I-Restricted Presentation Processing
Transcription
I-Restricted Presentation Processing
This information is current as of February 11, 2017. Exogenous Peptides Delivered by Ricin Require Processing by Signal Peptidase for Transporter Associated with Antigen Processing-Independent MHC Class I-Restricted Presentation Daniel C. Smith, Awen Gallimore, Emma Jones, Brenda Roberts, J. Michael Lord, Emma Deeks, Vincenzo Cerundolo and Lynne M. Roberts References Subscriptions Permissions Email Alerts This article cites 57 articles, 23 of which you can access for free at: http://www.jimmunol.org/content/169/1/99.full#ref-list-1 Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscriptions Submit copyright permission requests at: http://www.aai.org/ji/copyright.html Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/cgi/alerts/etoc The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 9650 Rockville Pike, Bethesda, MD 20814-3994. Copyright © 2002 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 J Immunol 2002; 169:99-107; ; doi: 10.4049/jimmunol.169.1.99 http://www.jimmunol.org/content/169/1/99 The Journal of Immunology Exogenous Peptides Delivered by Ricin Require Processing by Signal Peptidase for Transporter Associated with Antigen Processing-Independent MHC Class I-Restricted Presentation1 Daniel C. Smith,* Awen Gallimore,† Emma Jones,† Brenda Roberts,* J. Michael Lord,* Emma Deeks,* Vincenzo Cerundolo,† and Lynne M. Roberts2* rotective CD8⫹ CTL that recognize viral or tumor peptides in association with MHC class I are deemed valuable in controlling pathogen spread and, potentially, in reducing tumor progression. In recent years, several avenues have been investigated to pursue the challenge of artificially inducing protective CTL (1–3). These methods have used both live and nonlive vectors to deliver Ags to the cytosol for subsequent presentation of peptides that correspond to epitopes recognized by CTL. Indeed, several protein toxins constitute a class of non-live vectors whose intracellular trafficking and/or membrane translocation abilities can be exploited to allow delivery of peptides for the induction of a CTL response (4 –9). In a previous study we showed that the A subunit of a catalytically defective Shiga-like toxin 1 (SLT1)3 could be engineered so that the holotoxin could successfully deliver a viral peptide into the conventional MHC class I processing and P *Department of Biological Sciences, University of Warwick, Coventry, United Kingdom; and †Molecular Immunology Group, Nuffield Department of Medicine, John Radcliffe Hospital, Oxford, United Kingdom Received for publication January 17, 2002. Accepted for publication April 29, 2002. 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. 1 This work was supported by a Wellcome Trust Program Grant and a U.K. Biotechnology and Biological Sciences Research Council project grant (to L.M.R. and J.M.L.), a grant from the Wellcome Trust (to A.G.), and a grant from Cancer Research U.K. (to V.C.). 2 Address correspondence and reprint requests to Dr. Lynne M. Roberts, Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, U.K. E-mail address: [email protected] 3 Abbreviations used in this paper: SLT1, Shiga-like toxin 1; ER, endoplasmic reticulum; ERAD, ER-associated protein degradation; EndoH, Endo--N-glucosaminidase H; RTA, ricin A chain; RTB, ricin B chain; BFA, brefeldin A; LCMV, lymphocytic choriomeningitis virus; NP, peptide 366 –374 of the influenza nucleoprotein; gp33, peptide 33– 41 of the LCMV glycoprotein. Copyright © 2002 by The American Association of Immunologists, Inc. presentation pathway (7). However, due to the restricted cellular expression pattern of CD77 (10), the receptor for SLT1, the use of this protein as a generalized delivery vehicle may be limited. By contrast, the plant cytotoxin ricin has the capacity to bind to and be endocytosed into a vast number of different cell types (11). Ricin, like SLT1, has an unusual endocytic routing that takes it from the cell surface via the Golgi complex to the endoplasmic reticulum (ER) lumen (12–14). In a productive intoxication of the cell, the fraction of toxin that reaches the ER is somehow perceived as a substrate for ER-associated protein degradation (ERAD). This arm of the ER quality control process normally apprehends newly synthesized but aberrant proteins or orphan subunits of oligomers, and triggers their passage through Sec61 channels to the cytosol. In a tightly coupled sequence of events, such dislocated proteins are normally deglycosylated, certain lysyl residues become ubiquitinated, and the protein is then degraded by proteasomes (reviewed in Ref. 15). In the ER lumen, the catalytic subunit of ricin, though structurally native and endocytosed from the exterior of the cell, appears able to exploit this process to enter the cytosol (16). From our understanding of these events, we reasoned that genetic fusions of immunodominant epitopes with toxins such as SLT1 and ricin could be introduced directly into the MHC class I processing pathway after their retrotranslocation to the cytosol from the ER. In the present study we have investigated the potential of using a disarmed version of ricin to deliver exogenous epitopes and found that, in contrast to SLT1 and most other toxin carriers, delivery and presentation of viral peptides is not dependent on proteasome activity or TAP. We present evidence for release of viral peptides from the toxin carrier by ER signal peptidase due to the presence of a potential signal peptidase cleavage site in the surfaceexposed N-terminal segment of mature ricin A chain (RTA). Therefore, use of engineered ricin may be of particular value in delivering 0022-1767/02/$02.00 Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 In this study we demonstrate that a disarmed version of the cytotoxin ricin can deliver exogenous CD8ⴙ T cell epitopes into the MHC class I-restricted pathway by a TAP-independent, signal peptidase-dependent pathway. Defined viral peptide epitopes genetically fused to the N terminus of an attenuated ricin A subunit (RTA) that was reassociated with its partner B subunit were able to reach the early secretory pathway of sensitive cells, including TAP-deficient cells. Successful processing and presentation by MHC class I proteins was not dependent on proteasome activity or on recycling of MHC class I proteins, but rather on a functional secretory pathway. Our results demonstrated a role for signal peptidase in the generation of peptide epitopes associated at the amino terminus of RTA. We showed, first, that potential signal peptide cleavage sites located toward the N terminus of RTA can be posttranslationally cleaved by signal peptidase and, second, that mutation of one of these sites led to a loss of peptide presentation. These results identify a novel MHC class I presentation pathway that exploits the ability of toxins to reach the lumen of the endoplasmic reticulum by retrograde transport, and suggest a role for endoplasmic reticulum signal peptidase in the processing and presentation of MHC class I peptides. Because TAP-negative cells can be sensitized for CTL killing following retrograde transport of toxin-linked peptides, application of these results has direct implications for the development of novel vaccination strategies. The Journal of Immunology, 2002, 169: 99 –107. 100 TAP-INDEPENDENT DELIVERY OF PEPTIDE USING RICIN VECTOR peptides to MHC class I molecules in TAP-deficient cells, e.g., in tumors where there is a down-regulation of TAP (17). Materials and Methods Reagents Synthetic peptides ASNENMDAM (peptide 366 –374 of the influenza nucleoprotein (NP)) and KAVYNFATC (LCMV glycoprotein (gp33)), encompassing the 366 –374 residues of influenza nucleoprotein and the 33– 41 residues of lymphocytic choriomeningitis virus (LCMV), respectively, were obtained from Alta Bioscience (Birmingham, U.K.) and stored at ⫺20°C in PBS. Lactacystin (Calbiochem, La Jolla, CA), brefeldin A (BFA; Sigma, Poole, U.K.), chloroquine (Sigma), decanoyl-Arg-Val-LysArg-chloromethylketone (Bachem, St. Helens, U.K.) were used for inhibition studies at the concentrations indicated in the figures. Cell lines Production of recombinant RTA-peptide fusions DNAs encoding nNP-RTA and nNP-6K A chain proteins were constructed by PCR mutagenesis using synthetic oligonucleotides containing the mutations of choice in an overall PCR strategy as described previously (7). This resulted in the fusion of the NP sequence (encoding residues 366 –374 of the influenza nucleoprotein) to the mature 5⬘ termini of two RTA cDNA variants encoding RTAR180H (18) and RTA6K-R180H (19), respectively. The claNP-RTA was constructed by inserting annealed complementary oligonucleotides encoding the NP sequence into the ClaI restriction site (at base position ⫹338) in the RTA cDNA (20). PCR-directed mutagenesis was used, as described previously (7), to introduce arginyl point mutations into the DNA coding for the nNP-6K RTA variant. All variants were fully sequenced before expression and fusion constructs were then ligated into a pUC18 vector modified to optimize RTA expression in Escherichia coli (21). Expression and purification of RTA fusions and reassociation into ricin Recombinant RTA fusions were expressed in a 1-L culture of E. coli JM101 cells and purified by cation exchange chromatography as previously described (18). The purified fusions were reassociated to plant-derived ricin B chain (RTB; Vector Laboratories, Burlingame, CA) by dialysis against PBS, following mixing 100 g of each chain in the presence of the reducing agent 2-ME (2% in PBS containing 100 mM lactose), as described previously (22). EndoH treatment EndoH (Boehringer Mannheim, Indianapolis, IN) digestions were performed by resuspending the pellets of protein A-Sepharose-bound immunoprecipitates in 30 l of 50 mM sodium citrate buffer (pH 5.5) containing 0.2% SDS and heating to 95°C for 5 min before addition of 3 U (1.5 l) of EndoH and incubation overnight at 37°C. ER import assays In vitro transcription of mRNAs and translation in rabbit reticylocyte lysate (Promega, Madison, WI), in the presence or absence of canine pancreatic microsomes, was performed as described (24). Cotranslational import into microsomes and cleavage by signal peptidase was determined after a 40min posttranslational treatment with 0.5 mg/ml proteinase K on ice (⫾ 1% Nonidet P-40) and visualizing bands by SDS-PAGE and fluorography. Results Ricin is a heterodimeric protein composed of two subunits, RTA and RTB, that are normally disulfide bonded together. RTA has a rRNA-specific N-glycosidase activity and is therefore capable of inactivating cytosolic ribosomes should it encounter them, while RTB is a galactose-specific lectin able to bind the holotoxin to the target cell surface before endocytic uptake. We created genetic fusions of DNA for the H-2Db-restricted peptides, influenza nucleoprotein NP 366 –374 peptide (ASNENMDAM) (25), or LCMV glycoprotein 1 (gp33) peptide, residues 33– 41 (KAVYN FATC) (25), to either the 5⬘ end of the cDNA encoding mature RTA (20) or spliced into an internal ClaI site that permits exposure on the toxin surface (26) or onto the 3⬘end (data not shown). These proteins were expressed in and purified from E. coli as described previously (18) before being reassociated with glycosylated RTB purified from Ricinus communis (22). For most experiments, a catalytic site mutant (R180H) was used (18), and in one case a variant NP-ricin fusion was made using a lysine-enriched RTAR180H that is known to be more effectively degraded by proteasomes (19). The NP fusion proteins are schematically depicted in Fig. 1. Ag presentation assays Cells were pulsed with ricin fusions at the concentrations and durations stated in the text. Target cells (2 ⫻ 106) were radiolabeled with 100 Ci of 51 Cr for 2 h. Target cells were infected with 5 PFU/cell of recombinant vaccinia virus expressing the relevant viral protein for 90 min before labeling, or pulsed with 500 nM free peptide (NP or gp33) for 2 h during the 51 Cr labeling. Labeled cells were plated at 104 cells/well and CTL were mixed with the targets at the E:T ratios shown in the figures. Supernatants were harvested after 4 h. The percentage of specific lysis was calculated as described previously (7). To study the effect of various inhibitors and reagents on the processing, cells were treated at the concentrations and durations stated in the figures before performing the Ag presentation assay. Immunoprecipitations RMA-S cells (20 ⫻ 106) were treated with toxin (500 ng/ml) or mocktreated for 90 min at 37°C. Cells were washed and starved for 60 min in 1 ml methionine-free RPMI 1640 medium supplemented with 10% (v/v) FCS, 2 mM L-glutamine, 100 U/ml penicillin, and 100 g/ml streptomycin at 37°C. They were then metabolically labeled in 500 l fresh methioninefree RPMI 1640 medium with L-[35S]methionine (Amersham Pharmacia Biotech, Little Chalfont, U.K.) at a concentration of 10 Ci/l for 30 min FIGURE 1. Schematic representation of the ricin fusion proteins. A MHC class I-restricted antigenic peptide (NP) corresponding to residues 366 –374 (ASNENMDAM) from the influenza nucleoprotein was made as a set of recombinant fusions with RTA. Three different fusion proteins are shown, in all of which RTA contained an active site mutation (R180H). The purified recombinant fusions were associated to plant-derived RTB. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 RMA cells and the corresponding TAP-deficient RMA-S cell line, both derived from a Rauscher virus-induced T cell lymphoma, and MC57 cells, a C57BL/6-derived fibrosarcoma, were grown in RPMI 1640 (Life Technologies, Rockville, MD) containing 10% (v/v) FCS, 2 mM L-glutamine, 100 U/ml penicillin, and 100 g/ml streptomycin. For IFN-␥ stimulation, cells were incubated in the presence of 10 U/ml murine rIFN-␥ (PeproTech, London, U.K.) for 48 h. CTL lines were maintained in RPMI 1640 containing 10% (v/v) FCS, 2 mM L-glutamine, 100 U/ml penicillin, 100 g/ml streptomycin, 1 mM sodium pyruvate, and 50 M 2-ME by weekly restimulation using peptide-pulsed spleen cells. at 37°C. Cells were lysed for 30 min on ice in lysis buffer (50 mM Tris-HCl (pH 7.5) containing 1% (v/v) Triton X-100 and 150 mM NaCl), and their nuclei were removed by centrifugation for 15 min at 15,000 ⫻ g. For aliquots to be lysed in the presence of stabilizing peptide, 10 M NP366 –374 was added at this stage. Cell lysates were precleared overnight with 100 l fixed Staphylococcus aureus (Sigma) in PBS, then incubated with the B22 H-2Db conformational-specific Ab (23) for 90 min before adding a small volume of protein A-Sepharose for 45 min. All steps were performed at 4°C. Immunoprecipitates were washed three times in lysis buffer and subjected to Endo--N-glucosaminidose H (EndoH) digestion overnight (see EndoH treatment) and resolved by SDS-PAGE. Fluorographs were analyzed using a densitometer and ImageQuant software (Molecular Dynamics, Sunnyvale, CA). The Journal of Immunology 101 FIGURE 2. Ricin-mediated delivery of the NP366 –374 results in specific lysis by NP-restricted CTL. NP or gp33-specific CTL were tested for their ability to lyse target cells treated as indicated along the x-axis. MC57 target cells, with or without IFN-␥, were pulsed with control peptides, infected with vaccinia virus-expressing full-length Ags, or treated with the NP-ricin fusions. The percentage of specific lysis at three E:T ratios is shown. The data are representative of at least two similar experiments. the endosomal system becomes processed by endoprotease(s) to release the NP peptide such that it is able to interact with either surface recycling MHC class I proteins (28) or with MHC class I complexes that are being secreted. To investigate the possibility of ricin fusions being processed in the early endocytic pathway, we first studied the effect of inhibiting the ubiquitous furin-related serine proteases of the subtilisin family using the competitor inhibitor decanoyl-Arg-Val-Lys-Arg-chloromethylketone-CMK. Members of the furin family are frequently involved in activating proproteins in FIGURE 3. Presentation of ricin-delivered peptide is not dependent on functional proteasomes. NP-specific CTL were tested for their ability to lyse MC57 target cells treated as indicated along the x-axis. MC57 cells were incubated with either nNP-ricin or nNP-6K for 16 h. Where indicated, cells were pretreated with IFN-␥ and subsequently with 20 M clastolactacystin--lactone (L) in 0.2% DMSO for 24 h before the addition of toxin. MC57 cells were used as target cells. Controls included MC57 cells infected with a vaccinia virus-encoding influenza nucleoprotein (VacNP), or pulsed with either influenza nucleoprotein (NP pep) or LCMV glycoprotein (gp33 pep) peptides. The percentage of specific lysis at three E:T ratios is shown. The data are representative of at least two similar experiments. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 Ricin was able to deliver NP peptide to MC57 cells for processing and presentation by MHC class I molecules, because cells treated with nNP-ricin or nNP6K-ricin were sensitized for specific lysis by NP366 –374-restricted CTL, but not by the irrelevant gp33restricted CTL (Fig. 2). Controls show that these cells were able to process influenza nucleoprotein expressed in the cytosol from a vaccinia virus construct (VacNP). However, for both nNP-ricin or nNP6K-ricin, significant lysis was observed only in toxin-pulsed MC57 cells following pretreatment for 48 h with IFN-␥. Such treatment is known to up-regulate gene expression of MHC class I proteins, several proteasomal subunits, and the proteasome regulator PA28 (27). The fact that IFN-␥ is not known to up-regulate extracellular proteases and that presentation of the NP peptide was only seen after IFN-␥ treatment of MC57 cells suggests that extracellular processing of toxin is not occurring. Supporting this conclusion, peptide presentation was blocked when the nNP-ricin was prevented from binding and subsequently entering cells by pretreating the toxin with 100 mM lactose (data not shown). Interestingly, both the claNP-ricin fusion (Fig. 2) and a C-terminally located NP-ricin fusion (data not shown) failed to promote specific lysis. Therefore, it was decided to focus on the N-terminally located NP-ricin fusions. Surprisingly, however, peptide presentation was not dependent on active proteasomes (Fig. 3). When MC57 cells were treated for 24 h with clasto-lactacystin--lactone, such that processing and presentation of NP peptides from vaccinia-expressed nucleoprotein was significantly reduced (Fig. 3), surface presentation still occurred under these conditions when cells were challenged with free NP peptide or the fusions nNP-ricin or nNP-6K-ricin. We also observed that the extent of specific lysis between nNP-ricin and the normally more degradable nNP-6K-ricin, in either the presence or absence of proteasome inhibition, was consistently similar. Most strikingly of all, processing and presentation of ricin-delivered NP was independent of functional TAP (Fig. 4). The TAP-deficient RMA-S cells were significantly lysed when ricin delivered NP peptide via endocytosis. Furthermore, this presentation was not dependent on IFN-␥ (Fig. 4). Taken together, these data suggest that the processing of NP366 –374 from ricin was occurring within the endomembrane system of the cell rather than at the cell surface or within the cytosol. It was of interest to locate the site of peptide processing and MHC class I loading within the endomembrane system. Ricin is a promiscuous toxin that, on its journey, can harmlessly enter all compartments of the endosomal/lysosomal/trans-Golgi network system of the cell. Therefore, it is conceivable that ricin entering 102 TAP-INDEPENDENT DELIVERY OF PEPTIDE USING RICIN VECTOR the late secretory pathway, and they have been shown able generate class I peptides (29 –31). However, the furin inhibitor did not block the processing of the N-terminally placed NP peptide from ricin (Fig. 5A). Likewise, presentation was not sensitive to the disruption of endosome/lysosome pH by the lysomotropic amine, chloroquine, which affects furin activity (32) and which would be predicted to indirectly FIGURE 5. Presentation of toxin-derived NP peptide is both furin and chloroquine independent but is reliant on a functional secretory pathway. RMA-S cells (6 ⫻ 106) were incubated with nNP-6K ricin (nNP-6K) at 500 ng/ml for 2 h, following a 2-h pretreatment with increasing concentrations of the specific furin inhibitor decanoyl-Arg-Val-Lys-Arg-chloromethylketone (A), chloroquine (B), or BFA (C). Controls include RMA-S cells pretreated with the various inhibitors for 2 h, then pulsed with free peptide from either influenza nucleoprotein (NP pep) or LCMV glycoprotein (gp33 pep). The percentage of lysis is shown for three E:T ratios. The data are representative of at least two similar experiments. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 FIGURE 4. Ricin-dependent peptide presentation is more efficient in TAP-deficient RMA-S cells. MC57, RMA, and RMA-S cells were incubated with nNP-ricin or nNP-6K (50 ng/ml) for 16 h. NP-specific CTL were tested for their ability to lyse target cells treated as indicated along the x-axis. Controls include cells pulsed with free peptide from either influenza nucleoprotein (NP pep) or LCMV glycoprotein (gp33 pep). The percentage of lysis at three E:T ratios is shown. The data are representative of at least two similar experiments. reduce the proteolysis normally occurring within acidic compartments (Fig. 5B). To further investigate any processing in the early endocytic pathway, and to specifically rule out presentation by surface recycling MHC class I proteins, we studied the effect of disrupting both anterograde and retrograde traffic in the Golgi stack with BFA. Such treatment would prevent the toxin moving along its productive entry pathway to the ER and concomitantly block exocytosis of newly assembled MHC class I complexes from the ER. Fig. 5C shows that surface presentation of the NP peptide delivered into cells by ricin is extremely sensitive to BFA. These data support the idea that processing and loading were occurring in the early secretory pathway. To more precisely investigate whether peptide processed from ricin interacted with MHC proteins as they were being transported through the early secretory pathway, or whether it interacted to stabilize MHC complexes in the ER, we looked at the loading of newly synthesized, empty MHC class I H chains (Fig. 6). This was achieved by radiolabeling RMA-S cells with a short pulse of [35S]methionine after treating cells with or without ricin or the ricin-peptide fusion, followed by detergent lysis, immunoprecipitation of any peptide-loaded H-2Db complexes using the conformation-specific mAb B22, and EndoH treatment. Samples were compared with non-EndoH-treated controls. One hundred percent relative loading was seen when cell lysates were incubated with free NP peptide, while the background level of correctly assembled complexes was ⬃10%. The pulse-labeled MHC H chains were EndoH sensitive as judged by the faster migration of bands compared with a non-EndoH-treated sample (Fig. 6A). When cells were pretreated with nNP-6K ricin, the amount of immunoprecipitable peptide-loaded MHC complexes increased by ⬃180% relative to untreated or ricin-only controls (Fig. 6, B and C). A structural-based algorithm (33) used to predict any H-2Db MHC class I epitopes present in the nNP-6K ricin primary sequence identified a total of 23 putative epitopes. Of these only 12 had an energy score to indicate good binding interactions, with the NP epitope ranked first. Indeed, as shown in Fig. 6, B and C, there was no The Journal of Immunology increase above the background level in the loading of newly synthesized MHC proteins when the 6K-ricin lacking the NP epitope was tested. Collectively, the experimental and theoretical data in- dicated that the increased loading seen in TAP-negative RMA-S cells of newly synthesized, EndoH-sensitive MHC proteins could be attributed solely to the presence of the NP epitope within the ricin construct, rather than to processing and loading of internal RTA peptides. The significant increase in loading of newly synthesized, EndoH-sensitive MHC proteins when TAP-deficient cells were treated with nNP-ricin supported the idea that the ricin fusions were being processed in the ER or cis-Golgi. Interestingly, examination of the N terminus of the nNP-RTA revealed the existence of two potential signal peptidase cleavage sites that fit the consensus defined by von Heijne et al. (34) (Fig. 7A) but that are never used in the biosynthesis of the toxin in the producing plant (35). Theoretical prediction analysis indicated that the second site (AGA2TV) would be a more effective cleavage site than the first (IFP2KQ). When these sequences were engineered into the secretory protein preprolactin to replace the natural preprolactin signal peptidase cleavage site, they were indeed able to serve as cleavage sites during in vitro translation in the presence of canine microsomal membranes. After cotranslational import, a posttranslational incubation with protease showed that the lower, signal peptide-cleaved band was protected within the vesicles, unless the protease was added together with a membrane-disrupting detergent (Fig. 7B). As predicted, the first site, where incomplete cleavage was seen, was not as effective a substrate as the second site (Fig. 7B, ⴱ). To exclude the possibility that simply any sequence from the N terminus of RTA can serve to provide a signal peptidase cleavage site, a stretch of residues between the two putative cleavage sites (PIINF) was also engineered into preprolactin. Because the translation product in this case was resistant to proteinase K (i.e., protected within the membrane vesicles) but remained a full-length protein, we conclude that the sequence was not able to provide a cleavage site for signal peptidase (Fig. 7B). Signal peptidase is the best-characterized endoprotease within the ER. To investigate any Ag processing function of this protease, the two potential cleavage sites within the nNP-6K ricin were separately rendered noncleavable by replacing the ⫺3 and ⫺1 positions with arginine. As described previously (36), such changes would preclude processing by signal peptidase. Indeed, replacement of the preprolactin cleavage site with either RFRKQ or RGRTV, followed by in vitro translation in the presence of canine microsomal membrane, confirmed that such substitutions prevented cleavage (data not shown). Therefore, these mutant proteins carrying the NP peptide were purified from E. coli. They were shown to be structurally sound because the modified RTAs readily reassociated with RTB, and when the catalytically active RTA equivalent was used the reassembled holotoxins exhibited native FIGURE 7. Sites in the first 20 residues of mature RTA can function as substrates for signal peptidase. A, The first 20 amino acids of mature RTA contain two potential signal peptidase cleavage sites (in boldface and underlined, where arrowheads indicate potential cleavage sites). These sequences, along with a control sequence (boldface only), were inserted into preprolactin (ppl) to replace its natural signal peptide cleavage site. B, In vitro transcripts encoding preprolactin wild type, preprolactin-IFPKQ, preprolactin-PIINF, and preprolactin-AGATV were translated in vitro in rabbit reticulocyte lysates in the absence or presence of canine pancreas microsomes (RMs). Some samples were then treated with proteinase K (PrK) in the absence or presence of the detergent Nonidet P-40 (NP-40) as indicated. Translation products were separated by SDS-PAGE and visualized by fluorography. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 FIGURE 6. Toxin-derived NP peptide stabilizes newly synthesized H-2Db molecules. A, RMA-S cells (20 ⫻ 106) were incubated for 90 min with medium alone (lanes 1-3), nNP-6K ricin (nNP-6K; lanes 4 and 5), or 6K ricin (lanes 6 and 7) at 500 ng/ml. Following radioactive labeling for 30 min, cell lysates were precleared in the absence (lanes 3, 5, and 7) or presence (lanes 1, 2, 4, and 6) of a saturating concentration (20 M) of a stabilizing influenza nucleoprotein peptide (NP pep). Peptide-loaded H-2Db molecules were recovered by immunoprecipitation using the conformation-specific Ab B-22 that detected only loaded H chains, and immunoprecipitates were treated with EndoH (lanes 2-7) or were left untreated (lane 1). B, Resolved bands were quantified and expressed as the percentage of relative loading of newly synthesized H-2Db molecules. Integrated optical band intensities, expressed as a percentage of radiolabeled unstabilized to radiolabeled stabilized immunoprecipitations, were taken as a measure of the loading of peptide onto newly synthesized H-2Db molecules. C, The percentage of increase in loading upon incubation with nNP6K, compared with the average of the mock-treated and the 6K ricintreated controls, is shown. The data are means of duplicate samples and are representative of two similar experiments. 103 104 TAP-INDEPENDENT DELIVERY OF PEPTIDE USING RICIN VECTOR Discussion Ricin is an opportunistic toxin that can promiscuously bind to cell surfaces via the galactoside-binding RTB and be internalized by endocytosis using a multiplicity of pathways to reach endosomes, lysosomes, the Golgi complex, and the ER lumen (37). Only the FIGURE 8. Signal peptidase is involved in Ag processing from ricinmediated delivery. RMA-S cells (6 ⫻ 106) were incubated with 500 ng/ml nNP-6K ricin (nNP-6K), nNP(RFRKQ)-6K ricin (nNP-RFRKQ-6K), or nNP(RGRTV)-6K ricin (nNP-RGRTV-6K) for 16 h. NP-specific CTL were tested for their ability to lyse target cells treated as indicated along the x-axis. Control RMA-S cells were pulsed with free peptide from either influenza nucleoprotein (NP pep) or LCMV glycoprotein (gp33 pep). The percentage of lysis at three E:T ratios is shown. The data are representative of at least two similar experiments. FIGURE 9. Ricin-mediated delivery and TAP-independent surface presentation is not epitope specific. RMA-S cells (6 ⫻ 106) were incubated with either 500 ng/ml nNP-6K ricin (nNP-6K) or 500 ng/ml gp33-6K ricin (gp33-6K) for 16 h. NP- and gp33-specific CTL were tested for their ability to lyse target cells treated as indicated along the x-axis. Control RMA-S cells were pulsed with free peptide from either influenza nucleoprotein (NP pep) or LCMV glycoprotein (gp33 pep). The percentage of lysis is shown for three E:T ratios. The data are representative of at least two similar experiments. fraction of toxin that reaches the ER will retrotranslocate to the cytosol (14), where a small proportion evades proteasomal degradation to inactivate ribosomes (19). We reasoned that, because the productive intoxication pathway involves toxin dislocation from the ER as a pseudo-substrate of the ERAD pathway and an encounter with the ubiquitin/proteasome machinery, it may be possible to use ricin to deliver Ags directly into the MHC class I processing pathway. Our data now show that nanomolar amounts of a catalytically crippled form of ricin can indeed be used as a delivery vehicle in vitro. However, only peptides genetically fused at the N terminus of RTA were successfully processed and presented (Fig. 2), and processing was not dependent on active proteasomes (Fig. 3). Indeed, CTL-induced lysis was most efficient when ricin fusions were internalized by cells of a TAP-deficient cell line in the absence of IFN-␥ (Fig. 4). So, although ricin can retrotranslocate from the ER of TAP-deficient cells (known because the native version can kill RMA-S cells (data not shown)) processing of the fused peptide from nNP-ricin is clearly not dependent on this step or other downstream events within the cytosol. It is now recognized that the classical pathway for the presentation of endogenous antigenic peptides is not the only route by which peptides can reach MHC class I proteins. Indeed, processing of Ag and loading onto MHC class I can be quite flexible. For example, cross-presentation of exogenous Ag may occur. This may happen in some cell lines when exogenous Ags are internalized to intersect the classical processing pathway in the cytosol after release from endosomes or other intracellular compartments, or following endosomal degradation and recycling of peptide to bind cell surface class I molecules (38). However, raising endosomal pH to reduce proteolytic activity or blocking furin-related endoproteases did not prevent peptide presentation (Fig. 5, A and B). Furthermore, because protein traffic through recycling endosomes is known not to be affected by BFA, the effect of this drug in blocking the presentation of peptide delivered via the ricin carrier (Fig. 5C) would again preclude cross-presentation events in this particular case. Class I Ag processing in the cytosol is normally mediated by proteasomes (for an example, see Ref. 39), although this may not Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 potency (IC50, 0.015 nM) (19). This indicated that the primary structural changes at the N terminus of RTA did not perturb toxin trafficking to the ER. To test that signal peptidase could act posttranslationally to process peptides appended to the N terminus of RTA, we added the purified mutant proteins described above to TAP-deficient cells and looked for peptide presentation. We observed that surface presentation of the NP peptide did not occur when the first of the predicted signal peptidase sites (IFP2KQ) was deliberately disrupted by substitution with arginine (Fig. 8). By contrast, when the second predicted cleavage site (AGA2TV) was mutated to prevent signal peptidase cleavage, presentation was reproducibly higher than that seen with the control (nonmutated) nNP-ricin (Fig. 8; for a possible explanation, see Discussion). These findings support the idea that CD8⫹ peptide epitopes positioned at the N terminus of RTA may be cleaved from the endocytosed toxin posttranslationally by signal peptidase (or an undetermined protease with signal peptidase-like activity), loaded onto MHC class I proteins and subsequently presented, only when the sequence IFP2KQ is present at the start of mature RTA. Signal peptidase processing at this site would release the nine-residue N-terminal NP epitope with a short C-terminal extension of three residues of mature RTA. To test whether such processing is confined to one particular peptide at the N terminus of RTA, we tested an alternative peptide. When the H-2Db-restricted gp33 peptide from LCMV was positioned at the N terminus of RTA to generate gp33-ricin, specific lysis by gp33-specific CTL was observed using the standard assay with RMAS target cells (Fig. 9). The possibility that ricin may provide a signal peptidase site to allow the processing of N-terminally positioned Ags is addressed in Discussion. The Journal of Immunology the ricin-producing plant (35). According to theoretical predictions (34), the second of these sites would be the more effective substrate. Indeed, when the natural signal peptidase cleavage site of a preprolactin reporter was substituted with either of these sequences and tested in a translation system supplemented with ER-derived microsomes, it was clear that they could serve as substrates for signal peptidase with the predicted efficiencies (Fig. 7). Furthermore, when the first of these sites was replaced with the nonfunctional arginyl sequences in nNP-RTA, peptide presentation in TAP-deficient cells was completely abolished (Fig. 8). However, a similar mutation of the second cleavage site in nNP-RTA, predicted to be a more effective signal peptidase substrate, actually provoked an increase in peptide presentation. In the absence of a more extensive investigation we can only speculate as to why this happens. We propose that, although Nterminal peptide fusions to ricin are normally accessible to a signal peptidase-like activity for the production of both peptides, it is somehow more favorable to load only the shorter peptide of the two for subsequent surface presentation. Cleavage at the first site would release a peptide predicted to contain the nine-residue NP peptide with an additional three residues at its C terminus (Fig. 7). By contrast, cleavage at the second site would generate a longer peptide (the nine-residue peptide epitope with a 16-residue C-terminal extension of the RTA sequence; Fig. 7). This hypothesis is consistent with a need for further processing to remove the Cterminal extensions, because we consider it unlikely that 12- and 25-mer peptides generated by signal peptidase would be presented at the cell surface. Although it is known that the NP366 –374 nonamer with a C-terminal extension of three residues can directly assemble with class I molecules, this species has much faster off rates than the optimal nonamer (51). Therefore, it is likely that another enzyme(s) is involved, one that is only able to cut or trim the shorter 12-mer peptide to optimal size, possibly because the longer peptide possesses a structure incompatible for further signal peptidase cleavage and/or subsequent processing. In this scenario, successful generation and loading of NP366 –374 would normally be a competitive process. Abrogating cleavage at the second site, as in nNP-(RGRTV)-6K (Fig. 9), would in effect ensure the production of only the shorter peptide, rather than a mixed population, and thereby increase the opportunities for successful loading and presentation. Consistent with the hypothesis that extensive Ag presentation can occur within the ER lumen, it has previously been shown that, in cells devoid of TAP, the embedded NP366 –374 epitope within a full-length, ER-targeted NP mutant that lacked N-linked glycans was released for assembly with class I proteins (52). However, a full-length glycosylated NP targeted to the ER in an identical fashion was not capable of sensitizing TAP-negative target cells for lysis (44). These results strongly suggest that the tertiary conformation of polypeptides within the ER lumen may influence the generation of MHC class I epitopes. It should be noted that the longer peptide predicted to arise from the ricin fusion carries a glycosylation sequon. Whether the longer peptide becomes posttranslationally glycosylated (14) to similarly preclude further processing, or whether its structure is simply inappropriate for processing, remains unknown. Although proteases are known to exist in the ER lumen (Refs. 46 – 48 and reviewed in Ref. 53), the exact complement and role of resident carboxypeptidases, aminopeptidases, and endoproteases in Ag processing is unclear. Where Ags have been extensively processed in the ER (44), the enzyme(s) responsible have not usually been identified. One ER endoprotease that is known to be involved in the generation of HLA-E binding epitopes is signal Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 be an absolute requirement (40, 41). In this work we have shown that peptide processing from ricin and surface presentation of peptide can occur in cells where proteasomes are inhibited. Because presentation can also occur when TAP is absent, these data suggest that processing is occurring outside the cytosol. Although peptide transport through TAP is the optimal route by which peptides are loaded onto empty MHC proteins in the ER, it is now recognized that a physical interaction with TAP is not required for loading (42). Indeed, intracellular peptides can be presented on the surface of TAP-negative cells after delivery to the ER through Sec61 translocons via signal peptides (43, 44) or following direct penetration of exogenous peptides into cell organelles (45). Taken together, our data are consistent with processing and loading within the secretory pathway. Where in the secretory pathway could the processing occur? In addition to the possibility of endosomal processing outlined above, Ag processing may take place within the secretory pathway by unknown exoproteases, or by endoproteases of the furin family, for example. Furin is a member of the subtilisin family of serine proteases that is found predominantly in the TGN from where it cycles via the cell surface. The TAP-independent generation of MHC class I peptides by furin has been observed when placed at the C terminus of the hepatitis B virus core protein (29 –31). However, the specific furin inhibitor we used had no effect, and the lack of putative furin sites toward the N terminus of RTA would argue against a role for this protease in the release of epitopes joined to the N terminus of RTA. There are, of course, numerous other endoproteases within the late secretory pathway, most being cell type specific for the activation of proproteins such as hormones, growth factors, extracellular enzymes, etc. In our study, we saw a significant increase in peptide-stabilized MHC complexes that were still EndoH sensitive when TAP-deficient cells are treated with the toxin fusions (Fig. 6). This strongly suggested that loading was occurring within the ER lumen itself. If this were happening it would clearly make sense for the peptides to be proteolytically released within this compartment, in the vicinity of empty class I H chains. Indeed, this may help explain the IFN-␥-independent processing and presentation in TAP-deficient RMA-S cells (Fig. 4). Treatment of RMA-S cells with IFN-␥ would lead to an increase in MHC class I proteins in the ER lumen (27). However, such an increase would not be predicted to enhance presentation of the toxin-delivered NP peptide because, in the absence of other peptides entering from the cytosol, there is always a high concentration of available class I proteins within the ER. Conversely, in RMA cells, an IFN-␥-mediated increase in class I molecules within the ER would permit increased loading of the processed NP peptide in the face of competition from other peptides entering via TAP. The ER lumen is a major site of protein folding in an environment not compatible with protein degradation events. Therefore, the consequent likely dearth of endoproteases within the ER would appear to make endoproteolytic release of peptides from ricin unlikely. Nevertheless, some endoproteases do exist within the ER membrane and lumen (e.g., signal peptide peptidase (46) and the chaperones ERp60 and ERp72 (47, 48)). The best-characterized of these is the signal peptidase complex that cleaves short ER-targeting peptide sequences from newly synthesized polypeptides. However, in addition to the removal of signal peptides, this enzyme may also be responsible for the processing the H-2a subunit of the mammalian asialoglycoprotein receptor (49) and in the degradation of aberrant membrane proteins in yeast (50). Close examination of the mature N terminus of RTA revealed two potential signal peptidase cleavage sites, IFP⫹32KQ and AGA⫹162TV, although it should be noted that neither is used in 105 106 TAP-INDEPENDENT DELIVERY OF PEPTIDE USING RICIN VECTOR Acknowledgments We thank Prof. Tim Elliott (University of Southampton, Southampton, U.K.) and Prof. Eric Tartour (Institut Pasteur, Paris, France) for helpful discussions. References 1. Bona, C. A., S. Caseres, and T.-D. Brumenu. 1998. Towards development of T-cell vaccines. Immunol. Today 19:126. 2. Tartour, E., A. Ciree, N. Haicheur, F. Benchetrit, and W. H. Fridman. 2000. Development of non-live vectors and procedures as tools for cancer vaccines. Immunol. Lett. 74:45. 3. Bonnet, M. C., J. Tartaglia, F. Verdier, P. Kourilsky, A. Lindberg, M. Klein, and P. Moingeon. 2000. Recombinant viruses as a tool for therapeutic vaccination against human cancers. Immunol. Lett. 74:11. 4. Donnelly, J. J., J. B. Ulmer, L. A. Hawe, A. Friedman, X. P. Shi, K. R. Leander, J. W. Shiver, A. I. Oliff, D. Martinez, D. Montgomery, and M. M. Lui. 1993. Targeted delivery of peptide epitopes to class I major histocompatibility molecules by a modified Pseudomonas exotoxin. Proc. Natl. Acad. Sci. USA 90:3530. 5. Ballard, J. D., R. J. Collier, and M. N. Starnbach. 1996. Anthrax toxin-mediated delivery of a cytotoxic T-cell epitope in vivo. Proc. Natl. Acad. Sci. USA 93: 12531. 6. Lee, R.-S., E. Tartour, P. van der Bruggen, V. Vantomme, I. Joyeux, B. Goud, W. H. Fridman, and L. Johannes. 1998. MHC class I presentation of exogenous soluble tumor antigen fused to the B-fragment of Shiga toxin. Eur. J. Immunol. 28:2726. 7. Noakes, K. L., H. T. Teisserenc, J. M. Lord, R. D. Dunbar, V. Cerundolo, and L. M. Roberts. 1998. Exploiting retrograde transport of Shiga-like toxin 1 for the delivery of exogenous antigens into the MHC class I presentation pathway. FEBS Lett. 453:95. 8. Carbonetti, N. H., T. J. Irish, C. H. Chen, C. B. O’Connell, G. A. Hadley, U. McNamara, R. G. Tuskan, and G. K. Lewis. 1999. Intracellular delivery of a CTL epitope peptide by pertussis toxin to MHC class I without involvement of the cytosolic class I antigen processing pathway. Infect. Immun. 67:601. 9. Haicheur, N., E. Bismuth, S. Bosset, A. Adotevi, G. Warnier, V. Lacabanne, A. Regnault, C. Desaymard, S. Amigorena, P. Castagnoli, et al. 2000. The B subunit of Shiga toxin fused to a tumor antigen elicits CTL and targets dendritic cells to allow MHC class I-restricted presentation of peptides derived form exogenous antigens. J. Immunol. 165:3301. 10. Eiklid, K., and O. Olsnes. 1983. Interaction of Shigella dysenteriae cytotoxin with receptors on sensitive and insensitive cells. J. Recept. Res. 1:380. 11. Olsnes, O., and A. Pihl. 1982. Toxic lectins and related proteins. In Molecular Action of Toxins and Viruses. P. Cohen and S. van Heyningen, eds. Elsevier Biomedical, Amsterdam, p. 51. 12. Sandvig, K., O. Garred, K. Prydz, J. Kozlov, S. H. Hansen, and B. van Deurs. 1992. Retrograde transport of endocytosed Shiga toxin to the endoplasmic reticulum. Nature 358:510. 13. Simpson, J. C., C. Dascher, L. M. Roberts, J. M. Lord, and W. E. Balch. 1995. Ricin cytotoxicity is sensitive to recycling between the endoplasmic reticulum and the Golgi complex. J. Biol. Chem. 270:20078. 14. Rapak, A., P. Falsnes, and S. Olsnes. 1997. Retrograde transport of mutant ricin to the endoplasmic reticulum with subsequent translocation to cytosol. Proc. Natl. Acad. Sci. USA 94:3783. 15. Brodsky, J. L., and A. A. McCracken. 1999. ER protein quality control and proteasome-mediated protein degradation. Semin. Cell Dev. Biol. 10:507. 16. Simpson, J. C., L. M. Roberts, K. Römisch, J. Davey, J, D. H. Wolf, and J. M. Lord. 1999. Ricin A chain utilises the endoplasmic reticulum-associated protein degradation pathway to enter the cytosol of yeast. FEBS Lett. 459:80. 17. Kamarashev, J., S. Ferrone, B. Seifert, R. Boni, F. O. Nestle, G. Burg, and R. Dummer. 2001. TAP1 down-regulation in primary melanoma lesions: an independent marker of poor prognosis. Int. J. Cancer 95:23. 18. Day, P. J., S. R. Ernst, A. E. Frankel, A. F. Monzingo, J. M. Pascal, M. C. Molina-Svinth, and J. D. Robertus. 1996. Structure and activity of an active site substitution of ricin A chain. Biochemistry 35:11098. 19. Deeks, E. D., J. P. Cook, P. J. Day, D. C. Smith, L. M. Roberts, and J. M. Lord. 2002. The low lysine content of ricin A chain reduces the risk of proteolytic degradation after translocation from the endoplasmic reticulum to the cytosol. Biochemistry 41:3405. 20. Lamb, F. I., L. M. Roberts, and J. M. Lord. 1985. Nucleotide sequence of cloned cDNA coding for preproricin. Eur. J. Biochem. 148:265. 21. Ready, M. P., Y. Kim, and J. D. Robertus. 1991. Site-directed mutagenesis of ricin A-chain and implications for the mechanism of action. Proteins 10:270. 22. Simpson, J. C., J. M. Lord, and L. M. Roberts. 1995. Point mutations in the hydrophobic C-terminal region of ricin A chain indicate that Pro250 plays a key role in membrane translocation. Eur. J. Biochem. 232:458. 23. Townsend, A., T. Elliott, V. Cerundolo, L. Foster, B. Barber, and A. Tse. 1990. Assembly of MHC class I molecules analyzed in vitro. Cell 62:285. 24. Chaddock, J. A, L. M. Roberts, B. Jungnickel, and J. M. Lord. 1995. A hydrophobic region of ricin A chain which may have a role in membrane translocation can function as an efficient noncleaved signal peptide. Biochem. Biophys. Res. Comm. 217:68. 25. Elliott, T., M. Smith, P. Driscoll, and A. McMichael. 1993. Peptide selection by class I molecules of the major histocompatibility complex. Curr. Biol. 3:854. 26. Rutenber, E., B. J. Katzin, S. Ernst, E. J. Collins, D. Mlsna, M. P. Ready, and J. D. Robertus. 1991. Crystallographic refinement of ricin to 2.5 A. Proteins 10:240. 27. Fruh, K., and Y. Yang. 1999. Antigen presentation by MHC class I and its regulation by interferon ␥. Curr. Opin. Immunol. 11:76. 28. Chiu, I., D. M. Davies, and J. L. Strominger. 1999. Trafficking of spontaneously endocytosed MHC proteins. Proc. Natl. Acad. Sci. USA 96:13944. 29. Gil-Torregrosa, B. C., R. Castano, and M. Del Val. 1998. MHC class I viral antigen processing in the secretory pathway defined by the trans-Golgi network protease furin. J. Exp. Med. 188:1105. 30. Gil-Torregrosa, B. C., R. Castano, D. Lopez, and M. Del Val. 2000. Generation of MHC class peptide antigens by protein processing in the secretory route by furin. Traffic 1:641. 31. Zhang, Y., Y. Kida, K. Kuwano, Y. Misumi, Y. Ikehara, and S. Arai. 2001. Role of furin in delivery of a CTL epitope of an anthrax toxin-fusion protein. Microbiol. Immunol. 45:119. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 peptide peptidases (54). This enzyme cleaves transmembrane signal peptides within their hydrophobic domain, liberating the resulting N-terminal fragment to the cytosol (46). However, the peptides we predict to arise upon signal peptidase proteolysis in the present study appear neither sufficiently hydrophobic nor membrane associated to serve as good signal peptide peptidase substrates. Therefore, for the moment, questions regarding the precise mechanism of posttranslational signal peptidase cleavage of the N-terminal ricin fusions, any interplay between the peptides generated, and the exact nature and sequence of the processing steps remain unknown and are clearly beyond the scope of the present study. Signal peptidase has been previously speculated to posttranslationally process the endogenous ER membrane protein JAW1, releasing fused peptides for surface presentation by class I MHC (55). However, functional evidence for the involvement of signal peptidase was not presented in this study. By contrast, the data presented here show that a protein, physiologically able to reach the ER lumen from the exterior of the cell, can be processed there by signal peptidase or a signal peptidase-like activity. Such processing may be coupled with the protein disulfide isomerase-catalyzed unfolding that toxins appear to require pretranslocation (56) and/or the retrotranslocation process itself. By definition, the early stages of the ERAD of ricin would occur in proximity to the dynamic, bidirectional translocation machinery that contains, on its periphery, signal peptidase. Further characterization of these events is now required. To check that these events were not limited to presentation of the NP epitope, we created a similar N-terminal fusion of the T cell epitope gp33 with RTA. This was also processed and presented in a TAP-independent manner (Fig. 9), in marked contrast to the TAP-dependent presentation of gp33 from the LCMV glycoprotein (57, 58). This finding supports a model of signal peptidase processing within the sequence of mature RTA and suggests that, provided the fused peptide did not cover the cleavage site, any added peptide would be similarly processed. Administration of ex vivo toxin-treated APCs to promote protective CTL as part of a vaccination protocol remains an obvious long-term goal of toxin exploitation. Of the toxins tested for such an application, only pertussis toxin (8) and ricin (this study) appear able to deliver peptides in a BFA-sensitive, TAP-independent manner. Although the processing of an N-terminal peptide Ag borne by pertussis toxin clearly occurred within the secretory pathway, the involvement of signal peptidase or other ER protease, and the site of subsequent peptide loading, were not demonstrated (8). The behavior of the ricin vector is particularly striking. The strong TAP-independent bias exhibited by this vehicle may pave the way for developing disarmed ricin vectors directed solely against certain TAP-negative tumors in vivo. Theoretically, it follows that preferential TAP-independent presentation of a common viral epitope, such as the influenza nucleoprotein NP peptide, by a TAPdeficient tumor cell might allow tumor clearance by preexisting viral-specific CTL. The Journal of Immunology 45. Lu, J., P. J. Wettstein, Y. Higashimoto, E. Appella, and E. Celis. 2001. TAPindependent presentation of CTL epitopes by Trojan antigens. J. Immunol. 166: 7063. 46. Weihofen, A., M. K. Lemberg, H. L. Ploegh, M. Bogyo, and B. Martoglio. 2000. Release of signal peptide fragments into the cytosol requires cleavage in the transmembrane region by a protease activity that is specifically blocked by a novel cysteine protease inhibitor. J. Biol. Chem. 275:30951. 47. Urade, R., Y. Takenaka, and M. Kito. 1993. Protein degradation by Erp72 from rat and mouse liver endoplasmic reticulum. J. Biol. Chem. 268:22004. 48. Otsu, M., R. Urade, M. Kito, F. Omura, and M. Kikuchi. 1995. A role of ER-60 protease in the degradation of misfolded proteins in the endoplasmic reticulum. J. Biol. Chem. 270:14958. 49. Tolchinsky, S., M. H. Yuk, M. Ayalon, and H. F. Lodish. 1996. Membrane-bound versus secreted forms of human asialoprotein receptor subunits: role of a juxtamembrane pentapeptide. J. Biol. Chem. 268:14496. 50. Mullins, C., Y. Lu, A. Campbell, H. Fang, and N. Green. 1995. A mutation affecting signal peptidase inhibits degradation of an abnormal membrane protein in Saccharomyces cerevisiae. J. Biol. Chem. 270:17139. 51. Cerundolo, V., T. Elliott, J. Elvin, J. Bastin, H. G. Rammensee, and A. Townsend. 1991. The binding affinity and dissociation rates of peptides for class I major histocompatibility complex molecules. Eur. J. Immunol. 21:2069. 52. Wood, P., and T. Elliott. 1998. Glycan-regulated antigen processing of a protein in the endoplasmic reticulum can uncover cryptic cytotoxic T cell epitopes. J. Exp. Med. 188:773. 53. Rock, K. L., and A. L. Goldberg. 1999. Degradation of cell proteins and the generation of MHC class I-presented peptides. Annu. Rev. Immunol. 17:739. 54. Lemberg, M. K., F. A. Bland, A. Weihofen, V. M. Braud, and B. Martoglio. 2001. Intramembrane proteolysis of signal peptides: an essential step in the generation of HLA-E epitopes. J. Immunol. 167:6441. 55. Snyder, H. L., I. Bacik, J. R. Bennink, G. Kearns, T. W. Behrens, T. Bachi, M. Orlowski, and J. W. Yewdell. 1997. Two novel routes of TAP-independent MHC class I antigen processing. J. Exp. Med. 186:1087. 56. Tsai, B., C. Rodighiero, W. I. Lencer, and T. A. Rapoport. 2001. Protein disulfide isomerase acts as a redox-dependent chaperone to unfold cholera toxin. Cell 104:937. 57. Gallimore, A., K. Schwarz, M. van den Broek, H. Hengarter, and M. Groettrup. 1998. The proteasome inhibitor lactacystin prevents the generation of an endoplasmic reticulum leader-derived T cell epitope. Mol. Immunol. 35:581. 58. Hombach, J., H. Pircher, S. Tonegawa, and R. M. Zinkernagel. 1995. Strictly TAP-dependent presentation of an immunodominant CTL epitope in the signal sequence of a virus protein. J. Exp. Med. 182:1615. Downloaded from http://www.jimmunol.org/ by guest on February 11, 2017 32. Garten, W., S. Hallenberger, D. Ortman, W. Shafer, M. Vey, H. Angliker, E. Shaw, and H. D. Klenk. 1994. Processing of viral glycoproteins by the subtilisin-like endoprotease furin and its inhibition by specific peptidylchloroalkylketones. Biochimie 76:217. 33. Betancourt, M. R., and D. Thirumalai. 1999. Pair potentials for protein folding: choice of reference states and sensitivity of predicted native states to variations in the interaction schemes. Protein Sci. 8:361. 34. Nielsen, H., J. Engelbrecht, S. Brunak, and G. von Heijne. 1997. Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites. Protein Eng. 10:1. 35. Frigerio, L., A. Vitale, J. M. Lord, A. Ceriotti, and L. M. Roberts. 1998. Free ricin A chain, proricin and native toxin have different cellular fates when expressed in tobacco protoplasts. J. Biol. Chem. 273:14194. 36. Carlos, J. L., M. Paetzel, G. Brubaker, A. Karla, C. M. Ashwell, M. O. Lively, G. Cao, P. Bullinger, and R. E. Dalby. 2000. The role of the membrane spanning domain of type I signal peptidases in substrate cleavage site selection. J. Biol. Chem. 275:38813. 37. Lord, J. M., and L. M. Roberts. 1998. Toxin entry: retrograde transport through the secretory pathway. J. Cell Biol. 140:733. 38. den Haan, J. M. M., and M. J. Bevan. 2001. Antigen presentation to CD8⫹ T cells: cross priming in infectious diseases. Curr. Opin. Immunol. 13:437. 39. Cerundolo, V., A. Benham, V. Braud, S. Mukherjee, K. Gould, B. Macino, J. Neefjes, and A. Townsend. 1997. The proteasome-specific inhibitor lactacystin blocks presentation of cytotoxic T lymphocyte epitopes in human and murine cells. Eur. J. Immunol. 27:336. 40. Luckey, C., G. King, H. Marto, S. Venketeswaran, B. Maier, V. Crotzer, T. Colella, J. Shabanowitz, D. Hunt, and V. Engelhard. 1998. Proteasomes can either generate or destroy MHC class I epitopes: evidence for nonproteasomal epitope generation in the cytosol. J. Immunol. 161:112. 41. Glas, R., M. Bogyo, J. S. McMcMaster, M. Gaczynska, and H. L. Ploegh. 1998. A proteolytic system that compensates for loss of proteasome function. Nature 392:618. 42. Knittler, M. R., K. Gulow, A. Seelig, and J. C. Howard. 1998. MHC class I molecules compete in the endoplasmic reticulum for access to transporter associated with antigen processing. J. Immunol. 161:5967. 43. Wei, M. L., and P. Cresswell. 1992. HLA-A2 molecules in an antigen processing mutant cell contain signal sequence-derived peptides. Nature 356:443. 44. Elliott, T., A. Willis, V. Cerundolo, and A. Townsend. 1995. Processing of MHC class I-restricted antigens in the endoplasmic reticulum. J. Exp. Med. 181:1481. 107