VIROLOGY

Intrahost and Interhost Variabilityof the HIV Type 1 nef Gene in Brazilian Children

Elizabeth Cavalieri, Camila Florido, Elcio Leal, Daisy Maria Machado, Michelle Camargo,Ricardo S. Diaz, and Luiz Mario Janini

Abstract

Many aspects of HIV-1 pathogenesis are affected by Nef protein activity, and efforts have been made to studyvariation in the nef gene and how that variation relates to disease outcome. We studied the genetic diversity ofthe nef gene in distinct clones obtained from the same patient (intrahost) and in sequences obtained fromdifferent hosts (interhost). The set of sequences analyzed was obtained from HIV-1-infected Brazilian childrenand contained 112 clones from 25 children (intrahost samples), as well as 55 sequences from epidemiologicallyunlinked children (interhost samples). We found extensive site polymorphisms and amino acid length varia-tions, mainly in the amino terminal region of the nef gene, between the myristoylation motif (MGxxxS) and theMHC-1 downregulation motif (Rxx). Analysis of the sequences deposited in the Los Alamos HIV sequencesdatabase (www.hiv.lanl.gov) indicated that the most frequent motif at the MHC-1 downregulation site in thesubtype B strain is R86%A64%E82% (n¼ 1040) and R78%T74%E56% in the subtype C strain (n¼ 549). Conversely,the Brazilian subtype B isolates presented the motif R81%T62%E67% at this site (n¼ 64). A detailed analysis ofselective pressures identified a concentration of codons under strong positive selection in the amino terminalregion of the nef gene. We also determined that different sites are under positive selection in the subtype B andsubtype C viruses. The amino acid composition in the MHC-1 downregulation motif of the nef gene in oursequences may indicate a distinct adaptive pattern of HIV-1 subtype B to the Brazilian host population.

Introduction

The pleiotropic role of the Nef protein elicits an in-tense selective pressure from the host immune system.1–4

Immunodominant cytotoxic T lymphocyte (CTL) epitopeshave been identified in nearly all variations of the Nef pro-tein.5 In addition, it has been demonstrated that the adaptiveevolution of HIV-1 within the cellular environment is mostlyattributable to immune-mediated positive selection againstviral proteins.6,7 In this context, the CTL response plays acritical role in restricting HIV-1 replication in infected indi-viduals.8–13 Many studies have addressed the implications ofHIV diversification and their influence on disease outcomeand viral adaptation.14–17

Previous work has shown that the dynamics of the naturalcourse of HIV-1 infection differ in adults and children, as ex-emplified by viral loads that decline much more slowly inchildren.18–20 During the first years after infection, the mater-nally acquired virus maintains a high level of replication thatgradually decreases to a level similar to that observed in in-fected adults.20,21 Additionally, CTL responses are rare before

6 months of age and are generally narrow and less intensethan the responses measured in infected adults.22–27 Thislonger period before development of an efficient cell-mediatedresponse to HIV-1 epitopes in infected children might accountfor the higher viral loads encountered in pediatric infections,particularly when infected children share the same humanleukocyte antigen (HLA) alleles with their mothers.28–30 Fur-thermore, early exposure to viral peptides might induce anexacerbated and ineffective Th2-like response.26

Similarly, the evolutionary dynamics of HIV-1 in pediat-ric infection might be affected by the pediatric immune re-sponse against viral proteins.4,31,32 There is a notable lack ofstudies focused on HIV-1 diversification during pediatric in-fection.33–35 Pioneering works have provided a rough esti-mate of viral diversity and have tried to establish a linkbetween this diversity and the outcome of AIDS in chil-dren.36–39 Recently, other studies have focused on the loss ofviral diversity and the selective pressure conferred by ma-ternal antibodies during perinatal HIV transmission.40,41 Nostudies have yet provided a comprehensive analysis of se-lective pressures on the viral genome nor has any work

Federal University of Sao Paulo, Sao Paulo, Brazil.

AIDS RESEARCH AND HUMAN RETROVIRUSESVolume 25, Number 11, 2009ª Mary Ann Liebert, Inc.DOI: 10.1089=aid.2009.0061

1

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 1

examined the role of polymorphisms in pediatric diseaseprogression. It has been shown that the nef gene is underhigher positive selection in slow progressing children, ascompared to rapidly progressing pediatric patients.35 Thisstudy also found that nef polymorphisms might be related toonset of the disease. However, polymorphisms found in thenef gene might represent high genetic diversity of a particularstrain of HIV-1, without any relationship to pathogenesis.42 Inaddition, compelling evidence has also been discovered in-dicating that most intraindividual HIV-1 mutations are notnecessarily representative of the viral adaptation in a popu-lation of individuals.43,44 Moreover, a recent study showedthat the positive selection on the env gene of HIV-1 in infectedchildren might be equivalent to the selection observed in in-fected adults.34

For these reasons, we have analyzed the diversity of the nefin pediatric HIV-1 infection, accounting for both intrahostvariation and interhost differences. Investigation of selectiveregimens was performed using a maximum likelihoodmethod, through the analysis of the number of nonsynon-ymous substitutions (dN) and synonymous substitutions (dS),codon by codon.

Materials and Methods

Biological samples

All samples were obtained from HIV-1 pediatric patientswho regularly receive medical assistance in the ambulatorycare unit of the Federal University of Sao Paulo. Biologicalsamples were obtained in full accordance with signed in-formed consent. Our sample set of 55 interhost sequences (i.e.,sequences from epidemiologically unrelated individuals) wasdivided into two datasets: (3G) 16 vertically infected childrenranging from 0 to 3 years old and (4G) 39 vertically infectedchildren ranging from 4 to 6 years old. Another set used in thisstudy was composed of 25 pediatric patients, divided into twogroups: (1) pediatric children with undetectable virus loadand (2) pediatric children with detectable virus load.

Our sample set of intrahost sequences (i.e., multiple clonesobtained from the same individual) was composed of a totalof 112 HIV-1 nef clones, obtained by limiting dilution poly-merase chain reaction (PCR) amplification from peripheralblood mononuclear cells (PBMCs) of 25 pediatric patientsreceiving antiretroviral therapy.

HIV databank samples

Additional sequences collected from the Los Alamos HIVsequences database (www.hiv.lanl.gov) were also included inthe selective regimen analyses. They were organized into datasets with the following accession numbers:

Twenty-eight sequences of African children infected byHIV-1 subtype C35,45:

DQ398972 (South Africa), DQ398985 (South Africa), DQ-398983 (South Africa), DQ398994 (South Africa), DQ398978(South Africa), DQ398981 (South Africa), DQ398990 (SouthAfrica), DQ398973 (South Africa), DQ399002 (South Africa),DQ399001 (South Africa), DQ398977 (South Africa), DQ-399003 (South Africa), DQ398982 (South Africa), DQ398987(South Africa), DQ398974 (South Africa), DQ398976 (SouthAfrica), DQ398999 (South Africa), DQ398991 (South Africa),DQ398995 (South Africa), DQ398998 (South Africa), DQ-

399005 (South Africa), DQ398988 (South Africa), DQ398979(South Africa), DQ398993 (South Africa), DQ398975 (SouthAfrica), DQ398984 (South Africa), DQ398980 (South Africa),DQ398996 (South Africa).

Thirty sequences of adults infected by HIV-1 subtypeB46–48:

AF129358 (United Kingdom), AF129361 (United Kingdom),AF129362 (United Kingdom), AF129363 (United Kingdom),AF129370 (United Kingdom), AF129373 (United Kingdom), AF-129374 (United Kingdom), AF129375 (United Kingdom), AF-129377 (United Kingdom), AF129378 (United Kingdom),AF129379 (United Kingdom), AF129380 (United Kingdom), AF-129381 (United Kingdom), AF129382 (United Kingdom),AF129384 (United States), AF129388 (United States), AF129390(USA), AF129391 (USA), AF129394 (United Kingdom),AF462705 (South Korea), AF462709 (South Korea), AF462714(South Korea), AF462742 (South Korea), AF462758 (SouthKorea), AF462773 (South Korea), AF462778 (South Korea),AF462791 (South Korea), AF462792 (South Korea), AF462793(South Korean), AY819715 (Russia).

Seventy-eight sequences of adults infected by HIV-1 sub-type C49–53:

AF286223 (India), AF286224 (Zambia), AF286225 (Zam-bia), AF286227 (South Africa), AF286231 (India), AF286232(India), AF286233 (Israel), AF290027 (Botswana), AY265893(India), AY727522 (Brazil), AY727524 (Brazil), AY727525(Brazil), AY727527 (Brazil), AY838565 (South Africa), AY-838566 (South Africa), AY838567 (South Africa), AY838568(South Africa), AY838729 (South Africa), AY838730 (SouthAfrica), AY838731 (South Africa), AY838732 (South Africa),AY838733 (South Africa), AY838734 (South Africa), AY-838736 (South Africa), AY838737 (South Africa), AY838738(South Africa), AY838739 (South Africa), AY838740 (SouthAfrica), AY838741 (South Africa), AY838742 (South Africa),AY838743 (South Africa), AY838744 (South Africa), A-Y838745 (South Africa), AY838746 (South Africa), AY838748(South Africa), AY838749 (South Africa), AY838751 (SouthAfrica), AY838752 (South Africa), AY838754 (South Africa),AY838755 (South Africa), AY838756 (South Africa), AY-614502 (India), AY614503 (India), AY614504 (India),AY614505 (India), AY614506 (India), AY614507 (India), AY-614508 (India), AY614509 (India), AY614510 (India),AY614511 (India), AY614512 (India), AY614513 (India), AY-614514 (India), AY614515 (India), AY614516 (India),AY614517 (India), AY614518 (India), AY614519 (India), AY-614520 (India), AY614521 (India), AY614522 (India),AY614523 (India), AY614524 (India), AY614525 (India), AY-614526 (India), AY614527 (India), AY614528 (India),AY614529 (India), AY614530 (India), AY614531 (India), AY-614532 (India), AY614533 (India), AY614534 (India),AY614535 (India), AY614536 (India), AY61453 (India), AY-614538 (India), AY614539 (India), AY614540 (India),AY614541 (India), AY614542 (India).

PCR amplification and sequencing

Proviral DNA was extracted from buffy-coat using theQIAquick Kit (Quiagen Inc., Hilden, Germany) and used as atarget to amplify a 615-bp fragment of the HIV-1 nef gene. Endpoint nested PCR reaction was carried out for 25 sets ofsamples (intrahost analysis) and bulk nested PCR was per-formed for 55 sets of samples (interhost analysis). Primers

2 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 2

used for amplification of the nef sequences were Nef out 50

(50-CCGCCGCCTGAGAGACTTAC-30, positions to HXB28528? 8547) and Nef out 30 (50-ATCTGACCCTTAGCCACTCC- 30, positions 9299? 9318 LTR 50); inner primers Nefinner 50 (50-GATGCTGAGGGGACAGATAGG-30, positions8688? 8708) and Nef inner 30 (5’-CCTCCCCGGAAAGTCCCCAG- 30, positions 9339? 9358 LTR 50). The sameprimers were used for the amplification of nef bulk sequences.

PCR reactions were performed in a final volume of 50 ml,and contained 10 mM Tris–HCl, pH, 8.0, 50 mM KCl, 3.5 mMMgCl2, 0.2 mM of each dNTP, 20 pmol of each primer, and2.5 U of Taq DNA polymerase (Gibco BRL Life Technologies).The first and second rounds of the reactions were incubated at948C for 10 min, followed by 35 cycles of 948C for 30 s; 558C for30 s; 728C for 2 min, with a final extension of 728C for 10 min.For the second round, 5ml of the first PCR reaction was usedas template. All reactions were performed on a Veriti thermalcycler (Applied Biosystems Inc., Foster City, CA). Secondround PCR products were inspected by electrophoresis on1.5% agarose gels stained with ethidium bromide using0.5�TBE buffer.

PCR products were purified using a QIAquick spin PCRpurification kit (Quiagen Inc., Hilden, Germany) prior to thesequencing reactions. DNA sequences were determined bi-directionally on an ABI 3100 sequencer (Applied BiosystemsInc., Foster City, CA) using an ABI PRISM Big Dye TerminatorCycle Sequencing Ready Reaction Kit with AmpliTaq DNAPolymerase (Applied Biosystems Inc., Foster City, CA).

Sequence alignments

The sequences were first aligned using the Clustal X pro-gram.54 All sites with deletions and insertions were excluded,in order to preserve the reading frames of the genes. After thisediting process, the sequences were manually aligned usingthe SE-AL program, version 2.0 (Department of Zoology,Oxford University; http:==evolve.zoo.ox.ac.uk=software=).

Analysis of selective pressures

We used a maximum likelihood method to estimate selec-tion pressures.55,56 This approach compares the fit of the datato various models of codon evolution, which differ in thedistribution of dN=dS (o) between sites. This method alsotakes into account the phylogenetic relationships of the se-quences.

The model 0 (M0: one-ratio) assumes a single o for all sitesin the alignment and is thus the simplest model specified.

The model 1 (M1: neutral) allows for different proportionsof conserved sites (o¼ 0) and neutral sites (o¼ 1), both esti-mated from the data.

The model 2 (M2: selection) adds an additional class of sitesthat has a o ratio (which can be> 1) estimated from the data.

The model 3 (M3: discrete) allows for positive selection byincorporating three categories of codon sites, with the o valueat each site estimated from the data. Nested models can becompared using a standard likelihood ratio test (LRT). Sig-nificant evidence for positive selection is provided if M2, ormore normally M3, significantly rejects the null hypothesesM0 and M1, and if the favored models contain a class of co-dons with o>1. We also used a model that employs differento ratios between lineages.57 Briefly, the model B, which is anextension of the discrete model (M3), assumes that both the

o ratio and the positively selected sites vary between thebranches of a maximum likelihood tree. Therefore, the dis-crete model (M3 with K¼ 3) provides the null hypothesis tobe compared with model B. All of these models are im-plemented in the CODEML program from the PAML v. 3.14package (http:==abacus.gene.ucl.ac.uk=software=paml.html).

Phylogenetic inferences

Maximum likelihood trees were generated under the GTRmodel58 plus a gamma distribution and the proportion ofinvariable sites (GTRþGþPinvar). This model was used toobtain an initial tree through the neighbor-joining method;then it was swapped by the nearest-neighbor interchange(NNI) method. All phylogenetic inferences were done usingthe program PAUP*.59

Recombination analysis

We used a window-based phylogenetic tool to detect re-combination in the nef gene. The alignment was trimmed andall gaps were removed. The edited alignment contained a 540-bp stretch of the nef gene that was used to test for the presenceof recombination. Recombinants and breakpoints were iden-tified with the program Simplot v.2.5 (http:==sray.med.som.jhmi.edu=RaySoft), using sequences from established HIV-1subtypes as references. Recombination was checked byBootscan, using the neighbor-joining and the Kimura (two-parameter) substitution models, over a sliding window of140 bp with 20-bp increments. Window trees were replicated500 times in order to provide bootstrap support for permutedtrees.

Results

Amino acid composition of the nef gene from Brazilianpediatric patients

To characterize the sequences of the HIV-1 nef gene fromBrazilian pediatric patients, we analyzed amino acid substi-tutions on a site-by-site basis. nef gene sequences were con-verted into amino acid sequences and gathered into twoalignments. One alignment corresponded to the 112 clonesobtained from 25 pediatric patients under antiretroviraltherapy. The other alignment was composed of 55 Nef se-quences from children in groups 3G and 4G. The amino acidcomposition of the nef gene from Brazilian isolates presented ahigh degree of variation, including amino acid substitutionsand insertions. While amino acid polymorphisms were ob-served throughout the sequences, insertions were more fre-quently observed in the amino terminal region of the gene.We noticed that insertions in the amino terminal region werefrequently located just after the MHC-1 downregulationsite (RTE in our alignments), a region that may accommodateextensive sequence modifications without alteration of thenormal function of the nef gene. Remarkably, amino acid in-sertions with distinct characteristics were observed withinclones from single individuals. For example, insertions of 10amino acids with distinct compositions (i.e., PPAAEEMGNPand PPAAEGMETP) were observed in patient i3606.

We also observed amino acid insertions within the acidmotif (QEEEEV) of the nef gene. However, these insertionswere less frequent and the inserted amino acids were re-stricted in number and did not change the acid properties of

DIVERSITY OF NEF GENE IN BRAZILIAN CHILDREN 3

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 3

the motif. The only exception was one sequence, obtainedfrom patient d8238, which contained an insertion of six aminoacids within the acid motif (QEEEEV?QEELEAQEDEEV).In general, the function-associated motifs of the nef gene werequite conserved. For example, most sequences of the myr-istoylation motif, MGxxxS, were found to be MGGWKS, withthe exception of sequence 4G645147, containing theMGGNSMS sequence. In addition, the majority of subtype Bstrains contained the CD4 downregulation site CAWLEA,while a few clones presented the sequences VAWLEA (i2135,3G14438, 4G28442, 3G13528, and 4G44655) or LAWLEA(4G28441). One of the five clones from patient i2035 containedthe CD4 downregulation site sequence RAWLEA. BF recom-binant viruses presented either VAWLEA or LAWLEA in thesubtype F portion of the nef sequence. The protease cleavagesite, located within the CD4 downregulation motif(CAW=LEA), remained almost fully conserved; in two se-quences, leucine (L) was replaced by isoleucine (I) or valine(V), with no modification of the motif’s hydrophobic charac-teristics.

Table 1 shows the amino acid frequencies in the MHC-1downregulation motif of the nef gene of distinct HIV-1 sub-types. We compared our interhost sequences with interhostsequences of the subtype B strain from the Los Alamos HIVsequences database. Our examination determined that theconsensus of the MHC-1 downregulation motif of the nef genein all subtype B nef sequences (excluding the Brazilian iso-lates) is R86%A64%E82% (n¼ 1040). Conversely, the consensusat the MHC-1 downregulation site of the nef gene in the Bra-zilian subtype B isolates (which includes the sequences gen-erated in this work, excluding our recombinant sequences,plus the following sequences from GenBank: AY173956,DQ358805, DQ358808, DQ358809, DQ358810, EF637046,EF637047, EF637048, EF637049, EF637050, EF637051,EF637053, EF637054, EF637056, and EF637057) is theR81%T62%E67% (n¼ 64) motif. Interestingly, the most frequentamino acid at the second site of the MHC-1 downregulationmotif (RTE) of the nef gene in the Brazilian isolates is threonine(40=64) and the next most frequent is alanine (24=64). Incontrast, the most common amino acid at this position insubtype B isolates from other countries is alanine (667=1040)

and the second most common is threonine (330=1040). Ad-ditionally, the consensus sequences of the MHC-1 down-regulation motifs of the nef gene from subtype C, D, and Fisolates are R78%T74%E56% (n¼ 549), R87%T88%D47% (n¼ 96),and R81%T75%P94% (n¼ 16), respectively.

Phylogenetic analysis of intrahost nef molecular clones

To establish the phylogenetic relationship between in-trahost clones of the nef gene, a maximum likelihood tree wasconstructed. All molecular nef clones plus HIV-1 group Mreference sequences were included in this analysis. Allalignment gaps were removed and the tree was constructedusing a 615-bp fragment of the nef sequence. As expected, thetree topology shows that clones obtained from each patientclustered in individual groups (gray and black triangles inFig. 1). In addition, most of the samples belonged to sub-type B, and they formed a large cluster that included thesubtype B references. Interestingly, subtype D referencesequences did not form a separate group. Instead, they clus-tered within the main subtype B group, probably due to thelack of sufficient phylogenetic information in this portion ofthe nef gene to distinguish them from subtype B. One of thesequences (i2487) most likely belongs to subtype D, as itdisplayed the same pattern of clustering as observed forsubtype D references. For this reason, we sequenced the V3region from this sample and performed additional phyloge-netic analysis to demonstrate that it was indeed a subtype Bsample (tree not shown). Another sample (d7451) was locatedon the basal node, separated by a long branch from the main Bcluster.

Other sequences (d8238 and i3606) were located within acluster that included subtype F sequences and a CRF_12 ref-erence sequence. One key feature of most of the clones thatclustered within the subtype F clade was extensive amino acidinsertions between sites 25 and 35 in the intrahost align-ment. However, pairwise distances (measured assuming theKimura two-parameter model, data not shown) betweenthese clade F sequences compared with distances betweensequences that clustered within clade B were not substan-tially different (0.106� 0.008 and 0.105� 0.007, respectively).Molecular clones from these individuals formed subclus-ters in the maximum likelihood tree. It is worth noting thatthe maximum likelihood tree was constructed from a gaplessalignment. Thus, subclustering of clade F molecular clonesrelies on the additional diversity found along sequences of thenef gene.

Phylogenetic analysis of interhost samples

To establish the relationship between nef gene sequences atthe population level of our pediatric samples, we performed aphylogenetic analysis of epidemiologically unlinked individ-uals. We again removed all gaps from the alignment, andconstructed the tree using a 615-bp fragment of the nef gene.The same HIV-1 group M references used in the intrahost treewere also used for this analysis. As observed in the earlieranalysis, most samples in the interhost maximum likelihoodtree were classified as subtype B, as they clustered within thelarge clade that included all subtype B references (Fig. 2). Theclustering pattern of this tree also positioned subtype D refer-ences within the main subtype B clade, as seen in the intrahosttree. One sample (4G28670) was located in the basal position of

Table 1. Frequency of Amino Acids at the MHC-1Downregulation Motif (Rxx)

of the Nef Protein of HIV-1 Subtypesa

Subtype A (n¼ 132) Subtype C (n¼ 549)R78 Q42 others11 R427 Q79 others43

T84 A42 others6 T405 A138 others6

P113 others19 E309 others240

Subtype B (n¼ 1040) Subtype D (n¼ 96)R899 Q113 others28 R84 K12

A667 T330 others43 T84 A11 I1

E849 others191 D45 E20 others31

Subtype B Brazil (n¼ 64) Subtype F (n¼ 16)R52 Q10 others2 R13 Q3

T40 A24 T12 A3 P1

E43 others20 P15 S1

aR, arginine; T, threonine; P, proline; Q, glutamine; A, alanine; E,glutamic acid; K, lysine; D, aspartic acid; I, isoleucine; S, serine. HIV-1 sequences of the nef gene were obtained from the Los Alamos HIV-1 database (http:==www.hiv.lanl.gov=).

4 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 4

clade B, separated by a long branch. Two other sequences(4G43674, 4G644657) were located within the cluster formed bythe subtype F references. Interestingly, the subtype B cladefrom both the intrahost and interhost trees contained a variety

of subclusters within the main group. This might indicate dis-tinct introductions of subtype B viruses in Brazil. The tree wasevaluated by bootstrap analysis, and open circles indicatebootstrap values higher than 80%. We were unable to establish

FIG. 1. Maximum likelihood tree constructed with molecular clones of nef from intrahost samples. The tree was constructedwith the nearest-neighbor interchange method, using a 615-bp nef gene fragment. Reference sequences from the main HIV-1strains were included. To facilitate visualization, sequences isolated from each child were collapsed (gray and black triangles)in the tree. @ Symbols in the nodes of the tree indicate bootstrap values higher than 80%. Bootstrap support was obtainedafter 1000 replicates, using a neighbor-joining method.

DIVERSITY OF NEF GENE IN BRAZILIAN CHILDREN 5

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 5

any relationship between the clinical=laboratory profile (i.e.,viral load, CD4 counts) of patients and the structure of theseclusters. Filled triangles indicate sequences that presented anatypical clustering pattern, and were therefore considered asputative recombinant viruses (see below).

Recombination analysis

Since the phylogenetic analysis indicated many sequenceswith conflicting clustering patterns, all nef sequences werechecked for recombination. Our analysis identified 10 re-combinant nef sequences (due to space limitations, sequenceswith similar mosaic patterns were combined into the commonplots, Fig. 3). Three samples (i3606, d8238, and 4G43674) werefound to contain mostly sequence from subtype F, with shortsubtype D fragments between positions 380 and 420. Sincephylogenetic analyses showed that nef gene trees did not

clearly separate subtype D from the main subtype B, the smallfragment imbedded in the nef sequence from samples i3606,d8238, and 4G43674 was likely from the subtype B sequence.Three other samples (d6814, i3493, and 4G8239) displayed aBF mosaic structure, with the subtype F fragment ending nearposition 260 and the remainder of the fragment coming fromsubtype B (Fig. 3b). Samples d7451 and 4G28670 began withthe subtype B sequence until position 400, after which thefragment was composed of the subtype F sequence (Fig. 3c).The mosaic structure of sample 4G4453 is mostly subtype B,with a small subtype F fragment between positions 420 and460 (Fig. 3D). The nef gene sequence of sample 4G644657showed a complex pattern of recombination. The mostupstream region could not be classified, a subtype B frag-ment was identified from position 260 to 410, and the re-mainder of the sequence was classified as a subtype Ffragment (Fig. 3e).

FIG. 2. Maximum likelihood tree of nef sequences from interhost samples. The tree was constructed using the nearest-neighbor interchange method and a 615-bp nef gene fragment. HIV-1 references for the strains were included in the tree. Isolatesidentified by filled triangles are putative recombinants. The tree was rooted at the midpoint. Bootstrap support was obtainedafter 1000 replicates using a neighbor-joining method and nodes with values higher than 80% are denoted by @ symbols.

6 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 6

Selective regimen of the nef gene

Initially, we performed a pairwise analysis to estimate thedN=dS (o) ratio for the clones of each patient. This analysisindicated that the estimated dN=dS ratios were similar amongpatients. Next, we employed a maximum likelihood codon-based approach that uses a collection of models and takes intoaccount the phylogenetic relationship of samples to estimatethe selective regimen. We compared our interhost sequenceswith data sets obtained from the Los Alamos HIV sequencesdatabase. According to estimated likelihoods of each model,the hypothesis of positive selection based on either the M2 orthe M3 model was widely accepted to the detriment of the nullhypothesis of neutral evolution (which is based on the likeli-hood of the M1 model). Specifically, the M2 (selection) modelwas the best-fit model for datasets composed of both HIV-1subtype B sequences from Brazilian children (Children 3G

and Children 4G). This model identified five codons (10, 14,27, 84, and 197) under positive selection in the dataset Chil-dren 3G and two codons (8 and 84) in the data set Children4G. These codons correspond to amino acid positions in ournef gene consensus sequences (based on the alignment ofBrazilian isolates), as shown in Fig. 4. The percentages ofpositively selected sites were 0.64% and 0.58% and the esti-mated mean o values were 3.44 and 2.96 for the Children 3Gand Children 4G datasets, respectively. Interestingly, whensequences generated from all children were combined intoone data set (Children subtype B), the best-fit model indicatedby the LRT was the M3 model (discrete). In this case, a largernumber of positively selected sites were identified.

The analysis of the combined datasets identified 15 codonswith a high posterior probability of positive selection (8, 10,11, 14, 15, 28, 39, 50, 85, 102, 133, 152, 163, 182, and 198). Theproportion of positively selected sites was 0.985% and the

FIG. 3. Recombinant patterns of nef from HIV-1-infected children. Each panel (a through e) represents the most commonpatterns of nef recombination identified in Brazilian isolates. The x-axis represents the sequence length in base pairs (bp). They-axis represents the bootstrap support based on 500 replicates. Analyses were performed using a neighbor-joining methodand Kimura two-parameter model in windows of 140 bp sliding along sequences in increments of 20 bp. The GenBankaccession numbers of the nonrecombinant sequences used as parental references are subtype A (AF004885), subtype B(AY423387, M38432, and AY331295), D (K03454, AY253311, and U88824), G (AF084936 and AF061642), J (AF082394), K(AJ249235 and AJ249239), C (U52953, AF067155, and AY772699), H (AF005496), and F (AF005494). The bootscan analysis wasperformed using the program Simplot v.2.5 (http:==sray.med.som.jhmi.edu=RaySoft). This figure can be seen in color onlineat www.liebertpub.com.

DIVERSITY OF NEF GENE IN BRAZILIAN CHILDREN 7

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 7

mean o value was 2.91. Analysis of the dataset composed ofpediatric HIV-1 subtype C nef sequences indicated that the M2model was the best fit to explain the evolutionary rates ob-served. In this case, the percentage of positively selected siteswas 0.659% and the mean o value was 3.06. The M2 modeldetected four codons (49, 50, 151, and 187) under positiveselection in this dataset. We also analyzed the HIV-1 nef genesequences from infected adults. These sequences were orga-nized into two data sets, one containing subtype C sequences(subtype C Adults) and the other containing subtype B se-quences (subtype B Adults). The maximum likelihood-basedselective regimen analysis indicated the M2 model as the bestfit for both adult datasets. The model detected 0.756% of sitesunder positive selection with a mean o value of 2.76 andseven codons under positive selection (15, 98, 120, 151, 188,192, and 194) in the subtype C dataset. For the subtype Bdataset, the model estimated the percentage of sites underpositive selection as 0.549%, with a mean o value of 4.18 andeight codons under positive selection (9, 10, 11, 12, 14, 15, 85,and 133).

In summary, the intensity of positive selection (mean ovalues) ranged from 2.91 to 3.44 in children and from 2.76 to4.18 in adults. Additionally, the proportion of positively se-lected sites ranged from 0.584% to 0.985% and from 0.549% to0.756% in children and in adults, respectively. Although theM2 and M3 models allow for positive selection, the majority ofcodons in the HIV-1 nef genes from both adults and childrenare under purifying selection, as more than 60% of all sites hado values lower than 1 (dN=dS< 1).

Location of positively selected sitesin the nef gene consensus

To gain deeper insights into the issue of selective pressuresacting on HIV-1 during pediatric infection, we mapped pos-itively selected sites detected by the maximum likelihoodmethod onto the consensus sequences of the nef gene. It hasbeen hypothesized that the adaptive evolution of HIV-1 tothe host is largely caused by immune-mediated positive se-lection against viral proteins.7,60 Additionally, the cytotoxicT-lymphocyte (CTL) response plays a considerable role inrestricting HIV-1 replication in infected individuals.8–13

Based on these observations, we compared the positions ofcodons identified by the maximum likelihood analyses withthe locations of HIV-1 nef epitopes, as described in the LosAlamos National Laboratory Database (http:==www.hiv.lanl.gov=content=immunology=maps=maps.html). Figure 4summarizes the positions of the codons under positive se-lection and the locations of CTL epitopes in the nef gene.Specifically, positively selected codons detected in the pedi-atric subtype B datasets (Children 3G and Children 4G) areindicated by filled triangles and sites detected in the pediatricsubtype C dataset are indicated by open triangles. Sites underpositive selection identified in subtype B adults are indicatedby filled circles and sites in subtype C adults are indicated byopen circles. Horizontal bars above the consensus sequencesindicate nef CTL epitopes (Fig. 4). Since there are CTL epitopesmapped in nearly all regions of nef, most codons under pos-itive selection are located within CTL epitopes. In addition,

FIG. 4. Location of positively selected sites and CTL epitopes in the nef consensus sequence. Codons with the highestprobability of positive selection are represented by triangles and dots as follows: filled triangles represent codons identified inpediatric subtype B nef, open triangle represent codons identified in subtype C children, filled dots represent codons iden-tified in subtype B adults, and open dots represent positively selected codons identified in subtype C adults. Horizontal barsabove the consensus sequences define regions where CTL epitopes have been mapped in the Nef protein. The HLA hap-lotypes that recognize the epitopes are depicted above the horizontal bars. Numbers above the bars denote the followinghaplotypes: (1)¼B*4201, B*4202,B*8101, B35, B7, B81, B*3501, B51, A*0301, A*1101, A2, A2, B27, Cw4, B3; (2)¼B*5801, B57,A2, Cw*0802, Cw8,A*0201, A*0205, A2, B60, B62, Cw*0802, Cw3, Cw8; (3)¼A2, B8, A1, B61, B60, B*4002, B84001;(4)¼CW*0602, CW*0401, B37, A24, B7, B37, CW*0602, B*2705, B27, B18, CW*0701, CW*0702, CW7, B13, B37, CW*0602, B18,CW*0701, CW7, A3; (5)¼CW*0602, CW*0401, B35, A1, B17, B37, B62, B57, A1, A29, B*3701, B*5701, B15, B37, B51, B57, B63,Cw6, A1, B*5801, A29, A*3002, A24, B*5801; (6)¼B7, B57, B58, B63, B*0702, B7, B*0702, B*3501, B*4201, B88101, B35, B7, B35,B57, A*2402, A24, A33, B27, A*2402, A24, B35, B18, A2, B*0702, B*1801, B*5301, B*1517, B57, B63, A*6901, A2. The myr-istoylation motif (MGxxxS), the MHC-1 downregulation motif (RTE) and the CD4 downregulation site (CAWLEA) arehighlighted in the consensus sequence. Positions of the CTL epitopes in the Nef protein recognized by HLA haplotypes wereobtained from the HIV Molecular Immunology 2006: Maps of CTL=CD8þEpitopes Locations Plotted by Protein(http:==www.hiv.lanl.gov=content=immunology=maps=maps.html).

8 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 8

codons under positive selection were distributed along theentire length of the gene. Nevertheless, positively selectedcodons are more frequently found in the extremities of the nefgene.

This finding suggests that the amino and carboxy terminalregions of the translated protein are exposed to heavy selec-tive pressures. Interestingly, we found a cluster of positivelyselected codons in the amino terminal portion of nef, follow-ing the myristoylation motif MGxxxS, a region practicallydevoid of CTL epitopes. In fact, only the human leukocyteantigen (HLA)-A*2501 is located in this region of the nef gene(Fig. 4). In contrast, we detected only a few codons underpositive selection in the hydrophobic regions, where mostCTL epitopes are mapped. Perhaps our most impressive re-sult is that the selective regimen differs, in terms of sitespositively selected, between subtypes B and C. This is par-ticularly clear in the region located after the myristoylationmotif. This region contains a cluster of positively selected sitesin subtype B viruses, but not in subtype C sequences.

Discussion

Nef sequences obtained from HIV-1-infected Brazilianchildren presented great variation in terms of amino acidpolymorphisms and length of amino acid insertions. Wefound a large number of sequences with insertions betweenthe myristoylation motif (MGxxxS) and the MHC-1 down-regulation site, suggesting that this variation does not inter-fere with the pathogenic roles of the protein. In support of thishypothesis, this variation did not correlate with CD4 levels orviral loads in these children. Comparison with other se-quences from the Los Alamos HIV sequences database indi-cated that certain polymorphisms are more often encounteredin some HIV-1 subtypes and not in others. For example,subtype B viruses predominantly contain a CAWLEA se-quence in the CD4 down-regulation site, while the subtype Fportion of BF recombinants contains either the VAWLEA orLAWLEA sequences. Likewise, the MHC-1 downregulationsite (Rxx) displays a specific amino acid composition in eachsubtype. Subtype B contains an RAE motif and subtype Cpresents an RTE motif. Interestingly, nef gene sequences fromBrazilian subtype B strains presented the RTE consensus inthe MHC-1 downregulation site, instead of the RAE sequence(which is the most frequent in subtype B strains from othercountries).

Polymorphisms that may influence disease progressionhave been described in the nef gene of distinct HIV-1 sub-types.42,61–64 However, none of these studies has evaluatedthe impact of these polymorphisms on the evolution of HIV-1.Recently, it has been shown that certain polymorphisms in thenef gene may be associated with slow disease progression inchildren infected with subtype C viruses.35 This study alsodetected higher positive selection in slow progressing chil-dren as compared with rapid progressing children. Ad-ditionally, high positive selection has also been detected in theHIV-1 env gene of adults with slow progression.7,15,16 Theselective regimens may not differ between adults and childrenin terms of the intensity of positive selection.34 Thus, it isunlikely that a strong positive selection or lineage-associatedpolymorphisms in nef could be used to gauge disease pro-gression. Indeed, we found extensive polymorphisms in all ofour evaluated nef sequences.

These polymorphisms did not appear to correlate withvirus loads or CD4 levels. This relationship between nefpolymorphisms and clinical status was additionally verifiedby the analyses of intrahost sequences. We could not identifyany clear relationship at the amino acid level between nefsequences from children with low virus load (group i) orchildren with high virus load (group d) during antiretroviraltherapy. Furthermore, even within a single patient, highamino acid variation was encountered, including differencesin the number of amino acids. For example, patient i3606possessed a number of insertions of 1 to 10 amino acids in theregion following the MHC-1 downregulation site of nef.Similarly, one sequence from patient d8238 contained aninsertion of six amino acids within the acid motif(QEEEEV?QEELEAQEDEEV) of the gene. In addition to theamino acid variation, the nef sequences obtained also dis-played high nucleotide diversity within each patient, rangingfrom 0.002 (�0.001) to 0.032 (�0.008). Our investigation of nefrevealed extensive nucleotide diversity, amino acid poly-morphism, and length variation within each patient, as well asat the population level. Although we have detected insertionsand polymorphisms in the vicinity of nef gene motifs, thesemotifs were extremely conserved in most sequences.

In addition to the insertion in the acid motif of the sequencefrom patient d8238, another exception to the conservation ofmotifs was the sequence from patient 4G645147, which con-tained a peculiar myristoylation motif: MGGNSMS (theconsensus motif is MGGWKS). The CD4 downregulation sitesequence CAWLEA was identified in almost all subtype Bsequences; the exceptions were VAWLEA (i2135, 3G14438,4G28442, 3G13528, and 4G44655), LAWLEA (4G28441), andRAWLEA (clone i20351). The CD4 downregulation motif se-quence in the recombinant samples was either VAWLEA orLAWLEA, which seem to be the signature of subtype F nefgene sequences of the mosaic structure. Furthermore, oneclone from patient d8238 contained the VAWLEA motif andthree clones displayed the LAWLEA sequence, indicating thevariability of the nef gene within this conserved motif, evenwithin a single individual. Interestingly, both leucine (L) andvaline (V) are hydrophobic amino acids, whereas cysteine (C)forms S–S bonds.

It is likely that this type of amino acid replacement mayinterfere with the tridimensional shape of the protein. Ourresults showed that overall variation in the nef gene sequencewas not associated with either CD4 levels or viral loads in thestudied patients. In addition to polymorphisms, insertions,and deletions, another important aspect of nef variability toconsider was the high prevalence of recombinants. Attemptsto associate nef gene variation with disease outcome shouldconsider clade-specific polymorphisms and genome recom-bination as potential confounding factors. In fact, most HIV-1polymorphisms are not necessarily associated with immuneescape. Rather, these shared polymorphisms might be ex-plained solely by the founder effect of HIV-1 lineages.43

Our dN=dS analysis revealed that regions containing themost CTL epitopes were surprisingly poor in codons withevidence of positive selection. On the other hand, the aminoportion of the Nef protein contains a cluster of positively se-lected amino acids (8 to 15) located in a region recognized byonly three CTL haplotypes (A*2501, B8, and B*0801). Inter-estingly, putative epitopes recognized by A*2501, B8, andB*0801 haplotypes are regularly identified elsewhere along

DIVERSITY OF NEF GENE IN BRAZILIAN CHILDREN 9

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 9

the Nef protein. This might indicate that amino acids 8 to 15are located within a region that is poorly recognized by HLAhaplotypes. Alternatively, we believe that the amino portionof the nef gene might contain CTL epitopes that have not yetbeen fully identified. Finally, locations of positively selectedcodons in the Nef protein consensus sequences differed be-tween subtypes B and C.

Previous studies have indicated that codons sharing thesame location may be under different selective regimens,depending on distinct HIV subtypes.65 We identified posi-tively selected codons and compared their positions againstpreviously described CTL epitopes in the nef gene. Positivelyselected codons were dispersed along nef sequences from bothB and C subtypes. However, the amino terminal regionof subtype B nef contained several codons under selectivepressure. Notably, this region contains two important signalsfor HIV-1 pathogenesis (the MHC-1 downregulation signaland the myristoylation motif). Surprisingly, sequences fromsubtype C contained only one site under positive selection inthe same portion of the nef gene. Small differences in posi-tively selected codons have been previously observed insubtype C sequences that evolved in two distinct popula-tions.66 In our study, differences in positively selected codonswere greater because we identified a large number of sites thatare either selected only in subtype B or only in subtype C. Weanalyzed subtype C sequences from India and Africa andfound the same sites under positive selection.

This result suggests that geographic distribution mighthave little impact on the number and location of codons underpositive selection in the nef gene. In addition, nef gene se-quences from subtype B were obtained from Brazilian chil-dren and North American adults, and the location ofpositively selected sites also coincided between these twogroups. Therefore, the main differences observed in thenumber and position of positively selected codons in the nefgene appear to be a hallmark of HIV-1 subtypes, rather thanan indication of the geographic diversification of strains. It istempting to speculate that codons under distinct selectiveregimens in subtype B and C sequences might have distinctadaptive roles in different HIV-1 subtypes. Selective pressuresemerge from complex combinations between immune rec-ognition and viral escape. Certainly, our insights into HIV-1nef selective regimens will benefit from more detailed studies.

In summary, based on the amino acid frequencies in theMHC-1 downregulation motif of the nef gene observed in oursequences, it is likely that Brazilian HIV-1 subtype B has beensubjected to a process of local adaptation to its host popula-tion. Further evidence of distinct evolutionary adaptation wasprovided by the selective pressures that appear to follow asubtype-specific distribution. Consequently, it is possible thatcodons in the nef gene have distinct adaptive fitness roles inHIV-1 subtypes B and C. Finally, we did not find differencesin the positions of positively selected sites in the nef genebetween children and adults. This result disagrees with theidea of differential HIV-1 adaptation due to a distinct selectiveenvironment during pediatric infection.

Sequence Data

All sequences generated in this study have been depositedin the GenBank database under accession numbers FJ822712–FJ822876.

Acknowledgments

We would like to express our sincere gratitude to the chil-dren participating in this study. This work was partiallysupported by the Fundacao de Amparo a Pesquisa do Estado deSao Paulo (FAPESP, Foundation for the Support of Research inthe State of Sao Paulo; Grants 01=11086-6 and 05=53246-0);C.F. was supported by FAPESP (Grant 04=12044-3), E.L. is apostdoctoral researcher supported by FAPESP (Grants04=10372-3 and 07=52841-8). The first three authors contrib-uted equally to this study

Disclosure Statement

No competing financial interests exist.

References

1. Betts MR, Ambrozak DR, Douek DC, et al.: Analysis of totalhuman immunodeficiency virus (HIV)-specific CD4(þ) andCD8(þ) T-cell responses: Relationship to viral load in un-treated HIV infection. J Virol 2001;75:11983–11991.

2. Fausther-Bovendo H, Sol-Foulon N, Candotti D, et al.: HIVescape from natural killer cytotoxicity: nef inhibits NKp44Lexpression on CD4þT cells. AIDS 2009;23:1077–1087.

3. Turk G, Gundlach S, Carobene M, et al.: Single Nef proteinsfrom HIV type 1 subtypes C and F fail to upregulate in-variant chain cell surface expression but are active for otherfunctions. AIDS Res Hum Retroviruses 2009;25:285–296.

4. Chaudhuri R, Mattera R, Lindwasser OW, et al.: A basic patchon alpha-adaptin is required for binding of human immu-nodeficiency virus type 1 Nef and cooperative assembly of aCD4-Nef-AP-2 complex. J Virol 2009;83:2518–2530.

5. Shalekoff S, Meddows-Taylor S, Gray GE, et al.: Identifica-tion of human immunodeficiency virus-1 specific CD8þ andCD4þT cell responses in perinatally-infected infants andtheir mothers. AIDS 2009;27:789–798.

6. Leal ES, Holmes EC, and Zanotto PM: Distinct patterns ofnatural selection in the reverse transcriptase gene of HIV-1in the presence and absence of antiretroviral therapy. Vir-ology 2004;325:181–191.

7. Ross HA and Rodrigo AG: Immune-mediated positive se-lection drives human immunodeficiency virus type 1 mo-lecular variation and predicts disease duration. J Virol2002;76:11715–11720.

8. Draenert R, Verrill CL, Tang Y, et al.: Persistent recognitionof autologous virus by high-avidity CD8 T cells in chronic,progressive human immunodeficiency virus type 1 infec-tion. J Virol 2004;78:630–641.

9. Jamieson BD, Yang OO, Hultin L, et al.: Epitope escapemutation and decay of human immunodeficiency virus type1-specific CTL responses. J Immunol 2003;171:5372–5379.

10. Kalams SA, Buchbinder SP, Rosenberg ES, et al.: Associationbetween virus-specific cytotoxic T-lymphocyte and helperresponses in human immunodeficiency virus type 1 infec-tion. J Virol 1999;73:6715–6720.

11. McMichael AJ and Rowland-Jones SL: Cellular immune re-sponses to HIV. Nature 2001;410:980–987.

12. Musey L, Hughes J, Schacker T, et al.: Cytotoxic-T-cell re-sponses, viral load, and disease progression in early humanimmunodeficiency virus type 1 infection. N Engl J Med1997;337:1267–1274.

13. Sabbaj S, Edwards BH, Ghosh MK, et al.: Human immuno-deficiency virus-specific CD8(þ) T cells in human breastmilk. J Virol 2002;76:7365–7373.

10 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 10

14. Holmes EC, Zhang LQ, Simmonds P, et al.: Convergent anddivergent sequence evolution in the surface envelope gly-coprotein of human immunodeficiency virus type 1 within asingle infected patient. Proc Natl Acad Sci USA 1992;89:4835–4839.

15. Lemey P, Kosakovsky Pond SL, Drummond AJ, et al.: Sy-nonymous substitution rates predict HIV disease progres-sion as a result of underlying replication dynamics. PLoSComput Biol 2007;3:e29.

16. Shankarappa R, Margolick JB, Gange SJ, et al.: Consistentviral evolutionary changes associated with the progressionof human immunodeficiency virus type 1 infection. J Virol1999;73:10489–10502.

17. Wolinsky SM, Wike CM, Korber BT, et al.: Selective trans-mission of human immunodeficiency virus type-1 variantsfrom mothers to infants. Science 1992;255:1134–1137.

18. Mofenson LM, Korelitz J, Meyer WA 3rd, et al.: The rela-tionship between serum human immunodeficiency virustype 1 (HIV-1) RNA level, CD4 lymphocyte percent, andlong-term mortality risk in HIV-1-infected children. NationalInstitute of Child Health and Human Development In-travenous Immunoglobulin Clinical Trial Study Group.J Infect Dis 1997;175:1029–1038.

19. Pillay T and Phillips RE: Adaptive evolution in perinatalHIV-1. Best Pract Res Clin Obstet Gynaecol 2005;19:211–229.

20. Shearer WT, Quinn TC, LaRussa P, et al.: Viral load anddisease progression in infants infected with human immu-nodeficiency virus type 1. Women and Infants TransmissionStudy Group. N Engl J Med 1997;336:1337–1342.

21. McIntosh K, Shevitz A, Zaknun D, et al.: Age- and time-related changes in extracellular viral load in children verti-cally infected by human immunodeficiency virus. PediatrInfect Dis J 1996;15:1087–1091.

22. Buseyne F, Burgard M, Teglas JP, et al.: Early HIV-specificcytotoxic T lymphocytes and disease progression in childrenborn to HIV-infected mothers. AIDS Res Hum Retroviruses1998;14:1435–1444.

23. Chandwani R, Jordan KA, Shacklett BL, et al.: Limitedmagnitude and breadth in the HLA-A2-restricted CD8 T-cellresponse to Nef in children with vertically acquired HIV-1infection. Scand J Immunol 2004;59:109–114.

24. Luzuriaga K, Holmes D, Hereema A, et al.: HIV-1-specificcytotoxic T lymphocyte responses in the first year of life.J Immunol 1995;154:433–443.

25. McFarland EJ, Harding PA, Luckey D, et al.: High frequencyof Gag- and envelope-specific cytotoxic T lymphocyte pre-cursors in children with vertically acquired human immu-nodeficiency virus type 1 infection. J Infect Dis 1994;170:766–774.

26. Sandberg JK, Fast NM, Jordan KA, et al.: HIV-specificCD8þT cell function in children with vertically acquiredHIV-1 infection is critically influenced by age and the state ofthe CD4þT cell compartment. J Immunol 2003;170:4403–4410.

27. Scott ZA, Chadwick EG, Gibson LL, et al.: Infrequent de-tection of HIV-1-specific, but not cytomegalovirus-specific,CD8(þ) T cell responses in young HIV-1-infected infants.J Immunol 2001;167:7134–7140.

28. Cheynier R, Langlade-Demoyen P, Marescot MR, et al.: Cy-totoxic T lymphocyte responses in the peripheral blood ofchildren born to human immunodeficiency virus-1-infectedmothers. Eur J Immunol 1992;22:2211–2217.

29. Goulder PJ, Pasquier C, Holmes EC, et al.: Mother-to-childtransmission of HIV infection and CTL escape through

HLA-A2-SLYNTVATL epitope sequence variation. Im-munol Lett 2001;79:109–116.

30. Leslie A, Kavanagh D, Honeyborne I, et al.: Transmissionand accumulation of CTL escape variants drive negativeassociations between HIV polymorphisms and HLA. J ExpMed 2005;201:891–902.

31. Jin CZ, Zhao Y, Zhang FJ, et al.: Different plasma levels ofinterleukins and chemokines: Comparison between childrenand adults with AIDS in China. Chin Med J (Engl) 2009;122:530–535.

32. Palumbo P, Wu H, Chadwick E, et al.: Virologic response topotent antiretroviral therapy and modeling of HIV dynamicsin early pediatric infection. J Infect Dis 2007;196:23–29.

33. Carvajal-Rodriguez A, Posada D, Perez-Losada M, et al.:Disease progression and evolution of the HIV-1 env gene in24 infected infants. Infect Genet Evol 2008;8:110–120.

34. Leal E, Janini M, and Diaz RS: Selective pressures of humanimmunodeficiency virus type 1 (HIV-1) during pediatricinfection. Infect Genet Evol 2007;7:694–707.

35. Walker PR, Ketunuti M, Choge IA, et al.: Polymorphisms inNef associated with different clinical outcomes in HIV type 1subtype C-infected children. AIDS Res Hum Retroviruses2007;23:204–215.

36. Briant L, Wade CM, Puel J, et al.: Analysis of envelope se-quence variants suggests multiple mechanisms of mother-to-child transmission of human immunodeficiency virus type 1.J Virol 1995;69:3778–3788.

37. Ganeshan S, Dickover RE, Korber BT, et al.: Human immu-nodeficiency virus type 1 genetic evolution in children withdifferent rates of development of disease. J Virol 1997;71:663–677.

38. Pasquier C, Cayrou C, Blancher A, et al.: Molecular evidencefor mother-to-child transmission of multiple variants byanalysis of RNA and DNA sequences of human immuno-deficiency virus type 1. J Virol 1998;72:8493–8501.

39. Wade CM, Lobidel D, and Brown AJ: Analysis of humanimmunodeficiency virus type 1 env and gag sequence vari-ants derived from a mother and two vertically infectedchildren provides evidence for the transmission of multiplesequence variants. J Gen Virol 1998;79(Pt. 5):1055–1068.

40. Dickover R, Garratty E, Yusim K, et al.: Role of maternalautologous neutralizing antibody in selective perinataltransmission of human immunodeficiency virus type 1 es-cape variants. J Virol 2006;80:6525–6533.

41. Edwards CT, Holmes EC, Pybus OG, et al.: Evolution of thehuman immunodeficiency virus envelope gene is dominatedby purifying selection. Genetics 2006;174:1441–1453.

42. Casartelli N, Di Matteo G, Argentini C, et al.: Structuraldefects and variations in the HIV-1 nef gene from rapid,slow and non-progressor children. AIDS 2003;17:1291–1301.

43. Bhattacharya T, Daniels M, Heckerman D, et al.: Foundereffects in the assessment of HIV polymorphisms and HLAallele associations. Science 2007;315:1583–1586.

44. Herbeck JT, Nickle DC, Learn GH, et al.: Human immuno-deficiency virus type 1 env evolves toward ancestral statesupon transmission to a new host. J Virol 2006;80:1637–1644.

45. Walker PR, Pybus OG, Rambaut A, et al.: Comparativepopulation dynamics of HIV-1 subtypes B and C: Subtype-specific differences in patterns of epidemic growth. InfectGenet Evol 2005;5:199–208.

46. Cho YK, Lim JY, Jung YS, et al.: High frequency of grosslydeleted nef genes in HIV-1 infected long-term slow pro-gressors treated with Korean red ginseng. Curr HIV Res2006;4:447–457.

DIVERSITY OF NEF GENE IN BRAZILIAN CHILDREN 11

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 11

47. Galkin AN, Gagarina EY, Lukyanova NS, et al.: Full-lengthgenomic sequencing and analysis of four HIV type 1 subtypeB isolates circulating in the territory of Russia. AIDS ResHum Retroviruses 2006;22:1192–1197.

48. Kirchhoff F, Easterbrook PJ, Douglas N, et al.: Sequencevariations in human immunodeficiency virus type 1 Nef areassociated with different stages of disease. J Virol 1999;73:5497–5508.

49. Jere A, Tripathy S, Agnihotri K, et al.: Genetic analysis ofIndian HIV-1 nef: Subtyping, variability and implications.Microbes Infect 2004;6:279–289.

50. Kumar M, Jain SK, Pasha ST, et al.: Genomic diversity in theregulatory nef gene sequences in Indian isolates of HIV type1: Emergence of a distinct subclade and predicted implica-tions. AIDS Res Hum Retroviruses 2006;22:1206–1219.

51. Ndung’u T, Renjifo B, Novitsky VA, et al.: Molecular cloningand biological characterization of full-length HIV-1 subtypeC from Botswana. Virology 2000;278:390–399.

52. Sanabani S, Neto WK, de Sa Filho DJ, et al.: Full-length ge-nome analysis of human immunodeficiency virus type 1subtype C in Brazil. AIDS Res Hum Retroviruses 2006;22:171–176.

53. Rodenburg CM, Li Y, Trask SA, et al.: Near full-length clonesand reference sequences for subtype C isolates of HIV type 1from three different continents. AIDS Res Hum Retroviruses2001;17:161–168.

54. Thompson JD, Gibson TJ, Plewniak F, et al.: The CLUS-TAL_X windows interface: Flexible strategies for multiplesequence alignment aided by quality analysis tools. NucleicAcids Res 1997;25:4876–4882.

55. Nielsen R and Yang Z: Likelihood models for detectingpositively selected amino acid sites and applications to theHIV-1 envelope gene. Genetics 1998;148:929–936.

56. Yang Z: PAML: A program package for phylogenetic anal-ysis by maximum likelihood. Comput Appl Biosci 1997;13:555–556.

57. Yang Z and Nielsen R: Codon-substitution models for de-tecting molecular adaptation at individual sites along spe-cific lineages. Mol Biol Evol 2002;19:908–917.

58. Rodriguez F, Oliver JL, Marin A, et al.: The general stochasticmodel of nucleotide substitution. J Theor Biol 1990;142:485–501.

59. Swofford DL: PAUP* 4.0 Phylogenetic Analysis Using Par-simony (* and Other Methods). Sinauer Associates, Sun-derland, MA, 2002.

60. Williamson S: Adaptation in the env gene of HIV-1 andevolutionary theories of disease progression. Mol Biol Evol2003;20:1318–1325.

61. Brambilla A, Turchetto L, Gatti A, et al.: Defective nefalleles in a cohort of hemophiliacs with progressing andnonprogressing HIV-1 infection. Virology 1999;259:349–368.

62. Chakraborty R, Morel AS, Sutton JK, et al.: Correlates ofdelayed disease progression in HIV-1-infected Kenyan chil-dren. J Immunol 2005;174:8191–8199.

63. Deacon NJ, Tsykin A, Solomon A, et al.: Genomic structureof an attenuated quasi species of HIV-1 from a bloodtransfusion donor and recipients. Science 1995;270:988–991.

64. O’Neill E, Baugh LL, Novitsky VA, et al.: Intra- and inter-subtype alternative Pak2-activating structural motifs ofhuman immunodeficiency virus type 1 Nef. J Virol 2006;80:8824–8829.

65. Travers SA, O’Connell MJ, McCormack GP, et al.: Evidencefor heterogeneous selective pressures in the evolution of theenv gene in different human immunodeficiency virus type 1subtypes. J Virol 2005;79:1836–1841.

66. Pond SL, Frost SD, Grossman Z, et al.: Adaptation to dif-ferent human populations by HIV-1 revealed by codon-based analyses. PLoS Comput Biol 2006;2:e62.

Address correspondence to:Elcio Leal

Rua Pedro de Toledo781-168 andar

Vila ClementinoSao Paulo, SP Brazil, CEP 04039 032

E-mail: e.leal@unifesp.br

12 CAVALIERI ET AL.

AIDS-2009-0061-Cavalieri_2P.3D 11/05/09 10:42am Page 12