotide mutations responsiblefor HIV-1 resistance to ddl, we have cloned and sequencedthe complete RT-coding region of a number of ddI-resistantvariants of HIV-IIIB selected in vitro (12). The results ofsuch studies showed that several such viruses apparentlycontained mutations in the RT-coding region at codons 184(Met -- Val; ATG -* GTG) and 294 (Pro -) Ser; CCA --

TCA) in comparison with parental drug-sensitive HIV-IIIB.To confirm the biological significance of these mutations,

we used site-directed mutagenesis to introduce the valine-encoding triplet GTG into position 184 and the serine-encoding TCA into position 294 of the HIV-1 RT gene of theinfectious molecular clone HXB2-D, yielding the mutantclones HXB2-D184 and HXB2-D294. In addition, we gener-ated a construct, HXB2-D184+294, containing both of thesechanges. The susceptibilities of HXB2-D184, HXB2-D294,HXB2-D184+294, and wild-type HXB2-D to ddI and otherdrugs were assessed by viral replication in MT-4 cells; IC50swere calculated from levels of RT activity in culture fluids.Figure 3 presents data demonstrating the abilities of parentalHXB2-D and HXB2-D184 to replicate in the presence ofdifferent concentrations of ddI, ddC, and AZT. It is apparentthat the presence of the GTG substitution at position 184caused a significant diminution in susceptibility to both ddl(Fig. 3a) and ddC (Fig. 3b). However, little or no resistanceto AZT was seen when concentrations approaching the usualIC50s of this drug, e.g. 0.05 R,M, were used (Fig. 3c).These findings are summarized in Table 2, which shows

that the Val-184 mutation caused more than a fivefolddecrease in susceptibility to ddI and ddC but did not affectsensitivity to AZT. However, the substitution of serine forproline at position 294 did not affect drug susceptibility(Table 2). In addition, the combination of both of thesemutations, yielding HXB2-D184+294, did not give rise to alevel of drug resistance higher than that seen with themutation at codon 184 alone. Similar observations wereobtained on the basis of assays for detection of p24 antigenin culture fluids and indirect immunofluorescence assays forp24 antigen in infected cells (data not shown).To distinguish the wild-type codon from the mutated

codon at RT position 184, we employed specific primer pairsand PCR as described in Materials and Methods. Viruseswhich were isolated from the peripheral blood mononuclearcells of HIV-infected patients receiving prolonged (>6months) ddI therapy were grown in MT-4 cells in thecontinuous presence of ddI (40 ,M) for 3 days, after whichcellular DNA was extracted for purposes of PCR amplifica-tion. Similar analyses were performed as controls withddl-resistant viruses generated in tissue culture throughselection pressure as well as with wild-type control virusesderived from patients or grown in culture in the absence ofdrug. In these studies, we amplified all samples in whichviral DNA should have been present to determine thepresence of a 256-bp fragment corresponding to a conservedregion in pol as a positive control. We also investigated thepresence of either wild-type or mutated sequences at codon184, as described in Materials and Methods.

a 300000

:-.E

CU

tI-.tco

b

E

Ft.t

G100000

1000

ddl concentration (aiM)

1 0

ddC concentraton (riM)C 300000

E

0.k

I- 1000C

0.01 0.1

AZT concentration (toM)FIG. 3. Susceptibility of HXB2-D and HXB2-D184 to ddI (a),

ddC (b), and AZT (c). MT-4 cells were infected with these viruses inthe presence of different concentrations of drugs. Viral susceptibil-ity was assessed by measurement of RT activity in samples ofclarified culture supernatants. Symbols: O, HXB2-D; *, HXB2-D184.

The results in Fig. 4 show that, as expected, no viral DNAcould be detected in uninfected MT-4 cells (lanes A and B).In the case of wild-type HXB2-D DNA (lanes C and D), boththe 256-bp fragment and DNA with a wild-type codon 184were detected. However, DNA with the mutated form ofcodon 184 was not present. In contrast, each of two ddI-resistant viral samples, selected in tissue culture, i.e., HIV-IIIB-ddI-1 and HIV-IIIB-ddI-2, contained the mutated formof codon 184 (lanes F and H) but not the wild-type form(lanes E and G). The recombinant HXB2-D184 likewise did

I

)o

0 . .-I . ,a ...

VOL. 66, 1992

7132 GU ET AL.

TABLE 2. Antiretroviral sensitivity of HIV-1 variants

IC50 (,M) in MT-4 cellsaVirus tested

ddI ddC AZT

HXB2-D 7.8 + 0.5 0.44 + 0.075 0.01 ± 0.0002HXB2-D184 39.8 ± 2.7 2.4 + 0.2 0.018 + 0.0007HXB2-D294 6.9 ± 0.6 0.46 + 0.05 0.013 ± 0.0004HXB2-D184+294 37.5 ± 2.1 2.6 ± 0.3 0.014 ± 0.0003

a IC50s were obtained from plots of amounts of RT activity detected inculture fluids as a function of antiretroviral drug concentration. Each value isthe average of three separate determinations (- the standard deviation).

not possess wild-type DNA at codon 184 (lane I) but didpossess the mutated form (lane J). Two clinical samplesobtained from patients who had undergone ddI therapy for 6months were shown to have mixtures of viral quasispecies,in the sense that both wild-type sequences (lanes K and M)and mutated forms (lanes L and N) were present. Theseindividuals, designated patients 1566 and 1567, had receivedAZT therapy for approximately 9 months before beingswitched to ddl (500 mg per day) because of AZT intoler-ance. Neither patient was symptomatic for HIV-associateddisease. In each of these cases, the wild-type forms wereapparently predominant. All samples which contained eitherthe mutated or wild-type sequence at codon 184 also con-tained the 256-bp positive control.These analyses were extended to include a broader sample

of clinical isolates, as shown in Fig. 5. Virus isolated frompatient 1393, who had not been treated with antiviral che-motherapy, contained the wild-type form of codon 184 (laneB) but not the mutated form (lane H). The 256-bp segmentcorresponding to a conserved region of the pol gene wasdetected as a positive control in all samples tested. Patient1051, who had been treated with ddl for 12 months, pos-sessed virus in which only the mutated form of codon 184was present (lane I); no wild-type material at codon 184could be detected (lane C). Similar findings were obtained

ABCDE 'T(3)1 I JKLhoNO

/256 brp -

1 34 1)ii-S(cocon 184)

FIG. 4. Detection by PCR of a wild-type or mutated codon atposition 184 of the HIV-1 RT-coding region. Lanes A and B containuninfected MT-4 cell controls; lanes C and D contain HXB2-DDNA; lanes E and F contain DNA from a ddl-resistant virus,generated by in vitro selection from HIV-IIIB and designatedHIV-IIIB-ddl-1; lanes G and H contain DNA from a second HIV-IIIB-derived ddl-resistant virus, HIV-IIIB-ddI-2; lanes I and J con-tain DNA from HXB2-D184; lanes K and L contain DNA from cellsinfected with virus from patient 1566, who had been treated with ddIfor 6 months; and lanes M and N contain DNA from cells infectedwith virus from patient 1567, who had been treated with ddl for 6months. Lane 0 contains a DNA base pair ladder. In this study,primer 184W (wild type) was used to amplify samples in lanes A, C,E, G, I, K, and M. Primer 184G (to detect the site 184 mutation) wasused to amplify material in lanes B, D, F, H, J, L, and N. Materialin all lanes was also amplified with the 184U-184D primer pair toensure the presence of HIV-1 DNA as a positive control.

_34 bp) (cocl()n 1 84)

FIG. 5. Detection by PCR of a wild-type or mutated codon atposition 184 of the HIV-1 coding region in the peripheral bloodmononuclear cells of patients treated with ddI or other drugs. LaneA contains a DNA base pair ladder; lanes B and H contain DNAfrom cells infected with virus of patient 1393, who was not treatedwith antiviral chemotherapy; lanes C and I contain DNA from cellsinfected with virus from patient 1051, who had been treated with ddIfor 12 months; lanes D and J contain DNA from cells harboring virusfrom patient 1552, who had been treated with ddl for 14 months;lanes E and K contain DNA from cells infected with virus frompatient 1559, who had been treated with ddl for 11 months; lanes Fand L contain DNA from cells infected with virus from patient 1266,who had been treated with ddI for 4 months; and lanes G and Mcontain DNA from cells harboring virus of patient 1241, who hadbeen treated with ddI for 5 months. Primer 184W for detection ofwild-type codon 184 was used to amplify DNA in lanes A to G.Primer 184G for detection of the site 184 mutation was used in lanesH to M. Material in all lanes was also amplified with the 184U-184Dprimer pair to ensure the presence of HIV-1 DNA as a positivecontrol.

with virus from patient 1552, who had been treated with ddIfor 14 months, with the mutated form only being present(lanes D and J). Patient 1559, who had been treated with ddIfor 11 months, yielded virus that was apparently heteroge-neous for mutation 184, with both wild-type and mutatedforms being present (lanes E and K); however, the mutatedform appeared to be predominant. In contrast, virusesderived from subjects 1266 and 1241, who had been treatedwith ddl for 4 and 5 months, respectively, contained only thewild-type form of codon 184 (lanes F and G, respectively),with no apparent presence of mutated sequences (lanes Land M, respectively).Table 3 summarizes the therapeutic regimens received by

the patients studied as well as the detection of resistance-conferring mutation sites as determined by PCR analysis. Allsubjects had been monitored off protocol, and had receivedAZT for various times before being switched to ddl therapyfor reasons of AZT intolerance. Most of these individualswere still asymptomatic, although a few had progressed tomild forms of lymphadenopathy. Pretreatment isolates fromthese individuals were not available. Nonetheless, the re-sults in Table 3 show that viruses isolated from patientsreceiving prolonged ddI therapy were resistant to this com-pound in some cases, as determined in tissue culture.Although pretreatment isolates from the same individualswere not available for comparison, the ranges of IC50sobtained for most of the isolates are consistent with previousreports describing a drug-sensitive phenotype (12, 31). Incontrast, patients 1051, 1552, and 1559 possessed viruses forwhich the IC50s of ddI were 4 to 10 times higher than thosecommonly seen. On this basis and because ddI-resistantisolates generally require IC50s approximately fivefoldhigher than those required by paired pretreatment isolates(31), these viruses were considered to be resistant to ddI.

J. VIROL.

HIV RESISTANCE TO ddl 7133

TABLE 3. Summary of available information for patients monitored in this study

Genotype at:

Patient Health status History of antiviral Codon 184 Codon 41 ICd50 of

Wild type Mutated Wild type Mutated

1566 Asymptomatic AZT (9 mo), ddI (6 mo) + + + - 4.21567 Asymptomatic AZT (8 mo), ddI (6 mo) + + + - 2.91393 Asymptomatic None - + - 2.41051 Mild lymphadenopathy AZT (6 mo), ddI (12 mo) - + - + 23.61552 Asymptomatic AZT (5 mo), ddI (14 mo) - + + - 16.21559 Mild lymphadenopathy AZT (8 mo), ddI (11 mo) + + + - 12.81266 Asymptomatic AZT (4 mo), ddI (4 mo) + - - + 3.01241 Lymphadenopathy AZT (12 mo), ddI (5 mo) + - + - 2.1

Coincidentally, viruses from two of these subjects (1051 and1552) displayed only the site 184 mutation and no evidence ofwild-type DNA at this codon, while in the case of patient1559, a mixture of viral quasispecies was apparently present.

In order to possibly shed light on the relationship betweenthe mutation at site 41, which confers resistance to AZT(since all patients had been treated with this drug), and themutation at site 184, we also performed PCR analysis todetect the former codon change, as described in Materialsand Methods. Only two individuals studied (patients 1051and 1266) showed evidence of the site 41 mutation (Table 3).

Finally, we wished to determine whether the site 184mutation would persist after replication of ddI-resistantHIV-1 in CBL. Accordingly, both wild-type HXB2-D andHXB2-D184 were grown for 2 weeks in CBL in the absenceof drug pressure, as described in Materials and Methods.Cloning and sequencing showed that HXB2-D continued topossess a wild-type ATG (Met) codon at position 184 afterthis time, while HXB2-D retained the GTG (Val) mutationdescribed above. IC50 determinations performed on the basisof RT levels in culture fluids, after replication in CBL at avariety of drug concentrations, showed that the site 184mutation continued to account for resistance to both ddl andddC (Table 4).

DISCUSSION

The development of HIV-1 resistance to antiretroviraldrugs may become an important problem in the therapy ofHIV-1-infected individuals (10, 20, 30). This may be attrib-utable to the infidelity of HIV-1 RT, which leads to muta-genesis during reverse transcription; hence, mutations con-ferring resistance to antiviral drugs might easily emergeunder conditions of drug pressure. Indeed, multiple muta-tions in the RT-coding regions of AZT-resistant variants ofHIV-1 have been demonstrated. The biological relevance ofthese mutations has been confirmed by site-directed muta-genesis, and combinations of certain mutations have been

TABLE 4. Antiretroviral sensitivity of viruses grown in CBL

IC50 (LM) in CBLUVirus tested

ddI ddC AZT

HXB2-D 5.4 + 0.4 0.52 + 0.06 0.02 ± 0.0004HXB2-D184 42.6 ± 3.8 3.8 ± 0.4 0.02 ± 0.0006

a IC50s were obtained from plots of RT activity in culture fluids as afunction of drug concentration. Each value is the average of three determi-nations (+ the standard deviation).

shown to yield higher degrees of resistance than singlemutations (18). In addition, a single-amino-acid substitutionhas been reported for resistance to ddI and cross-resistanceto ddC (10, 31). Interestingly, the presence of this Val-74mutation in the RT of ddI-resistant HIV-1, isolated frompatients who had previously received prolonged AZT ther-apy and who possessed AZT-resistant virus in their blood,caused an apparent increase in sensitivity to AZT (31).

This paper reports a novel mutation at codon 184 (MetVal) associated with HIV-1 resistance to ddI and cross-resistance to ddC. This work grew out of cloning andsequencing studies performed with the complete RT-codingregion of a HIV-IIIB derivative previously shown to beresistant to ddI. We identified two mutations at codons 184(ATG -- GTG) and 294 (CCA -- TCA) as potentiallyresponsible for the observed resistance. Amino acid 184 islocated in a highly conserved region of the RT (1, 2, 13, 20,33), in which the sequence consisting of amino acids 183 to186 has been reported to be crucial to RT enzymatic function(4, 19, 20). Single mutations at position 184, introduced bysite-directed mutagenesis, caused reductions in RT activityof 80% in the case of Met -- Tyr and 95% in the case of Met

Leu (4, 19).In our studies, a change from the hydrophobic amino acid

methionine at this site to the aliphatic amino acid valine ledto resistance to both ddI and ddC. This cross-resistance maybe due to the fact that both ddI and ddC possess 2',3'-dideoxy moieties. We have also found the Val-184 substitu-tion in variants of HIV-1 that have been selected in vitro forresistance to ddC (not shown). Our failure to demonstratecross-resistance to AZT may be due to the 3'-azido moiety ofthe AZT molecule. It is consistent as well that the Val-184substitution is not present in variants of HIV-1 selected forresistance to AZT (12). As mentioned above, a mutation atVal-74 is also associated with cross-resistance to both ddIand ddC (31). Interestingly, a mutation at site 181 that causesa Tyr-to-Cys alteration in the RT primary structure isapparently responsible for the generation of resistance tononnucleoside inhibitors of viral RT that act by noncompet-itive inhibition (22, 25). The fact that we were unable toconfirm the biological significance of the mutation at site 294indicates the necessity to perform site-directed mutagenesisin this work.Our previous report on a ddI-resistant derivative of HIV-

I"B generated through in vitro selection showed that thisvirus required an IC50 over 20-fold higher than that requiredby the parental virus and was not resistant to ddC (12). Thecurrent study on site-directed mutagenesis has revealed onlya fivefold difference in IC50 as well as extensive cross-

VOL. 66, 1992

7134 GU ET AL.

resistance to ddC. Further studies to address this issue areunder way. It is significant that HXB2-D184, when passagedonto CBL, continued to possess the codon 184 mutation andto maintain resistance to both ddI and ddC. This indicatesthat this mutation can persist in cells other than those of theMT-4 line, which we used in our in vitro selection protocol.

It is important to ask whether the demonstration of HIVdrug resistance, as revealed through in vitro tissue cultureselection procedures, is clinically relevant. For this reason,we used the PCR to amplify the Val-184 mutation site fromclinical isolates obtained from patients undergoing prolongedddl therapy. Five of seven such isolates possessed theVal-184 mutation site, while two did not.The two individuals whose isolates lacked the site 184

mutation had received ddl therapy for less than 6 months. Ofthe five individuals positive for mutation 184, three alsopossessed the wild-type codon for this site, indicating thatmixtures of viruses were present. This finding is consistentwith previous observations of mixtures of viral quasispeciesthat showed heterogeneity for mutation sites relevant toAZT. Interestingly, our study showed that those individualsin whom virus with the site 184 mutation was present to thegreatest extent were also the donors from whom ddI-resis-tant viruses were most easily isolated. This indicates that thesite 184 mutation is likely to occur in at least some patientswho develop ddI-resistant variants of HIV-1.The patients studied had received AZT for various periods

before being switched to ddl. Therefore, we wanted toascertain the relationship between mutation site 184 and theAZT resistance-conferring mutation at site 41 in terms ofpossible synergy with regard to resistance to ddM. No appar-ent relationship was shown to exist. Nonetheless, similarstudies of the presence or absence of the other sites knownto confer resistance to AZT are in order, as is work onpossible synergy between sites 184 and 74 in the develop-ment of resistance to ddl.

ACKNOWLEDGMENTS

We are grateful to F. Busschaert for preparation of the manuscriptand to E. Faust for its critical review. We thank N. Sonenberg forassistance with site-directed mutagenesis and R. Beaulieu and J.Montaner for supply of certain of the blood samples used in thiswork.

This research was supported by grants from Health and WelfareCanada and from the Medical Research Council of Canada.

REFERENCES1. Argos, P. 1988. A sequence motif in many polymerases. Nucleic

Acids Res. 16:9909-9916.2. Barber, A., A. Hizi, J. V. Maizel, Jr., and S. H. Hughes. 1990.

HIV-1 reverse transcriptase: structure predictions for the poly-merase domain. AIDS Res. Hum. Retroviruses 6:1061-1072.

3. Boulerice, F., S. Bour, R. Geleziunas, A. Lvovich, and M. A.Wainberg. 1990. High frequency of isolation of defective humanimmunodeficiency virus type 1 and heterogeneity of viral geneexpression in clones of infected U-937 cells. J. Virol. 64:1745-1755.

4. Boyer, P. L., A. L. Ferris, and S. H. Hughes. 1992. Cassettemutagenesis of the reverse transcriptase of human immunode-ficiency virus type 1. J. Virol. 66:1031-1039.

5. Broder, S. 1990. Pharmacodynamics of 2',3'-dideoxycytidine:an inhibitor of human immunodeficiency virus. Am. J. Med.88:5B-2S-5B-7S.

6. Chu, G., H. Hayakawa, and P. Berg. 1987. Electroporation forthe efficient transfection of mammalian cells with DNA. NucleicAcids Res. 15:1311-1326.

7. Fischl, M. A., D. D. Richman, D. M. Causey, M. H. Grieco, Y.Bryson, D. Mildvan, 0. L. Laskin, J. E. Groopman, P. A.

Volberding, R. T. Schooley, G. G. Jackson, D. T. Durack, J. C.Andrews, S. Nusinoff-Lehrman, D. W. Barry, and the AZTCollaborative Working Group. 1989. Prolonged zidovudine ther-apy in patients with AIDS and advanced AIDS-related complex.JAMA 262:2405-2410.

8. Fischl, M. A., D. D. Richman, M. H. Grieco, M. S. Gottlieb,P. A. Volberding, 0. L. Laskin, J. M. Leedom, J. E. Groopman,D. Mildvan, R. T. Schooley, G. G. Jackson, D. T. Durack, D.King, and the AZT Collaborative Working Group. 1987. Theefficiency of azidothymidine (AZT) in the treatment of patientswith AIDS and AIDS-related complex: a double-blind placebo-controlled trial. N. Engl. J. Med. 317:185-191.

9. Fisher, A. G., E. Collalti, L. Ratner, R. C. Gallo, and F.Wong-Staal. 1985. A molecular clone of HTLV-III with biolog-ical activity. Nature (London) 316:262-265.

10. Fitzgibbon, J. E., R. M. Howell, C. A. Haberzettl, S. J. Sperber,D. J. Gocke, and D. T. Dubin. 1992. Human immunodeficiencyvirus type 1 pol gene mutations which cause decreased suscep-tibility to 2',3'-dideoxycytidine. Antimicrob. Agents Chemo-ther. 36:153-157.

11. Furman, P. A., J. A. Fyfe, M. H. St. Clair, K. Weinhold, J. L.Rideout, G. A. Freeman, S. N. Lehrman, D. P. Bolognesi, S.Broder, H. Mitsuya, and D. W. Barry. 1986. Phosphorylation of3'-azido-3'-deoxythymidine and selective interaction of the 5'-triphosphate with human immunodeficiency virus reverse tran-scriptase. Proc. NatI. Acad. Sci. USA 83:8333-8337.

12. Gao, Q., Z. Gu, M. A. Parniak, X. Li, and M. A. Wainberg.1992. In vitro selection of variants of human immunodeficiencyvirus type 1 resistant to 3'-azido-3'-deoxythymidine and 2',3'-dideoxyinosine. J. Virol. 66:12-19.

13. Johnson, M. S., M. A. McClure, D.-F. Feng, J. Gray, and R. F.Doolittle. 1986. Computer analysis of retroviral pol genes:assignment of enzymatic functions to specific sequences andhomologies with nonviral enzymes. Proc. Natl. Acad. Sci. USA83:7648-7652.

14. Kellam, P., C. A. B. Boucher, and B. A. Larder. 1992. Fifthmutation in human immunodeficiency virus type 1 reversetranscriptase contributes to the development of high-level resis-tance to zidovudine. Proc. Natl. Acad. Sci. USA 89:1934-1938.

15. Kunkel, T. A. 1985. Rapid and efficient site-specific mutagenesiswithout phenotypic selection. Proc. Natl. Acad. Sci. USA82:488-492.

16. Larder, B. A., K. E. Coates, and S. D. Kemp. 1991. Zidovudine-resistant human immunodeficiency virus selected by passage incell culture. J. Virol. 65:5232-5236.

17. Larder, B. A., G. Darby, and D. D. Richman. 1989. HIV withreduced sensitivity to zidovudine (AZT) isolated during pro-longed therapy. Science 243:1731-1734.

18. Larder, B. A., and S. D. Kemp. 1989. Multiple mutations inHIV-1 reverse transcriptase confer high-level resistance tozidovudine (AZT). Science 246:1155-1158.

19. Larder, B. A., S. D. Kemp, and D. J. M. Purifoy. 1989.Infectious potential of human immunodeficiency virus type 1reverse transcriptase mutants with altered inhibitor sensitivity.Proc. Natl. Acad. Sci. USA 86:4803-4807.

20. Larder, B. A., D. J. M. Purifoy, K. L. Powell, and G. Darby.1987. Site-specific mutagenesis of AIDS virus reverse tran-scriptase. Nature (London) 327:716-717.

21. Merigan, T. C. 1990. Summary of the symposium. Rev. Infect.Dis. 12:S576.

22. Moore, R. D., J. Hildago, B. W. Sugland, and R. E. Chaisson.1991. Zidovudine and the natural history of the acquired immu-nodeficiency syndrome. N. Engl. J. Med. 324:1412-1416.

23. Nunberg, J. H., W. A. Schleif, E. J. Boots, J. A. O'Brien, J. C.Quintero, J. M. Hoffman, E. A. Emini, and M. E. Goldman.1991. Viral resistance to human immunodeficiency virus type1-specific pyridinone reverse transcriptase inhibitors. J. Virol.65:4887-4892.

24. Preston, B. D., B. J. Poiesz, and L. A. Loeb. 1988. Fidelity ofHIV-1 reverse transcriptase. Science 242:1168-1171.

25. Richman, D., C.-K. Shih, I. Lowy, J. Rose, P. Prodanovich, S.Goff, and J. Griffin. 1991. Human immunodeficiency virus type1 mutants resistant to nonnucleoside inhibitors of reverse tran-

J. VIROL.

HIV RESISTANCE TO ddl 7135

scriptase arise in tissue culture. Proc. Natl. Acad. Sci. USA88:11241-11245.

26. Roberts, J. D., K. Bebenek, and T. A. Kunkel. 1988. Theaccuracy of reverse transcriptase from HIV-1. Science 242:1171-1173.

27. Rooke, R, M. Tremblay, H. Soudeyns, L. DeStephano, X.-J.Yao, M. Fanning, J. S. G. Montaner, M. O'Shaughnessy, K.Gelmon, C. Tsoukas, J. Ruedy, and M. A. Wainberg. 1989.Isolation of drug-resistant variants of HIV-1 from patients onlong-term zidovudine (AZT) therapy. AIDS 3:411-415.

28. Rooke, R., M. Tremblay, and M. A. Wainberg. 1990. AZT(zidovudine) may act postintegrationally to inhibit generation ofHIV-1 progeny virus in chronically infected cells. Virology176:205-215.

29. Saiki, R. K., D. H. Gelfand, S. Stoffel, S. J. Scharf, R. Higuchi,G. T. Horn, K. B. Muilis, and H. A. Erlich. 1988. Primer-directed enzymatic amplification of DNA with a thermostableDNA polymerase. Science 239:487-491.

30. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

31. St. Clair, M. H., J. L. Martin, G. Tudor-Williams, M. C. Bach,C. L. Vavro, D. M. King, P. Kellam, S. D. Kemp, and B. A.Larder. 1991. Resistance to ddI and sensitivity to AZT inducedby a mutation in HIV-1 reverse transcriptase. Science 253:1557-1559.

32. Takeuchi, Y., T. Nagumo, and H. Hoshino. 1988. Low fidelity ofcell-free DNA synthesis by reverse transcriptase of humanimmunodeficiency virus. J. Virol. 62:3900-3902.

33. Toh, H., H. Hayashida, and T. Miyata. 1983. Sequence homol-ogy between retroviral reverse transcriptase and putative poly-merase of hepatitis B virus and cauliflower mosaic virus. Nature(London) 305:827-829.

34. Yarchoan, R., R. W. Klecker, K. J. Weinhold, P. D. Markham,H. K. Lyerly, D. T. Durack, E. Gelmann, S. N. Lehrman, R. M.Blum, D. W. Barry, G. M. Shearer, M. A. Fischl, H. Mitsuya,R. C. Gallo, J. M. Collins, D. P. Bolognesi, C. H. Myers, and S.Broder. 1986. Administration of 3'-azido-3'-deoxythymidine, aninhibitor of HTLV-III/LAV replication, to patients with AIDSor AIDS-related complex. Lancet i:575-580.

35. Yarchoan, R., H. Mitsuya, R. V. Thomas, J. M. Pluda, N. R.Hartman, C.-F. Perno, K. S. Marczyk, J.-P. Allain, D. G. Johns,and S. Broder. 1989. In vivo activity against HIV and favorabletoxicity profile of 2',3'-dideoxyinosine. Science 245:412-415.

36. Yarchoan, R., C. F. Perno, R. V. Thomas, R. W. Klecker, J.-P.Allain, R. J. Wills, N. McAtee, M. A. Fischl, R. Dubinsky, M. C.McNeely, H. Mitsuya, J. M. Pluda, T. J. Lawley, M. Leuther, B.Safai, J. M. Collins, C. E. Myers, and S. Broder. 1988. Phase Istudies of 2',3'-dideoxycytidine in severe human immunodefi-ciency virus infection as a single agent and alternating withzidovudine (AZT). Lancet i:76-81.

VOL. 66, 1992