The Role of MHC-E in T Cell Immunity Is Conserved among Humans, Rhesus Macaques, and Cynomolgus Macaques

This article is featured in In This Issue, p.1

Major histocompatibility complex E is a nonclassical MHC-Ib molecule encoded by the MHC-E locus. Similar to other MHC-Ib molecules, the human MHC-E molecule HLA-E exhibits limited polymorphism; there are only two known functional HLA-E alleles that differ by a single amino acid (1–3). HLA-E binds and presents a subset of 9-mer peptides derived from the signal sequences of HLA-A, B, C, and G molecules (4–6). These HLA-E/signal peptide complexes bind to CD94/NKG2 receptors on NK cells, regulating NK cell activation (7–9). However, HLA-E also binds and presents other self- and pathogen-derived peptides to HLA-E–restricted CD8⁺ T cells, which recognize HLA-E–bound peptide through the TCR (10–16). Pathogen-specific HLA-E–restricted CD8⁺ T cell responses are elicited by a number of bacterial and viral pathogens, including Mycobacterium tuberculosis, Salmonella enterica, EBV, human CMV, and hepatitis C virus (12, 13, 15, 17–21). Pathogen-specific, HLA-E–restricted CD8⁺ T cells secrete antiviral cytokines and recognize and kill infected cells (12, 14, 15, 17, 22), but the impact of these unconventional MHC-E–restricted CD8⁺ T cell responses on pathogen control has not been thoroughly investigated. Such studies would be facilitated by the identification of a nonhuman primate (NHP) model that mirrors the salient features of HLA-E immunobiology.

Orthologs of HLA-E have been identified in many species, including mice (Qa-1b), rats (RT-BM1), and NHPs, including rhesus macaques (RM) (Mamu-E) (23–28). The role of MHC-E in both NK cell regulation and Ag presentation to CD8⁺ T cells is conserved in mice (29–34), but few studies have investigated MHC-E function in physiologically relevant NHP models. Recently, we described the induction of pathogen-specific MHC-E–restricted CD8⁺ T cell responses in RM (Macaca mulatta) after vaccination with rhesus CMV (RhCMV)–based vaccine vectors (35), confirming the role of Mamu-E in Ag presentation to CD8⁺ T cells. RhCMV-based vaccination with SIV Ags (RhCMV/SIV) elicits SIV-specific, Mamu-E–restricted CD8⁺ T cells, and results in robust control and clearance of SIV infection in approximately 50% of vaccinated RM (36), suggesting pathogen-targeted MHC-E–restricted CD8⁺ T cells might serve as effective antiviral immune responses. Although these findings suggest macaques could be used to model the impact of HLA-E–restricted CD8⁺ T cell responses on infection and disease, it is unclear whether RM accurately model human MHC-E immunobiology.

The classical MHC-Ia molecules that typically present antigenic peptides to CD8⁺ T cells are highly polymorphic (37, 38), particularly in the amino acids lining the peptide-binding groove. In contrast, MHC-E molecules exhibit relatively limited diversity within and across species, including complete conservation of the peptide-binding groove among nearly all primate MHC-E molecules identified to date (26, 28, 39). Indeed, on the sequence level, the MHC-E locus is the most well conserved of all known primate MHC-I genes (2, 39). However, previous studies have demonstrated increased MHC-E diversity in RM compared with humans (26), suggesting potential functional diversity between macaque and human MHC-E. In this study, we investigated the degree to which macaque MHC-E mirrors HLA-E functionality, to evaluate NHP models that could be employed to study HLA-E–restricted CD8⁺ T cells.

In this study, we describe MHC-E immunobiology in two distinct populations of macaques commonly used in biomedical research: outbred Indian-origin RM (Macaca mulatta) and Mauritian-origin cynomolgus macaques (MCM; Macaca fascicularis). RM possess extremely complex MHC genetics compared with humans, with individual RM expressing up to 20 distinct MHC-Ia molecules that present pathogen epitopes to CD8⁺ T cells (40, 41). In contrast, MCM are an insular population of cynomolgus macaques descended from a small founder population (42, 43), and thus possess simplified MHC genetics relative to other NHPs (44). MCM express MHC molecules from seven defined MHC haplotypes, termed M1 through M7, each of which is fully defined for MHC-Ia and MHC class II (MHC-II) alleles (45, 46). Thus, we hypothesized that, compared with Indian-origin RM, MCM would possess less MHC-E genetic diversity and thus more closely mirror the limited diversity of HLA-E in humans. In addition, we hypothesized that species-specific sequence differences among primate MHC-E molecules would confer functional differences, such as differences in the binding of peptide ligands. To address these hypotheses, we evaluated Indian RM, MCM, and human MHC-E for differences in genetic diversity, expression on immune cells, peptide binding, Ag presentation to CD8+ T cells, and modulation by viruses.

Overall, our data revealed striking sequence similarity among RM, MCM, and human MHC-E allomorphs. Despite differences in the level of MHC-E diversity among the three species, primate MHC-E molecules exhibited a high degree of sequence conservation in the α1 and α2 domains that form the peptide-binding groove and contact the TCR, including complete conservation of the peptide-binding residues in 35 of 39 MHC-E alleles. We observed similar expression of primate MHC-E on the surface of peripheral blood T cells, B cells, NK cells, and monocytes. In addition, MHC-E was similarly upregulated on the surface of SIV-infected RM, MCM, and human CD4⁺ T cells, in contrast to MHC-I, which was downregulated. In functional studies, primate MHC-E molecules bound and presented identical peptides to MHC-E–restricted CD8⁺ T cells, which exhibited allogeneic and cross-species Ag recognition of MHC-E–bound epitopes. The data described in this study demonstrate functional conservation of primate MHC-E and support the use of RM and MCM as physiologically relevant models of HLA-E immunobiology.

All Indian-origin RM and MCM described in this study were used with the approval of the Oregon National Primate Research Center Institutional Animal Care and Use Committee or the Wisconsin National Primate Research Center Institutional Animal Care and Use Committee, under the standards of the U.S. National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Sequencing of Mamu-E genomic DNA to identify novel Mamu-E alleles was performed using NimbleGen HLA SeqCap reagents to capture genomic MHC sequences from 31 RM as described by Cao et al. (47). The captured MHC products were subsequently sequenced with Illumina HiSeq following the manufacturer’s instructions. The resulting 90 bp paired-end reads are assembled into ∼1.1 kb contigs containing exons 1–3 of MHC-E. Characterization of full-length Mafa-E transcripts was achieved by isolating RNA from peripheral mononuclear cells homozygous for each of the four most common MHC haplotypes (M1, M2, M3, and M4) in the MCM population. Briefly, cDNA generated from these RNA samples with the Superscript III First-Strand Synthesis System (Invitrogen). Complete Mafa-E open reading frames from these cDNAs were amplified with Phusion high-fidelity polymerase (New England Biolabs) using the following pair of locus-specific primer sequences that target conserved regions of the 5′ and 3′ untranslated regions: 5′-CACGACTCCCGACTATAAAG-3′ and 5′-TTGCACACAAGGCAGCTGTC-3′. Unique 16 bp barcode sequences (Pacific Biosciences) were used for each cDNA sample with the barcodes incorporated on the 5′ end of the locus-specific primer sequences during oligonucleotide synthesis. An amplicon library was created by the addition of SMRTbell adaptors according to the manufacturer’s protocols and subjected to circular consensus sequencing on a PacBio RSII instrument (Pacific Biosciences). The resulting sequences were analyzed as described recently by Karl et al. (48) using a series of open-source bioinformatics tools and the Geneious Pro v9.1.7 software (Biomatters). Next, 454 MHC-E typing of 51 RM was performed as previously described for MHC-I (41). Briefly, amplicons of Mamu class I sequences were generated via amplification of cDNA by PCR using high-fidelity Phusion polymerase (New England Biolabs) and a pair of universal MHC-I–specific primers with the following thermocycling conditions: 98°C for 3 min, (98°C for 5 s, 57°C for 1 s, 72°C for 20 s) for 23 cycles, and 72°C for 5 min. Each PCR primer contained a unique 10 bp Multiplex Identifier (MID) tag along with an adaptor sequence for 454 Sequencing (5′-GCCTCCCTCGCGCCATCAG-MID-GCTACGTGGACGACACG-3′; 5′-GCCTTGCCAGCCCGCTCAG-MID-TCGCTCTGGTTGTAGTAGC-3′). Resulting amplicons span 190 bp of a highly polymorphic region within exon two. The primary cDNA-PCR products were purified using AMPure XP magnetic beads (Beckman Coulter Genomics). Emulsion PCR and pyrosequencing procedures were carried out with Genome Sequencer FLX instruments (Roche/454 Life Sciences) as per the manufacturer’s instructions. Data analysis was performed using Geneious-Pro bioinformatics software (Biomatters) for sequence assembly. An additional 64 RM were typed for MHC-E molecules by unbiased RecA-mediated capture of MHC-I cDNAs, cloning, and subsequent Sanger sequencing following the procedure outlined by Zhumabayeva et al. (49), using the following biotinylated capture probe that binds to a highly conserved region of the MHC-I α3 domain (5′-CGGAGATCAYRCTGACVTGGC-3′). GenBank accession numbers for novel MHC-E alleles are: Mafa-E*02:01:02 (MF004403), Mafa-E*02:03:02 (MF004404), Mafa-E*02:13 (MF004405), Mafa-E*02:14 (MF004406), Mamu-E*02:24 (MF004407), Mamu-E*02:25:01 (MF004408), Mamu-E*02:25:02 (MF004411), Mamu-E*02:26 (MF004409), Mamu-E*02:27 (MF004410), Mamu-E*02:28 (MF04412), Mamu-E*02:29 (MF004413), Mamu-E*02:30 (MF004414) (https://www.ncbi.nlm.nih.gov/genbank/). Sequences were submitted to the IPD-MHC database (50) and given official designations (https://www.ebi.ac.uk/ipd/mhc/). MHC-E α1- α2 aa sequences were aligned using Geneious 7.1 software (Biomatters). Phylogenetic trees were constructed using PHYML 3.0 (51), using the LG amino acid substitution model (52), evaluated using 1000 bootstrap replicates.

The creation of single-chain trimer constructs has been previously described in detail (35, 53). Briefly, each construct encodes a fusion protein of MHC-E signal peptide, peptide of interest, human β-2-microglobulin (β2M), the mature form of MHC-E of interest or Mamu-A1*001:01 (Mamu-A*01) (α1 through cytoplasmic domain), and EGFP connected by flexible linker regions [(GGGGS)_X]. Transfections of HEK 293T cells with single-chain trimer constructs were conducted as previously described (35). Briefly, transfections were carried out in six-well plates using GeneJuice (Millipore) as per the manufacturer’s instructions. Twenty-four hours posttransfection, 293T cells were stained with MHC-E primary Ab (clone 4D12) in 100 μl of PBS at 4°C for 15 min, washed twice with PBS, stained with live/dead fixable yellow dead cell stain and secondary Ab [Allophycocyanin Goat-Anti-Mouse (H+L) F(ab′)₂ Fragment], washed twice with PBS, and fixed in 100 μl of 2% PFA (Electron Microscopy Sciences). As a background control, a second aliquot of sample was stained in the absence of MHC-E primary Ab. Cells were collected on an LSR-II instrument (BD Biosciences) and analyzed using FlowJo (Tree Star).

All peptides used in these studies were synthesized by Genscript. B lymphoblastoid cell lines (BLCL) were generated by infecting macaque and human PBMC with herpesvirus papio or EBV, respectively, as previously described (54–56). A mammalian expression vector for Mafa-E*02:01:02 was generated by ligating the full-length Mafa-E*02:01:02 coding sequence into pCEP4 HindIII/NotI restriction sites. Plasmid was cloned in DH5α Escherichia coli (Life Technologies), sequence confirmed, and electroporated into MHC-I–negative K562 cells (57) using the Amaxa Nucleofector kit C (Lonza), program G-016. Transfected cells were maintained on drug selection (Hygromycin B; Corning), and routinely confirmed for surface expression of MHC-I by staining with pan–MHC-I Ab clone W6/32 alongside negative control K562 cells. Throughout use in T cell assays, RNA from Mafa-E*02:01:02 transfectants was extracted using the AllPrep DNA/RNA Mini Kit (Qiagen), amplified by RT-PCR using primer pairs flanking a highly polymorphic region within exon 2, and sequence confirmed by Sanger sequencing. APCs were pulsed with peptides of interest at a final concentration of 10 μM for 90 min then washed three times with warm PBS and once with warm R10 (RPMI 1640 supplemented with 10% FBS, l-glutamine, and antibiotic or antimycotic) to remove unbound peptide before combining with freshly isolated PBMC at an E:T ratio of 10:1 (PBMC/APC). To stabilize MHC-E surface expression (8, 58), the Mafa-E*02:01:02 transfectants were incubated at 27°C overnight prior to use in assays and maintained at 27°C throughout peptide incubation until combined with effectors. SIV-infected CD4⁺ cells were generated by isolation of CD4⁺ T cells from PBMC with CD4 microbeads and LS columns (Miltenyi Biotec), activation with a combination of IL-2 (National Institutes of Health AIDS Reagent Program), Staphylococcus enterotoxin B (Toxin Technologies), and anti-CD3 (Nonhuman Primate Reagent Resource), anti-CD28, and anti-CD49d mAbs (BD Biosciences), and spinoculation with sucrose-purified SIVmac239. Prior to use in T cell assays, SIV-infected target cells were purified using CD4 microbeads and LS columns (Miltenyi Biotec), as previously described (59). Infected target-cell preparations were confirmed to be >95% CD4⁺ T cells prior to infection and >50% SIV-infected following enrichment, and were used at an E:T ratio of 40:1 (effectors: CD8β⁺ T cells) or 8:1 (effectors: CD8⁺ T cell lines). In these experiments, uninfected, activated CD4⁺ T cells served as negative control APCs [uninfected targets from SIV+ macaques and HIV+ patients were cultured with 400 μM tenofovir (National Institutes of Health AIDS Reagent Program)].

SIV-specific CD8⁺ T cell responses in PBMCs or CD8⁺ T cell lines were measured by flow cytometric intracellular cytokine staining (ICS). PBMC were isolated from anticoagulant-treated whole blood by Ficoll density gradient centrifugation (GE Healthcare). In SIV recognition experiments, CD8β⁺ effectors were isolated from PBMC via Miltenyi sorting using CD8β-PE and anti-PE beads. CD8⁺ T cell lines were generated as previously described (59). Briefly, PBMC from a Mamu-A*01 Ad5/gag-vaccinated RM were stimulated with Gag peptide-pulsed, irradiated autologous BLCL weekly, and cultured in R15 supplemented with 100 U/ml IL-2 (National Institutes of Health AIDS Reagent Program). PBMC, isolated CD8β⁺ T cells, or CD8⁺ T cell lines were incubated with peptide-pulsed APCs or SIV-infected CD4⁺ T cells, and the costimulatory molecules CD28 and CD49d (BD Biosciences) for 1 h, followed by addition of brefeldin A (Sigma-Aldrich) for an additional 8 h. Costimulation with unpulsed APCs or uninfected CD4⁺ T cells served as background controls. In SIV recognition assays, the MHC association (MHC-Ia, MHC-E, MHC-II) of a response was determined by preincubating SIV-infected CD4⁺ cells for 1 h at room temperature (prior to combining effector and target cells and incubating per the standard ICS assay) with the following blockers: 1) the pan anti–MHC-I mAb W6/32 (10 mg/ml), 2) the MHC-II–blocking CLIP peptide (MHC-II–associated invariant chain, aa 89–100; 20 μM), and 3) the MHC-E–blocking VL9 peptide (VMAPRTLLL; 20 μM), alone or in combination (blocking reagents were not washed, but remained throughout the assay). Stimulated cells were stained, collected, and analyzed as previously described (35). Briefly, cells were washed with 1× PBS, surface stained for 30 min, washed with FACS (1× PBS supplemented with 10% FBS), fixed with 2% paraformaldehyde, permeabilized with saponin buffer, and stained intracellularly for 45 min. Sample collection was performed on an LSR-II instrument (BD Biosciences), and analysis was conducted with FlowJo software (Tree Star).

MHC-E surface stains of PBMC and SIV-infected CD4⁺ T cells were performed by washing cells once with PBS, staining with MHC-E primary Ab (clone 4D12) for 30 min at room temperature, washing twice with FACS (1× PBS supplemented with 10% FBS), staining with secondary Ab (PE-Cy7 anti-mouse IgG1), and washing twice with PBS. The cells were then stained for additional surface and intracellular markers. For analysis of MHC-E stains of PBMC, initial gating was performed on live singlets. Subsequent gating of subsets was done as follows: CD4⁺ T cells (CD3⁺, CD4⁺, CD8⁻), CD8⁺ T cells (CD3⁺, CD4⁻, CD8⁺), B cells (CD20⁺, CD3⁻), NK cells (CD3⁻, CD20⁻, CD8⁺), and monocytes (CD3⁻, CD20⁻, CD8⁻). Subsets of CD4⁺ and CD8⁺ T cells were further defined as follows: naive (CD95⁻, CD28⁺), central memory (CD95⁺, CD28⁺), and effector memory (CD95⁺, CD28⁻). Subsets of monocytes were further defined as follows: classical (CD14⁺, CD16⁻), intermediate (CD14⁺, CD16⁺), and nonclassical (CD14⁻, CD16⁺). For MHC-E stains of HIV/SIV-infected CD4⁺ T cells, infected CD4⁺ T cells were generated as described above without postinfection purification and stained for surface MHC-E, MHC-I, or Mamu-A*01, CD3, and CD4, and intracellular SIVgag p27 or HIVgag p24. As a background control, a second aliquot of sample was stained in the absence of MHC-E primary Ab. For all stains, sample collection was performed on an LSR-II instrument (BD Biosciences), and analysis was done using FlowJo software (Tree Star). HIV type 1 (HIV-1) LAI was used for HIV-1 infections (AIDS Reagent Resource).

The following conjugated Abs were used in these studies: 1) from BD Biosciences, SP34-2 (CD3; PacBlu, Alexa700), SK1 (CD8a; TruRed, AmCyan), 25723.11 (IFN-γ; APC), MAb11 (TNF; FITC, Alexa700), DX2 (CD95; FITC), 28.2 (CD28; PE), 2H7 (CD20; APC-H7), RPA-T8 (CD8; PacBlu), L200 (CD4; PerCP-Cy5.5), 3G8 (CD16; FITC), 2) from BioLegend, W6/32 (pan-MHC-I; PerCP-Cy5.5), OKT-4 (CD4; PE-Cy7), 3) from Miltenyi Biotec, M-T466 (CD4; APC), 4) from eBioscience, M1-14D12 (mouse IgG1; PE-Cy7), 5) from Life Technologies, Allophycocyanin Goat-Anti-Mouse (H+L) F(ab′)₂ Fragment, 6) from Beckman Coulter, RMO52 (CD14; ECD), and 7) from the Nonhuman Primate Reagent Resource, Mamu-A*01-PE (catalog no. PR-1102). The following unconjugated Abs were used in these studies: 1) from Advanced BioScience Laboratories, 4324 (SIV Gag p27), conjugated in-house to FITC using FluoReporter FITC Protein Labeling Kit (Invitrogen), 2) 4D12 (HLA-E), grown and purified in-house, and 3) W6/32 (pan–MHC-I), grown and purified in-house. Live/dead Fixable Yellow Dead Cell Stain (Life Technologies) was used to assess cell viability.

The following statistical analyses were conducted with PRISM software (GraphPad) unless otherwise stated. One-way ANOVA was used to test for differences in surface MHC-E modulation among HIV/SIV-infected target types. The nonparametric Kruskal–Wallis test was used to test for differences in immune cell–subset MHC-E surface levels among primate species. Paired t tests were used to test for differences in immune cell–subset MHC-E surface levels pre- and post-SIV infection. The Fisher–Pitman permutation test (R3.3.2 software, Comprehensive R Archive Network) was used to test for differences in peptide-induced stabilization among primate MHC-E molecules. Mixed-effects ANOVA (SAS9.4 software, SAS Institute) was used to test for differences in SIV-specific CD8⁺ T cell response frequencies among target types for each blocking condition, as well as differences between CLIP blockade and other blockade conditions within each target type (Bonferroni-adjusted for multiple comparisons).

The human population encodes only two HLA-E alleles that differ by a single amino acid outside the peptide-binding region, and thus individuals express one or two HLA-E molecules of extremely limited diversity (1–3). In contrast, 25 RM MHC-E (Mamu-E) alleles encoding 23 distinct Mamu-E molecules have been identified in RM, including two Mamu-E molecules that differ at predicted peptide-binding residues (50). Further, one study found individual RM expressing greater than two Mamu-E molecules (27), suggesting more than one MHC-E gene locus exists in RM. To further investigate the diversity of RM MHC-E, we employed multiple sequencing methods. First, we enriched for MHC-I genomic DNA with NimbleGen HLA SeqCap regents, deep-sequenced 90 bp paired-end reads with Illumina HiSeq, and assembled into contigs corresponding to exons 1–3 of MHC-I. Using this technique, we identified eight novel Mamu-E alleles, two of which encoded the same amino acid sequence (Mamu-E*02:25:01 and Mamu-E*02:25:02) (Supplemental Fig. 1). Thus, the RM population encodes at least 33 Mamu-E alleles corresponding to 30 distinct Mamu-E molecules.

To assess the diversity of Mamu-E expressed in individual RM, we typed 105 animals for Mamu-E molecules via two independent methods, 454 deep-sequencing of 190 bp amplicons of MHC-I (41) or RecA capture of MHC-I cDNAs, and subsequent Sanger sequencing (49). Both Mamu-E typing techniques demonstrated that individual RM express between one and four distinct Mamu-E transcripts (Supplemental Fig. 2). Cumulatively, these data confirm that RM possess increased MHC-E allelic diversity compared with humans at both the population and individual level (Fig. 1A).

Next, we assessed MHC-E diversity in MCM. We focused on the most common MHC haplotypes, M1, M2, M3, and M4, which are carried by 88% of MCM (45, 60). Using primers that bind in the 5′ and 3′ untranslated regions of MHC-E transcripts, we sequenced full-length Mafa-E cDNAs from MCM homozygous for each of the M1, M2, M3, and M4 MHC haplotypes. Sequencing identified one Mafa-E transcript expressed from the M1 haplotype (Mafa-E*02:01:02), another shared and expressed from both the M2 and M3 haplotypes (Mafa-E*02:03:02), and two additional Mafa-E transcripts expressed from the M4 haplotype (Mafa-E*02:13 and Mafa-E*02:14) (Supplemental Fig. 1). Thus, individual MCM within the M1-M4-carrying population express between one and three Mafa-E molecules, a similar level of MHC-E diversity as individual RM (Fig. 1A). However, the Mafa-E*02:01:02 (M1) and Mafa-E*02:03:02 (M2M3) molecules differ by only 6 aa, including only 1 aa difference in the α1 and α2 domains that form the peptide-binding region (Supplemental Fig. 1). Thus, individual MCM within the M1-M3–carrying population express one or two MHC-E molecules of extremely limited diversity, closely resembling both population and individual HLA-E diversity in humans (Fig. 1A).

To understand the relationship of MHC-E alleles from RM, MCM, and humans, we constructed a phylogenetic tree based on the α1-α2 aa sequences of the 33 Mamu-E, four Mafa-E, and two HLA-E alleles (Fig. 1B). We noted a minimum of 85% aa sequence identity in the α1-α2 region among the 39 primate MHC-E alleles, with 100% identity among a number of Mamu-E and Mafa-E molecules (e.g., Mafa-E*02:03:02, Mamu-E*02:07, and Mamu-E*02:19). Examining the sequence diversity at each amino acid position along the α1-α2 region of the 39 MHC-E molecules revealed that most differences occur in amino acids not predicted to line the peptide-binding groove (Fig. 1C, Supplemental Fig. 1) (61, 62). Indeed, 16 of 22 predicted peptide-binding residues are completely conserved in all 39 primate MHC-E molecules with the remaining six residues differing in only 4 of the 39 MHC-E molecules (Mamu-E*02:15, Mamu-E*02:20, Mamu-E*02:30, and Mafa-E*02:14) (Fig. 1C, Supplemental Fig. 1). Of note, every novel MHC-E molecule identified possesses the conserved acid residues required for interaction with β2M and conformation of a typical Ag-presenting MHC-I molecule: positions 3 (histidine), 10 (threonine), 29 (aspartic acid), 96 (glutamine), 101 (cysteine), 120 (glycine), and 164 (cysteine) (27) (Supplemental Fig. 1). Thus, we next investigated the ability of these Mamu-E, Mafa-E, and HLA-E molecules to bind and present peptides.

The majority of previously described pathogen-derived HLA-E–binding peptides closely resemble the predominant HLA-E–stabilizing MHC-I signal sequence-derived peptides, which possess hydrophobic residues at positions 2 and 9 (4–6, 13–15, 17–19, 21, 63). However, we previously demonstrated that SIV-specific, Mamu-E–restricted CD8⁺ T cells in RhCMV/SIV-vaccinated RM target a broad range of diverse peptides that do not resemble the canonical signal sequence-derived peptides (35). In addition, HLA-E*01:03 was able to bind and present these diverse Mamu-E–binding peptide epitopes (35), suggesting primate MHC-E molecules share a peptide-binding repertoire composed of a more diverse pool of peptides than previously appreciated. To compare binding of peptide ligands among primate MHC-E molecules, we used single-chain trimer constructs that encode a single fusion protein whereby the peptide of interest is covalently linked to both the MHC-E H chain and β2M via flexible peptide linkers (35). As MHC-E surface expression is dependent upon binding an appropriate ligand (4, 6), transfection of single-chain trimer constructs into 293T cells assesses the ability of a peptide to stabilize surface expression of the encoded MHC-E molecule. Thus, we tested the ability of four previously described SIVgag-derived Mamu-E–binding peptide epitopes to stabilize surface expression of four primate MHC-E molecules, HLA-E*01:01, HLA-E*01:03, Mamu-E*02:04, and Mafa-E*02:01:02 (Fig. 2). We compared the level of MHC-E stabilization to positive control constructs containing VMAPRTLLL (VL9), a canonical signal sequence-derived MHC-E–binding peptide, and negative control constructs containing TVCVIWCIH, an SIVgag epitope not targeted by Mamu-E–restricted CD8⁺ T cells. As expected, VL9 strongly enhanced surface levels of each MHC-E molecule tested, whereas TH9 failed to stabilize MHC-E surface expression (Fig. 2, Supplemental Fig. 3). Transfection of single-chain trimer constructs encoding peptides Gag_65–73(AE9), Gag_276–284(RL9), and Gag_389–397(KG9) consistently increased surface expression of all four primate MHC-E molecules, whereas Gag_132–140(GI9) only consistently increased surface expression of macaque MHC-E molecules Mamu-E*02:04 and Mafa-E*02:01:02. Of note, stabilization of HLA-E*01:01 by Gag_65–73(AE9), Gag_276–284(RL9), Gag_389–397(KG9), and control peptide VL9 was weaker than that observed for the other MHC-E molecules. This is consistent with previous studies demonstrating HLA-E*01:01 is expressed at lower levels and exhibits a weaker affinity for MHC-I–derived signal peptides compared with HLA-E*01:03 (64–66). As expected, transfection of single-chain trimer constructs encoding RM MHC-Ia molecule Mamu-A*01 revealed that none of the tested peptides were able to stabilize Mamu-A*01 on the cell surface, with the exception of known Mamu-A*01–binding peptide Gag_181–189(CM9) (Fig. 2, Supplemental Fig. 3). Together, these data indicate that although primate MHC-E molecules differ in sequence, Mamu-E, Mafa-E, and HLA-E are able to bind the same peptide ligands.

Each T cell recognizes antigenic peptide in the context of a particular MHC molecule; the TCR contacts both the peptide and the presenting MHC molecule (67–69). Thus, T cell recognition of peptide is restricted to a particular MHC molecule, which typically precludes Ag recognition of cells not expressing the restricting MHC molecule. However, the low level of diversity among primate MHC-E molecules and their ability to bind identical peptides suggested that MHC-E–restricted CD8⁺ T cells might be uniquely able to recognize antigenic peptides presented across distinct primate MHC-E molecules. Indeed, we previously demonstrated that SIV-specific, Mamu-E*02:04–restricted CD8⁺ T cells from RM responded to SIV peptide presented by Mamu-E*02:11, Mamu-E*02:20, and HLA-E*01:03 (35). To extend this finding to a wider range of primate MHC-E molecules, including Mafa-E molecules, we tested the ability of SIV-specific, Mamu-E–restricted CD8⁺ T cells from RhCMV/SIV-vaccinated RM to recognize SIVgag peptide presented by a K562 cell line transfected to singly express Mafa-E*02:01:02, as well as panels of human, RM, and MCM BLCLs expressing a variety of MHC-E molecules (Fig. 3). We focused on the presentation and T cell recognition of SIVgag_276–284(RL9), an epitope targeted by Mamu-E–restricted CD8⁺ T cells in every RhCMV/SIV-vaccinated RM (35). APCs were pulsed with RL9, washed to remove unbound peptide, and combined with PBMC from RM mounting an RL9-specific CD8⁺ T cell response. ICS for TNF-α and IFN-γ revealed that Mamu-E–restricted CD8⁺ T cells were able to recognize RL9 when pulsed onto Mafa-E*02:01:02–expressing K562 cells, but not when pulsed onto negative control untransfected K562 cells, demonstrating the ability of an MCM MHC-E molecule to present peptide to RM CD8⁺ T cells. In addition, RL9-pulsed BLCLs derived from humans, RM, and MCM were each able to activate Mamu-E–restricted RL9-specific CD8⁺ T cells, whereas BLCLs lacking peptide were not, confirming the ability of diverse human and macaque MHC-E molecules to present the same peptide to MHC-E–restricted CD8⁺ T cells. These data also demonstrate the promiscuity of Mamu-E–restricted CD8⁺ T cells, which do not appear to be restricted by specific MHC-E molecules, but rather recognize their antigenic peptide presented across HLA-E, Mamu-E, and Mafa-E molecules.

Our studies thus far have demonstrated that SIV-specific, Mamu-E–restricted CD8⁺ T cells from RhCMV/SIV-vaccinated RM recognized SIV peptide pulsed onto heterologous RM, MCM, and human APCs (Fig. 3). However, we sought to confirm that presentation of antigenic peptide by primate MHC-E occurs naturally upon SIV infection. We coincubated CD8⁺ T cells from RhCMV/SIVgag-vaccinated RM with SIVmac239-infected RM, MCM, or human CD4⁺ T cells in vitro, and assayed for CD8⁺ T cell recognition of infected targets by TNF-α and IFN-γ induction (Fig. 4). Remarkably, we observed consistent CD8⁺ T cell recognition of SIV-infected CD4⁺ T cells regardless of the individual or species from which infected targets were derived, confirming allogeneic and cross-species recognition of naturally processed SIV Ag by RhCMV/SIV-elicited CD8⁺ T cells. Importantly, in addition to induction of SIV-specific, Mamu-E–restricted CD8⁺ T cell responses, RhCMV/SIV vaccination also elicits a subset of SIV-specific CD8⁺ T cell responses restricted by MHC-II molecules (54). Thus, to confirm Mamu-E–restricted CD8⁺ T cells participate in the observed cross-recognition of nonautologous SIV-infected targets, we preincubated SIV-infected CD4⁺ T cell targets with reagents that specifically block MHC-II (CLIP), bulk MHC-I (mAb W6/32), or MHC-E (VL9). In every case, we observed incomplete blocking of the SIV-specific CD8⁺ T cell response with CLIP alone, but complete blocking with a combination of VL9 and CLIP, confirming that Mamu-E–restricted CD8⁺ T cells cross-recognize SIV-infected CD4⁺ T cells derived from heterologous RM, MCM, and humans. In contrast, recognition by Mamu-A*01–restricted CD8⁺ T cells specific for SIVgag_181–189(CTPYDINQM) was limited to SIV-infected CD4⁺ T cells derived from Mamu-A*01⁺ RM (Supplemental Fig. 4). Thus, whereas Ag recognition by conventional MHC-Ia–restricted CD8⁺ T cells is limited to APCs derived from a small subset of individuals within the same species, Mamu-E–restricted CD8⁺ T cells recognize Ag across heterologous APCs, even those derived from other primate species.

In humans, HLA-E is ubiquitously expressed in virtually all nucleated cells, with particularly high levels expressed on the surface of immune cell subsets (8, 25, 70). To compare surface expression levels of macaque MHC-E to that of humans, we stained PBMC subsets with an HLA-E–specific mAb (clone 4D12) previously shown to cross-react with Mamu-E (35) (Fig. 5). Similar to the ubiquitous pattern of HLA-E expression in humans, we detected MHC-E on the surface of all major immune cell subsets of RM and MCM. Although surface MHC-E expression varied among individuals, we observed no statistically significant difference in levels of MHC-E among the three primate species on the surface of T cells, B cells, NK cells, and monocytes. Thus, human, RM, and MCM MHC-E are similarly expressed on the surface of immune cell subsets.

A number of pathogens possess mechanisms to upregulate HLA-E surface expression on infected cells to evade NK cell killing (21, 63, 71). Whereas the Nef protein encoded by HIV and SIV downregulates surface expression of MHC-Ia molecules to evade CD8⁺ T cell recognition, HLA-E and Mamu-E are impervious to Nef and instead are upregulated on the surface of HIV-1– and SIV-infected cells, respectively (35, 72, 73). Thus, we sought to compare the impact of SIV infection on surface expression of Mamu-E, Mafa-E, and HLA-E. Infection of RM, MCM, and human CD4⁺ T cells with SIVmac239 in vitro led to increased levels of MHC-E, but decreased levels of bulk MHC-I, on the surface of infected cells (Gag⁺) compared with uninfected cells (Gag⁻) in the same well (Fig. 6A, 6C). In agreement with previous studies, HIV-1 infection of human CD4⁺ T cells also led to increased levels of surface HLA-E. Thus, HLA-E, Mamu-E, and Mafa-E are similarly upregulated by SIV infection, and MHC-E upregulation is a shared consequence of both HIV and SIV infection. As the pan–MHC-I mAb clone W6/32 binds to all MHC-I molecules, including MHC-E, we also tested a mAb that binds specifically to the RM MHC-Ia molecule Mamu-A*01 (Fig. 6C). This allowed for more accurate and robust detection of SIV-induced MHC-Ia downregulation in Mamu-A*01⁺ RM cells, and further highlighted the disparate modulation of MHC-E and MHC-Ia by SIV (Fig. 6C).

To investigate SIV-induced MHC-E modulation in vivo, we examined MHC-E expression on PBMC subsets pre- and post-SIVmac239 infection in a cohort of RM acutely infected with SIVmac239 (Fig. 6D). In agreement with our in vitro data from SIV-infected CD4⁺ T cells, we observed upregulation of MHC-E on the surface of CD4⁺ T cells 14 d after in vivo SIV infection. In addition, CD8⁺ T cells, NK cells, B cells, and monocytes also expressed higher levels of surface MHC-E after in vivo SIV infection, demonstrating surface MHC-E upregulation occurs in bystander cells that are not direct targets of SIV infection. In line with this finding, during our in vitro SIV infections, we observed elevated levels of surface MHC-E on bystander uninfected cells (Gag⁻) cultured with infected cells (Gag⁺) compared with cells in uninfected wells (Fig. 6A, 6B). Together, these data suggest that MHC-E upregulation is not necessarily restricted to SIV-infected cells, but rather represents a global response to viral infection.

To our knowledge, this is the first study investigating the functionality of MHC-E across humans and NHPs. We conducted analyses of macaque MHC-E surface expression, binding, and presentation of peptide ligands to CD8⁺ T cells, and modulation by HIV and SIV. Together, the data presented in this study demonstrate striking a similarity in the sequence, expression, and function of MHC-E in Indian-origin RM, MCM, and humans.

In contrast to classical MHC-Ia molecules, primate MHC-E molecules share a remarkable degree of sequence conservation, particularly in the peptide-binding groove. Whereas previously identified macaque MHC-E alleles were discovered via cloning and Sanger sequencing of MHC-E cDNAs, in this study we used multiple deep-sequencing techniques to sequence MHC-E from RM and MCM, and identified eight novel Mamu-E and four novel Mafa-E alleles. Of note, we were able to assign novel Mafa-E alleles to MCM MHC haplotypes M1, M2, M3, and M4, and thus these haplotypes are now fully defined for MHC-Ia, MHC-II, and MHC-E alleles (45, 46), further increasing the value of MCM as an MHC-defined macaque resource. In addition, this sequence analysis identified a population of MCM, those carrying the common M1, M2, and M3 haplotypes, whose MHC-E diversity closely mirrors HLA-E diversity in humans. Similar to humans, individual M1-M3–carrying MCM express one or two MHC-E molecules of extremely limited diversity. In contrast, RM express a greater diversity of MHC-E molecules compared with humans at both the population and individual level. The observed expression of >2 MHC-E molecules by individual RM and MCM suggests potential duplications of the MHC-E locus in these species, similar to those observed with MHC-Ia loci (40, 45). Although this differs from the single MHC-E locus present in humans (74), the implications of macaque MHC-E locus duplication are currently unclear, as we did not observe significant functional differences among Mamu-E or Mafa-E molecules in this study. However, this could reflect differences among MHC-E molecules not examined in this study, such as the ability to ligate NK cell receptors.

Perhaps the strongest evidence for the functional conservation of primate MHC-E molecules presented in this study is the ability of Mamu-E–restricted CD8⁺ T cells to recognize peptide Ag presented across RM, MCM, and human MHC-E molecules. Such broad cross-species recognition by Mamu-E–restricted CD8⁺ T cells demonstrates similarity among primate MHC-E molecules in both the binding of peptides and the ligation of TCRs. Indeed, transfection of single-chain trimer constructs encoding various combinations of MHC-E and linked peptide revealed stabilization of surface Mamu-E*02:04, Mafa-E*02:01:02, HLA-E*01:01, and HLA-E*01:03 by the same peptides, providing evidence that primate MHC-E molecules bind a similar repertoire of peptide ligands. Although we, and others, observed weaker peptide-induced stabilization of HLA-E*01:01 compared with the other MHC-E molecules (64), human BLCLs expressing only HLA-E*01:01 were nevertheless able to effectively present Gag_276–284(RL9) to SIV-specific CD8⁺ T cells. Indeed, RM, MCM, and human BLCLs encoding a wide range of MHC-E molecules were each consistently able to present Gag_276–284(RL9) to the Mamu-E–restricted CD8⁺ T cells of four RhCMV/SIVgag-vaccinated RM. Further, this phenomenon of allogeneic and cross-species recognition by Mamu-E–restricted CD8⁺ T cells held true for naturally processed Ag presented by SIV-infected CD4⁺ T cells, suggesting similar pathways of MHC-E Ag presentation exist in macaque and human cells. Data presented in this study only reflect recognition by primed MHC-E–restricted CD8⁺ T cells, and thus subtle differences among MHC-E molecules may impact MHC-E–restricted CD8⁺ T cell priming among species or individuals.

Finally, an appropriate animal model of HLA-E immunobiology must mirror pathogen-induced MHC-E modulation, particularly if used to test the impact of HLA-E–restricted CD8⁺ T cells on infection. Many pathogens, including HIV and SIV, interfere with Ag processing and presentation on MHC-Ia to evade CD8⁺ T cell recognition (reviewed in Ref. 75). In contrast to downregulation of bulk MHC-I, we observed consistent increases in surface MHC-E levels on SIV-infected RM, MCM, and human CD4⁺ T cells and on HIV-infected human CD4⁺ T cells, demonstrating primate MHC-E molecules are similarly upregulated by HIV/SIV infection. Of note, we also observed surface MHC-E upregulation on bystander CD4⁺ T cells after SIV infection in vitro and on CD8⁺ T cells, B cells, NK cells, and monocytes during acute in vivo SIV infection of RM. Shang et al. (76) found increases in Mamu-E transcripts and protein in the female reproductive tract of RM within 4 d of intravaginal SIV infection, primarily in Langerhans cells. Given these findings, we hypothesize that MHC-E expression is globally induced after SIV infection, perhaps due to increases in IFN-γ, which is previously demonstrated to stimulate HLA-E transcription via a mechanism distinct from that of MHC-Ia induction (77, 78). IFN-γ–mediated induction of Mamu-E has also been observed with RM PBMC (79). However, as we observed the highest level of MHC-E surface expression on productively SIV-infected CD4⁺ T cells, we cannot rule out the existence of additional virally encoded mechanisms of MHC-E upregulation, such as surface MHC-E stabilization by SIV-derived MHC-E–binding peptides.

Together, our findings demonstrate robust functional conservation of MHC-E in primates, to the extent that MHC-E molecules can effectively present naturally processed Ag to the CD8⁺ T cells of other primate species. These studies identify two appropriate animal models of HLA-E–restricted CD8⁺ T cells, RM and MCM, which can be used to test the role of unconventional HLA-E–restricted CD8⁺ T cells in pathogen control. As SIV-specific Mamu-E–restricted CD8⁺ T cell responses are engendered upon protective vaccination with RhCMV-based vaccine vectors, eliciting MHC-E–restricted CD8⁺ T cell responses represents a potentially effective vaccine strategy against HIV/SIV. Indeed, conservation of MHC-E Ag presentation to CD8⁺ T cells in humans, RM, and MCM supports the possibility that CMV-based vaccine vectors will elicit MHC-E–restricted CD8⁺ T cells in other primates, including humans. As many viruses upregulate surface MHC-E while downregulating MHC-Ia to simultaneously evade NK cell and MHC-Ia–restricted CD8⁺ T cell immunity, vaccine-induced pathogen-specific MHC-E–restricted CD8⁺ T cells might be effective against other viruses, particularly those well versed in immune evasion. Finally, the limited sequence and functional diversity among MHC-E molecules suggests MHC-E–based CD8⁺ T cell vaccines would elicit similar immune responses among vaccinated individuals, in contrast to conventional CD8⁺ T cell vaccines that rely on the expression of particular MHC-Ia molecules carried by only a fraction of individuals. In conclusion, our data demonstrate that MHC-E is functionally conserved among humans, RM, and MCM, and thus NHPs represent powerful, physiologically relevant animal models of HLA-E immunobiology.

We thank Daniel E. Geraghty for the MHC-E mAb 4D12 hybridoma, David Watkins for sharing MHC RecA capture database, and Steven G. Deeks, Phillip J. Goulder, and Lishomwa C. Ndhlovu for human PBMC samples. We thank the Nonhuman Primate Reagent Resource for providing Mamu-A*01-specific antibody and anti-CD3 antibody, as well as the AIDS Reagent Resource for providing recombinant IL-2 and HIV-1 LAI.

Oregon Health and Science University, J.B.S., S.G.H., and L.J.P. have a financial interest in Vir Biotechnology, Inc., a company that may have a financial interest in the results of this research and technology. This potential individual and institutional conflict of interest has been reviewed and managed by Oregon Health and Science University. The other authors have no financial conflicts of interest.