Proteasome inhibitors reconstitute the presentation of cytotoxic T-cell epitopes in Epstein-Barr virus–associated tumors

Authors

  • Riccardo Gavioli,

    1. Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden
    2. Dipartimento di Biochimica e Biologia Molecolare, Università di Ferrara, Ferrara, Italy
    Search for more papers by this author
  • Simona Vertuani,

    1. Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden
    2. Dipartimento di Biochimica e Biologia Molecolare, Università di Ferrara, Ferrara, Italy
    Search for more papers by this author
  • Maria G. Masucci

    Corresponding author
    1. Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden
    • Microbiology and Tumor Biology Center, Karolinska Institutet, Box 280, S-171 77 Stockholm, Sweden
    Search for more papers by this author
    • Fax: +46-8-331399


Abstract

EBV-infected cells and EBV-associated tumors may evade CTL recognition by defective antigen processing, resulting in poor presentation of CTL epitopes. Since the proteasome is the major source of MHC class I–presented peptides, we analyzed the effect of proteasome inhibitors on the expression of surface HLA class I and the generation of EBV-derived CTL epitopes presented by the HLA-A2 and HLA-A11 alleles. Treatment with covalent and reversible inhibitors of the proteasome partially reduced the total and allele-specific expression of surface HLA class I in EBV-carrying LCLs. HLA-A2 expression was also decreased by treatment with leupeptin and bestatin, while HLA-A11 expression was affected by treatment with phenanthroline. Despite their general inhibitory effect on HLA class I expression, all proteasome inhibitors tested enhanced the presentation of 2 subdominant HLA-A2 epitopes from EBV LMP1 and LMP2, while the presentation of the immunodominant HLA-A11-restricted epitope from EBNA4 was inhibited by MG132 and lactacystin and increased by ZL3VS. Treatment with ZL3VS restored the presentation of endogenously expressed EBNA4 in 1 HLA-A11-positive BL cell line. These findings suggest that specific inhibitors of the proteasome may be used to increase the antigenicity of virus-infected and malignant cells that are per se inefficient at generating particular CTL target epitopes. © 2002 Wiley-Liss, Inc.

CTLs recognize antigenic peptides expressed at the surface of target cells in association with MHC class I molecules.1 Class I–presented peptides are 8–10 amino acids long and derived from the intracellular degradation of cytosolic, nuclear and membrane proteins. The major enzymatic activity responsible for the generation of class I–associated peptides is the proteasome, an essential component of the ATP-dependent pathway of protein degradation.2 The enzymatic core of the proteasome is the 20S subunit, a 700 kDa cylindrical particle with multiple peptidase activities that is composed of 2 outer rings of 7 α subunits and 2 inner rings of 7 β subunits.3 The 20S proteasome is capped at each end by a 19S cap and/or the PA28 regulatory complex. Proteolysis occurs in the inner chamber of this particle and is catalyzed through a nucleophilic attack on the peptide bond by the N-terminal threonine of the β1, β2 and β5 subunits.3 These subunits can be replaced by the IFN-regulated Lmp2, Lmp7 and MECL-1 subunits, which alter the proteolytic activity of the proteasome and favor the generation of peptides with appropriate C termini for binding to MHC class I molecules.4

The critical involvement of the proteasome in MHC class I– restricted antigen processing was established with the help of specific inhibitors.5 It is now widely recognized that treatment with membrane-permeable inhibitors of the proteasome strongly affects the generation of antigenic peptides and impairs MHC class I surface expression in different cell types.6 However, not all class I epitopes require the proteasome for generation. The TAP-independent expression of HLA-A2 and expression of HLA-A11 appear to be largely insensitive to the effects of proteasome inhibition.7 Furthermore, the presentation of certain CTL epitopes appears to be insensitive to proteasome inhibition, and some epitopes are generated in greater amount in cells subjected to partial inhibition of proteasomes.8, 9, 10, 11

Virus-infected cells and tumors may evade CTL recognition by altering the expression of HLA class I or the activity of the antigen-processing pathway, resulting in poor presentation of CTL target epitope.12 EBV-associated BL provides one of the best examples of a tumor that evades CTL recognition.13 Several consistent features of the BL cell phenotype may contribute to CTL resistance, including HLA class I and TAP downregulation.14, 15, 16 However, correction of these defects failed to reconstitute antigen presentation in some BL lines, suggesting that additional defects may be involved.16 This possibility is further corroborated by studies on other EBV-associated malignancies, such as NPC and HD. Although these tumors are obviously capable of escaping EBV-specific CTL control in vivo, the malignant cells do not exhibit consistent defects of TAP or HLA class I expression.15, 16, 17 We previously reported that the proteasome is functionally impaired in BL and HD cell lines.18, 19, 20 In line with these findings, BL cells are poorly antigenic and fail to present MHC class I– restricted epitopes from endogenous proteins.16

In this investigation, we analyzed the effect of different proteasome inhibitors on the generation of subdominant and immunodominant viral epitopes in EBV-transformed LCLs of normal B-cell origin and in EBV-carrying tumors. We show that different inhibitors could enhance or impair the presentation of an immunodominant HLA-A11-restricted epitope from EBNA4, while the presentation of 2 subdominant HLA-A2-restricted epitopes from LMP1 and LMP2 was in every case enhanced. Thus, selective and partial inhibition of the proteasome may be a strategy to modulate the presentation of antigenic epitopes in poorly immunogenic tumors.

Abbreviations:

Ac-YVAD-AMC, H-Ala-Ala-Phe-aminomethylcoumarin; BCA, bicinchoninic acid; BFA, bifluoroacetic acid; BL, Burkitt's lymphoma; Boc-LRR-AMC, Boc-Leu-Arg-Arg-aminomethylcoumarin; CLG, CLGGLLTMV; CTL, cytotoxic T lymphocyte; DTT, dithiothreitol; EBNA4, Epstein-Barr virus nuclear antigen 4; EBV, Epsein-Barr virus; FITC, fluorescein isothiocyanate; HD, Hodgkin's disease; IVT, IVTDFSVIK; LCL, lymphoblastoid cell line; LMP1, latent membrane protein 1; MAb, monoclonal antibody; MG132, carboxybenzyl-leucyl-leucyl-leucinal; NLVS, 4-hydroxy-5-iodo-3-nitrophenyl-Leu-Leu-Leu-vinyl sulfone; NPC, nasopharyngeal carcinoma; PBL, peripheral blood lymphocyte; PHA, phytohemagglutinin; Suc-LLVY-AMC, succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin; TAP, transporter associated with antigen processing; YLL, YLLEMLWRL; ZL3VS, carboxybenzyl-Leu-Leu-Leu vinyl sulfone.

MATERIAL AND METHODS

Chemicals

The proteasome inhibitors MG132, lactacystin and epoxomicin were from Affiniti (Exeter, UK); NLVS was kindly provided by Dr. H. Ploegh (Harvard Medical School, Boston, MA), and ZL3VS was synthesized by Dr. M. Marastoni (University of Ferrara, Ferrara, Italy). The aminopeptidase inhibitor bestatin, the metalloproteinase inhibitor 1-10-phenanthroline and the serine-cysteine protease inhibitor leupeptin were from Sigma (St. Louis, MO). Brefeldin A was from Affiniti. Suc-LLVY-AMC, Boc-LRR-AMC and Ac-YVAD-AMC, used to assay the chymotryptic, tryptic- and caspase-like activities of the proteasome, respectively, were from Sigma.

Synthetic Peptides

All peptides were synthesized by a solid-phase method using a continuous-flow instrument with on-line UV monitoring, as previously described.21 Crude deprotected peptides were purified by HPLC to >98%. Structure verification was performed by elemental and amino acid analyses and mass spectrometry. Peptide stocks were dissolved in DMSO at a concentration of 10–2 M, kept at –20°C and diluted in PBS before use.

Cell Lines and CTLs

LCLs were established by in vitro infection of normal B lymphocytes from healthy donors with the B95.8 strain of EBV and expanded in RPMI-1640 medium supplemented with 10% FCS. The E95B-BL28 cell line was obtained by infection of the EBV-negative BL28 line with the B95.8 virus.22 The .174/T2 cell line (T2) was obtained by fusion of the peptide transporter mutant .174 LCL with the T-cell line CEM.23 PHA-activated blasts were obtained by stimulation of PBLs with 1 μg/ml purified PHA (Wellcome Diagnostics, Dartford, UK) for 3 days and expanded in medium supplemented with 10 U/ml IL-2, as described previously.24 HLA-A11-restricted CTL cultures reacting against the EBV epitope EBNA4416–424 were obtained by stimulation of lymphocytes from EBV-seropositive donor BK (HLA-A2,11, -B7,35) with the autologous B95.8 virus–transformed LCL. After 2 or 3 consecutive restimulations, cultures were expanded in complete medium supplemented with 10 U/ml recombinant IL-2. HLA-A2-restricted EBV-specific CTL cultures reacting against the epitopes LMP2426–434 and LMP1125–133 were obtained by stimulation of monocyte-depleted PBLs from EBV-seropositive donor RG (HLA-A2, -B8,44) with peptide-pulsed T2 cells.21, 25, 26 The second and third stimulations were performed under the same conditions on days 7 and 14, respectively. Medium was supplemented from day 8 with 10 U/ml recombinant IL-2. The specificity of CTL cultures was tested against a panel of EBV-positive and-negative targets, including the autologous LCLs and PHA-activated blasts, allogeneic LCLs sharing HLA-A2 and HLA-A2-mismatched LCLs.

Peptide Stripping and Flow-cytometric Assays

For peptide stripping, cells were washed in PBS and pellets were gently resuspended in a citrate buffer phosphate (pH 3) buffer (0.131 M citric acid, 0.066 M Na2HPO4) for 2 min on ice. The suspension was then neutralized by dilution with RPMI/10% FCS, and cells were washed twice. Aliquots of cells were resuspended in 2 ml complete medium in the presence or absence of the indicated concentrations of proteasome inhibitors. After 5 hr at 37°C, cells were washed once and resuspended in 50 μl of PBS containing saturating concentrations of the indicated MAbs and incubated on ice for 30 min. Binding of MAbs was detected by incubation with FITC-conjugated rabbit antimouse IgG (Dako, Copenhagen, Denmark). Fluorescence intensity was measured with a FACSort analyzer (Becton Dickinson, Mountain View, CA).

In Vitro Effect of Proteasome Inhibitors

Cells (5 × 108) were washed in cold PBS and resuspended in buffer containing 50 mM TRIS-HCl (pH 7.5), 5 mM MgCl2, 1 mM DTT (Sigma), 2 mM ATP and 250 mM sucrose. Glass beads equivalent to the volume of the cell suspension were added, and the mixture was vortexed for 1 min at 4°C. Beads and cell debris were removed by 5 min centrifugation at 1,000g, followed by 20 min centrifugation at 10,000g.27 Lysates were cleared by ultracentrifugation for 1 hr at 100,000g, and supernatants were further ultracentrifuged for 5 hr at 100,000g. Proteasome-containing pellets were resuspended in 0.5 ml of homogenization buffer [50 mM TRIS-HCl (pH 7.5), 100 mM KCl, 15% glycerol]. Protein concentration was determined using the BCA protocol (Pierce, Rockford, IL). Fluorogenic substrates Suc-LLVY-AMC, Boc-LRR-AMC and Ac-YVAD-AMC were used to measure chymotryptic-like, tryptic-like and caspase-like activities, respectively. Semipurified proteasomes (10 μl), pretreated or not with inhibitors for 30 min at 37°C, were assayed at 37°C for 45 min using the different peptide substrates in a buffer containing 50 mM TRIS-HCl (pH 7.5), 5 mM MgCl2 and 1 mM DTT (final volume 100 μl). The reaction was quenched with 1 ml 1% SDS and fluorescence determined by fluorimeter (Perkin-Elmer, Beaconsfield, UK) with excitation at 380 nm and emission at 440 nm.18 Data are expressed as the percent inhibition relative to untreated proteasomal preparations.

In Vivo Effect of Proteasome Inhibitors

LCLs (5 × 107 cells) were treated or not with the indicated concentrations of the different proteasome inhibitors for 4 hr at 37°C. Cells were washed in cold PBS and resuspended in buffer containing 50 mM TRIS-HCl (pH 7.5), 5 mM MgCl2, 1 mM DTT, 2 mM ATP and 250 mM sucrose. Proteasomes were semipurified as described above. Proteasome-containing pellets were resuspended in 0.5 ml of homogenization buffer, and protein concentration was determined using the BCA protocol. Fluorogenic substrates Suc-LLVY-AMC, Boc-LRR-AMC and Ac-YVAD-AMC were used to measure the activities of the different proteasomal preparations as described above. Data are expressed as the percent inhibition relative to equal amounts of proteasomal preparations isolated from untreated LCLs.

Cytotoxicity Assay

Target cells were labeled with Na51CrO4 for 90 min at 37°C. Cytotoxicity tests were routinely run at different E:T ratios in triplicate. Percent specific lysis was calculated as 100 × (cpm sample – cpm medium)/(cpm Triton X-100 – cpm medium).24 Spontaneous release was always <20%. Where indicated, labeled cells were peptide-stripped and treated with proteasome inhibitors as described above and 51Cr-release assays were run in the presence of BFA 5 μg/ml to block any further transport of MHC class I molecules.

RESULTS

Effect of Proteasome Inhibitors on HLA Class I Expression

Proteasome inhibitors are a valuable tool to study the role of the proteasome in the generation of MHC class I–presented peptide epitopes.5 We used some of these compounds to study the contribution of the proteasome to the generation of EBV-derived HLA class I epitopes in virus-infected normal and malignant cells. We first asked whether the proteasome plays a major role in the generation of the bulk of HLA class I-bound peptides in EBV-transformed LCLs. Surface class I molecules were disrupted by exposing LCLs to a pH 3 buffer, and the appearance of de novo synthesized class I molecules was monitored over time. As shown in Figure 1, acid treatment induced complete loss of HLA class I–peptide complexes, while re-expression to normal levels occurred within 4–6 hr of incubation at 37°C. The effect of the proteasome inhibitor lactacystin5, 28 on the re-expression of class I molecules was tested by incubating acid-stripped LCLs for 4 hr in the presence of different concentrations of inhibitor. HLA class I expression and cell viability were monitored by FACS analysis and trypan blue exclusion, respectively. As illustrated by the representative experiment in Figure 2, lactacystin induced a dose-dependent inhibition of HLA class I expression with 50% inhibition at 20 μM lactacystin, a concentration that had little effect on cell viability (>25%).

Figure 1.

Kinetics of reappearance of HLA class I molecules. FY-B1 LCLs were treated or not (–) for 2 min at pH 3 on ice. HLA class I expression was evaluated by immunofluorescence at the indicated time points using the HLA class I-specific W6/32 MAb. Data are expressed as mean fluorescence intensity from 1 representative experiment of 3.

Figure 2.

Re-appearance of HLA class I molecules in presence of the proteasome inhibitor lactacystin. FY-B1 LCLs were peptide-stripped for 2 min at pH 3 on ice and then cultured with the indicated concentrations of proteasome inhibitor for 4 hr at 37°C. HLA class I expression was evaluated by immunofluorescence and cell viability by trypan blue exclusion. One representative experiment of 3.

Having defined the kinetics of MHC class I reconstitution, we repeated this type of assay on a panel of HLA-A2- and -A11-positive LCLs using other proteasome inhibitors.28, 29, 30, 31, 32 In addition, parallel experiments were performed using different protease inhibitors,33, 34, 35, 36 including inhibitors of metalloproteases since this class of enzymes contributes to the generation of class I binding peptides.37 The inhibitors listed in Table I were used at concentrations that did not affect cell viability (not shown).

Table I. PROTEASE INHIBITORS USED IN OUR STUDY
InhibitorSpecificityReferences
1,10-PhenanthrolineAll metalloproteases and caspase-135
BestatinMetalloaminopeptidases35,36
EpoxomicinProteasome chymotrypsin and trypsin activities31,32
LactacystinProteasome chymotrypsin and trypsin activities5,28
LeupeptinTrypsin-like proteases and cystein proteases33,34
MG132Proteasome29
NLVSProteasome chymotrypsin and trypsin activities30
ZL3VSProteasome chymotrypsin and trypsin activities30

Expression of total HLA class I, HLA-A2 and HLA-A11 was measured by FACS analysis in acid-stripped LCLs that had been cultured for 4 hr in the presence of the various protease inhibitors. Treatment with the proteasome inhibitors NLVS, ZL3VS, MG132, lactacystin and epoxomicin induced only partial inhibition of total class I expression (Fig. 3), and similar levels of inhibition were also observed for HLA-A2 and HLA-A11. To exclude the possibility that the partial effect was due to the presence of an intracellular pool of class I binding peptides generated before the addition of proteasome inhibitors, cells were pretreated with proteasome inhibitors for 2 hr before acid stripping. This treatment did not affect the overall class I recovery (not shown). HLA-A2 expression was also partially affected by pretreatment of LCLs with 100 μM of the serine-cysteine proteinase inhibitor leupeptin or the aminopeptidase inhibitor bestatin, while expression of HLA-A11 was not affected. Conversely, the metalloproteinase inhibitor 1–10-phenanthroline affected HLA-A11, but not HLA-A2, expression. Taken together, these results indicate that the proteasomes and other intracellular peptidases participate in the generation of MHC class I–presented peptides. Different combinations of peptidases appear to contribute to the production of peptides binding to a specific class I allele.

Figure 3.

Effect of protease inhibitors on the expression of newly synthesized HLA class I molecules. FY-B1 and JAC-B2 LCLs were treated for 2 min at pH 3 on ice and then cultured for 4 hr at 37°C with the indicated inhibitors. Total HLA class I expression (a), HLA-A2 expression (b) and HLA-A11 expression (c) were evaluated using the specific W6/32, HB54 and HB164 MAbs, respectively. Data are expressed as percent inhibition of mean fluorescence intensity relative to untreated control cells. Mean ± SD of 3 independent experiments.

Proteasome Inhibitors Affect the Generation of HLA-A2- and HLA-A11-restricted Epitopes

Acid-stripped LCLs were used to evaluate the role of the proteasome in the generation of different EBV-derived epitopes. For this purpose, we chose 2 HLA-A2-restricted epitopes derived from viral antigens that are expressed in LCLs as well as in some EBV-associated tumors, the LMP2 epitope CLG and the LMP1 epitope YLL,13 and 1 HLA-A11-restricted epitope derived from EBNA4, IVT, that is expressed in LCLs and in EBV-positive BL lines with latency III.24 Lysis of HLA-A2- and HLA-A11-positive LCLs was abolished by acid treatment, while 70–80% of maximal killing was reconstituted after incubation for 5 hr. As expected, this reconstitution was blocked by addition of BFA, which inhibits the surface export of newly synthesized class I molecules (not shown). Treatment with all 3 proteasome inhibitors tested strongly enhanced the lysis of LCLs by CLG- and YLL-specific, HLA-A2-restricted CTLs, while leupeptin, bestatin and phenanthroline had no significant effect (Fig. 4a,b). In contrast, lysis of LCLs by HLA-A11-restricted, IVT-specific CTLs was inhibited by treatment with the proteasome inhibitors MG132 and lactacystin and by leupeptin, bestatin and phenanthroline, while treatment with the proteasome inhibitor ZL3VS resulted in significantly higher target cell lysis, indicating enhanced re-expression of the IVT epitope (Fig. 4c).

Figure 4.

Effect of protease inhibitors on presentation of MHC class I–restricted CTL epitopes. After treatment at pH 3, cells were incubated for 4 hr at 37°C with or without (–) the indicated inhibitors and used as target in cytotoxicity assays in the presence of 5 μg/ml brefeldin A to block additional transport of class I molecules from the endoplasmic reticulum. Results are expressed as percent specific lysis. (a) Re-expression of the HLA-A2-restricted LMP2-derived epitope CLG in BK-B5 LCLs (HLA-A2, -A11, -B7, -B35) was tested using the CLG-specific CTL culture RG. One representative experiment of 3. The range of killing of untreated BK-B5 LCLs was 10–22 % specific lysis. (b) Re-expression of the HLA-A2-restricted, LMP1-derived epitope YLL in RG-B1 LCLs (HLA-A2, -B8, -B44) was tested using the YLL-specific CTL culture RG. One representative experiment of 2 (lactacystin and Zl3VS) or 3 (MG132) experiments. The range of killing of untreated RG-B1 LCLs was 10–19 % specific lysis. (c) Re-expression of the HLA-A11-restricted, EBNA4-derived IVT epitope at the cell surface of JAC-B2 LCLs (HLA-A1, -A11, -B49, -B55) was tested using the IVT-specific BK CTL culture. One representative experiment of 3. Range of killing of untreated BK-B5 LCLs was 20–40% specific lysis.

The effect of proteasome inhibitors was further evaluated in titration experiments. Treatment of HLA-A2-positive LCLs with increasing amounts of proteasome inhibitors resulted in a concentration-dependent increase of cell lysis by CLG- and YLL-specific CTLs, with an up to 100% increase at concentrations ≥15 μM (Fig. 5a,b). Lysis of HLA-A11-positive LCLs was inhibited by increasing concentrations of lactacystin and MG132, while a dose-dependent increase was detected after treatment with ZL3VS (Fig. 5c). The different effect of proteasome inhibitors on the generation of the IVT epitope could be due to different inhibition of the 3 proteolytic activities of proteasomes. To test this possibility, we evaluated the effect of MG132, lactacystin and ZL3VS on proteasome activity in vitro and in vivo.In vitro assays were performed with semipurified proteasomes isolated from LCLs and pretreated with the indicated concentrations of inhibitors for 30 min before the assay. As shown in Table II, similar levels of inhibition were obtained with all compounds. In vivo assays were performed with semipurified proteasomes isolated from LCLs that were pretreated for 4 hr with the indicated concentrations of inhibitors. Also, in this case, all proteasomal preparations showed comparable activity, suggesting that the compounds have similar effects in vitro and in vivo (Table II).

Figure 5.

Effect of proteasome inhibitors on presentation of EBV-derived CTL epitopes. After treatment at pH 3, cells were incubated for 4 hr at 37°C with increasing concentrations of inhibitors and used as target in cytotoxicity assays in the presence of 5 μg/ml brefeldin A to block additional transport. Results are expressed as percent specific lysis. One representative experiment of 3.

Table II. EFFECT OF INHIBITORS ON THE MAJOR PROTEOLYTIC ACTIVITIES OF SEMIPURIFIED PROTEASOMES
 % Inhibition in cell lysates% Inhibition in treated cells
Chymotryptic1Tryptic2Caspase-like3ChymotrypticTrypticCaspase-like
  1. MG132, 30 μM; lactacystin, 20 μM; ZL3VS, 30 μM. Peptidase activity was measured using the following fluorescence labeled substrates:–1Suc-LLVY-AMC,–2Boc-LRR-AMC and–3Ac-YVAD-AMC.

MG132905266927713
Lactacystin94494890465
ZL3VS844144964113

Proteasome Inhibitors Reconstitute Presentation of the HLA-A11-associated IVT Epitope in BL Cells

We have previously shown that presentation of the IVT epitope is impaired in the EBV-positive E95B-BL28 cell line, which expresses EBNA4 at levels comparable to regular LCLs.16 Upregulation of HLA-A11 molecules and reconstitution of TAP activity by 1-, 2- or 3-day treatment with IFN-γ did not restore presentation of the IVT epitope, suggesting a defect in epitope generation (not shown). Indeed, proteasomes from BL cells have different enzymatic activities compared to proteasomes isolated from LCLs and exhibit a different pattern of cleavage of synthetic peptides in vitro.18, 19, 20 To test the effect of proteasome inhibitors on the presentation of the IVT epitope, E95B-BL28 cells were cultured overnight in the presence of 1,000 U/ml INF-γ and then treated for 4 hr with the previously described panel of inhibitors (Fig. 6). In accordance with previous results, the E95B-BL28 cell line was not lysed by IVT-specific CTLs.16 Treatment with the proteasome inhibitors lactacystin and MG132 or with phenanthroline, bestatin and leupeptin did not affect the presentation of this epitope. In contrast, a significant level of lysis was regularly observed in cells treated with ZL3VS.

Figure 6.

Effect of proteasome inhibitors on the presentation of endogenously expressed EBNA4 epitopes in the E95B-BL28 cell line. Cells were treated overnight with 1,000 U/ml INF-γ, incubated for 4 hr at 37°C with the indicated inhibitors and used as target in cytotoxicity assays. Results are expressed as percent specific lysis. One representative experiment of 3.

DISCUSSION

We used different inhibitors to evaluate the role of the proteasome and other cellular proteases on the production of peptides that allow surface expression of HLA class I complexes and on the generation of HLA-A2- and HLA-A11-restricted epitopes in EBV-infected normal and malignant cells. Treatment with proteasome inhibitors delayed the reappearance of properly folded class I molecules at the surface of acid-stripped LCLs, and some inhibition was also observed when re-expression of HLA-A2 was monitored in cells treated with inhibitors of metalloaminopeptidases and cysteine peptidases, such as leupeptin and bestatin. Partial inhibition was also observed when the effect of proteasome inhibitors was monitored on HLA-A11 re-expression, but a different inhibitor of metalloproteases, phenanthroline, appeared to be effective. As already suggested by others, these data indicate that cellular proteases may participate, together with proteasomes or by parallel pathways, in the generation of peptides for MHC class I loading.7–9, 37–39 Given the diversity of HLA class I binding motifs, it is perhaps not surprising that the production of peptides binding to a specific class I allele appears to be dependent on different combinations of proteases.

Our finding that the presentation of 2 HLA-A2-restricted epitopes from LMP1 and LMP2 was enhanced in proteasome inhibitor–treated cells while the presentation of 1 HLA-A11- restricted epitope was inhibited by treatment with lactacystin and MG132 but not ZL3VS reinforces the notion of a complex interplay between different proteolytic activities in the generation of class I binding peptides. Production of certain epitopes appears to be more efficient in cells with partially blocked proteasomes. This suggests that the proteasome may participate both in the production and in the destruction of antigenic peptides and the final outcome of antigen presentation may be the result of a dynamic balance between positive and negative regulators of proteolysis. These may change in different cells or different stages of cell activation/differentiation. There is a discrepancy between our finding that proteasome inhibitors enhance the presentation of the HLA-A2-restricted CLG epitope and the report by Lautscham et al.40 that presentation of this epitope is dependent on the proteasome. This apparent contradiction is likely explained by the use of different cell types, LCLs in our case and epithelial cells in the experiments of Lautscham et al.,40 which express proteasome with different subunit composition and enzymatic activity. LCL cells express high levels of the IFN-regulated β subunits Lmp2, Lmp7 and MECL-1 and the alternative regulator PA28α/β.18 Indeed, in different model systems, these immunoproteasomes failed to generate HLA-A2 binding CTL epitope.41, 42 Furthermore, 2 subdominant Db-restricted epitopes were presented with more efficiency in Lmp2 knockout compared to Lmp2-expressing mice, demonstrating that epitope generation strongly depends on proteasome subunit composition and activity.43 The molecular mechanisms of this peptide selectivity are not fully understood. However, association with the PA28α/β regulator may constitutively open access to the proteolytic chamber of the 20S proteasome.44, 45 This is likely to cause radical changes in the flow-through of substrates, with possible consequences on the length of the peptide products and the type or frequency of the proteolytic cleavages.

Our observation that the presentation of several EBV-derived epitopes is enhanced in virus-transformed LCLs by treatment with proteasome inhibitors offers a likely explanation for one puzzling paradox of the EBV system. EBV is a strong immunogen, and high levels of specific CTL precursors are detected in the blood during primary infection and in healthy virus carriers.13 EBV-carrying LCLs are efficient stimulators of EBV-specific CTL responses in vivo and in vitro, yet these cells are often resistant to lysis by EBV-specific effectors in standard 51Cr-release assays and demonstration of specific recognition requires overexpression of the endogenous antigen by infection with recombinant vaccinia viruses or sensitization with the relevant synthetic peptide.46 In light of our present findings, this may be explained by excessive activity of the proteasome, resulting in rapid destruction of the majority of antigenic peptides generated in cells. The low residual amount of antigenic peptides may be sufficient for triggering memory T-cell responses since a lower density of MHC peptide complexes is required for induction of T-cell proliferation.47, 48 It is interesting to speculate on the implications of this finding for the CTL control of virus-infected cells in vivo. Adoptive transfer of EBV-specific CTLs has been successfully used to prevent and cure the expansion of LCL-like EBV-carrying blasts in immunosuppressed individuals.49, 50 It remains to be seen whether the in vivo growing blasts are more sensitive to CTL-mediated lysis than their in vitro growing counterparts, perhaps due to selective modulation of proteasome activity, or whether other effector functions are more relevant in vivo.

Treatment with the proteasome inhibitor ZL3VS restored the presentation of endogenously expressed EBNA4 in HLA-A11-positive BL cells. This is in line with our previous findings that a synthetic peptide corresponding to the immunodominant HLA-A11-restricted epitope from EBNA4 is efficiently cleaved by purified proteasome from BL cells while proteasomes from LCLs have a significantly lower activity in vitro.18 It remains to be seen why reconstitution of antigen presentation was achieved with ZL3VS while other inhibitors either had no effect or inhibited the presentation of this HLA-A11-restricted epitope in regular LCLs. It should be stressed that, while the specific inhibitors used in our study had similar effects on the 3 proteolytic activities of the proteasome when tested with fluorogenic tripeptides (Table II), the type of assay is not strictly representative of proteasome-dependent proteolysis since the true substrates of the proteasome are proteins or long peptides. Thus, the effect of different inhibitors on the hydrolysis of IVT precursors or the entire EBNA4 would be more relevant. Indeed, in a preliminary set of experiments, we observed the that ZL3VS was more effective than lactacystin at blocking the degradation of a synthetic IVT peptide by semipurified proteasomes (not shown), suggesting that the degradation of long peptides may be differentially affected by the 2 proteasome inhibitors.

Defects of antigen processing were thought to be of minor importance in the immunoescape of EBV-carrying BLs since tumor cells were believed to express only EBNA1, which escapes CTL recognition due to the presence of a Gly-Ala repeat that protects it from proteasomal processing.51 However, some EBV-carrying BLs stably express a broader repertoire of viral antigens, including the high m.w. EBNAs 3, 4 and 6 and LMP2 in vivo and in vitro (A.B. Rickinson, personal communication). Collectively, our findings stress the great importance of developing proteasome inhibitors that selectively affect the activity of single catalytic β subunits. While lacking the toxic effect of general inhibitors of the proteasome, these compounds may allow, alone or in combination, selective modulation of antigen processing in virus-infected and malignant cells, providing a new strategy to boost CTL-mediated protection.

Acknowledgements

S.V. was supported by a fellowship awarded by the European Commission Training and Mobility Program (grant ERBFMRXCT960026).

Ancillary