Human Immunodeficiency Virus Mutations during the First Month of Infection Are Preferentially Found in Known Cytotoxic T-Lymphocyte Epitopes (original) (raw)

Abstract

The full protein coding region of human immunodeficiency virus (HIV) genomes were sequenced using plasma collected from nine African-Americans prior to seroconversion and 7 to 28 days later. HIV mutations emerged in seven of these subjects at a genomewide rate of 2% per year. The location of nonsynonymous (NS) HIV mutations within these subjects was compared to their potential HLA-A and B types restricted CTL epitopes reported in the Los Alamos National Laboratory HIV immunology database. A statistically significant (P < 0.005) number of the early NS mutations (13.5%) were found within previously reported CTL epitopes. A virus sequencing and reported CTL epitopes database analysis therefore support a model where a significant proportion of very early nonsynonymous HIV mutations are selected by CTL.

Cytotoxic T-lymphocytes (CTL) play an important role in controlling viremia during primary human immunodeficiency virus (HIV) infection (6, 29). CTL recognize viral peptides presented in a complex with HLA class I molecules at the surface of infected cells, inducing T-cell activation and clonal expansions followed by lysis of infected target cells (37, 40). The resulting selection pressure on the replicating virus can lead to the emergence of CTL escape variants (7, 9, 18-20, 22, 25, 26, 36). CTL targeted epitopes have been reported in every HIV protein in the Los Alamos Database (28). The breadth of HIV-1-specific CTL response increases during chronic versus primary infection (4, 8, 13, 19). CTL responses appear to favor epitopes within regulatory and accessory proteins during primary infection, whereas later responses appear to preferentially recognize structural proteins (1, 8, 9).

CTL escape has been powerfully illustrated by the identification of an immunodominant Mamu-A*01-restricted Tat-specific CTL response rapidly selecting for viral escape variants during acute simian immunodeficiency virus (SIV) infection (3, 32). In the SIV model system, peptide reagents of the infecting SIV strain have allowed all CTL epitopes restricted by various HLA class I allele to be identified. As many as 64% of NS mutations selected over the entire course of SIV infection outside of the envelope gene have been located with known CTL epitopes (33). In HIV-1-infected subjects, viral mutations preventing CTL recognition were described in immunodominant epitopes and were correlated with a steady rise in viral load and disease progression (18, 25).

On a population basis, Moore et al. compared HIV protease and reverse transcriptase sequences of chronically infected patients to their presumably ancestral subtype consensus sequence and showed that consensus deviating sequences were overrepresented in known HLA-restricted CTL epitopes (31). When phylogenetically closely related but epidemiologically unlinked HIV genomes were analyzed positive selection in potential CTL target sites was also detected (35). Particular HLA types have also been associated with susceptibility or resistance to HIV infection (14, 30, 38) as well as with slow or rapid disease progression (10, 11, 16, 17, 21, 23, 24, 39).

To directly identify HIV sequence changes and characterize HIV-1 regions undergoing possible selection very early following infection, we sequenced plasma HIV genomes of nine African-American plasma donors at two time points (Fig. 1). Subjects reported no symptoms of acute infection at the time of plasma collection and were not febrile. The initial donations were p24 antigen negative, anti-HIV-1 antibody negative, and HIV-1 RNA negative (using the National Genetics Institute UltraQual HIV-1 reverse transcription-PCR assay with a detection limit of five HIV-1 genome copies/ml) except for subjects 1057 and PRB959, who had an initial plasma viremia of 3.9 × 104 and 2 × 105 HIV-1 RNA copies/ml, respectively (Fig. 1). Longitudinally collected plasma samples were then quantified with the NGI SuperQuant HIV-1 reverse transcription-PCR assay. Anti-HIV antibodies were detected using the HIV-1 enzyme immunoassay (Abbott, Abbott Park, IL) and p24 antigenemia was detected with the Coulter HIV-1 p24 antigen assay (Beckman Coulter, Brea, CA) (Fig. 1). Antiretroviral therapies were not initiated during the period of study. Because donations were of plasma only, no peripheral blood mononuclear cells were available for functional analyses of the CTL response. These studies were approved by the University of California, San Francisco Committee on Human Research.

FIG. 1.

FIG. 1.

Viral load, p24 antigenemia, and anti-HIV-1 serology of acute infections. Nine African-Americans were characterized as undergoing primary infection. Day 0: first plasma sample with detectable viral RNA using the UltraQual assay or the SuperQuant assay (threshold of 5 and 100 RNA copies/ml, respectively). -/+: negative and positive reactivity, respectively, for p24 and for anti-HIV-1 antibodies.

Seroconversion was observed during the second or third week following initial RNA detection for seven subjects, during week 1 for 1001, and during the fourth week for 1057. The initial plasma donations of subject 1057 and PRB959 were already viremic but subsequent seroconversion confirmed primary infection status.

All open reading frames of plasma HIV-1 RNA from two time points separated by 7 to 28 days (arrows in Fig. 1) were sequenced. For genome amplification plasma HIV-1 RNA was purified, reverse transcribed, and subjected to nested PCR (5). The resulting half-genome amplicons were directly sequenced. The PCR and sequencing protocols have been described in detail (5). Sequences were manually edited using EditView, assembled into a single contig, and aligned using the Seqman and Megalign programs, respectively (DNAstar Inc., Madison, WI). The wide variation in the location of HIV mutations evolving within each subject is shown in Fig. 2.

FIG. 2.

FIG. 2.

Genomewide distribution of early HIV mutations. The vertical bar represent one nucleotide substitution.

Both complete and partial nucleotide sequence changes were observed by direct amplicon sequencing. Partial nucleotide changes were detected when either time point showed a mixed base consisting of two overlapping sequencing electrophoregram peaks (where the minority peak reached ≥25% of the dominant peak) while the other time point showed only a single base peak. These mixed bases are recorded using the IUB nomenclature in the deposited sequences (GenBank accession numbers AY331282 to AY331297, AY332236, and AY332237). Mixed bases, commonly seen when directly sequencing HIV amplicons, reflect polymorphisms in the viral quasispecies that are not yet fixed in the entire viral population. The short time interval between sampled plasma pairs is likely responsible for the frequent detection of such partial nucleotide changes. Partial mutations were analyzed in the same way as full nucleotide substitutions (i.e., positions changing from single into mixed residues, and vice versa, were counted as one mutation). No codon contained more than a single mixed base, and therefore all codons with a mixed base could be translated into only one or two amino acids.

No mutations were observed in subjects 1006 and 1027, who were sampled only 7 and 12 days apart (Fig. 1). For the seven other genome pairs, a total of 107 nucleotide substitutions were detected, resulting in intrasubject mutation frequencies of 0.08% to 0.37%. The accumulation of nucleotide substitutions occurred at an average yearly rate of 2.05%. The genomewide dN/dS ratios, calculated using the SNAP program from the HIV sequence database [www.hiv.lanl.gov (27)], ranged from 0.249 to 0.87, reflecting overall purifying selection against amino acid changes. The gene distribution of NS mutations varied widely between donors. Overall the highest number of NS mutations was located in env (n = 19), pol (n = 15), gag (n = 14), nef (n = 6), vif (n = 3), and vpr (n = 3). In these subjects no NS mutations were detected in the early expressed accessory genes tat, rev, and vpu. When adjusted for gene length, the highest density of NS mutations was found with the vpr, nef, and gag genes (The high mutation rate for the short open reading frame of vpr was the result of a single NS mutation in three subjects [1013, 1018, and 1057] located within Vpr amino acids 34 to 40).

The total number of nucleotide substitutions was correlated with both the time interval between the two sequenced time points and the time interval between the initial detection of HIV RNA and the second sequenced time point (P = 0.05, _r_2 = 0.32 and P = 0.02, _r_2 = 0.48, respectively). The number of nucleotide substitutions was also positively correlated with changes in plasma viremia between the two sequenced time points (P = 0.04, _r_2 = 0.37) indicating that increasing viral loads were associated with a higher number of mutations. Similar trends were observed using NS mutations only (P = 0.09, _r_2 = 0.24; P = 0.006, _r_2 = 0.62; and P = 0.06, _r_2 = 0.3, respectively).

HIV-1 evolution is influenced by numerous virus-host interactions including changes in antibody and CTL selection pressures upon infection of a new host. In order to evaluate the influence of the HLA class I types on the location of early NS mutations, high-resolution HLA-A and -B types were determined by PCR and sequencing of human DNA purified from plasma samples. DNA was isolated in a clean room designed for pre-PCR manipulation using the QIAamp DNA mini kit (QIAGEN, Valencia, CA). Locus-specific PCR amplifications were performed for exons 2 and 3; exon 4 was amplified as required to resolve ambiguities (Perkin Elmer Life Sciences, Shelton, CT) (L. A. Baxter-Lowe, unpublished). Ambiguities involving polymorphism in the leader peptide (HLA-A*7401/02 and B*2705/13) were not resolved, because these mutations are not predicted to alter the peptide binding characteristics of the protein product. These are listed in Table 1 as HLA-A*7401 and HLA-B*2705. For subject 1013, there was insufficient DNA to resolve the exon 4 ambiguity of HLA-A*2301/07N. This has been listed in Table 1 as HLA-A*2301 as HLA-A*2307N is extremely rare. Eighteen different HLA-A and -B types were detected among the seven plasma donors whose HIV evolved between the sampled time points (Table 1). Many of these types are frequently observed in African-Americans (10).

TABLE 1.

High-resolution HLA class I typing

Subject HLA-A HLA-B
1001 0202 0301 0702 1516
1006 2601 6802 1510 1510
1012 6601 6802 4201 5301
1013 2301 3001 4201 5301
1018 6801 7401 1503 5802
1027 3002 3201 2703 2705
1057 0301 2301 0702 1503
1058 2301 6801 5801 5802
PRB959 0308 3001 0705 4501

When published CTL epitopes for these 18 HLA class I molecules were sought in the existing literature only 13 of 18 HLA types had any CTL epitopes described (Los Alamos HIV immunology database, January 2005). A total of 62 HIV CTL epitopes with a range of 1 to 16 underlined epitopes per HLA type were identified. Eight of the 59 HIV NS mutations (13.5%) were located within reported HLA class I restricted epitopes (Table 2). These eight NS mutations, found in three subjects, were within epitopes restricted by one or in two cases by two of their host's class I molecules (Table 2).

TABLE 2.

Mutations located in reported CTL epitopes

Subject no. Position (HxB2)a Defined CTL epitope sequence Subject's HIV sequenceb Mutation Restricting HLA
1001 Env 211 VSFEPIPPHYCA VSFKPIPIHYCA K to E A2 serotype
Env 379 (381) SFNCGGEFF SFNCGGEFF E to G B1516
KNCGGEFFYCNS FNCGGEFFYCNT E to G A2 serotype
Env 501 (504) VKIEPLGVAPTKAKRRVVQR VKIEPLGVAPTRAKRKVVQR K to R A2 serotype
1057 p6 39/Gag 492 YPLTSLRSLF YPLTSLRSLF T to A B7 serotype
p6 42/Gag 494 YPLTSLRSLF YPLTSLRSLF R to K B7 serotype
Vpr 40 FPRIWLHGL FPRIWLHGL H to Y B0702
Nef 184 WRFDSRLAF WKFDSRLAL K to E B1503
DPEKEVLQWK DSEGEVLQWK K to E B7 serotype
1058 Env 588 RYLKDQQLL RYLKDQQLL K to E A2301

We calculated the probability of 8 out of 59 NS mutations occurring within previously reported CTL epitopes for the seven subjects with HIV mutations. All 62 reported epitopes restricted by their hosts' class I HLA-A and HLA-B molecules were compiled from the database. The total numbers of amino acids included within reported CTL epitopes were calculated individually for each of the seven primary infection subjects and then added up (the same amino acids found in overlapping epitopes within the same subject were counted once). A total of 943 amino acids were found within reported CTL epitopes. The total number of amino acids encoded by HIV in these seven subjects was 22,048 (Table 3). The fraction of virally encoded amino acids within reported CTL epitopes was therefore 943/22,048 (4.3%). The proportion of NS mutations found within these reported CTL epitopes was 8/59 (13%), which was significantly greater than expected by chance, considering the low density (4.3%) of reported CTL epitopes (P < 0.005, test of proportion and Fisher’s exact test). This result indicated that NS mutations evolving within the first month of infection were significantly more likely to be found within than outside a reported CTL epitope.

TABLE 3.

Odds ratios of NS mutations within reported CTL epitopea

Subject no. No. of NS mutations No. of NS mutations in reported CTL epitopes % of NS mutations in reported CTL epitopes Total no. of aa in genome Total no. of aa within reported CTL epitopes % of total aa within reported CTL epitopes Odds ratio for NS mutations being within reported CTL epitopes 95% confidence interval
1001 7 3 42.9 3,163 237 7.5 5.7 2.4, 13.6
1012 5 0 0 3,155 134 4.25 0
1013 13 0 0 3,131 92 2.94 0
1018 10 0 0 3,150 88 2.8 0
1057 11 4 36.36 3,147 293 9.3 3.9 1.8, 8.6
1058 8 1 12.5 3,155 80 2.5 4.9 0.8, 31.2
959 5 0 0 3,147 19 0.6 0
Total 59 8 13.5 22,048 943 4.3 3.1 3.0, 11.1

We also calculated separately for each subject and for all subjects together the odds ratio of finding an NS mutation in a predicted CTL epitope (Table 3). Odds ratio values in three subjects were greater than one, ranging from 3.9 to 5.7, reflecting the preferential location of NS mutations within reported CTL epitopes. An odds ratio of 3.1 was calculated using all seven subjects with HIV mutations.

The number of CTL epitopes restricted by the HLA class I molecules of these African American plasma donors in the Los Alamos databases is likely to be a large underestimate of potential epitopes, because many of these HLA types have not been studied in detail (i.e., no CTL epitopes were reported for 5 of the 18 HLA types). Furthermore most studies mapping epitopes have done so using only a subset of HIV proteins, and epitopes have typically been identified using reference or subtype consensus strains rather than autologous strains. Mutational escape from CTL can also occur at different steps during peptide processing and HLA binding. Viral protein degradation by the proteasome, transport of peptides into the endoplasmic reticulum, peptide trimming, and peptide loading may all be affected by sequence changes flanking rather than within the final antigenic peptide (2, 12, 15). Finally, some NS mutations linearly distant from the CTL epitope may also be selected as compensatory mutations to reduce the fitness cost imposed by CTL escape mutations (34).

The immature state of the immune response so early following primary infection may also have limited the amount of CTL selection pressure on the replicating quasispecies although in the SIV model system immunodominant epitope escape mutations have been seen within 4 weeks of infection (3, 32). The percentage of early NS mutations reported here that may affect CTL epitopes may therefore be a minimum value. Indeed we noted that the two subjects with the highest number of HLA-A and -B types restricted CTL epitopes in the database (1001 and 1057) were among the three subjects with an overabundance of NS mutations within reported epitopes (Tables 2 and 3). As more CTL epitopes to less frequent HLA-A and -B types are reported, a greater proportion of very early NS mutations may therefore be found within CTL epitopes.

In summary, the location of NS mutations occurring within the first month of acute HIV infection varied greatly among subjects. The influence of the host's class I HLA type was reflected within weeks of infection by the preferential distribution of 13.5% of NS mutations within previously reported CTL epitopes.

Acknowledgments

Support for this study was provided to E.L.D. by the UC Universitywide AIDS Research Program (ID02-SF-075) and NIAID (AI47320 and AI43261).

REFERENCES