HLA-E is an MHC class Ib molecule that binds nonamer peptides derived from the leader sequence of MHC class 1a molecules and is the major ligand for CD94/NKG2 receptors found on NK and T cells. Using the MHC class Ia-null cell line 721.221, we find that surface HLA-E increases following heat shock at 42°C and NK cell-mediated lysis is inhibited using heat-stressed 721.221 targets. We have used mass spectrometry to identify and sequence a novel peptide from HLA-E following heat shock, ALALVRMLI, derived from the transmembrane domain of the human ATP-binding cassette protein, multidrug resistance-associated protein, MRP7. Pulsing 721.221 targets with synthetic MRP7 peptide results in strong inhibition of NK cell-mediated lysis that is reversible using anti-CD94 and anti-class I mAbs. This report is the first to identify a non-MHC leader inhibitory peptide bound to HLA-E and provides insight into the immunoregulatory role of HLA-E during cell stress.
Human histocompatibility leukocyte Ag-E (HLA-E) is an MHC class Ib molecule that is distinguished by very limited polymorphism and low levels of cell surface expression. Unlike the highly polymorphic MHC class Ia molecules that bind a wide array of peptides, HLA-E binds a limited set of highly conserved hydrophobic nonamer peptides originating from class Ia leader sequences (1, 2, 3, 4) for presentation to the CD94/NKG2 family of receptors found on the surface of NK and T cells (5, 6, 7). Multispecies genetic analysis has identified the MHC-E locus is the most conserved of all primate class I genes, especially in the peptide-binding region, underscoring the high degree of evolutionary conservation and the crucial biological function of the HLA-E-peptide complex (8, 9).
Crystallagraphic examination of HLA-E together with a peptide derived from the class Ia HLA-B8 leader sequence (VMAPRTVLL) revealed that HLA-E possesses a groove that necessitates interaction with amino acid residues along the entire length of the peptide, implying that only a limited set of highly homologous hydrophobic peptides are effectively accommodated by HLA-E (10). This high degree of peptide specificity is optimized for binding class Ia leader sequences, and in the event of insufficient leader peptide, as may occur following viral infection or during tumorogenesis, surface HLA-E is diminished and renders a cell susceptible to NK cell-mediated lysis due to insufficient ligand density to engage the CD94/NKG2A receptor. The HLA-E-CD94/NKG2A interaction is therefore a superb and critical mechanism by which the entire MHC class Ia-processing pathway is monitored for deficiencies using a single ligand-receptor pair, and exemplifies the tenets set forth in the “missing self” hypothesis of Ljunggren and Karre (11).
Qa-1b is the murine MHC class Ib functional homologue of HLA-E and several reports have found that both molecules bind a nearly identical repertoire of leader-derived peptides (1, 2, 3, 12, 13, 14, 15, 16). Early studies using L cells transfected with the Qa-1b-encoding gene T23 showed that the level of surface Qa-1b is selectively increased following elevated temperature at 42°C (17). Given our current understanding of the functional relatedness of Qa-1 and HLA-E, it seemed plausible to consider that HLA-E may also be modulated following stress and that surface HLA-E would also be increased in response to heat shock. In the present study, we have used the MHC class Ia-null cell line 721.221 (.221) that expresses only endogenous HLA-E due to the absence of class Ia leader peptide (18). We find that surface HLA-E is increased on .221 cells following heat shock and that these stressed cells are less susceptible to CD94/NKG2A-regulated NK cell-mediated lysis. Isolation and characterization of peptides eluted from HLA-E expressed in heat stressed .221 cells identified a single peptide, ALALVRMLI, derived from the human ATP-binding cassette (ABC)5 transporter protein, multidrug resistance-associated protein 7 (MRP7), also termed ABCC10. We show that this peptide binds to HLA-E by its ability to inhibit NK cell-mediated lysis in a CD94 and class I-dependent fashion. This report is the first to identify a self non-leader inhibitory HLA-E-binding peptide and provides a new and expanded avenue of investigation to explore the previously unknown connection between HLA-E, cell stress, and the ATP transporter protein, MRP7.
Materials and Methods
The human B cell line .221 was obtained from American Type Culture Collection (ATCC) (CRL-1855) and grown in RPMI 1640 with 10% FCS (Invitrogen Life Technologies). The NKL cell line (19) was kindly provided by M. Robertson (University of Indiana, Indianapolis, IN) and was grown as above with the addition of 30 U/ml recombinant human IL-2 (Endogen).
For cell surface staining, W6/32 was purified from mouse ascites (provided by R. F. Siliciano, Johns Hopkins University, Baltimore, MD) using a Protein-A/G affinity column (Pierce) or anti-HLA-E mAb, 3D12 (D. Geraghty, University of Washington, Seattle, WA), was added followed by FITC-conjugated sheep anti-mouse F(ab′)2 (Accurate Biochemical). Cells were resuspended in buffer containing 2% propidium iodide for live cell gating using a FACSCalibur (BD Biosciences) with analysis using CellQuest software.
High purity synthetic peptides (Macromolecular Resources); HLA-A2 leader, (VMAPRTLVL); HLA-G leader (VMAPRTLFL); murine leader Qdm, (AMAPRTLLL); mammalian heat shock protein 60 (hsp60; GMKFDRGYI); MRP7 (ALALVRMLI); and HLA-B27-binding peptide (RRYQKSTEL) were rehydrated at concentrations of 1 and 5 mg/ml using sterile and endotoxin-free water (Invitrogen Life Technologies).
NKL cells were tested for the ability to lyse heat-shocked or peptide-pulsed .221 targets using a standard 4-h 51Cr release assay. For heat shock, .221 targets were placed at 42°C for 10, 20, 40, or 60 min, 51Cr added and cells were allowed to recover for 1.5 h at 37°C. For peptide pulsing, 51Cr-labeled targets were incubated in the presence of synthetic peptides for 3 h at 25°C before adding NKL effectors. For HLA-E blocking, W6/32 or control polyclonal mouse IgG2a (BD Pharmingen) were added to target cells. For anti-CD94 blocking, DX22 (provided by L. Lanier, University of California, San Francisco, CA) or control mAb was added to NKL effectors.
Immunoprecipitation of HLA-E and acid extraction of peptides
HLA-E-peptide complexes were isolated and peptides extracted using a method previously described (13, 20, 21). Briefly, 2 × 109 heat shocked .221 cells were lysed and nuclei were separated by centrifugation at 10,000 × g for 15 min at 4°C. The postnuclear supernatant was precleared using mouse IgG-sepharose beads and HLA-E/peptide complexes were isolated by incubating the extract with W6/32-coupled sepharose beads for 2 h at 4°C. W6/32 beads were washed sequentially in PBS, followed by water, and peptides were acid extracted by adding 1% trifluoroacetic acid (TFA) (Pierce). Extracts were size fractionated using Amicon Centriprep-10 membrane filters, then frozen and lyophilized for HPLC purification and mass spectral analysis.
HPLC and mass spectrometry
Lyophilized samples were resuspended in 0.5 ml 0.1% TFA/10% acetonitrile (J. T. Baker) in water and fractionated by reverse-phase HPLC (Waters) using a C18 column (Vydac). Peptides were separated using a gradient of solution A, 0.1% TFA/water, and solution B, 0.1% TFA/acetonitrile at a flow rate of 1 ml/min, with 1-ml fractions collected. Mass spectral analysis was performed using a Kratos Axima-CFR mass spectrometer with 0.3 μl of saturated ammonium sulfate and 0.3 μl of matrix (saturated α-cyano-4-hydroxycinnamic acid in 50% ethanol). Trypsin and carboxypeptidase Y (Calbiochem) digestions were performed on HPLC fraction 57 using enzyme concentrations of 0.05 μg/ml. All spectra were obtained in positive ion mode. Protein database searches were conducted using Protein Prospector, MS Pattern program, University of California (San Francisco, CA; 〈www.proteinprospector.ucsf.edu〉).
Results and Discussion
Surface HLA-E is increased on .221 cells following heat shock at 42°C
To test the possibility that surface HLA-E increases following heat stress, .221 cells were placed at 42°C for varying lengths of time, allowed to recover, and assayed via flow cytometry. Results of this experiment revealed that although unstressed .221 cells express low levels of class I, the shortest duration of heat stress, 10 min, lead to the most substantial increase in a class I molecule, while prolonged heat stress, e.g., 20, 40, or 60 min, caused a decrease in the level of surface class I protein (Fig. 1,A). An analysis of HLA-DR expression revealed no such heat shock-induced change in surface expression (data not shown). To determine whether this increased class I molecule was HLA-E, experiments were done using equal concentrations of the pan-class I mAb W6/32 and the HLA-E-specific mAb 3D12, and the results showed identical fluorescence intensity on .221 cells before and after heat shock (Fig. 1,A at 10′, and Fig. 1 B). This finding demonstrates that surface HLA-E on .221 cells is increased in response to elevated temperature, and indicates the generation of an HLA-E-epitope with kinetics that are consistent with the rapid protein translocation and Ag processing of a stress response protein.
Inhibition of NK cell-mediated lysis using heat-shocked .221 targets
Increased surface HLA-E on .221 cells following heat stress led us to examine whether these cells would be less susceptible to NK cell-mediated lysis. Cytotoxicity assays were conducted using the CD94/NKG2A+ NKL cell line and .221 targets ± heat shock and the results of this experiment showed a 48% decrease in NK cell-specific lysis at three different E:T ratios (Fig. 1 C). The inhibition was reversed using anti-CD94 Ab indicating the involvement of the inhibitory receptor CD94/NKG2A and its ligand HLA-E. This result implies that only a modest increase in the surface density of HLA-E is sufficient to effectively engage the CD94/NKG2A receptor and lessen susceptibility to NK cell cytotoxicity. These findings also infer that the de novo peptide bound to surface HLA-E following heat stress possesses characteristics consistent with an inhibitory class I leader-derived sequence.
Identification of novel inhibitory peptide bound to HLA-E
The results from these initial experiments suggested that a non-class I leader inhibitory peptide was bound to HLA-E following heat shock. To identify the peptide or peptides, HLA-E complexes were isolated from heat-stressed .221 cells, followed by acid extraction and separation of peptides using reverse-phase HPLC. Fractions were analyzed for peptides using MALDI mass spectrometry together with mass gating. A single peptide-containing fraction was identified, no. 57, with two peptide mass species of 951.2 and 999.1 m/z (Fig. 2,A). Postsource decay (PSD) of the m/z 951.2 peptide revealed a large c8 ion, corresponding to a loss of 113.2 from the C terminus, indicating a leucine or isoleucine in position 9 (Fig. 2,B). Further analysis of PSD spectra suggested a y-ion series of peaks corresponding to the sequence; Ala in position 1, Leu or Ile in position 2, and Ala in position 3. The loss of a single amino acid from either end suggested the presence of a basic residue, Lys or Arg, in the P4, P5, or P6 position of the peptide (Fig. 2,B). To confirm the presence of Lys or Arg, enzymatic digestion of fraction 57 peptide was conducted using both trypsin and carboxypeptidase Y, and the mass spectra revealed peaks at 642.57 and 642.52 m/z, respectively. Several y-series ions in the PSD mass spectra of these truncated peptides enabled us to establish the sequence of the first six amino acids as ALALVR. (Fig. 2 C).
Further biochemical analyses were performed and confirmed that the mass difference of 48 between the two peptides was due to a decomposed carboxymethylated methionine residue (22), with the peptide species at 951.2 being a degradation product of the intact methionine-containing parent peptide at m/z 999.1 (Fig. 2 A). Human protein database searches were conducted and one peptide matched the experimental data well, ALALVRMLI, with a [MH]+ (average) mass of 999.6, derived from aa 496–504 of the first transmembrane domain of the human ABC protein, MRP7, encoded on the short arm of chromosome 6, centromeric to HLA-DR (23, 24).
Peptide 496–504 from MRP7 has a number of structural features that are compatible with binding to the HLA-E groove. The presence of an isoleucine in the P9 primary anchor position is consistent with the requirements for an HLA-E-binding peptide, as confirmed by random peptide libraries showing that leucine is the predominant amino acid in this position (〈www.biomedcentral.com/1471-2172/2/5〉; Refs.25 and 26). The alanine in P1 and P3 of the MRP7 peptide are in excellent agreement with amino acids that are accommodated by HLA-E as evidenced by the fact that the P1 and P3 residues of all leader-derived inhibitory peptides are also alanine. Although crystallographic studies indicate that the B pocket of HLA-E may be optimized for a P2 methionine, (10) the data presented here together with the observations from other groups using a soluble refolding assay, NK inhibition, and random peptide libraries clearly substantiate that leucine in P2 is also well-suited for both HLA-E and Qa-1-binding peptides that inhibit NK cell lysis (15, 16, 26).
The presence of a positively charged amino acid within the middle region of an HLA-E-binding peptide is a key component of the sequence selectivity imposed by the HLA-E structure (10). Studies using HLA-E tetramers refolded using peptides randomized at P5 show a strong preference for Arg and Lys in this position (26). The findings presented here imply that the positive charge of a P6 arginine in the MRP7-derived peptide is also capable of providing an electrostatic environment allowing effective binding to HLA-E and engagement of CD94/NKG2A.
Inhibition of NK cell-mediated lysis using synthetic MRP7-derived peptide
To verify that the peptide sequence derived from MRP7 protein is able to bind to HLA-E and inhibit NK cell-mediated lysis, .221 cells were pulsed with increasing amounts of synthetic peptides; MRP7, HLA-A2 leader, Qdm, hsp60, and a control HLA-B27-binding sequence, and tested in a 51Cr-release assay using the NKL cell line. The results of this experiment demonstrated a remarkable ability of the MRP7-derived peptide to inhibit NK cell-mediated lysis with near parity to leader-derived peptides at equimolar concentrations (Fig. 3,A). To validate that the measured inhibition is due to the interaction of CD94/NKG2A with HLA-E, Ab-blocking experiments were performed and showed that blocking of either CD94 or HLA-E in the presence of peptide prevents inhibition (Fig. 3 B), confirming that MRP7-mediated inhibition is the result of the direct interaction of HLA-E with CD94/NKG2A.
The identification of a unique inhibitory peptide extracted from HLA-E following heat stress expands our understanding of the repertoire of peptides capable of binding to HLA-E for recognition by CD94/NKG2 molecules. Although the connection between HLA-E and the MRP7 protein is not immediately evident, one intriguing possibility is that this recognition may serve to protect cells during periods of nonlethal stress. Under such conditions, class Ia peptides may be limiting, and a mechanism that sustains steady-state levels of surface HLA-E ensures that CD94/NKG2-bearing cells are not cytolytic. Thus, the generation of an HLA-E epitope derived from MRP7 protein provides insight into the molecular details of the mechanisms that act to maintain tissue integrity and function during periods of transient stress.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants R21ES10309 and RO1AI047286 (to M.J.S.) and Grant 1R01GM/RR64402-01 (to R.J.C.).
Abbreviations used in this paper: ABC, ATP-binding cassette; MRP7, multidrug resistance-associated protein 7; TFA, trifluoroacetic acid; hsp60, heat shock protein 60; PSD, postsource decay.