Comprehensive Immunological Evaluation Reveals Surprisingly Few Differences between Elite Controller and Progressor Mamu-B*17-Positive Simian Immunodeficiency Virus-Infected Rhesus Macaques

Abstract

The association between particular major histocompatibility complex class I (MHC-I) alleles and control of human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) replication implies that certain CD8⁺ T-lymphocyte (CD8-TL) responses are better able than others to control viral replication in vivo. However, possession of favorable alleles does not guarantee improved prognosis or viral control. In rhesus macaques, the MHC-I allele Mamu-B*17 is correlated with reduced viremia and is overrepresented in macaques that control SIVmac239, termed elite controllers (ECs). However, there is so far no mechanistic explanation for this phenomenon. Here we show that the chronic-phase Mamu-B*17-restricted repertoire is focused primarily against just five epitopes—VifHW8, EnvFW9, NefIW9, NefMW9, and env_ARFcRW9—in both ECs and progressors. Interestingly, Mamu-B*17-restricted CD8-TL do not target epitopes in Gag. CD8-TL escape variation occurred in all targeted Mamu-B*17-restricted epitopes. However, recognition of escape variant peptides was commonly observed in both ECs and progressors. Wild-type sequences in the VifHW8 epitope tended to be conserved in ECs, but there was no evidence that this enhances viral control. In fact, no consistent differences were detected between ECs and progressors in any measured parameter. Our data suggest that the narrowly focused Mamu-B*17-restricted repertoire suppresses virus replication and drives viral evolution. It is, however, insufficient in the majority of individuals that express the “protective” Mamu-B*17 molecule. Most importantly, our data indicate that the important differences between Mamu-B*17-positive ECs and progressors are not readily discernible using standard assays to measure immune responses.

The search for a successful vaccine for AIDS has been hampered by an inability to understand how an effective immune response controls human immunodeficiency virus/simian immunodeficiency virus (HIV/SIV) replication. Understanding the complete CD8⁺ T-lymphocyte (CD8-TL) response against an AIDS virus in the setting of durably controlled viral replication would be an important step forward. Rhesus macaques that spontaneously control SIV replication to fewer than 1,000 viral RNA (vRNA) copy equivalents (cEq) per milliliter of blood plasma (elite controllers [ECs]) may offer a glimpse into such a response.

Expression of particular MHC class I (MHC-I) alleles is correlated with reduced virus loads and delayed disease progression in HIV-infected humans (5, 9, 21, 26, 46). Likewise, we previously demonstrated that the MHC-I allele Mamu-B*17 was associated with decreased plasma virus concentrations in two independent cohorts of Indian rhesus macaques infected with clonal SIVmac239 (34, 62). In fact, 9 of 16 (56%) ECs were Mamu-B*17 positive (36). However, the expression of Mamu-B*17, like that of protective alleles in humans, does not guarantee in vivo viral suppression and EC status (61, 62). Most Mamu-B*17-positive individuals (33/42 [79%]) experience continued high-level viral replication and progression to AIDS. Understanding why certain Mamu-B*17-positive macaques suppress viral replication whereas others do not may provide insights into how to control AIDS virus replication (14).

Despite clear associations between MHC-I alleles and disease progression, the mechanisms underlying these associations are largely unknown. One possible exception is the case of HLA-B27-positive, HIV-infected humans. Loss of viral control in HLA-B27-positive ECs is often associated with mutational escape in the HLA-B27-restricted Gag KK10 epitope (17, 23, 29, 33). This implicates CD8-TL responses against this single immunodominant epitope in control of HIV replication. Viral escape from this response is difficult due to dramatic reductions in viral fitness associated with this escape, which must be overcome by the appearance of compensatory mutations outside the epitope (54). Such a mechanistic explanation of HIV control in HLA-B57-positive and other ECs has not been demonstrated (14, 45), despite extensive effort. Similarly, the mechanisms of control of SIV replication associated with Mamu-B*17 expression have not been defined. In this study, we conducted a thorough investigation of the Mamu-B*17-restricted response repertoire against SIVmac239 in an attempt to understand the nature of the enhanced viral control exhibited by animals that express this molecule.

Escape from CD8-TL responses is a major cause of HIV (2) and SIV (51) evolution. Certain CD8-TL responses select for viral variants that elude immune recognition and lead to viral breakthrough, as exemplified by the HLA-B27-restricted Gag KK10 epitope in humans (17, 23). It is not known whether this phenomenon is generally the rule or an exception to it. In addition, CD8-TL might select for viral variants that have diminished replicative capacity. Although viral fitness deficits stemming from escape from CD8-TL responses are well documented (1, 18, 31, 32, 38, 44, 54), it is unclear whether this leads to in vivo control of viral replication and/or improved disease prognosis. The role of viral escape in disease progression is also unclear because variant peptides (including escape variants) are often reactive in cytokine secretion assays such as enzyme-linked immunospot (ELISPOT) (6, 13, 40, 57), which may not accurately predict physiologically relevant recognition or viral suppression in cells infected with the variant viruses (34, 57). Further complicating the role of escape in viral control, certain viral escape mutations can stimulate de novo CD8-TL responses specific for the variant epitope (3, 16), provided the variant maintains the ability to bind the MHC-I molecule.

Here we show that Mamu-B*17-positive ECs and progressors both have the same narrowly focused CD8-TL response repertoire in the chronic phase of infection and that the targeted epitopes escape in similar fashion. Furthermore, we show that ECs and progressors are similarly able to recognize escape variants of the epitopes. Together, our data indicate that the mechanisms underlying elite SIV control in Mamu-B*17-positive macaques, like that in HLA-B57-positive humans, are not evident by conventional measures of CD8-TL efficacy and, further, that understanding the important phenomenon of elite control will require novel approaches.

MATERIALS AND METHODS

Animals and MHC typing.

SIVmac239-infected Indian rhesus macaques (Macaca mulatta) were housed at the Wisconsin National Primate Research Center. Animals were cared for according to the regulations outlined in the Guide for the Care and Use of Laboratory Animals of the National Research Council. The University of Wisconsin Institutional Animal Care and Use Committee approved all procedures involving animals. Animals were genotyped for the MHC class I alleles Mamu-A*01, -A*02, -A*08, -A*11, -B*01, -B*03, -B*04, -B*17, and -B*08 as previously described (28, 36).

Virus loads.

Plasma virus concentrations were determined as described previously (12, 19) and were compiled for a previous study (36).

ELISPOT assays.

Measurements of cellular immune responses were made using gamma interferon (IFN-γ) ELISPOT. ELISPOT assays were performed as described previously (60) using 100,000 peripheral blood mononuclear cells (PBMCs) and 10 μM synthetic peptide (or 10-fold serial dilutions of peptide) per well in precoated ELISpot^plus kits (Mabtech, The Netherlands). All samples were run in duplicate, and negative (no peptide) and positive (concanavalin A) controls were included on each plate.

Viral sequencing.

Viral RNA sequencing was performed as described previously (51) on a 3730 automated DNA sequencer (Applied Biosystems, Foster City, CA) using primer pairs that produced amplicons spanning the regions of the SIVmac239 genome that encode the four Mamu-B*17-restricted epitopes carried by functional proteins.

Peptide binding.

Peptide binding studies were done as described previously (47) using quantitative assays to measure the binding affinity of peptides to purified Mamu-B*17 molecules based on the inhibition of binding of a radiolabeled standard peptide.

Clone production and viral suppression assays.

CD8-TL clones were produced as described previously (11) using three rounds of limiting dilution of PBMCs that were stimulated weekly using irradiated autologous B-lymphoblastoid cells pulsed with the peptide of interest. Clones were maintained in RPMI medium containing 15% fetal calf serum, 1% l-glutamine, 1% antibiotic/antimycotic (all from HyClone, Logan, UT), and 100 U/ml interleukin 2 (NIH AIDS Reference and Reagent Program, Germantown, MD). Virus suppression assays were performed essentially as described previously (37).

Statistical analysis of selection.

Numbers of synonymous substitutions per synonymous site (d_S) and numbers of nonsynonymous substitutions per nonsynonymous site (d_N) were estimated using the method of Nei and Gojobori (49), and the results were analyzed as described previously (27, 35).

RESULTS

Mamu-B*17-positive rhesus macaques exhibit a range of control over SIVmac239 replication.

The MHC-I allele Mamu-B*17 is associated with decreased plasma viremia and is overrepresented in animals that control virus below 1,000 cEq/ml, termed ECs (62). Mamu-B*17-positive macaques exhibit a wide range of plasma virus concentrations during the chronic phase of infection. Despite this, the average chronic-phase virus loads of Mamu-B*17-positive macaques are markedly reduced compared to Mamu-B*17-negative animals (Fig. 1). It was shown previously that neither the inheritance of the Mamu-B*17-containing haplotype (61) nor polymorphisms in particular host genes (59) could predict control of SIV, but that control was likely associated with possession of the Mamu-B*17 allele itself. Therefore, in an attempt to understand Mamu-B*17-associated control of SIV and the discordance between Mamu-B*17-positive ECs and progressors, we conducted a thorough investigation of the repertoire of Mamu-B*17-restricted responses against SIV.

FIG. 1.

Mamu-B*17-positive animals exhibit a range of chronic virus loads. The mean chronic-phase (>10 weeks postinfection) virus loads of 185 SIV-infected Indian rhesus macaques were divided into Mamu-B*17-positive and -negative groups. Animals that express Mamu-B*08 were removed from analysis because they are highly disposed to control SIVmac239 replication (36). Animal r98016, which expresses both Mamu-B*17 and -B*08, is included here. The horizontal lines through each grouping represent the mean virus loads of that group. The horizontal line at 1 × 10³ vRNA cEq/ml plasma is the cutoff for EC status, as previously reported (36).

Mamu-B*17-restricted CD8-TL responses focus on four epitopes in the viral proteins Nef, Vif, and Env and one in an alternate reading frame (ARF) of Env, but do not target Gag.

We previously conducted a survey of SIVmac239-derived peptides that bound the Mamu-B*17 molecule (47). This provided us with a set of 36 optimal Mamu-B*17-bound peptides with which to screen the epitope repertoire of our SIV-infected, Mamu-B*17-positive rhesus macaques. We tested archived chronic-phase samples from Mamu-B*17-positive progressors and ECs for responses to the 36 known Mamu-B*17-restricted epitopes using IFN-γ ELISPOT. Strikingly, we found that the Mamu-B*17-restricted response was focused almost entirely on just five epitopes in the chronic phase of infection, Vif HW8 (HW8), Nef IW9 (IW9), Nef MW9 (MW9), Env FW9 (FW9), and a cryptic epitope derived from the +2 reading frame of the envelope gene (41) (Fig. 2). Interestingly, the only Gag-encoded peptide known to bind to Mamu-B*17 (Gag_406-415CW10; CRAPRRQCW), binds poorly (50% inhibitory concentration [IC₅₀] of 406 nM) (47) and was not targeted by any animal in this study. Although there is no observable difference in the chronic-phase immunodominance hierarchies between ECs and progressors, the lack of sufficient acute-phase EC samples precludes a thorough comparison at these critical early time points. Here we report results obtained only from the PBMC. It will be important to compare the tissue-specific distribution of antigen-specific CD8-TL between ECs and progressors, as tissue-specific differences in antigen-specific responses have been established (53).

FIG. 2.

The Mamu-B*17-restricted response is focused primarily on four epitopes. Mamu-B*17-positive macaques were tested in the chronic phase of SIV infection for responses to 35 previously defined Mamu-B*17-restricted epitopes (47) using IFN-γ ELISPOT. WPI, weeks postinfection. The y axis represents the number of animals making a response (as defined below) to the given epitope. Only epitopes that showed at least one positive response in chronic infection are shown. All ELISPOTS were performed in duplicate. ELISPOT responses were measured as spot-forming cells (SFC) per million PBMCs. The mean number of spots in unstimulated (no peptide) wells was subtracted from each well. ELISPOT responses were considered positive if the number of spots (per million PBMCs) in replicate wells exceeded background plus two times the standard deviation and was >50. An asterisk indicates that data for the cRW9 epitope were previously published (41) and represent data from 15 progressors and 5 ECs.

Viral suppression and evolution in Mamu-B*17-restricted epitopes.

We next determined whether Mamu-B*17-specific CD8-TL, cultured from EC Mamu-B*17-positive macaques, suppressed SIV replication better than those derived from their progressor counterparts. Since the chronic-phase Mamu-B*17-restricted repertoires targeted by ECs and progressors are essentially identical, we sought to characterize the functional abilities of CD8-TL directed at Mamu-B*17-restricted epitopes. To examine this, we first isolated CD8-TL clones from ECs and progressors, directed against the four Mamu-B*17-restricted epitopes encoded by functional SIV proteins. We examined the abilities of these clones to suppress virus replication in a functional assay. We showed previously that CD8-TL specific for the cRW9 epitope were potent at limiting virus replication in vitro (41). In accordance with previous findings (11), we found clonal variation in suppressive abilities, but, strikingly, we found that the majority of EC-derived clones specific for all four epitopes were effective at restricting virus replication, defined as >80% suppression as measured by intracellular p27 Gag staining. This cutoff for “effective” suppression is in accordance with our previous studies (11). We have found that an 80% reduction in p27 Gag staining is roughly equivalent to a 10-fold decrease in virus present in culture supernatants. In contrast, a smaller fraction of clones derived from progressors were effective at virus suppression (Table 1). It is unclear, however, whether this is due to preservation of an intact CD4 T-helper compartment in ECs, prevention of CD8-TL exhaustion stemming from reduced antigenic stimulation, or some other phenomenon.

TABLE 1.

EC-derived CD8-TL clones specific for the four primary Mamu-B*17-restricted epitopes effectively suppress virus replication

Epitope	No. of clones/no. testeda:
	EC effective	Progressor effective
Nef IW9	7/7	6/7
Nef MW9	12/17	2/5
Vif HW8	7/7	4/13
Env FW9	11/13	2/16
Mamu-B*17 total	37/44	14/41

^aCD8-TL clones were used as effector cells in an in vitro assay at an effector/target ratio of 1:10. “Effective” suppression is defined as >80% suppression by intracellular p27 Gag staining after 7 days in culture.

Since CD8-TL directed at Mamu-B*17-restricted epitopes suppressed virus replication in vitro, we wished to determine whether CD8-TL directed against Mamu-B*17-restricted epitopes could exert sufficient selective pressure to drive viral evolution and, furthermore, if patterns of epitope evolution were different between ECs and progressors. To test this, we sequenced the regions of the vif and nef open reading frames that encode the HW8, IW9, and MW9 epitopes. The Env FW9 epitope was excluded from this analysis because the envelope protein is subject to potent selective forces aside from CD8-TL pressure and is generally far less conserved than other SIV proteins. Since viral loads were too low to sequence in ECs, SIV sequences from four ECs were obtained by sequencing the recrudescent virus that replicated after experimental, in vivo CD8 cell depletion (19). We showed previously that escape occurs in the IW9 (50) and cRW9 (41) epitopes. Amino acid changes were observed in all epitopes but were less common in HW8 (Fig. 3). Analysis of viral escape revealed no clear difference between ECs and progressors. Three of four ECs harbored SIV with wild-type sequence in the HW8 epitope. However, the wild-type sequence (and responses against it [Fig. 2]) tends also to be preserved in progressors. Additionally, of the five primary Mamu-B*17-restricted epitopes, the HW8 epitope sequence is the only one that is entirely conserved between the viral isolate SIVmac239 and the distantly related SIVsmmE660, as published in the Los Alamos HIV databases (http://www.hiv.lanl.gov). These data indicate that patterns of viral evolution in Mamu-B*17-restricted epitopes do not distinguish ECs from progressors.

FIG. 3.

Amino acid variation was observed in three Mamu-B*17-restricted epitopes. Most of the coding regions for Vif and Nef were sequenced at the time of euthanasia or late chronic SIV infection in 31 Mamu-B*17-positive and 31 Mamu-B*17-negative animals. Of 31 Mamu-B*17-positive animals, four were ECs. An asterisk indicates that viral sequences from ECs represent the recrudescent virus after experimental CD8 cell depletion in four Mamu-B*17-positive ECs (19). We could not obtain sequences from other ECs due to their extremely low viral loads. Double asterisks indicate that macaque r95003 (Mamu-B*17 negative) made an HW8-specific response, as well as a cRW9-specific response, and harbored SIV with escape mutations in both epitopes (41). Triple asterisks indicate that this variant confers escape from an overlapping Mamu-A*02-restricted epitope, Nef_159-167YY9 (58).

We next determined if the observed patterns of epitope variation were due to positive selection acting on the epitopes. To do this, we compared the predicted amino acid sequences of the epitopes and surrounding regions in 31 chronically infected Mamu-B*17-positive macaques to previously obtained sequences from 31 Mamu-B*17-negative, MHC-defined, SIVmac239-infected macaques (51). Variation was significantly associated with positive selection in Mamu-B*17-positive animals in both Nef epitopes but not in HW8 (Table 2), likely due to the small number of animals harboring virus with mutations in this epitope. However, our analysis did show that rates of nonsynonymous substitutions were significantly elevated in all three epitopes in Mamu-B*17-positive versus Mamu-B*17-negative animals, indicating that viral evolution in these epitopes is associated with expression of Mamu-B*17 and, therefore, likely represents viral escape from CD8-TL responses directed at Mamu-B*17-restricted epitopes.

TABLE 2.

Comparisons of the mean d_S and d_N between the inoculum and time of death in virus sequences in rhesus macaques

Mamu-B*17 expression	Gene or region	Mean dS ± SE	Mean dN ± SE or significance	dS = dN?a
Positive (31 macaques)	vif
	Vif HW8	0.0113 ± 0.0083	0.0308 ± 0.0049	NSc
	Remainder	0.0023 ± 0.0008	0.0061 ± 0.0007	P < 0.01
	nef
	Nef IW9	0.0263 ± 0.0070	0.0789 ± 0.0083	P < 0.001
	Nef MW9	0.0104 ± 0.0054	0.0426 ± 0.0042	P < 0.001
	Remainder	0.0041 ± 0.0015	0.0114 ± 0.0016	P < 0.01
Negative (31 macaques)	vif
	Vif HW8	0.0000 ± 0.0000	0.0182 ± 0.0018	NS
	Remainder	0.0010 ± 0.0006	0.0033 ± 0.0007	NS
	nef
	Nef IW9	0.0000 ± 0.0000	0.0099 ± 0.0039	NS
	Nef MW9	0.0000 ± 0.0000	0.0023 ± 0.0017	NS
	Remainder	0.0011 ± 0.0006	0.0080 ± 0.0019	P < 0.01
B*17+dN = B*17−dNb	vif
	Vif HW8		P < 0.001
	Remainder		P < 0.05
	nef
	Nef IW9		P < 0.001
	Nef MW9		P < 0.001
	Remainder		NS

^aTests of the hypothesis (two-sample t test, Bonferroni corrected) that d_S was equal to d_N in Mamu-B*17-restricted epitopes and in the remaining gene sequence.

^bTests of the hypothesis (two-sample t test, Bonferroni corrected) that d_N for Mamu-B*17⁺ animals equals the corresponding value for Mamu-B*17⁻ animals.

^cNS, not significant.

Variation in Mamu-B*17-restricted epitopes resulting from positive selection indicates that SIV evolved to escape CD8-TL responses directed against these epitopes. Because of this, we carried out further experiments to understand the nature of this escape. Strikingly, mutations changing the C-terminus tryptophan anchor residues were never observed, which contrasts with escape from the cryptic, ARF-derived epitope cRW9 (41). Escape in cRW9 always involved changing the tryptophan anchor, which may have been tolerated in an ARF-derived peptide without known function. Since tryptophan is a rare amino acid, it is perhaps more constrained in functional proteins. In fact, amino acid substitutions only rarely involved changing the position 2 anchor, possibly indicating that escape in Mamu-B*17-restricted epitopes often involves escape from the T-cell receptor (TCR). To test if common escape variants of these epitopes were capable of binding the Mamu-B*17 molecule, we carried out competition assays using purified Mamu-B*17 molecules (Table 3). As expected, the observed substitutions in the Nef epitopes IW9 and MW9, which are largely clustered at (non-anchor residues) position 1 (for both epitopes) and position 6 (for IW9), reduce binding to some degree, but the peptides still bind the Mamu-B*17 molecule, with an IC₅₀ of <100 nM. In contrast, both variants of HW8, where substitutions generally occur in the N-terminal anchor residue, show drastic reductions in binding but still bind under the previously defined threshold of physiological relevance (500 nM) (55).

TABLE 3.

Many common escape variants of Mamu-B*17 epitopes retain the ability to bind the Mamu-B*17 molecule

Epitope	Sequencea	IC50 (nM)b	Binding reduction (fold)
Vif HW8	HLEVQGYW	2.9
H1Y	YLEVQGYW	434	150
H1Q	QLEVQGYW	492	170
Nef IW9	IRYPKTFGW	3.5
I1T	TRYPKTFGW	82	23
T6M	IRYPKMFGW	94	27
Nef MW9	MHPAQTSQW	3.7
M1I	IHPAQTSQW	65	18
M1V	VHPAQTSQW	294	79
Env FW9	FHEAVQAVW	7
H2Q	FQEAVQAVW	8.6	1.2
V8I	FHEAVQAIW	13	1.9
V8F	FHEAVQAFW	14	2
V5A	FHEAAQAVW	15	2.1
H2Y	FYEAVQAVW	1,313	188

^aMutations (underlined) that result in substantial binding reductions (>100-fold) are shown in boldface.

^bRelative binding of the four Mamu-B*17-restricted peptides and common escape variants of the epitopes. IC₅₀ is the concentration of peptide needed to reduce binding of a radiolabeled reference peptide by 50%. The cutoff for physiologically relevant binding is 500 nM (55).

Since many of the variant epitopes maintained the ability to bind the Mamu-B*17 molecule, we hypothesized that animals might mount responses against the variants. To test this, we examined whether animals could recognize these variants in the chronic phase of SIV infection. The ability to recognize escape variant epitopes could be due to cross-reactive TCRs or to de novo responses specific for the escape variant peptides. We were particularly interested in whether the ability to recognize epitope escape variants was associated with elite control. We used IFN-γ ELISPOT to examine fresh PBMCs from five animals, two ECs (r98016 and r95071), two progressors (r90092 and r02039), and one former EC that had since experienced breakthrough viremia and progression to AIDS (r96112). Since autologous viral sequences were not known prior to testing, we examined responses to common variants of these epitopes. In several instances, the strongest response was directed at a variant of the epitope, particularly against specific variants of the IW9, MW9, and FW9 epitopes (Fig. 4). Responses to the M1I variant (IHPAQTSQW [the variant amino acid is underlined]) of MW9 were dominant in progressor r90092 (Fig. 4b, second panel) and EC r95071 (Fig. 4d, second panel) and codominant with the response to the wild-type peptide in animal EC r98016 (Fig. 4e, second panel). The response to the I1T (TRYPKTFGW) variant of IW9 was dominant in progressor r02039 (Fig. 4a, third panel) and codominant (at the highest peptide concentration) with the wild type in progressor (former EC) r96112 (Fig. 4c, third panel) and nearly codominant with the wild type in EC r98016 (Fig. 4e, third panel). In addition, the responses of progressor r90092 and EC r95071 were most strongly directed at the T6M variant (IRTPKMFGW) (Fig. 4b and d, respectively, third panel).

FIG. 4.

Mamu-B*17-positive animals recognize common escape variants of Mamu-B*17-restricted epitopes. IFN-γ ELISPOT responses to titrated peptides representing wild-type and common escape variants of the four primary Mamu-B*17-restricted epitopes, Vif HW8 (left panels), Nef MW9 (second from left), Nef IW9 (second from right), and Env FW9 (far-right panels). The autologous (aut) viral sequences (the epitope sequences at the time of ELISPOT testing) are in insets in each panel, aligned to wild-type (wt) SIVmac239. Five chronically infected animals were tested: progressors r02039 (a) and r90092 (b), former EC r96112 (c), and ECs r95071 (d) and r98016 (e). (f) Animal r95071 harbored SIV with an M-to-I variant of the MW9 epitope at position 1 after experimental in vivo CD8 cell depletion (experiment done in reference 19), but currently harbors an M-to-T variant. The M1I-specific response greatly expanded upon return of CD8⁺ cells, while wild-type-specific cells did not (g). All ELISPOTS (in panels a to e and g) were performed in duplicate. Responses were measured as spot-forming cells (SFC) per million PBMCs. The mean number of spots in unstimulated (no peptide) wells was subtracted from each well. ELISPOT responses were considered positive if the number of spots (per million PBMCs) in replicate wells exceeded the background plus two times the standard deviation and was >50. Error bars represent the mean ± standard error for each measurement. The horizontal line in panels a thru e is the cutoff for what is considered positive.

To determine if responses directed at escape variants might be due to de novo responses, we sequenced the regions of the SIV genome encoding the four Mamu-B*17-restricted epitopes in the same five animals (depicted at the top of each panel in Fig. 4, aligned with wild-type SIVmac239). We were able to obtain viral sequences from the two ECs because these two animals harbored detectable replicating virus, yet at a level below the threshold of EC status. We detected instances where the dominant response was directed at the autologous epitope variant, possibly indicating de novo responses directed at these variants. Specifically, both progressor animals (r02039 and r90092) directed their IW9-specific responses at the autologous variant: I1T in r02039 and T6M in r90092. Additionally, the autologous SIV sequence in the IW9 epitope in EC r98016 was I1T, while the responses were nearly equally distributed between I1T and the wild type. The MW9 epitope in EC r98016 was a mixture of the wild type and M1I variant, reflecting exactly the codominant response.

Strikingly, we also found several instances in which the dominant response was clearly directed against a nonautologous viral variant. In these cases, we hypothesized that this might be due to de novo, variant-specific responses driving viral evolution away from the corresponding variant sequence. To test this, we sequenced virus from earlier time points to determine if such a variant had arisen at some point in the past, eliciting the variant-specific response. Progressor r90092 (Fig. 4b, second panel) and EC r95071 (Fig. 4d, second panel), both of which made dominant responses against the M1I variant of the MW9 epitope, did harbor SIV with this mutation at some point in their SIV infection history. SIV from progressor r90092 harbored this mutation at 15 weeks postinfection. EC r95071 harbored SIV with this mutation after its CD8⁺ cells were experimentally depleted in a previous study (19) (Fig. 4f). To determine whether the response to this epitope expanded upon return of CD8⁺ cells in animal r95071, we tested PBMC samples frozen from 5 weeks before and 5 weeks after depletion for responses to the M1I variant peptide. Indeed, the M1I-specific response, which was undetectable before depletion, strongly expanded upon the return of CD8⁺ cells after experimental depletion (Fig. 4g), which might explain evolution away from this variant to the current sequence. In addition, progressor r02039 harbored SIV with a V5A variant (FHEAAQAVW) of the FW9 epitope at 5 weeks postinfection (Fig. 4a, fourth panel). This was the only animal to make a response to this variant in this study. Finally, EC r95071 made a stronger response to the T6M variant of the IW9 epitope (Fig. 4d, third panel). However, we were unable to detect such a variant in the autologous viral sequence in the time points examined (data not shown). It is a reasonable hypothesis that such a variant did arise, eliciting the response, but was simply not detected. Together, our ELISPOT data indicate that recognition of Mamu-B*17-associated escape variants is a consistent phenomenon and that some variant-specific responses are capable of driving viral evolution away from the specific variant they recognize, suggesting a dynamic coevolution of host response and virus in both ECs and progressors.

DISCUSSION

A comprehensive understanding of an effective CD8-TL response against HIV/SIV will enable better design and testing of candidate anti-HIV vaccines. Recent data indicating that multiple Gag-specific responses are important in control of HIV replication (26, 30) are encouraging for vaccine design. However, data from the SIV/macaque model demonstrate that vaccination with Gag alone affords only short-term attenuation of viral replication, and only in individuals that express certain MHC-I molecules (10). The beneficial effect was greatly enhanced by the addition of Tat, Rev, and Nef in the vaccine regimen (60). Likewise, in this study we found that the rhesus macaque MHC-I allele, Mamu-B*17, which is highly correlated with control of SIVmac239 (36, 62), does not restrict epitopes in the Gag protein. Instead the Mamu-B*17-restricted repertoire focuses on just four epitopes in the viral proteins, Vif, Nef, and Env, and one, cRW9, in an ARF of env (41). In fact, only one Gag gene-encoded peptide can bind the Mamu-B*17 molecule (47), and we have never seen responses directed at this epitope. In contrast, the human HLA-B27 and HLA-B57 molecules restrict several epitopes in the HIV Gag protein (45). Our results do not contradict the HIV findings: rather, they imply that responses directed at other AIDS virus proteins, particularly Nef and Vif, might also be important in control of viral replication.

The narrow focus of the Mamu-B*17-restricted response suggests that SIV might frequently evolve to evade these responses and that this evolution might be important in the control of viral replication in Mamu-B*17-positive animals in general or in Mamu-B*17-positive ECs in particular. The role of viral escape in abolishing control of AIDS viruses is controversial. In HLA-B27-positive EC humans, escape in the immunodominant Gag KK10 epitope can lead to breakthrough viremia and progression to AIDS (17, 23). These results suggest that a strong response to the Gag KK10 epitope may be sufficient to maintain control of HIV replication. However, in ECs that express other protective MHC molecules, such as HLA-B57 in humans and Mamu-B*17 in macaques, such a straightforward explanation for control has not been found, despite extensive efforts to understand the HLA-B57-restricted repertoire (15, 24, 32, 39, 45, 48).

To determine the role of viral evolution in SIV control in Mamu-B*17-positive macaques, we tested whether ECs and progressors harbored SIV with differential patterns of viral evolution in three of the Mamu-B*17-restricted epitopes. Sequence variation associated with expression of Mamu-B*17 was detected in Vif HW8, Nef IW9, and Nef MW9. Patterns of sequence variation, however, were largely the same between ECs and progressors. However, three of four ECs harbored SIV with a wild-type HW8 epitope. The HW8 epitope is the only one of the five targeted Mamu-B*17-restricted epitopes that is completely conserved in the distantly related SIVsmmE660. In addition, the anchor residues (position 1 histidine and position 8 tryptophan) are well conserved among HIV-2 isolates (www.hiv.lanl.gov). It is tempting to speculate that it is difficult to escape from the HW8-directed response and that preservation of the HW8 epitope is associated with control of SIV; however, further experiments will be necessary to test this hypothesis.

Surprisingly, much of the sequence variation did not substantially reduce the epitopes’ ability to bind the Mamu-B*17 molecule. Because of this, we examined whether ECs and progressors recognized escape variants of the Mamu-B*17-restricted epitopes. De novo generation of escape variant-specific CD8-TL responses has been documented in HIV (3, 16). In addition, Bailey et al. (6) showed that HLA-B57-positive EC humans in the chronic phase of HIV infection often show stronger reactivity to the autologous variant of several HLA-B57-restricted epitopes than to the wild-type sequence, implying either cross-reactive or variant-specific de novo responses. They suggest that this might be important in maintaining virus control in ECs. However, they did not test progressors to determine if this phenomenon was exclusive to ECs. Likewise, in this study, we also found dominant responses to escape variant peptides. Unexpectedly, variant-specific responses were not always directed at the autologous variant, but at a variant previously present in the host. Further experiments using more animals will be required to determine exactly how often this occurs, but this finding strongly suggests that variant-specific responses are capable of driving viral evolution away from that particular variant. This phenomenon has been reported in HIV (25), but recent data indicate that variant-specific responses often have poor antiviral efficacy (20, 56). Importantly, however, we found recognition of variants in both ECs and progressors, suggesting that this is likely not the primary mechanism maintaining EC status but might be crucial in reducing viremia in Mamu-B*17-positive animals overall.

We also found that CD8-TL clones derived from ECs were more effective at suppressing virus replication than those derived from progressors. This observation is similar to that by Betts et al. (8), who noted that HIV-infected nonprogressing humans maintain CD8-TL with a more multifunctional phenotype than do progressors. However, it is difficult to discern whether this is the cause of EC status or a consequence of reduced antigenic stimulation resulting from reduced viral loads in ECs. Hence, since Mamu-B*17-associated elite control of SIV replication was not due to chronic-phase immunodominance patterns, viral escape, or recognition of variant epitopes, the mechanisms underlying this control remain unknown. An intriguing possibility is that Mamu-B*17-positive ECs express certain killer cell immunoglobulin-like receptor alleles that, in concert with Mamu-B*17, afford extraordinary control of SIV replication. In fact, epistatic relationships between specific killer cell immunoglobulin-like receptor and MHC-I alleles that enhance control of HIV replication are well established (4, 7, 42, 43, 52). It is also possible that ECs have CD4⁺ T-helper cells better able to maintain effective CD8-TL responses. We showed previously that Mamu-B*17-positive ECs are significantly enriched for two MHC-II alleles, Mamu-DRB1*1003 and -DRB1*0306 (22), which provides another possibility about the important difference between ECs and progressors that both express Mamu-B*17.

In the SIV/macaque model, EC status is defined as control of chronic phase (≥10 weeks postinfection) SIV viremia to below 1,000 cEq/ml (22, 36). This threshold represents a >2-log reduction in chronic-phase viremia compared to the chronic-phase geometric mean of the entire cohort of 196 SIV-infected macaques (223,800 cEq/ml) (36). Although the threshold for EC status in HIV-infected humans is more stringent (usually defined as ≤50 cEq/ml), the mean chronic-phase viremia in HIV-infected humans is also much lower (∼30,000 cEq/ml), necessitating a less-stringent threshold for SIV-infected macaque EC status. Although the level of viral control clearly distinguishes Mamu-B*17-positive ECs from Mamu-B*17-positive progressors, it is important to note that SIV-infected, Mamu-B*17-positive macaques exhibit a wide range of viral control from less than 100 to greater than 10,000,000 cEq/ml in the chronic phase. It is possible that certain immunological or virological features, such as specific CD8-TL escape mutations that lead to elite control of SIV, are only manifested in animals with extremely low (<100 cEq/ml) or undetectable virus loads, and so were not discerned in this study. However, this study represents a comprehensive test of several common hypotheses used to explain elite control and, surprisingly, no correlates of control were identified.

In this study, we found that Mamu-B*17-positive ECs and progressors both recognize the same epitopes, which escape in similar fashion. In addition, ECs and progressors alike recognize escape variants of the targeted epitopes. Most importantly, our data indicate that conventional assays that measure immune responses by cytokine secretion and the subsequent viral evolution are likely inadequate to answer some of the most important questions in AIDS vaccine research.