Low accumulation of L90M in protease from subtype FHIV-1 with resistance to protease inhibitors is caused by the L89M polymorphism

(1)

M A J O R A R T I C L E

Low Accumulation of L90M in Protease

from Subtype F HIV-1 with Resistance to Protease

Inhibitors Is Caused by the L89M Polymorphism

Alexandre Calazans,1_{Rodrigo Brindeiro,}1_{Patrı´cia Brindeiro,}1_{Hugo Verli,}3_{Moˆnica B. Arruda,}1

Luis Manuel Felipe Gonzalez,1_{Jorge A. Guimara˜es,}3_{Ricardo Sobhie Diaz,}4_{Octavio A. C. Antunes,}2

and Amilcar Tanuri1

1

Department of Genetics, Biology Institute, and2

Department of Inorganic Chemistry, Chemistry Institute, Federal University of Rio de Janeiro, Rio de Janeiro,3

Center of Biotechnology, Federal University of Rio Grande do Sul, Porto Alegre, and4

Retrovirology Laboratory, Federal University of Sa˜o Paulo, Sa˜o Paulo, Brazil

Background. This work evaluates the role of subtype F human immunodeficiency virus type 1 (HIV-1) protease (PR) substitutions L89M and L90M in viral replication and resistance to PR inhibitors (PIs).

Methods. Subtype B and F PR genes were subjected to site-directed mutagenesis, to create and reverse the methionine at positions 89 and 90. Viruses were re-created in cell culture, and their replicative capacity was assessed by fitness assay. Generated viruses were also phenotyped for PI resistance.

Results. The subtype F clone (89M90L) showed a replicative capacity comparable to that of the PI-susceptible subtype B clone (89L90L) and was more fit than the L89M mutated subtype B clone (89M90L). Both 89M90M subtype B and F clones presented the lowest fitnesssvalues. The L89M mutation impacted phenotypic resistance to all PIs in half of the subtype F isolates but not in the subtype B isolates. Subtype F isolates presented a phenotypic profile similar to that of subtype B isolates when the M89L mutation was introduced.

Conclusion. The L89M mutation in subtype F viruses is a high genetic barrier to the accumulation of the L90M resistance mutation and can function as a resistance mutation, depending on the presence of other poly-morphisms in the subtype F PR backbone.

HIV-1 protease (PR) inhibitors (PIs) are crucial com-ponents in combined antiretroviral therapy (or high-ly active antiretroviral therapy [HAART]) for patients with AIDS [1–4]. Seven different PIs have been ap-proved by the United States Food and Drug Admin-istration (FDA) and are available for clinical use: sa-quinavir (SQV), ritonavir (RTV), indinavir (IDV), nel-finavir (NFV), amprenavir (APV), lopinavir (LPV), and, more recently, atazanavir (ATV) [5–7]. Despite the great potency of HAART that includes at least 1 PI, resistant strains can emerge from the viral population circulating

Received 21 May 2004; accepted 10 December 2004; electronically published 28 April 2005.

Financial support: Brazilian Program of STD/AIDS; Coordenaçaõ de Aper-feiçoamento de Pessoal de Ensino Superior.

Reprints or correspondence: Dr. Amilcar Tanuri, Laborato´rio de Virologia Mo-lecular, Departamento de Gene´tica, Universidade Federal do Rio de Janeiro, CCS– Bloco A–Cidade Universita´ria–Ilha do Funda˜o, 21944-970 Rio de Janeiro, RJ, Brazil ([email protected]).

The Journal of Infectious Diseases 2005; 191:1961–70

in the infected individual and cause therapy failure [8– 11]. These resistant strains harbor the enzymes PR and reverse transcriptase (RT) with amino acid substitutions that confer resistance to the PIs and RT inhibitors (RTIs), respectively [1, 10, 12–16].

Drug-resistance mutations have been characterized extensively for subtype B strains of HIV-1 [10, 17]. The primary resistance mutations appear to modify the amino acids that interact directly with the PIs inside of the catalytic pocket of the enzyme [18–20], whereas the secondary, or compensatory, mutations occur mostly outside of the active cleft of the enzyme, restoring its plasticity and catalytic capacity, which is generally di-minished by the presence of primary mutation(s) [12, 18, 21, 22].

Genotyping and phenotyping assays play an impor-tant role in the characterization of the mutational pat-terns responsible for drug resistance. Different inter-pretation algorithms for drug-resistance genotyping have been established on the basis of both phenotyping as-says and clinical correlations between therapy failure

at Escola Paulista Medicina on October 16, 2015

http://jid.oxfordjournals.org/

(2)

1962 • JID 2005:191 (1 June) • Calazans et al.

Figure 1. Amino acid sequences of protease (PR) from viral clones. A, Consensus sequences of PR from HIV-1 subtypes B and F (sequences

obtained from Los Alamos HIV sequence databank), illustrating the molecular signatures of subtype F PRs.B,The polymorphic regions of amino acids 89 and 90, generated from PR inhibitor–susceptible subtype B (pNL4-3) and F (F13) clones by the mutagenized clones.C,Amino acid sequences from 3 treatment-naive isolates of subtype F (F18, F17, and F22) with their respective mutants at amino acid 89. The amino acids at positions 89 and 90 are depicted to the right of the clone name.

and viral strains carrying mutations [23–27]. The algorithms are nevertheless based on observations made using subtype B strains of HIV-1, and there is little information on non–subtype B strains [6, 28–30]. The usefulness of these algorithms to the study of isolates of non–subtype B strains, as well as to the characterization of new, subtype-specific PI-resistance muta-tions, still needs to be addressed. Substitutions frequently found in PRs of non–subtype B isolates, so-called “molecular signa-tures,” may also interfere with the PI-resistance phenotype pro-duced by other mutations that are well characterized in subtype B isolates.

Subtype F is the most prevalent (prevalence, 15%) non-B HIV-1 subtype found in Brazil [31–33]. Recently, we genotyp-ically and phenotypgenotyp-ically characterized Brazilian clinical isolates of HIV-1 of subtypes F and B infecting adults [34] and children [35] in whom antiretroviral therapy failed. Surprisingly, we observed discordant genotypic profiles among subtype F iso-lates that were phenotypically resistant to NFV. In total, 7 (53.8%) of 13 NFV-resistant subtype B isolates carried the L90M substitution, which was characterized as conferring in vitro and in vivo primary resistance to SQV [16, 36] and NFV [15]. Another 3 isolates (23%) carried the D30N mutation, which is also related to NFV resistance [37]. Of the 5 NFV-resistant subtype F isolates found, however, only 1 (20%) car-ried the L90M substitution (none carcar-ried the D30N mutation). All of the subtype F isolates carried a polymorphism at neigh-boring amino acid position 89 (L89M) identical to that found at position 90, which confers resistance to NFV. No subtype B isolate harbored this latter polymorphism. In the present study,

the role of the L89M substitution in preventing the accumu-lation of L90M in subtype F isolates is demonstrated, by use of a series of recombinant viruses carrying different mutations in fitness experiments.

MATERIALS AND METHODS

Cells and viruses. MT-4 cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum and an-tibiotics (Invitrogen). The HIV-1 subtype B proviral clone pNL4-3 [38] was used as the reference virus. The cDNA of HIV-1 viral isolate TVGG13 [35] was used as the representa-tive of HIV-1 subtype F (F13; GenBank accession number AY145824). This isolate carries 11 amino acid differences in comparison with the consensus subtype B PRs (figure 1), 7 of which are the molecular signatures of subtype F PRs (F1 con-sensus). In addition, 3 other subtype F clinical isolates from PI-experienced patients were used: F17 (TVGG17), F18 (TVGG18), and F22 (TVGG22) (GenBank accession numbers AY145825, AY145826, and AY145832, respectively) (figure 1C). The re-combination vector pHXB-2-Dprowas used for the phenotyp-ing assay [39].

Polymerase chain reaction (PCR), cloning, and site-directed mutagenesis. A 651-bp PR-coding sequence was amplified by PCR from both F13 cDNA and the pNL4-3 plasmid, by use of the primers RVP5 (5′_{-gggaagatctggccttcctacaaggg-3}′_{) and RVP3} (5′_{-ggcaaatactggagtattgtatgg-3}′_{), with a standard protocol [39].} The amplification products were cloned into the pCR4-TOPO vector as instructed by the TOPO TA Cloning Kit manual

(3)

vitrogen). The cloned PR fragments were subjected to site-directed mutagenesis, by use of the Quick Change Mutagenesis Kit (Stratagene), to create or reverse the 89M polymorphism, as well as to introduce the L90M mutation (primers are shown in table A1 of the Appendix, which appears only in the elec-tronic edition of the Journal). The other 3 PR genes from subtype F clinical isolates (F17, F18, and F22) were cloned and mutagenized at amino acid position 89, to revert the methi-onine to leucine, by use of primers 89L/90L F and R. The L89I mutation was also introduced into clones F13 and Bwt, which carry either leucine or methionine at PR position 90 (table A1 of the Appendix, which appears only in the electronic edition of the Journal). The putative mutagenized clones were con-firmed through DNA sequencing in both directions, by use of the BigDye Terminator Sequencing Kit and an ABI 3100 au-tomated sequencer (Applied Biosystems).

Virus PR genotyping. Viral RNA was extracted with Trizol Reagent (Invitrogen), and the PR gene was amplified and se-quenced as described elsewhere [33]. PR sequences were edited with DNAStar software and were submitted to the HIV-ge-notyping algorithm of the HIV RT and Protease Sequence Da-tabase (available at: http://hivdb.stanford.edu) [23].

Phenotyping assay. The determination of HIV-1 phenotypic resistance to FDA-approved PR inhibitors (except ATV) was per-formed using recombinant virus assay technology, as described elsewhere [40], by use of PCR fragments of the PR region of the clones generated by site-directed mutagenesis (figure 1). These fragments were cotransfected into CD4+

T lymphocytes (MT-4) with the linearized pHXB-2-Dpro plasmid [39]. Homologous recombination led to generation of chimerical virus containing the different PR sequences. The virus containing the Bwt 89L90L PR was used to obtain the reference 50% effective concentration (EC50) values during the phenotyping assay. The susceptibility of

chimerical virus to all PIs was determined, in 2 independent assays with duplicates, by an MT-4-3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazoliumbromide (MTT)–based cell viability as-say, as described elsewhere [40].

Growth competition experiments (fitness assay). Experi-ments were performed as described elsewhere [41, 42], in the absence of drugs in MT-4 cells. The assay consists of the co-infection of cultures with known amounts of the virus to be tested, together with a reference clone. These cocultures were subjected to repeated bottleneck passages [43]. The ratio of the 2 viruses was quantified by a genotypic assay as follows: the proportion of the competing variant with respect to the ref-erence strain (Rn) is divided by its proportion in the initial

mixture (Ro), and its value (Rn/Ro) is plotted against the

com-petition passage, to derive the fitness vector [43, 44]. Each competition assay was performed in duplicate by infecting MT-4 cells with mixtures of the viruses to be tested (Bwt- and F13-derived mutagenesis clones) with Bwt 89L90L (the reference

virus), at an initial ratio of 1:1. For each competition, 105

MT-4 cells were infected with 104

50% cell culture infectious doses (CCID50) of the virus tested and of the reference virus (MOI

for each virus, 0.05). The culture supernatant was harvested when a cytophatic effect was evident (∼3–7 days after infection). Fresh MT-4 cells were then infected with 50mL of this super-natant. A total of 5 passages were performed, as described be-low, and aliquots were removed from the cultures at days 0, 7, 10, 13, 16, and 19 and were centrifuged; the supernatants were stored at⫺80C for further quantitative analysis of genotypes.

We have developed a quantitative assay to determine the ratio of the different competing genotypes found in each culture passage by use of the ABI Prism Snapshot Multiplex Kit (Ap-plied Biosystems). This kit is based on the dideoxy single-base extension of an unlabeled oligonucleotide (or a set of oligo-nucleotides; see table A1 of the Appendix, which appears only in the electronic edition of theJournal). Each primer binds to a complementary template in the presence of the 4 ddNTPs labeled with different fluorophores andTaqDNA polymerase. The polymerase extends the primer by 1 nt, adding a single labeled ddNTP to its 3′_{end. The sample reactions are} electro-phoresed in an ABI Prism 310 Genetic Analyzer, and data are analyzed with GeneScan Analysis Software (version 3.1; ABI). This methodology was validated using an artificial mixture of wild-type and mutant cloned material, and a linear regression curve was obtained for the different mixture ratios, with a coefficient of correlation (R2

)10.98.

Segregation of fitness assay PRs and analysis of recom-bination. Supernatant harvested on day 19 of the fitness assay (viral competitions Bwt 89L90L⫻F13 89L90L and Bwt 89L90L

⫻F13 89M90M) were submitted to viral RNA extraction, fol-lowed by reverse-transcription PCR and cloning in a TOPO TA Cloning kit (Invitrogen). Colony PCR for the PR gene was performed and submitted toBclI endonuclease digestion. Re-striction-negative clones were submitted to sequencing, fol-lowed by sequence alignment and analysis of recombination, by use of the RIP 1.9 beta test (http://www.hiv.lanl.gov/content/ hiv-db/RIPPER/RIP.html).

Clinical samples. Analysis of clinical samples was per-formed with internal databases from the Laboratory of Retro-virology of the Federal University of Sa˜o Paulo and the Mo-lecular Virology Laboratory of the Federal University of Rio de Janeiro. These databases include clinical information on treated patients from different regions of Brazil, including treatment regimens, CD4+

T lymphocyte and viral load counts, and re-sistance genotyping profiles. For the viral isolates from infected individuals without PI exposure (treatment-naive individuals), we used the database of these 2 laboratories from BResNet [31]. The data on viral isolates from HIV-1–infected individuals with exposure to NFV and/or SQV was also obtained from Caride et al. [34] and Brindeiro et al. [35]. All data were collected

(4)

Table 1. Distribution of amino acid mutations at positions 89 and 90 in

treatment-naive and treated HIV-1 subtype B and F isolates.

Mutation

Treatment naive Treated

Subtype Ba Subtype Fa Pb Subtype B Subtype F Pb

89L 1997 49 50 4

89M 24 69 !.0001 4 18 !.0001

90L 2021 118 41 21

90M 0 0 !_.0001 ₁₃ ₁ _.04

NOTE. Data are no. of isolates found with each mutation.

a

Calculation based on data from the Stanford HIV sequence database (available at: http:// hivdb.stanford.edu/).

b

P(Fisher’s exact test),ap0.05, for2⫻2contingency tables.

under the supervision of the Universities’ respective institu-tional review boards.

Statistical analysis. All statistical analyses were performed using Analyse-it (version 1.62) for the Microsoft Excel statistics package. Sigmaplot software (version 8.0) was used to perform linear regressions for the fitness assay and nonlinear regressions (Hill’s dose-response curve) for phenotyping. We established the significant values of resistance by use of a nonparametric Mann-Whitney U test, for comparison between the quadru-plicates of EC50for each mutagenized virus and the

quadru-plicates of EC50for the Bwt L89L90 drug-susceptible clone.

RESULTS

Genotypic frequencies of the L89M and L90M polymorphisms in subtype B and F clinical samples. Seventy-six previously genotyped clinical isolates of virus from NFV- or SQV-treated individuals (54 from subtype B and 22 from subtype F) were analyzed for the presence of the L89M and L90M polymor-phisms (table 1). Since L89M is a characteristic molecular sig-nature of subtype F PRs, we also analyzed its frequency (as well as the frequency of L90M) in genotyped clinical samples from treatment-naive infected individuals (2021 from subtype B and 118 from subtype F), obtained from the Stanford HIV Drug Resistance Database (http://hivdb.stanford.edu/) [23].

To avoid misinterpretation of the L89M frequencies in differ-ent subtypes due to its presence as a natural polymorphism of subtype F, the frequencies were also analyzed in separate groups of treatment-naive and PI (NFV and SQV)–treated individuals. A significant difference was noted between the subtype B and F samples from treated individuals (8% and 82% with L89M, in subtype B and F samples, respectively;P!.0001, Fisher’s exact test), and a significant accumulation of L90M in subtype B sam-ples was noted, with a low frequency of this mutation in subtype F isolates (subtype B vs. subtype F,Pp.04). There was a sig-nificant increase in the prevalence of L89M when treatment-naive subtype F isolates were compared with their treated counterparts (P_p.03). Of note, the 89M90M PR genotype could not be found in our study, for any subtype. Some other studies (R.S.D.,

un-published data) have found this genotype in the subtype F back-bone, at a frequency of!1%. Nevertheless, the subtype F isolates from treated individuals described here as carrying the L90M re-sistance mutation have also accumulated the M89I polymorphism. Relative replicative fitness of mutated viruses. After gen-erating the viral stocks in cell culture by cotransfection of PR amplicons with a pHXB-2-Dpro full-length clone, we per-formed replicative competition experiments (fitness assay) be-tween the PI-susceptible Bwt 89L90L viral clone and the other clones herein described, in the absence of any PI (figure 2). The Bwt 89M90L and Bwt 89M90M viruses replicate with a lower capacity than does the reference virus, whereas the Bwt 89L90M mutant had minimally increased replicative fitness, compared with that of Bwt 89L90L (figure 2A). The F13 89M90L and F13 89L90M clones had a replicative capacity similar to that of the Bwt reference virus, but the latter reached a positive fitness value (s) slightly higher than that of its F 89M90L counterpart (figure 2B). Both the F13 89L90L and F13 89M90M mutants showed a significant decrease in their rep-licative capacity, compared with that of the Bwt reference virus (sp⫺0.072and⫺0.1622, respectively). Surprisingly, after

bot-tleneck passages 1 (F13 89M90M) and 2 (F13 89L90L), these clones recovered their replicative capacity and reached positive fitness values (s_p0.0459and 0.0499, for F13 89L90L and F13 89M90M, respectively).

To assess possible genetic causes of the fitness shift of viruses F13 89L90L and F13 89M90M during the replicative compe-tition experiment (fitness assay), we isolated and sequenced clones of subtype F PR genes amplified from the mixture of subtype B and F viruses at the last bottleneck event of the fitness assay (day 19 virus passage). After PCR amplification of cDNA obtained from viral RNA of culture supernatant, PR gene amplicons were cloned directly into a pCR4.3 vector. Bac-terial colonies carrying segregated PR genes were submitted to colony PCR, followed byBclI enzyme digestion screening for subtype F PRs. ABclI cut site can be found at nucleotide po-sition 177 of the subtype B—but not subtype F—PR genes. Eight of 22 clones and 12 of 24 clones analyzed for the F13

(5)

Figure 2. Fitness vectors and corresponding fitness values of HIV-1 Bwt(A)and F13(B)89L90L, 89L90M, 89M90L, and 89M90M and of 89I90L and 89I90M(C).In each competing passage, viral proportion (Rn) was compared with the ratio of the 2 viruses in the initial mixtures (Ro). The value

in each passage was used to derive the fitness vectors by linear regression. Details of all procedures involved are given in Materials and Methods. For all lineages, the initial 1:1 viral competition rate was used in the calculation of the vectors. Vectors were drawn with data from 3 competing passages. Slopes of fitness vectors (fitness values,s) for the HIV-1 clones are indicated for each clone in the plotting areas. The pNL4-3 Bwt L89L90 clone was used as a reference, under the assumption of ansvalue of 0.

(6)

Figure 3. Recombination analysis (RIP 1.9 beta; Los Alamos HIV

se-quence database) in 3 sese-quences of 2 subtype F clones(AandB)segregated from fitness assay Bwt 89L90L⫻F13 89L90L and 1 subtype F clone(C)

segregated from fitness assay Bwt 89L90L⫻F13 89M90M. The cloning segregation was performed at bottleneck point 5 (day 19). The background sequences for comparison were Bwt 89L90L–pNL4-3 (subtype B) and F13 89M90L (subtype F). RIP 1.9 beta analysis was performed with a window size of 30 nt and a 90% threshold for statistical significance.

89L90L and F13 89M90M fitness assays, respectively, did not digest withBclI and were subjected to DNA sequencing. Sub-type F viruses isolated from the F13 89L90L fitness assay showed 2 different patterns of recombination with the standard Bwt 89L90L virus present in culture (figure 3Aand 3B), as revealed by the RIP 1.9 beta sequence analysis of recombination. For the F13 89M90M fitness assay, only 1 pattern of recombination between the subtype B and F viral clones originally present in the culture was found (figure 3C).

The L89I mutation negatively impacted the replicative ca-pacity of subtype B and F viruses carrying this mutation. This effect was more drastic in subtype B viruses (Bwt 89I90L and Bwt 89I90M) than in subtype F viruses (F13 89I90L and F13 89I90M), and it was also independent of the presence or ab-sence of the L90M mutation (figure 2C). Both subtype F viruses carrying the L89I substitution had a replicative capacity coef-ficient (s) similar to that found for virus F 89M90M.

Phenotyping of clones from subtypes B and F. Recom-binant viruses carrying the mutant PR genes were generated in MT-4 cells, and all of the constructs were phenotyped in quad-ruplicate for PI susceptibility by MT-4 cell–MTT-based cell viability assay. The resulting EC50values are shown in table 2.

To facilitate the comparison between the different clones from subtypes B and F, we chose to establish the statistical signifi-cance of resistance on the basis of the quadruplicate sampling comparing the EC50in the Bwt L89L90 clone (subtype B

wild-type clone) with those in the other clones tested. The L89M mutation conferred significant (5.6-, 4.7-, and 6.2-fold) in-creases in EC50 for IDV, NFV, and RTV, respectively, on F13

89M90L virus, compared with Bwt 89L90L virus. These val-ues are comparable to the resistance valval-ues obtained with Bwt 89L90M virus: 2.8-, 3.8-, and 5.5-fold resistance, respectively. In contrast, the Bwt 89M90L clone behaved in a way that was phenotypically similar to the susceptible Bwt 89L90L virus.

The F13 89L90M clone showed a 6.4-, 6.5-, 7.5-, 23-, 5.1-, and 3.9-fold resistance increase for SQV, IDV, NFV, RTV, APV, and LPV, respectively. These values represent a 3.0-, 3.7-, 1.9-, 4.3-, 1.8-, and 2.6-fold increase over its counterpart, Bwt 89L90M, for the same drugs. The Bwt 89M90M and F13 89M90M clones showed divergent phenotypic behavior. There was a significant increase in resistance in the Bwt 89M90M clone, compared with that in the Bwt 89L90L clone (table 2), with the exception of resistance to LPV. In contrast, there was a significant decrease in resistance to RTV, NFV, and APV in the Fwt 89M90M clone, compared with that in the Fwt 89M90L clone. (P!.05, Mann-WhitneyU test).

Virus clones of both subtypes carrying an isoleucine at amino acid position 89 were also phenotypically characterized (table 2). This mutation was not able to confer any significant level of resistance to the PIs analyzed. The coexistence of an L90M

substitution with the 89I residue did not modify this pattern of PI susceptibility, for either subtype analyzed.

To explore the unexpected phenotype coded by the L89M polymorphism in subtype F PR, we cloned the PR genes of 3 other clinical isolates, and the 89M signature (wild-type subtype F signature) was reversed to 89L by site-directed mutagenesis. All 89L reversed clones were phenotyped for PI resistance (table

(7)

Table 2. Results of protease inhibitor phenotying assay for Bwt and F13 clones.

Virus, clone

Fold resistance,meanSDa

SQV IDV RTV NFV APV LPV

Bwt

89L90L 1.000.127 1.000.625 1.000.124 1.000.314 1.000.192 1.000.159 89M90L 1.000.329 2.960.095 0.770.135 1.100.069 1.470.136 0.770.240 89L90M 2.120.480 2.800.871 3.820.819 5.460.956 2.900.173 1.480.239 89M90M 1.670.399 2.040.020 2.710.303 4.440.410 2.570.195 1.320.150 89I90L 0.710.085 0.900.076 0.770.136 0.930.067 0.910.075 NDb 89I90M 1.310.108 1.600.070 2.560.152 1.810.210 1.230.067 NDb F13

89L90L 0.510.182 1.230.331 0.330.068 0.820.154 0.740.143 0.780.104 89M90L 2.060.237 5.600.150 4.741.387 6.200.556 4.470.337 3.440.607 89L90M 6.420.967 6.510.430 7.451.901 23.394.605 5.131.880 3.870.344 89M90M 0.850.256 2.421.101 2.480.212 2.800.828 2.080.528 0.900.111 89I90L 0.680.008 0.770.008 0.960.080 0.800.105 0.700.071 NDb 89I90M 1.590.165 1.330.238 3.910.481 2.470.104 1.000.053 NDb F18

89L90L 1.330.126 0.600.215 2.910.681 4.780.885 1.510.278 1.940.300 89M90L 2.030.128 2.290.151 3.710.832 5.840.714 1.210.644 2.020.070 F17

89L90L 0.950.054 0.890.057 0.450.019 2.170.199 1.840.017 1.590.021 89M90L 0.820.060 0.630.049 1.740.167 1.380.119 1.220.140 1.090.023 F22

89L90L 0.790.026 0.490.020 1.320.060 1.000.106 0.44018 0.930.688 89M90L 0.380.032 0.360.036 0.590.056 0.380.038 0.330.024 0.510.171

a

APV, amprenavir; IDV, indinivir; LPV, lopinavir; ND, not determined; NFV, nelfinavir; RTV, ritonavir; SQV, saquinavir. Fold resistance is calculated as (50% effective concentration [EC50] of sample virus)/(EC50of reference virus Bwt 89L90L). Data represent the results

of quadruplicate experiments. Cut-off values of fold resistance with clinical significance, according to Harrigan et al. [45], are as follows: SQV, 2.5; IDV, 3.0; RTV, 3.5; NFV, 4.0; APV, 2.5; LPV, 4.0. Fold resistance values higher than the cut-off value are highlighted in bold.

b

Phenotyping values for 89I clones were not determined for LPV.

2), together with their wild-type 89M counterparts. Isolates F13 and F18 showed a significant decrease in EC50values for IDV,

RTV, NFV, APV, and IDV alone when 89M was reversed to 89L. In contrast, isolates F17 and F22 showed no significant difference in EC50values when 89M was reversed.

DISCUSSION

HIV-1 subtype F is found throughout South America [18], in Romania, and in Central Africa. Since the beginning of the 1990s, surveys conducted in Brazil have identified subtype F as the main non–subtype B isolate in that country. Subtype F strains, together with their recombinant forms with subtype B, account for 15% of HIV-1 infections in Brazil. The subtype F consensus sequence differs from the US and Brazilian subtype B consensus sequence at 8 positions (I15V, E35D, M36I, R41K, R57K, Q61N, L63P, and L89M) [33].

Subtype F isolates are infrequently found to carry the L90M NFV and SQV–resistance mutation in HIV-1–infected individ-uals in whom therapy has failed [34, 35, 46]. The relative rep-licative capacity of viruses carrying subtype F PRs with M89L and L90M substitutions appears to be higher than those with the PI-susceptible prototypic subtype F PR genotype 89M90L,

as shown here by fitness assay. However, the accumulation of L90M appears to be constrained to the change in the 89M sub-type F polymorphism. This hypothesis is supported by the poor-est relative replicative capacity being found for F13 89M90M, compared with that in the F13 89L90M and F13 89M90L virus clones (sp⫺0.1622, vs. 0.0165 and⫺0.0018, respectively; see

figure 2Bfor details). This effect was more noticeable in subtype F clones, since thesvalues obtained for the subtype B clone counterparts did not show this difference (figure 2A). Fur-thermore, a detailed search of the PI-susceptibility data in the Stanford HIV sequence database [23] reveals the low clinical prevalence (0.36%, or 6/1682 samples analyzed) of viruses car-rying the 89M90M polymorphisms. None of these isolates be-longed to subtype F, and, of 109 subtype F isolates deposited, only 1 carried the L90M mutation together with a M89I poly-morphism (isolate TVGG27 [35]).

Since the F13 89L90L and F13 89M90M viruses had low replicative fitness, the M89I mutation could represent the best in vivo alternative for the transient mechanism of generation of L90M PRs from the original 89M90L subtype F PRs [8, 34, 35, 47]. Nevertheless, the fitness assay showed both F13 89I90M and F13 89I90L viruses with no advantage over their

(8)

part F13 89M90M in their fitness s values (all surrounding

⫺0.16). Moreover, there was no fold increase, in these 2 89I

F13 clones, in resistance to any PI analyzed. Together, the fitness assay and PI-resistance data indicate the existence of L90M and M89I mutations occurring in subtype F PRs as independent genetic events. The emergence of an M89I mutation in the backbone of a subtype F 89M90M virus is favored by the low genetic barrier of a single GrA transition, which changes the ATG codon of methionine to the ATA codon of isoleucine, against a less probable ArT or ArC transversion, which could generate a highly adapted subtype F 89L90M virus. The com-mon GrA hypermutation that occurs in HIV sequences and is caused by the activity of cellular enzymes such as APOBEC 3G [48–50] is another possible mechanism that could favor the ATGrATA transition at codon 89 in subtype F PRs, generating the 89I mutation, although at a low frequency.

A recovery of relative replicative capacity for the F13 89M90M and 89L90L clones was seen after bottleneck passages 2 and 3, respectively, in the fitness assay. The sequences obtained from different segregated clones of PRs from these subtype F virus subpopulations revealed a recombination event between 2 viruses of different subtypes (B and F). These recombinant events have probably yielded more fit viruses, explaining the fitness recovery by consequent alteration of the PR structure and its residue interactions. The loss of some PR molecular signatures of the subtype F backbone through genetic recom-bination could be involved with the fitness recovery, but this hypothesis needs to be further analyzed.

The Bwt 89L90M virus was more fit than the PI-susceptible Bwt 89L90L, in contrast to the findings of Martinez-Picado et al. [44]. They found a low relative replicative fitness of subtype B viruses carrying L90M, compared with a wild-type virus (10% less fit than wild type), although their theoretical mathematical model used for fitness calculation may not adequately quantify relative growth rates of different HIV-1 mutants, as stated in their article.

The L89M signature of F13 in its original sequence could confer a significant level of resistance to IDV, as expected [13, 51], but also, surprisingly, to RTV and NFV. The reversion of the L89M mutation in subtype F isolates was able to confer a significant decrease in EC50 to all 6 PIs analyzed. The same

phenotypic behavior was not observed for the subtype B clone carrying L89M, which remained susceptible to PIs, with the exception of IDV. The role of the L89M polymorphism in IDV resistance in the subtype B backbone has already been dem-onstrated by others [13, 49]. In contrast, the LrM change in PR amino acid position 89 was previously described to have no correlation with PI regimen failure [16], although only clin-ical samples from subtype B were analyzed.

Interestingly, this 89M-related resistance is observed in only half of the subtype F isolates (F13 and F18) analyzed in our

work, which indicates that other polymorphisms are involved in this phenotypic behavior. Samples F17 and F22 did not present significant levels of resistance to IDV, RTV, or NFV, unlike their subtype F counterparts F13 and F18. This fact can be partially explained by other polymorphisms found to differ from those common subtype F signatures, such as the I72T mutation, present in both F17 and F22, or the V15I and K70R mutations (reversing the F molecular signature), present in F17 and F22, respectively. The question of whether the differences between the sequences of clones F17 and F22 and the sequences of clones F13 and F18 account for the phenotypic divergence found still remains to be addressed.

Several studies [28–30, 34–47] have already demonstrated different adaptive strategies of resistance-mutation accumula-tion between subtype B and non–subtype B viruses, although the amino acid substitutions described for non–subtype B vi-ruses and conferring resistance to antiretrovirals were basically the same as those described for subtype B viruses. The L89M polymorphism in the subtype F PR backbone is mostly related to the maintenance of viral fitness, and its presence as a mo-lecular signature in the subtype F PR backbone confers a higher genetic barrier to the accumulation of the L90M resistance mutation. The L89M polymorphism is the most prevalent sig-nature among treatment-naive non–subtype B isolates and ac-counts for 98% of all PR sequences in the Stanford HIV se-quence database. The same genetic barrier posed by L89M could play a role in other non–subtype B isolates and needs to be addressed in specific studies.

Acknowledgment

We thank Charles Boucher, who kindly provided the clone

pHXB-2-Dpro.

References

1. Carpenter C, Fischi M, Hammer S, et al. Antiretroviral therapy for HIV infection. JAMA1996; 276:146–54.

2. Gulick RM, Mellors JW, Havlir D, et al. Simultaneous vs sequential initiation of therapy with indinavir, zidovudine and lamivudine for HIV-1 infection. JAMA1998; 280:35–41.

3. Hammer SM, Squires KE, Hughes MD, et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immu-nodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. N Engl J Med1997; 337:725–33.

4. Ruiz L, Nijhuis M, Boucher TC, et al. Efficacy of adding indinavir to previous reverse transcriptase nucleoside analogues in relation to ge-notypic and phege-notypic resistance development in advanced HIV-1 infected patients. J Acquir Immune Defic Syndr Hum Retrovirol1998; 19:19–28.

5. Haas DW, Zala C, Schrader S, et al.Therapy with atazanavir plus sa-quinavir in patients failing highly active antiretroviral therapy: a ran-domized comparative pilot trial. AIDS2003; 17:1339–49.

6. Piliero PJ. Atazanavir: a novel HIV-1 protease inhibitor. Expert Opin Investig Drugs2002; 11:1295–301.

7. Sanne I, Piliero P, Squires K, Thiry A, Schnittman S. AI424-007 Clinical Trial Group. Results of a phase 2 clinical trial at 48 weeks (AI424-007):

(9)

a dose-ranging, safety, and efficacy comparative trial of atazanavir at three doses in combination with didanosine and stavudine in antiret-roviral-naive subjects. J Acquir Immune Defic Syndr2003; 32:18–29. 8. Condra HJ, Schleif AW, Blahy OM, et al. In vivo emergence of

HIV-1 variants resistant to multiple protease inhibitors. Nature1995; 374: 569–71.

9. Hirsch MS, Brun-Ve´zinet F, Da´quila RT, et al. Antiretroviral resistance testing in adult HIV-1 infection: recommendations of an international AIDS Society-USA panel. JAMA2000; 283:2417–26.

10. Iversen A, Shafer R, Wehrly K, et al. Multidrug-resistant human im-munodeficiency virus type 1 strains resulting from combination an-tiretroviral therapy. J Virol1996; 70:1086–90.

11. Schmit JC, Cogniaux J, Hermans P, et al. Multiple drug resistance to nucleoside and nonnucleoside reverse transcriptase inhibitors in an efficiently replicating human immunodeficiency virus type 1 patient strain. J Infect Dis1996; 174:962–8.

12. Boden D, Markowitz M. Resistance to human immunodeficiency virus type 1 protease inhibitors. Antimicrob Agents Chemother 1998; 42: 2775–83.

13. Condra HJ, Holder JD, Schleif AW, et al. Genetic correlates of in vivo viral resistance to indinavir, a human immunodeficiency virus type 1 protease inhibitor. J Virol1996; 70:8270–6.

14. Molla A, Korneyeva M, Gao Q, et al. Ordered accumulation of mu-tations in HIV protease confers resistance to ritonavir. Nat Med1996;2: 760–6.

15. Patick AK, Duran M, Cao DY, et al. Genotypic and phenotypic char-acterization of human immunodeficiency virus type 1 variants isolated from patients treated with the protease inhibitor nelfinavir. Antimicrob Agents Chemother1998; 42:2637–44.

16. Shafer RW, Hsu P, Patick AK, Craig C, Brendel V. Identification of biased amino acid substitution patterns in human immunodeficiency virus type 1 isolates from patients treated with protease inhibitors. J Virol1999; 73:6197–202.

17. Wu TD, Schiffer CA, Gonzales MJ, et al. Mutation patterns and structural correlates in human immunodeficiency virus type 1 protease following different protease inhibitor treatments. J Virol2003; 77:4836–47. 18. Muzammil S, Ross P, Freire E. A major role for a set of non-active

site mutations in the development of HIV-1 protease drug resistance. Biochemistry2003; 42:631–8.

19. Rose RB, Craik CS, Stroud RM. Domain flexibility in retroviral pro-teases: structural implications for drug resistant mutations. Biochem-istry1998; 37:2607–21.

20. Scott WRP, Schiffer CA. Curling of flap tips in HIV-1 protease as a mechanism for substrate entry and tolerance of drug resistance. Struc-ture2000; 8:1259–65.

21. Kantor R, Fessel WJ, Zolopa AR, et al. Evolution of primary protease inhibitor resistance mutations during protease inhibitor salvage ther-apy. Antimicrob Agents Chemother2002; 46:1086–92.

22. Nijhuis M, Schuurman R, de Jong D, et al. Increased fitness of drug resistant HIV-1 protease as a result of acquisition of compensatory mutations during suboptimal therapy. AIDS1999; 13:2349–59. 23. Kantor R, Machekano R, Gonzales MJ, Dupnik K, Schapiro JM, Shafer

RW. Human immunodeficiency virus reverse transcriptase and protease sequence database: an expanded data model integrating natural lan-guage text and sequence analysis programs. Nucleic Acids Res2001; 29:296–9.

24. Kijak GH, Rubio AE, Pampuro SE, et al. Discrepant results in the interpretation of HIV-1 drug-resistance genotypic data among widely used algorithms. HIV Med2003; 4:72–8.

25. Schmidt B, Walter H, Moschik B, et al. Simple algorithm derived from a geno-/phenotypic database to predict HIV-1 protease inhibitor re-sistance. AIDS2000; 14:1731–8.

26. Schmidt B, Walter H, Zeitler N, Korn K. Genotypic drug resistance interpretation systems—the cutting edge of antiretroviral therapy. AIDS Rev2002; 4:148–56.

27. Van Laethem K, De Luca A, Antinori A, Cingolani A, Ban˜adme AM. A genotypic drug resistance interpretation algorithm that significantly

predicts therapy response in HIV-infected patients. Antivir Ther2002; 7:123–9.

28. Fleury H, Recordon-Pinson P, Caumont A, et al. HIV type 1 diversity in France, 1999–2001: molecular characterization of non-B HIV type 1 subtypes and potential impact on susceptibility to antiretroviral drugs. AIDS Res Hum Retroviruses2003; 19:41–7.

29. Kantor R, Katzenstein D. Polymorphism in HIV-1 non-subtype B pro-tease and reverse transcriptase and its potential impact on drug sus-ceptibility and drug resistance evolution. AIDS Rev2003; 5:25–35. 30. Pieniazek D, Rayfield M, Hu DJ, et al. Protease sequences from

HIV-1 group M subtypes A-H reveal distinct amino acid mutation patterns associated with protease resistance in protease inhibitor-naive individ-uals worldwide. HIV Variant Working Group. AIDS2000; 14:1489–95. 31. Brindeiro RM, Diaz RS, Sabino EC, et al. Brazilian network for HIV drug resistance surveillance (HIV-BResNet): a survey of chronically in-fected individuals. AIDS2003; 17:1063–9.

32. Morgado MG, Sabino EC, Shpaer EG, et al. V3 region polymorphism in HIV-1 from Brazil: prevalence of subtype B strains divergent from North American/European prototype and detection of subtype F. AIDS Res Hum Retroviruses1994; 10:569–76.

33. Tanuri A, Vicente ACP, Otsuki K, et al. Genetic variation and suscep-tibilities to protease inhibitors among subtypes B and F isolates in Brazil. Antimicrob Agents Chemother1999; 43:253–8.

34. Caride E, Hertogs K, Larder B, et al. Genotypic and phenotypic evi-dence of drug-resistance mutation patterns between B and non-B sub-type isolates of human imunodeficiency virus sub-type 1 found in Brazilian patients failing HAART. Virus Genes2001; 23:193–202.

35. Brindeiro PA, Brindeiro RM, Mortensen C, et al. Testing genotypic and phenotypic resistance in human immunodeficiency virus type 1 isolates of clade B and other clades from children failing antiretroviral therapy. J Clin Microbiol2002; 40:4512–9.

36. Ives KJ, Jacobsen H, Galpin SA, et al. Emergence of resistant variants of HIV in vivo during monotherapy with the proteinase inhibitor saquinavir. J Antimicrob Chemother1997; 39:771–9.

37. Kaldor SW, Kalish VJ, Davies JF 2nd, et al. Viracept (nelfinavir mes-ylate, AG1343): a potent, orally bioavailable inhibitor of HIV-1 pro-tease. J Med Chem1997; 40:3979–85.

38. Adachi A, Gendelman HE, Koenig S, et al. Production of acquired immunodeficiency syndrome-associated retrovirus in human and non-human cells transfected with an infectious molecular clone. J Virol1986; 59:284–91.

39. Maschera B, Furfine E, Blair ED. Analysis of resistance to human im-munodeficiency virus type 1 protease inhibitors by using matched bac-terial expression and proviral infection vectors. J Virol1995; 69:5431–6. 40. Hertogs K, De Bethune MP, Miller V, et al. A rapid method for si-multaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Antimicrob Agents Chemother1998; 42:269–76.

41. Holland JJ, de la Torre JC, Clarke DK, Duarte E. Quantitation of relative fitness and great adaptability of clonal populations of RNA viruses. J Virol1991; 65:2960–7.

42. Holland J, Spindler K, Horodyski F, Grabau E, Nichol S, Vanderpol S. Rapid evolution of RNA genomes. Science1982; 215:1577–85. 43. Maree AF, Keulen W, Boucher CA, De Boer RJ. Estimating relative

fitness in viral competition experiments. J Virol2000; 74:11067–72. 44. Martinez-Picado J, Savara AV, Sutton L, D’Aquila RT. Replicative fitness

of protease inhibitor-resistant mutants of human immunodeficiency virus type 1. J Virol1999; 73:3744–52.

45. Harrigan PR, Montaner JS, Wegner SA, et al. World-wide variation in HIV-1 phenotypic susceptibility in untreated individuals: biologically relevant values for resistance testing. AIDS2001; 15:1671–7. 46. Pillay D, Walker AS, Gibb DM, et al. Impact of human

immunode-ficiency virus type 1 subtypes on virologic response and emergence of drug resistance among children in the Paediatric European Network for Treatment of AIDS (PENTA) 5 trial. J Infect Dis2002; 186:617–25. 47. Abecasis A, Gomes P, Derdelinckx I, et al. L89I/V: a novel mutation

(10)

selected by protease inhibitor therapy in subtype G, but not in subtype B-infected patients [abstract]. Antivir Ther2003; 8:S140.

48. Caride E, Brindeiro RM, Kallas EG, et al. Sexual transmission of HIV-1 isolate showing GrA hypermutation. J Clin Virol2002; 23:179–89. 49. Zhang H, Yang B, Pomerantz RJ, Zhang C, Arunachalam SC, Gao L. The cytidine deaminase CEM15 induces hypermutation in newly syn-thesized HIV-1 DNA. Nature2003; 424:94–8.

50. Mangeat B, Turelli P, Caron G, Friedli M, Perrin L, Trono D. Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. Nature2003; 424:99–103.

51. Zhang YM, Imamichi H, Imamichi T, et al. Drug resistance during indinavir therapy is caused by mutations in the protease gene and in its Gag substrate cleavage sites. J Virol1997; 71:6662–70.