|Home | About | Journals | Submit | Contact Us | Français|
Murine infection with the Gram-positive intracellular bacterium Listeria monocytogenes activates CD8+ T cells that recognize bacterially derived N-formyl methionine peptides in the context of H2-M3 MHC class Ib molecules. Three peptides, fMIGWII, fMIVIL, and fMIVTLF, are targets of L. monocytogenes-specific CD8+ T cells. To investigate epitope cross-recognition by H2-M3-restricted CD8+ T cells, we deleted the sequence encoding fMIGWII from a virulent strain of L. monocytogenes. Infection with fMIGWII-deficient L. monocytogenes unexpectedly primed CD8+ T cells that stain with fMIGWII/H2-M3 tetramers and lyse fMIGWII-coated target cells in vivo. Because the fMIGWII sequence is nonredundant, we speculated that other bacterially derived Ags are priming these responses. HPLC peptide fractionation of bacterial culture supernatants revealed several distinct L. monocytogenes-derived peptides that are recognized by fMIGWII-specific T cells. Our results demonstrate that the dominant H2-M3-restricted CD8+ T cell population, although reactive with fMIGWII, is primed by other, non-fMIGWII peptides derived from L. monocytogenes. Although this degree of Ag receptor promiscuity is unusual for the adaptive immune system, it may be a more common feature of T cell responses restricted by nonpolymorphic MHC class Ib molecules.
To provide effective defense and immunity against a plethora of pathogens, mammals have developed rapid innate and relatively delayed adaptive immune responses to infection. Innate immune responses are triggered by engagement of germline-encoded receptors, such as Toll-like receptors (1) and nucleotide-binding oligomerization domain proteins (2), with pathogen-associated molecular patterns. Signals mediated by these receptors initiate early antimicrobial defense mechanisms and fortify the developing adaptive immune response. During the adaptive immune response, pathogen-specific B and T cell clones expand and differentiate following recognition of specific Ags.
CD8+ T cells recognize pathogen-derived peptides presented on MHC class I molecules and are important for clearance of many invading pathogens (3). MHC class I molecules can be divided into two groups. MHC class Ia molecules in mice are encoded by the highly polymorphic H-2K, D, and L loci, and their principal function is to present pathogen-derived peptides to CD8+ T cells. MHC class Ib molecules, in contrast, are much less polymorphic, and, in mice, are encoded by genes in the Q, T, and M region of the MHC. Of these, the H2-M3 MHC class Ib molecule has the most completely defined role in antimicrobial immunity. H2-M3 was originally discovered through studies of a minor histocompatibility Ag, maternally transmitted Ag (4), which is a mitochondrial peptide derived from the N terminus of ND1, a subunit of NADH dehydrogenase (5). Mitochondrial protein synthesis initiates with N-formyl methionine, and H2-M3 selectively binds peptides containing this modified amino acid (6–8). The crystal structure of H2-M3 revealed a hydrophobic peptide-binding groove that specifically accommodates N-formyl methionine as an essential anchor residue (9). The specificity for N-formylated peptides positions H2-M3 especially well for presentation of bacterial ligands, because prokaryotic protein synthesis initiates with N-formyl methionine. Several peptides derived from Mycobacterium tuberculosis, for example, can be bound by H2-M3 and presented to CD8+ T cells (10). In the setting of murine infection with Listeria monocytogenes, three H2-M3-restricted peptides, fMIGWII (11), fMIVIL (Fr38) (12), and fMIVTLF (13), have been identified (11, 14). Although most recent studies have demonstrated peptide presentation, one study suggested that bacterial glycolipids may be presented by H2-M3 to L. monocytogenes-specific CD8+ T cells (15).
H2-M3-restricted CD8+ T cells differ in several respects from their MHC class Ia-restricted counterparts. During thymic development, H2-M3-restricted CD8+ T cells are predominantly selected on bone marrow-derived cells rather than thymic cortical epithelial cells (16), resulting in naive T cells with a partially activated (CD44+) phenotype (17). In addition, positive selection of H2-M3-restricted T cells is mediated by a very small subset of mitochondrially encoded peptides, some of which were recently defined as weak agonists (18–20). Thus, compared with MHC class Ia-restricted T cells, the number of selecting ligands is very limited (21).
Distinct thymic selection may have consequences for H2-M3-restricted T cell responses in the peripheral immune compartment. For example, H2-M3-restricted T cells are distinguished by their relative peptide promiscuity. Although most MHC class Ia-restricted T cells are specific to a single pathogen-derived epitope, H2-M3-restricted CD8+ T cells display a much higher degree of peptide cross-reactivity (22). H2-M3-restricted CD8+ T cells also manifest distinct primary and memory responses during bacterial infection. Following primary infection with L. monocytogenes, H2-M3-restricted CD8+ T cells expand and contract more rapidly than H2-Kd-restricted CD8+ T cells (23). Upon rechallenge with L. monocytogenes, H2-M3-restricted memory CD8+ T cells undergo very limited expansion, in contrast to the explosive expansion of MHC class Ia-restricted T cells (17, 23, 24).
In this study, we explored the extent of H2-M3-restricted promiscuity by generating a mutant strain of L. monocytogenes lacking the immunodominant fMIGWII epitope (L. monocytogenes fMIGneg). Surprisingly, immunization with fMIGWII-deficient L. monocytogenes primes fMIGWII-specific CD8+ T cells, as determined by H2-M3/fMIGWII-tetramer staining and in vivo CTL assays. This suggests that other cross-reacting ligands activate CD8+ T cells during L. monocytogenes infection. Indeed, HPLC fractionation of L. monocytogenes culture supernatants revealed several N-formylated peptides that activate fMIGWII-specific CD8+ T cells. Our results suggest that H2-M3-restricted CD8+ T cell responses are primed by a complex mixture of N-formylated bacterial peptides that induce multiple cross-reactive T cell clones. Promiscuous recognition by clonal, MHC class Ib-restricted CD8+ T cells of peptides sharing a similar molecular pattern is an example of the adaptive immune system pirating a fundamental innate immune strategy for antimicrobial defense.
C57BL/6 and CB6/F1 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Eight- to 10-wk-old females were used in all experiments and maintained under specific pathogen-free conditions.
The P815 mastocytoma cell line, which expresses H2-M3, was obtained from the American Type Culture Center (Manassas, VA) and maintained in RPMI/10% FCS.
Bacteria were cultured in brain heart infusion (BHI).3 Wild-type L. monocytogenes strain 10403s was originally provided by D. Portnoy (University of California, Berkeley, CA). To generate the fMIGneg L. monocytogenes strain, the LemA region of L. monocytogenes was mutated in the fMIGWII epitope to fMIVIL by the PCR overlap extension method. The mutation was then incorporated into the chromosome of L. monocytogenes 10403s by homologous recombination, as described previously (25). To confirm the presence of the mutation, the LemA gene region was amplified from the genomic DNA preparation of the L. monocytogenes wild-type and L. monocytogenes fMIGneg strain by PCR (PCR kit; Clontech, Palo Alto, CA) using primers outside of the 2-kb region (Keck Oligonucleotide Synthesis Facility, Yale University): 5′-LEMFAR 2 (GGACAGGCTTTCGGACT) and 3′-LEMFAR2 (GCTGATGCTAGTGCGGT). The PCR products were purified (QIAQUICK PCR purification kit) and sequenced using primers for the mutational region of LemA (~300 bp up-/downstream of the mutation; Keck Oligonucleotide Synthesis Facility, Yale University): 2KB-SEQ5 (CGCTACTTTACAACAACG).
Viable bacterial counts within spleen and liver of infected mice were determined by homogenizing the tissue in PBS containing 0.1% Triton X-100 and plating on BHI agar plates.
Mice were immunized by i.v. injection of 5 × 103 L. monocytogenes 10403s or L. monocytogenes fMIGneg, respectively, into the lateral tail vein. Spleens were harvested 6 days after immunization, and splenocytes were dissociated through a wire mash. Erythrocytes were lysed with ammonium chloride, and splenocytes were resuspended in RPMI/10% FCS (Life Technologies, Gaithersburg, MD).
PE-conjugated streptavidin tetramers of H2-M3 class Ib MHC complexed with various H2-M3 peptide ligands (fMIGWII, fMIVTLF, fMIVIL) for detecting epitope-specific T cell populations were generated, as previously described (26). For flow cytometric analysis, ~5 × 106 cells were aliquoted per staining well of a 96-well plate. After incubation at 4°C for 20 min with unconjugated streptavidin (0.5 mg/ml; Molecular Probes, Eugene, OR) and Fc-block (BD PharMingen, San Diego, CA) in FACS staining buffer (SB; PBS, pH 7.4, 0.5% BSA, and 0.02% sodium azide), cells were triple stained with FITC-conjugated anti-CD62L, PE-linked H2-M3 (fMIGWII, fMIVTLF, fMIVIL) tetramers, and allophycocyanin-conjugated anti-CD8α (clone 53-6.7; BD PharMingen), or double stained with FITC-conjugated mAbs specific for TCR-β (clone H57–597; BD PharMingen), or with 15 different TCR Vβ segments (Vβ 2, 3, 4, 5.1/5.2, 6, 7, 8.1/8.2, 8.1–3 (pan), 9, 10, 11, 12, 13, 14, 15, and 17 (all obtained from BD PharMingen)) in SB for 60 min at 4°C Subsequently, cells were washed three times in SB and then fixed in 10% paraformaldehyde/PBS (pH 7.4). Flow cytometric analysis was performed using a BD LSR flow cytometer, and data were further analyzed with CellQuest software (BD Biosciences, Mountain View, CA).
C57BL/6 mice were infected i.v. with 5 × 103 L. monocytogenes 10403s or with the fMIGneg strain. Six days after infection, splenocytes were harvested and stimulated with syngeneic fMIGWII-pulsed irradiated (3000 rad) splenocytes for 14 days. CTL clones were prepared using a limiting dilution technique. A total of 30, 300, 3,000, and 30,000 CTLs, respectively, was cultured with 2 × 106 syngeneic fMIGWII-pulsed irradiated splenocytes in RPMI/10% FCS containing IL-7 (5 ng/ml) and 5% T-Stim (rat Con A supernatant). Twelve distinct, H2-M3-restricted CTL clones were derived.
The CTL cell line and CTL clones were restimulated in vitro with naive, irradiated (3000 rad) syngeneic splenocytes (1/2 spleen/clone) that had been coated with 1 × 10−6 M fMIGWII for 1 h and washed twice in RPMI/10% FCS (27). For maintenance of the CTL cultures, the medium was additionally supplemented with recombinant mouse IL-7 (5 ng/ml; R&D Systems), rat Con A supernatant (5% v/v; Collaborative Biomedical Products, Bedford, MA), and α-methyl-mannoside (5% v/v; Calbiochem).
Synthetic peptides for the generation of CTL lines and clones, CTL assays, and tetramer generation were obtained from Research Genetics (Huntsville, AL).
L. monocytogenes 10403s and the fMIGWII epitope-deficient strain (fMIGneg) were grown overnight in 5 ml BHI medium. Ten liters of modified Welshimers’ Broth (MWB) peptide-free minimal medium (28) were inoculated with 500 µl of the bacterial overnight cultures and incubated at 37°C without agitation. After 72 h, the bacteria were pelleted and the supernatants were first passed through a 0.22-µm filter and then through a YM-10 membrane (molecular mass cutoff 5 kDa). The resulting filtrate was applied to a 20-ml preparative C18 column at a rate of 1 L/day. The columns were washed with 40 ml water (HPLC grade), and peptides were eluted with 30 ml of 10–40% acetonitrile. The eluates were lyophilized (Speedvac SPD111V; Savant/E-C Apparatus, Holbrook, NY), resuspended in 1 ml water (HPLC grade), and passed through 0.22-µm Millex filter (Millipore, Bedford, MA). These filtrates were applied to a C18 300Å reversed-phase HPLC column and fractionated on a 0–60% acetonitrile gradient. One-milliliter fractions were collected, lyophilized, and resuspended in 500 µl of PBS + 5% DMSO The fractions were tested for their targeting activity in CTL assays.
A total of 1 × 106 P815 target cells was labeled with 3.7 MBq of 51Cr sodium chromate in 200 µl RPMI/10% FCS for 1–1.5 h. Cells were washed twice with RPMI + 10% FCS and resuspended at 5 × 103 per 50 µl. To assay the HPLC fractions, 50 µl of each fraction and 50 µl of the target cells were placed in the wells of a 96-well tissue culture plate. CTLs were added to the wells in a volume of 100 µl of RPMI/10% FCS at an E:T ratio of 10:1. After 4-h incubation at 37°C, the plates were centrifuged at 1500 × g for 5 min, and 50 µl of the supernatant from each well was assayed for 51Cr release with a gamma counter (Top Count NXT; Packard Instruments, Meriden, CT). The percentage of specific lysis was determined, as previously described (29).
Cells were washed with PBS and resuspended at 5 × 107 cells/ml in PBS containing CFSE (Molecular Probes). After incubating at 37°C for 10 min in the dark, cells were immediately washed with cold RPMI/10% FCS before resuspension in PBS for i.v. injection into mice.
Analysis of in vivo cytolytic activity was conducted, as described previously (30, 31). Splenocytes from C57BL/6 mice were divided into two populations and labeled with either a high concentration (4 µM) or a low concentration (0.25 µM) of CFSE. Next, CFSEhigh cells were pulsed with 1 × 10−6 M fMIGWII peptide for 1 h at 37°C in the dark, while CFSElow cells remained nonpulsed. After washing, CFSEhigh cells were mixed with equal numbers of CFSElow cells and 2 × 107 cells were injected i.v. into individual mice. Recipient spleens were harvested 18 h later for flow cytometric analysis to measure in vivo killing, as indicated by loss of the CFSEhigh Ag-pulsed population relative to the control CFSElow population. Percentage of specific lysis was calculated according to the formula: (1 – (ratio unprimed/ratio primed) × 100), in which the ratio unprimed = percentage of CFSElow/percentage of CFSEhigh cells remaining in noninfected recipients, and ratio primed = percentage of CFSElow/percentage of CFSEhigh cells remaining in infected recipients.
LLO91–99 peptide was resuspended in PBS at a concentration of 10−8 M. The HPLC fractions from MWB bacterial cultures were prepared, as described above. Aminopeptidase M was used at an enzyme concentration of 0.2 U/ml, and carboxypeptidase Y was used at 400 µg/ml. Peptides were digested at 37°C for 1 h and then heated to 55°C for 5 min. The digest was used directly in CTL assays. The concentrations of aminopeptidase M and carboxypeptidase Y used in these experiments did not interfere with CTL function, as shown previously (32).
To test degree of cross-reactivity of MHC class Ib-restricted CD8+ T cells, we generated a mutant strain of L. monocytogenes, which lacks the major H2-M3-restricted epitope fMIGWII (L. monocytogenes fMIGneg). Because introduction of a new peptide sequence might inadvertently result in the generation of a novel epitope, which might interfere with our analysis, we mutated the fMIGWII sequence to the previously described fMIVIL sequence (12), which is a subdominant H2-M3-restricted L. monocytogenes-derived epitope (Fig. 1A). The L. monocytogenes fMIGneg strain was passaged twice through mice, and the mutation was confirmed by DNA sequencing of bacterial colonies isolated from the spleen. The mutation did not affect the growth of L. monocytogenes fMIGneg because it grew with similar kinetics as parental L. monocytogenes 10403s (Fig. 1B). Virulence was marginally affected by deletion of the fMIGWII epitope, with slightly lower bacterial loads in liver, but equivalent bacterial numbers in spleen (Fig. 1C).
Conventional MHC class Ia-restricted T cells are not primed by L. monocytogenes strains lacking their cognate epitope (25). To test whether deletion of an MHC class Ib-restricted epitope also leads to abrogation of priming of fMIGWII-specific CD8+ T cells, we infected C57BL/6 mice with a sublethal dose (5000 bacteria, i.v.) of wild-type or fMIGneg L. monocytogenes. H2-M3-restricted CD8+ T cell responses were measured by tetramer staining and flow cytometry 6 days following primary infection. Surprisingly, the fMIGWII-specific CD8+ T cell population was equivalent, as measured by staining with H2-M3/fMIGWII tetramers, in mice infected with either wild-type or fMIGneg L. monocytogenes (Fig. 2). Because no additional fMIGWII sequences can be found in the L. monocytogenes proteome (33), we speculated that other H2-M3-restricted epitopes must exist that prime CD8+ T cells in vivo and cross-react with fMIGWII. Introduction of a second fMIVIL sequence did not detectably increase the frequency of fMIVIL-specific CD8+ T cells following infection with L. monocytogenes fMIGneg (Fig. 2).
To test whether fMIGWII-specific CD8+ T cells derived from mice immunized with wild-type or fMIGneg L. monocytogenes recognize fMIGWII with similar affinities, we derived CD8+ T cell lines from mice immunized with these two strains. After two rounds of in vitro restimulation with fMIGWII peptide, both CTL lines lysed fMIGWII-coated target cells equivalently, as measured using a titration of peptide (Fig. 3). Remarkably, CD8+ T cells primed in the presence or absence of fMIGWII recognize this epitope with similar affinities. To determine whether in vivo cytolytic activity against fMIGWII is generated by immunization with L. monocytogenes fMIGneg, we measured epitope-specific cytolytic activity in mice infected with wild-type or fMIGneg L. monocytogenes using an in vivo cytotoxicity assay (30, 31). C57BL/6 splenocytes labeled with a high concentration of CFSE were pulsed with fMIGWII peptide and coinjected with similar numbers of control splenocytes labeled with a lower concentration of CFSE into mice 5 days following primary L. monocytogenes infection. In vivo CTL activity, as measured by disappearance of fMIGWII-pulsed CFSEhigh target cells 18 h posttransfer, was readily detectable in L. monocytogenes 10403s-infected mice (average decrease 82%) (Fig. 4). In mice infected with L. monocytogenes fMIGneg, in vivo cytolysis of fMIGWII-bearing target cells was also detected; however, the degree of lysis was diminished compared with mice infected with wild-type bacteria (average decrease 59%). The slight reduction in peptide-coated targets in uninfected mice resulted from a slight inequity of the transferred cells. In repeat experiments, there was no evidence for lysis of peptide-coated target cell lysis in the absence of L. monocytogenes infection. This result confirms that fMIGWII-specific cells are primed in the absence of fMIGWII, but also suggests that fMIGWII presentation enhances the magnitude of the fMIGWII-specific response.
To characterize the bacterial Ags that prime fMIGWII cross-reactive CD8+ T cells, we generated 12 CTL clones from mice infected with L. monocytogenes fMIGneg by limiting dilution. A CD8+ T cell line from L. monocytogenes 10403s-infected mice was also generated. All T cell clones and the T cell line stained with fMIGWII:H2-M3 tetramers (Fig. 5, A and B) and recognized fMIGWII with similar affinities, as determined by chromium release assays using a peptide titration (Fig. 5, C and D). Analysis of the TCR Vβ repertoire by staining with Vβ-specific mAbs demonstrated the complexity of the CTL line (Fig. 5E) and supported the clonality of the T cell clones (Fig. 5, F and G).
The fMIGWII peptide was originally identified by expression cloning (11), but can also be purified from bacterial supernatants. fMIVIL was identified by mass spectrometric sequencing of HPLC-fractioned Listeria supernatants (12). To further characterize the ligand(s) which primes fMIGWII-specific T cells in the absence of fMIGWII, we HPLC fractionated bacterial culture supernatants and assayed fractions for recognition by fMIGWII-specific CTL clones. L. monocytogenes fMIGneg was grown in peptide-free medium, and filtered culture supernatants were applied to C-18 reverse-phase column. The bound material was eluted with increasing concentrations of acetonitrile into four fractions (10, 20, 30, 40%). These bulk eluates were lyophilized and subjected to reverse-phase HPLC fractionation. The HPLC fractions were tested in CTL assays with all 12 CTL clones and demonstrated complex and distinct recognition pattern for each clone. Multiple fractions derived from the 20 and 30% bulk eluates contained fMIGWII cross-reactive ligands and induced cytolytic activity in three representative CTL clones derived from L. monocytogenes fMIGneg-infected mice (Fig. 6).
All identified L. monocytogenes-derived, H2-M3-restricted epitopes are N-formyl methonine-containing peptides (21). However, the heat-killed L. monocytogenes-associated Ag was reported to be highly protease resistant, but periodate sensitive, suggesting that it contains a carbohydrate component (15). Subsequently, it was shown that the heat-killed L. monocytogenes-associated Ag consists of a complex of immunogenic fMIGWII associated with bacterial cardiolipin A (34). To determine whether the antigenic HPLC fractions recognized by fMIGWII-specific CTL clones are peptides, as opposed to glycolipids, we selected HPLC fractions from L. monocytogenes fMIGneg supernatants that induced the highest cytolytic activity and subjected them to peptidase digestion. As expected, carboxypeptidase Y degrades both LLO91–99 (Fig. 7A) and fMIGWII (Fig. 7B), thereby destroying their ability to be detected by CTL. Similarly, the targeting activity contained in the HPLC fractions is also strongly reduced by C-terminal peptidase digestion (Fig. 7, C and D). This suggests that peptides, not glycolipids, induce cytolytic activity in the absence of fMIGWII. The N-formyl group of the amino terminus renders fMIGWII (Fig. 7B), but not LLO91–99 (Fig. 7A), resistant to aminopeptidase M digestion. Interestingly, the peptides in the HPLC fraction are similarly resistant to degradation from the N terminus, which indicates that their N termini are also protected from peptidase digestion (Fig. 7, C and D). Given the formyl-methionine specificity of H2-M3, it is likely that the HPLC fractions recognized by fMIGWII-specific CD8 T cell clones contain other N-formyl methionine-containing peptides that are secreted by L. monocytogenes.
In this work, we describe the generation of a strain of L. monocytogenes lacking the dominant, fMIGWII/H2-M3 CTL epitope. To our surprise, murine infection with this epitope-deficient strain primed a large, dominant CD8+ T cell population that stains with fMIGWII/H2-M3 tetramers. A database search proved that there are no additional N-terminal MIGWII sequences in the L. monocytogenes proteome. Only two other N-terminal peptides (fMIGYGK, fMIGPGS) are present in the L. monocytogenes proteome that share identity with the first three residues of fMIGWII. Despite this sequence similarity, neither peptide induced detectable cytolytic response by chromium release assay with the H2-M3-restricted CTL clones (data not shown). Hence, we suspect that other L. monocytogenes-derived peptides cross-react with fMIGWII and primed fMIGWII-specific T cells.
fMIGWII-specific CD8+ T cell clones generated from mice following infection with L. monocytogenes fMIGneg recognize the fMIGWII peptide in vitro with similar affinities as T cell lines generated from mice immunized with wild-type bacteria. Because it is possible that even a few rounds of in vitro restimulation with fMIGWII peptide provide a growth advantage for high affinity clones, we assessed the cytolytic activity of fMIGWII-specific T cells directly in vivo. Using this sensitive assay, we found that fMIGWII-coated target cells were killed slightly less effectively in L. monocytogenes fMIGneg-infected mice than in mice infected with wild-type L. monocytogenes. This suggests that fMIGWII is a bona fide epitope that primes a subset of CTLs with specificity for this epitope. However, our results also indicate that a substantial fraction of T cells, specific for fMIGWII, is primed by other L. monocytogenes-derived Ags.
Generation of CTL clones from L. monocytogenes fMIGneg-infected mice allowed us to distinguish several secreted peptides that could be fractionated from supernatants of wild-type L. monocytogenes cultures (results not shown). Interestingly, most of the targeting activity was also detected in L. monocytogenes fMIGneg culture supernatants, demonstrating that deletion of fMIGWII removes only one of many potential ligands. The large number of targeting fractions can be attributed to either several distinct ligands or a series of truncation variants of one or several Ags.
Glycolipid bacterial Ags may induce cytolytic activity by H2-M3-restricted CD8+ T cells (15). To prove that other peptides, rather than lipids or glycolipids, primed fMIGWII-specific CD8+ T cells during infection with L. monocytogenes fMIGneg, we subjected HPLC fractions from L. monocytogenes fMIGneg culture supernatants that induced the strongest cytolytic activity to enzymatic digestion. Upon incubation with carboxy-, but not amin-opeptidases, cytolytic activity was markedly reduced, demonstrating that peptides are responsible for the cytolytic activity and that their amino terminus is blocked.
H2-M3 and several other MHC class Ib molecules are known to bind a narrower range of peptides than conventional MHC class Ia molecules. This can be explained in part by the distinct structural features of MHC class Ib molecules. For example, the binding groove of H2-M3 and other MHC class I-like molecules such as CD1 is narrower and more hydrophobic than that of most MHC class Ia molecules (9, 35). Only a few of the 13 mitochondrial proteins, the only source for endogenous H2-M3 ligands, and some bacterially derived N-terminal sequences fulfill the requirements for this molecular pattern (21, 36). A restricted ligand specificity for endogenous peptides was also reported for the nonclassical MHC class Ib molecule Qa-2 (37), which is encoded in the Q region of the H-2 gene locus. The GPI-anchored Qa-2 molecule, implicated in innate and adaptive responses, specifically appears to be a resistance gene for murine cysticercosis (38). Qa-2 can still associate with a substantially more diverse array of peptides than other non-classical MHC class Ib molecules, but due to special constraints in the hydrophobic binding groove, the number of ligands is more limited compared with MHC class Ia molecules (39).
Interactions between H2-M3 and its peptide ligands, crucial for assembling a stable complex, have been investigated carefully. To identify those peptides that bind best to H2-M3, phage display libraries were screened. Amino acids with hydrophobic side chains preferentially occupied positions P2, P3, P4, and P6, which is consistent with the very hydrophobic pocket and the sequences of the natural ligands for H2-M3 (40). Furthermore, ~70% of the hydrophobic interactions are formed between the first four residues of the ligand and residues in the H2-M3-binding groove. Taking this into consideration, the requirements that are essential for binding to H2-M3, namely hydrophobic amino acid composition and interactions with essentially only four N-terminal residues, the number of N-terminal sequence permutations for H2-M3 ligands is limited. Of those hydrophobic peptides that do bind to H2-M3, the amino acid side chains are buried within the H2-M3-binding groove and therefore inaccessible to TCR (9, 41). Hence, similar surface characteristics that are formed by the H2-M3 molecule with different ligands might result in cross-reactive recognition by a given TCR.
Cross-reactive recognition of multiple ligands has also been reported for T cells specific for other MHC class Ib molecules. Following Salmonella typhimurium infection, Qa-1-restricted CTLs specific for an epitope derived from the bacterial GroEL molecule also cross-react with and expand in response to a peptide derived from self heat shock protein 60 (42). CD1-restricted CD8+ T cells recognize glycolipids derived from endogenous and bacterial sources. Various ubiquitous lipid structures in mammalian cells can be presented by CD1 to T cells, resulting in a considerable degree of autoreactivity in the CD1-restricted T cell pool (43). For example, CD1b presents ganglioside self Ags with complex carbohydrate structures to CD8+ T cells. Remarkably, the same TCR also recognizes carbohydrate epitopes shared by a variety of self glycosphingolipids. The GM1 ganglioside epitope that is recognized by autoreactive T cells can also be found in bacterial carbohydrates (44). Thus, CD1b-restricted T cells also appear to cross-react with several distinct Ags, in this case both self and microbial glycolipid structures (45).
MHC class Ib-restricted T cell responses remain an important field of investigation. Clearly, MHC class Ib-restricted T cells contribute to the clearance of intracellular pathogens in mice (17, 46, 47). Moreover, it was demonstrated recently that CD8+ T cells in humans recognize a M. tuberculosis-derived Ag in the context of the nonclassical, monomorphic MHC class I molecule HLA-E (48). Eliciting MHC class Ib-restricted T cells against infectious pathogens remains an exciting option for vaccine development.
We thank Rielle Giannino, An Tran, Ewa Menet, and Jessica Vega for excellent technical support.
1This work was supported by National Institutes of Health Grant RO AI49602. G.L was supported by a Human Frontier Science Program postdoctoral fellowship K.M.K. was supported by National Institutes of Health Training Grant 5T32AI07O19.
3Abbreviations used in this paper: BHI, brain-heart infusion; LLO, listeriolysin O; MWB, modified Welshimer’s broth; neg, negative; SB, staining buffer.