Viral evolution in HLA-B27-restricted CTL epitopes in human immunodeficiency virus type 1-infected individuals

Transcription

1 Journal of General Virology (0), 9, 7 80 DOI 0.099/vir Viral evolution in HLA-B7-restricted CTL epitopes in human immunodeficiency virus type -infected individuals Laurentia C. Setiawan, Esther F. Gijsbers, Adrianus C. van Nuenen and Neeltje A. Kootstra Correspondence Neeltje A. Kootstra n.a.kootstra@amc.uva.nl Department of Experimental Immunology, Sanquin Research, Landsteiner Laboratory, and Center for Infectious Diseases and Immunity Amsterdam (CINIMA), Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands Received 8 January 0 Accepted April 0 The HLA-B7 allele is over-represented among human immunodeficiency virus type -infected long-term non-progressors. In these patients, strong CTL responses targeting HLA- B7-restricted viral epitopes have been associated with long-term asymptomatic survival. Indeed, loss of control of viraemia in HLA-B7 patients has been associated with CTL escape at position in the immunodominant KK0 epitope. This CTL escape mutation in the viral Gag protein has been associated with severe viral attenuation and may require the presence of compensatory mutations before emerging. Here, we studied sequence evolution within HLA-B7-restricted CTL epitopes in the viral Gag protein during the course of infection of seven HLA-B7-positive patients. Longitudinal gag sequences obtained at different time points around the time of AIDS diagnosis were obtained and analysed for the presence of mutations in epitopes restricted by HLA-B7, and for potential compensatory mutations. Sequence variations were observed in the HLA-B7-restricted CTL epitopes IK9 and DR, and the immunodominant KK0 epitope. However, the presence of sequence variations in the HLA-B7-restricted CTL epitopes could not be associated with an increase in viraemia in the majority of the patients studied. Furthermore, we observed low genetic diversity in the gag region of the viral variants throughout the course of infection, which is indicative of low viral replication and corresponds to the low viral load observed in the HLA-B7-positive patients. These data indicated that control of viral replication can be maintained in HLA-B7-positive patients despite the emergence of viral mutations in HLA-B7-restricted epitopes. INTRODUCTION Expression of certain HLA class I alleles is known to be associated with lower viral load and prolonged survival after human immunodeficiency virus type (HIV-) infection, of which HLA-B7 and HLA-B7 are the most studied (Carrington & O Brien, 00; Deeks & Walker, 007; Gao et al., 00; Kaslow et al., 99; The International HIV Controllers Study, 00). CTLs play an important role in the protective effect of these HLA alleles, and the presence of HIV-specific CTLs restricted by HLA-B7 and HLA-B7 in early infection better defines disease progression than HLA genotype alone (Gao et al., 00). However, HIV- is able to escape from CTL responses through selection of mutations that abrogate CTL-mediated killing (Leslie These authors contributed equally to this paper. Present address: Department of Epidemiology and Surveillance of Infectious Diseases, National Institute of Public Health and the Environment, PO Box, 70 BA Bilthoven, The Netherlands. et al., 00). Individuals carrying HLA-B7 show CTL escape from the immunodominant TW0 epitope through the TN CTL escape mutation early after infection. The TN mutation is known to have a negative effect on the replication capacity of the virus, leading to lower viral load (Leslie et al., 00; Martinez-Picado et al., 00). Patients carrying the HLA-B7 allele also have low viral loads for long periods of time, indicative of efficient control of viral replication (Ammaranond et al., 0; Goulder et al., 997). The importance of CTLs in the control of viral replication is underscored by the observation that viral escape from HLA-B7-restricted CTL epitope KK0 has been associated with loss of control of viraemia and disease progression (Ammaranond et al., 0; Appay et al., 00; Feeney et al., 00; Goulder et al., 997; Kelleher et al., 00; Schneidewind et al., 007). The CTL escape mutation in this epitope is located at the R position and is most usually mediated by a mutation toward a lysine (K); however, mutations toward glycine (G), G 0 The Authors Printed in Great Britain

2 HIV- evolution in HLA-B7-positive patients threonine (T) or glutamine (Q) have also been described (Ammaranond et al., 00; Kelleher et al., 00; Nietfield et al., 99; O Connell et al., 00b; Schneidewind et al., 008). It has been hypothesized that the R escape mutation alone does not result in a viable virus and that compensatory mutations are essential before the escape variant can emerge (Kelleher et al., 00). Here, we studied the dynamics of mutations in HLA-B7- restricted CTL epitopes in gag from seven HLA-B7-positive patients during disease progression. Longitudinal gag sequences obtained at different time points during the course of infection and around AIDS diagnosis were analysed for the presence of mutations in HLA-B7-restricted CTL epitopes and potential compensatory mutations. Additionally, we analysed the impact of these mutations on viral load and disease progression. RESULTS Characteristics of HLA-B7-positive patients Seven HLA-B7-positive patients, who were followed longitudinally in the Amsterdam Cohort Studies (ACS), were analysed in the present study. Patient characteristics are summarized in Table. Two patients were already HIV--positive upon entry into the cohort and five patients seroconverted during follow-up. All patients progressed to AIDS without the use of antiretroviral therapy, an increasing viral load and declining number of CD + T-cells were observed over the course of infection (Fig. ). For all patients, serum samples from time points close to disease progression (between years prior and months after AIDS diagnosis) were available for analysis. Table. Patient characteristics AIDS diagnosis was according to the Centers for Disease Control AIDS definition 99 (Centers for Disease Control, 99) or CD + T-cell counts,00 cells ml. S, Date of seroconversion for HIV- antibodies during the active follow-up; E, HIV- + entry into the cohort. NA, Not available. Patient AIDS diagnosis (months after seroconversion or entry into the cohort) HLA allele ACH (S) A*0 A*0 B*7 B*00 ACH (E) A* A* B*7 B* ACH990 (E) NA NA B*70 B*080 ACH989 8 (E) A* A* B*7 B*8 ACH (S) A*007 A*007 B*7 B*0 ACH9 (E) A* A* B*7 B* ACH999 (E) A*0 A* B*7 B* Sequence variation in CTL epitopes restricted by HLA-B7 The Gag region of HIV- present in serum from the patients was analysed. Per time point, between one and gag sequences were generated to study sequence evolution within HLA-B7-restricted CTL epitopes over time. In the viral sequences obtained from patient ACH997, we observed three mutations (EG, RQ and VL) in the EL9 epitope, which were only present in a single viral sequence obtained at a single time point. Of these mutations, only RQ was located at an anchor residue and may have disrupted presentation of the epitope by HLA-B7. Furthermore, a single viral sequence containing a mutation in the DA9 epitope (F00L) was observed at 7 months after seroconversion. However, the viral sequences containing these mutations were not positively selected during the course of infection, indicating that there was no selective advantage for the viral variants. In epitope DR, the R8K mutation was present in all viral sequences obtained at all time points studied. In patient ACH997, the increase in the viral load observed at 77 months after seroconversion was not associated with emergence of mutations in HLA-B7-restricted epitopes in gag (Fig. a). In patient ACH9778, all of the observed viral sequences contained the VL mutation in the EL9 epitope; however, this mutation was not associated with escape from presentation by HLA-B7. Additionally, the L0P mutation, which may represent an escape mutation located at the P9 anchor residue of the HLA-B7 EL9 epitope, was observed in a single viral sequence obtained at months after seroconversion. However, viral variants containing this mutation apparently had no selective advantage throughout the course of infection. In the immunodominant epitope KK0, the L8M mutation was observed in all viral sequences from and 8 months after seroconversion. The emergence of the L8M mutation was observed along with a 0-fold increase of viral load at months after seroconversion, prior to the emergence of the most common RK mutation (Fig. b). In patient ACH990, no viral sequences were detectable in plasma obtained at the first two time points analysed (7 and 8 months after seroconversion). The KS/A mutation in the IK9 epitope was observed in viral sequences obtained at 0 and months after seroconversion. The RG mutation in the KK0 epitope was first observed in a minority of viral sequences at 97 months after seroconversion and was present in all viral sequences obtained at later time points (Table ). Additionally, the I7V mutation in the KK0 epitope was observed in two out of four viral sequences obtained at 97 months after seroconversion; however, this mutation was not positively selected during the course of infection. The R8K mutation in the DR epitope was present in the majority of viral sequences obtained at 0 and months after seroconversion in patient ACH

3 L. C. Setiawan and others (a) ACH997 (b) ACH9778 (c) ACH Viral load [log 0 (copies ml )] CD + T-cell count (cells μl ) Viral load [log0 (copies ml )] CD + T-cell count (cells μl ) Viral load [log0 (copies ml )] 0. 0 CD + T-cell count (cells μl ) (d) ACH989 Viral load [log 0 (copies ml )] CD + T-cell count (cells μl ) (e) ACH Viral load [log 0 (copies ml )] CD + T-cell count (cells μl ) (f) ACH9 Viral load [log 0 (copies ml )] CD + T-cell count (cells μl ) (g) ACH Viral load [log 0 (copies ml )] CD + T-cell count (cells μl ) Fig.. Longitudinal CD + T cell counts (grey) and HIV- RNA viral load measurements (black) for all patients. Arrows indicate the time points at which serum was obtained, and at which viral sequences were detectable (black) and not detectable (grey). Dotted lines indicate time of AIDS diagnosis according to the Centers for Disease Control AIDS definition 99 (Centers for Disease Control, 99) or CD + T cell counts,00 cells ml. The emergence of mutations in the HLA-B7-restricted epitopes was not associated with a clear increase in viral load in this patient (Fig. c). In patient ACH989, no viral sequences could be obtained from the first time point included in this study (70 months after seroconversion). In the IK9 epitope, the R0F mutation was observed in a single viral sequence obtained at 9 months after seroconversion; however, this mutation was not associated with a selective advantage for the virus and was not observed at later time points (Table ). In the KK0 epitope, the RG mutation was first observed in a minority of viral sequences at 9 months after seroconversion and was positively selected in all viral sequences obtained at later time points. Additionally, we observed the L8I and the L8M mutations in the KK0 epitope. The L8I mutation was present in the majority of viral sequences isolated from 8 months after seroconversion, but was not positively selected in the course of the disease. The L8M mutation was only observed at 9 months after seroconversion in three out of six viral sequences. This mutation has previously been associated with early escape from TCR recognition (Lichterfeld et al., 007) and has been described to have a negative impact on replication kinetics of the RG variant, but not the RK variant (Schneidewind et al., 008). Here, mutations L8M and RG were never seen in combination in a single viral sequence. Furthermore, we observed N7T/H mutation in the KK0 epitope at 9 months after seroconversion; these mutations were only observed in a minority of the viral sequences and were not positively selected. In the DR epitope, the mutation R8K was present in three out of four viral sequences observed at 8 months after seroconversion and was only present in the minority of viral sequences obtained at 9 months after seroconversion. Moreover, the emergence of mutations in CTL-restricted epitopes did not result in a strong increase in viral load during follow-up in patient ACH989 (Fig. d). In patient ACH8897, the KR mutation was observed in the majority of viral sequences isolated at months after seroconversion and in one out of two sequences at months after seroconversion in the IK9 epitope. In the immunodominant epitope KK0, the IV mutation in the KK0 was only observed in one out of four viral variants at months after seroconversion. The I7V 7 Journal of General Virology 9

4 HIV- evolution in HLA-B7-positive patients Table. Sequence variation in CTL epitopes restricted by HLA-B7 and at positions 7 and 0 that were associated with RK and RG mutations, respectively NA, Not available. Patient Time after seroconversion (months) Plasma RNA load (ml )* No. of sequences IK9 EL9 KK0 DR DA IRLRPGGKK ERFAVNPGL S E KRWIILGLNK DIRQGPKEPFR DRFYKTLRA ACH / K / K /0... G K / K /9....Q K / K.....L NA / K /......L K / K ACH / L /......L M /......L...P.....M / L... A..K...M ACH / V / G / /...S..... D.G.....K NA /...A..... D.G.....K /...A G ACH / I.....K / D.G NA / M / M.....K / T / H...K /.F G NA / D.G / D.G ACH / T......K / T......K / T....VM.....K /...R.... T....V...T...K /...R.... T....V.....K /...R.... T....VM..T...K / T....V...T...K ACH / M / / M / M / M ACH999 0/ M / M G / M / M / H / M *Plasma load was determined at or within months of the indicated time point of virus isolation. mutation in the KK0 epitope was observed in one out of three viral variants sequenced at 8 months after seroconversion and was positively selected throughout the course of infection (Table ). Moreover, the L8M mutation was observed at 8 months after seroconversion. This mutation was lost at the next time point studied ( months after seroconversion), but was observed again (one out of two viral sequences) at months after seroconversion. Furthermore, the N7T mutation in the KK0 epitope was observed at and months after 7

5 L. C. Setiawan and others seroconversion. The emergence of mutations in patient ACH8897 in the immunodominant KK0 epitope was first detectable in the minority of the viral population at 8 months after seroconversion and observed in all viral variants obtained later in infection after a 00-fold increase in viral load had occurred (Fig. e). Furthermore, all viral sequences obtained from patient ACH8897 carried the R8K mutation in the DR epitope. In patient ACH9, no viral sequences could be obtained at 80 months after seroconversion. At the following time points studied, sequence variations were only observed in the KK0 epitope. The L8M mutation in the KK0 epitope was obtained in the majority of the detectable viral sequences at all time points analysed (Table ). Finally, in patient ACH999, no viral sequence variation was observed in the IK9, EL9 and DR epitopes, whereas a single mutation (R0G) was found in the DA9 epitope in one out of viral sequences obtained at months after seroconversion. The mutation L8M in the KK0 epitope was observed in the majority of viral sequences at all time points. At 0 months after seroconversion, viral sequences with the N7H mutation in the KK0 epitope emerged in three out of sequences, but this mutation was not observed at the later time point (Table ). Compensatory mutations associated with the RK/G mutation The KK0 epitope has been described to be the immunodominant CTL epitope in most HLA-B7-positive patients (Goulder et al., 997). In this study, we showed that six out of seven patients had sequence variations in the KK0 epitope, yet only three patients had a mutation at the previously described position R (Nietfield et al., 99). The mutation at position R has been demonstrated to come at a fitness cost for the virus, which can be compensated by mutation S7A for RK or E0D for RG (Kelleher et al., 00; Schneidewind et al., 007). Indeed, the S7A (patient ACH9778) and E0D (patients ACH990 and ACH989) mutations emerged in viral sequences obtained from time points close to disease progression. In patient ACH9778, the RK mutation was always seen in the presence of the known S7A compensatory mutation, whereas the RG mutation in patients ACH990 and ACH989 was observed alone and in combination with the E0D compensatory mutation. This suggested that the fitness cost associated with the RG mutation was less severe as compared with the RK mutation, confirming previous observations (Schneidewind et al., 007). Viral evolution in gag sequences of HLA-B7 patients Next, we analysed viral evolution in the gag sequences. We compared the number of amino acid changes in Gag relative to the HIV- subtype B consensus sequence in the viral sequences obtained from all patients over time (Fig., Table ). We observed that the viral sequences studied had aa differences as compared with the subtype B Gag consensus, which were located mostly outside of the HLA-B7- restricted CTL epitopes (Table ). The number of amino acid differences relative to the subtype B consensus in Gag did not change significantly throughout disease progression in six out of seven HLA-B7-positive patients included in this study (Fig. ). However, in patient ACH9778, the number of changes in Gag steadily increased over the time points analysed (Fig. ). At 8 months after seroconversion, a significant increase (P0.00) in the total number of amino acid changes in Gag was observed (Fig. ). This increase in amino acid changes coincided with the emergence of the RK mutations in the HLA-B7-restricted epitope KK0. Moreover, we analysed the genetic divergence of viral sequences isolated between consecutive time points (Table ). In all patients, genetic distances between viral isolates obtained from two closest time points were low, ranging No. of amino acid changes ** *** ** ACH997 ACH9778 ACH990 ACH989 ACH8897 ACH9 ACH999 Time after seroconversion (months) Fig.. Number of amino acid changes in Gag relative to the HIV- subtype B consensus sequence in the viral sequences obtained from all patients over time. Statistical significance was assessed by Student s t-test (**P,0.00; ***P,0.000). 7 Journal of General Virology 9

6 HIV- evolution in HLA-B7-positive patients Table. Viral evolution in Gag compared with the subtype B consensus Genetic divergence and dn/ds ratios of sequences were obtained from two consecutive time points. Patient Time after seroconversion or entry (months) No. of amino acid changes No. of epitopes Divergence dn/ds Outside the epitopes Inside the epitopes ACH ACH ACH ACH ACH ACH ACH from to 0.09 %. Additionally, we calculated the ratio of non-synonymous and synonymous mutations (dn/ds ratio) to determine whether evolution was random due to error-prone reverse transcriptase and high viral turnover. A dn/ds ratio v indicates negative selection pressure to retain protein structure. A neutral dn/ds ratio * indicates an equal amount of non-synonymous and synonymous mutations showing no specific selective pressure on the protein. When the dn/ds ratio is w there is a positive selection, indicating pressure on the protein to change amino acid composition to gain fitness or evade immune response. When comparing gag sequences in HLA-B7-positive patients, all viral sequences showed negative selection pressure (0.09vdN/dSv) to retain their protein structure (Table ). DISCUSSION HLA-B7 is enriched amongst HIV controllers compared with non-controllers in the Caucasian population (Bailey et al., 00a; Lambotte et al., 00), and this association is believed to be a consequence of strong CTL responses (Carrington & O Brien, 00; Deeks & Walker, 007; Gao et al., 00). However, in patients carrying this HLA allele, CTL escape in the immunodominant KK0 epitope can occur late in the infection, and has been associated with loss of control of viraemia and disease progression (Ammaranond et al., 0; Appay et al., 00; Feeney et al., 00; Goulder et al., 997; Kelleher et al., 00; Schneidewind et al., 007). Here, we studied sequence evolution in HLA- B7-restricted CTL epitopes in the viral Gag protein during disease progression in seven HLA-B7-positive patients. In agreement with previous reports (Ammaranond et al., 00, 0; Appay et al., 00; Feeney et al., 00; Goulder et al., 997; Kelleher et al., 00; Nietfield et al., 99; O Connell et al., 00b; Schneidewind et al., 007, 008), mutation in the HLA-B7-restricted CTL epitopes was predominant in the immunodominant HLA-B7-restricted KK0 epitope in six out of seven patients. Nevertheless, sequence variations were also observed in the other HLA- B7-specific CTL epitopes. Escape from the immunodominant KK0 CTL response restricted by HLA-B7 via the R position can be accomplished by mutation toward lysine (K), glycine (G), 77

7 L. C. Setiawan and others threonine (T) or glutamine (Q) (Ammaranond et al., 00; Kelleher et al., 00; Nietfield et al., 99; O Connell et al., 00b; Schneidewind et al., 008). All these mutations decrease the binding affinity of the KK0 epitope to the HLA-B7 molecule, thereby limiting epitope presentation and preventing subsequent CTL-mediated lysis. In the present study, we only observed a mutation at this amino acid position in three patients: two patients developed the RG mutation, whereas only one patient harboured the RK mutation. This finding is in agreement with previous observations, where it was demonstrated that the RK/G mutations appear in 0 0 % of the included HLA-B7- positive patients (Ammaranond et al., 00; Cornelissen et al., 009; Goulder et al., 997; Kelleher et al., 00; Schneidewind et al., 007). The emergence of an escape mutation at position R in the KK0 epitope has been associated with an increase of viral load, indicating a loss of viraemic control (Ammaranond et al., 0; Appay et al., 00; Feeney et al., 00; Goulder et al., 997; Kelleher et al., 00; Schneidewind et al., 007). In patient ACH9778, an increase in the viral load was observed * months after seroconversion, whereas the RK mutation emerged 8 months after seroconversion. This indicates that the emergence of the RK mutation is not associated with the increase in viral load in this patient. The emergence of the L8M mutation at months after seroconversion could contribute to the observed increase in viral load at that time point. Moreover, CXCR-using viral variants were first detected at 7 months after seroconversion in this patient and the presence of these CXCR-using viral variants with high replicating capacity may additionally contribute to the increase in viral load (Van t Wout et al., 998). In agreement with this, we also observed an increase in amino acid changes outside of the HLA- B7-restricted epitopes. In patient ACH8897, sequence variations in the KK0 epitope were observed in a minority of the viral sequences obtained 8 months after seroconversion (0 %) prior to 00-fold increase in viral load, and sequence variation in the KK0 epitope was observed in all viral sequences during disease progression of this patient. These data indicate that emergence of mutations in the KK0 epitope might have contributed to disease progression in two out of seven HLA-B7-positive patients included in this study. Furthermore, in patients ACH990 and ACH989, the RG mutations were observed in viral sequences obtained from all time points studied. Although the frequency of viruses carrying the mutation continuously increases over time, the viral load at the time points studied remained stable. Our findings confirm previous observations demonstrating that the emergence of a mutation at R in KK0 does not always result in an increase in plasma viral load (Ammaranond et al., 0; Kelleher et al., 00). In vitro studies have shown that the RK escape mutation severely reduced viral replication capacity in comparison with the RG mutation and that introduction of compensatory mutations can rescue this fitness cost (Kelleher et al., 00; Schneidewind et al., 007). Interestingly, mutations that are able to compensate for the fitness cost associated with either RK or RG are located at different positions, with S7A and E0D serving as compensatory mutations for RK and RG, respectively. Here, we observed that the RK mutation was only seen in combination with the S7A compensatory mutation. However, the emergence of the RG mutation was not always accompanied by the E0D compensatory mutation at the initial emergence of this viral variant in these patients. This could suggest that the fitness cost associated with the RG mutation is less severe as compared with the RK mutation, confirming previous observations (Schneidewind et al., 007). Additionally, we observed other mutations in the KK0 epitope in the viral sequences. However, the presence of these mutations was not associated with an increase in viral load nor disease progression. Although the KK0 CTL epitope is usually immunodominant in patients carrying HLA-B7, here we observed mutation R8K in the DR epitope in the viral sequences studied from four out of seven patients. The DR epitope with the R8K mutation is poorly presented by HLA-B7 according to the Stabilized Matrix Method (Tenzer et al., 00). In two patients, the R8K mutation was only observed at several time points during the course of infection, but did not have a selective advantage. In the two other patients (ACH997 and ACH8897), the R8K mutation was present in all viral variants isolated at all time points, which suggests that this variant might be selected as a result of DR-specific CTLs. However, this mutation is observed in 0 % of the viral sequences obtained from participants of the ACS (data not shown) regardless of HLA typing and therefore it cannot be excluded that this mutation was already present in the virus at the moment of transmission. In two out of seven patients, a mutation in the IK9 epitope emerged during the course of infection. In patient ACH990, a mutation at position (KS/A) was positively selected during the course of infection. A mutation at the same position (KR) was also observed in patient ACH8897 in all viral sequences obtained at months after seroconversion and remained present in the viral population at months after seroconversion. The emergence of mutations in DR and IK9 could also not be linked to an increase in the viral load. Collectively, our data show maintenance of viral suppression despite emergence of mutations in HLA-B7 CTL-restricted IK9, DR or KK0 epitopes, which is in agreement with previous observations (Ammaranond et al., 0; Kelleher et al., 00; O Connell et al., 00a). Similarly, control of virulence was retained despite emergence of CTL escape mutations in HLA-B7-positive elite controllers and recently infected individuals (Bailey et al., 00b; Durand et al., 00; Goonetilleke et al., 009; O Connell et al., 00a). 78 Journal of General Virology 9

8 HIV- evolution in HLA-B7-positive patients Notably, the viral sequences obtained from all patients show very low genetic diversity in gag, which is most likely a result of the low level of viral replication in these patients, allowing fewer errors to be made by the enzyme reverse transcriptase throughout disease progression (Edo-Matas et al., 0). In conclusion, we analysed the sequence variation in HLA- B7-restricted CTL epitopes in longitudinal gag sequences from seven HLA-B7-positive patients during the course of infection and around the time of AIDS diagnosis. Sequence variations were mainly observed in the IK9, KK0 and DR CTL epitopes. However, the emergence of mutations could not be linked to an increase in viral load in the majority of the patients studied. The low genetic diversity and preservation of the Gag protein throughout the course of infection is suggestive of low-level viral replication in the HLA-B7-positive patients, which coincides with the low viral loads observed. Our findings indicate that control of viral replication can be preserved despite the emergence of mutations in the CTL-restricted epitopes in HLA-B7-positive patients. METHODS Patient selection. Seven participants of the ACS, who were followed longitudinally and from whom serum samples were available, were analysed in this study (Table ). All participants had routine monthly visits for blood donation and physical examination. All patients reached AIDS diagnosis [Centers for Disease Control 99 definition (Centers for Disease Control, 99) or CD + T-cell counts v00 cells ml ] without the use of antiretroviral treatment. Selection of serum samples was done on the basis of a viral load w000 copies ml around the time of disease progression (between years before and months after AIDS diagnosis). The ACS have been conducted in accordance with the ethical principles set out in the Declaration of Helsinki and written informed consent was obtained prior to data collection. This study was approved by the Amsterdam Medical Center institutional medical ethics committee. RNA isolation, reverse transcription-pcr and PCR amplification. Viral RNA was isolated from 0 ml serum with a QIAmp Viral Mini kit according to the manufacturer s protocol. Subsequently, viral RNA was transcribed into cdna using M-MLV Reverse Transcriptase (Invitrogen) and the Gag-specific primer gag outer reversed (9-GCCTGTCTCTCAGTAC-9). DNA was amplified by a nested PCR with outer primers gag forward (9-CGACGCAGGACTCGGCTTGCTG-9) and gag outer reversed (9-GCCTGTCTCTCAGTAC-9) and inner PCR primer combinations: gag Bss HII fw (9-TGCTGAAGCGCCCGCACGGC-9) in combination with gag Apa I rev (9-TTCCTAGGGGCCCTGCAA-9). High-fidelity Taq polymerase (Roche) was used for amplification to limit possible amplification errors. Molecular cloning and sequencing. Second-round PCR products were cloned into a pgem-t Easy Vector System (Promega), transformed into competent DHa Escherichia coli (Invitrogen) and plated on LB agar using blue/white screening. White colonies were picked at random. The vector primers T7 and SP were used to amplify the cloned gag regions. PCR products were purified with ExoSap-IT (USB) and sequenced with an ABI Prism BigDye Terminator sequencing kit (Perkin Elmer) on an ABI 0xl DNA sequencer according to the manufacturer s protocol. Sequences were analysed using CodonCode Aligner software. The nucleotide sequences of the gag region were translated and analysed with the BioEdit program (version 7.0.). Statistical analysis of the number of changes in the gag region was done using Student s t-test (GraphPad Prism version.0). Genetic divergence and selection pressure. Genetic divergence between the viral sequences isolated from two successive time points was analysed using the Kimura two-parameter model of evolution (Kimura, 980) in the MEGA software package. Genetic selection pressure on the viruses was analysed by determining the number of non-synonymous (dn) and synonymous (ds) mutations with the SNAP tool from the Los Alamos database website ( (Korber, 000). dn/ds ratios were calculated comparing viral variants isolated from two successive time points. ACKNOWLEDGEMENTS This work was supported by the Aids Fonds Netherlands (grants 0080 and 0008). The Amsterdam Cohort Studies on HIV infection and AIDS, a collaboration between the Amsterdam Health Service, the Academic Medical Center of the University of Amsterdam, Sanquin Blood Supply Foundation, the University Medical Center Utrecht, and the Jan van Goyen Clinic are part of The Netherlands HIV Monitoring Foundation and financially supported by the Center for Infectious Disease Control of The Netherlands National Institute for Public Health and the Environment. We are indebted to all participants for their continuous participation in the study. REFERENCES Ammaranond, P., Zaunders, J., Satchell, C., van Bockel, D., Cooper, D. A. & Kelleher, A. D. (00). A new variant cytotoxic T lymphocyte escape mutation in HLA-B7-positive individuals infected with HIV type. AIDS Res Hum Retroviruses, Ammaranond, P., van Bockel, D. J., Petoumenos, K., McMurchie, M., Finlayson, R., Middleton, M. G., Davenport, M. P., Venturi, V., Suzuki, K. & other authors (0). HIV immune escape at an immunodominant epitope in HLA-B*7-positive individuals predicts viral load outcome. J Immunol 8, Appay, V., Papagno, L., Spina, C. A., Hansasuta, P., King, A., Jones, L., Ogg, G. S., Little, S., McMichael, A. J. & other authors (00). Dynamics of T cell responses in HIV infection. J Immunol 8, 0. Bailey, J. R., Lassen, K. G., Yang, H. C., Quinn, T. C., Ray, S. C., Blankson, J. N. & Siliciano, R. F. (00a). Neutralizing antibodies do not mediate suppression of human immunodeficiency virus type in elite suppressors or selection of plasma virus variants in patients on highly active antiretroviral therapy. J Virol 80, Bailey, J. R., Williams, T. M., Siliciano, R. F. & Blankson, J. N. (00b). Maintenance of viral suppression in HIV--infected HLA-B*7 + elite suppressors despite CTL escape mutations. J Exp Med 0, 7 9. Carrington, M. & O Brien, S. J. (00). The influence of HLA genotype on AIDS. Annu Rev Med,. Centers for Disease Control (99). 99 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Recomm Rep (RR-7), 9. Cornelissen, M., Hoogland, F. M., Back, N. K. T., Jurriaans, S., Zorgdrager, F., Bakker, M., Brinkman, K., Prins, M. & van der Kuyl, A. C. (009). Multiple transmissions of a stable human leucocyte antigen-b7 cytotoxic T-cell-escape strain of HIV- in The Netherlands. AIDS,

9 L. C. Setiawan and others Deeks, S. G. & Walker, B. D. (007). Human immunodeficiency virus controllers: mechanisms of durable virus control in the absence of antiretroviral therapy. Immunity 7, 0. Durand, C. M., O Connell, K. A., Apuzzo, L. G., Langan, S. J., Imteyaz, H., Ahonkhai, A. A., Ceccato, C. M., Williams, T. M., Margolick, J. B. & Blankson, J. N. (00). HIV- Gag evolution in recently infected human leukocyte antigen-b*7 patients with low-level viremia. AIDS, Edo-Matas, D., Lemey, P., Tom, J. A., Serna-Bolea, C., van den Blink, A. E., van t Wout, A. B., Schuitemaker, H. & Suchard, M. A. (0). Impact of CCRdelta host genetic background and disease progression on HIV- intrahost evolutionary processes: efficient hypothesis testing through hierarchical phylogenetic models. Mol Biol Evol 8, 0. Feeney, M. E., Tang, Y., Roosevelt, K. A., Leslie, A. J., McIntosh, K., Karthas, N., Walker, B. D. & Goulder, P. J. (00). Immune escape precedes breakthrough human immunodeficiency virus type viremia and broadening of the cytotoxic T-lymphocyte response in an HLA-B7-positive long-term-nonprogressing child. J Virol 78, Gao, X., Bashirova, A., Iversen, A. K., Phair, J., Goedert, J. J., Buchbinder, S., Hoots, K., Vlahov, D., Altfeld, M. & other authors (00). AIDS restriction HLA allotypes target distinct intervals of HIV- pathogenesis. Nat Med, Goonetilleke, N., Liu, M. K., Salazar-Gonzalez, J. F., Ferrari, G., Giorgi, E., Ganusov, V. V., Keele, B. F., Learn, G. H., Turnbull, E. L. & other authors (009). The first T cell response to transmitted/ founder virus contributes to the control of acute viremia in HIV- infection. J Exp Med 0, 7. Goulder, P. J. R., Phillips, R. E., Colbert, R. A., McAdam, S., Ogg, G., Nowak, M. A., Giangrande, P., Luzzi, G. & Morgana, B. & other authors (997). Late escape from an immunodominant cytotoxic T-lymphocyte response associated with progression to AIDS. Nat Med, 7. Kaslow, R. A., Carrington, M., Apple, R., Park, L., Muñoz, A., Saah, A. J., Goedert, J. J., Winkler, C., O Brien, S. J. & other authors (99). Influence of combinations of human major histocompatibility complex genes on the course of HIV- infection. Nat Med, 0. Kelleher, A. D., Long, C., Holmes, E. C., Allen, R. L., Wilson, J., Conlon, C., Workman, C., Shaunak, S., Olson, K. & other authors (00). Clustered mutations in HIV- gag are consistently required for escape from HLA-B7-restricted cytotoxic T lymphocyte responses. J Exp Med 9, 7 8. Kimura, M. (980). A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J Mol Evol, 0. Korber, B. (000). Computational analysis of HIV molecular sequences. In HIV Signature and Sequence Variation Analysis, pp. 7. Edited by A. G. Rodrigo & G. H. Learn. Dordrecht: Kluwer. Lambotte, O., Boufassa, F., Madec, Y., Nguyen, A., Goujard, C., Meyer, L., Rouzioux, C., Venet, A., Delfraissy, J. F. & other authors (00). HIV controllers: a homogeneous group of HIV--infected patients with spontaneous control of viral replication. Clin Infect Dis, 0 0. Leslie, A. J., Pfafferott, K. J., Chetty, P., Draenert, R., Addo, M. M., Feeney, M., Tang, Y., Holmes, E. C. Allen, T. & other authors (00). HIV evolution: CTL escape mutation and reversion after transmission. Nat Med 0, Lichterfeld, M., Kavanagh, D. G., Williams, K. L., Moza, B., Mui, S. K., Miura, T., Sivamurthy, R., Allgaier, R., Pereyra, F. & other authors (007). A viral CTL escape mutation leading to immunoglobulinlike transcript -mediated functional inhibition of myelomonocytic cells. J Exp Med 0, 8 8. Martinez-Picado, J., Prado, J. G., Fry, E. E., Pfafferott, K., Leslie, A., Chetty, S., Thobakgale, C., Honeyborne, I., Crawford, H. & other authors (00). Fitness cost of escape mutations in p Gag in association with control of human immunodeficiency virus type. J Virol 80, 7. Nietfield, W., Bauer, M., Fevrier, M., Maier, R., Holzwarth, B., Frank, R., Maier, B., Riviere, Y. & Meyerhans, A. (99). Sequence constraints and recognition by CTL of an HLA-B7-restricted HIV- gag epitope. J Immunol, O Connell, K. A., Brennan, T. P., Bailey, J. R., Ray, S. C., Siliciano, R. F. & Blankson, J. N. (00a). Control of HIV- in elite suppressors despite ongoing replication and evolution in plasma virus. J Virol 8, O Connell, K. A., Pelz, R. K., Dinoso, J. B., Dunlop, E., Paik-Tesch, J., Williams, T. M. & Blankson, J. N. (00b). Prolonged control of an HIV type escape variant following treatment interruption in an HLA-B*7-positive patient. AIDS Res Hum Retroviruses, 07. Schneidewind, A., Brockman, M. A., Yang, R., Adam, R. I., Li, B., Le Gall, S., Rinaldo, C. R., Craggs, S. L., Allgaier, R. L. & other authors (007). Escape from the dominant HLA-B7-restricted cytotoxic T-lymphocyte response in Gag is associated with a dramatic reduction in human immunodeficiency virus type replication. J Virol 8, 8 9. Schneidewind, A., Brockman, M. A., Sidney, J., Wang, Y. E., Chen, H., Suscovich, T. J., Li, B., Adam, R. I., Allgaier, R. L. & other authors (008). Structural and functional constraints limit options for cytotoxic T-lymphocyte escape in the immunodominant HLA-B7- restricted epitope in human immunodeficiency virus type capsid. J Virol 8, 9 0. Tenzer, S., Peters, B., Bulik, S., Schoor, O., Lemmel, C., Schatz, M. M., Kloetzel, P. M., Rammensee, H. G., Schild, H. & Holzhütter, H. G. (00). Modeling the MHC class I pathway by combining predictions of proteasomal cleavage, TAP transport and MHC class I binding. Cell Mol Life Sci, The International HIV Controllers Study (00). The major genetic determinants of HIV- control affect HLA class I peptide presentation. Science 0, 7. Van t Wout, A. B., Blaak, H., Ran, L. J., Brouwer, M., Kuiken, C. & Schuitemaker, H. (998). Evolution of syncytium-inducing and non-syncytium-inducing biological virus clones in relation to replication kinetics during the course of human immunodeficiency virus type infection. J Virol 7, Journal of General Virology 9