Previous Article | Next Article ![]()
Antimicrobial Agents and Chemotherapy, July 2002, p. 2087-2094, Vol. 46, No. 7
0066-4804/02/$04.00+0 DOI: 10.1128/AAC.46.7.2087-2094.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.
McGill University AIDS Centre, Lady Davis Institute-Jewish General Hospital, Montreal, Quebec, Canada,1 Hadassah Hospital, Jerusalem, Israel2
Received 29 October 2001/ Returned for modification 22 January 2002/ Accepted 8 April 2002
|
|
|---|
|
|
|---|
Although there are overall similarities in genomic arrangement among HIV-1 clades, there is marked interstrain divergence with variations of 30 to 40% in env amino acid sequences, whereas intrastrain heterogeneity ranges from 5 to 20% (4, 7, 30). Characterization of the genotypic divergence of pol sequences among different HIV-1 subtypes is not yet complete, although the reverse transcriptase (RT) and protease (PR) enzymes are the major targets of antiretroviral therapy (3, 8, 10, 11). Both in vitro and in vivo evolution of RT polymorphism and the appearance of resistance mutations have been extensively documented for subtype B viruses (8, 16, 17, 31, 33, 35). Little information is available on the impact of viral subtype diversity on natural susceptibility to antiretroviral drugs. Moreover, it is not known whether preexisting polymorphisms of RT and PR can influence the development of drug resistance patterns through various sequence evolution pathways and have an impact on the outcome of antiretroviral therapy (3, 10-13, 37).
We analyzed here RT sequences from five drug-naive Ethiopian émigrés to Israel infected with clade C HIV-1. These sequences were compared to the RT sequences of subtype B and to subtype C reference strains from various regions of the world, as well as to reference strains of other clades. The phenotypic susceptibility of these strains was compared to clade B clinical isolates. We characterized phenotypic and genotypic drug resistance patterns in these clade C Ethiopian clinical isolates grown in increasing concentrations of three different members of the nonnucleoside reverse transcriptase inhibitor (NNRTI) family of drugs, e.g., nevirapine (NVP), delavirdine (DLV), and efavirenz (EFV).
(This research was conducted mostly by Hugues Loemba in partial fulfillment of the requirements for a Ph.D. degree, Faculty of Graduate Studies and Research, McGill University, Montreal, Quebec, Canada.)
|
|
|---|
DNA was also extracted from 106 CBMC for viral genotyping. The genotypes of the amplified viral stocks were identical to the original plasma viral RNA with respect to silent mutations, polymorphisms, and resistance mutations.
HMA.
Subtype determination of the different clinical isolates was performed by heteroduplex mobility assays (HMAs) by using the protocol and reagents from the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Disease, National Institutes of Health. DNA was extracted from CBMC with the QIAamp DNA purification kit (Qiagen, Inc., Chatsworth, Calif.). A region of the env gene spanning the C2-V5 sequence was amplified by using two rounds of PCR. A
-32P-labeled 3' primer was used during the second round of PCR to generate 0.7-kb labeled fragments. A 5-µl sample from the nested PCR product of each uncharacterized HIV-1 isolate was mixed with 5 µl of corresponding DNA fragments, amplified from plasmids containing env genes from reference strains representing the different HIV-1 subtypes A2, B1, B2, B3, C1, C2, C3, and E1. Then, 1.1 µl of HMA annealing buffer (100 mM NaCl, 10 mM Tris [pH 7.2], 2 mM EDTA) was added to each sample mixture. Homo- and heteroduplexes were formed between sample and reference strains by thermal denaturation at 94°C for 2 min and reannealing after rapid cooling on wet ice. The duplexes were mixed with loading dye and then loaded onto a 5% acrylamide gel and separated by electrophoresis on a gel apparatus (160 by 160 by 1.5 mm; Protean II Cell; Bio-Rad, Hercules, Calif.) at a constant voltage of 150 V for 4 h. The mobility of heteroduplexes was visualized by autoradiography.
HIV-1 RT sequencing. Viral RNA was isolated from culture supernatants of infected cells by using the QIAamp viral extraction kit (Qiagen). The TruGene HIV-1 Assay Gene Kit was used in conjunction with the Open Gene automated DNA sequencing system (Visible Genetics, Inc., Toronto, Ontario, Canada) to sequence the PR and RT regions of HIV-1 cDNA. Testing involved simultaneous clip sequencing of PR and codons 35 to 244 of RT from amplified cDNA in both the 3' and the 5' directions. Sequences were aligned and compared to a LAV-1 consensus sequence by using Gene Librarian software (Visible Genetics, Toronto, Ontario, Canada).
Phylogenetic analysis. A multiple alignment of five drug-naive Ethiopian clade C isolates was performed with four reference clade C strains: ETH2220 (from Ethiopia), 92BR025.8 (from Brazil), IN21068 (from India), and 96BW05.02 (from Botswana). In addition, genotypic variations of the Ethiopian clinical isolates were compared to reference isolates of subtype A (U455 and 92UG037.1 from Uganda and Q2317 from Kenya), subtype D (NDK from the Democratic Republic of Congo), subtype E (CM240 from Thailand, 90CF402.1 from Central African Republic, and 93TH253.3 from Thailand), subtype F (93BR020.1 from Brazil), subtype G (SE61165 from both Sweden and the Democratic Republic of Congo), subtype H (90CF056.1 from Central African Republic), subtype J (SE9280.9 from Sweden), subtype O (MVP5180 from Cameroon), subtype CPZ (CPZGAB from Gabon), and subtype B (LAV from France and JRFL from the United States). The sequences of all subtype reference strains were obtained from the HIV sequence database (http://hiv-web.lanl.gov). All alignments were gap stripped, and a total of 25 sequences, each with 397 bp corresponding to the same polymerase region (mainly the fingers and palm subdomains) of RT, were generated by using Genetool and Peptool software (both from Biotools, Inc., Edmonton, Alberta, Canada). A multiple alignment pairwise matrix, based on the percent identity of the sequences, was performed. Phylogenetic trees based on distances between sequences were constructed by the neighbor-joining method and by using the phylogenetic programs Dnadist, Neighbor, and Drawtree/Drawgram (HIV-WEB Treemaker interface [http//hiv-web.lanl.gov]).
Phenotypic drug susceptibility assay. Drug susceptibility was measured by determining the extent to which antiretroviral drugs inhibit in vitro HIV replication (15, 16). Clade C and B isolates were amplified and quantified by RT enzyme assays in order to generate defined and titrated clinical isolates with a minimum of interinoculum effects (16, 17). CBMC infected with patient isolates were then plated in 96-well plates both in the absence and in the presence of a variety of ARV concentrations. After 7 days, RT enzyme assays were used to determine the 50% drug inhibitory concentration (IC50) (18, 19; P. Herman, 1st IAS Conf. HIV Pathogen. Treatment, abstr. 396, 2001). The observed IC50 values of patient viral isolates were then compared to the known IC50 values of treatment-naive clade B isolates and clinical isolates known to possess resistance to select ARVs as drug-resistant controls (16, 17).
Selection for NNRTI resistance. By using procedures previously described in our laboratory, we selected for resistance to the NNRTIs, i.e., NVP, DLV, and EFV, by growing cells in the presence of increasing concentrations of drugs. In these experiments, clade B and C isolates were grown in parallel by repeated passage of wild-type clinical isolates in CBMC in the presence of increasing concentrations of NNRTIs for 8 to 15 weeks for NVP or DLV and for 13 to 30 weeks for EFV. Suboptimal doses of the drugs, i.e., <IC50 values, were used at the beginning of the selection process, i.e., 0.001 to 0.01 µM for NVP and DLV and 0.001 µM for EFV. Final concentrations reached levels of 2.0 to 10.0 µM for NVP and DLV and of 0.01 to 1.0 µM for EFV. RT assays were performed weekly to assess viral replication. At the times of RT peaks, genotyping was performed to identify changes associated with drug resistance. The times to development of resistance and the genotypic profiles were compared in clade B and C isolates.
Nucleotide sequence accession numbers. The RT sequences of the five Ethiopian isolates4742, 4743, 4761, 4762, and 4766described here are available under GenBank accession numbers AF492618, AF492619, AF492620, AF492621, and AF492622, respectively.
|
|
|---|
![]() ![]() View larger version (94K): [in a new window] |
FIG. 1. Phylogenetic profiles of clinical isolates from five drug-naive Ethiopian individuals. (A) HMA of isolates from five Ethiopian drug-naive patients. HMAs were performed as described in Materials and Methods. Heterocomplexes were formed by mixing the amplified DNA from the viral isolates with the PCR-amplified env sequences of reference strains. A more rapid migration on acrylamide gels indicates the relative degree of similarity between the unknown isolate and the reference strain sequences. For comparative purposes, the results with a wild-type clade B virus (i.e., 4746 virus) is included. (B) Phylogenetic analysis of reverse transcriptase sequences from HIV-1 Ethiopian isolates. Phylogenetic analysis comparing the RT regions of HIV-1 pol genes from five Ethiopian clinical isolates and 20 different reference strains. Tree topology was inferred from the neighbor-joining method and was based on an alignment of 397 nucleotides from which columns containing gaps were deleted. The subtype O prototype isolate was treated as an outgroup.
|
RT genotypic analysis. Direct sequencing of the RT region was performed by using Visible Genetics technology. The RT sequences of five Ethiopian isolates were aligned with a panel of reference strain RT sequences from different geographic regions. The Ethiopian isolate RT sequences had an average divergence of 6.8 to 10% from the different subtype B reference strains, whereas RT sequence variation was just 3.5 to 5.8% between the Ethiopian viruses and the clade C reference strains.
The inter- and intraspecies diversity of RT is depicted on the phylogenetic tree in Fig. 1B. The neighbor-joining tree was constructed by using multiple alignment of 25 RT sequences of the Ethiopian viruses and a broad panel of reference strains. As shown, the RT sequences of the Ethiopian clinical isolates obtained from the Israeli immigrants were more closely related to the clade C reference strain from Botswana, southern Africa, than to the Ethiopian reference strain 2220 (northeastern Africa). The RT regions of the clade C viruses clustered together apart from clade B and from the majority of other non-subtype B viruses.
The amino acid diversity at codons associated with drug resistance for the five Israeli Ethiopian isolates, compared to clade C isolates from four other geographic locations, is summarized in Table 1. As shown, natural polymorphisms were present in some of these subtype C isolates at key codon sites associated with resistance to NNRTIs and ZDV (16, 38). Amino acid substitutions at positions 98, 138, 139, and 190, which have been associated with phenotypic or secondary (low-level) resistance to NNRTIs, were also observed. Additional mutations were detected, including K70R implicated in resistance to ZDV (8, 35) and the common polymorphism L214F that has been previously suggested to play an accessory role in the presence of certain mutations that confer dual resistance to ZDV and lamivudine (37).
|
View this table: [in a new window] |
TABLE 1. Genotypic diversity in the RT nucleotide sequences of five treatment-naive clade C HIV-1 Ethiopian isolates versus the RT sequences of clade C reference strains from Botswana, Ethiopia, and Indiaa
|
Phenotypic drug susceptibility. The sensitivities of three drug-naive Ethiopian isolates to a panel of nucleoside reverse transcriptase inhibitor (NRTI) (i.e., ZDV and 3TC) and NNRTIs (i.e., NVP, DLV, and EFV) were investigated in cell culture as indicated in Material and Methods. The presence of the primary G190A resistance mutation in the clinical isolate 4743 resulted in an approximately 100-fold resistance to NVP (data not shown). This isolate remained relatively sensitive to the other NNRTIs, i.e., DLV and EFV. The overall susceptibility of the Ethiopian clade C isolates, compared to the clade B control, as judged by the IC50 values, was relatively low with respect to NVP and was higher for EFV, DLV, and the NRTIs. The presence of the K70R and L214F polymorphisms did not significantly reduce phenotypic susceptibility to ZDV in our study (data not shown).
Selection of drug-resistant variants. Selection of resistance to NVP, EFV, and DLV was performed to identify genotypic variations that may arise in clade C versus clade B isolates. As shown in Tables 2, 3, and 4, some of the mutations that arose in clade C viruses selected for resistance against NNRTIs were the same as those seen in subtype B, although a few clade C mutations may have appeared through either synonymous or nonsynonymous codon change. As indicated, the concentrations of NVP, EFV, and DLV that generated primary resistance mutations were 10, 1, and 10 µM, respectively, for the subtype B controls and 2 to 4, 0.01, and 4 µM, respectively, for the Ethiopian subtype C isolates. The numbers of weekly passages needed to generate mutations associated with phenotypic resistance to NVP, EFV, and DLV were 15, 30, and 15, respectively, for the clade B isolates and 9, 13, and 8, respectively, for the Ethiopian subtype C viruses. At the mid-point of the selection period, the clade B viruses still contained a mixture of wild-type and mutated types, whereas the clade C isolates already harbored primary mutations. Several previously unreported amino acid changes were noted in the clade C isolates, e.g., A98I in the 4742 clade C NVP-resistant virus and V106M in the EFV-resistant virus (Tables 2 and 4); the latter mutation is usually detected in clade B viruses as V106A and is associated with resistance to NVP. During DLV selection, a codon change from GCC to GTA was generated at position 62 as previously noted in the Ethiopian clade C isolate 4762 with a silent mutation at codon 62 (Table 1). This, in turn, yielded a secondary mutation, A62V, associated with multinucleoside resistance (Table 3). Another mutation V75E was detected in the Ethiopian isolate 4743 after selection with DLV (Table 3). A similar substitution is also associated with multidrug resistance against NRTIs in clade B strains (V75T).
|
View this table: [in a new window] |
TABLE 2. Selection of viruses resistant to NVP by using HIV-1 clade C and B clinical isolatesa
|
|
View this table: [in a new window] |
TABLE 3. Selection of resistance to EFV by using HIV-1 clade C and B clinical isolatesa
|
|
View this table: [in a new window] |
TABLE 4. Selection of resistance to DLV by using HIV-1 clade C and B isolatesa
|
![]() View larger version (12K): [in a new window] |
FIG. 2. Progression to resistance against NVP in cell culture. Ethiopian clade C and clade B control viruses were selected for resistance against NVP by growing cells in the presence of increasing concentrations of drugs. Concentrations of NVP selecting for primary resistance mutations and the amount of time required (in weeks) are shown.
|
|
View this table: [in a new window] |
TABLE 5. Phenotypic drug susceptibility values for NNRTIs with clade C isolates selected for resistance to designated NNRTI drugsa
|
|
|
|---|
Our findings show that five Ethiopian clinical isolates studied were of clade C origin. Of interest, these isolates were more closely related to Botswana clade C variants than to the Ethiopian strains previously described (1, 23, 27). Although the predominant subtype in Ethiopia is clade C, phylogenetic divergence of some Ethiopian isolates from other HIV-1 strains has previously been observed (1, 14, 27). Similarly, HIV-1 strains from India were reported to be highly divergent from prototypic African and U.S.-European strains but linked to the South African reference strain (14, 26).
HMAs confirmed that the Ethiopian isolates were of C subtype. Interestingly, the envelope of isolate 4743 appeared to be a clade B and C mosaic, but RT region sequencing revealed homology to other Ethiopian isolates and to the different subtype C reference strain RTs. The circulation of mosaic viruses that contain resistance mutations argues for extensive RT screening and drug resistance surveillance for non-subtype B viruses. Recombination has been reported to be a common feature among retroviruses and particularly among various HIV-1 strains (19, 28). In addition, mutation and recombination may both contribute to rescuing high-fitness HIV-1 variants that harbor phenotypically relevant genetic alterations. The recent identification of individuals infected with HIV-1 isolates of two subtypes and intersubtype recombinants suggests that this phenomenon may be common among viruses cocirculating in specific regions such as parts of Africa or Asia (19, 21, 28). Recently, a study in Ivory Coast showed that almost all HIV-1 patients were infected with non-B subtypes, with a predominance of recombinant A/G viruses (2). In addition, a high prevalence of 57.4% HIV-1 drug-resistant strains were reported among 68 patients who were treated with NNRTIs, NRTIs, and PR inhibitors between 1998 and 1999 (2). Since various non-subtype B strains are currently being reported to carry resistance mutations, intersubtype mosaics may pose problems for the application of antiviral therapies in populations where the predominant HIV-1 subtypes are non-clade B.
The analysis of Ethiopian isolate RT sequences showed that these strains clustered phylogenetically with clade C reference strain RTs; a KVEQ specific motif of silent mutations (amino acids 65, 106, 138, and 161) at resistance sites has been found in the RT polymerase region. In addition, several silent mutations at codons 183 and 186 were detected in the highly conserved YMDD motif of all studied Ethiopian isolates and previously reported subtype C prototype strains (Table 1). However, the emergence of a mutation at position 186 in RT is unlikely, since the encoded aspartic residue D186 plays a key role in catalysis and in coordinating the presence of the required metal ion (36). To date, the only mutations associated with drug resistance in this region of RT have occurred at residue M184 (39, 40).
Our evaluation of Ethiopian isolate drug susceptibility showed that most viruses displayed similar drug sensitivities, confirming observations reported for clade C strains from Zimbabwe (34). However, numerous resistance mutations, polymorphisms, and silent mutations in RT have been linked to resistance to NNRTIs and NRTIs. Phenotypic drug testing revealed resistance to NVP in isolate 4743, which carried the G190A primary mutation. As confirmed in our studies, this mutation is not associated with primary resistance to DLV, and the IC50 of EFV was only slightly higher than that of the clade B control (5, 6). It is interesting that the Ethiopian clade C isolates 4742 and 4762 initially harbored an A98S secondary mutation associated with resistance to NVP. After cell culture selection with NVP, a novel S98I mutation appeared in isolate 4742. In subtype B HIV-1 strains, the mutation at this position has been reported to be A98G and has been observed in vivo (5, 6). This is the first report of the S98I mutation in RT and suggests that S and not A is the naturally occurring residue at position 98 in the RT of subtype C viruses.
The final drug concentration that selected for primary resistance mutations was significantly higher for the clade B than clade C viruses for each of NVP (10 versus 2 µM), EFV (1 versus 0.01 µM), and Del (10 versus 1 µM), respectively. Furthermore, resistant variants were fully selected more rapidly in the clade C isolates (8 or 9 weeks with NVP or DLV and 13 weeks with EFV) compared to the clade B control (at least 15 weeks with NVP or DLV and 30 weeks with EFV).
In the middle interval of the selection period, the subtype B virus harbored a mixture of both wild-type and mutated forms in regard to all of the NNRTIs (Tables 2 to and 4). These findings suggest that clade C viruses can be rapidly selected for resistance to NNRTIs. Non-B viruses may be more prone to development of resistance after highly active antiretroviral therapy and may show different mutational patterns than B isolates (Herman, abstr. 396). Furthermore, several polymorphisms have been reported at high frequency at resistance sites in clade C RT and PR from individuals who received treatment in Israel (19). These in vivo reports correlate with our cell culture observations on the development of NNRTI resistance mutations not previously observed in vitro for HIV-1 subtype C.
Other polymorphisms and silent mutations in RT may also be linked to the emergence of resistance to NNRTIs and NRTIs. After selection with DLV, a silent mutation, A62A, initially observed in Ethiopian isolate 4762, became A62V, a secondary mutation previously associated with multidrug resistance to NRTIs (24). At late stages of selection with EFV, the novel V106M mutation was detected in the clade C isolate 4742. Another substitution, i.e., V106A, has been reported in clade B viruses after treatment failure with NNRTIs, and its emergence is associated with
120-fold resistance to NVP, intermediate levels of resistance to DLV, and low-level resistance to EFV (6). V106M has emerged at a site at which a silent mutation (GTA
GTG) was detected in clade C isolate 4742 (Table 1). The V106M codon change may be more facilitated in clade C than clade B viruses, because only a single nucleotide change is required in this instance (GTG
ATG), as opposed to the situation in clade B viruses, which requires two such events (GTA
ATG). It is possible, therefore, that silent mutations at sites related to drug resistance in clade C RT may facilitate the emergence of resistance.
A similar impact of clade C genetic background on the development of resistance to NNRTIs has been reported in patients failing therapy (29; D. Pillay, K. Sinka, P. Rice, B. Peters, J. Clarke, J. Workman, B. Evans, and P. A. Cane, Antiviral Ther. 5[Suppl. 3]:128, abstr. 163, 2000). Moreover, a different secondary mutation at a site associated with cross-resistance among multiple NRTIs, i.e., V75E, was generated in an Ethiopian clade C isolate during selection with DLV (25). This suggests that a divergent genotypic resistance profile of clade C RT may exist, which may result in enhanced development of resistance to NNRTIs on the part of clade C viruses. Prospective studies need to be conducted in order to assess the incidence of drug resistance-related mutations in populations infected with subtype C strains and undergoing drug therapy.
As stated, the emergence of some NVP resistance mutations may be more accelerated in certain non-B subtypes and facilitated by preexisting polymorphisms. In a recent clinical trial in which NVP was used in Uganda for prevention of mother-to-child transmission of HIV-1, the K103N mutation was generated in 20% of treated women by 6 weeks after a single dose of NVP at the onset of the labor (20).
In our study, the presence of certain secondary mutations associated with resistance to NNRTIs and to ZDV did not significantly decrease the susceptibility of Ethiopian clade C strains to RT inhibitors, except for strain 4743 which harbored a NVP resistance primary mutation. The natural genotypic diversity of HIV RT among different subtypes, variations in drug susceptibility, and the development of resistance to certain drugs all indicate a need for global genotypic and phenotypic surveillance of non-subtype B strains. The emergence of recombinant viruses in areas where various HIV-1 subtypes are endemic may accelerate the selection of highly resistant mosaics and represent another challenge for the treatment of HIV disease.
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»