Cytosol Is Mediated by Two Peptidases Class I-Restricted Antigenic ...

4 downloads 10348 Views 315KB Size Report
Receive free email-alerts when new articles cite this article. Sign up at: Print ISSN: ... and the nucleus of eukaryotic cells, which degrades the bulk of intracellular ...
The Journal of Immunology

The Final N-Terminal Trimming of a Subaminoterminal Proline-Containing HLA Class I-Restricted Antigenic Peptide in the Cytosol Is Mediated by Two Peptidases1 Fre´de´ric Le´vy,2,3* Lena Burri,2* Sandra Morel,4† Anne-Lise Peitrequin,* Nicole Le´vy,‡ Angela Bachi,§ Ulf Hellman,¶ Benoıˆt J. Van den Eynde,† and Catherine Servis‡ The proteasome produces MHC class I-restricted antigenic peptides carrying N-terminal extensions, which are trimmed by other peptidases in the cytosol or within the endoplasmic reticulum. In this study, we show that the N-terminal editing of an antigenic peptide with a predicted low TAP affinity can occur in the cytosol. Using proteomics, we identified two cytosolic peptidases, tripeptidyl peptidase II and puromycin-sensitive aminopeptidase, that trimmed the N-terminal extensions of the precursors produced by the proteasome, and led to a transient enrichment of the final antigenic peptide. These peptidases acted either sequentially or redundantly, depending on the extension remaining at the N terminus of the peptides released from the proteasome. Inhibition of these peptidases abolished the CTL-mediated recognition of Ag-expressing cells. Although we observed some proteolytic activity in fractions enriched in endoplasmic reticulum, it could not compensate for the loss of tripeptidyl peptidase II/puromycin-sensitive aminopeptidase activities. The Journal of Immunology, 2002, 169: 4161– 4171.

L

igands of the MHC class I molecules are composed of peptides of 8 –10 aa in length. Normally, these peptides are derived from the pool of intracellular polypeptides translated in the various cell types. During viral infection or in tumor cells, the pool of MHC class I-restricted peptides also includes those derived from proteins encoded by virus or tumorspecific genes. Those peptides represent a crucial component of the specific recognition and lysis of the abnormal cells by CTL. An essential step in the production of the MHC class I-restricted peptides is the degradation of proteins by the proteasome (1). The proteasome is a large multicatalytic protease, present in the cytosol and the nucleus of eukaryotic cells, which degrades the bulk of intracellular proteins and generates peptides ranging from 3 to 22 aa in length (2). Whereas some antigenic peptides are directly produced by the proteasome in their final form, others are produced as precursor peptides (3, 4). Those precursor peptides display the exact C terminus of the final antigenic peptides, but carry N-terminal amino acid extensions of various lengths. It is therefore assumed

*Ludwig Institute for Cancer Research, Lausanne Branch, and †Institute of Biochemistry, University of Lausanne, Epalinges, Switzerland; ‡Ludwig Institute for Cancer Research, Brussels Branch, Universite´ Catholique de Louvain, Brussels, Belgium; § DIBIT, San Raffaele Scientific Institute, Milan, Italy; and ¶Ludwig Institute for Cancer Research, Uppsala Branch, Biomedical Center, Uppsala, Sweden Received for publication May 1, 2002. Accepted for publication August 7, 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.

that other peptidases, in the endoplasmic reticulum (ER)5 or in the cytosol, are involved in the N-terminal trimming of precursor peptides (5). Although ER resident proteases have been inferred to function in the Ag-processing pathway, none have yet been described at the molecular level (6 –12). In contrast, several cytosolic peptidases potentially involved in this process have been identified. Those include puromycin-sensitive aminopeptidase, bleomycin hydrolase, thimet oligopeptidase, and leucyl aminopeptidase (13–15). A common feature of these peptidases is that they display a broad specificity and do not lead to an enrichment of the exact antigenic peptide. In some instances, the generation of the accurate N terminus of antigenic peptides can be mediated by more than one peptidase in a redundant manner (13). The human gene RU1 codes for a ubiquitously expressed intracellular protein of unknown function. A CTL clone, originally raised against an autologous renal carcinoma, recognized an HLAB51-restricted peptide derived from the RU1 region spanning aa 34 – 42 (RU134 – 42), with sequence VPYGSFKHV (16). In accordance with the predicted HLA-B51-specific anchor motifs, this peptide contained a proline (P) at position 2 and valine (V) at position 9 (17). In vitro studies using precursor peptides encompassing the antigenic region demonstrated that the exact C terminus of this CTL-defined epitope was directly produced by the standard proteasome, but not the final N terminus. Consequently, other peptidases could act on the N-terminal extension to produce the antigenic nonamer. Based on several in vitro studies indicating that antigenic peptides containing a subaminoterminal P are poor substrates for the human TAP transporters (7, 18 –20), it has been inferred that such peptides are transported through the TAP as

1 This work was supported in part by a grant of the Swiss National Fund (to F.L.), by an Investigator Award of the Cancer Research Institute (to F.L.), and by a grant of the Swiss Cancer League (to L.B.). 2

F.L. and L.B. contributed equally to this study.

3

Address correspondence and reprint requests to Dr. Fre´de´ric Le´vy, Ludwig Institute for Cancer Research, Ch. des Boveresses 155, CH-1066 Epalinges, Switzerland. E-mail address: [email protected] 4 Current address: Institute of Experimental Immunology, Department of Pathology, University of Zu¨rich, CH-8091 Zurich, Switzerland.

Copyright © 2002 by The American Association of Immunologists, Inc.

5 Abbreviations used in this paper: ER, endoplasmic reticulum; AMC, 7-amido-4methylcoumarin; CMK, chloromethylketone; 2-D, two-dimensional; EBNA, EBVencoded nuclear Ag; EGFP, enhanced GFP; GFP, green fluorescent protein; HA, hemagglutinin; MALDI-TOF, matrix-assisted laser desorption ionization-time of flight; MS, mass spectrometry; PSA, puromycin-sensitive aminopeptidase; RCC, renal cell carcinoma; RT, room temperature; TFA, trifluoroacetic acid; TPP II, tripeptidyl peptidase II; Ub, ubiquitin; VSV, vesicular stomatitis virus.

0022-1767/02/$02.00

4162

CYTOSOLIC N-TERMINAL EDITING OF ANTIGENIC PEPTIDE PRECURSORS

N-terminal extended precursors and that the final N terminus is generated by peptidases localized within the ER (7, 21). Contrary to this prediction, our results indicate that the antigenic peptide RU134 – 42, which contains P at second position, can be produced in the cytosol, before TAP-mediated transport. By using a substrate-based assay, we identified two cytosolic peptidases that trim the N terminus of the RU134 – 42 precursors produced by the proteasome. These two peptidases, tripeptidyl peptidase II (TPP II) and puromycin-sensitive aminopeptidase (PSA), act on the N-terminally extended precursors to produce, and transiently enrich for, the exact N terminus of the antigenic peptide. We also searched for proteolytic activities in membranes enriched in ER. Although we observed a detectable proteolytic activity against one of the proteasomal products, RU131– 42, it could apparently not rescue the loss of CTL recognition of tumor cells resulting from the inhibition of proteasome and TPP II/PSA activities. Our data suggest that the production of RU134 – 42/HLA-B51 is a cytosolic process, involving the proteasome to generate the exact C terminus of the antigenic peptide and TPP II and PSA to trim the N-terminal extensions produced by the proteasome.

Materials and Methods Protein purification For the identification of TPP II, BB64-renal cell carcinoma (RCC) (6.3 ⫻ 108 cells) were mechanically disrupted by douncing in a Dounce homogenator. Sucrose was immediately added to the homogenate to a final concentration of 250 mM. Debris were removed by centrifugation at 13,000 ⫻ g for 15 min at 4°C. The supernatant was transferred to ultracentrifuge tubes and subjected to ultracentrifugation at 80,000 ⫻ g for 45 min at 4°C. Proteasomes were removed from the clear supernatant (complete cytosol) by affinity purification, using the mAb anti-proteasome MCP21 (22). Proteasome-depleted cytosol was loaded onto a high performance ion-exchange Source 15Q PE 4.6/100 Sepharose column (Amersham Pharmacia Biotech, Piscataway, NJ) at a flow rate of 1 ml/min in buffer A (20 mM Tris-HCl, pH 7.6), washed with five-column volumes of buffer A, and pre-eluted with 30% buffer B (20 mM Tris-HCl, pH 7.6, 1 M NaCl). The adsorbed material was then eluted, using a linear gradient of 30 –70% buffer B, in 35 fractions of 1 ml each. An aliquot of each fraction (50 ␮l) was incubated with 4 nmol peptide RU129 – 47 (TGSTAVPYGSFKH VDTRLQ, in one-letter code, in which the underlined sequence corresponds to the final CTL-defined antigenic peptide) for 1 h at 37°C. The reaction was stopped by adding trifluoroacetic acid (TFA) to a final concentration of 2%. The samples were then analyzed by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (MS), as previously described (22). The positive fractions (fractions 10 – 14) were pooled, concentrated to a final volume of 500 ␮l, and precipitated with 20% TCA overnight at 4°C. The precipitated proteins were pelleted at 13,000 ⫻ g for 10 min at 4°C, washed three times with 100% cold (⫺20°C) acetone, and lyophilized. The proteins were then subjected to two-dimensional gel electrophoresis (2-D gel). For the identification of PSA, proteasome-depleted cytosol from 100 ml concentrated human erythrocytes was loaded onto a DEAE-32 column, pre-equilibrated in 10 mM phosphate buffer, pH 7.5. The column was extensively washed, and the adsorbed material was eluted in 30 ⫻ 1-ml fractions with 200 mM phosphate buffer, pH 7.5. The eluted material was then desalted on a Sephacryl S-300 column, in 20 mM Tris-HCl, pH 7.6, and loaded onto a Source 15Q PE 4.6/100 column, at a flow of 1 ml/min. After extensive washing with buffer A, the adsorbed material was eluted in 60 ⫻ 1-ml fractions, using a linear gradient of 0 – 65% buffer B. An aliquot (20 ␮l) of each fraction was tested with 100 ␮M H-Ala-Ala-Phe-amidomethylcoumarin (AAF-AMC) for 15 min at 37°C. Proteolytic activity was monitored by the increased fluorescence resulting from the release of the AMC group (excitation/emission 380/440 nm). A total of 4 nmol peptide RU132– 47 was incubated with 12 ␮l fractions 29, 31, and 32, respectively, for 20 min at 37°C. The digestion was stopped with 2% TFA and analyzed by MS as above. Fraction 30 was precipitated with 20% TCA and processed as above. After lyophilization, the protein pellet was resuspended in SDS-sample buffer containing DTT and boiled for 3 min at 95°C. After cooling, iodoacetamide was added to the samples before separation by SDS-12% PAGE. The gel was stained with 0.2% Coomassie brilliant blue R in 20% methanol/0.5% acetic acid and destained with 20% methanol, and the visible bands were excised and treated, as described below.

TPP II was purified from 5 ⫻ 107 HEK293 cells by affinity purification, using 5 ␮l polyclonal chicken Ab anti-human TPP II (Immunsystem, Uppsala, Sweden), 5 ␮g anti-chicken biotin conjugate Ab (Promega, Madison, WI), and 20 ␮l streptavidin-coated agarose beads (Pierce, Rockford, IL). The plasmid PSA-vesicular stomatitis virus (VSV), directing the synthesis of PSA carrying a C-terminal Ab epitope from the vesicular stomatitis virus (a generous gift of A. Fontana, University Hospital, Zu¨ rich, Switzerland), was transiently transfected into 4 ⫻ 107 HEK293 cells using Fugene (Roche, Basel, Switzerland) and the manufacturer-supplied protocol. Twenty-four hours posttransfection, cells were lysed in 1% Triton X-100, and PSA-VSV was immunoprecipitated using 5 ␮g anti-VSV tag mAb (Fluka, Buchs, Switzerland) and 20 ␮l protein G-Sepharose slurry (Pierce). Due to the limited number of cells used in this assay, the final purity of the isolated peptidases could not be ascertained. However, the specificities of both Abs have been described by others (23, 24), and the proteolytic activities of the TPP II and PSA preparations could be completely blocked by butabindide and puromycin, respectively (data not shown). The immunoprecipitated material was washed four times and was used to digest 4 nmol peptides of interest. At the end of the digestion, the supernatant was collected and the digestion products were analyzed by MS.

Preparative 2-D gel electrophoresis The protein pellets obtained from the above-mentioned fractions 10 –14 were resuspended in 600 ␮l loading buffer (40 mM Tris-HCl, pH 8.0, 8 M urea, 4% CHAPS (3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate), 65 mM dithioerythritol, 0.01% bromphenol blue), loaded onto 18-cm-long nonlinear, pH 3–10 gradient strips (Amersham Pharmacia Biotech, Piscataway, NJ), and separated overnight by electrophoresis. During the initial 3 h, the voltage was linearly increased from 300 to 3500 V, followed by 3 h at 3500 V, to reach the final voltage of 5000 V. After separation in the first dimension, the strips were equilibrated in 50 mM Tris-HCl, pH 8.4, 6 M urea, 30% glycerol, 2% SDS, and 2% dithioerythritol for 12 min. Thiol groups were subsequently blocked with 2.5% iodoacetamide. Separation in the second dimension was conducted using a vertical gradient slab gel with a modified Laemmli-SDS discontinuous system (10% acrylamide-piperazine diacrylyl gel) and run at 200 V for ⬃5 h. Gels were silver stained according to a protocol compatible with MS (25).

Destaining, in-gel protein digestion, extraction, and purification Each visible spot of the 2-D gel was cut and lyophilized in a sterile Eppendorf tube. The silver stain was removed by covering the gel piece with 30 mM K-ferricyanide and 100 mM Na-thiosulfate (1:1, v/v), shaking for some minutes, and observing the destaining (26). Each gel piece was then washed with water three times, covered with 0.2 M ammonium bicarbonate, and incubated for at least 20 min at room temperature (RT). The ammonium bicarbonate was then removed, replaced with 100% acetonitrile, and washed three times. Each gel piece was dried, and the digestion was started by adding trypsin (Promega; 0.5 ␮g/gel piece) in 0.2 M ammonium bicarbonate and kept on ice for 15–20 min. More buffer was added in small aliquots to allow a slow uptake of the protease into the gel. The digestion was conducted overnight at 37°C. The reaction was stopped by adding TFA to a final concentration of 1%, and the sample was sonicated for 10 min. The supernatant was saved, 0.1% TFA/60% acetonitrile was added to cover each gel piece, and the tube was incubated for at least 30 min at 37°C. This extract was combined with the previous supernatant, and extraction was repeated. A final extraction was performed, using 100% acetonitrile. All extracts were then combined, and the volume was reduced by vacuum centrifugation to about one-third of the original volume. The samples were desalted by passing them though Zip-Tips (C18; Millipore, Bedford, MA). The samples were lyophilized and resuspended in 3 ␮l 0.1% TFA/H20. One microliter of the sample was mixed (1:1, v/v) with matrix (a saturated solution of ␣-cyano-4-hydroxycinnamic acid in 50% acetonitrile/0.1% TFA) and spotted on the target of the mass spectrometer. Treatment of the bands excised from the Coomassie-stained gel was identical, except that the K-ferricyanide/Na-thiosulfate treatment was omitted.

Synthetic peptide digestion assays A total of 4 nmol peptide was used in digestion assays with cytosol, its derived fractions, and purified peptidases. The digestions were allowed to proceed for the time indicated in the figures, and the reactions were stopped by adding 2% TFA. After lyophilization, the samples were analyzed by MS, as previously described (22). In digestion performed in the presence of peptidase inhibitors, the inhibitors were added to the cytosol and incubated for 15 min at RT prior to the addition of peptide. The concentration of butabindide (a kind gift of J.-C. Schwartz, Institut National de la Sante´ et

The Journal of Immunology de la Recherche Me´ dicale Unite´ 109, Paris, France) was 5 ␮M, AAFchloromethylketone (CMK) 50 ␮M, and puromycin 50 ␮M.

MALDI-TOF mass-spectroscopic analysis and database search For protein identification, all analyses were performed using a Perseptive Biosystems MALDI-TOF Voyager DE-RP or a Voyager DE-STR mass spectrometer (Framingham, MA) operated in the delayed extraction and reflector mode. The search program ProFound, developed by The Rockefeller Mass Spectrometry Laboratory and New York University (New York, NY), was used for database searches (27). Peptides were selected in the mass range of 800-4000 Da. Spectra were calibrated using a matrix and tryptic autodigestion ion peaks as internal standards. For regular peptide digestion assays, the settings of the instrument were as reported (22).

Cell fractionation BB64-RCC (2 ⫻ 108 cells) were mechanically disrupted in hypotonic buffer, and the membranes were immediately equilibrated with 250 mM sucrose. Cell debris were pelleted at 13,000 ⫻ g for 10 min. The supernatant was subjected to high speed centrifugation (80,000 ⫻ g) for 45 min at 4°C. Incomplete cytosol (6 ml) was obtained by the removal of the proteasome from the clear supernatant, using immobilized anti-proteasome Ab MCP21. Membranes were resuspended in 2 ml RM buffer (50 mM HEPES, pH 7.2, 250 mM sucrose, 50 mM potassium acetate, 2 mM magnesium acetate) and pelleted at 80,000 ⫻ g for 45 min at 4°C. After repeating this procedure twice, the membranes were resuspended in 1 ml RM buffer containing 2% Triton X-114, maintained on ice for 10 min, and subjected to phase separation (28). This was performed by placing the samples at 37°C for 10 min and by subsequent centrifugation at 12,000 ⫻ g for 10 min at RT. The upper phase (fraction A), corresponding to the detergent-poor fraction and containing hydrophilic luminal proteins, was transferred into a new tube, and the remaining detergent-rich lower phase (fraction B), containing membrane-anchored proteins, was washed by adding fresh RM buffer lacking Triton X-114. The tube was placed on ice for 10 min and the phase separation was repeated. After two washing cycles, the volume of the detergent-rich phase was readjusted to the original volume (1 ml). Twenty microliters of incomplete cytosol, 4 ␮l fraction A, and 4 ␮l fraction B, which corresponds to the material obtained from an equivalent number of cells, were incubated with 4 nmol RU1 peptides for 20 min at 37°C. The reaction was stopped by adding 2% TFA, and the samples were analyzed by MS, as described above. Forty microliters of incomplete cytosol, 8 ␮l phase A, and 8 ␮l phase B were incubated with 100 ␮M AAF-AMC for 10 min at 37°C. Where indicated, the peptidase inhibitors butabindide (5 ␮M) and puromycin (50 ␮M) were added 15 min before the addition of the fluorogenic substrate.

Western blot analysis Western blot analysis was performed according to standard procedures. A total of 20 ␮l cytosol, 4 ␮l fraction A, and 4 ␮l fraction B was separated on SDS-8% PAGE for the detection of TPP II and SDS-10% PAGE for the detection of PSA, and blotted onto nitrocellulose. The polyclonal chicken anti-human TPP II Ab (Immunsystem), anti-chicken biotin conjugate Ab (Promega), and the streptavidin-HRP conjugate (Invitrogen, San Diego, CA) were used at a 1/1000 dilution to reveal TPP II. For the detection of PSA, a goat Ab anti-PSA (1:1000) Ab and a second Ab anti-goat peroxidase conjugate (1:3000) (Fluka) were used. The protein signals were revealed with ECL Western blotting detection reagent (Amersham Pharmacia Biotech).

Cloning of RU1 minigenes The pEGFP-Ub vector is similar to the one described previously (22), except that the green fluorescent protein (GFP) moiety has been replaced by enhanced GFP (EGFP). Details are available upon request. The cDNA fragments corresponding to the minimal HLA-B*5101-restricted epitope RU134 – 42 and its C-terminally extended version, RU134 – 47, were obtained by annealing complementary oligonucleotides encoding the two RU1-derived peptide fragments. The oligonucleotides were designed so as to reconstitute a SacII site at the 5⬘ end and an AvaI site at the 3⬘ end, and included a stop codon immediately upstream of the AvaI site. Two additional codons, specifying two Gly residues, were added at the 5⬘ end of the minigene so as to reconstitute the exact C terminus of the Ub moiety. Upon annealing, the fragments were inserted between the SacII/AvaI sites of pEGFP/Ub, resulting in plasmid pEGFP/Ub-RU134 – 42 and pEGFP/UbRU134 – 47. The same approach was used to construct the plasmid pEGFP/ Ub⌬G75,76-RU134 – 42 coding for an uncleavable fusion EGFP-Ub-RU134 – 42, except that the codons specifying the Gly at the 3⬘ end of Ub were omitted. The

4163 EGFP-Ub plasmid control coded for EGFP-Ub-Melan-AMART1, in which Melan-AMART1 is a melanoma-associated protein of 118 aa.

Cell lines, transient transfections, and metabolic labeling The human renal carcinoma BB64-RCC and the human embryonic kidney HEK293 and HEK293EBNA (EBV-encoded nuclear Ag) cell lines were maintained in DMEM medium (Invitrogen), supplemented with 10% FCS, antibiotics, and 20 mM Na HEPES, pH 7.3. Transient transfections of the HEK293-EBNA cells were performed using the Lipofectamine reagent (Invitrogen). Fifty thousand cells were plated into flat-bottom microwells and transfected with 1.5 ␮l Lipofectamine, 20 ng plasmid pcDNA3 containing the HLA-B*5101 cDNA, and 12.5 ng of either pcDNA3.1TOPO plasmid (Invitrogen) containing the RU1 full-length cDNA, pEGFP/Ub plasmid containing the cDNA-encoding RU134 – 42, or pEGFP/Ub-encoding RU134 – 47. A total of 100 ng pBJ1neo construct encoding the herpes simplex-derived TAP inhibitor ICP47 molecule (a kind gift of H. G. Rammensee, Tu¨ bingen, Germany) was added to this mix in one-half of the wells. After 20 h, the transfected cells were incubated with 104 anti-RU134 – 42/HLA-B51 CTL 381/84, along with 25 U/ml IL-2. The amount of TNF-␣ secreted in the supernatant was assessed 24 h later by ELISA (Endogen, Woburn, MA). For metabolic labeling, 2 ⫻ 106 HEK293 cells were transfected with 4 ␮g plasmid using the Fugene reagent (Roche) and following the manufacturer’s protocol. Sixteen hours posttransfection, the cells were starved for 45 min in Met/Cys-free DMEM medium (ICN Biomedicals, Aurora, OH) at 37°C. The cells were metabolically labeled for 20 min at 37°C in fresh Met/Cys-free medium containing 150 ␮Ci [35S]Met/Cys (Pro-mix; Amersham Pharmacia Biotech). Cells were washed once and lysed in 1 ml lysis buffer containing 1% Triton X-100 and 30 mM iodoacetamide to prevent postlysis deubiquitylation. Unsolubilized material was removed by centrifugation, and the supernatant was incubated with saturating amounts of the mAb anti-hemagglutinin (HA) epitope (Berkeley Antibody, Richmond, CA) and 20 ␮l protein G-Sepharose. The immunoprecipitates were washed and treated, as described earlier (22), subjected to SDS-12% PAGE, followed by autoradiography.

Treatment of target cells with inhibitors of Ag processing BB64-RCC cells were acid treated as follows to discard peptides from the surface HLA molecules. One million cells were incubated for exactly 30 s at 37°C in 500 ␮l 300 mM glycine buffer (pH 2.5), supplemented with 1% sterile BSA, and then washed several times with culture medium. Eight thousand cells seeded in microplates were incubated for 14 h with 50 ␮M lactacystin (Calbiochem, La Jolla, CA) or 100 ␮M AAF-CMK in culture medium. Thereafter, some of the cells were pulsed for 30 min with the RU1 antigenic peptide VPYGSFKHV at a final concentration of 10 ␮M, washed, and incubated with 3000 CTL 381/84 and 25 U/ml IL-2. The ability of the treated cells to stimulate the CTL was assessed by measuring the production of TNF by the CTL after 20 h of incubation. The TNF (␣ and ␤) content of the supernatants was evaluated by testing their cytotoxic effect on WEHI-164 clone 13 cells (29).

Results TPP II mediates the initial trimming of RU129 – 47 precursors, but fails to produce the final N terminus of the antigenic peptide Earlier work, using the precursor peptide RU124 – 47, demonstrated that the exact C terminus of the antigenic peptide was directly produced by purified standard proteasome, but that the N terminus always carried additional 3 (RU131– 42) and 5 (RU129 – 42) aa (16) (data not shown). This suggested that other peptidases might be necessary to trim the N-terminal extensions to the final nonamer RU134 – 42. We therefore sought to identify the cytosolic peptidases involved in the N-terminal editing of the RU134 – 42 epitope precursors. Because we were not able to determine the relative abundance of each of the two fragments produced by the proteasome due to their coelution from the HPLC columns (data not shown), we investigated the editing of both species. We first focused our attention on the precursor RU129 – 47, which carries an N-terminal extension of 5 aa, and whose C-terminal processing was studied in our previous work (Fig. 1) (16). To identify the peptidase(s) involved in the N-terminal trimming of

4164

CYTOSOLIC N-TERMINAL EDITING OF ANTIGENIC PEPTIDE PRECURSORS

FIGURE 1. Sequence of the RU1 peptides used in this study. The numbers above the sequence refer to the relevant positions within the fulllength protein. The antigenic peptide is underlined, and the filled box corresponds to the constant sequence 34 – 47 shared by the different peptides. The open box corresponds to the sequence 34 – 42. TGST and STA are the two N-terminal extensions produced by the proteasome.

this precursor, we adopted the following strategy: proteasome-depleted cytosol isolated from the renal carcinoma cell line BB64RCC was fractionated by ion-exchange chromatography. The presence of a proteolytic activity was assayed by incubating an aliquot of each fraction with the peptide RU129 – 47 and by subsequent analysis by MS. The 5-aa extension at the C terminus of the antigenic peptide sequence was included in the precursor so as to monitor for the possibility that a peptidase other than the proteasome may directly generate the CTL-defined epitope. No fraction was found to generate the exact N terminus of the antigenic peptide. Rather, fractions 10 –14, eluting at ⬃420 – 460 mM NaCl, contained an activity that resulted in the trimming of the first 3 N-terminal aa (Fig. 2, A and B). No proteolytic activity on the C-terminal extension of the antigenic peptide was detected in any fraction (data not shown). Fractions 10 –14 were pooled and separated by 2-D gel electrophoresis. After silver staining of the gel, visible spots (218) were manually excised, and 77 of those were subjected to in-gel trypsin digestion. The tryptic peptide fragments were extracted from the gel and analyzed by MS. The m.w. values of the extracted peptides were introduced into the program ProFound (http://129.85.19.192/profound_bin/WebProFound.exe) and used for peptide mass fingerprinting. Based on the pattern of its tryptic fragments and its migrating properties in the 2-D gel, one spot was unambiguously identified as TPP II. No other peptidases were identified among the analyzed spots. TPP II is a very large homomultimeric peptidase (molecular mass 5000 –9000 kDa), with subunit molecular mass 138 kDa, that removes tripeptides from the N terminus of peptides and also displays endoproteolytic activities (30, 31). Based on indirect evidence, TPP II has been suggested to participate in the MHC class I Ag-processing pathway, as it may partially compensate for the generation of MHC class I ligands, in situations in which proteasomes are pharmacologically inactivated (32). To confirm that the proteolytic activity producing the fragment RU132– 47 could be ascribed to TPP II, we purified TPP II by immunoadsorption and assayed its activities on the precursor RU129 – 47 (Fig. 2C). After numerous unsuccessful attempts to purify active rTPP II from prokaryotic as well as eukaryotic expression systems, we developed a new purification scheme that yielded active TPP II from 5 ⫻ 107 cells. As demonstrated by the MS analysis, the peak profile of the digested peptide, using immunoadsorbed TPP II, was identical with the one detected after incubation of the same precursor with cytosolic fraction 12 (Fig. 2, compare B and C). Because the final N terminus of the antigenic peptide was not detected, two possibilities were envisaged: either the generation of the final N terminus was sequential and required another peptidase for the removal of the last 2 aa, or

FIGURE 2. Isolation and characterization of TPP II. A, Proteasomedepleted cytosol from the renal carcinoma cell line BB64-RCC was separated by ion-exchange chromatography and eluted, in 1-ml fractions, using a linear gradient from 300 to 650 mM NaCl (dotted line). Protein content was monitored by UV detection at 280 nm, in milliabsorbance units (mAU). Fractions are separated by the vertical narrow line, and the fractions containing the relevant activity are marked by the thick horizontal line (fractions 10 –14). B, An aliquot of fraction 12 was incubated with the precursor peptide RU129 – 47, resulting in the production of a fragment lacking the 3 N-terminal aa (RU132– 47). In each panel, the filled box corresponds to the sequence VPYGSFKHVDTRLQ, as described in Fig. 1. C, Peptide RU129 – 47 was incubated with purified TPP II for 0 or 2 h and subsequently analyzed by MS. After 2 h, a fragment lacking the 3 Nterminal aa was visible. The fragments detected with purified TPP II match those shown in B. Peaks of higher mass than RU129 – 47 correspond to salt adducts.

the N-terminal processing occurred entirely within membranes, which were absent from this preparation. Puromycin-sensitive aminopeptidase generates the exact N terminus of RU134 – 42 To address the first hypothesis, we investigated the possibility that a second cytosolic peptidase might be responsible for the final trimming of the N-terminal extension. To exclude the possibility of contamination from peptidases located in subcellular compartments, proteasome-depleted cytosol isolated from erythrocytes (these cells do not contain internal membranes (33)) was subjected to fractionation. The proteolytic activities of the fractions obtained after separation on Q-Sepharose were tested using the fluorogenic peptide AAF-AMC (Fig. 3A). Two major peaks of activity were detected in fractions 29 –33 and 38 – 45. The activity of the first peak was able to digest the N terminus of RU132– 47, resulting in the enrichment of a fragment corresponding to RU134 – 47, which displays the exact N terminus of the antigenic peptide (Fig. 3A, inset). In light of this result, fraction 30, the proteolytically most active fraction, was precipitated by TCA, separated on SDSPAGE, and stained by Coomassie blue. Each of the 12 visible protein bands was excised from the gel, digested by trypsin, and further processed as above. The tryptic peptides obtained from a protein band with apparent molecular mass 100 kDa led to the identification of the peptidase puromycin-sensitive aminopeptidase (PSA). This peptidase, originally found in brain tissues (see Ref. 34 and references therein), has recently been identified as playing

The Journal of Immunology

4165

FIGURE 3. Isolation and characterization of PSA. A, The proteolytic activity of the fractions obtained after ion-exchange chromatography was tested using 100 ␮M AAF-AMC and monitoring the increased fluorescence, in arbitrary units (A.U.), emitted by the released fluorogenic group AMC (excitation/emission of 380/440 nm). Fluorescence released after incubation of AAF-AMC with unfractionated cytosol (T) or in buffer (⫺) is indicated on the right. Inset, Depicts the proteolytic activity of fraction 31, using RU132– 47 as precursor. This activity leads to the production of peptide RU134 – 47. The filled box is as before. B, Peptides RU132– 47 and RU129 – 47 were incubated with purified PSA-VSV for 0 or 2 h and subsequently analyzed by MS. After 2 h, RU132– 47 was completely converted into a fragment lacking the 2 N-terminal aa (middle panel). This fragment matches the one shown in the inset of A and corresponds to a fragment displaying the exact N terminus of the antigenic peptide. In contrast, RU129 – 47 was barely degraded (right panel). The peak labeled ⫺1 corresponds to a fragment lacking 1 N-terminal aa. See Materials and Methods for details.

a role in the N-terminal trimming of another antigenic peptide precursor (13). To ascertain the role of PSA in the final N-terminal trimming of RU134 – 42, we transfected and immunoadsorbed VSVtagged PSA in the human embryonic kidney cell line HEK293. Using the same technique developed for the purification of active TPP II, immunoadsorbed and proteolytically active PSA was used to digest the precursor peptide RU132– 47. Analysis of the digested material by MS revealed that, after 2-h incubation at 37°C, the peak corresponding to the original peptide RU132– 47 had completely disappeared and a single peak corresponding to peptide RU134 – 47 could be detected (Fig. 3B). This fragment corresponds to a species displaying the exact N terminus of the antigenic peptide (Fig. 1). In contrast, the precursor peptide RU129 – 47 was barely degraded by purified PSA during the same time frame (Fig. 3B). The fractions obtained after separation on Q-Sepharose were also incubated with RU129 – 47, and the proteolytic activity detected in fractions 38 – 45 was compatible with the one of TPP II. However, we did not manage to unambiguously identify the peptidase responsible for this activity. We concluded from these in vitro experiments that PSA trims the N-terminal extension of RU132– 47 to yield the exact N terminus of the antigenic peptide. Although PSA is a peptidase with broad specificity (34), it is noteworthy that no fragment shorter than RU134 – 47 could be detected, indicating that PSA cannot completely degrade the peptide precursor, thereby leading to an enrichment of a peptide with the exact N terminus.

precursor that already displayed the exact C terminus of the antigenic peptide. Using the same assay as described above, we incubated the precursor peptide either with immunoadsorbed TPP II (Fig. 4A) or PSA (Fig. 4B). Analysis of the digested products by MS revealed that both peptidases were capable of trimming the N-terminal extension of the precursor peptide to a fragment corresponding to the exact antigenic peptide. The dichotomy between the processing of RU129 – 47 and RU131– 42 led us to conclude that the sequential trimming of the first one by TPP II and PSA may be caused by the presence of particular amino acids that resist cleavage by PSA (see Discussion). The N-terminal trimming of RU134 – 42 precursors is sensitive to specific TPP II and PSA inhibitors To determine the contribution of the two identified cytosolic peptidases to the N-terminal trimming of RU134 – 42 precursors in unfractionated cytosol, we performed a series of digestions of the

TPP II and PSA can both edit the N terminus of peptide RU131– 42 to its final size Because neither TPP II nor PSA alone could efficiently trim RU129 – 47 to the exact N terminus of the CTL-defined epitope, we tested whether this was also the case for the second N-terminally extended precursor produced by the proteasome, RU131– 42 (Fig. 1). Contrary to the first RU1 precursor, we used in this study a

FIGURE 4. Both TPP II and PSA can trim RU131– 42. A, Purified human TPP II was incubated with peptide RU131– 42 for 2 h at 37°C, and the digested material was analyzed by MS. The open box corresponds to the sequence VPYGSFKHV, displaying the final N and C terminus of the antigenic peptide. B, Same as A, except that the precursor peptide was digested with purified PSA-VSV.

4166

CYTOSOLIC N-TERMINAL EDITING OF ANTIGENIC PEPTIDE PRECURSORS

various peptide precursors in the presence or absence of specific TPP II and PSA inhibitors (Fig. 5). The peptides RU129 – 47, RU131– 42, and RU132– 47 (Fig. 1) were incubated with proteasomedepleted cytosol for 10 min at 37°C (Fig. 5, A–C). The reaction was stopped by the addition of 2% TFA and analyzed by MS. The same digestions were also performed in the presence of butabindide, a specific TPP II inhibitor (Fig. 5, D–F) (35); AAF-CMK, an inhibitor of TPP II, PSA, and bleomycin hydrolase (G–I) (13, 30); and puromycin, a specific PSA inhibitor (J–L) (34). In the case of RU129 – 47, a fragment displaying the final N terminus of the antigenic peptide was readily detectable after incubation with untreated cytosol (Fig. 5A). No such fragment could be detected when the peptide was incubated with cytosol pretreated with butabindide or AAF-CMK (Fig. 5, D and G, respectively). Interestingly, not only was the fragment with the exact N terminus absent, but the intermediate corresponding to RU132– 47 was not detectable either. The result obtained after treatment with butabindide confirms and extends the findings shown in Figs. 2 and 3, namely that the processing of the N-terminal extension of RU129 – 47 in unfractionated cytosol is a sequential process that requires the activity of TPP II. Finally, puromycin-treated cytosol led to an accumulation of the intermediate RU132– 47 by blocking the second proteolytic event that normally leads to the generation of the final N terminus (Fig. 5J). Peptide RU132– 47, which corresponds to the fragment produced after the digestion of the longer precursor by TPP II, was also incubated with cytosol. As expected, the fragment displaying the

FIGURE 5. Effects of TPP II and PSA inhibitors on the processing of RU1 precursor peptides. Peptides RU129 – 47 (A, D, G, and J), RU131– 42 (B, E, H, and K), and RU132– 47 (C, F, I, and L) were incubated with unfractionated proteasome-depleted cytosol of BB64-RCC for 10 min at 37°C. Reactions were stopped by adding 2% TFA. The samples were lyophilized and analyzed by MS. For peptide RU129 – 47 (A, D, G, and J), only the region of the spectrum encompassing RU132– 47 and RU134 – 47 is shown. A–C, Peptides digested by untreated cytosol. D–F, As A–C, but in presence of 5 ␮M butabindide, a specific inhibitor of TPP II. G–I, As A–C, but in presence of 50 ␮M AAF-CMK, a second inhibitor of TPP II. J–L, As A–C, but in presence of 50 ␮M puromycin, a specific inhibitor of PSA. The stars identify contaminating peaks (already present at time 0, not shown). In B, E, H, and K, the peaks labeled ⫺1 and ⫺2 correspond to fragments lacking 1 and 2 N-terminal aa. See text for details.

exact N terminus of the antigenic peptide was efficiently produced in untreated cytosol (Fig. 5C) and was insensitive to butabindide and only partially sensitive to AAF-CMK (F and I, respectively). However, no digestion was observed when the cytosol was pretreated with puromycin (Fig. 5L), confirming the essential role of PSA in the final trimming of this intermediate. Finally, we analyzed the trimming of the precursor RU131– 42. As already predicted from the result shown in Fig. 4, a fragment corresponding to the final size of the antigenic peptide was produced both in untreated cytosol and in cytosol treated with inhibitors (Fig. 5, B, E, H, and K). We conclude that this precursor can be edited simultaneously either by TPP II or PSA, but the contribution of yet another peptidase on the editing of this particular peptide cannot be ruled out, as peaks corresponding to fragments lacking 1 and 2 N-terminal aa could be detected in all conditions. The N-terminal processing of RU134 – 42 precursors occurs predominantly in the cytosol Several reports have suggested that the N-terminal trimming of antigenic peptides may occur within the ER. Although we presented evidence that the processing of RU134 – 42 precursors could occur in the cytosol (Fig. 5), we nevertheless tested whether the trimming could also take place within ER membranes. We purified membranes from the BB64-RCC cell line, separated integral membrane proteins from soluble luminal proteins by Triton X-114 extraction (see Materials and Methods for details), and tested the proteolytic

The Journal of Immunology

4167

FIGURE 6. The N-terminal trimming of RU134 – 42 precursors occurs mainly in cytosolic extract. BB64-RCC cells were separated into cytosol and membranes. The microsomal fraction was further fractionated so as to separate membrane-associated proteins (fraction B) from soluble luminal proteins (fraction A). An aliquot of each fraction was used to digest RU129 – 47 (A, D, and G), RU132– 47 (B, E, and H), and RU131– 42 (C, F, and I). A–C, Fragments obtained after digestion of the precursors with incomplete cytosol. As in Fig. 6, the stars identify contaminating peaks. D–F, Same as A–C with fraction A, containing soluble luminal proteins. G–I, Same as A–C with fraction B, containing membrane-associated proteins. The peaks ranging from mass 800 to 1000 (I) correspond to the mass of detergent micelles. The filled and open boxes are as before. J, Proteolytic activity of the various fractions was tested using 100 ␮M AAF-AMC and monitoring the increased fluorescence, in arbitrary units (A.U.), emitted by the released fluorogenic group. Where indicated, the fractions were preincubated with 5 ␮M butabindide and 50 ␮M puromycin before addition of the fluorogenic substrate. K, Western blot analysis of the various fractions detecting the presence of TPP II and PSA. See text for details.

activity of each of these fractions using the three peptide precursors. As before, we could clearly detect proteolytic cleavage of all precursors in the cytosolic fraction, resulting in the formation of a fragment displaying the exact N terminus of the antigenic peptide (Fig. 6, A–C). Very little proteolytic activity was detected in the fraction A, containing soluble luminal proteins (Fig. 6, D–F), and no activity was detected in the fraction B, containing membraneembedded proteins (Fig. 6, G–I). Proteolytic activity present in the three fractions was also independently monitored by the release of AMC from the fluorogenic tripeptide AAF-AMC (Fig. 6J). The cytosolic extract was proteolytically active against AAF-AMC, and these activities could be partially blocked either by butabindide or puromycin, indicating that: 1) TPP II and PSA are present in this fraction, and 2) other peptidases are also active against this fluorogenic peptide. The presence of the two peptidases was further confirmed by Western blot analysis (Fig. 6K). Some proteolytic activity could also be detected in fraction A. However, this activity was totally resistant to puromycin (confirming the absence of PSA from this fraction), but was completely blocked by butabindide, indicating the presence of contaminating TPP II in this fraction. This was independently confirmed by Western blot analysis using an anti-TPP II Ab (Fig. 6K). The presence of TPP II could be responsible for the small peak of RU132– 47 and RU132– 42 detected in D and F, respectively. The presence of TPP II in fraction A could be explained by the fact that a fraction of this very large peptidase (5–9 Md) cosediments with microsomal membranes. Indeed, we found that a 60-min centrifugation at 350,000 ⫻ g efficiently sediments most of the TPP II present in the cell lysate (data not shown). A close analysis of the fragments produced from RU131– 42 by the proteolytic activity present in fraction A suggests that at least

one other peptidase can digest this peptide. Indeed, as already detected in Fig. 5, B, E, H, and K, peaks corresponding to fragments missing 1or 2 N-terminal aa could also be observed in digested material obtained from fraction A (Fig. 6F). This proteolytic activity is insensitive to butabindide, puromycin, and AAF-CMK (Fig. 5). It is noteworthy that both other precursors do not seem to be substrate of this peptidase. We do not have any information on this peptidase at the moment. The presentation of RU134 – 42/HLA-B51 is blocked by lactacystin and AAF-CMK In an attempt to correlate our results obtained in an acellular system with the cellular pathway leading to the presentation of RU134 – 42 by HLA-B51, we treated BB64-RCC with specific inhibitors and tested their recognition by CTL. Cells treated with the proteasome inhibitor lactacystin were poorly recognized by specific CTL, as indicated by the low level of TNF produced by the CTL clone (Fig. 7). Recognition was partially restored when the antigenic peptide was added exogenously. In comparison, untreated cells were efficiently recognized by the same CTL clone, confirming that the proteasome played an essential role in the presentation of this CTL-defined epitope. We also tested whether the recognition of RU134 – 42/HLA-B51 was influenced by the presence of the inhibitor AAF-CMK. As with lactacystin, cells treated with AAF-CMK were poorly recognized by specific CTL. Again, exogenously added peptide led to a partial restoration of the recognition of BB64-RCC cells. The lack of recognition resulting from the AAF-CMK treatment could not be ascribed to an inhibition of proteasome because, as tested by us and reported by others, proteasomal activities were not influenced by AAF-CMK (13, 30) (data not shown). In our hands, 100 ␮M lactacystin did not inhibit

4168

CYTOSOLIC N-TERMINAL EDITING OF ANTIGENIC PEPTIDE PRECURSORS

FIGURE 7. Lactacystin and AAF-CMK block the recognition of tumor cells by specific CTL. BB64-RCC cells were treated with the proteasome inhibitor lactacystin and the protease inhibitor AAF-CMK or left untreated. CTL clone 381/84 was added to the cells, and TNF release was measured (open bars). As control, saturating amounts of the antigenic peptide RU134 – 42 were added exogenously (filled bars).

the activity of TPP II (data not shown). Taken together, these results indicate that, in cells, the proteasome most likely does not directly produce the final antigenic peptide RU134 – 42, but that additional peptidases, like TPP II and/or PSA, are necessary to generate the antigenic peptide. We were unable to test the effect of butabindide on the presentation of RU134 – 42 because this drug does not cross the cell membrane (data not shown). Puromycin could not be tested either, due to its excessive cell toxicity (data not shown). Cells transfected with a minigene encoding the minimal antigenic peptide RU134 – 42/HLA-B51 are efficiently recognized by specific CTL The antigenic peptide RU134 – 42 contains a subaminoterminal P (Fig. 1) and is thus predicted to be poorly transported by TAP in its final form. However, we have shown that the production of the final N terminus can occur in the cytosol. To test whether the N-terminally trimmed antigenic peptide can be transported into the ER, the human embryonic kidney cells HEK293 were transiently transfected with the cDNA-encoding HLA-B51 and either the RU1 cDNA or two minigenes coding for, respectively, peptide RU134 – 47, displaying the exact N terminus, but carrying a C-terminal extension, and RU134 – 42, the minimal HLA-B51-restricted epitope (Fig. 1). The vectors used for the expression of the minigenes were designed based on the ubiquitin/protein/reference technique described previously (22, 36). In short, the plasmid codes for the tripartite linear fusion protein EGFP-ubiquitin (Ub) minigene. The EGFP-Ub moiety is cotranslationally cleaved after the last residue of Ub by the cytosolic Ub peptidase, thereby liberating the peptides in the cytosol (37, 38). In addition, this gene arrangement also allows the expression of minigenes without the need of the initiation codon for methionine. As shown in Fig. 8A, cells transfected with the three RU1 constructs were recognized by the specific CTL. It is noteworthy that the cells transfected with the vector encoding the minimal antigenic peptide sequence were more efficiently recognized by the CTL, supporting the notion that the proteasomal processing may limit the efficiency of Ag presentation. More efficient recognition of DNA-encoded RU134 – 42 occurred when cells were transfected with plasmid concentrations ranging from 1.25 to 50 ng (data not shown). To confirm that the antigenic peptide was transported by the TAP complex in our experimental conditions, the gene encoding the natural TAP inhibitor ICP47 from herpes simplex was cotransfected with the plasmids described above (39). As expected, these cells were not recognized

FIGURE 8. Peptide RU134 – 42 is transported by the TAP complex. A, HEK293-EBNA cells were transfected with a plasmid coding for HLAB51 and constructs containing the RU1 full-length cDNA or minigenes coding for the CTL-defined epitope RU134 – 42 with a C-terminal extension (pEGFP/Ub-RU134 – 47) or the minimal epitope RU134 – 42 (pEGFP/UbRU134 – 42). The peptides encoded by the latter two constructs are produced as a result of the proteolytic cleavage after the last residue of Ub, a system that allows to bypass the need for an N-terminal methionine. Some of the cells were also transfected with a plasmid encoding the herpes simplex protein ICP47, which blocks the TAP-mediated transport of antigenic peptides. The presentation of RU134 – 42 by HLA-B51 in absence (open bars) or presence (filled bars) of ICP47 was monitored by measuring the amount of TNF-␣ released by CTL 381/84 incubated for 24 h with the transfected cells. B, HEK293 cells expressing EGFP-Ub-RU134 – 47 and EGFPUb⌬G75,76-RU134 – 47, respectively, were metabolically labeled with 35S and lysed, and the lysate was immunoprecipitated, using an Ab against a peptide sequence located between the C terminus of EGFP and the N terminus of Ub. The precipitated material was separated by SDS-12% PAGE and subjected to autoradiography. As control, EGFP-Ub was immunoprecipitated from radiolabeled cells expressing the 118-aa-long protein Melan-A instead of the RU1 peptides. The arrow indicates the position of EGFP-Ub, and the star indicates unidentified species. See text for details.

anymore, confirming the necessity of functional TAP for the recognition of target cells (Fig. 8A). The slightly reduced inhibition of the recognition of cells expressing the minimal epitope (RU134 – 42) in presence of ICP47 could be explained by the incomplete effects of ICP47 on TAP, a TAP-independent peptide transport, or the fact that a small percentage of the peptide may be released into the extracellular medium and recaptured by HLA-B51 molecules at the cell surface. In other experiments, the effect resulting from the expression of ICP47 was confirmed to be specific, and not due to dilution by the additional plasmid DNA transfected (data not shown). We conclude from these experiments that the antigenic peptide RU134 – 42, carrying P at position 2, can be translocated across the ER membrane in its final form via the TAP transporters. We could exclude that the fusion GFP-Ub could be cleaved within the ER because it lacks an ER-targeting signal sequence and because the Ub-specific peptidase is absent from the ER (38) (F. Le´ vy, unpublished data). We also tested the possibility that a significant amount of uncleaved EGFP-Ub peptide could be present in transfected cells and

The Journal of Immunology could lead to the production of N-terminally extended antigenic peptides. Therefore, we compared, by SDS-PAGE, the mobility of two EGFP-Ub minigene constructs. The first one codes for EGFPUb-RU134 – 47, which is cleaved by Ub-specific peptidase to produce EGFP-Ub and RU134 – 47. The second plasmid encodes EGFP-Ub⌬G75,76-RU134 – 42, which lacks the two C-terminal Gly residues of Ub, thereby producing an uncleavable form of Ub. Indeed, the C-terminal Gly have been shown to be essential for the Ub peptidase-mediated cleavage of Ub (40). Both constructs contained a sequence derived from the influenza A HA between the C terminus of EGFP and the N terminus of Ub, which can be recognized by a specific Ab (22). Cells were transfected with both plasmids, metabolically labeled with [35S]Met/Cys, and lysed in the presence of alkylating agent to prevent postlysis deubiquitylation, and the cleared lysate was immunoprecipitated using the antiHA mAb linked to protein G-Sepharose. The length of the RU1 minigenes was selected so as to maximize the migration differences after separation on SDS-PAGE. As shown in Fig. 8B, the SDS mobility of EGFP-Ub-RU134 – 47 carrying the wild-type Ub moiety was indistinguishable from EGFP-Ub derived from a plasmid encoding a melanoma-associated protein fused to EGFP-Ub. In contrast, EGFP-Ub⌬G75,76-RU134 – 42 migrated with a slightly slower mobility. This result indicated that the cleavage at the Ub minigene junction had occurred in the wild-type construct and produced a polypeptide that differs from the mutant construct by 7 aa. A very faint band with slower mobility could be observed for the nonmutated construct, but its precise identity could not be determined because a band of identical mobility was also visible in the construct carrying the mutated Ub and could be sometimes detected in the EGFP-Ub control plasmid. Even though the vast majority of EGFP-Ub-RU134 – 42 molecules was cleaved, the possibility that a small fraction of EGFP-Ub-RU134 – 42 may not be cleaved by Ub-specific proteases and produces an amino-terminally extended antigenic peptide precursor cannot be completely ruled out. Nevertheless, the data presented in this work support the notion that the final processing of RU134 – 42 can take place in the cytosol.

Discussion In this work, we investigated the N-terminal trimming of the two RU134 – 42 precursor peptides, RU129 – 42 and RU131– 42, liberated by the proteasome in vitro. We show that two cytosolic peptidases can trim the N-terminal extensions of these peptide precursors. The peptidases, TPP II and PSA, acted sequentially on peptide RU129 – 42 and redundantly on peptide RU131– 42. In all cases, this process led to the transient accumulation of a species displaying the final N terminus of the antigenic peptide. Moreover, we showed that proteasome and TPP II/PSA inhibitors blocked the presentation of RU134 – 42 by HLA-B51⫹ tumor cells. Taken together, these data suggest that the production of the epitope RU134 – 42/HLA-B51 may be a cytosolic process involving at least three distinct peptidases, the proteasome, TPP II, and PSA. The N-terminal trimming peptidases identified to date share the properties of having broad specificities, of being redundant, and of trimming antigenic peptide precursors without detectable accumulation of a species with the correct N terminus. We were therefore surprised that the generation and the transient accumulation of the final N terminus of RU134 – 42 from RU129 – 47 required the sequential action of two distinct peptidases. The two identified peptidases, TPP II and PSA, have already been implicated, in different experimental conditions, in the MHC class I Ag-processing pathway. The first one, TPP II, was identified using cells that had been adapted to a proteasome inhibitor (32). This treatment induced the overexpression of TPP II, and indirect evidences, such as cell sur-

4169 face expression of MHC class I and HPLC profile of peptides eluted from MHC class I molecules, suggested that TPP II may compensate for the lack of proteasome activity (32, 41). However, the exact contribution of TPP II to this process remains to be elucidated, as the activities of the proteasome may not be completely blocked in those cells (42). Another study demonstrated that TPP II did not only remove tripeptides from the N terminus of peptides, but also had some endoproteolytic activities, which could potentially produce antigenic peptides (30). Of note is that in this study we did not observe any other endoproteolytic activity of human TPP II than the one removing tripeptides from free N termini (data not shown). However, in none of the two cases mentioned above was TPP II directly identified in the trimming of a specific antigenic peptide precursor. Contrary to the role of TPP II in the production of antigenic peptides, the contribution of PSA to the processing of a CTL-defined epitope was recently revealed by an experimental approach similar to the one described in this work (13). In that approach, PSA was shown to degrade N-terminally extended precursors of the antigenic peptide VSV nuclear protein52–59 and to generate, among many other fragments, a peptide with the final N terminus. Using the precursor peptide RU129 – 47, we showed that the production of the final N terminus of the antigenic peptide was an ordered process, which required the sequential activities of the peptidases TPP II and PSA (Figs. 3, 4, and 6). This was also the case in a cytosolic extract, which contained other active peptidases (Figs. 6 and 7J). A molecular explanation for this resides probably in the specificity of the individual peptidases. It appears that neither TPP II nor PSA can cleave a peptide bond before or after Pro (P). In addition, PSA does not cleave efficiently peptide bonds involving a Gly (G) residue (34). Because the sequence of the precursor peptide is TGSTAVP. . . , in which VP corresponds to the first 2 aa of the CTL-defined epitope (Fig. 1), this may explain why TPP II has to remove the first 3 aa (it cannot cleave further because this would involve the residue P) before PSA, which does not cleave the full-length precursor because of the amino acid G, can trim the last 2 aa. PSA will then stop at the exact N terminus of the antigenic peptide because of the residue P. Such a process may be specific for antigenic peptides carrying the residue P at position 2, as is the case for certain ligands of the HLA-B7, HLA-B8, HLA-B15, HLA-B51, and other class I molecules. However, the fact that the same peptidase (PSA) has been identified using two different peptide precursors may not be coincidental, and leads us to postulate that a limited number of peptidases will be responsible for the final editing of MHC class I ligands. At present, we do not know whether this sequence of action is identical in all cell types nor in situations in which the expression of other peptidases is induced, as is the case for leucyl aminopeptidase induced by IFN-␥ (15). The production of the final antigenic peptide from the N-terminally extended precursor RU131– 42 could be achieved by TPP II and PSA in a redundant fashion. This result is in agreement with the findings reported on the generation of the N-terminal end of VSV nuclear protein52–59, namely that the final N terminus of this antigenic peptide could be obtained by two redundant processes, mediated either by PSA or, in that case, by bleomycin hydrolase (13). The likely explanation for the different processing of RU131– 42 and RU129 – 47 resides in the length and the sequence of the N-terminal amino acid extension. Therefore, the choice of peptidase(s) responsible for the postproteasomal trimming of antigenic peptide precursors may be dictated by the nature of the N-terminal extension produced by the proteasome. We have analyzed the processing of several antigenic peptide precursors in vitro and have noticed that the proteasome can, in some instances, directly generate the exact antigenic peptide (our unpublished data). However,

4170

CYTOSOLIC N-TERMINAL EDITING OF ANTIGENIC PEPTIDE PRECURSORS

this species is frequently accompanied by the presence of peptides that carry additional N-terminal amino acids. It will be interesting to study whether the final antigenic peptide produced within a cell will be more efficiently loaded onto MHC class I molecules than the precursors requiring further N-terminal processing. N-terminal trimming in the ER has also been described (5). However, because no specific peptidase has been identified at the molecular level, it is difficult to speculate on the role of this ER trimming. The fact that most peptides isolated from the HLA-A2 molecules of the TAP-deficient cell line T2 are considerably longer in size than those isolated from normal cells (43) indicates that if aminopeptidases are active in the ER, they may be very specific or play a minor role in the processing of antigenic peptides derived from signal sequences released by the signal peptidase in the ER. In this study, we directly searched for membrane-embedded or luminal peptidases that would mediate the final trimming of our precursor (Fig. 7). Although we could detect a proteolytic activity other than the one mediated by TPP II in our membrane preparation, it was only acting on the precursor RU131– 42 and was also found in the cytosol. Moreover, this activity was not inhibited by AAF-CMK even though this drug efficiently blocked the CTL recognition of APCs, suggesting that this peptidase is not participating to the final processing of RU134 – 42. One reason for this is that this peptidase is localized in another subcellular compartment, which contaminated our microsomal preparation. Alternatively, the product RU131– 42 detected in our in vitro digestion assay may constitute only a minor species that is not produced intracellularly. Irrespective of the role of this peptidase in this process, our data suggest that the antigenic peptide RU134 – 42 is made in the cytosol and can be transported as such across the ER membrane. After prolonged incubation of the various RU134 – 42 precursors in cytosol, the peptides were eventually completely degraded (data not shown). This may be explained by the fact that our in vitro digestion assay does not contain any membrane, which, in cells, can offer a physical barrier against the attack of other cytosolic peptidases. However, the mechanism by which antigenic peptides produced in their final form in the cytosol reach the TAP transporters has not been elucidated. Interestingly, a membrane-associated protein of 100 kDa, p100, has been identified, which binds peptides, possibly transiently, on the cytoplasmic side of the ER membrane (44). The association between this protein and the membrane appears to be mediated by an unknown protein. Because the m.w., pI, and peptide interaction properties of p100 are similar to those of PSA, it is tempting to speculate that p100 could be PSA. If this were the case, it would provide a mean to simultaneously edit the N terminus of longer peptide precursors and increase the chance of having the final antigenic peptide transported by the TAP1/2 complex. Experiments aimed at elucidating the intracellular localization of PSA are underway.

Acknowledgments We thank Dr. A. Fontana for the generous gift of the plasmid coding for human PSA-VSV, Dr. H.-G. Rammensee for the plasmid encoding ICP47, Dr. J.-C. Schwartz for butabindide, Dr. J.-C. Cerottini and L. Chapatte for critical reading of the manuscript, F. Piette for expert technical help, and F. Penea for peptide synthesis.

References 1. 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. 2. Kisselev, A. F., T. N. Akopian, K. M. Woo, and A. L. Goldberg. 1999. The sizes of peptides generated from protein by mammalian 26 and 20 S proteasomes. J. Biol. Chem. 274:3363. 3. Cascio, P., C. Hilton, A. F. Kisselev, K. L. Rock, and A. L. Goldberg. 2001. 26S proteasomes and immunoproteasomes produce mainly N-extended versions of an antigenic peptide. EMBO J. 20:2357.

4. Lucchiari-Hartz, M., P. M. van Endert, G. Lauvau, R. Maier, A. Meyerhans, D. Mann, K. Eichmann, and G. Niedermann. 2000. Cytotoxic T lymphocyte epitopes of HIV-1 Nef: generation of multiple definitive major histocompatibility complex class I ligands by proteasomes. J. Exp. Med. 191:239. 5. Shastri, N., S. Schwab, and T. Serwold. 2002. Producing nature’s gene-chips: the generation of peptides for display by MHC class I molecules. Annu. Rev. Immunol. 20:463. 6. Komlosh, A., F. Momburg, T. Weinschenk, N. Emmerich, H. Schild, E. Nadav, I. Shaked, and Y. Reiss. 2001. A role for a novel luminal endoplasmic reticulum aminopeptidase in final trimming of 26 S proteasome-generated major histocompatibility complex class I antigenic peptides. J. Biol. Chem. 276:30050. 7. Serwold, T., S. Gaw, and N. Shastri. 2001. ER aminopeptidases generate a unique pool of peptides for MHC class I molecules. Nat. Immun. 2:644. 8. Lobigs, M., G. Chelvanayagam, and A. Mu¨ llbacher. 2000. Proteolytic processing of peptides in the lumen of the endoplasmic reticulum for antigen presentation by major histocompatibility class I. Eur. J. Immunol. 30:1496. 9. Paz, P., N. Brouwenstijn, R. Perry, and N. Shastri. 1999. Discrete proteolytic intermediates in the MHC class I antigen processing pathway and MHC I-dependent peptide trimming in the ER. Immunity 11:241. 10. Snyder, H. L., I. Bacik, J. R. Bennink, G. Kearns, T. W. Behrens, T. Ba¨ chi, M. Orlowski, and J. W. Yewdell. 1997. Two novel routes of transporter associated with antigen processing (TAP)-independent major histocompatibility complex class I antigen processing. J. Exp. Med. 186:1087. 11. Craiu, A., T. Akopian, A. Goldberg, and K. L. Rock. 1997. Two distinct proteolytic processes in the generation of a major histocompatibility complex class I-presented peptide. Proc. Natl. Acad. Sci. USA 94:10850. 12. Hammond, S. A., R. P. Johnson, S. A. Kalams, B. D. Walker, M. Takiguchi, J. T. Safrit, R. A. Koup, and R. F. Siliciano. 1995. An epitope-selective, transporter associated with antigen presentation (TAP)-1/2-independent pathway and a more general TAP-1/2-dependent antigen-processing pathway allow recognition of the HIV-1 envelope glycoprotein by CD8⫹ CTL. J. Immunol. 154:6140. 13. Stoltze, L., M. Schirle, G. Schwarz, C. Schro¨ ter, M. W. Thompson, L. B. Hersh, H. Kalbacher, S. Stevanovic, H.-G. Rammensee, and H. Schild. 2000. Two new proteases in the MHC class I processing pathway. Nat. Immun. 1:413. 14. Silva, C. L., F. C. Portaro, V. L. Bonato, A. C. de Camargo, and E. S. Ferro. 1999. Thimet oligopeptidase (EC 3.4.24.15), a novel protein on the route of MHC class I antigen presentation. Biochem. Biophys. Res. Commun. 255:591. 15. Beninga, J., K. L. Rock, and A. L. Goldberg. 1998. Interferon-␥ can stimulate post-proteasomal trimming of the N terminus of an antigenic peptide by inducing leucine aminopeptidase. J. Biol. Chem. 273:18734. 16. Morel, S., F. Le´ vy, O. Burlet-Schiltz, F. Brasseur, M. Probst-Kepper, A.-L. Peitrequin, B. Monsarrat, R. Van Velthoven, J.-C. Cerottini, T. Boon, et al. 2000. Processing of some antigens by the standard proteasome but not by the immunoproteasome results in poor presentation by dendritic cells. Immunity 12:107. 17. Rammensee, H.-G., J. Bachmann, N. P. N. Emmerich, O. A. Bachor, and S. Stevanovic. 1999. SYFPEITHI: database for MHC ligands and peptide motifs. Immunogenetics 50:213. 18. Gubler, B., S. Daniel, E. A. Armandola, J. Hammer, S. Caillat-Zucman, and P. M. van Endert. 1998. Substrate selection by transporters associated with antigen processing occurs during peptide binding to TAP. Mol. Immunol. 35:427. 19. Uebel, S., W. Kraas, S. Kienle, K.-H. Wiesmu¨ ller, G. Jung, and R. Tampe´ . 1997. Recognition principle of the TAP transporter disclosed by combinatorial peptide libraries. Proc. Natl. Acad. Sci. USA 94:8976. 20. Van Endert, P. M., D. Riganelli, G. Greco, K. Fleischhauer, J. Sidney, A. Sette, and J. F. Bach. 1995. The peptide-binding motif for the human transporter associated with antigen processing. J. Exp. Med. 182:1883. 21. Lauvau, G., K. Kakimi, G. Niedermann, M. Ostankovitch, P. Yotnda, H. Firat, F. V. Chisari, and P. M. van Endert. 1999. Human transporters associated with antigen processing (TAPs) select epitope precursor peptides for processing in the endoplasmic reticulum and presentation to T cells. J. Exp. Med. 190:1227. 22. Valmori, D., U. Gileadi, C. Servis, P. R. Dunbar, J.-C. Cerottini, P. Romero, V. Cerundolo, and F. Le´ vy. 1999. Modulation of proteasomal activity required for the generation of a CTL-defined peptide derived from the tumor antigen MAGE-3. J. Exp. Med. 189:895. 23. Tobler, A. R., D. B. Constam, A. Schmitt-Gra¨ ff, U. Malipiero, R. Schlapbach, and A. Fontana. 1997. Cloning of the human puromycin-sensitive aminopeptidase and evidence for expression in neurons. J. Neurochem. 68:889. 24. Tomkinson, B., and F. Nyberg. 1995. Distribution of tripeptidyl-peptidase II in the central nervous system of rat. Neurochem. Res. 20:1443. 25. Shevchenko, A., M. Wilm, O. Vorm, and M. Mann. 1996. Mass spectrometric sequencing of protein silver-stained polyacrylamide gels. Anal. Chem. 68:850. 26. Gharahdaghi, F., C. R. Weinberg, D. A. Meagher, B. S. Imai, and S. M. Mische. 1999. Mass spectrometric identification of proteins from silver-stained polyacrylamide gel: a method for the removal of silver ions to enhance sensitivity. Electrophoresis 20:601. 27. Zhang, W., and B. T. Chait. 2000. ProFound: an expert system for protein identification using mass spectrometric peptide mapping information. Anal. Chem. 72:2482. 28. Bordier, C. 1981. Phase separation of integral membrane proteins in Triton X-114 solution. J. Biol. Chem. 256:1604. 29. Traversari, C., P. van der Bruggen, B. Van den Eynde, P. Hainaut, C. Lemoine, N. Ohta, L. Old, and T. Boon. 1992. Transfection and expression of a gene coding for a human melanoma antigen recognized by autologous cytolytic T lymphocyte. Immunogenetics 35:145. 30. Geier, E., G. Pfeifer, M. Wilm, M. Lucchiari-Hartz, W. Baumeister, K. Eichmann, and G. Niedermann. 1999. A giant protease with potential to substitute for some functions of the proteasome. Science 283:978.

The Journal of Immunology ¨ . Zetterqvist. 1986. Purifi31. Bålo¨ w, R.-M., B. Tomkinson, U. Ragnarsson, and O cation, substrate specificity, and classification of tripeptidyl peptidase II. J. Biol. Chem. 261:2409. 32. Glas, R., M. Bogyo, J. S. McMaster, M. Gaczynka, and H. L. Ploegh. 1998. A proteolytic system that compensates for loss of proteasome function. Nature 392: 618. 33. Lodish, H., D. Baltimore, A. Berk, S. L. Zipursky, P. Matsudaira, and J. Darnell. 1995. Molecular Cell Biology. Scientific American, New York. 34. Barrett, A. J., N. D. Rawlings, and J. F. Woessner. 1998. Handbook of Proteolytic Enzymes. Academic Press, London. 35. Rose, C., F. Vargas, P. Facchinetti, P. Bourgeat, R. B. Bambal, P. B. Bishop, S. M. T. Chan, A. N. J. Moore, C. R. Ganellin, and J.-C. Schwartz. 1996. Characterization and inhibition of a cholecystokinin-inactivating serine peptidase. Nature 380:403. 36. Le´ vy, F., N. Johnsson, T. Ru¨ menapf, and A. Varshavsky. 1996. Using ubiquitin to follow the metabolic fate of a protein. Proc. Natl. Acad. Sci. USA 93:4907. 37. Turner, G. C., and A. Varshavsky. 2000. Detecting and measuring cotranslational protein degradation in vivo. Science 289:2117. 38. Johnsson, N., and A. Varshavsky. 1994. Ubiquitin-assisted dissection of protein transport across membranes. EMBO J. 13:2686.

4171 39. Fru¨ h, K., K. Ahn, H. Djaballah, P. Sempe, P. M. van Endert, R. Tampe, P. A. Peterson, and Y. Yang. 1995. A viral inhibitor of peptide transporters for antigen presentation. Nature 375:415. 40. Butt, T. R., M. I. Khan, J. Marsh, D. J. Ecker, and S. T. Crooke. 1988. Ubiquitinmetallothionein fusion protein expression in yeast: a genetic approach for analysis of ubiquitin functions. J. Biol. Chem. 263:16364. 41. Wang, E. W., B. M. Kessler, A. Borodovsky, B. F. Cravatt, M. Bogyo, H. L. Ploegh, and R. Glas. 2000. Integration of the ubiquitin-proteasome pathway with a cytosolic oligopeptidase activity. Proc. Natl. Acad. Sci. USA 97:9990. 42. Princiotta, M. F., U. Schubert, W. Chen, J. R. Bennink, J. Myung, C. M. Crews, and J. W. Yewdell. 2001. Cells adapted to the proteasome inhibitor 4-hydroxy5-iodo-3-nitrophenylacetyl-Leu-Leu-leucinal-vinyl sulfone require enzymatically active proteasomes for continued survival. Proc. Natl. Acad. Sci. USA 98:513. 43. Henderson, R. A., H. Michel, K. Sakaguchi, J. Shabanowitz, E. Appella, D. F. Hunt, and V. H. Engelhard. 1992. HLA-A2.1-associated peptides from a mutant cell line: a second pathway of antigen presentation. Science 255:1264. 44. Marusina, K., G. Reid, R. Gabathuler, W. A. Jefferies, and J. J. Monaco. 1997. Novel peptide-binding proteins and peptide transport in normal and TAP-deficient microsomes. Biochemistry 36:856.