Evolution of HIV-1 in an HLA-B*57-Positive Patient during Virologic Escape

Some HIV-1-infected individuals, termed “elite suppressors,” maintain viral loads (VLs) of <50 copies of HIV-1 RNA/mL of plasma and normal CD4⁺ T cell counts without antiretroviral therapy [1]. The mechanisms by which these patients control viremia remain unclear. The major histocompatibility complex class I allele group HLA-B*57 is overrepresented in elite suppressors, suggesting a role for the CD8⁺ T cells [1]. Interestingly, although some HLA-B*57-positive individuals suppress viremia early after infection [2], 10% of patients with progressive disease are HLA-B*57 positive [3]. Control of viral replication in some HLA-B*57-positive individuals may involve host immune responses, characteristics of the infecting virus, or both.

We present here the first report of an HLA-B*57-positive patient who spontaneously controlled viremia and then developed virologic escape. Successful virus isolation while the patient's VL was undetectable allowed us to compare this virus with the virus present after virologic escape and analyze the virologic factors associated with control of viremia.

Patient, materials, and methods. The patient was a 35-year-old bisexual male who had a 1-year relationship with an HIV-1-positive male partner who was taking zidovudine, lamivudine, abacavir, efavirenz, and lopinavir/ritonavir. The patient was initially unaware of his partner's HIV status and engaged in unprotected sex. On learning of his partner's status in September 2002, he abstained from further sexual activity and obtained an HIV test, which was negative. A follow-up test obtained on 30 December 2002 was positive. Infection was thus dated at approximately 10/1/02 (day 0). The patient remained asymptomatic and never took antiretroviral therapy. Three months later, his CD4⁺ cell count was 918 cells/mL, and his VL was <400 copies/mL. VLs obtained at months 5 and 7 were <50 copies/mL. VLs obtained at months 12 and 13 were 69 and 477 copies/mL, respectively. At 24 months, his VL was 13,870 copies/mL, and his CD4⁺ cell count was 875 cells/μL.

Virus was obtained from the latent reservoir in resting CD4⁺ T cells as described elsewhere [4]. Replication capacity was determined as described elsewhere [5]. The Phenosense assay (Virologic) was also performed to assess pol function.

Genomic DNA was purified from peripheral blood mononuclear cells (PBMCs), and viral RNA was isolated from plasma or culture supernatants as described elsewhere [4]. Viral long terminal repeats and genes were amplified as described elsewhere [4] (accession numbers EF125548-EF125659).

CCR5 was amplified by polymerase chain reaction from PBMCs and sequenced. Titers of autologous neutralizing antibodies were determined as described elsewhere [6]. HLA-B*57 tetramers were purchased from Beckman Coulter. Staining with tetramers and fluorescein isothiocyanate-conjugated anti-CD8 antibody was performed on cryopreserved PBMCs. Five hundred thousand events were analyzed for each tetramer (by F. Kos, Johns Hopkins and Immunology Core). Interferon (IFN)- g enzyme-linked immunspot assays were performed as described elsewhere [7].

Results. Resting CD4⁺ T cells (35 × 10⁶ cells) were cultured at month 6 when VL was <50 copies/μL. Two independent isolates, 25 and 29, were obtained. The frequency of latently infected cells was 0.02 infectious units per million (IUPM) resting CD4⁺ T cells, much lower than the median frequency in patients receiving highly active antiretroviral therapy (0.5 IUPM).

Sequence analysis of the pol gene from isolates 25 and 29 revealed multiple drug-resistance mutations, including M184V and T215Y in reverse transcriptase (RT) (table 1). These results indicate infection with multiple drug-resistant HIV-1. After virologic escape at month 12, reversion of the M184V mutation to wild type was observed in the plasma virus, and later a Y to D substitution at position 215 of RT appeared in the majority of isolates (table 1).

Because the M184V and T215Y mutations may decrease fitness in the absence of therapy [8, 9], these mutations could have led to lower levels of viremia in this patient. However, in single-round fitness assays using patient-derived pol sequences, the replication capacity of clones 25 and 29 was well within the normal range of HIV-1 isolates (figure 1). A similar replication capacity was obtained by the Phenosense assay (Virologic) (37% and 24% for clones 25 and 29, respectively). This range of replication capacity is comparable to that of isolates with improved fitness due to reversion of M184V or mutation obtained from patients with high VLs [8, 10]. More impor-of T215Y to T215D. As shown in figure 1A, there was no tantly, isolates 25 and 29 replicated as well in vitro, as did 2 consistent difference in replication capacity between isolates 25 isolates (1C and 2C) obtained from an acute seroconverter who and 29 and isolates obtained from the plasma and latent reshad wild-type virus as determined by a commercial genotype ervoir after virologic escape. Growth curves for isolates 1A, 1B, assay and a VL of >750,000 copies/mL (figure 1B) and as well and 2B, obtained from CD4⁺ T cells after breakthrough, were as the standard laboratory strains BAL and IIIB (figure 1C).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

Fitness of virus isolated from study patient. A, Replication capacity of the patient's isolates obtained at different time points. NL4–3 is used as a reference. The replication capacity of the isolates are given as a fraction of that of NL4–3. Clones 25 and 29 were obtained from the latent reservoir at month 6 while the viral load (VL) was still undetectable. The remaining isolates were obtained at 24 months. P2–1 and P2–5 were obtained from plasma. The remaining isolates were obtained from the latent reservoir. The amino acids present at positions 184 and 215 in reverse transcriptase are shown for each isolate (B-D). Growth assays comparing clones 25 (B-D) and 29 (B and C) to HIV-1 isolates obtained from the latent reservoir of an acute seroconverter (AS) who had a VL of 1750,000 copies/mL and wild-type virus (B), to the laboratory strains BAL and IIIB (C), or to isolates (1A, 1B, and 2B) obtained from the patient 24 months after infection when his VL was 13,870 copies/mL (D). E and F, Enzyme-linked immunospot assays looking at the interferon-g response of peripheral blood mononuclear cells (PBMCs) (E) or a cell line specific for wild-type TW10 (F) to different variants of TW10.

We also asked whether the virologic escape was associated with improved fitness due to reversion of M184V or mutation of T215Y to T215D. As shown in figure 1A, there was no consistent difference in replication capacity between isolates 25 and 29 and isolates obtained from the plasma and latent reservoir after virologic escape. Growth curves for isolates 1A, 1B, and 2B, obtained from CD4+ T cells after breakthrough, were similar to that of isolate 25 (figure 1D).

To assess whether breakthrough was associated with changes in other viral genes, we sequenced all genes from isolates 25 and 29 as well as full-length virus isolated from plasma when the VL was 13,870 copies/mL. Sequence analysis revealed no stop codons or major deletions in any open reading frame. As with all primary isolates, there were scattered unique polymorphisms, none of which appeared to explain the initial control or subsequent virologic escape (table 1).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1.

Features of viral isolates.

Many of the polymorphisms in pol were drug-resistance mutations, and, aside from M184V and T215Y, they were maintained after loss of viral suppression. This was also the case for most of the other polymorphisms (table 1). The only novel polymorphism in the genome of clones 25 and 29 that reverted to consensus as VL increased was R37G in Vpu (table 1).

Plasma from month 12 had very low titers (>1:20) of neutralizing antibodies to autologous envelope from clones 25 and 29 (data not shown). HLA typing revealed A*310102, A*6603, B*440302, B*570301, C*07, C*040101, DR-010201, and DR1503. Because HLA-B*57 is associated with elite suppressor status [1], we measured the frequency of CD8⁺ T cells specific for 4 well-characterized B*57 epitopes in Gag using B*57 tetramers at month 7. For 3 of 4 epitopes, tetramer-positive CD8⁺ T cells were readily detectable. The targeted epitopes were ISP-RTLNAW (IW9, 1.1%), TSTLQEQIGW (TW10, 1.1%), and KAFSPEVIPMF (KF11, 1.0%). These levels did not change appreciably at month 20.

To determine whether virologic escape correlated with mutations at these epitopes, we compared the sequence of gag from the initial isolates, clones 25 and 29, obtained at month 6, with that of plasma isolates obtained during and after breakthrough. Interestingly, the TW10 epitope (Gag 241–249) in clones 25 and 29 had alanine at position 248. In contrast, a month later, when the VL was still <50 copies/mL, 8 of 10 clones from PBMCs contained glycine, the amino acid found most frequently in this position in clade B isolates. Subsequently, when the VL was 69 copies/mL, there was a striking replacement of glycine/alanine at this position with glutamic acid (13/13 clones) (table 1). Simultaneously, 8 of 13 clones developed an alanine to proline mutation in the position preceding the IW9 epitope. This is a well-characterized processing escape mutation of the IW9 epitope [11]. It was seen in all 12 isolates obtained from plasma at month 24 when the VL was 13,870 copies/mL. At that time, further variation in the TW10 epitope was noted, with an aspartic acid at position 248 becoming dominant.

As shown in figure 1E, PBMCs from month 20 had a stronger IFN-g response to wild-type TW10 epitope than to variants with alanine, glutamic acid, or aspartic acid at position 248. It is thus most likely that the patient was infected with virus that had glycine at position 248 and that the G248A substitution in isolates 25 and 29 represents an early escape mutation. The fact that a cell line specific for wild-type TW10 had diminished responses to variants containing G248A, G248E, and G248D is consistent with this hypothesis (figure 1F). The shift from G248A to G248E probably represents further evolution of the virus in response to selective pressure on the TW10 epitope. The breadth of the patient's immune response before virologic escape is unknown, but HLA-B*57-restricted epitopes are dominant in acute infection [2]. The development of escape mutations in 2 of these epitopes could thus have been critical. No mutations developed in the KF11 epitope, but immune responses to this peptide alone are neither necessary [7] nor sufficient [3] for the control of viremia in HLA-B*57-positive patients. No new mutations developed in other genes in known class I epitopes restricted by the patient's HLA alleles. Taken together, these results suggest that the loss of virologic control in this individual was temporally associated with amino acid changes in HLA-B*57-restricted epitopes in Gag.

Discussion. Viral escape from cytotoxic T lymphocytes has been described in HIV/simian immunodeficiency virus infection but has rarely been associated with loss of control of viral replication [12–14]. We isolated 2 replication-competent HIV1 clones from this patient when his VL was undetectable and compared the sequences with plasma virus after virologic escape.

Reversion of the M184V mutation in RT and the R37G mutation in Vpu was seen as VL increased. Although the M184V lamivudine resistance mutation has some fitness cost [8, 9], and although Vpu enhances virion release [15], reversion of these mutations to consensus was not associated with a marked increase in in vitro replication capacity. Given that no deficits in gene function due to R37G, M184V, or other polymorphisms present in isolates 25 and 29 were detectable in vitro, their effect on viral fitness may be relatively minor and insufficient to explain

initial suppression to <50 copies/mL. Reversion of these mutations to consensus may nevertheless have contributed to virologic breakthrough in this individual.

It is likely that the mutations in HLA-B*57-restricted epitopes were the major cause of the virologic escape. Substitutions were present in the TW10 epitope in both isolates recovered when the VL was still undetectable. These substitutions may represent early escape mutations that preceded the actual virologic escape. This is the first time the development of escape mutations has been temporarily associated with breakthrough viremia in a patient who spontaneously maintained a VL of <50 copies/μL.

In summary, this HLA-B*57-positive male, infected with replication-competent, multiple drug-resistant X4-tropic clade B virus who controlled viral replication for ∼1 year before virologic escape. This is the first study that has analyzed the entire HIV-1 genome before and after virologic escape. The results suggest that HLA-B*57-restricted CD8⁺ T cell responses play a key role in the control of viremia. Mutations that may have subtle effects on viral fitness may also play a significant role.

• Received December 13, 2006.
• Accepted February 1, 2007.

• © 2007 by the Infectious Diseases Society of America