HIV Evolution: CTL Escape Mutation and Reversion After Transmission

A J Leslie; K J Pfafferott; P Chetty; R Draenert; M M Addo; M Feeney; Y Tang; E C Holmes; T Allen; J G Prado; M Altfeld; C Brander; C Dixon; D Ramduth; P Jeena; S A Thomas; A St John; T A Roach; B Kupfer; G Luzzi; A Edwards; G Taylor; H Lyall; G Tudor-Williams; V Novelli; J Martinez-Picado; P Kiepiela; B D Walker; P J R Goulder

Disclosures

Nat Med. 2004;10(3) 

In This Article

Abstract and Introduction

Within-patient HIV evolution reflects the strong selection pressure driving viral escape from cytotoxic T-lymphocyte (CTL) recognition. Whether this intrapatient accumulation of escape mutations translates into HIV evolution at the population level has not been evaluated. We studied over 300 patients drawn from the B- and C-clade epidemics, focusing on human leukocyte antigen (HLA) alleles HLA-B57 and HLA-B5801, which are associated with long-term HIV control and are therefore likely to exert strong selection pressure on the virus. The CTL response dominating acute infection in HLA-B57/5801-positive subjects drove positive selection of an escape mutation that reverted to wild-type after transmission to HLA-B57/5801-negative individuals. A second escape mutation within the epitope, by contrast, was maintained after transmission. These data show that the process of accumulation of escape mutations within HIV is not inevitable. Complex epitope- and residue-specific selection forces, including CTL-mediated positive selection pressure and virus-mediated purifying selection, operate in tandem to shape HIV evolution at the population level.

Evasion of the host CTL response through mutation of key epitopes is a major challenge to natural or vaccine-induced immune control of HIV.[1,2,3,4] This phenomenon of CTL escape mutation in immunodeficiency virus infection has been well documented.[1,2,3,4,5,6,7,8,9,10] Recent studies suggest that, through escape mutation, CTL has a major role in driving evolution of HIV.[5,9,10] However, the extent to which this actually occurs remains to be determined. It is clear that CTL escape mutations are common, and are selected by CTL responses restricted by a wide array of different HLA molecules.[9] However, the question of whether these mutations are transmissible and then stable in the absence of the HLA allele that drove their selection has yet to be addressed. These two parameters need to be established to quantify the role of the CTL response in driving HIV evolution at a population level. It is also of vital importance in terms of vaccine design to determine which immune responses will remain relevant and which will be irrevocably lost through escape mutations.

To address this question, we sought examples of HLA-B57- and HLA-5801-restricted CTL escape, and observed the outcome of transmission of these mutations to HLA-B57/5801-negative individuals. These HLA alleles were chosen because HLA-B57 and the closely related HLA-B5801 are associated with effective control of HIV[11,12,13,14,15] (P.J.R.G. et al., unpublished data). We therefore hypothesized that the HLA-B57/5801-restricted CTL response represents a strong selection force that would be associated with escape mutation. We focused here on the Gag polyprotein, as this is a highly immunogenic region of HIV-1 (refs. 16-18).

HLA-B57/5801

Gag sequences from 311 subjects with chronic HIV infection were analyzed. Variation in the epitope TSTLQEQIAW (designated TW10; Gag HXB2 residues 240-249) was observed in association with expression of HLA-B57 or HLA-B5801 ( Table 1 ). TW10 dominates the CTL response in acute HIV infection in HLA-B57 individuals,[19] and is presented by the closely related HLA-B5801 allele.[14] The dominant change in TW10 is a substitution of Thr by Asn at residue 242 (T242N). This mutation occurs in C-clade infections, arising in 84% and 63% of HLA-B57-positive and HLA-B5801-positive subjects, respectively, and in 0% of B57/5801-negative subjects (P < 0.0001 in each case). Data from B-clade-infected subjects reveal a similar picture ( Table 1 ). Again, the T242N mutation is highly prevalent in HLA-B57-positive and HLA-B5801-positive subjects (79% and 93%, respectively), and is not found in any B57/5801-negative individuals (P < 0.0001 in each case).

An interclade difference exists within the TW10 epitope at residue 248, where the consensus is Gly (G) in B-clade strains and Ala (A) in C-clade strains. In C-clade-infected subjects, variation at this position is weakly associated with HLA-B57 only (P = 0.03). By contrast, there is a high degree of variation at this position in B-clade-infected subjects, predominantly involving the G248A change, again strongly associated with HLA-B57 (P < 0.0001) but not HLA-B5801. Unlike T242N, G248X is found in substantial numbers in HLA-B57/5801-negative individuals ( Table 1 ). Together, these results indicate that TW10 may be under immune selection pressure that results predominantly in a T242N substitution in both B-clade and C-clade infection, as well as an additional G248X substitution, particularly in B-clade infections.

To show that the TW10 variants arise after transmission to HLA-B57/5801-positive subjects, we sought examples of vertical and horizontal transmission in which HLA-B57/5801-negative donors, infected with wild-type virus, transmit to subjects expressing HLA-B57 or HLA-B5801. Six such mother-child transmission pairs were identified, with viral sequences obtained from all six mothers encoding wild-type TW10 and from five of the six HLA-B57/5801-positive children carrying the T242N and/or G248A mutations ( Table 1 ). In the two adult pairs in which the direction of horizontal transmission was unambiguous, only the HLA-B57/5801-positive subjects possessed the T242N mutation. Likewise, in three additional adult pairs in which the direction of horizontal transmission was ambiguous, only virus isolated from the HLA-B57/5801-positive subjects possessed the T242N mutation ( Table 1 ). In this last group, the data could indicate either transmission of wild-type virus to the B57/5801-positive individual followed by escape, or transmission of the T242N mutant virus followed by reversion to wild type in the HLA-B57/5801-negative recipient. The authenticity of all of these transmission pairs was confirmed by phylogenetic analysis (Supplementary Fig. 1 online).

To determine whether TW10 is under positive selection from HLA-B57/5801-restricted CTLs, we examined selection pressures using a maximum-likelihood method.[20] This method revealed nine sites under positive selection when B57/5801-positive and non-B57/5801 subjects were grouped together (Supplementary Table 1 online), including residues 242 (mean ratio of nonsynonymous to synonymous nucleotide substitutions (d N/d S) = 3.216) and 146 (mean d N/d S = 3.215). Both results were highly significant, with Bayesian posterior probabilities of 1.000 and 0.999, respectively. When HLA-B57/5801-positive subjects were excluded from the analysis, only residues 242 and 146 were no longer identified as being under positive selection, suggesting that selection at these sites is operating through the HLA-B57/5801 alleles. The same analysis conducted using the B-clade sequence data similarly identified residues 242 and 146 as being under positive selection (mean d N/d S = 2.34 and 2.23, respectively).

B57/5801 CTL Escape Mutants

The optimal epitope TW10 (TSTLQEQIGW) was previously defined in B-clade-infected subjects expressing HLA-B57 or B5801 (ref. 14). TW10 (TSTLQEQIAW) was confirmed as the optimal epitope in C-clade infection (Fig. 1a). Recognition of the commonly occurring variants at residues 242 and 248 was evaluated in B57- and B5801-positive subjects with B- or C-clade infection, respectively (Fig. 1b-e). The pattern of recognition was similar in each case, with reduced recognition equivalent to 1-2 logs peptide concentration of variants with single amino-acid substitutions, and complete abrogation of a response to peptides bearing double mutations. Thus, these HLA-B57- and B5801-associated mutations within TW10, which dominate the viral population in chronically infected individuals expressing these alleles, represent escape mutations.

TW10 variants are escape mutations. IFN- production was recorded as spot-forming cells (SFCs) per million PMBCs. Results shown are representative of assays using PBMCs from seven subjects expressing either HLA-B57 or HLA-B5801. (a) Recognition of TW10 and suboptimal 9-mer and 11-mer overlapping peptides using PBMCs from C-clade-infected donor A-005-M (HLAB57/7). (b-d) TW10 variant recognition. (b) HLA-B57-positive, C-clade-infected subject A-005-M (c), HLA-B57-positive, B-clade-infected subject SWS (HLA-B57; d), HLA-B5801-positive, C-clade-infected subject PS-032-M (B5801; e) and HLA-B5801-positive, B-clade-infected subject, 90-92851-RI (B5801). Sequence of wild-type TW10 is shown in full in each key.

TW10 variants are escape mutations. IFN- production was recorded as spot-forming cells (SFCs) per million PMBCs. Results shown are representative of assays using PBMCs from seven subjects expressing either HLA-B57 or HLA-B5801. (a) Recognition of TW10 and suboptimal 9-mer and 11-mer overlapping peptides using PBMCs from C-clade-infected donor A-005-M (HLAB57/7). (b-d) TW10 variant recognition. (b) HLA-B57-positive, C-clade-infected subject A-005-M (c), HLA-B57-positive, B-clade-infected subject SWS (HLA-B57; d), HLA-B5801-positive, C-clade-infected subject PS-032-M (B5801; e) and HLA-B5801-positive, B-clade-infected subject, 90-92851-RI (B5801). Sequence of wild-type TW10 is shown in full in each key.

TW10 variants are escape mutations. IFN- production was recorded as spot-forming cells (SFCs) per million PMBCs. Results shown are representative of assays using PBMCs from seven subjects expressing either HLA-B57 or HLA-B5801. (a) Recognition of TW10 and suboptimal 9-mer and 11-mer overlapping peptides using PBMCs from C-clade-infected donor A-005-M (HLAB57/7). (b-d) TW10 variant recognition. (b) HLA-B57-positive, C-clade-infected subject A-005-M (c), HLA-B57-positive, B-clade-infected subject SWS (HLA-B57; d), HLA-B5801-positive, C-clade-infected subject PS-032-M (B5801; e) and HLA-B5801-positive, B-clade-infected subject, 90-92851-RI (B5801). Sequence of wild-type TW10 is shown in full in each key.

TW10 variants are escape mutations. IFN- production was recorded as spot-forming cells (SFCs) per million PMBCs. Results shown are representative of assays using PBMCs from seven subjects expressing either HLA-B57 or HLA-B5801. (a) Recognition of TW10 and suboptimal 9-mer and 11-mer overlapping peptides using PBMCs from C-clade-infected donor A-005-M (HLAB57/7). (b-d) TW10 variant recognition. (b) HLA-B57-positive, C-clade-infected subject A-005-M (c), HLA-B57-positive, B-clade-infected subject SWS (HLA-B57; d), HLA-B5801-positive, C-clade-infected subject PS-032-M (B5801; e) and HLA-B5801-positive, B-clade-infected subject, 90-92851-RI (B5801). Sequence of wild-type TW10 is shown in full in each key.

TW10 variants are escape mutations. IFN- production was recorded as spot-forming cells (SFCs) per million PMBCs. Results shown are representative of assays using PBMCs from seven subjects expressing either HLA-B57 or HLA-B5801. (a) Recognition of TW10 and suboptimal 9-mer and 11-mer overlapping peptides using PBMCs from C-clade-infected donor A-005-M (HLAB57/7). (b-d) TW10 variant recognition. (b) HLA-B57-positive, C-clade-infected subject A-005-M (c), HLA-B57-positive, B-clade-infected subject SWS (HLA-B57; d), HLA-B5801-positive, C-clade-infected subject PS-032-M (B5801; e) and HLA-B5801-positive, B-clade-infected subject, 90-92851-RI (B5801). Sequence of wild-type TW10 is shown in full in each key.

B57/5801

We noted above that T242N is seen in 0 of 187 HLA-B57/5801-negative subjects with chronic HIV infection. The most likely explanation for this is that T242N mutant viruses are not transmitted, or that the T242N mutant is outcompeted by wild-type virus ('reverts') after transmission to B57/5801-negative individuals. To determine whether T242N is transmitted, instances of transmission from HLA-B57/5801-positive subjects to individuals lacking these alleles were sought. In one such case of intrapartum mother-to-child transmission (Fig. 2a), the B-clade-infected mother (SMH-05M; HLA-B57-positive) carried both the T242N and G248A mutations (46 of 46 clones) at 5 months, 7 months and 8 years postpartum, indicating that both mutations are stable in a B57/5801-positive individual for this duration. In the child (SMH-05C; HLA-B7 homozygous) at 2 months postpartum, all clones possessed the T242N and G248A mutations, confirming that both mutations were transmitted. By 5-7 months of age, the frequency of the T242N mutation declined to ~50%, and by 5 years none of the 16 clones isolated from plasma RNA possessed the T242N mutation. ( Table 2 and Fig. 2a). Although 3 of 17 clones isolated from proviral DNA at 8 years of age encoded T242N, phylogenetic analysis confirmed that these are closely related to the transmitted virus, and thus represent archival sequences (Fig. 2b). In contrast to T242N, the G248A mutation was present in all SMH-05C clones from all time points, suggesting that G248A is stable (does not revert) in the absence of the selecting HLA allele.

Reversion of TW10 variants after transmission to HLA-B57/5801-negative subjects. (a,c,d) Viral loads of three subjects, SMH-05C (a), 997C (c) and AC-33 (d), showing intrapartum mother-to-child, postpartum mother-to-child and horizontal transmission, respectively. (b) Maximum-likelihood phylogenetic tree of SMH-05-Mother and SMH-05-Child proviral DNA clones. Blue, SMH-05-Child clones from 2 months (C2), 5 months (C5) and 8 years (C8Y); pink, SMH-05-Mother clones from 5 months (M5) and 8 years (M8Y); T242N mutation is mapped onto the tree. Lineages in which ancestral state is equivocal are highlighted in gray. Yellow arrows indicate major sublineages supported by high bootstrap values.

Reversion of TW10 variants after transmission to HLA-B57/5801-negative subjects. (a,c,d) Viral loads of three subjects, SMH-05C (a), 997C (c) and AC-33 (d), showing intrapartum mother-to-child, postpartum mother-to-child and horizontal transmission, respectively. (b) Maximum-likelihood phylogenetic tree of SMH-05-Mother and SMH-05-Child proviral DNA clones. Blue, SMH-05-Child clones from 2 months (C2), 5 months (C5) and 8 years (C8Y); pink, SMH-05-Mother clones from 5 months (M5) and 8 years (M8Y); T242N mutation is mapped onto the tree. Lineages in which ancestral state is equivocal are highlighted in gray. Yellow arrows indicate major sublineages supported by high bootstrap values.

Reversion of TW10 variants after transmission to HLA-B57/5801-negative subjects. (a,c,d) Viral loads of three subjects, SMH-05C (a), 997C (c) and AC-33 (d), showing intrapartum mother-to-child, postpartum mother-to-child and horizontal transmission, respectively. (b) Maximum-likelihood phylogenetic tree of SMH-05-Mother and SMH-05-Child proviral DNA clones. Blue, SMH-05-Child clones from 2 months (C2), 5 months (C5) and 8 years (C8Y); pink, SMH-05-Mother clones from 5 months (M5) and 8 years (M8Y); T242N mutation is mapped onto the tree. Lineages in which ancestral state is equivocal are highlighted in gray. Yellow arrows indicate major sublineages supported by high bootstrap values.

Reversion of TW10 variants after transmission to HLA-B57/5801-negative subjects. (a,c,d) Viral loads of three subjects, SMH-05C (a), 997C (c) and AC-33 (d), showing intrapartum mother-to-child, postpartum mother-to-child and horizontal transmission, respectively. (b) Maximum-likelihood phylogenetic tree of SMH-05-Mother and SMH-05-Child proviral DNA clones. Blue, SMH-05-Child clones from 2 months (C2), 5 months (C5) and 8 years (C8Y); pink, SMH-05-Mother clones from 5 months (M5) and 8 years (M8Y); T242N mutation is mapped onto the tree. Lineages in which ancestral state is equivocal are highlighted in gray. Yellow arrows indicate major sublineages supported by high bootstrap values.

A second mother-to-child transmission pair, 997-Mother (HLA-B18/57) and 997-Child (HLA-B18/42), was identified ( Table 2 and Fig. 2c). This pair was C-clade infected. T242N was present in the mother antenatally and at all time points postpartum, as expected. In the child, who was infected 3-9 months postpartum, 0 of 22 clones sequenced from RNA had the T242N mutation at the 9-month time point. These data again support the hypothesis that the T242N mutation reverts in the absence of HLA-B57/5801.

Analysis of 187 chronically infected HLA-B57/5801-negative individuals revealed none with virus expressing the T242N mutation. In contrast, virus expressing the T242N mutation was isolated from 4 of 19 HLA-B57/5801-negative subjects with acute HIV infection ( Table 2 ). This further demonstrates that T242N is transmitted horizontally, and is in itself evidence that T242N reverts at some point between acute and chronic infection in HLA-57/5801-negative individuals in whom the transmitted virus carries the T242N mutation.

One of these acutely infected individuals, AC-33, was studied longitudinally, revealing that T242N also reverts over time in this situation ( Table 2 and Fig. 2d). This subject was placed on antiretroviral therapy (ART) immediately after diagnosis of HIV infection. During ART, viral replication was suppressed to undetectable levels (<50 copies/ml). One might thus anticipate that T242N reversion would take longer in this setting than in patient SMH-05C, who did not receive ART until 4 years after infection, or in patient 997-C, who never received ART.

B57/5801
'Footprints'

To further substantiate the hypothesis that T242N undergoes reversion in the absence of HLA-B57/5801, we next sought T242N-linked mutations that are stable in the absence of B57/5801. The term 'footprints'[21] was previously coined to describe sequence polymorphisms associated with particular HLA alleles. Thus, the presence of characteristic polymorphisms can indicate the impact of specific HLA alleles on HIV. In HLA-B57/5801-negative individuals, the association of such footprints of the T242N mutation with variation in TW10, which we have shown is strongly associated with HLA-B57/5801, would support the hypothesis that the ancestral virus in these subjects originated in HLA-B57/5801-positive individuals (Fig. 3). Two sites were identified at which variation from the consensus sequence was associated with T242N in the HLA-B57/5801-positive group: H219X (X = Q, P or R; P < 0.0001) and A146X (X = P, T, S, V, N or H; P = 0.0008). In the case of A146P, this polymorphism represents a processing escape mutation (R.D. et al., unpublished data). In the case of H219X, no HLA-B57 or B5801-restricted epitope has been identified for this region,[14,18,22] and there is no potential B57 or B5801-restricted epitope that would satisfy the defined HLA-B57/5801-binding motif.[23]

Patterns of HLA-B57/5801-associated mutation and reversion before and after transmission. (1) Transmission of wild-type virus to HLA-B57- or HLA-B5801-positive subject, in whom the T242N mutation arises early. (2) Accumulation of HLA-B57/5801- and T242- linked mutations. (3) Transmission of HLA-B57/5801-adapted virus to B57/5801-negative individual. (4) T242N mutation reverts, whereas stable B57/5801- and T242N-linked mutations remain as footprints of the HLA-B57/5801 allele. Shown are structural features of N-terminal portion (residues 131-270) of p24 Gag: -helices 1-7 (h1-h7), cyclophilin A binding loop, and HLA-B57/5801-restricted epitopes ISPRTLNAW (ISW9; Gag residues 147-155) and TSTLQEQIGW (TW10; residues 240-249).

The H219X variant appears subsequent to the development of the T242N mutation. In B57/5801-positive subjects, H219X is seen only in the presence of T242N (32 of 32 instances; Table 3 ), compared with 64 cases where T242N is found in the absence of H219X and 28 cases where neither variant is found (P > 0.0001). In addition, longitudinal sequencing of virus in SMH-05M demonstrated the acquisition of this H219X variant after fixation of the T242N mutation ( Table 3 ). The hypothesis that the H219 variant may be maintained after transmission, whereas the T242N variant reverts, is also supported by analysis of two husband-wife pairs (6007/6008 and 068D/068M; Table 3 ). In both cases, the T242N mutation present in the putative donors (6007 and 068D; both HLA-B5801-positive) was linked with the H219Q variant, whereas in the spouses (6008 and 068M; both HLA-B5801-negative), only the footprints of these variants remain.

We next examined the B57/5801-negative subjects to test the hypothesis that the T242N footprints H219X and A146X are associated with TW10 mutations other than T242N, as T242N would be expected to revert in the absence of HLA-B57/5801. We found that both H219X and A146X were associated with variation in TW10 (P = 0.0009 and P = 0.0002, respectively), in spite of the absence of T242N from these subjects. These data, which show an association between the two identified T242N footprints, H219X and A146X, and TW10 variation within HLA-B57/5801-negative subjects, support the hypothesis that the ancestral virus in these individuals originated from HLA-B57/5801-positive subjects in which the T242N variant was present. Taken together, these data provide compelling evidence that the T242N mutation is transmitted to, and subsequently reverts in, B57/5801-negative subjects.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....