HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Yamano, T. Articles by Udono, H. Search for Related Content PubMed PubMed Citation Articles by Yamano, T. Articles by Udono, H.

Allele-Selective Effect of PA28 in MHC Class I Antigen Processing¹

Taketoshi Yamano^2,*, Hidetoshi Sugahara^2,*, Shusaku Mizukami^*, Shigeo Murata, Tomoki Chiba, Keiji Tanaka, Katsuyuki Yui^¶ and Heiichiro Udono^3,*

^* Laboratory for Immunochaperones, Research Center for Allergy and Immunology, RIKEN Yokohama Institute, Tsurumi, Yokohama, Japan; Laboratory of Protein Metabolism, Graduate School of Pharmaceutical Sciences, the University of Tokyo, Hongo, Bunkyo-ku, Tokyo, Japan; Graduate School of Life and Environmental Sciences, University of Tsukuba, Tennodai, Tsukuba City, Ibaraki, Japan; Laboratory of Frontier Science, Tokyo Metropolitan Institute of Medical Science, Honkomagome, Bunkyo-ku, Tokyo, Japan; and ^¶ Division of Immunology, Department of Molecular Microbiology and Immunology, Graduate School of Biomedical Sciences, Nagasaki University, Sakamoto, Nagasaki, Japan

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
PA28 is an IFN--inducible proteasome activator and its genetic ablation causes complete loss of processing of certain Ags, but not all of them. The reason why this occurs and how PA28 influences the formation of peptide repertoires for MHC class I molecules remains unknown. In this study, we show the allele-specific role of PA28 in Ag processing. Retrovirus-transduced overexpression of PA28 decreased expression of K^d (D^d) while it increased K^b and L^d on the cell surface. By contrast, overexpression of PA28C5, a mutant carrying a deletion of its five C-terminal residues and capable of attenuating the activity of endogenous PA28, produced the opposite effect on expression of those MHC class I molecules. Moreover, knockdown of both PA28 and β by small-interfering RNA profoundly augmented expression of K^d and D^d, but not of L^d, on the cell surface. Finally, we found that PA28-associated proteasome preferentially digested within epitopic sequences of K^d, although correct C-terminal flankings were removed, which in turn hampered production of K^d ligands. Our results indicate that whereas PA28 negatively influences processing of K^d (D^d) ligands, thereby, down-regulating Ag presentation by those MHC class I molecules, it also efficiently produces K^b (L^d) epitopes, leading to up-regulation of the MHC molecules.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Major histocompatibility complex class I ligands are produced mainly by proteasomes (1, 2, 3). The proteasome activator PA28 ( and β), which is strongly induced by the major immunomodulatory cytokine IFN- (1, 4), has been implicated in the regulation of MHC class I Ag processing (5). PA28 accelerates the production of MHC class I ligands from longer precursor peptides by the 20S proteasome in vitro (6). The C-terminal flanking region is critical for efficient production of the T cell epitope (7). It is possible that PA28 activates the 20S proteasome by opening its -ring (8) that is usually closed and through which substrates can pass into the core catalytic portion. In vivo analysis has also shown that the processing of several, but not all, Ags is stimulated by overexpression of PA28 and PA28β (9). Likewise, the lack of PA28 impairs the ability to process a melanoma Ag TRP2-derived peptide, but does not apparently result in a deficient processing of other Ags such as OVA (10, 11). This indicates that PA28/β is not a prerequisite for Ag presentation in general, but plays an essential role for the processing of certain Ags. So far, the reason why PA28 is crucial in the processing of certain Ags remains unknown.

IFN- stimulation increases expression of the "homo-PA28 proteasome" and the "hybrid proteasome" (12). The former proteasome is a complex where PA28 is attached to both ends of the central 20S proteasome and the latter comprises the 20S proteasome flanked by PA28 on one side and a 19S cap (alias regulatory particle RP or PA700) on the other, functioning as a new ATP-dependent protease, similar to 26S proteasomes, which have a 19S cap on both sides (13). It has been suggested that hybrid proteasomes play a major role in IFN--induced peptide supply for MHC class I molecules, because they can directly process ubiquitylated proteins into MHC class I ligands or into the shortest precursor peptide (14). Indeed, PA28 deficiency suppressed up-regulation of cell surface MHC class I molecules by IFN-, even though immunoproteasomes could be induced (10). Thus, PA28, possibly as a hybrid proteasome, is a prerequisite for IFN--induced enhancement of MHC class I expression. Cascio et al. (14) have shown that the peptide repertoire produced in vitro by a hybrid proteasome from insulin growth factor 1 protein was very different from that produced by the 26S proteasome. Considering the essential role of PA28 in IFN--induced enhancement of MHC class I, it is possible that the repertoire of MHC class I ligands changes in response to IFN-; however, the PA28-induced peptide repertoire in vivo has not yet been studied and neither has the role of allelic polymorphism on the activity of PA28.

We noticed that expression levels of cell surface K^d and D^d, but not L^d, in BALB/c PA28^–/–β^–/– cells were slightly higher than those measured on wild-type cells. Knocking down both PA28 and β by small-interfering RNA (siRNA)⁴ revealed that K^d and D^d, but not L^d, molecules were extremely up-regulated on the cell surface. To clarify the role of PA28, we devised a PA28 mutant lacking the five C-terminal residues, designated as PA28C5 and capable of competing with the endogenous PA28. Using this mutant PA28C5 together with knocking down PA28 by siRNA and retrovirus-transduced overexpression of PA28, we examined the role of PA28 on the cell surface expressions of various MHC molecules. Furthermore, we performed digestion assays with the 20S proteasome mixed with recombinant (r)PA28 (or PA28C5) plus PA28β for several synthetic peptides harboring K^b or K^d ligands, and liquid chromatography/mass spectrometry (LC/MS) analysis revealed that whereas the homo-PA28 proteasome is prone to digest within sequences of K^d ligands even with removing correct C-terminal flanking, it was partly able to produce K^b ligands. Our results indicate that the effect of PA28 in Ag processing, be it positive or negative, is allele specific.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Cells and cell culture

RL1 is a BALB/c mouse T cell leukemia. EL4 is a methylcholanthrene-induced C57BL/6 mouse thymoma. E.G7 is an OVA cDNA-transfected EL4 cell line (15). PA28^–/–β^–/– tumor cells is a methylcholanthrene-induced BALB/c fibrosarcoma. CTLs specific for each peptide were generated and maintained as described previously (16). E.G7 was cultured in RPMI 1640 supplemented with 10% FCS (Invitrogen), 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 20 mM L-glutamic acid, 5 x 10^–5M 2-ME, and penicillin-streptomycin supplemented with 400 µg/ml G418. Cells transfected with a retroviral gene (pMSCVpuro encoding murine PA28, PA28C5) were selected and maintained with 2 µg/ml puromycin.

Abs, peptides, and reagents

Rabbit polyclonal anti-histidine tag was obtained from MBL. Mouse monoclonal anti-20S2, Rpt1, PA28β were obtained from BIOMOL. mAb to PA28 was produced from a hybridoma clone 1G11 in our laboratory. Rabbit polyclonal anti-actin Ab was obtained from Sigma-Aldrich. Mouse monoclonal anti-K^b (AF6-88.5, IgG2a, biotin-conjugated) and D^d (34-5-8S, IgG2a, biotin-conjugated) were obtained from BD Pharmingen. Mouse monoclonal anti-K^d (31-3-4, IgM), anti-L^d (30-5-7, IgG2a) were purified from ascitic fluid. These Abs recognize peptide-bound folded MHC class I molecules. 25D1.16 mAb specific for OVA_257–264-K^b was provided by Dr. R. Germain (17). Abs specific to heat shock protein 90 were purchased from Stressgen.

The substrate suc-LLVY-amc was obtained from the Peptide Institute. The peptides OVA_248–269 (EVSGLEQLESIINFEKLTEWTS: underlined residues represent the K^b-restricted epitope), OVA_257–269 (SIINFEKLTEWTS: K^b), OVA_248–264 (EVSGLEQLESIINFEKL: K^b), OVA_257–264 (SIINFEKL: K^b), TRP2_181–193 (VYDFFVWLHYYSV; K^b), TRP2_181–188 (VYDFFVWL; K^b) (18), circumsporozoite protein (CSP)_281–289 (SYVPSAEQI; K^d), and pRL1a (IPGLPLSL; L^d) (19) were purchased from Sawady Technology. For LC/MS analysis, OVA_252–269, TRP2_181–193, vesicular stomatitis virus (VSV) NP_47–66, HSV glycoprotein B_493–512, Tum-P198_814–828, Listeria monocytogenes p60_212–231, Influenza A HA_513–532, and Plasmodium yoelii CSP_278–295 peptides (with 90% purity) were purchased from Scrum. The sequences of those peptides are indicated in Figs. 6 and 7. As TRP2_181–193 peptide contains a cysteine residue (Cys¹⁷⁹) in its N-terminal flanking, we used only C-terminally extended peptide. Also, Tum-P198_14–28 peptide has an uncertain amino acid in its N-terminal flanking (XM_214370; HEVGXKYQAVTATLEEKRKE); we used only C-terminally extended peptide. ATP was purchased from Sigma-Aldrich.

Recombinant proteins and plasmids

Purification of murine PA28 was described previously (10). cDNA of murine PA28β was amplified by RT-PCR from the mRNA of EL4 cells and cloned into the pQE31 expression vector (Qiagen) (5'SacI and 3'KpnI), which incorporates a 6x His tag at the N terminus of the protein. Primers for amplification used were: forward, ATGAGCTCCATGGCCAAGCCTTGTG, and reverse, ATGGTACCTCAGTACATCGATGGCTTTT. A series of reverse primers in which the 3' codon was serially deleted while the 5' codon was elongated and was used to produce the PA28C1 to PA28C9 deletion mutants (5'BamHI and 3'KpnI). Protein expression was induced by 1 mM isopropyl-β-D-thiogalactoside, and purified with Ni-NTA as described previously (10). Murine PA28 and PA28C5 were also cloned into sites of 5'XhoI and 3'HpaI of pMSCVpuro (Takara Bio).

Loading of peptides and proteins, and Ag-presentation assay

rPA28 and its deletion mutants (100 µg) were osmotically loaded into 2 x 10⁶ EL4 cells or LPS blasts with or without synthetic peptides (4 nmol) as described previously (10). The cells were used for CTL assay as described previously (10).

Electrophoresis and peptidase assay

For preparation for cell lysates, EL4 or PA28^–/–β^–/– cells were lysed with 26S buffer (25 mM Tris-HCl (pH 7.5), 250 mM sucrose, 1 mM DTT, 1 mM PMSF) containing 1% Nonidet P-40 for 30 min on ice, and centrifuged. The supernatants were obtained for electrophoresis. Native PAGE was performed with a 3–10% gradient gel (Wako Pure Chemical). SDS-PAGE was performed using a 5–20% gradient gel. The in-gel hydrolysis assay was performed as described previously (20). Peptidase activity of the proteasome after separation by native PAGE was measured using suc-LLVY-amc (chymotrypsin-like activity). After electrophoresis, the gels were incubated with the substrates (0.1 mM) at 37°C for 10 min. Proteasome bands were then visualized by exposure of the gel to UV light at 360 nm and detected with a 460-nm filter.

Flow cytometry

Cells were suspended in FACS buffer (PBS, 1% FCS, and 0.02% azide) with specific primary Ab and incubated for 30 min at 4°C. They were then washed twice and stained with a second Ab, FITC-conjugated rabbit anti-mouse IgG (H+L), or avidin-conjugated FITC (Jackson ImmunoResearch Laboratories). Flow cytometric analysis was performed on a FACScan (BD Biosciences), and the data were analyzed by CellQuest software (BD Biosciences). The same experiments were also performed with cells treated with 1 ng/ml IFN- (R&D Systems) for 48 h. The acid-wash recovery assay was performed as described previously (10).

Small-interfering RNA

Target sequences for PA28 and PA28β were AAGCCAAGGTGGATGTGTT and AGCGAGCAAGGCCAGAAGC, respectively. Oligonucleotides were cloned into piGENE PUR hU6 plasmid vector (Toyobo). A total of 1 x 10⁶ cells were transfected with the plasmids (2 µg) encoding siRNA by using the Nucleofector device (Amaxa Biosystems).

Purification of the 20S proteasome and its use for peptide digestion assay with LC/MS

Normal mouse livers (10 ml) were homogenized in buffer A (25 mM Tris-HCl (pH 7.5), 1 mM DTT, 0.25 M sucrose, 1 mM PMSF), and centrifuged at 10,000 x g for 20 min. The supernatant was further centrifuged at 100,000 x g for 1 h, and then, the resulting supernatant was resolved into AKTA FPLC connected with a RESOURCE Q column (Amersham Biosciences) equilibrated with buffer A, and eluted with 01 M NaCl gradient at 4 ml/minute for 30 min and collected at 1 ml/fraction/minute. The active fractions (Fr. 7–15) against suc-LLVY-amc with 0.02% SDS were mixed and concentrated with a Centrifugal Filter Device (Ultrafree-0.5; Millipore) and loaded on a heparin column (HiTrap Heparin HP; Amersham Biosciences), followed by elution with 01 M NaCl gradient at 2 ml/min for 20 min. The active fractions against suc-LLVY-amc were eluted with <240 mM NaCl, and were further subjected to a hydrophobic column (RESOURCE PHE for HIC; Amersham Biosciences) equilibrated with buffer B (10 mM potassium phosphate (pH 7.5), 1 mM DTT, 0.25 M sucrose) and eluted with 10500 mM potassium phosphate. The active fractions were collected and concentrated (and washed) by a centrifugal filter device with buffer A. The purified material was prepared at 20 µg/ml with buffer A and used for peptide digestion assay. Synthetic peptide with 90% purity (1 µg) was incubated with 50 ng of 20S proteasome with or without indicated doses of recombinant PA28, PA28(C5), and PA28β in a 20 µl total volume for 3 h, and 40 µl of 0.2% trifluoroacetic acid was added. After mild pipetting, the sample was kept on ice for 30 min and then applied onto a LC/MS system with electrospray ionization.

For LC/MS, an Alliance 2695 Separation Module attached to an Atlantis dC18 column (Waters) HPLC, online connected to MICROMASS ZQ (Waters), was used. The buffer A and B for HPLC was 0.1% formic acid/water and 0.08 formic acid/water plus 80% acetonitrile, respectively. The acetonitrile gradient was performed from 0 to 80% buffer B for 30 min. Scans were acquired every 1.5 s over a mass range m/z 400 to 1500.

MassLynx and BioLynx protein software were used for data analysis. To monitor hydrolysis activity for suc-LLVY-amc, the 20S proteasome (50 ng), mixed with indicated doses of recombinant PA28, PA28(C5), and PA28β in a total volume of 100 µl, was subjected onto a MultiDetection Microplate Reader, POWERSCAN HT (Dainippon Pharmaceutical).

	Results

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
PA28 deletion mutants capable of interfering with the activity of PA28 in Ag processing

Initially, we tried to obtain PA28 variants that could compete with the action of endogenous PA28 in terms of Ag processing. Because the C-terminal amino acid residues of PA28 are required for binding to the 20S proteasome (21, 22), we examined how various deletion mutants of PA28 influence Ag processing. To this end, we expressed rPA28 mutants whose residues were serially deleted, ranging from 1 to 9 aa at the C terminus, termed PA28C1 to PA28C9, respectively, in Escherichia coli, and purified them to near homogeneity. These mutants were osmotically loaded with OVA_248–269, a precursor polypeptide harboring the CTL epitope OVA_257–264 of OVA, into EL4 cells (H-2^b), and the cytolysis of the loaded cells by CTLs specific for OVA_257–264 was monitored. There was still enhanced cytolytic activity with the PA28 mutants carrying deletions of one or two C-terminal residues, similar to wild-type PA28 (Fig. 1A). However, no significant enhancement was observed when 3- or 4-aa deletion mutants of PA28 were used. Interestingly, PA28 mutants lacking five to nine residues from the C terminus markedly inhibited the Ag processing. These effects, i.e., enhancement by PA28C1 and inhibition by PA28C5 of the OVA epitope processing, were also observed in C57BL/6 LPS blasts (Fig. 1B). Importantly, however, both mutants did not have any influence on Ag processing by LPS blasts derived from PA28^–/–/β^–/– mice (Fig. 3B), indicating that these PA28 variants exert their effects through their association with endogenous PA28 and/or PA28β.

View larger version (54K): [in this window] [in a new window]   FIGURE 1. Effects of PA28 deletion mutants capable of interfering with PA28 action in Ag processing. A, A series of rPA28 deletion mutants ranging from C-terminal residues 1–9 (designated C1 to C9) were osmotically co-introduced with OVA248–269 (4 nM) into EL4 cells. The cells were used as target cells in a 51Cr-release assay (E:T ratio 10, ; E:T ratio 5, ). B, PA28 deletion mutants affected Ag processing of PA28+/+β+/+ cells but not PA28–/–β–/– cells. Intact PA28, PA28C1, and PA28C5 were co-introduced with OVA248–269 (4 nM) into PA28+/+β+/+ or PA28–/–β–/– LPS blasts. The cells were used as target cells in the 51Cr-release assay, as shown in A. C, Ag processing of COOH- but not NH2-terminally extended precursor peptides was decreased by PA28C5 in PA28+/+β+/+ cells. OVA257–269 and OVA248–264 (4 nM each) were introduced with () or without (•) PA28C5 into EL4 and LPS blasts. The cells were used as target cells in the 51Cr-release assay. D, PA28C5 associates with the homo-PA28 proteasome as well as with the hybrid proteasomes. Either PA28 or PA28C5 was osmotically introduced into EL4 (upper panel) or PA28–/–β–/– cells (lower panel). Two hours later, cell extracts were separated by native PAGE and chymotrypsin-like activity was examined using suc-LLVY-amc. In addition, the native PAGE gels were subjected to Western blotting with specific Abs to polyhistidine, 20S2, PA28, and PA28β. Quantities of loaded proteins were checked by anti-actin Ab (SDS-PAGE). E, Intact PA28 and PA28C5 were co-introduced with OVA248–269 (4 nM) into EL4 cells. Four hours later, after introduction, the cell surface Kb-OVA257–264 complex was examined with 25D1.16 mAb.

View larger version (21K): [in this window] [in a new window]   FIGURE 3. Less efficient activity of PA28C5 in hydrolysis of suc-LLVY-amc and production of TRP2181–188 from the C-terminally longer precursor. A, The 20S proteasome (50 ng) purified from normal mouse liver was mixed with indicated doses of recombinant PA28, PA28C5, and PA28β in a final 100 µl volume containing 0.1 mM suc-LLVY-amc and was then subjected onto a MultiDetection Microplate Reader to detect fluorescence activity. B, The same preparation as in A, but total volume was 20 µl, was added with synthetic peptide TRP2181–193 (1 µg) and incubated for 3 h at 37°C. The produced Kb epitope, TRP2181–188, was detected by LC/MS analysis. The quantity of detected epitope was indicated as areas of [M+H] m/z. The result was a representative of three independent experiments.

To understand the activity of PA28C5, we subsequently examined its association with the 20S proteasome in the cells. PA28C5 or PA28 were loaded into EL4 cells or PA28^–/–β^–/– cells, and 2 h later, the cell lysates were subjected to native PAGE to analyze enzyme activities and structures of the proteasome. PA28 increased the chymotrypsin-like activity of the hybrid as well as of the homo-PA28 proteasome, whereas PA28C5 slightly decreased the activity (Fig. 1D, left end panel). Both PA28 and PA28C5 associated with the hybrid as well as with the homo-PA28 proteasome in EL4 cells, but only PA28 associated with these proteasomes in PA28^–/–β^–/– cells as judged by Western blotting with anti-histidine Ab and other Abs indicated in Fig. 1D. PA28C5 should not be able to bind to the 20S proteasome because several C-terminal residues of PA28 are critical for binding. Deletion of only one residue at the C terminus of PA28 prevents its association with the -ring of the 20S proteasome, as previously indicated (22). Exogenously introduced PA28 does not need endogenous PA28 to bind to the 20S proteasome, i.e., the homopolymeric PA28 complex can associate with the 20S proteasome in vivo. This is consistent with previous findings showing that it functions as an activator in vitro (5). We also confirmed that osmotic cointroduction of OVA_248–269 with PA28 but not PA28C5 into EL4 cells increased the cell surface K^b-OVA_257–264 complex, compared with OVA_248–269 alone (Fig. 1E).

PA28 augments the cell surface expression of the K^b-OVA_257–264 complex and of total K^b

Subsequently, we prepared stable lines of E.G7 cells expressing a full-length OVA gene transfected with PA28, PA28C5, and mock plasmid (pMSCV empty vector), designated E.G7(PA28), E.G7(PA28C5), and E.G7mock, respectively. Western blot analysis revealed that E.G7(PA28) and E.G7(PA28C5) produced 2- and 3-fold more PA28 and PA28C5, respectively, than endogenous PA28 measured in E.G7mock cells (Fig. 2A). We also examined the amount of cell surface of K^b-OVA_257–264 and total K^b for each transfectant cultured with or without IFN-. In the absence of IFN-, PA28 up-regulated the K^b-OVA_257–264 complex and total K^b but PA28C5 markedly suppressed expression of both K^b-OVA_257–264 and total K^b (Fig. 2, B and D). The inhibitory effect by PA28C5 was still visible after 3 days of culture with IFN- (data not shown). The down-regulation of K^b expression was completely restored by a pulse of E.G7 (PA28C5) with OVA_257–264 (Fig. 2, C and E), indicating that the diminished supply of endogenous peptides was responsible for the down-regulated expression. Expression of D^b by those transfectants was almost the same as that of K^b (data not shown).

View larger version (32K): [in this window] [in a new window]   FIGURE 2. PA28 positively influences cell surface expression of Kb-OVA257–264 complex and total Kb molecules. A, Levels of expressed proteins. The amount of PA28 and PA28C5 in total cell extracts of E.G7mock, E.G7(PA28), and E.G7 (PA28C5) cells was examined by Western blotting using specific Abs. B, Cell surface expression of Kb-OVA257–264 complex on OVA-expressing E.G7mock (), E.G7(PA28) (), and E.G7(PA28C5) () cells were analyzed by FACS with mAb 25D1.16. C, The cell surface expression of Kb molecules on three transfectants was analyzed by FACS. OVA257–264 (10–5 M) was also pulsed on cells for 12 h before analysis. D, E.G7mock and E.G7(PA28C5) cells treated with or without IFN- (1 ng/ml) were analyzed for the cell surface level of Kb-OVA257–264 complex. E, The level of total Kb molecules on the transfectants, treated with or without IFN- (1 ng/ml), was examined. Results of B–E are shown as mean fluorescence subtracted of control fluorescence (staining with only second Ab or FITC-conjugated streptavidin) (mean ± SEM; n = 3). The result was a representative of three independent experiments.

PA28C5 was less efficient in production of TRP2_181–188 from C-terminally longer precursors, compared with intact PA28

PA28 enhanced the expression of K^b-OVA_257–264 and total K^b, in contrast, PA28C5 suppressed those, as shown in Fig. 2. Therefore, we investigated in the vitro effect of PA28 plus PA28β and PA28C5 plus PA28β together with the 20S proteasome on production of TRP2_181–188 from C-terminally longer precursors whose processing was previously shown to be dependent on PA28 (13). The 20S proteasome (50 ng) purified from mouse liver was mixed with graded doses of recombinant PA28 (or PA28C5) and PA28β in a volume of 20 µl, and then added with 1 µg of TRP2_181–193, followed by incubation for 3 h at 37°C. A total of 40 µl 0.2% trifluoroacetic acid was added to the mixture and a 10 µl total volume was injected into LC/MS to detect exact K^b-epitope TRP2_181–188. Simultaneously, hydrolysis activity for suc-LLVY-amc was examined. The results showed that both hydrolysis activity and production of TRP2_181–188 by PA28C5 were significantly lower than those by PA28 (Fig. 3).

PA28 attenuates cell surface expression of K^d and D^d but not L^d

We next examined the effects of PA28 and PA28C5 on the expression of MHC class I molecules other than K^b and D^b. The cell surface expression of K^d and D^d on peritoneal macrophages (M) of BALB/c PA28^–/–β^–/– mice was significantly higher than on M of wild-type mice, whereas, in contrast, L^d expression was slightly lower in M of PA28^–/–β^–/– (Fig. 4A). To assess directly the effects of PA28 and PA28C5, we established BALB/c RL1(H-2^d) expressing PA28, PA28C5, and mock plasmid, designated RL1 (PA28), RL1 (PA28C5), and RL1mock, respectively, and cell surface MHC class I molecules of those transfectants were examined. Surprisingly, expression of K^d was down-regulated but that of L^d was enhanced by PA28, whereas PA28C5 induced the opposite effects (Fig. 4B). The expression of K^d was restored to normal level by a pulse with CSP_281–289 derived from P. yoelii (23) (Fig. 4C). In an acid-wash recovery assay, PA28 delayed the recovery of K^d, whereas PA28C5 accelerated the recovery (Fig. 4D). A pulse with CSP_281–289 restored the delayed recovery of K^d (Fig. 4D). These results clearly show that PA28 is responsible for the shortage of K^d ligands in the cells. By contrast, expression of L^d was restored by a pulse with an exact L^d epitope, pRL1a (19) (Fig. 4C). In acid-wash recovery, PA28 accelerated, whereas PA28C5 delayed, the recovery of L^d (Fig. 4D). The delayed recovery of L^d was restored by a pulse with pRL1a (Fig. 4D). The results indicate that PA28 stimulates the production of L^d ligands in cells, in contrast to its effect on the production of K^d ligand.

View larger version (43K): [in this window] [in a new window]   FIGURE 4. PA28 negatively influences cell surface expression of Kd, Dd, but not Ld. A, Peritoneal M derived from BALB/c PA28+/+β+/+ and PA28–/–β–/– mice were analyzed for expression of Kd, Dd, and Ld. B, Cell surface expression levels of Kd and Ld on RL1(PA28), RL1(PA28C5), and RLmock cells are shown as histograms. Gray shadows indicate staining with only avidin-FITC or second Abs conjugated with FITC. C, Data for B are shown as bar graphs. MHC class I-specific ligands (CSP281–289 for Kd, pRL1a for Ld: 10–5 M) were also pulsed on those transfectants for 12 h before analysis. D, Acid-wash recovery of Kd and Ld molecules. RL1(PA28), RL1(PA28C5), and RLmock cells were treated with acid (pH 3.0), and incubated with or without specific ligands for 8 h to examine the recovery of Kd and Ld molecules. The mean fluorescence was plotted. E, IFN- (1 ng/ml, 48 h)-treated RL1(PA28), RL1(PA28C5), and RLmock cells were analyzed as in C. F, Acid-wash recovery assay with IFN--treated (1 ng/ml, 48 h) RL1(PA28), RL1(PA28C5), and RLmock cells were performed as in D (mean ± SEM; n = 3). The result was a representative of three independent experiments.

Next, we knocked down the expression of both PA28 and β of MEF/3T3 by siRNA to confirm the effects of PA28. As shown in Fig. 5A, expression of both PA28 and β was specifically repressed (80 90%) but expression of heat shock protein 90 and the 20S2 subunit was not altered, and hydrolysis activity to the substrate suc-LLVY-amc was decreased especially in the homo-PA28 proteasome. Native-PAGE followed by Western blotting with Abs to PA28, β, and 20S2, showed a decrease of the hybrid proteasome as well as the homo-PA28 proteasome but not of the singly capped 26S proteasome (RC) and the probably empty 20S proteasome (C) (Fig. 5A). Intriguingly, expression of K^d and D^d but not L^d was enhanced by depletion of PA28 and β (Fig. 7B) and those alterations are shown in Fig. 5C, which confirms the negative effect of PA28 on the expression of K^d and D^d.

View larger version (44K): [in this window] [in a new window]   FIGURE 5. Depletion of PA28 and β by siRNA augments cell surface expression of Kd, Dd, but not Ld. A, MEF/3T3 cells were transfected with siRNA against PA28 and β, or with a vector not encoding siRNA. After 48 h, expressed proteins were examined by SDS-PAGE followed by Western blotting with the indicated Abs. The peptidase activity for suc-LLVY-amc and the structure of the proteasome were examined. B, Expression of MHC class I Kd, Dd, Ld of MEF/3T3 cells in A was examined by FACS analysis. C, Mean fluorescence intensities of Kd, Dd, and Ld in B are shown as bar graphs.

View larger version (35K): [in this window] [in a new window]   FIGURE 7. Digesting patterns of precursor peptides harboring Kb, Kd, and Ld ligands by the proteasome, the same as in Fig. 1, except peptides were digested by the proteasome. A, HSV glycoprotein B493–512; B, VSV NP47–66; C, Influenza A HA513–532; D, P. yoelii CSP278–295. The exact MHC class I epitope is depicted with bold (and/or underlining). Note that each retention time (RT) of the exact epitope produced by digestion of A and B, was precisely the same as that of the synthetic corresponding peptides. None of the exact epitope was observed among the digested materials of C and D. N.D., Not detected or <100.

So far, our results consistently suggest that for the ligands used in this study, PA28 augments the generation of ligands for D^b, K^b, and L^d, while it attenuates the production of K^d and D^d ligands. It is crucial to know the reason why PA28 negatively influences K^d and D^d ligand production. To this end, in the next series of experiments, we investigated whether PA28 has a different influence on in vitro digestion using several precursor peptides. For this, we prepared four K^b, four K^d, and one L^d ligands extending C-terminal (and/N-terminal) flanking regions, although these peptides are not necessarily the relevant or native precursors. We digested them with the 20S proteasome alone or plus recombinant PA28 and PA28β, or plus PA28C5 and PA28β.

We observed exact K^b ligands produced from OVA_252–269, TRP2_181–193, HSV 16 glycoprotein B_493–512, VSV NP_47–66, and also observed N-terminal extended L^d precursor peptide from murine CMV pp89_163–182 (Figs. 6 and 7), although some of these epitopes were also partly digested. The 20S proteasome alone did not produce epitope fragments from OVA_252–269, TRP2_181–193, and VSV NP_47–66, but in the presence of PA28, exact epitopes were cleaved. An epitope from HSV glycoprotein B_493–512 was produced by the 20S proteasome alone and the quantity was further enhanced by PA28 (Fig. 7). Thus, in these cases, PA28 positively influenced cleavage of the K^b and L^d epitopes. In contrast, exact K^d ligands or its precursors were not recovered by the same proteasome preparations from L. monocytogenesis p60_212–231, Tum-P198_14–28, influenza A HA _513–532, P. yoelii CSP_278–295 (Figs. 6 and 7). Correct C-terminal flanking was not removed from L. monocytogenesis p60_212–231, which resulted in improper processing of the epitope. Even when correct C-terminal flanking was removed from Tum-P198_14–28, influenza A HA _513–532, and P. yoelii CSP_278–295, strong digestion within the epitopes occurred in the presence of PA28, which might result in destruction of the epitopes. Importantly, even the 20S proteasome alone efficiently digested within the epitope of influenza A HA_513–532, but the presence of PA28C5 diminished production of those fragments, while in contrast, intact PA28 augmented those cleavages (Fig. 7). This evidence suggested that PA28C5 could regulate (or suppress) the 20S proteasome-induced overdigestion of particular peptides, while intact PA28 could stimulate it. Although there was a number of reason(s) for the improper processing, this evidence accounted for the reason why PA28 negatively influenced production of K^d ligands by the proteasome.

View larger version (35K): [in this window] [in a new window]   FIGURE 6. Digesting patterns of precursor peptides harboring Kb, Kd, and Ld ligands by the proteasome. One microgram of each peptide, A–E, was incubated with the 20S proteasome (50 ng) mixed with 100 ng each of PA28 plus PA28β, or PA28C5 plus PA28β in 20 µl total volume for 3 h at 37°C. A, OVA252–269; B, TRP2181–193; C, L. monocytogenesis p60212–231; D, Tum-P19814–28; E; MCMV pp89163–182. The digestion mixture was subjected onto LC/MS analysis to detect peptide fragments. Amino acid sequences, their retention time (RT), and area determined by m/z values of [M+H] and [M+2H] of digested fragments were indicated. Horizontal lines under peptide sequences indicated the recovered fragments after digestion. Quantity of each fragment produced by the 20S proteasome and PA28 and PA28β is visualized as follows: >1,000,000 (thickest black bar), 100,000999,999 (second thickest black bar), 10,00099,999 (third thickest black bar), 1,0009,999 (thinnest black bar). Cleavage points by the proteasome were also indicated by inverted triangles () whose sizes paralleled with obtained quantity (area) of the digested fragments. The exact MHC class I epitope is depicted with bold (and/or underlining). Note that each retention time (RT) of the exact epitope produced by digestion of A, B, and E was precisely the same as that of the synthetic corresponding peptide. None of the exact epitope was observed among the digested materials of C and D. N.D., Not detected or <100.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Ag processing and presentation is crucial for the initiation of the immune response. Over the past decade, there is growing evidence that the proteasome, a large multisubunit protein degradative machinery in eukaryotes, plays an important role as a processing enzyme responsible for the generation of MHC class I ligands (1, 2, 3). This processing system is elaborately regulated by various immunomodulatory cytokines. In particular, IFN- induces the formation of the immunoproteasomes, in which three IFN--inducible subunits (i.e., β1i, β2i, and β5i) can replace the constitutive catalytic 20S subunits (i.e., β1, β2, and β5) during proteasome biogenesis (24). Furthermore, IFN- also induces PA28, producing the homo-PA28-20S proteasome and the hybrid proteasome, which contributes importantly to efficient production of MHC class I ligands (6, 12). Furthermore, it has been shown that the immunoproteasome and the PA28-containing proteasome in concert or independently play a critical role in the generation of the MHC class I ligands (1). However, the molecular mechanisms underlying the correct generation of CTL epitopes by those different types of proteasomes remain a mystery.

In the present study, we tried to clarify the role of the IFN--inducible proteasome activator PA28 in the Ag-processing and -presentation pathway, and surprisingly found different effects of PA28 on the MHC class I epitope generation depending on the allelic polymorphism. Indeed, whereas PA28 is unable to produce many of K^d ligands and thereby attenuates cell surface expression of those MHC class I molecules, it is able to produce most (if not all) K^b (also L^d) epitopes, leading to up-regulation of the corresponding MHC molecules on the cell surface. Our previous observation that IFN--induced up-regulation of K^b was canceled in PA28-deficient cells (10) is consistent with the present findings and indicate that PA28 plays a prominent role in IFN--stimulated peptide supply. Conversely, knockdown of both PA28 and PA28β by siRNA significantly increased the expression of K^d and D^d (Fig. 5, B and C), indicating that PA28 negatively influences the presentation of those ligands. Indeed, constitutive expression of K^d and D^d in peritoneal M from the BALB/c PA28^–/–β^–/– mouse was slightly higher than on wild-type M, whereas that of L^d was lower (Fig. 4A). Acid-wash recovery of cell surface K^d of RL1 cells was accelerated in the presence of PA28C5, especially upon IFN- treatment in which endogenous PA28 was extremely induced; in contrast, recovery of L^d was mostly retarded by PA28C5 but accelerated by intact PA28 (Fig. 4F). These results clearly support the negative and positive influence of PA28 on the processing of K^d and L^d ligands, respectively.

Why PA28 contributes differently to the MHC class I ligand generation in an allelic polymorphism-dependent fashion is largely unknown. We therefore analyzed a series of synthetic peptides harboring various epitopes whose in vitro digestion was conducted by the PA28-20S proteasome to evaluate from which types of peptides correct epitopes are produced. The PA28-20S proteasome was unable to process any of the K^d ligands tested in this study from their longer precursors, although K^b ligands were efficiently processed by the same proteasome preparation (Figs. 6 and 7). This finding, however, is not in agreement with previous observations that a K^d ligand of JAK1 kinase, SYFPEITHI, was efficiently produced from a longer precursor peptide by the PA28-20S proteasome (6). In fact, we confirmed the enhancing effect of PA28 on processing of this epitope in vivo by osmotically loading the precursor peptides and rPA28 into P815 (H-2^d) cells and also in vitro peptide digestion assay with PA28-20S proteasome (data not shown). The reason for this discrepancy is unknown at present. Nonetheless, considering the clear negative effect of siRNA-mediated knockdown of PA28 and β on K^d (D^d) expression, we emphasize that the majority of those ligands, if not all, are likely to be improperly or inefficiently processed by the PA28-20S proteasome. In vitro production of mouse T cell epitopes from longer precursors by the PA28-20S proteasome was mainly observed in the context of L^d and K^b, although only the JAK1 kinase-derived epitope was K^d restricted as described (6). For example, the mouse leukemia peptide pRL1a (7, 25) and the MCMV pp89-derived peptide (6) were for L^d, and the Moloney murine leukemia virus gag-derived peptide (9) and the melanoma Ag TRP2 peptide (10, 11) are K^b ligands. Our current observation that PA28 stimulates the processing of K^b and L^d is consistent with those reports, especially results of in vitro peptide digestion supporting the positive influence of PA28 on the production of K^b ligands by the PA28-20S proteasome (Figs. 6 and 7).

It should be pointed out that production of K^d ligands in vivo was absolutely dependent on the proteasome because acid-wash recovery of K^d as well as L^d was significantly delayed by treatment with the proteasome inhibitor lactacystin (data not shown), suggesting that the 26S proteasome rather than the PA28-20S or another type of proteasome flanked with the newly identified regulatory particle like PA200 (26, 27) and/or the immunoproteasome might be responsible for processing of K^d ligands. To this point, for the processing of P. yoelii CSP_278–295 and Tum-P198_14–28, we observed that a fraction of the 26S proteasome purified from the liver of the PA28^–/–β^–/– mouse could exactly produce the K^d epitopes, from the same precursors, although these epitopes were not efficiently produced by the PA28-20S proteasome purified from a wild-type mouse liver (data not shown). The other two K^d ligands used in this study were not processed by the 26S proteasome (data not shown), which might suggest the involvement of the aforementioned different proteasomes or of other so far unidentified molecule(s) in the processing of those peptides. However, the processing mechanisms mediated by the proteasome are intriguing in general, because while it is generally accepted that the immunoproteasome is able to dominantly generate a diverse array of epitopes (1), the processing of some Ags is catalyzed specifically by the standard proteasome (alias constitutive proteasome), but not the immunoproteasome (28). Yet, no one knows the mechanistic reason(s) for those different processing profiles.

Our results clearly demonstrate the allele-specific effect of PA28 on the expression of MHC class I, regardless of the fact that these effects are positive or negative. Pool sequencing of peptides eluted from MHC class I molecules of cells, or large sequencing of individual peptides, would provide more direct information on the effect of PA28 in epitope generation. We could not identify a common reason for the inefficient processing of K^d ligands by the PA28-20S proteasome in this study. However, anchor residues of MHC class I ligands (29) might be, at least in part, involved in the allele-specific effect of PA28. Thus, it is possible that tyrosine anchor residue of K^d ligand is responsible for inefficient processing of many of K^d ligands. We are currently focusing on this issue. Mouse K^d-binding peptides have motif xYxxxxxxL. Because human HLA-A24 ligands also contain the same binding motif, it will be interesting to investigate whether processing of human HLA-A24 ligands is also down-regulated by the PA28 proteasome. Should these findings apply also to Ag processing and MHC class I expression in humans, they would have a great impact on our understanding of the immune system, and have practical implications especially for vaccination strategies in cancer and infectious diseases. Further studies are necessary to fully demonstrate the role of PA28 in allele-specific Ag processing.

	Acknowledgments

 
We are grateful to T. Hiroiwa and C. Kajiwara for preparation of recombinant proteins. We also thank D. Tsubokawa for assistance in proteasome purification and for conducting LC/MS.

	Disclosures

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
The authors have no financial conflict of interest.

	Footnotes

 
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 Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports and Culture, Japan.

² T.Y. and H.S. contributed equally to this work.

³ Address correspondence and reprint requests to Dr. Heiichiro Udono, Laboratory for Immunochaperones, Research Center for Allergy and Immunology, RIKEN Yokohama Institute, Tsurumi, Yokohama 230-0045, Japan. E-mail address: udonoh{at}rcai.riken.jp

⁴ Abbreviations used in this paper: siRNA, small-interfering RNA; LC, liquid chromatography; MS, mass spectrometry; VSV, vesicular stomatitis virus; M, macrophage; MCMV, murine CMV; CSP, circumsporozoite protein; NP, nucleoprotein.

Received for publication April 23, 2008. Accepted for publication May 14, 2008.

	References

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 

1. Tanaka, K., M. Kasahara. 1998. The MHC class I ligand-generating system: roles of immunoproteasomes and the interferon--inducible proteasome activator PA28. Immunol. Rev. 163: 161-176. [Medline]
2. Kloetzel, P. M., F. Ossendorp. 2004. Proteasome and peptidase function in MHC-class-I-mediated antigen presentation. Curr. Opin. Immunol. 16: 76-81. [Medline]
3. Rock, K. L., I. A. York, T. Saric, A. L. Goldberg. 2002. Protein degradation and the generation of MHC class I-presented peptides. Adv. Immunol. 80: 1-70. [Medline]
4. Fruh, K., Y. Yang. 1999. Antigen presentation by MHC class I and its regulation by interferon . Curr. Opin. Immunol. 11: 76-81. [Medline]
5. Rechsteiner, M., C. Realini, V. Ustrell. 2000. The proteasome activator 11 S REG (PA28) and class I antigen presentation. Biochem. J. 345: (Pt. 1):1-15. [Medline]
6. Dick, T. P., T. Ruppert, M. Groettrup, P. M. Kloetzel, L. Kuehn, U. H. Koszinowski, S. Stevanovic, H. Schild, H. G. Rammensee. 1996. Coordinated dual cleavages induced by the proteasome regulator PA28 lead to dominant MHC ligands. Cell 86: 253-262. [Medline]
7. Shimbara, N., H. Nakajima, N. Tanahashi, K. Ogawa, S. Niwa, A. Uenaka, E. Nakayama, K. Tanaka. 1997. Double-cleavage production of the CTL epitope by proteasomes and PA28: role of the flanking region. Genes Cells 2: 785-800. [Abstract]
8. Whitby, F. G., E. I. Masters, L. Kramer, J. R. Knowlton, Y. Yao, C. C. Wang, C. P. Hill. 2000. Structural basis for the activation of 20S proteasomes by 11S regulators. Nature 408: 115-120. [Medline]
9. van Hall, T., A. Sijts, M. Camps, R. Offringa, C. Melief, P. M. Kloetzel, F. Ossendorp. 2000. Differential influence on cytotoxic T lymphocyte epitope presentation by controlled expression of either proteasome immunosubunits or PA28. J. Exp. Med. 192: 483-494. [Abstract/Free Full Text]
10. Yamano, T., S. Murata, N. Shimbara, N. Tanaka, T. Chiba, K. Tanaka, K. Yui, H. Udono. 2002. Two distinct pathways mediated by PA28 and hsp90 in major histocompatibility complex class I antigen processing. J. Exp. Med. 196: 185-196. [Abstract/Free Full Text]
11. Murata, S., H. Udono, N. Tanahashi, N. Hamada, K. Watanabe, K. Adachi, T. Yamano, K. Yui, N. Kobayashi, M. Kasahara, et al 2001. Immunoproteasome assembly and antigen presentation in mice lacking both PA28 and PA28β. EMBO J. 20: 5898-5907. [Medline]
12. Tanahashi, N., Y. Murakami, Y. Minami, N. Shimbara, K. B. Hendil, K. Tanaka. 2000. Hybrid proteasomes: induction by interferon- and contribution to ATP-dependent proteolysis. J. Biol. Chem. 275: 14336-14345. [Abstract/Free Full Text]
13. Baumeister, W., J. Walz, F. Zuhl, E. Seemuller. 1998. The proteasome: paradigm of a self-compartmentalizing protease. Cell 92: 367-380. [Medline]
14. Cascio, P., M. Call, B. M. Petre, T. Walz, A. L. Goldberg. 2002. Properties of the hybrid form of the 26S proteasome containing both 19S and PA28 complexes. EMBO J. 21: 2636-2645. [Medline]
15. Moore, M. W., F. R. Carbone, M. J. Bevan. 1988. Introduction of soluble protein into the class I pathway of antigen processing and presentation. Cell 54: 777-785. [Medline]
16. Udono, H., T. Yamano, Y. Kawabata, M. Ueda, K. Yui. 2001. Generation of cytotoxic T lymphocytes by MHC class I ligands fused to heat shock cognate protein 70. Int. Immunol. 13: 1233-1242. [Abstract/Free Full Text]
17. Porgador, A., J. W. Yewdell, Y. Deng, J. R. Bennink, R. N. Germain. 1997. Localization, quantitation, and in situ detection of specific peptide-MHC class I complexes using a monoclonal antibody. Immunity 6: 715-726. [Medline]
18. Bloom, M. B., D. Perry-Lalley, P. F. Robbins, Y. Li, M. el-Gamil, S. A. Rosenberg, J. C. Yang. 1997. Identification of tyrosinase-related protein 2 as a tumor rejection antigen for the B16 melanoma. J. Exp. Med. 185: 453-459. [Abstract/Free Full Text]
19. Uenaka, A., T. Ono, T. Akisawa, H. Wada, T. Yasuda, E. Nakayama. 1994. Identification of a unique antigen peptide pRL1 on BALB/c RL1 leukemia recognized by cytotoxic T lymphocytes and its relation to the Akt oncogene. J. Exp. Med. 180: 1599-1607. [Abstract/Free Full Text]
20. Glickman, M. H., D. M. Rubin, V. A. Fried, D. Finley. 1998. The regulatory particle of the Saccharomyces cerevisiae proteasome. Mol. Cell. Biol. 18: 3149-3162. [Abstract/Free Full Text]
21. Ma, C. P., P. J. Willy, C. A. Slaughter, G. N. DeMartino. 1993. PA28, an activator of the 20 S proteasome, is inactivated by proteolytic modification at its carboxyl terminus. J. Biol. Chem. 268: 22514-22519. [Abstract/Free Full Text]
22. Song, X., J. von Kampen, C. A. Slaughter, G. N. DeMartino. 1997. Relative functions of the and β subunits of the proteasome activator, PA28. J. Biol. Chem. 272: 27994-28000. [Abstract/Free Full Text]
23. Weiss, W. R., S. Mellouk, R. A. Houghten, M. Sedegah, S. Kumar, M. F. Good, J. A. Berzofsky, L. H. Miller, S. L. Hoffman. 1990. Cytotoxic T cells recognize a peptide from the circumsporozoite protein on malaria-infected hepatocytes. J. Exp. Med. 171: 763-773. [Abstract/Free Full Text]
24. Aki, M., N. Shimbara, M. Takashina, K. Akiyama, S. Kagawa, T. Tamura, N. Tanahashi, T. Yoshimura, K. Tanaka, A. Ichihara. 1994. Interferon- induces different subunit organizations and functional diversity of proteasomes. J. Biochem. 115: 257-269. [Abstract/Free Full Text]
25. Shimbara, N., K. Ogawa, Y. Hidaka, H. Nakajima, N. Yamasaki, S. Niwa, N. Tanahashi, K. Tanaka. 1998. Contribution of proline residue for efficient production of MHC class I ligands by proteasomes. J. Biol. Chem. 273: 23062-23071. [Abstract/Free Full Text]
26. Ustrell, V., L. Hoffman, G. Pratt, M. Rechsteiner. 2002. PA200, a nuclear proteasome activator involved in DNA repair. EMBO J. 21: 3516-3525. [Medline]
27. Ortega, J., J. B. Heymann, A. V. Kajava, V. Ustrell, M. Rechsteiner, A. C. Steven. 2005. The axial channel of the 20S proteasome opens upon binding of the PA200 activator. J. Mol. Biol. 346: 1221-1227. [Medline]
28. Morel, S., F. Levy, 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: 107-117. [Medline]
29. Rammensee, H. G., K. Falk, O. Rotzschke. 1993. Peptides naturally presented by MHC class I molecules. Annu. Rev. Immunol. 11: 213-244. [Medline]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Yamano, T. Articles by Udono, H. Search for Related Content PubMed PubMed Citation Articles by Yamano, T. Articles by Udono, H.