Report

Broadly targeted CD8+ T cell responses restricted by major histocompatibility complex E

See allHide authors and affiliations

Science  12 Feb 2016:
Vol. 351, Issue 6274, pp. 714-720
DOI: 10.1126/science.aac9475

An unconventional route to protection

One promising approach toward an HIV-1 vaccine involves infecting people with cytomegalovirus engineered to express proteins from HIV-1. This approach, which works by eliciting virus-killing CD8+ T cells, provides robust protection in nonhuman primate models. Hansen et al. have found out why this approach is so effective. Normally, peptide antigens presented by major histocompatibility complex-1a (MHC-Ia) activate CD8+ T cells. In vaccinated monkeys, however, CD8+ T cells reacted to peptide antigens presented by MHC-E molecules instead. Moreover, MHC-E could present a much wider range of peptides than MHC-Ia.

Science, this issue p. 714

Abstract

Major histocompatibility complex E (MHC-E) is a highly conserved, ubiquitously expressed, nonclassical MHC class Ib molecule with limited polymorphism that is primarily involved in the regulation of natural killer (NK) cells. We found that vaccinating rhesus macaques with rhesus cytomegalovirus vectors in which genes Rh157.5 and Rh157.4 are deleted results in MHC-E–restricted presentation of highly varied peptide epitopes to CD8αβ+ T cells, at ~4 distinct epitopes per 100 amino acids in all tested antigens. Computational structural analysis revealed that MHC-E provides heterogeneous chemical environments for diverse side-chain interactions within a stable, open binding groove. Because MHC-E is up-regulated to evade NK cell activity in cells infected with HIV, simian immunodeficiency virus, and other persistent viruses, MHC-E–restricted CD8+ T cell responses have the potential to exploit pathogen immune-evasion adaptations, a capability that might endow these unconventional responses with superior efficacy.

Adaptive cellular immunity against intracellular pathogens is the primary responsibility of CD8+ T cells that recognize short pathogen-derived peptide epitopes presented by highly polymorphic major histocompatibility complex class Ia (MHC-Ia) molecules on the surface of infected cells (1, 2). MHC-Ia allomorphs vary considerably in their peptide-binding properties, and therefore the particular pathogen-derived peptides targeted by pathogen-specific CD8+ T cells are largely determined by the peptide-binding specificity of the limited number of MHC-Ia allomorphs expressed by the infected individual (3). Consequently, the epitopes recognized by CD8+ T cells responding to the same pathogen are highly diverse across individuals, resulting in heterogeneity among individuals in the ability to clear or control various infections—in particular, agents such as HIV with a high intrinsic capacity for mutational immune escape (2, 4). Such variation can hamper efforts to develop universally effective vaccines based on CD8+ T cell responses (4, 5).

We recently reported that simian immunodeficiency virus (SIV)–targeted vaccine vectors based on rhesus cytomegalovirus (RhCMV) strain 68-1 [in which genes Rh157.5 and Rh157.4 are deleted (ΔRh157.5/.4)] violate the rules of MHC-Ia–restricted CD8+ T cell recognition (6) and offer a potential solution to MHC-Ia–dependent response diversity in vaccination targeted at CD8+ T cells. Cellular immune responses elicited by RhCMV/SIV vectors are able to stringently control and ultimately clear a highly pathogenic SIV challenge in slightly over half of vaccinated rhesus macaques (RMs) (7, 8). These vectors elicit SIV-specific CD8+ T cell responses that are entirely nonoverlapping with conventional MHC-Ia–restricted CD8+ T cells, despite manifesting three times the breadth of CD8+ T cells elicited by conventional vaccines. Part of this lack of epitope overlap has been explained by the finding that many of the epitopes are restricted by MHC-II, rather than MHC-I, molecules (6). However, a 68-1 RhCMV vector expressing SIV gag protein (SIVgag) also elicited CD8+ T cells that recognized MHC-I–dependent epitopes that were common to most, or even all, MHC-disparate RMs, an unprecedented degree of cross-recognition for MHC-Ia–restricted CD8+ T cell responses. Two SIVgag epitopes [SIVgag276–284(69) and SIVgag482–490(120)] were targeted by 42 of 42 RMs immunized with the 68-1 RhCMV/SIVgag vector in our previous study (6), and we have since documented these “supertope” responses in 130 of 130 RMs inoculated with this vector (fig. S1).

To understand the basis of this unusual MHC-I–dependent recognition, we selected four RMs vaccinated with the 68-1 RhCMV/SIVgag vector for detailed MHC-I restriction analysis. We first sequenced the expressed MHC-I genes, including both classical MHC-Ia and nonclassical MHC-Ib (9), in each RM, and constructed a panel of MHC-I transfectants that individually expressed these MHC-I molecules (fig. S2). These single–MHC-I–molecule transfectants were then used in a flow-cytometric intracellular cytokine staining (ICS) assay to present 12 commonly recognized 15-mer (15–amino acid oligomer) peptides to the CD8+ T cells induced by the 68-1 RhCMV/SIVgag vector in these RMs (Fig. 1, A and B, and fig. S3). Classical MHC-Ia allomorph groups were able to present only 3 of the 12 epitopic peptides to these T cells [Mamu-A1*001:01, epitopes SIVgag69–83(18) and SIVgag197–211(50); Mamu-A1*002:01, epitope SIVgag129–143(33)], and expression of these allomorphs in RMs did not track with these epitope-specific CD8+ T cell responses (fig. S4). However, all 12 peptides stimulated epitope-specific CD8+ T cells when presented by nonclassical major histocompatibility complex E (MHC-E) molecules, and all peptides were presented by transfectants singly expressing three different RM MHC-E allomorphs (Mamu-E*02:04, -E*02:11, and -E*02:20), as well as by a transfectant expressing human MHC-E (HLA-E*01:03; HLA, human leukocyte antigen) (Fig. 1, A and B, and figs. S3 and S5).

Fig. 1 MHC restriction of CD8+ T cells elicited by 68-1 RhCMV/SIVgag.

(A and B) Peripheral blood mononuclear cells (PBMCs) from a representative RM vaccinated with 68-1 RhCMV/SIVgag (RM Rh22034, one of four similarly analyzed; fig. S3) were stimulated with the indicated epitopic 15-mer peptides pulsed onto the surface of parental MHC-I–negative (neg) cell lines (.221 and K562 cells; negative controls), autologous B lymphoblastoid cell lines (BLCL; positive controls), or the indicated MHC-I transfectants, with CD8+ T cell recognition determined by detection of interferon-γ (IFN-γ) and/or tumor necrosis factor–α (TNF-α) production by flow-cytometric ICS assay (response frequencies of gated CD8+ T cells are shown in each quadrant). The MHC-I molecules tested included both those expressed by Rh22034 (A) and additional RM and human MHC-E molecules not expressed by Rh22034 (B). (C) Mamu-E*02:04, Mamu-E*02:20, and HLA-E*01:03 transfectants were pulsed with the serially diluted concentration of the indicated optimal SIVgag 9-mer epitopic peptides. The transfectants were combined with PBMCs from three to four 68-1 RhCMV/SIVgag–vaccinated RMs for flow-cytometric ICS determination of the frequency of responding CD8+ T cells (IFN-γ– and/or TNF-α–positive). Response frequencies at each peptide dose were normalized to the response observed for the transfectant pulsed with the highest concentration (10 μM) of peptide.

MHC-E is known to avidly bind canonical VMAPRTL(LVI)L peptides, which are derived from positions 3 to 11 of MHC-Ia leader sequences, for presentation to NKG2A/C receptor molecules on natural killer (NK) cells (1014). This conserved interaction predominantly inhibits NK cell activity when cells express normal levels of MHC-Ia. However, on interference with MHC-Ia biosynthesis by viral infection or neoplastic transformation, this inhibitory signal is reduced, facilitating NK cell responses to virally infected or neoplastic cells (15, 16). Although a subset of CD8+ T cells can express NKG2A/C (17), phenotypic analysis of MHC-E–dependent CD8+ T cells elicited by the 68-1 RhCMV/SIVgag vector revealed that the vast majority of responding cells were CD8αβ+, TCRγ/δ T cells lacking NKG2A/C expression (fig. S6). Moreover, preincubation of MHC-E transfectants or other antigen-presenting cells with the canonical MHC-E–binding VMAPRTLLL (VL9) peptide before epitopic peptide loading specifically blocked CD8+ T cell recognition of all 12 epitopic peptides (fig. S7). The epitopic SIVgag 15-mers could be truncated to optimal 9-mer peptides that were common among different 68-1 RhCMV/SIVgag vector–vaccinated RMs with MHC-E-associated CD8+ T cell responses to the parent 15-mer (fig. S8) (6). These optimal 9-mer peptides could trigger CD8+ T cells when pulsed on MHC-E transfectants (both Mamu-E and HLA-E) or autologous B cell lines at doses of ≤1 nM (Fig. 1C and fig. S9); such functional avidities are comparable to T cell recognition of classically MHC-Ia–restricted epitopes (18). Taken together, these data strongly suggest that the unconventional, MHC-I-dependent CD8+ T cell responses elicited by the 68-1 RhCMV/SIVgag vector reflect SIVgag 9-mer epitope–specific, MHC-E–restricted CD8αβ+ T cells.

MHC-E–restricted CD8+ T cell responses previously have been identified in human cytomegalovirus (HCMV), hepatitis C virus, Mycobacterium tuberculosis (M. tb.), and Salmonella enterica infections, typically involving epitopes that are structurally related to the canonical MHC-Ia leader-sequence peptides but foreign to the host (13, 14, 19, 20). To determine the extent to which MHC-E restricts RM CD8+ T cell responses in different settings, we applied blocking with high-affinity MHC-E–binding peptide VL9 [in conjunction with blocking with the MHC-II–associated invariant chain peptide (CLIP) and the MHC-I monoclonal antibody (mAb) W6/32 (6)] to restriction-classify all SIVgag epitope–specific CD8+ T cell responses in RMs vaccinated with the 68-1 RhCMV/SIVgag vector (ΔRh157.5/.4), a strain 68-1.2 RhCMV/SIVgag vector (Rh157.5 and Rh157.4 repaired), a 68-1.2 RhCMV/SIVgag vector from which the Rh157.5 and Rh157.4 genes were specifically reexcised (fig. S10), and a modified vaccinia Ankara (MVA)/SIVgag vector, as well as RMs infected with SIV itself (Fig. 2A and figs. S11 and S12). This analysis revealed that essentially all SIVgag epitope–specific CD8+ T cells in RMs vaccinated with the 68-1 RhCMV/SIVgag or ΔRh157.5/.4 68-1.2 RhCMV/SIVgag vectors were either MHC-II– or MHC-E–restricted. In contrast, 98 to 100% of SIVgag-specific CD8+ T cell responses in the RMs vaccinated with the MVA/SIVgag or 68-1.2 RhCMV/SIVgag vectors, or infected with SIV, were classically MHC-Ia–restricted. In keeping with these patterns of epitope restriction, recognition of autologous SIV-infected CD4+ T cells by the SIVgag-specific CD8+ T cells elicited by the 68-1 RhCMV/SIVgag vector was completely blocked by a combination of the MHC-II–blocking CLIP and the MHC-E–blocking VL9 peptides; however, SIV-infected autologous cell recognition by the SIVgag-specific CD8+ T cells elicited by MVA/SIVgag vector vaccination, 68-1.2 RhCMV/SIVgag vector vaccination, or SIV infection was insensitive to the CLIP-plus-VL9 peptide combination (Fig. 2B). In addition, the VL9 peptide alone completely blocked SIV-infected cell recognition by a MHC-E–restricted SIVgag482–490(120)–specific CD8+ T cell line (fig. S14).

Fig. 2 MHC-E restriction is limited to CD8+ T cell responses elicited by ΔRh157.5/.4 RhCMV vectors.

(A) CD8+ T cell responses to SIVgag were epitope-mapped using flow-cytometric ICS to detect recognition of 125 consecutive 15-mer SIVgag peptides [with an overlap of 11 amino acids (a.a.)] in RMs vaccinated with the indicated SIVgag-expressing viral vectors or infected with SIVmac239 (N = 4 to 6 per group shown; see fig. S12 for other studied RMs). Peptides resulting in specific CD8+ T cell responses are indicated by a box, with the color of the box designating MHC restriction as determined by blocking with the pan–MHC-I–blocking mAb W6/32, the MHC-E–blocking peptide VL9, and the MHC-II–blocking peptide CLIP (see the supplementary materials). The minimal number of independent epitopes in these MHC restriction categories is shown at right for each RM. (B) Analysis of SIV-infected CD4+ cell recognition by CD8β+ cells isolated from RMs that were vaccinated with the indicated SIVgag-expressing viral vectors or infected with SIV. The flow profiles at left show IFN-γ and TNF-α production after CD8β+ T cell incubation with autologous SIVmac239-infected CD4+ T cells alone (no block), in the presence of mAb W6/32 plus CLIP, or in the presence of VL9 plus CLIP. All plots are gated on live CD3+ CD8+ cells. The bar graph at right shows the results from all studied RMs.

Taken together, these data confirm that the 68-1 RhCMV/SIVgag vector elicits SIVgag-specific CD8+ T cell responses that are either MHC-II– or MHC-E–restricted, and that this unusual immunobiology is a specific consequence of the deletion of the RhCMV Rh157.5 and Rh157.4 genes, which are orthologs of the HCMV UL128 and UL130 genes and encode two components of the pentameric receptor complex involved in CMV infection of nonfibroblasts (21). Moreover, these data confirm that although SIV infection does not efficiently prime MHC-E–restricted CD8+ T cell responses, at least some of the SIV epitopes recognized by the CD8+ T cells elicited by the 68-1 RhCMV/SIVgag vector are naturally presented by SIV-infected cells, providing for their effective recognition by these heterologous responses.

Among 42 RMs vaccinated with the 68-1 RhCMV/SIVgag vector, we identified a median of 20 distinct CD8+ T cell–recognized, MHC-E–restricted SIVgag 15-mer epitopes per RM, a breadth that exceeds the median 11 and 14.5 distinct MHC-Ia–restricted SIVgag-specific epitopes identified within SIVgag-specific CD8+ T cell responses elicited by conventional vaccines or SIV infection, respectively (Fig. 3A). The ability of ΔRh157.5/.4 RhCMV vectors to elicit MHC-E– and MHC-II–restricted CD8+ T cells is not limited to SIVgag-specific responses (fig. S13), and the density of MHC-E–restricted epitopes (~4 independent MHC-E–restricted epitopes per 100 amino acids of protein length) is similar among all CD8+ T cell responses elicited by 68-1 RhCMV vectors, regardless of the nature of the antigen analyzed (Fig. 3B). Among the same 42 RMs vaccinated with the 68-1 RhCMV/SIVgag vector, 109 of the 125 overlapping SIVgag 15-mer peptides (87%) were recognized by MHC-E–restricted CD8+ T cells in at least one RM (Fig. 3C). Although MHC-E has been shown to bind a broader array of peptides than the canonical leader-sequence peptides (14, 22), the extent of this epitope diversity and breadth is unexpected, especially given the limited polymorphism of MHC-E and the observation that all tested Mamu-E and HLA-E variants present all tested MHC-E–restricted epitopes (Fig. 1, B and C, and figs. S5 and S9). In keeping with this previously unrecognized MHC-E epitope diversity, sequence analysis of 11 optimal MHC-E–restricted SIVgag 9-mer epitopes showed only one epitope, the Gag273–287(69) supertope, with a canonical MHC-E–binding motif (methionine at position 2, leucine at position 9), whereas the other 10 optimal epitopes not only lacked this motif but manifested no statistically significant overlap with previously characterized sets of MHC-E–bound peptides (Fig. 3, D and E, and fig. S15) (22). Statistical analysis of a larger group of 125 SIVgag 15-mer peptides also failed to show any enrichment for previously reported MHC-E–binding motifs (figs. S16 and S17). Despite this amino acid sequence diversity, 10 of the 11 immunologically defined optimal MHC-E–restricted SIVgag 9-mer epitopes were shown to physically associate with and stabilize MHC-E, either in an ELISA (enzyme-linked immunosorbent assay)–based HLA-E–plus–peptide refolding assay or in a single-chain MHC-E–peptide expression assay, or both (figs. S18 and S19).

Fig. 3 Diversity of MHC-E–restricted epitopes.

(A) Comparison of the total number of distinct MHC-E (green)– versus MHC-Ia (red)–restricted SIVgag epitopes recognized by circulating CD8+ T cells in individual RMs vaccinated with 68-1 RhCMV/SIVgag or conventional viral vectors—the latter including MVA/SIVgag (N = 11), adenovirus type 5/SIVgag (N = 3), and electroporated DNA/gag plus interleukin-12 (N = 4)—or in RMs with controlled SIVmac239 infection (N = 12). The horizontal bars indicate median values (P values are from unpaired, two-tailed Mann-Whitney tests). (B) Comparison of the number of distinct MHC-E–restricted epitopes (per 100 amino acids of protein length) recognized by circulating CD8+ T cells in individual RMs vaccinated with 68-1 RhCMV vectors expressing each of the indicated antigens (RhCMV IE1 responses were evaluated in CMV-naïve RMs that were administered 68-1 RhCMV/SIVgag). The horizontal bars indicate median values for each group. (C) Population-level analysis of the breadth of MHC-E–restricted SIVgag epitope–specific CD8+ T cell responses across 125 consecutive 15-mer gag peptides (with an overlap of 11 amino acids) in 42 RMs vaccinated with the 68-1 RhCMV/SIVgag vector. (D) Sequence logo indicating the frequency of each amino acid in a given position by the height of the letter, based on 11 optimal, MHC-E–restricted SIVgag 9-mer peptide epitopes recognized by CD8+ T cells in RMs vaccinated with the 68-1 RhCMV/SIVgag vector. Blue indicates significant amino acid enrichment in a given position relative to its background frequency in SIVmac239 gag (see supplementary materials). Green highlights the M2 and L9 of the canonical MHC-E–binding motif. (E) The same logo as in (D), colored according to enrichment (blue or green) or underrepresentation (red) among 551 peptides eluted from HLA-E*01:03 in a TAP-deficient setting by Lampen et al. (22) (fig. S15). Amino acids enriched in the second and C-terminal anchor positions among the 551 peptides from (22) were rare among our 11 optimal SIVgag peptides, whereas those that were significantly underrepresented in (22) were enriched in our SIVgag epitopic peptides, highlighted in the actual optimal epitopes listed on the right. The percentage of RMs vaccinated with 68-1 RhCMV/SIVgag (N = 42) that responded to each optimal peptide is noted as the recognition frequency.

These data suggest that the loading and binding of epitopic peptides to MHC-E must involve novel molecular mechanisms. To explore this possibility, we performed long–time scale, all-atom molecular dynamics simulations of canonical peptide bound and unbound forms of the solved structure of HLA-E*01:03 (23) in comparison with simulations of the classical HLA-Ia protein HLA-A*02:01, which overlaps with HLA-E in peptide-binding specificity (22). This analysis revealed that HLA-E*01:03, but not HLA-A*02:01, preserves a relatively rigid peptide binding groove that remains open even in the absence of canonical peptide during the simulation time scales (Fig. 4, A and B). Docking analysis further suggests that the 11 optimal MHC-E–restricted SIVgag epitopes described above can bind to both HLA-E*01:03 and Mamu-E*02:04 by adopting a similar backbone configuration as that of the canonical peptide VL9, with diverse side-chain rotamers being accommodated in the same buried positions (Fig. 4C and fig. S20). Further characterization using all-atom molecular dynamics simulations of HLA-E–peptide complexes revealed that the SIVgag epitopic peptides bind at slightly elevated positions in the binding groove relative to VL9, which mainly uses hydrophobic amino acid side-chain interactions to stabilize its deep positioning (Fig. 4D). Consequently, in agreement with direct MHC-E–peptide binding analysis (figs. S18 and S19), computationally determined binding affinities of the 11 optimal MHC-E–restricted SIVgag epitopes are lower than that of VL9 (fig. S21). However, at these elevated positions in the binding groove, epitopic peptides encounter a much more varied binding environment in terms of available polar or charged binding sites (Fig. 4E), and charged residues appear to contribute favorably to the binding of many of these peptides to MHC-E (figs. S22 and S23). Thus, in vivo conditions such as pH, ionic strength, and/or macromolecular crowding can modulate these interactions, possibly explaining some of the differences between in vitro and intracellular MHC-E–peptide binding measurements (figs. S18 and S19). There are limits to the peptide-binding breadth of MHC-E, because not all peptides can be accommodated by MHC-E in the same conformation as the canonical peptide (fig. S24). These structural analyses suggest that distinctive conformational properties of MHC-E may drive the observed epitope diversity and breadth. The open MHC-E binding groove would provide access to binding for an unusually diverse population of peptides, the side chains of which could interact with chemically varying environments at different depths within the binding groove. At the same time, the framework of the relatively rigid MHC-E binding groove would enforce a similar “canonical” backbone conformation on the bound peptides, providing a stable MHC-E–peptide complex for T cell recognition.

Fig. 4 Structural analysis of MHC-E–peptide binding.

(A) Structural changes in the binding groove of canonical peptide-bound and -unbound HLA-A*02:01 and HLA-E*01:03. Calculations of changes in the volume of the binding groove indicate that, unlike in HLA-A*02:01, the HLA-E*01:03 binding groove does not collapse when the peptide is not bound during the 0.5-µs all-atom molecular dynamics simulations. Error bars indicate the 95% confidence interval, determined using standard error estimates from five independent simulations for each case. The total volume (in cubic angstroms) is calculated from the cavity volumes from the N, F, and C regions, as shown in the three-dimensional structure on the right and described in the supplementary materials. (B) Root-mean-square fluctuations of the backbone atoms of unbound HLA-A*02:01 and HLA-E*01:03 are mapped on the x-ray structure (-ter, terminus). Consistent with (A), the binding groove of HLA-E*01:03 is less flexible compared with that of HLA-A*02:01. The binding groove helices partially unfold in unbound HLA-A*02:01, whereas the unbound HLA-E*01:03 binding groove remains relatively stable. Increasing flexibility is captured by the change in color gradient from blue to white to red. (C) HLA-E*01:03 binding profile obtained from a Rosetta-based docking approach (30) of 11 optimal, MHC-E–restricted, SIVgag epitopic peptides. The backbones of these peptides adopt a similar conformation, as shown by the colors in front of the binding groove cross-section. The bound conformations for these 11 peptides are shown in the insets. Residues buried in HLA-E and exposed are marked in red and white, respectively. (D) Molecular dynamics simulations of docked complexes show that the 11 epitopic peptides are bound in a slightly elevated position relative to VL9 in the MHC-E binding groove and are more solvent-exposed (inset bar graph). (E) Cross sections (viewed from above) of the MHC-E binding groove at two different depths show the differences in the chemical environment recognized by the buried residues of epitopic peptides and VL9. Unlike the hydrophobic environment experienced by the buried residues of the VL9 peptide, epitopic peptides experience a chemically heterogeneous environment at their slightly elevated position in the binding groove (figs. S22 and S23).

Both HCMV and RhCMV encode proteins with a strategically embedded canonical VL9 peptide within the UL40 and Rh67 genes, respectively (24, 25). The VL9 peptide of UL40 has been shown to be loaded onto nascent MHC-E chains by a transporter associated with antigen processing (TAP)–independent mechanism, and it therefore functions to stabilize and up-regulate MHC-E expression in HCMV-infected cells in the face of virus-mediated TAP inhibition and the profound MHC-Ia down-regulation mediated by the HCMV US2–11 gene products (15, 24). We have demonstrated a similar function for RhCMV Rh67 (fig. S25). MHC-E up-regulation is therefore thought to be a key viral strategy for evading the NK cell response to virally infected cells that lack MHC-Ia expression (15, 16). However, this evasion strategy would have the consequence of enhancing MHC-E expression in such infected cells, increasing the opportunity for loading and presentation of novel peptides to MHC-E–restricted T cells. In this regard, the canonical MHC-E–binding VL9 peptide might act as a chaperone that facilitates stable high expression of MHC-E and delivery to an endosomal compartment (26) that would facilitate peptide exchange, analogous to the invariant chain–associated CLIP peptide and MHC-II (1). The rigid open structure of the MHC-E peptide-binding groove would be expected to facilitate peptide exchange, consistent with this putative mechanism. Also, consistent with such a peptide exchange mechanism, MHC-E peptide loading has been directly demonstrated in the M. tb. phagolysosome (27).

CMV is not the only intracellular pathogen to up-regulate MHC-E expression. Hepatitis C virus also up-regulates MHC-E expression (28), and both HIV and SIV up-regulate MHC-E in concert with MHC-Ia down-regulation (fig. S26) (29). This common adaptation suggests that for these and probably other intracellular pathogens, the evolutionary pressure to up-regulate MHC-E to counter NK cell responses outweighs the potential risk of increased susceptibility to MHC-E–restricted CD8+ T cells, perhaps because MHC-E–restricted CD8+ T cells are poorly primed during infection with these agents. The reason why MHC-E–restricted CD8+ T cell responses are such a minor component of the modern mammalian immune system is unclear, especially given our finding that such responses can be diverse and broad (although arguably less diverse and broad on a population level than polymorphic MHC-Ia; fig. S27). However, ΔRh157.5/.4 RhCMV vectors are able to bypass the intrinsic constraint of MHC-E–restricted CD8+ T cell priming, perhaps by modulation of vector cell tropism and the priming microenvironment (6). The ability of these vectors to strongly elicit broad, diverse MHC-E–restricted CD8+ T cell responses therefore offers the potential opportunity to develop vaccines that exploit MHC-E up-regulation, an intrinsic vulnerability in the immune-evasion strategies of many highly adapted persistent pathogens. Moreover, because of limited MHC-E polymorphism, a vaccine targeted at MHC-E–restricted CD8+ T cell responses would elicit largely similar responses in all or most vaccinees, potentially providing for efficacy in all individuals regardless of MHC-I genotype. Evolution may have disfavored MHC-E as a primary restricting molecule for CD8+ T cells in modern mammals, but if HCMV vectors are able to replicate in humans the biology of ΔRh157.5/.4 RhCMV vectors in RMs (or, alternatively, if non–CMV-based strategies to elicit broadly targeted MHC-E–restricted CD8+ T cell responses can be developed), vaccinologists may be able to resurrect this largely dormant MHC-E–based adaptive immune system to attack pathogens with novel immune responses that they are not adapted to effectively evade.

Supplementary Materials

www.sciencemag.org/content/351/6274/714/suppl/DC1

Materials and Methods

Figs. S1 to S27

Tables S1 and S2

References (3154)

References and Notes

Acknowledgments: We are grateful to T. van Hall for the HLA-E*01:03 transfectant; D. Geraghty for the 4D12 mAb; D. O’Connor and R. Wiseman for 454-based MHC typing; and Y. Guo, D. Laddy, and M. Stone (Aeras) for provision of the 68-1 RhCMV construct expressing the M. tb. RpFA protein. We thank A. Townsend for help with the graphics; E. McDonald, A. Klug, A. Bhusari, G. Xu, J. Bae, C. Kahl, S. Hagen, A. Sylwester, and L. Boshears for technical or administrative assistance; J. Theiler for coding assistance; and P. Borrow for helpful discussions on HLA-E biology and peptide binding. This research was supported by NIH grants P01-AI094417, R37-AI054292, R01-DE021291, R01-AI095113, R01-AI117802, R01-AI059457, U24-OD010850, P51-OD011092, and P50-GM065794 and contract HHSN272201100013C; the Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery (grant UM1-AI100645-01) of the National Institute of Allergy and Infectious Diseases; the Bill and Melinda Gates Foundation (Global Health grants OPP1108533 and OPP1133649); the Aeras Global TB Vaccine Foundation; and the Center for Nonlinear Studies at the Los Alamos National Laboratory (LANL). LANL institutional computing was used for carrying out all-atom molecular dynamics simulations. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH or other funders. The Oregon Health & Science University (OHSU) has submitted a provisional patent entitled “Methods and compositions useful in generating non-canonical CD8+ T cell responses” (inventors: L.J.P., K.F., J.B.S, D.M, and S.G.H.). OHSU and L.J.P., S.G.H., K.F., and J.A.N. have a substantial financial interest in TomegaVax, a company that may have a commercial interest in the results of this research and technology. L.J.P., K.F., and J.A.N. serve on the board of TomegaVax, and K.F. is also an uncompensated officer of the company. The potential individual and institutional conflicts of interest have been reviewed and managed by OHSU. Additional data used in this report are available as supplementary materials on Science Online.
View Abstract

Navigate This Article