The subtype B of HIV-1 was initially characterized in developed countries, such as the US and in the Western Europe, and was considered the major variant outside Africa and in the rest of the world. As the pandemic extended over the globe, other subtypes showed occurrences in distinct regions. The more efficient control and treatment of HIV infection in the developed world, together with the explosion of AIDS epidemics in developing countries, has shifted major incidences to other subtypes in these areas. In the beginning of this new millennium, subtype C is the most prevalent in the globe, accounting for 56% of the infections world-wide . Subtype C was first detected in South Africa and Ethiopia in retrospectively analyzed samples of 1984 and 1986, respectively [2,3], and it has been found in the majority of South African countries, such as South Africa, Botswana, Tanzania and Kenya [4–9]. Recombinant strains between subtype C and previous subtypes prevalent in these countries were also documented in Zambia  and Tanzania . Outside Africa, India is the largest population infected by subtype C viruses , and despite its recent introduction in the country , it is estimated that India will have the highest number of HIV-1 infections in the world by 2010 . Finally, China also has a high prevalence of subtype C, notoriously by B/C recombinant genomes .
Brazil is the biggest South American country and it is the most affected by the HIV/AIDS epidemics. A large number of individuals have already been infected, and prevalence rates are around 0.6% of the population (Brazilian Ministry of Health, http://www.aids.gov.br). Brazil has shown a constant change in its HIV-1 epidemic regarding gender infection ratio and risk behavior. Since the beginning of the epidemics, HIV infection patterns have shifted towards women and heterosexuals.
The HIV-1 subtype distribution in Brazil is complex when compared with other South American countries. The major circulating subtype of HIV-1 in the country is B, but other subtypes such as F, C and B/C and B/F recombinants have been documented [16–20]. Subtype C was first detected in Brazil by Csillag et al. in the cities of Porto Alegre and São Paulo . Retrospective samples from 1991 and 1992 were sequenced by the World Health Organization HIV Network, and the first complete genome of a subtype C Brazilian virus, 92BR025, was generated , showing that by 1992 this subtype was already present in the country. More recently, another Brazilian full length C virus has been sequenced . Recent HIV-1 genetic diversity surveillance studies in Brazil have shown a small incidence of subtype C, around 3% [16,23]. These viruses were found in the southern and south-eastern regions of Brazil, mostly in the states of Rio Grande do Sul, São Paulo and Rio de Janeiro.
In Brazil, we have recently started a national network for drug resistance surveillance in the drug-naive population sponsored by the Brazilian Ministry of Health (BMoH), and we had the opportunity of surveying a great number of asymptomatic, drug-naive HIV-1-infected individuals in the southern and south-eastern regions of the country in 2001. Since these geographic areas are the most frequently described areas of subtype C incidence in Brazil, we were interested in assessing the contribution of this subtype to the overall infections in the area, as well as comparing the subtype C circulating strains with other world-wide subtype C isolates.
Materials and methods
Plasma from 112 HIV-1-positive individuals confirmed by serology was isolated at different voluntary counseling and testing (VCT) centers of the BMoH. These centers spanned five different Brazilian central-southern states, Rio Grande do Sul, Paraná, São Paulo, Rio de Janeiro and Mato Grosso do Sul. Table 1 summarizes all relevant epidemiological data of the individuals analyzed, such as age, sex, state and risk behavior. Since the study was originally focused on maximizing inclusion of recently seroconverted individuals, all but two individuals had no symptomatic manifestations of any kind. None of these subjects has ever been exposed to any antiretroviral treatment, according to their written statement. The study was approved by the Brazilian IRB (project no. 526 – CONEP) as an anonymous unlinked study.
RNA isolation, polymerase chain reaction and sequencing
Virus RNA was isolated as previously described . Following cDNA generation with random primers, nested polymerase chain reaction (PCR) was conducted for individual amplification of protease (PR, whole region) and reverse transcriptase (RT, nt 105 to 651). Primers used and PCR conditions used were as described elsewhere . PCR fragments were sequenced in an ABI 310 automated sequencer (Applied Biosystems, Foster City, California, USA), with the same primers used in the second round of the amplifications. All sequences obtained were subjected to quality control assessments to ensure there were no sample mix-ups or contamination from other sources . Sequences were reported to the GenBank database (accession numbers pending). BLAST searches of sequences from each subject identified best matches in the HIV sequence database  that were always other sequences from Brazil. However, each sequence was divergent from those in the database, and among themselves, by more than 3%, suggesting an absence of sample mix-ups with previously characterized sequences.
Phylogenetic, recombination and sequence analyses
The PR and RT sequences from all samples were submitted to phylogenetic analysis for HIV-1 subtype determination. Sequences were aligned using CLUSTAL W  and manually edited by codon alignment using the Genetic Data Environment (GDE) package . The alignment was then realigned against the reference set for subtyping analysis from the Los Alamos database (http://hiv-web.lanl.gov/). HIV-1 group O sequences were used as outgroups. Phylogenetic inferences were performed by the neighbor-joining method using the F84 model of substitution implemented in PAUP version 4.0b2a . Sequences from which both genomic regions clearly clustered inside one of the subtype reference groups were compiled and their respective subtypes assigned. Sequences suggestive of recombination by the above phylogenetic inferences that have shown unclear or discordant subtype classification in PR and RT were subjected to recombination analysis. The bootscanning method implemented in the Simplot software  was used. In order to increase the number of informative sites, PR and RT sequence fragments from query samples were concatenated into one larger fragment of approximately 850 bp. The subtype reference sequences used in the analyses were representative of the three most prevalent subtypes circulating in Brazil, B (BUS83RF), F (F1BR93BR0201) and C (CBR92 BR025). The parameters used were window = 250 bp, step = 20 bp, GapStrip = on, Reps = 100, Kimura, T/t = 2, and neighbor. Based on the bootscanning suggested breakpoints, phylogenetic analyses were further performed with fragments of the sequences that represented distinct subtypes, using the same subtype reference strains and methods described above.
Further phylogenetic characterization of subtype C sequences was done using maximum likelihood analysis. A subset of subtype C sequences, including 19 sequences from our new dataset from which both PR and RT sequences were available, three previously characterized Brazilian subtype C viruses, and multiple sequences from South Africa, Zimbabwe, Tanzania, Zambia, Ethiopia and Eastern India were aligned together using CLUSTAL W. An appropriate evolutionary model for these sequences was selected using the Akaike information criteria  as implemented in MODELTEST version 3.0 . Parameters of the chosen method (TVM+I+G) were as follows: freqA = 0.3784, freqC = 0.1791, freqG = 0.1925, freqT = 0.2499; R matrix values, R[A-C] = 3.1987, R[A-G] = 11.7310, R[A-T] = 0.9773, R[C-G] = 1.4501, R[C-T] = 11.7310, R[G-T] = 1.0000; proportion of invariable sites = 0.3889, and heterogeneous variable sites distribution gamma with alpha shape = 0.6691.
Pairwise distances were calculated for different geographic clusters of subtype C sequences, including Brazilian, Indian and African groups using the Kimura 2-parameter model  implemented in the software MEGA version 2.0 . Both intra- and intercluster distances were computed and compared. Amino acid signature analysis using VESPA  was also performed for each of these groups of sequences, and they were compared to the subtype C world consensus. For the PR coding region, sequences from Zambia, Tanzania and South Africa (n = 53) were pooled together, because they are more closely related to each other, whereas sequences from Botswana (n = 8) were left as a separate group. Sequences from India (n = 8), which were also suggested to form a monophyletic group , were also analyzed together. The Brazilian cluster was represented by 22 sequences. The Brazilian subtype B sequences generated in the study were also subjected to VESPA analysis. All previously published sequences and consensi used in the pairwise distance and VESPA analyses were obtained at the Los Alamos HIV Database (http://hiv-web.lanl.gov).
The differential incidences of polymorphisms between subtypes B and C were statistically evaluated using a two-tailed Fisher exact test implemented in the software Analyze-it® for Microsoft Excel.
Molecular epidemiology of HIV infection
One hundred and twelve HIV-1-positive samples were collected and processed. Protease and reverse transcriptase (nt 105 to 651) were separately PCR amplified from plasma through RT-PCR. Seven of the samples (6.2%) did not generate PCR fragments for either genomic region (protease or reverse transcriptase) and were excluded from the analysis. All fragments obtained were sequenced and subjected to phylogenetic analysis for subtype determination. In this manner we identified 58 (55.2%) subtype B, 30 (28.6%) subtype C, and seven (6.7%) subtype F viruses. Ten isolates (9.5%) were subjected to recombination analysis by bootscanning implemented in the SIMPLOT software . All 10 sequences showed evidence of recombination between two or more of these subtypes. A schematic representation of all mosaic viruses found can be seen in Figure 1. Of the 10 sequences, five have shown a pattern of C/B recombination. Despite this large number, each of the recombinants showed breakpoints at different sites, and these sequences do not meet the criteria to classify them as a circulating recombinant form. Two of the sequences showed a B/C recombination pattern, again with different breakpoints. Interestingly, we found a virus in which three different subtypes were represented in the bootscanning analysis. The B/C/F recombinant is consistent with the three major circulating subtypes of HIV-1 in Brazil. Despite the previous characterization of individuals infected with multiple subtypes in Brazil [36,37], this is the first time a triple recombinant has been observed in the country. Finally, we found two viruses, one B and one F, in which part of the sequence could not be assigned to any subtype, and they were classified as undetermined (U).
Brazil has its own HIV-1 subtype C virus
Since this study has generated the largest collection of this subtype yet in Brazil, we wanted to further examine the relationships of all subtype C viruses co-circulating in the country. All Brazilian subtype C isolates from which both PR and RT sequences were available, three previously characterized Brazilian subtype C viruses, and multiple sequences from South Africa, Zimbabwe, Tanzania, Zambia, Ethiopia and Eastern India were aligned together using CLUSTAL W, and subjected to phylogenetic analysis using the maximum likelihood method and parameters as described in Methods. Two isolates (BRP2024 and BRP2073) were highly indicative of G←A hypermutation, and were excluded from the analysis. All subtype C sequences from Brazil, including the ones previously described, form a monophyletic cluster in the tree (Fig. 2). This cluster was supported by an 82% bootstrap analysis value. The cluster of Brazilian sequences still remained intact when the two hypermutated isolates were included or when trees spanning only RT fragments, for which more subtype C isolates are available – see Table 1 (data not shown). Consistent with the maximum likelihood analysis, all sequences from Brazil also clustered together when analyzed by neighbor-joining analysis (data not shown). Of note, the Brazilian sequences were more tightly clustered than any other country's in the tree. No sequence fell outside the cluster, such as happened for India, which has also been recently suggested to comprise an HIV-1 subtype C monophyletic group .
In order to further analyze the relationships among HIV-1 isolates of subtype C nucleotide pairwise distances were calculated for each geographic cluster (Brazil, India and Africa). The results of this analysis are depicted in Table 2. The Indian cluster had the smallest average distance (3.7%), followed by the Brazilian (4.5–4.6%) and the African (5.1–5.2%) clusters. When distances between each cluster were computed, however, slightly larger values ranging from 5 to 7.5% were documented. These values further strengthen the geographic clustering of the clade C sequences analyzed.
We next wanted to more deeply characterize the Brazilian subtype C cluster (CBR) as a distinct group. We therefore looked for specific signature amino acids in our group of subtype C sequences that were not present in other subtype C viruses from around the world. Signature analyses using VESPA  were performed against the world consensus C individually for PR and RT regions. CBR was represented by the 22 sequences included in Figure 2. Figure 3a shows the results of the VESPA analysis for PR. Only amino acid differences that appeared in more than 50% of the sequences from the world consensus C for each set were computed and shown. None of the consensi from Africa had any signature differences from the world consensus C. On the other hand, sequences from India showed three differences from the consensus, namely an arginine at position 14 (a lysine in the consensus; 50%), a valine at position 36 (an isoleucine in the consensus; 62.5%), and a proline at position 63 (a leucine in the consensus, 62.5%). The analysis of CBR sequences showed four significant amino acid signatures which are not present in the world consensus C: a threonine at position 12, a leucine at position 19, a lysine at position 37, and an asparagine at position 41. These amino acid changes are highly prevalent among CBR sequences, with incidences of 95, 85, 90 and 100%, respectively, in the sequences analyzed. Of those, the first two changes, S12T and I19L, are commonly observed in subtype B sequences, and in fact are present in the world consensus B from Los Alamos. The change I36V, a signature of CIN and also prevalent in the Chinese C/B recombinant viruses (Fig. 3c), was not found in any of the CBR sequences. In the RT region analyzed, CBR (n = 28) did not show any signature amino acid changes when compared to the world consensus C. Sequences from African and Indian viruses showed diverse patterns of signatures in that region (Fig. 3b). Of note, we have found a number of signature amino acids which are specific of the Indian and the Ethiopian strains in the RT region, namely E39D, V60I, D121Y, G123D, and K211R. The last one is also found in some African, but not South African and Brazilian strains. These results show that the Brazilian, as well as Indian subtype C cluster, is further supported by the presence of specific amino acid signature patterns.
We also looked for possible signature sequences in our subtype B subset (53 sequences for PR and 42 sequences for RT), which would be different from the Los Alamos’ world subtype B consensus. The analysis showed that only two amino acid differences were present: L63P (47.8%) and C95S (67.4%). Although L63P has not reached the threshold of 50% used in our analysis, this mutation has been described before with a high incidence in subtype B viruses from Brazil [17,38,39], and its position is associated with secondary drug resistance. The biological significance of the C95S change is unknown. As for subtype CBR, no specific signature patterns of subtype BBR were observed in the RT region (Fig. 3b).
When PR sequences representing these two subtypes were analyzed separately, we observed differential frequencies of mutations in a number of secondary residues associated with protease inhibitor resistance. Notably, the mutations M36I and L63P/H/A/T had significant differences. The M36I mutation had incidences of 24.5% (13 of 53) and 68.2% (15 of 22) for subtypes B and C, respectively (P = 0.001). Similarly, L63P/H/A/T had incidences of 52.8% (28 of 53) and 13.6% (3 of 22) for subtypes B and C (P = 0.003). In contrast to that, the mutation L10I/V, which was present in 24.5% (13 of 53) of the subtype B sequences and in 36.4% (8 of 22) of the subtype C sequences, as well as the mutation K20R, with an incidence of 5.7% (3 of 53) in B sequences, and of 9.1% (2 of 22) in C sequences, had no significant subtype-specific differences (P = 0.445 and 0.921, respectively).
Recombinant viruses are composed of genuine Brazilian subtypes
Since we had observed specific signature patters in Brazilian HIV-1 subtype C and B sequences, we wanted to see if the recombinant forms we have described had been generated by truly local parental viruses. To accomplish that, we looked for the CBR and BBR signature amino acids in our B/C and C/B recombinants. The results of this analysis for the protease region are shown in Figure 3. The reverse transcriptase region was not subjected to analysis since no specific Brazilian signatures were found in it (see above). Brazilian B/C viruses (labeled BR.BC in the Fig. 3c) retain the signature pattern of BBR in their subtype B portions, such as L63P and C95S in the PR region. Similarly, the C/B recombinant viruses showed specific CBR signatures in their subtype C-derived sequences, such as N37K and K41N in the PR region. These results strongly suggest that the recombinant forms observed in our study were generated locally by recombination events of subtypes B and C parental strains, and do not represent introductions by migration of external recombinant viruses.
In this study, a large proportion of HIV-1 subtype C in the south of Brazil was observed. Previous surveys in Brazil have shown frequencies of around 3% for subtype C viruses in the country [16,23]. In this new dataset, we have found that almost 30% of the viruses circulating in south and south-eastern Brazil are of subtype C. In fact, the local incidence of this subtype in the city of Porto Alegre, RS, where it was first detected, reaches remarkable 37% of the viruses, and can increase to 44% when the recombinant viruses containing subtype C-related sequences are considered. Although there are no previous reports of subtype C incidence in the south of Brazil, an analysis of AIDS patients in Porto Alegre in 2001 using env HMA has shown a frequency of 22% for HIV-1 subtype C-infected individuals (A.M.B. Martinez, personal comm.). Despite our analysis has been based on the pol region, we could speculate an increase in the prevalence of C subtype in that city.
We have shown evidence that subtype C HIV-1 viruses from Brazil form a monophyletic group when compared with other viruses of the same subtype from around the world, suggesting that subtype C has entered the country as a single introduction, or at least as a very small group of genetically related viruses. Alternatively, a more rapid spread of this subtype throughout susceptible individuals, when compared with the subtype B counterparts that co-circulate in the country, could also explain the tighter relatedness of subtype C Brazilian viruses. The intra-clade genetic diversity of CBR also supported the observed clustering of the sequences (Table 2). The Brazilian cluster showed an intermediary mean genetic divergence (4.5%) when compared with the Indian and African subtype C sequences (3.7 and 5.0%, respectively). Inter-cluster distances were higher, ranging from 5 to 7.5%, showing that viruses within each group are more genetically related to each other than to viruses from other groups.
The high incidence of subtype C HIV-1 viruses in the south and south-eastern part of the country, a region with equally high incidences of subtype B viruses, has enabled a high number of recombination events between viruses from these two subtypes. This is demonstrated by a large number of B/C and C/B recombinants seen among the mosaic viruses analyzed in our dataset. Eight out of 10 recombinant viruses showed subtype B- and C-related sequences in their genomes. We cannot completely exclude the possibility that PR and RT sequences were amplified from two different viruses co-infecting an individual. However, as the methods we have used amplify the predominant strain of the individual, strains of different subtypes would have to be present in approximately equal amounts in order to explain the obtained results. As the raw sequence data obtained were clear and unambiguous, we believe that the majority of the described strains are likely to represent true mosaic viruses. The absence of a defined prevailing CRF suggests that these recombination events are recent and probably an ongoing process in Brazil. The recombinant viruses seem to have been generated locally, and not imported from another country, as shown by the presence of CBR and BBR signatures in their sequences (Fig. 3c).
We have shown that subtype CBR has its own signature patterns which are not seen in any other subtype C group, as we could determine by Vespa analysis. Those included the changes S12T, I19L, N37K and R41N of the protease amino acid sequence. Velazquez-Campoy et al.  have recently reported specific amino acid differences between HIV-1 proteases of subtypes A, C and B that account for a higher vitality of the first two enzymes when compared with the third one in in vitro assays. These changes were M36I, R41K, H69K and L89M in subtype C PR, and I13V, E35D and R57K together with the previous four changes in subtype A PR. Moreover, these amino acid changes also make HIV-1 proteases more resistant to most of the commercially available protease inhibitors, such as indinavir, ritonavir, saquinavir and nelfinavir . Our subtype CBR contains three out the four substitutions reported for subtype C PR, but it lacks R41K. It shows instead an R41N mutation, which is in fact one of the CBR signatures. Although this change is not as conservative as R41K, the effect of that mutation in the vitality of subtype C protease is not known. In addition, the mutation N37K, which is represented in 91% of Brazilian subtype C sequences and is not associated to any specific phenotype, remains to be determined. When we analyzed our new subtype B sequences, we also found that positions E35, N37 and R41 are subjected to a very high degree of amino acid variation (data not shown). Although they did not meet our threshold of 50% to be included as subtype B signatures, they had percentages of variation ranging from 30 to 43%. The position E35 had changed to D and the position R41 had changed to K both in 94% of the isolates where variation was detected. In interpreting these data, we could envisage a scenario in which proteases from viruses circulating in drug-naive subjects, independent of their subtype, could incorporate amino acid changes that enable them to evolve into a higher fitness state, with a higher catalytic efficiency. However, experimental evidence to support this hypothesis remains to be demonstrated. Additionally, a number of secondary residues associated with protease inhibitor drug resistance such as M36I and L63P/H/A/T were clearly associated with different subtype, representing signature sequences of subtypes C and B, respectively. Although experimental evidences of the impact of these substitutions in subtype C viruses on resistance and clinical outcome are largely unknown, we could speculate that these mutations may decrease their genetic barrier to resistance acquisition.
The study of HIV-1 genetic variability in distinct regions of the globe is of pivotal importance in the design of more efficient, customized HIV vaccines which must include locally circulating subtypes. Despite the recent report pointing out the elevated incidence of subtypes B and F, and the existence of B/F CRFs in South America , a major concern provided by our study is the high, and perhaps increasing, prevalence of subtype C in Brazil. Moreover, differences in vitality and fitness of subtype C viral proteases, and their relative in vitro resistance to commercially available and widely used PIs when compared to subtype B, should be of special concern to clinicians and epidemiologists. The differential polymorphisms between PR and RT of clades C and B associated with drug resistance justify the setting up of clinical trials comparing efficiencies of antiretroviral drug treatment and vaccine intervention in which these two subtypes circulate.
We would like to thank all participants of the Brazilian Network for Drug Resistance Surveillance (HIV-BResNet) who contributed to this study.
Sponsorship: this work was supported by the AIDS/STD National Program, Brazilian Ministry of Health, the State Science Foundation of Rio de Janeiro Grant E-26/151.970/00, the Brazilian Council for Scientific and Technologic Development Grant 462394/00-0, and Wellcome Trust Grant 061238/Z/00/Z.
1.Esparza J, Bhamarapravati N. Accelerating the development and future availability of HIV-1 vaccines: Why, when, where, and how? Lancet
2.Johansson B, Sherefa K, Sonnerborg A. Multiple enhancer motifs in HIV type 1 strains from Ethiopia. AIDS Res Hum Retrovir
3.Zacharova V, Becker ML, Zachar V, Ebbesen P, Goustin AS. DNA sequence analysis of the long terminal repeat of the C subtype of human immunodeficiency virus type 1 from Southern Africa reveals a dichotomy between B subtype and African subtypes on the basis of upstream NF-IL6 motif. AIDS Res Hum Retrovir
4.Novitsky VA, Montano MA, McLane MF, Renjifo B, Vannberg F, Foley BT, et al
. Molecular cloning and phylogenetic analysis of human immunodeficiency virus type I subtype C: a set of 23 full-length clones from Botswana. J Virol
5.Renjifo B, Chaplin B, Mwakagile M, Shah P, Vamberg F, Msamanga G, et al
. Epidemic expansion of HIV type 1 subtype C and recombinant genotypes in Tanzania. AIDS Res Hum Retrovir
6.van Harmelen JH, Wood R, Lambrick M, Rybicki EP, Williamson AL, Williamson C. An association between HIV-1 subtypes and mode of transmission in Cape Town, South Africa. AIDS
7.van Harmelen JH, Van der Ryst E, Loubser AS, York D, Madurai S, Lyons S, et al
. A predominantly HIV-1 subtype C-restricted epidemic in South African urban populations. AIDS Res Hum Retrovir
8.Neilson JR, John GC, Carr JK, Lewis P, Kreiss JK, Jackson S, et al
. Subtypes of human immunodeficiency virus type 1 and disease stage among women in Nairobi, Kenya. J Virol
9.Robbins KE, Kostrikis LG, Brown TM, Anzala O, Shin S, Plummer FA, et al
. Genetic analysis of human immunodeficiency virus type 1 strains in Kenya: a comparison using phylogenetic analysis and a combinatorial melting assay. AIDS Res Hum Retrovir
10.Salminen MO, Carr JK, Roberston DL, Hegerich P, Gotte D, Koch C, et al
. Evolution and probable transmission of intersubtype recombinant human immunodeficiency virus type 1 in a Zambian couple. J Virol
11.Koulinska IN, Ndung'u T, Mwakagile D, Masamanga G, Kagoma C, Fawzi W, et al
. A new human immunodeficiency virus type 1 circulating recombinant form from Tanzania. AIDS Res Hum Retrovir
12.Shankarappa R, Chatterjee R, Learn GH, Neogi D, Ding M, Roy P, et al
. Human immunodeficiency virus type 1 env sequences from Calcutta in Estearn India: identification of features that distinguish subtype C sequences in India from other subtype C sequences. J Virol
13.Delwart EL, Shpaer EG, McCutchan FE, Louwagie J, Grez M, Rübsamen-Waigmann H, et al
. Genetic relationships determined by a DNA heretroduplex mobility assay: analysis of HIV-1 env genes. Science
14.Bollinger RC, Tripathy SP, Quinn TC. The human immunodeficiency virus epidemic in India: current magnitude and future projections. Medicine (Baltimore)
15.Rodenburg CM, Li Y, Trask SA, Chen Y, Decker J, Robertson DL, et al
. Near full-length clones and reference sequences for subtype C isolates of HIV type 1 from three different continents. AIDS Res Hum Retrovir
16.Bongertz V, Bou-Habib DC, Brigido LF, Caseiro M, Chequer PJ, Couto-Fernandez JC, et al
. HIV-1 diversity in Brazil: genetic, biologic, and immunologic characterization of HIV-1 strains in three potential HIV vaccine evaluation sites. J Acquir Immune Defic Syndr
17.Caride E, Brindeiro R, Hertogs K, Larder B, Dehertogh P, Machado E, et al
. Drug-resistance reverse transcriptase genotyping and phenotyping of B and non-B subtypes (F and A) of human immunodeficiency virus type 1 found in Brazilian patients failing HAART. Virology
18.Cornelissen M, Kampinga G, Zorgdrager F, Gousmit J,Unaids Network for HIV Isolation and Characterization. Human immunodeficiency virus type 1 subtypes defined by env show high frequency of recombinant gag genes. J Virol
19.Couto-Fernandez JC, Morgado MG, Bongertz V, Tanuri A, Andrade T, Brites C, et al
. HIV-1 subtyping in Salvador, Bahia, Brazil: a city with African sociodemographic characteristics. J Acquir Immune Defic Syndr
20.Sabino EC, Shpaer EG, Morgado MG, Korber BT, Diaz RS, Bongertz V, et al
. Identification of human immunodeficiency virus type 1 envelope genes recombinant between subtypes B and F in two epidemiologically linked individuals from Brazil. J Virol
21.Csillag C. HIV-1 subtype C in Brazil. Lancet
22.Gao F, Robertson DL, Carruthers CD, Morrison SG, Jian B, Chen Y, et al
. A comprehensive panel of near full-length clones and reference sequences for non-subtype B isolates of human immunodeficiency virus type 1. J Virol
23.Brindeiro R, Vanderborght B, Caride E, Correa L, Oravec RM, Berro O, et al
. Sequence diversity of the reverse transcriptase of human immunodeficiency virus type 1 from Brazilian untreated individuals. Antimicrob Agents Chemother
24.Stuyver L, Wyseur A, Rombout A, Louwagie J, Scarcez T, Verhofstede C, et al
. Line probe assay for rapid detection of durg-selected mutations in the human immunodeficiency virus type 1 reverse transcriptase gene. Antimicrob Agents Chemother
25.Korber BT, Learn G, Mullins JI, Hahn BH, Wolinsky S. Protecting HIV databases. Nature
26.Gaschen B, Kuiken C, Korber B, Foley F. Retrieval and on-the-fly alignment of sequence fragments from the HIV database. Bioinformatics
27.Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucl Acids Res
28.Smith SW, Overbeek CR, Woese W, Gilbert W, Gillevet PM. The Genetic Data Environment: An expandable GUI for multiple sequence analysis. Comput Appli Biosci
29.Swofford D. PAUP 4.0: Phylogenetic analysis using parsimony (and other methods), 4.0b2a.
Sunderland, MA, USA: Sinauer Associates, Inc.; 1999.
30.Salminen MO, Carr JK, Burke DS, McCutchan FE. Identification of breakpoints in intergenotypic recombinants of HIV type 1 by bootscanning. AIDS Res Hum Retrovir
31.Akaike H. A new look at statistical model identification. IEEE Trans Automatic Control
32.Posada D, Crandall KA. MODELTEST: testing the model of DNA substitution. Bioinformatics
33.Kimura M. A simple method for estimating evolutionary rates of base substitution through comparative studies of nucleotide sequences. J Mol Evol
34.Kumar S, Tamura K, Jakobsen IB, Nei M. MEGA2: Molecular Evolutionary Genetics Analysis software, Bioinformatics
35.Korber B, Myers G. Signature pattern analysis: a method for assessing viral sequence relatedness. AIDS Res Hum Retrovir
36.Janini LM, Pieniazek D, Peralta JM, Schechter M, Tanuri A, Vicente AC, et al
. Identification of single and dual infections with distinct subtypes of human immunodeficiency virus type 1 using restriction fragment length polymorphism analysis. Virus Genes
37.Janini LM, Tanuri A, Schechter M, Peralta JM, Vicente AC, De La Torre N, et al
. Horizontal and vertical transmission of human immunodeficiency virus type 1 dual infections caused by viruses of subtypes B and C. J Infect Dis
38.Dumans AT, Soares MA, Pieniazek D, Kalish M, De Vroey V, Hertogs K, et al
. Prevalence of protease and reverse transcriptase drug resistance mutations over time in drug-naïve HIV-1-positive individuals in Rio de Janeiro, Brazil. Antimicrob Agents Chemother
39.Pieniazek D, Rayfield M, Hu DJ, Knengasong J, Wiktor SZ, Downing R, 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-naïve individuals worldwide. HIV Variant Working Group. AIDS
40.Velazquez-Campoy A, Todd MJ, Vega S, Freire E. Catalytic efficiency and vitality of HIV-1 proteases from African viral subtypes. Proc Natl Acad Sci
41.Carr JK, Avila M, Carrillo MG, Salomon H, Hierholzer J, Watanaveeradej V, et al
. Diverse BF recombinants have spread widely since introduction of HIV-1 into South America. AIDS
Participants in the Brazilian Network for Drug Resistance Surveillance
A. Tanuri, M. Soares, R. Brindeiro, M. Arruda, and E. Soares (UFRJ, Rio de Janeiro); E. Caride (IOC, FIOCRUZ), C. Lauria (Hospital Pedro Ernesto, UERJ, Rio de Janeiro), F. Esperanza, S. Ishii, F. Oliveira, and I. Pires (IBEX, Rio de Janeiro); M. Morgado, J. Fernandez, S. Fernandez (FIOCRUZ, Rio de Janeiro); R. Diaz, L. Costa and E. Cavalieri (EPM/UNIFESP, São Paulo); E. Sabino, N. Gaburo Jr., and A. Shoko (Fundação Pró-Sangue, São Paulo); R. Rodrigues, L. Brigido, and R. Custódio (Instituto Adolfo Lutz, São Paulo), M. Dantas, D. Barreira and P. Teixeira (CN/DST-AIDS, Brasília).