The Antiviral Immune Response in Mice Devoid of Immunoproteasome Activity

Antigen recognition by CTLs occurs through the interaction of their TCRs with peptide–MHC class I complexes. The proteasome is the main protease in the cytoplasm and the nucleus, which is responsible for the generation of most peptide ligands of MHC-I molecules (1, 2). It consists of α and β subunits that build a barrel-shaped complex of four rings with seven subunits each (called 20S proteasome) (3). The outer two rings consist of seven α subunits, and the inner two rings of seven β subunits, which bear the proteolytically active centers. Depending on the cell type and the presence or absence of IFN-γ or TNF-α, the three inducible β subunits, low molecular mass polypeptide (LMP) 2/β1i, LMP7/β5i, and multicatalytic endopeptidase complex-like–1 (MECL-1)/β2i, can, in addition to the corresponding constitutive subunits δ/β1, MB1/β5, and MC14/β2, enrich the cellular assortment of catalytically active β subunits.

Mice lacking one or more of the inducible immunosubunits have been generated and infected with commonly used laboratory strains of viruses, bacteria, and fungi. These studies have revealed a role for the immunoproteasome in shaping the CTL repertoire and pathogen clearance that has been ascribed to alterations in MHC-1 ligand generation. Recently, a novel function of immunoproteasomes in autoimmune diseases and T cell expansion has been proposed (4–7).

MECL-1–, LMP2-, and LMP7-deficient mice have been analyzed intensively (reviewed in Ref. 5). LMP2 (8) and MECL-1 gene-targeted mice (9) displayed in contrast to LMP7^−/− mice (10) no change in MHC-I cell surface expression, but the numbers of CD8⁺ cells were reduced in the spleen. Due to the altered proteasomal cleavage patterns in the presence of the inducible immunosubunits, some MHC class I ligands are generated preferentially by immunoproteasomes (8, 11), some only by constitutive proteasomes (12), and some epitopes are processed to the same extent by constitutive and immunoproteasomes (13), but the bulk of MHC-I ligands can still be generated in LMP2-, LMP7-, and MECL-1–deficient mice (8–10). The altered MHC-I peptidome on thymic APCs during positive and negative selection of CD8⁺ T cells leads to a different CD8⁺ T cell repertoire in LMP2-, LMP7-, and MECL-1–deficient mice (9, 14–16).

The requirement for immunoproteasomes for pathogen elimination varies markedly between infection models. Although immunoproteasome-deficient mice can cope with most pathogens, a role for β5i in the clearance of pathogens was shown in knockout mice postinfection with Listeria monocytogenes (17). The clearance of bacteria from the liver was not apparent by day 10, by which time bacterial burden in the spleen of wild-type mice had decreased. Infection with the protozoan parasite Toxoplasma gondii revealed an even more prominent phenotype (18). In contrast to wild-type mice, β5i-deficient mice were highly susceptible to infection with T. gondii and showed a reduced number of functional T cells. Testing of immunoproteasome-deficient mice is required to fully appreciate the contribution of individual proteasome subunits to the immune response to infectious agents.

In this study, we characterize LMP2^−/−/MECL-1^−/− double-deficient mice with the help of the well-described lymphocytic choriomeningitis virus (LCMV). With the help of these mice and the recently published LMP7-selective inhibitor ONX 0914, we generated mice that do not contain active immunoproteasome subunits. We show in this study that different immunoproteasome subunits have opposing effects on the same MHC-I epitope.

C57BL/6 mice (H-2^b) were originally purchased from Charles River. MECL-1 (9), LMP2 (8), and LMP7 (10) gene-targeted mice were provided by J. Monaco (Department of Molecular Genetics, Cincinnati Medical Center, Cincinnati, OH). LMP2^−/−/MECL-1^−/− double-deficient mice were generated by crossing the F₁ generation of LMP2^−/− × MECL-1^−/− mice. Mice were kept in a specific pathogen-free facility and used at 6–10 wk of age. Animal experiments were approved by the review board of Regierungspräsidium Freiburg. LCMV-WE was originally obtained from F. Lehmann-Grube (Hamburg, Germany) and propagated on the fibroblast line L929. LCMV was titrated on MC57 cells, as previously described (19). Mice were infected with 200 PFU LCMV-WE i.v. All media were purchased from Invitrogen-Life Technologies (Karlsruhe, Germany) and contained GlutaMAX, 10% FCS, and 100 U/ml penicillin/streptomycin.

The lysis of organ tissues, the purification of 20S proteasomes from liver, and the quantification of the 20S proteasome from LCMV-infected (8 d postinfection with 200 PFU LCMV-WE i.v.) C57BL/6 or LMP2^−/−/MECL-1^−/− double-deficient mice were performed as described previously (20).

Nonequilibrium pH-gradient gel electrophoresis/SDS-PAGE was performed exactly as described previously (21).

A total of 1 μg purified proteasomes was separated by SDS-PAGE (15% gel), blotted onto nitrocellulose (Schleicher & Schuell BioSciences, Dassel, Germany), blocked (PBS/5% w/v low-fat dry milk/0.2% Tween 20) for 1 h, and agitated overnight at 4°C with an affinity-purified polyclonal rabbit Ab recognizing MECL-1 (22). The blots were washed three times and incubated for 2 h with peroxidase-conjugated anti-rabbit Ab. After extensive washing with PBS/0.2% Tween 20, proteins were visualized by ECL.

The synthetic peptides gp33–41 (KAVYNFATC), gp276–286 (SGVENPGGYCL), nucleoprotein (NP)396–404 (FQPQNGQFI), NP205–212 (YTVKYPNL), gp92–101 (CSVNNSHHYI), and gp118–125 (LNHNFCNL) were obtained from P. Henklein (Charité, Berlin, Germany).

The β5i (ONX 0914; formerly called PR-957; Onyx Pharmaceuticals)-specific inhibitor was dissolved at a concentration of 10 mM in DMSO and stored at −20°C (4). For proteasome inhibition in mice, ONX 0914 was formulated in an aqueous solution of 10% (w/v) sulfobutylether-β-cyclodextrin and 10 mM sodium citrate (pH 6) and administered to mice as an i.v. bolus dose of 10 mg/kg (in a volume of 100 μl).

Splenocytes from LCMV-infected C57BL/6 mice were incubated for 30 min with anti-CD4, anti-CD8, or anti-CD19 (all BD Biosciences) Abs at 4°C. After two washes, cells were acquired with the use of the Accuri 6 flow cytometer system.

Splenocytes (10⁶/well) of C57BL/6 or LMP2^−/−/MECL-1^−/− double-deficient mice were incubated with 300 nM ONX 0914 or DMSO and stimulated with 2 μg/ml LPS. After 20 h, IL-6, IL-10, IFN-γ (all BD Biosciences), and TNF-α (eBioscience) in the supernatant were determined by ELISA, according to the manufacturer’s protocol.

Analysis of T cell responses was performed as previously detailed (2). Briefly, splenocytes were incubated in round-bottom 96-well plates with 10⁻⁶ M specific peptide in 100 μl IMDM 10% plus brefeldin A (10 μg/ml) for 5 h at 37°C. The staining, fixation, and permeabilization of the cells were performed exactly as detailed previously (23).

T cells from splenocytes of Thy1.1-positive mice were isolated with the Pan T cell isolation kit (Miltenyi Biotec). Purified T cells (2.5 × 10⁷) were transferred i.v. into naive mice on day −1. On day 0, mice were infected with 200 PFU LCMV-WE i.v. Eight days later, splenocytes were analyzed as described in intracellular cytokine staining (ICS) for IFN-γ with anti-Thy1.1 and anti-CD8.

LCMV-specific CTL lines were generated exactly as previously described (24). An additional density centrifugation step was conducted 1–2 d before using CTL in Ag presentation experiments. CTL were used in ICS at an effector:stimulator ratio of 0.2 in the first dilution, and serial 3-fold dilution of stimulators was performed.

The statistical significance of the differences was determined using the Student t test or one-way ANOVA. For ANOVA, we used the Bonferroni post hoc analysis to compare treatment groups. All statistical analyses were performed using GraphPad Prism Software (version 4.03; GraphPad, San Diego, CA). Statistical significance was achieved when p < 0.05. *p < 0.05, **p < 0.01, ***p < 0.001.

To investigate the physiological role of the IFN-γ–inducible proteasome subunit LMP2^−/− and MECL-1^−/−, we crossed these two mouse strains to generate LMP2^−/−/MECL-1^−/− double-deficient mice. The mice showed no visible abnormalities, were fertile, and lived to at least 1 y of age (data not shown). The absence of MECL-1 in LMP2^−/−/MECL-1^−/− double-deficient mice was confirmed by Western blot (Fig. 1B). To analyze the incorporation of LMP7 in LMP2^−/−/MECL-1^−/− double-deficient mice, these mice were infected with 200 PFU LCMV-WE. Eight days later, the proteasome subunit composition of infected livers was analyzed by two-dimensional nonequilibrium pH gradient electrophoresis/SDS-PAGE (Fig. 1A). Incorporation of LMP7 into proteasomes was similar in LMP2^−/−/MECL-1^−/− double-deficient mice compared with wild-type C57BL/6 mice. It has been reported that in LMP7^−/−/MECL-1^−/− double-deficient mice incorporation of LMP2 is reduced in these mice compared with wild-type mice (25). Hence, analysis of LMP2^−/−/MECL-1^−/− double-deficient mice allows us to exclusively investigate the absence of LMP2 and MECL-1.

It has been reported that LMP2^−/− and MECL-1^−/− mice have a reduced number of CD8⁺ splenocytes (8, 9). To determine whether a similar phenomenon can be found in LMP2^−/−/MECL-1^−/− double-deficient mice, splenocytes from naive MECL-1^−/−, LMP2^−/−, LMP2^−/−/MECL-1^−/−, and C57BL/6 control mice were stained for CD4, CD8, and CD19 and analyzed by flow cytometry. Splenocytes of LMP2^−/−/MECL-1^−/− showed a 50% reduction of CD8⁺ cells (Fig. 2), whereas the percentage of CD4⁺ and CD19⁺ cells was not affected. Percentage of CD8⁺ cells in spleens of naive MECL-1^−/− (30% reduction) and LMP2^−/− (36% reduction) was reduced, as previously reported (8, 9).

To characterize LMP2^−/−/MECL-1^−/− double-deficient mice in a model system of viral infection, we infected these mice with LCMV-WE and determined viral titer on day 4 postinfection in the spleen (Fig. 3A). No difference in viral titer could be observed between C57BL/6 and LMP2^−/−/MECL-1^−/− double-deficient mice. To compare the anti-LCMV T_CD8+ response of wild-type and LMP2^−/−/MECL-1^−/− double-deficient mice, we analyzed the CTL response directed against gp33–41/D^b/K^b, NP396–404/D^b, gp276–286/D^b, gp92–101/D^b, gp118–125/K^b, and NP205–212/K^b on day 8 post–LCMV-WE infection by ICS for IFN-γ (Fig. 3B). CTL responses to the dominant epitopes gp33 and NP396 and the subdominant epitopes gp118 were slightly increased, whereas responses to gp276 and NP205 were markedly reduced in LMP2^−/−/MECL-1^−/− double-deficient mice. Although the CTL response to several epitopes was drastically altered in LMP2^−/−/MECL-1^−/− double-deficient mice, LCMV induced a robust antiviral CTL response comparable to C57BL/6 wild-type mice.

LMP2, LMP7, and MECL-1 single-knockout mice were first described several years ago (8–10). Because MECL-1 is encoded on a different chromosome than LMP2 and LMP7, which are located in the MHC-II gene locus, it is possible to cross MECL-1^−/− with LMP7^−/− (25–27) or with LMP2^−/− (described in this work), respectively, to create immunoproteasome double-deficient mice. However, the generation of mice lacking all three immunoproteasome subunits is not feasible by intercrossing immunoproteasome-deficient mice. Therefore, we used the recently published LMP7-selective inhibitor ONX 0914 (previously PR-957) (4) in LMP2^−/−/MECL-1^−/− double-deficient mice to produce mice lacking all immunoproteasome activity. First, we studied the effect of the lack of immunoproteasome activity on IL-6 and IL-10 production of LPS- or TNF-α and IFN-γ secretion of αCD3/CD28-stimulated splenocytes. Recently, we demonstrated that inhibition of LMP7 by the selective inhibitor ONX 0914 reduced cytokine production in LPS-stimulated splenocytes or CD3/CD28-stimulated T cells. To investigate whether cytokine production of splenocytes deficient for all three immunosubunits is altered compared with wild-type mice, LMP2^−/−/MECL-1^−/− double-deficient mice and C57BL/6 mice were treated with ONX 0914 and stimulated with LPS or plate-bound Abs specific for CD3/CD28 to induce cytokine secretion. Analysis of supernatants revealed similar reduction in IL-6, TNF-α, and IFN-γ secretion of splenocytes lacking all functional immunosubunits compared with mice lacking only functional LMP7 (Fig. 4). IL-10 production was affected in neither wild-type nor double-deficient mice. Hence, it seems that the lack of LMP2 and MECL-1 has no contribution to the recently described function of LMP7 in cytokine production (4).

To investigate the effect of the lack of all three immunoproteasome subunits on viral infection, LMP2^−/−/MECL-1^−/− double-deficient mice were treated for consecutive 5 d with ONX 0914 (days −1, 0, 1, 2, and 3) or were left untreated. On day 0, mice were infected with 200 PFU LCMV-WE, and CD4⁺, CD8⁺, and CD19⁺ lymphocyte populations were analyzed in the spleen on day 8 postinfection (Fig. 5A). No differences could be observed between ONX 0914-treated and untreated mice. Hence, it seems that the immunoproteasome has no crucial role in the general expansion of CD8⁺ cells in LCMV-infected mice. Similar viral titers on day 4 postinfection were determined in the spleens of ONX 0914-treated and untreated LMP2^−/−/MECL-1^−/− double-deficient mice (Fig. 5B).

To compare the specific anti-LCMV T_CD8+ response in mice lacking functional immunoproteasome subunits, C57BL/6 and ONX 0914-treated or untreated LMP2^−/−/MECL-1^−/− double-deficient mice were infected with LCMV-WE, and the responses to six defined LCMV epitopes were analyzed by ICS for IFN-γ. ONX 0914 treatment reduced the CTL response to gp33 and NP396 (Fig. 6A), as previously reported for C57BL/6 wild-type mice (4). The poor CTL response to gp276 in double-deficient mice compared with wild-type C57BL/6 mice (Figs. 3, 6A) was not altered by ONX 0914 treatment, whereas the CTL response to NP205 was increased from background levels to a 2.5% specific response (Fig. 6A). Comparison of the CTL response in ONX 0914-treated or untreated C57BL/6 mice to treated or untreated LMP2^−/−/MECL-1^−/− double-deficient mice revealed an increase of the NP205 response in ONX 0914-treated LMP2^−/−/MECL-1^−/− double-deficient mice to the level of wild-type mice (Fig. 6B). Interestingly, ONX 0914 treatment of MECL-1^−/− single-knockout mice did not improve the NP205-specific T_CD8+ response in these mice (Fig. 6C). Hence, it seems that all three immunosubunits are involved in the generation or destruction of the NP205 epitope.

It has been reported that LMP2 (14)-, LMP7 (15)-, and MECL-1 (9)–deficient mice possess an altered T cell repertoire. The reduced T cell response observed for gp276 in LMP2^−/−/MECL-1^−/− double-deficient mice is similar to MECL-1^−/− single-deficient mice, for which an altered gp276-specific T cell repertoire has been reported (9). To investigate whether the increased response to NP205 in ONX 0914-treated LMP2^−/−/MECL-1^−/− double-deficient mice compared with untreated mice is a presentation or a T cell-intrinsic phenomenon, an adoptive transfer experiment was performed. Magnetically purified T cells from Thy1.1 wild-type mice were adoptively transferred (on day −1) into LMP2^−/−/MECL-1^−/− double-deficient mice. Mice were treated for consecutive 5 d with ONX 0914 (days −1, 0, 1, 2, and 3) or were left untreated. On day 0, mice were infected with 200 PFU LCMV-WE, and the cytotoxic T cell response was analyzed on day 8 by ICS for IFN-γ. To discriminate between transferred and endogenous T cells, splenocytes were stained for Thy1.1 (transferred cells). The endogenous CTL response (Thy1.1⁻) (Fig. 7B) to NP205 in untreated LMP2^−/−/MECL-1^−/− double-deficient mice was comparable to the transferred response (Fig. 7A). Hence, the reduced NP205 response in LMP2^−/−/MECL-1^−/− double-deficient mice is due to an altered presentation of this CTL epitope in the double-deficient mice. Treatment of LMP2^−/−/MECL-1^−/− double-deficient mice with ONX 0914 increased the CTL response to NP205 of both endogenous and transferred cells, whereas the response to gp33 was reduced in ONX 0914-treated compared with untreated mice. Therefore, the increased CTL response observed in Fig. 6A and 6B is not a T cell-intrinsic phenomenon, but is due to an altered presentation of NP205 in ONX 0914-treated LMP2^−/−/MECL-1^−/− double-deficient mice.

To further investigate the contribution of Ag presentation to the increased presentation of NP205 by ONX 0914 in LMP2^−/−/MECL-1^−/− double-deficient mice, an in vitro Ag presentation assay was performed. Splenocytes from LMP2^−/−/MECL-1^−/− double-deficient mice as well as C57BL/6 control mice were infected in vitro with LCMV-WE and were treated with ONX 0914. These splenocytes were used as stimulators for IFN-γ production (ICS) by monospecific CTL lines specific for gp33 or NP205. As shown in Fig. 8, ONX 0914 reduced presentation of gp33 in C57BL/6 wild-type (Fig. 8A) and LMP2^−/−/MECL-1^−/− double-deficient mice (Fig. 8B) to a similar extent. In contrast, whereas LMP7 inhibition slightly reduced NP205 presentation in C57BL/6 mice, ONX 0914 treatment increased the presentation of the NP205 CTL epitope in LMP2^−/−/MECL-1^−/− double-deficient splenocytes. Hence, inhibition of LMP7 by ONX 0914 in LMP2^−/−/MECL-1^−/− double-deficient mice increases the amount of NP205 peptides presented on MHC-I on APCs.

Mice lacking one of the immunoproteasome subunits LMP2, LMP7, and MECL-1 are overall able to mount a powerful T cell response and clear the majority of all to date analyzed bacterial and viral infections as efficiently as immunoproteasome-proficient wild-type mice (9, 14, 16, 17, 23), suggesting the constitutive counterparts are sufficient not only for housekeeping processes, but also for ensuring health and survival during immune responses. Nevertheless, it has recently been shown that LMP7 and to a lesser extent LMP2 are crucial for mice to combat Toxoplasma gondii infection (18). Additionally, the lack of LMP7 protracted and lowered the severity of LCMV-induced meningitis in LMP7-deficient mice (22). These two findings point to a specific role for the immunoproteasome in regulating other facets of immune responses, independent of class I Ag presentation. Indeed, we recently demonstrated that LMP7 is involved in the regulation of the production of proinflammatory cytokines (4, 6). In mouse models of rheumatoid arthritis and colitis, an LMP7-selective inhibitor (ONX 0914) reversed signs of disease and resulted in reductions in cellular infiltration and cytokine production. These studies reveal a unique role for LMP7 in controlling pathogenic immune responses.

To investigate the effect of the lack of all immunoproteasome activities on cytokine production, we stimulated splenocytes from LMP2^−/−/MECL-1^−/− double-deficient mice with LPS or plate-bound Abs specific for CD3/CD28 and treated these cells with the LMP7-selective inhibitor ONX 0914 (Fig. 4). Comparison of inhibitor-treated C57BL/6 wild type with LMP2^−/−/MECL-1^−/− showed a similar reduction in IL-6, IFN-γ, and TNF-α production, whereas IL-10 secretion was not altered in both types of mice. Hence, it seems that the regulation of cytokine production by immunoproteasomes is only dependent on LMP7 and not influenced by MECL-1 and LMP2. A recent publication demonstrating that mixed proteasomes expressed in LMP2^−/− mice decrease cytokine production by DCs (28) supports the notion of immunoproteasomes playing a role in cytokine production. To shed light on the precise role of immunoproteasome subunits, analyses of different immunoproteasome subunit-deficient mice with different pathogens and investigation of these mice in different mouse models for autoimmune disorders are warranted.

The assembly of both constitutive and immunoproteasomes is a multistep process assisted by the concerted activity of specific chaperons (25, 29). A considerable number of investigations focused on the order of β-propeptide assembly. LMP2-deficient mice display an impaired MECL-1 incorporation, and the absence of LMP7 reduced the level of both LMP2 and MECL-1 and resulted in an accumulation of their precursors (25). MECL-1^−/−/LMP7^−/− double-deficient mice showed a reduced incorporation of LMP2 into proteasomes. In contrast, proteasomes isolated from livers of LCMV-WE–infected LMP2^−/−/MECL-1^−/− double-deficient mice and C57BL/6 mice display comparable incorporation of LMP7 (Fig. 1). Therefore, LMP2^−/−/MECL-1^−/− double-deficient mice truly permit the study of the function of MECL-1 and LMP2. Both LMP2- and MECL-1–deficient mice have reduced numbers of CD8 in the spleen (Fig. 2) (8, 9), although MHC-I surface expression is, in contrast to LMP7^−/− mice (10), not reduced in these mice. Using bone marrow chimeras, it could be demonstrated that the reduced number of CD8⁺ T cells in MECL-1^−/− and LMP2^−/− mice is not due to an altered Ag presentation in these mice, but due to a T cell-intrinsic phenomenon (26). Analysis of LMP2^−/−/MECL-1^−/− double-deficient mice revealed an even more pronounced phenotype regarding percentage of CD8⁺ T cells in these mice compared with single knockout mice (Fig. 2). It seems that both LMP2 and MECL-1 contribute to the T cell-intrinsic phenomenon of reduced CD8⁺ T cells. During CD8⁺ T cell development, bone marrow-derived T cell precursors migrate into the thymus, where they develop to functional CD8⁺ T cells. After egress from the thymus, homeostatic proliferation in the periphery takes place. At what stage in CD8⁺ T cell development and how LMP2 and MECL-1 interfere have to be determined. It has been reported that LMP2-deficient mice also have reduced numbers of CD4⁺ T cells and B cells (28). In contrast to this study, we could not find a reduced percentage and numbers of CD19⁺ cells in spleens of LMP2^−/− mice (Fig. 2), but we observed that the size of the naive spleen is dramatically reduced in these mice (our unpublished observation). The infection of LMP2^−/−/MECL-1^−/− double-deficient mice with LCMV induced a robust CTL response comparable to that of wild-type mice (Fig. 3B). Virus titers on day 4 postinfection of LMP2^−/−/MECL-1^−/− double-deficient and wild-type mice (Fig. 3A) and the percentage of CD8⁺ cells in the spleen of infected mice were similar (data not shown). It seems that the reduced number of CD8⁺ in naive LMP2^−/−/MECL-1^−/− double-deficient mice (Fig. 2) does not affect bulk CD8⁺ T cell expansion in LCMV-infected mice. The CTL response to gp276 and NP205 was strongly reduced in LMP2^−/−/MECL-1^−/− double-deficient mice compared with wild-type mice. Similar results were obtained in MECL-1–deficient mice, for which we could demonstrate that the reduced CTL response to gp276 is due to an altered T cell repertoire (9). Adoptive transfer experiments of wild-type cells into LMP2^−/−/MECL-1^−/− double-deficient mice showed a similar CTL response to NP205 of endogenous and transferred cells, indicating an altered processing of NP205–212 in LMP2^−/−/MECL-1^−/− double-deficient mice (Fig. 7).

Analysis of mice deficient for all three immunoproteasome subunits is crucial to investigate the concerted function of these subunits. Nevertheless, generation of these mice is rather difficult, because LMP2 and LMP7 are located on the same chromosome in the MHC locus. Therefore, we applied the recently developed LMP7-selective inhibitor ONX 0914 to LMP2^−/−/MECL-1^−/− double-deficient mice to generate mice lacking functional immunoproteasomes. Infection of these mice with LCMV demonstrated unaltered CD8⁺ T cell expansion and virus titers on day 4 postinfection (Fig. 5). Although it has recently been shown that immunoproteasomes are essential for survival and expansion of adoptively transferred T cells in virus-infected mice (7), expansion of CD8⁺ T cells in LCMV-infected mice lacking all three functional immunoproteasome subunits was not affected (Fig. 5A), as can be seen by similar numbers of CD8⁺ in the spleen. The specific CTL response to gp33 and NP396 was reduced, whereas LMP7 inhibition in LMP2^−/−/MECL-1^−/− double-deficient mice increased the response to NP205 (Fig. 6A). Comparison of C57BL/6 and LMP2^−/−/MECL-1^−/− double-deficient mice showed that LMP7 inhibition increased the NP205 CTL response in LMP2^−/−/MECL-1^−/− to levels of wild-type mice (Fig. 6B). Interestingly, ONX 0914 had no influence on the NP205-specific CTL response in wild-type mice (Fig. 6B), demonstrating that LMP7 inhibiton only affects the NP205-specific CTL response in the absence of MECL-1 and LMP2. Adoptive transfer experiments (Fig. 7) and in vitro Ag presentation assays (Fig. 8) revealed that the processing and presentation of the NP205 CTL epitope are affected in mice lacking functional immunosubunits. It seems that this CTL epitope is dependent on all three immunoproteasome subunits in a concerted manner. Inhibition of LMP7 in MECL-1 single knockouts could not revert the reduced NP205 response to levels of wild-type mice (Fig. 6C). Hence, MECL-1 and LMP2 are needed for the generation of this epitope (Fig. 3B), whereas LMP7 destroys NP205 in the LMP2^−/−/MECL-1^−/− double-deficient mice (Figs. 6–8). Interestingly, LMP7 inhibition in C57BL/6 mice did not further increase the CTL response in wild-type mice (Fig. 6B). Therefore, LMP7 only affects NP205 peptide production under limiting conditions as present in LMP2^−/−/MECL-1^−/− double-deficient mice. Due to the different cleavage pattern of constitutive and immunoproteasomes, some MHC class I ligands were shown to be preferentially, if not exclusively, produced by immunoproteasomes (8, 11) or constitutive proteasomes (12), respectively. Attempts to develop algorithm to predict constitutive and immunoproteasome cleavage pattern have been developed (30–32). The opposing effects revealed in this study by the analysis of NP205 epitope, which is destroyed by LMP7 and dependent on LMP2 and MECL-1, indicate that the cleavage patterns of immunoproteasome-dependent ligands will be rather difficult to predict by computer-based algorithms. Analysis of mice lacking all three immunoproteasome subunits will probably differ from LMP2^−/−/MECL-1^−/− double-deficient mice treated with an LMP7-selective inhibitor. In mice lacking all three immunoproteasome subunits, β5 is incorporated instead of LMP7, and, therefore, full chymotrypsin-like activity remains. In contrast, the chymotrypsin-like activity is blocked in LMP7-containing proteasomes inhibited with ONX 0914. Hence, generation of mice lacking all three immunoproteasome subunits is still warranted, especially if immunoproteasomes play a crucial role during development of mice.

To our knowledge, we describe for the first time a viral CTL epitope that is dependent on the opposing effects of immunoproteasome subunits. Additionally, to our knowledge, this is the first study investigating a viral CTL response in mice simultaneously lacking all three immunoproteasome activities.

We thank J. Monaco for providing gene-targeted mice.

The authors have no financial conflicts of interest.