|Home | About | Journals | Submit | Contact Us | Français|
Peptides that are presented by MHC class I (MHC-I) are processed from two potential sources, as follows: newly synthesized endogenous proteins for direct presentation on the surface of most nucleated cells and exogenous proteins for cross-presentation typically by professional APCs. In this study, we present data that implicate the nonclassical HLA-F and open conformers of MHC-I expressed on activated cells in a pathway for the presentation of exogenous proteins by MHC-I. This pathway is distinguished from the conventional endogenous pathway by its independence from TAP and tapasin and its sensitivity to inhibitors of lysosomal enzymes, and further distinguished by its dependence on MHC-I allotype-specific epitope recognition for Ag uptake. Thus, our data from in vitro experiments collectively support a previously unrecognized model of Ag cross-presentation mediated by HLA-F and MHC-I open conformers on activated lymphocytes and monocytes, which may significantly contribute to the regulation of immune system functions and the immune defense.
It is well established that exogenous proteins are processed by professional APCs in endosomes for MHC class II presentation to CD4+ cells, whereas endogenous proteins are processed in the cytoplasm for MHC class I presentation to CD8+ T cells (1–3). Both intracellular transport and cell surface expression of MHC-I proteins are dependent on the availability of peptides within the endoplasmic reticulum, actively transported there by the TAP complex (4). MHC-I can also present exogenous Ag by a process termed cross-presentation, which, as with MHC-II presentation, has been shown to operate primarily in professional APCs (5). Many of the details of the specific pathways through which extracellular proteins are internalized and associate with MHC-I remain to be established (6). Attention has largely been focused on how peptides traffic to sites where MHC-I molecules reside during cross-presentation. Several models have been proposed for transport of peptides from extracellular derived compartments to the cytoplasm, including the endoplasmic reticulum (ER)–phagosome fusion model (7) and the physical disruption of lysosomal membranes model (8). Other studies have focused on pathways such as the ER-associated degradation pathway (9), protein channels, and autophagy (10). The uptake and cross-presentation of extracellular Ag may represent an important means to stimulate T cell responses to Ag otherwise not available through endogenous MHC presentation, as often occurs in tumor transformation or viral infection.
Whereas the presentation of antigenic epitopes by classical MHC-I molecules is a critical component of the adaptive immune response, the human MHC also contains three nonclassical class I genes (HLA-E, F, and G) with divergent immune function. The HLA-E and G proteins evidently do not, at least as a primary function, participate directly in Ag presentation, but rather interact with immunoregulatory receptors expressed on lymphocytes. HLA-G, expressed on placental trophoblast cells, may function as an important tolerogenic immunoregulator during pregnancy (11, 12). HLA-E complex, expressed ubiquitously in coordination with classical MHC class I, interacts with the lectin heteroduplexes CD94 combined with different NKG2 subunits to inhibit and activate NK cells and subsets of T cells (13, 14).
HLA-F is expressed as a protein independent of bound peptide (15, 16), and surface expression is upregulated in monocytes and most lymphocyte subsets upon activation, including NK cells, B cells, and all T cell subsets, except regulatory T cells (17). MHC-I is also expressed on proliferating lymphoid cells as a stable pool of MHC-I H chain devoid of peptide and/or β2-microglobulin (β2m) (18). These so-called open conformers (OC) have been implicated in a number of interactions with other receptors on the cell surface both in trans and in cis, including the formation of homodimers (19). Most MHC-I proteins bind HLA-F as OC without peptide, but not as peptide-bound complex. Thus, the physical interaction between HLA-F and MHC-I H chain may a play role in the function of MHC-I H chain OC.
In this study, we extended our observations of HLA-F and MHC-I interactions to investigate their potential to act as mediators in the presentation of exogenous Ag by MHC class I. Viral, tumor, and minor histocompatibility Ags were tested for their capacity to stimulate class I–restricted cellular responses when presented by HLA-F–positive, MHC-I OC–positive cells. The involvement of MHC-I OC and HLA-F interaction in this pathway was tested through direct binding of Ag and specific interference with their surface expression and transport. Mutant cell lines that lack a functional endogenous pathway and drugs that interfere with intracellular trafficking and protein degradation were used to distinguish the pathway from endogenous MHC-I presentation. These data collectively support a model for a general mode of exogenous MHC-I Ag uptake and presentation by activated lymphocytes and monocytes that differ in significant detail from the presentation of endogenous Ag, suggesting a role for cooperation between HLA-F and MHC-I OC in this pathway.
NKL and KMA were all obtained from American Type Culture Collection (Manassas, VA) and cultured according to the product information sheet provided. B-LCL cell lines were previously collected and analyzed by the International Histocompatibility Workshops and Conference and obtained directly from the International Histocompatibility Working Group in Seattle (20). LCL 721.221 was obtained from the American Type Culture Collection and maintained in RPMI 1640 medium supplemented with 10% v/v FCS, 2 mM l-glutamine, and 1 mM sodium pyruvate. The related B-LCL 721 derivative TAP and tapasin mutant cells, 0.134, 0.134C2 (TAP restored), 0.174, and T2 (hemizygous chr 6 derived from 0.174), were the gift of T. Spies (Fred Hutchinson Cancer Research Center, Seattle, WA). Other B-LCL cell lines were grown in RPMI 1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 U/ml streptomycin. Genotyping for the presence of the HA-1H and R alleles was carried out, as described (21). The CTL clones specific for HIV, CMV, and minor H Ags were derived in prior studies (22–24), and CTL clones specific for tumor Ag gp100 were also isolated previously (25). CTL clones were expanded using the Rapid Expansion Protocol (26). Typically, 1–2 × 105 freshly thawed CTL clone cells were added to 25 × 106 irradiated PBMC and 5 × 106 irradiated TM LCL in 25 ml CTL medium (RPMI 1640–HEPES +10% human AB serum and 55 μM 2-ME) with 30 ng/ml anti-CD3 Ab. After 24 h, 50 U/ml rIL-2 (R&D Systems, Minneapolis, MN) was added, and, at 2 d, the cells were washed and resuspended in CTL medium with 50 U/ml rIL-2. At days 7 and 8, and every 2–3 d thereafter, the cells were fed by replacing 50% of media with new CTL media plus 50 U/ml rIL-2. Cells were typically assayed between days 10 and 14.
mAbs 3D11, 4A11, 4B4, and 6A4 specific for HLA-F were generated in our laboratory, as previously described (15, 17, 27). HCA2 was a gift from Thomas Spies (Fred Hutchinson Cancer Research Center, Seattle, WA). Other mAb were purchased from suppliers, including MA2.1 (American Type Culture Collection), Rab 5 (Abcam, Cambridge, MA), Rab 7 (Cell Signal, Danvers, MA), and CMV pp65 Tegument protein (UL83) mAb (Fitzgerald Industries International, Acton, MA). Proteins were obtained from the following sources: recombinant HIV-p24 (24 kDa; Prospecs, Ness-ziona, Israel) and recombinant pp65 protein (65 kDa; Miltenyi Biotec, Auburn, CA). Unbiotinylated and biotinylated 50-aa proteins were synthesized (Biosynthesis, Lewisville, TX), as follows: gp100#1, YVWKTWGQYWQVLGGPVSGLSIGTGRAMLGTHTMEVTVYH RRGSRSYVPL; gp100#2, Bio-SSGTLISRALVVTHTYLEPGPVTAQVVLQAAIPLTSCGSSPVP GTTDGHR; HA-1H, Bio-DISHLLADVARFAEGLEKLKECVLHDDLLEARRPRAHECLGEALR VMHQII; HA-1R, DISHLLADVARFAEGLEKLKECVLRDDLLEARRPRAHECLGEALRVMHQII; P17, Bio-ASVLSGGKLDRWEKIRLRPGGKKKYKLKHIVWASRELERFAVNPGLLETS; P24, Bio-IGWMTNNPPIPVGEIYKRWIILGLNKIVRMYSPTSILDIRQGPKEPFRDY. Peptides were synthesized (Anaspec, Fremont, CA), as follows: pp65, HLA-A*0201, NLVPMVATV; gp100#1, HLA-A*0201, KTWGQYWQV; gp100#2(g280), HLA-A*0201, YLEPGPVTA-K; HA-1H, HLA-A*0201, VLHDDLLEA; HIV-gag (p17), HLA-A*0301, RLRPGGKKK; HIV-gag (p24), HLA-B*2705, KRWIILGLNK.
Liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis of peptides and proteins was carried out at the Proteomics Shared Resource at the Fred Hutchinson Cancer Research Center, essentially as described previously (15).
B-LCL cells (BM9) were incubated with 100 μg/ml biotinylated gp100#2 or HA-1H in RPMI 1640 plus 2% BSA for 4 h at 37°C. After incubation, cells were washed and surface stained with pan class I mAb W6/32 labeled with Alexa-488 (BioLegend, San Diego, CA). The cells were incubated for another 60 min with anti–HLA-F mAb 3D11 or MHC-I H chain mAb HCA2. After incubation, cells were washed and surface stained with W6/32 labeled with Alexa-488 (BioLegend). The cells were then washed, and intracellular staining was performed with anti–IgG1-Dylight 649 (Jackson ImmunoResearch Laboratories, West Grove, PA) for detection of anti–HLA-F and streptavidin Alexa-594 (Invitrogen, Eugene, OR) for detection of intracellular biotinylated Ags, as per instructions (Intraprep; Beckman Coulter, Brea, CA). After intracellular staining, the cells were fixed with 1% paraformaldehyde, washed, resuspended in Prolong Gold antifade reagent with DAPI (Invitrogen), and mounted onto optical slides. The cells were imaged using a Deltavision RT Wide-Field Deconvolution microscope (Applied Precision, Issaquah, WA), and the images were analyzed with ImageJ.
B-LCL cell lines were prelabeled with 50 μCi 51Cr for 1 h at 37°C and washed and incubated in the absence or presence of 10 μg/ml brefeldin A (BfA; BioLegend), 100 μM N-ethylmaleimide (Sigma-Aldrich, St. Louis, MO), 200 μM chloroquine (Sigma-Aldrich), or 200 μM leupeptin (Sigma-Aldrich) for 1 h prior to addition of Ag. A total of 5 × 105 51Cr-labeled B-LCL cells suspended in 500 μl RPMI 1640 plus 2% BSA was incubated with peptide, vaccinia virus-pp65, recombinant protein (p24, pp65), or synthesized recombinant 50-aa proteins derived from their respective parent sequences for 4 h at 37°C. These experiments resulted in establishing a protocol for proteins that required denaturing the proteins prior to the sensitization of cells, similar to that previously described (28). Following incubation with Ag, cells were washed three times and resuspended in RPMI 1640 plus 10% FBS and plated out at 5 × 103 cells/well with effectors at the indicated ratios. At 4 h, 30 μl supernatant was collected and applied to lumaplates, dried, and counted using a TopCount scintillation counter (Perkin Elmer, San Jose, CA).
Temperature shift experiments were performed in an essentially similar protocol with cells incubated at either 4°C or 37°C for 1 h before labeling with control peptide, p24, or p17 at the indicated concentration for an additional 2 h at 4°C or 37°C. For the centricon experiments, Ags were prepared by adding a 2× concentration of Ag to cells at 106 per ml suspended in AIM-V serum-free medium or RPMI 1640 plus 2% FBS (no differences were observed) for 2 h at 37°C. After incubation, cells were spun down at 1300 rpm, and the supernatant was collected and spun through a 3000 m.w. cutoff centricon device. The flowthrough was applied at 1:2 starting dilution to the 51Cr-labeled targets, in parallel to 51Cr-labeled targets pulsed with control peptide or protein at the indicated concentration through serial dilutions. After incubation, the cells were treated and analyzed, as above. In peptide competition experiments, B-LCL or NKL cells were prelabeled with 51Cr for 1 h at 37°C prior to pulsing with Ag. After labeling, the cells were incubated for 1 h with 100 μM A*02 peptide (C1R; Anaspec, Fremont, CA) at 37°C before the addition of positive control peptide or recombinant proteins for an additional 4 h.
Intracellular cytokine staining was performed by directly adding recombinant pp65 to CTL clone 1C7-31 or the HA-1H 50-aa protein to CTL clone GAS#9, each resuspended in RPMI 1640 plus 10% human AB serum. The cells were incubated for 1 h at 37°C, followed by the addition of 10 μg/ml BfA and further incubation for 4 h at 37°C. After incubation, cells were fixed and stained for intracellular IFN-G as per instructions (Intraprep; Beckman Coulter).
Lentiviral vector (29, 30) was used to construct β2m and HLA-F–specific knockdowns using synthetic oligonucleotides cloned into the XbaI/EcoRV cloning sites. Short hairpin RNA (shRNA) construct targeting β2m had targeting sequence 5′-CAGCAGAGAATGGAAAGTCAA-3′ with forward oligonucleotide 5′-CTAGACAGCAGAGAATGGAAAGTCAACTCGAGTTGACTTTCCATTCTCTGCTG TTTTTTGAT-3′ and reverse oligonucleotide 5′-ATCAAAAAACAGCAGAGAATGGAAAGTCAACTCGAGTTGACTTTCCATTCTCTGCTGT-3′; the construct targeting HLA-F had targeting sequence 5′-TGGTCGCTGCTGTGATGTGGAGGAAGAAG-3′, with forward oligonucleotide 5′-CTAGATGGTCGCTGCTGTGATGTGGAGGAAGAAG TCAAGAGCTTCTTCCTCCACATCACAGCAGCGACCATTTTTTGAT-3′ and reverse oligonucleotide 5′-ATCAAAAAATGGTCGCTGCTGTGATGTGGAGGAAGAAGCTCTTGACTTCTTCCTCCACATCACAGCAGCGACCAT-3′. Italicized font represents the loop. A termination sequence (TTTTTT) is located immediately downstream of the reverse complementary sequence to terminate the transcription by RNA pol III. Construct was cotransfected with gag-pol transfer and envelope helper plasmids into 293T cells by calcium phosphate. Viral particles were harvested at 16 and 40 h, filtered by 0.45 μm pore size, and concentrated 100-fold by PEG-it (SBI, Mountain View, CA). All transductions were performed over consecutive 4 d. At day 1, 60 × 104 B-LCL cells were seeded in 24-well plates in 1.5 ml RPMI 1640 with 10% PBS; 20 μl 100-fold concentrated virus supernatant was added in the presence of 8 μg/ml protamine sulfate; and the plate was centrifuged at 2500 rpm at room temperature for 1 h. A total of 1 ml culture medium was replaced with 1 ml fresh medium containing 20 μl concentrated virus and protamine sulfate and centrifuged once per day for the next 3 d. GFP-positive cells were analyzed and sorted by flow cytometry.
T2 cells were washed twice with RPMI 1640 medium plus 10% BSA and resuspended in RPMI 1640–10% BSA at a concentration of 50 × 104 cells/ml, and 200 μM each of leupeptin and chloroquine were added. After 1 h, 30 μg/ml human CMV (HCMV) protein pp65 or 2 μM pp65 peptide (NLVPMVATV) in DMSO was added. Urea buffer containing 50 mM NaH2PO4, 300 mM NaCl, and 4 M urea or DMSO was added to control reactions. After 16 h, cells were incubated with 10 μg/ml BfA for 4 h and stained with MHC-I mAbs by FACS.
KOSE cells were suspended in hypo-osmotic buffer containing 10 mM HEPES (pH 7.5), 15 mM KCl, 1.5 mM MgAc, and 1 mM DTT at a concentration of 35 × 106 cells/ml. Cells were homogenized with 15 strokes in a Pyrex glass homogenizer, and 1/10 (v/v) of hyperosmotic buffer containing 10 mM HEPES (pH 7.5), 700 mM KCl, 40 mM MgAc, and 1 mM DTT was added to the final homogenate. After centrifugation at 800 × g for 10 min at 4°C, the nuclear pellet was washed with 10 mM HEPES (pH 7.5), 85 mM KCl, 5.5 mM MgAc, and 1 mM DTT and collected by centrifugation. Combined postnuclear supernatants were centrifuged at 145,000 × g at 4°C for 25 min, and the resulting crude membrane pellet was resuspended in 1 ml homogenizer buffer containing 10 mM Tris (pH 8.0), 0.25 M sucrose, and 1 mM EDTA.
A 50% (w/v) sucrose stock solution in 10 mM Tris (pH 7.5), 1 mM EDTA was used to prepare a 50–10% sucrose gradient using 11 dilution steps of 1 ml each. The gradient was equilibrated overnight at 4°C. The membrane suspension prepared above was layered on top of the gradient and centrifuged overnight at 100,000 × g at 4°C in a Beckman ultracentrifuge equipped with a SW-41Ti rotor. The 1-ml fractions were collected and analyzed for marker enzymes by enzyme activity assays. Marker proteins (Rab 5, Rab 7), pp65 protein, HLA-F, and class H1 chain were analyzed by Western blotting. The activity of lysosomal enzyme β-hexaminidase was assayed fluorometrically using 4-methylumbelliferyl substrates (Apollo Scientific), as described (31). Plasma membrane 5′-nucleotidase activity was measured, as described (32). For Western blot analysis, a 60 μl aliquot of each fraction was reduced by 5× sample buffer and separated by 12% Tris-glycine gel (Invitrogen, Carlsbad, CA). Specific protein was detected by mAb, as indicated, and visualized with a chemiluminescence system (Roche Applied Science, Penzberg, Germany).
T2- or T2/B35-transfected cells were conditioned in RPMI 1640 supplemented with 5% protease-free BSA for 30 min at 37°C with 5% CO2. Cells were incubated with 50 μg/ml biotinylated 50-aa proteins for 2 h, washed with ice-cold Dulbecco’s PBS, and lysed in Dulbecco’s PBS containing 1% Nonidet P40, as described previously (15). Proteins of interested were pulled down by streptavidin and eluted by 8 M guanidine HCl (pH 1.5). Proteins were separated on 10% Bis-Tris gels (Invitrogen, Grand Island, NY) and analyzed by Western blot using indicated Abs.
Our previous work demonstrated that HLA-F was downmodulated on the surface of B-LCLs upon addition of MHC-I mAb specific for H chain or OC (15). Because H chain–specific mAb bind within the cleft region of MHC-I, we hypothesized that such mAb binding either structurally mimics peptide binding and thus alters the structure of MHC-I H chain to resemble complex resulting in dissociation and downmodulation of HLA-F, or, alternatively, causes cross-linking of HLA-F/MHC-I heterodimers triggering internalization. Because the postulated structure of MHC-I when associated with HLA-F is open and thus peptide receptive, we considered the possibility that long polypeptides (>30 aa) with internal MHC-I epitopes may also bind to open class-I MHC, and that the binding of long polypeptides containing multiple MHC epitopes would produce cross-linking in the same way as mAb and thus downmodulation of HLA-F.
To test this hypothesis, HLA-F levels were compared before and after the addition of denatured viral proteins HIV-1 p24 or HCMV pp65. Marked decreases in HLA-F levels were indeed observed in both cases (Fig. 1A). The observed downmodulation suggested that HLA-F was being internalized in response to the addition of exogenous protein potentially interacting with MHC-I, HLA-F, or a related complex or structure on the cell surface. Fluorescence microscopy was then used to visualize the fate of HLA-F, MHC-I, and exogenous protein directly. A synthesized 50-aa polypeptide (biotinylated gp100#2) derived from melanoma Ag gp100 (33) was incubated with target HLA-F–positive B-LCL cell lines and visualized by costaining with MHC-I–specific mAb HC10 and HLA-F–specific mAb 3D11. The biotinylated gp100 protein colocalized with HLA-F or MHC-I both on the cell surface and within the cell, supporting the idea that the molecules are internalized together (Fig. 1B). The overlapping intracellular signals may also indicate that both molecules remain colocalized during the initial stages of processing.
Given the colocalization of HLA-F, open MHC, and Ag, we next attempted to trace the passage of Ag from the extracellular space, through internalization, and back to the surface as MHC-I peptide complex. First, we examined complex formation on the surface of T2 cells in the presence or absence of added Ag using conformation-specific mAbs mA2.1 (HLA-A*02 specific) and W6/32 (pan MHC complex specific). These experiments were designed to evaluate an increase in MHC complex formation after addition of Ag as a means of gauging cross-presentation of target Ag and are similar in design to a previously described assay that measured relative MHC-I peptide affinities (34). This assay uses the TAP-deficient T2 cell line to maximize the ability to detect the formation of new complex above background when using the conformation-specific mAbs W6/32 and mA2.1. To optimize the concentrations of control nonamer peptide and target Ag to be used before and after drug treatment, titrations of peptide and protein were carried out in the absence of drugs (Supplemental Fig. 1). The midpoint or half-maximal concentration of Ag for both protein and peptide was used in triplicate experiments, and the mean fluorescence index (MFI) of mAb binding before and after addition of peptide and protein was compared in the presence of two lysosomal inhibitors and BfA. Although the change in MFI for control nonamer peptide was not significantly affected by any of the drugs, the increase in MFI for both mA2.1 and W6/32 observed after addition of exogenous pp65 protein was virtually eliminated by all three inhibitors tested (Fig. 1C).
Next, to observe the pathway and processing of Ag more directly, we fractionated cells after the addition of exogenous pp65 Ag and analyzed the proteins by Western blot with pp65-specific mAb. Enzyme activity for each fraction was measured to identify the subcellular compartments contained within each fraction. The enzyme markers present in fractions 1–4 and 9–10 were consistent with the presence of early endosomes and lysosomes, respectively. These fractions also contained pp65 protein, which migrated at a reduced m.w. in all fractions. The species of pp65 in the gradient fractions containing lysosomes had further reduced m.w. relative to the fractions containing early endosomes, suggesting additional processing of protein (Fig. 1D). These data, taken together with the drug sensitivity data, are consistent with Ag entering cells, passing through early endosomes, and proceeding through lysosomes before the derivative peptide is generated. Derivative peptide then presumably combines with MHC-I, and the resultant complex is expressed on the surface (as detected by W6/32 and mA2.1).
Finally, to confirm and clarify the biochemical evidence for localization, we performed additional fluorescence microscopy using costaining with endosomal markers (Fig. 2). Four markers were used, as follows: Rab5, a marker for early endosomes regulating the fusion between endocytic vesicles and early endosomes (35); Rab7, a marker found on a major organelle in the endolysosomal pathway (36); LAMP1, the lysosomal-associated membrane protein 1 (37); and EEA1, the early endosome Ag 1 (38). Initial stains of gp100 with EEA1 and LAMP1 showed isolated surface staining for gp100, but after internalization showed largely overlapping staining with both of the endosomal markers. Furthermore, costains of Ag gp100 with 3D11 and either Rab5 or Rab7 showed a clear overlap in signals among the sets of trios, indicating that as HLA-F and gp100 were internalized, they transited from the endocytic vesicles to the lysosomal pathway. Similarly, MHC-I and gp100 could be observed to internalize and colocalize with both Rab5 and LAMP1, indicating their coordinate transit from the endosome to the lysosome pathway.
Given coincident internalization of exogenous Ag with MHC-I and HLA-F, we examined whether direct physical interaction occurs between exogenous Ag and MHC-I in an epitope-specific manner. We used a previously established experimental system that identified HLA-A*0201–specific high-affinity mutant peptides derived from tumor Ag gp100 (39). We chose three peptides differing from one another by successive single amino acid changes from that study, that bound with low (ELE), medium (the naturally occurring sequence YLE), or high (YLF) affinities to HLA-A*0201. These peptides were used to measure relative increases in MHC-I complex formation using the T2 system described above. The ability of the peptides to increase levels of MHC-I complex was in direct relation to their binding affinity for HLA-A*0201 as was a corresponding reduction in the levels of MHC-I OC (through complex formation) and HLA-F (Fig. 3A). These data suggested that higher affinity peptides spontaneously formed complexes with MHC-I more readily, resulting in increased levels of HLA-F/MHC-I OC dissociation and consequent reduced levels of surface HLA-F.
To test whether specific MHC epitopes contained within an extended polypeptide affected their ability to bind OC of MHC-I, we synthesized three N-terminally biotinylated 50-aa protein fragments of gp100, as follows: one that included the native nonamer sequence and two derivatives that contained the low- and high-affinity mutant sequences. The relative binding of each polypeptide to T2 cells reflected the binding affinity of the epitope sequence contained within it, suggesting a direct interaction of the extended polypeptide with open MHC-I, as seen with the corresponding nonameric peptides. In support of this interpretation, mAbs against MHC-I OC were able to block relative binding of the high-affinity protein to either T2 cells or B-LCL HOM2 (Fig. 3B). To examine the interaction of the peptide sequence with MHC-I and possibly HLA-F, we used the low- and high-affinity polypeptides in comparative precipitation experiments using T2 cells and T2-HLA-B*35 transfectants. Western blot analysis of precipitated material showed that the high-affinity polypeptide consistently bound to quantitatively higher levels of both MHC-I and HLA-F, reflecting the surface-binding abilities and relative affinities of the epitopes contained within the polypeptides (Fig. 3C).
The coincident internalization of exogenous Ag with MHC-I and HLA-F combined with epitope specificity of MHC-I binding suggested a pathway for Ag presentation distinct from existing models for cross-presentation by MHC-I (40–43). To explore a role for the interaction of MHC-I OC and HLA-F in Ag uptake for processing and presentation, we used available CTL clones specific for epitopes HIV-gag p17 and CMV pp65 and tested whether B-LCL target cells that express both HLA-F and the MHC-1–restricting allele could be specifically sensitized to lysis by CTL effectors upon the addition of exogenous Ag. For a number of different combinations of target cell, exogenous protein (HIV-gag p17 and CMV pp65), class I HLA-restricting allele (HLA A*0201, A*0301, B*2705), and CTL, we were able to define conditions in which exogenous proteins optimally sensitized B-LCL to lysis by specific CTL. BfA was used to distinguish between two possibilities that could confer sensitivity to CTL lysis: internalization and processing of exogenous protein or spontaneous formation of MHC-I complex by direct addition of peptide (Fig. 4A). BfA inhibits transport of proteins through the Golgi and induces retrograde protein transport from the Golgi to the ER, thus distinguishing presentation of internalized and processed protein from spontaneous complex formation by direct addition of peptide.
Although inhibition of lysis by BfA discriminated between exogenous protein and direct peptide addition, three additional control experiments were performed to exclude the possibilities that the protein preparations were contaminated with degenerate peptides or were degraded to release corresponding nonamer peptides during the course of the experiments. First, we tested for the presence of small amounts of contaminating peptide by subjecting each protein preparation to LC-MS/MS and examining the spectra for the targeted specific peptides and extended peptides containing the specific peptide sequence (for up to 4 aa extending in both directions). No peptides containing the target epitopes were detectable in any of the preparations (Supplemental Fig. 2A, 2B). Next, to examine the possibility that peptides might be spontaneously generated from proteins during incubation with target cells, we performed a mock incubation identical to experimental conditions used for sensitization, followed by fractionation. Peptide titrations before or after centricon pass-through were overlapping, demonstrating effective recovery of pM concentrations of peptide. In contrast, the pass-through from long polypeptide preparations was ineffective for sensitization to lysis at all concentrations tested, down to the lowest concentrations of protein effective for sensitization (Supplemental Fig. 2C).
The third control experiment was based on our observation that surface binding of biotinylated p17 and p24 Ag to B-LCL was reduced substantially at 4°C versus 37°C for both polypeptides (Supplemental Fig. 3A). Thus, if Ag uptake and processing were required, sensitization to lysis would be impaired at 4°C, but not at 37°C. Conversely, direct addition of peptide should not be impaired at 4°C as peptide uptake is not required for complex formation (28). For both polypeptides, the ability of exogenous Ag to sensitize B-LCL to lysis was impaired at lower temperature, whereas direct addition of peptide was unaffected (Supplemental Fig. 3B). In combination with the blockade of presentation by BfA, these control experiments support the conclusion that specific peptide is neither present in the original protein preparations nor generated external to cells during the course of the experiments prior to exposure of targets to effectors.
The data support a pathway in which exogenous Ags are internalized, processed, and presented by MHC-I by B-LCLs—one potentially involving HLA-F. However, surface expression of HLA-F is upregulated in most lymphocyte subsets upon activation, including activated T cell clones (17), suggesting that this pathway might also function in other cell types that express HLA-F. To test this in T cells, Ag-specific, HLA-F–positive CTL clones were pulsed with Ag directly in the absence of a target cell line and assayed for their ability to act as self-stimulators. The observed increase in IFN-G expression after exposure to extracellular Ag suggested that T cell clones apparently acquire, process, and present exogenous Ag, suggesting that the effector cell itself can be involved in expansion of a memory response by cross-presenting Ag once stimulated (Fig. 4B). The ability of activated effectors to recruit Ag via this pathway suggests that Ag cross-presentation of fragmented target proteins may occur at the site of inflammation during effector responses.
To examine the involvement of HLA-F in the novel pathway directly, we designed shRNA knockdowns targeting HLA-F, using lentivirus constructs to express the shRNA (29). B-LCL KOSE cells expressing HLA-A*0201 were transfected with the construct and with control vector. We screened four distinct sequence constructs for HLA-F and selected one that effected maximal downregulation of HLA-F. Surface levels of HLA-F were markedly decreased using the F4 shRNA construct, which also coincidentally reduced MHC-I H chain expression (Fig. 5A). The downregulation of MHC-I in the F4 transductants may be related to the intracellular interactions previously detected between HLA-F and MHC-I (15). It was not possible in our experience to separate MHC-I OC expression from HLA-F expression; class I–deficient 0.221 cells do not express surface HLA-F, whereas all precursor B-LCL expressing other MHC-I do coexpress HLA-F. Based on this limitation and their observed biophysical interactions, our working hypothesis is that HLA-F and MHC-I OC are expressed codependently.
Binding of biotinylated proteins to the surface of cells transduced with F4 was examined as an indirect measure of the effect that HLA-F and MHC-I H chain levels might have on Ag binding. Both 50-aa polypeptides derived from gp100 and HA-1H showed marked reductions in surface binding on cells treated with HLA-F–specific shRNA (Fig. 5B). Furthermore, when tested for sensitization to lysis by a pp65-specific CTL clone, uptake and processing of exogenous pp65 protein were significantly impaired in F4 transfectants. This was in contrast to the recognition of target cells that occurred with endogenously synthesized pp65 after vaccinia/pp65 infection or those pulsed with specific peptide. In those cases, both knockdowns compared similarly to vector only–transfected control (Fig. 5C). Sensitization assays incorporating mAb HCA2 or 3D11 blocking also demonstrated that specific interference of MHC-I OC or HLA-F prior to addition of exogenous protein affected Ag uptake and subsequent processing for MHC-I presentation (Fig. 5D).
The endogenous class I Ag presentation pathway and its dependence on TAP and tapasin are well characterized (44, 45). As a logical step toward characterization of exogenous MHC-I presentation, we examined two Ag sources in which we could compare endogenous presentation and exogenous presentation in TAP and tapasin mutant lines. We first established that the HA-1H Ag was TAP dependent when presented through the endogenous pathway. B-LCL 721 was typed for HA-1 alleles using an established protocol (21) and found to be HA-1H homozygous. Dependence on TAP was confirmed when a panel of 721 and derivative.134 (TAP negative), 0.134C2 (TAP restored), and class II deletion mutant 0.174 (TAP and tapasin deficient) were tested with HA-1H–specific CTL. Only 721 and TAP-restored 0.134C2 cells were sensitive to lysis by HA-1H–specific CTL (Fig. 6A). We next tested the same panel for access by exogenous Ag using the HA-1H 50-aa polypeptide. In contrast to endogenous Ag, HA-1H CTL lysed both 0.134 and 0.174 mutant lines pulsed with HA-1H protein, and, consistent with prior experiments, BfA inhibited presentation of exogenous protein, but not peptide (Fig. 6B).
Confirmation of the independence of this pathway from TAP and tapasin was obtained with pp65-specific CTL in which the presentation of endogenous pp65, expressed with a vaccinia virus construct, could be compared directly with exogenous protein using the same panel of LCL mutants. CTL specific for a HLA-A*0201–restricted pp65 epitope only recognized the TAP-restored mutant 0.134C2 infected with vaccinia-pp65, consistent with prior studies showing that endogenous pp65 Ag presentation is TAP and tapasin dependent (46). In contrast, exogenous pp65 protein sensitized all targets, including the mutant cell lines. For all targets, sensitization to lysis by protein was inhibited by BfA, whereas peptide sensitization was not altered (Fig. 6C).
Our investigations into the biological function of HLA-F support a novel role for the interaction between HLA-F and OC of MHC-I in the uptake of extracellular Ag for cross-presentation. Several lines of evidence suggest an active role for OC of MHC-I in the uptake of Ag as whole or partial protein that contains epitopes specific to receptive MHC-I expressed on target cells. Exogenous Ag was internalized together with MHC-I, and downmodulation of surface MHC-I resulted in reduced binding, uptake, and presentation of Ag. Differential binding of polypeptides containing low- and high-affinity MHC-I–binding epitopes to MHC-I and HLA-F suggested that exogenous Ag binds to the surface of activated cells to a structure that includes MHC-I OC and HLA-F and that is in contact with the MHC-I–specific epitope sequence found within the extended polypeptide. The requirement of this model for exposure of linear epitopes is consistent with the observation that in our experiments protein required denaturation to enter this pathway, as it did in previously described experiments for cross-presentation of BZLF-1 and pp65 (28). We speculate that MHC-I OC, possibly stabilized by HLA-F, retains rudimentary peptide–binding specificity conferred by the peptide-binding cleft and can bind epitopes in an open-ended fashion without size limitations, possibly similar to MHC class II.
It is well known, based on many crystal structures and considerable binding and T cell recognition data, that a folded class I MHC molecule binds 8- to 10-mer peptides, depending on the allele and epitope, and that, in contrast to MHC-II, the ends of the pocket are closed (4). There are few reported exceptions to this rule, but, whereas the suggestion that extended polypeptide chains are capable of binding to MHC-I OC may be considered controversial, peptide length specificity of some HLA class I alleles has been shown to be very broad and includes peptides of up to 25 aa in length (47). The structure of the MHC-I OC is unknown and is most certainly distinct from the classical structure consisting of H chain, β2m, and peptide. Solving the MHC-I OC structure could address possibilities for an alternative Ag-binding mechanism related to our proposed model for interaction of the MHC-I–binding cleft with epitopes within extended polypeptides.
In addition to MHC-I OC, several experiments directly implicated HLA-F in this pathway. First, the observed overlapping internalization and localization of Ag and HLA-F were coincident with the same observation with MHC-I OCs. Secondly, HLA-F was coprecipitated with MHC-I by polypeptide containing MHC-I–binding epitopes at relative levels approximately paralleling those of MHC-I. Third, downmodulation of HLA-F resulted in interference with Ag binding, uptake, and processing for presentation, and, although downmodulation was coincident with downmodulation of MHC-I, blocking with HLA-F–specific mAb alone interfered with Ag cross-presentation. Indeed, the fact that HLA-F and MHC-I OC, and not MHC-I complex, physically interact in cells relates the findings of MHC-I OC function directly to HLA-F (15). One possible role for HLA-F could be in the stabilization and transport of MHC-I OC to, on, and from the surface. The physical interaction between HLA-F and classical MHC-I OC, their coincident surface expression on activated lymphocytes, and their coincident downmodulation in HLA-F knockdowns combine to suggest that they are interdependent for surface transport. HLA-F may stabilize MHC-I OC as it is formed and the two proteins, possibly as a heterodimer, transit to the surface. Other not mutually exclusive roles include cooperation in the internalization of Ag and MHC-I.
The evidence presented in this work suggests that Ag, MHC-I OC, and possibly HLA-F transit from the surface through the endosomal pathway into lysosomes or lysosome-like structures, where protein is degradated to produce target peptide independently of TAP or tapasin. After complex formation, MHC-I–containing specific peptide derived from the exogenous Ag source is transported to the surface. This model has evident similarities with the MHC-II Ag-presenting pathway in particular, and there is already good evidence that MHC-I molecules visit phagolysosomal compartments to acquire peptides prior to surface expression (48, 49). MHC-I proteins have been shown to reside in endosomes and lysosomes of dendritic cells, and exchange of MHC-I between the cell membrane and endosomal compartments has been demonstrated in both T cells and macrophages (50–52). Furthermore, in the mouse, exogenous MHC-I Ag loading has been associated with endosomal and lysosomal trafficking in dendritic cells (53).
Although previous studies have provided good evidence in support of a pathway for class I loading that is shared with class II molecules (49), our work presents at least three major new findings. First, no evidence implicating the participation of MHC-I OC nor any including HLA-F has been reported. Second, no studies or proposed models have suggested an allelic dependence on MHC-I–specific epitopes for Ag uptake. Third, most cross-presentation pathways studied were operating in professional APCs, primarily dendritic cells (40, 41, 43, 54, 55). In that regard, evidence that other cell types can take up and process Ags has been accumulating. Voeten et al. (56) showed that exogenous recombinant influenza A virus nucleoprotein is processed and presented via MHC-I by EBV-transformed B-LCLs. Also, not only dendritic cells and monocytes, but also B cells can cross-present uBZLF1 in vitro (28).
As noted above, the experimental responses reported in this study were dependent on the use of denatured proteins or polypeptides as opposed to intact proteins, which generally did not evoke similar responses. The likelihood that denatured polypeptides arise in vivo would appear to be high, considering the plethora of proteases active during apoptosis and in cells subjected to cytolysis, thereby producing fragmented proteins containing class I MHC epitopes (57). Because the requirement for HLA-F in this pathway implies it is operating primarily when cells are activated (17) and thus during an inflammatory response, cell lysis at the site of activation arising either from the lytic effects of virus or from the cytolytic activity of innate or adaptive cells would expose the local environment to resultant degradation products. Activated cells exposed to such degradation products could benefit from the ability to take up and process extracellular Ag for a number of different reasons. For example, the possibility that monocyte-derived cells can participate critically in processing Ag for cross-presentation has been suggested, even if they do not present that Ag to T cells themselves (55). Also, cross-presentation by nonprofessional APCs has been demonstrated for HLA class I epitopes from exogenous NY-ESO-1 polypeptides (58). Membrane transfer of Ags from activated B cells to bystander B cells was recently demonstrated (59). Transfer from other activated cells could also be explained by the process of intercellular communication of biomolecules through exosomes (60). This points to the possibility of transfer of lipid rafts containing processed Ag, presented by MHC-I from activated NK or T cells to monocytes, dendritic cells, or B cells (60–63). Other more speculative possibilities include NK cells acquiring Ag from target cells for subsequent stimulation of T cells (in this case memory T cells), particularly when a target cell is lacking in MHC expression. Very little detail has been reported regarding how memory CD8+ T cells are activated, including their potential activation by amateur APCs (64).
Whereas the details of cross-presentation for MHC-I have not been well elucidated, several lines of evidence suggest this immune pathway could play a fundamental role in protection from pathogens (40, 43, 45). If MHC-I OC and HLA-F function in MHC-I Ag cross-presentation, then uncovering the biochemical rules for Ag uptake, processing, and presentation particular to this pathway could provide new strategies for the design and optimization of novel immunogens.
We thank Thomas Spies and Veronika Groh for HCA2 and mutant 721 cell lines and for helpful discussions and critical reading of the manuscript. This work was greatly facilitated by expert technical help from Tori Yamamoto, Lilien Voong, and Ivy Lai and reagents and instruction from Aude Chapuis.
This work was supported by a Burroughs Wellcome Fund Translational Scientist Award and National Institutes of Health Grant CA128283 (to C.Y.), National Institutes of Health Grants CA18029 and AI053193 (to S.R.R.), and National Institute of Child Health and Human Development Grant HD45813 (to D.E.G.). This work was also supported by the University of Washington Center for AIDS Research, a National Institutes of Health–funded program (Grant P30 AI027757 to J.P.G. and D.E.G.) that is supported by the following: National Institute of Allergy and Infectious Diseases, National Cancer Institute, National Institute of Mental Health, National Institute on Drug Abuse, National Institute of Child Health and Human Development, National Heart, Lung, and Blood Institute, and National Center for Complementary and Alternative Medicine. Interim funding support from the Clinical Research Division at the Fred Hutchinson Cancer Research Center is gratefully acknowledged.
The online version of this article contains supplemental material.
Abbreviations used in this article:
The authors have no financial conflicts of interest.