Proteasomes generate peptides bound by major histocompatibility complex (MHC) class I molecules. Avoiding proteasome inhibitors, which in most cases do not distinguish between individual active sites within the cell, we used a molecular genetic approach that allowed for the first time the in vivo analysis of defined proteasomal active sites with regard to their significance for antigen processing. Functional elimination of the δ/low molecular weight protein (LMP) 2 sites by substitution with a mutated inactive LMP2 T1A subunit results in reduced cell surface expression of the MHC class I H-2Ld and H-2Dd molecules. Surface levels of H-2Ld and H-2Dd molecules were restored by external loading with peptides. However, as a result of the active site mutation, MHC class I presentation of a 9-mer peptide derived from a protein of murine cytomegalovirus was enhanced about three- to fivefold. Our experiments provide evidence that the δ/LMP2 active site elimination limits the processing and presentation of several peptides, but may be, nonetheless, beneficial for the generation and presentation of others.
Using proteasome-specific inhibitors, the proteasome system has been shown to be involved in antigen processing and to represent the major source for the generation of MHC class I peptides (1–4). The 20S proteasome is an NH2-terminal nucleophile hydrolase possessing an active site threonine residue (5). It is a cylinder-shaped particle composed of four stacked rings of seven subunits each. In eukaryotes, the seven different α type subunits occupy positions in the two outer rings, whereas the two inner rings are formed by seven different β type subunits (6). The proteolytic activity is restricted to the lumen of the cylinder and is mediated by three of the seven β type subunits, i.e., subunits δ (β1), MB1 (β5), and Z (β2) (parentheses, new nomenclature according to Groll et al., reference 7). Therefore, in total, the 20S proteasome complex possesses six active sites within the two inner β rings. By induction with the cytokine IFN-γ, the active site bearing constitutive β subunits are replaced by their IFN-γ–inducible counterparts low molecular weight protein 2 (LMP2)1 (iβ1), LMP7 (iβ5), and MECL-1 (iβ2) during proteasome assembly (3, 8, 9). Of these, LMP2 (iβ1) and LMP7 (iβ5) are encoded within the MHC class II region in the direct neighborhood of the TAP1 and TAP2 peptide transporter genes (10, 11). MECL-1 (iβ2) is encoded outside the MHC locus, but its incorporation into the 20S proteasome complex is guaranteed through the presence of LMP2 (iβ1; reference 12). The IFN-γ–induced replacement of subunit δ (β1) by LMP2 (iβ1), subunit MB1 (β5) by LMP7 (iβ5), and Z (β2) by MECL-1 (iβ2) results in changes of the hydrolytic activities as monitored with short fluorogenic peptide substrates (13, 14). In addition, the incorporation of these subunits strongly alters the cleavage site preferences of the 20S proteasome in vitro (14, 15). As a consequence, a different set of peptides products is generated by the 20S proteasome. Under physiological conditions, the ratio between constitutive and cytokine-modified proteasomes complexes changes only slowly. Accordingly, the abundance of certain peptide products as well as their quality will gradually change during the time course of IFN-γ induction. Indeed, targeted deletion of LMP2 (iβ1) and LMP7 (iβ5) in mice caused alterations in antigen presentation, emphasizing the importance of these subunits for the generation of at least certain MHC class I antigens (16, 17). Using proteasome inhibitors, it has been shown that the inhibition of some of the proteasomal peptidase activities affects the processing of MHC class I antigens. (1, 18). However, there exists little active site specificity of the available proteasome aldehyde inhibitors. Even the active site specificity of lactacystin demonstrated in vitro is difficult to control in cell experiments since, depending on the experimental condition, lactacystin affects more than one type of active site (19, 20). Therefore, experimental setups using proteasome inhibitors in most cases do not allow one to draw any conclusions on the functional importance of a specific active site for the generation of a defined MHC class I antigen. Such knowledge is, however, important to better understand the basic rules of antigen processing and to develop strategies that may allow either up- or downregulation of the generation of a defined antigenic peptide.
To overcome these problems, we made use of a recently described mutation in the nonconstitutive LMP2 (iβ1) subunit in which the NH2-terminal active site threonine was replaced by alanine (21). This T1A mutation resulted in the impairment of correct maturation by autocatalytic processing of the subunit and rendered an proteolytically inactive LMP2 subunit. In this study, we used the inactive mutant to study the functional importance of the δ/LMP2 (β1/ iβ1) active sites with regard to MHC class I antigen presentation. Overexpression of the mutant LMP2 T1A subunit in mouse fibroblast cells resulted in an effective replacement of the proteolytically active δ (β1) subunit. As a consequence, the mutant LMP2 T1A cells contain proteasomes in which two of the six active sites of the 20S proteasome complex are eliminated. Our experiments demonstrate that the deletion of these active sites generally limits the production of peptides bound to H-2Ld and H-2Dd molecules. At the same time, the mutation enhances the generation and presentation of an H-2Ld epitope derived from the cytomegalovirus pp89 protein.
Materials And Methods
The BALB/c-derived mouse fibroblast cell lines C4 and B8 were used. The B8 cell line, which is derived from the C4 cells, constitutively expresses the IE I pp89 of the murine cytomegalovirus (22). The B8 cell line was subcloned by limiting dilution and one clone was chosen as recipient for the transfection experiments. The generation of the cDNA constructs of LMP2 and LMP2 T1A, transfection by conventional calcium phosphate precipitation and selection are described in detail in reference 21.
Purification of 20S Proteasomes and Assay of Proteolytic Activity.
20S proteasomes were purified using standard procedures (21). Vmax and Km values were determined using the fluorogenic peptide substrates Bz-Val-Gly-Arg-7 Amido-4-methylcoumarin (MCA), Z-Gly-Gly-Leu-MAC, Suc-Leu-Leu-Val-Tyr-MCA, and Methoxysuccinyl-Gly-Leu-Phe-MCA (Bachem, Heidelberg, Germany). The peptides were used in concentrations from 5 to 200 μM and incubated with 1 μg/ml proteasome in 200 μl of 50 mM Tris-HCl, pH 7.5, 25 mM KCl, 10 mM NaCl, and 0.1 mM EDTA (assay buffer) for 1 h at 37°C as described before (14). Fluorescence intensity was measured at an excitation wave length of 390 nm and an emission wave length of 460 nm in a SLT Fluostar spectrofluorometer. Data were analyzed according to Lineweaver and Burk. All assays were performed in triplicate and repeated three times.
15 ng of purified proteasomes were separated by SDS-PAGE and blotted as described (21). The blots were incubated with either LMP2- or δ-reactive antisera. Immunoreactive proteins were detected by enhanced chemiluminescence.
The poly A+ mRNA of the cell lines C4, B8, B8-LMP2, control, and B8-LMP2 T1A was prepared using commercially available kits (Quiagen, Darmstadt, Germany). 3 μg of mRNA was applied to each lane. Agarose gel electrophoresis, blotting, labeling of the cDNA fragments, and hybridization was performed according to standard procedures (23). We used three different cDNA fragments simultaneously, a PstI/HindIII fragment (bp 297–815), a HindIII/PstI fragment (bp 815–1,314), and a PstI to 3′ fragment (bp 1,314–1,788). After hybridization and washing the blot was quantitated and visualized with a PhosphorImager (Molecular Dynamics, Krefeld, Germany).
Cells were removed from culture dishes with calcium- and magnesium-free medium, washed, and stained according to standard protocols with the monoclonal antibodies 19/191 (anti–H-2Dd), 3-25.4 (anti–H-2Dd; PharMingen, San Diego, CA), 28-14-8S (anti–H-2Ld), 28-14-8 (anti–H-2Ld; PharMingen), and 15-5-5S (anti–H-2Kd), and a sheep anti–mouse F(ab)2–FITC conjugate as a second-stage reagent. The analysis was performed with a FACSCAN® flow cytometer and LYSIS II™ software (Becton Dickinson, Heidelberg, Germany).
Acid Elution of Natural Peptides and MHC Class I Peptide Binding Assay.
For peptide extraction and external-loading peptides we followed the procedure as described previously (24, 25). B8 cells (2 × 109) were separated from the culture dishes with calcium- and magnesium-free medium and washed with PBS to remove serum proteins. The cells were incubated for 15 min on ice in 0.1% TFA/H2O, sonicated, and kept on ice for another 15 min. Cells were centrifuged for 15 min at 15,000 rpm and the supernatant was collected. High molecular mass material was removed by centrifugation at 4°C through a 10-kD Centricon filter (Amicon Corp., Easton, TX). The peptides were concentrated by Speed Vac centrifugation to a final concentration of ∼5 mg/ml as judged by OD 280. B8 cells and the transfectants were cultured for 18 h at 27°C in the presence of either 25 μg/ml of the synthetic peptide YPHFMPTNL (H-2Ld epitope of pp89) or ∼25 μg/ml of peptides extracted from B8 cells by acidic elution. Cells were removed from culture dishes with calcium- and magnesium-free medium and incubated at a density of 106 /ml for 1 h on ice in PBS containing the peptides in concentrations as described above. After 1 h, cells were resuspended in serum-free medium containing the peptides and left for 2 h at 37°C at a density of 2 × 105 /ml. Staining of the surface level of MHC class I molecules was performed as described above.
Target cells for cytolytic assays were labeled for 90 min with Na251CrO4. A standard 4-h cytolytic assay was performed in triplicate with 1,000 target cells and the indicated numbers of effector cells in two- or fourfold dilution steps as detailed in reference 26. All experiments were performed three times in triplicate cultures with two clones of each transfectant, except for the LMP2 T1A transfectant where four clones were analyzed.
Results And Discussion
Efficient Incorporation of the LMP2 T1A Subunit into the 20S Proteasomes.
To investigate the consequence of a defined active site elimination in the mouse 20S proteasome complex, the murine fibroblast cell line B8 was stably transfected with cDNAs either encoding a wild-type LMP2 or a mutated LMP2 subunit in which the active site threonine 1 residue was exchanged against alanine by site-directed mutagenesis (21). Overexpression of the LMP2 or LMP2 T1A subunits results in efficient incorporation of these subunits into the 20S proteasome complex (Fig. 1). The incorporation of the LMP2 proteins is associated with an almost complete exchange against subunit δ. Accordingly, by immunoblotting with anti-δ antibody only after overexposure of the enhanced chemiluminescence blot, negligible amounts of residual δ subunit could be identified in 20S proteasomes of B8-LMP2 and B8-LMP2 T1A cells. The slower electrophoretic mobility of LMP2 T1A also demonstrates that the NH2-terminal prosequence is only partially cleaved, resulting in an NH2-terminal extension of the subunit. Thus, overexpression of both the functional LMP2 subunit and the LMP2 T1A active site mutant subunit and the concomitant elimination of the active site bearing δ subunit from the 20S proteasome complexes allows production of a B8 cell line whose proteasome population possesses only four, instead of six, active sites. Interestingly, the functional elimination of this active site had no obvious phenotypic effect on the B8 fibroblast cells and appeared not to affect their growth rate.
Active Site Mutation Influences the Proteolytic Activity of 20S Proteasomes In Vitro.
The effect of the active site mutation on the proteolytic activities of the 20S proteasome was tested by analyzing the peptide-hydrolyzing activities of 20S proteasomes from B8, B8-LMP2, and B8-LMP2 T1A with short fluorogenic peptide substrates (Table 1). Independent of the substrates used, neither the substitution of subunit δ by LMP2 nor the elimination of this active site by incorporation of LMP2 T1A has a significant effect on the Vmax of the 20S proteasome. This holds true for the trypsin-like activity monitored with the substrate Bz-Val-Gly-Arg-MCA as well as for the chymotrypsin-like activity measured with Z-Gly-Leu-Leu-MCA and Suc-Leu-Leu-Val-Tyr-MCA. Only in using MeOsuc-Gly-Leu-Phe-MCA was a reduction in Vmax by a factor of 2.9 measured in the B8-LMP2 T1A mutant. Also, the Km value, the measure for the binding affinity of substrates, was only moderately influenced by subunit substitution or active site mutation. For all substrates, we monitored an approximately twofold increase in the Km for the LMP2 T1A proteasome. One possible reason for the observed increase in Km values in the LMP2 T1A mutant could be that the NH2-terminal extension of 8–10 amino acids of the mutant subunit influences the accessibility of the other active sites and hence the substrate binding affinity. Apart from this, the data suggest that the active site under investigation has little effect upon the trypsin and chymotrypsin-like peptide substrates, which is in agreement with the previous finding that the δ/LMP2 site affects the peptidyl glutamyl peptide-hydrolyzing activity (PDGH activity) of the proteasome complex. This activity is completely eliminated in these cells (data not shown). On the other hand, these data show that the different hydrolyzing activities, as monitored with unphysiologically short substrates, are in fact overlapping and that the attractive model of three different proteolytic specificities each mediated by one of three pairs of active sites is perhaps too simple. In a recent investigation, Eleuteri and coworkers (27) came to a similar conclusion by showing that short peptide hydrolyzing activities are overlapping and that different active sites cleave more than one type of short fluorogenic substrate. Interestingly, the incorporation of the LMP2 subunit into 20S proteasomes as such and not its activity seems to affect the enzymatic characteristics of the active sites of the neighboring subunit Z (β2) and the more distant subunit MB1(β5). This suggests once more (14) that the incorporation of this IFN-γ–inducible subunit may also influence the structure function relationship within the proteasome complex.
Active Site Deletion Leads to Reduced Surface Levels of MHC Class I Alleles.
To investigate the effect of the δ/LMP2 active site mutation on the generation of antigenic peptides in vivo, we analyzed the cell surface expression of MHC class I molecules whose assembly and efficient transport to the cell surface is dependent on the loading with suitable peptides. Flow cytometric analysis of several independent B8-LMP2 T1A cell clones with allele-specific antibodies revealed an ∼40% reduction in the cell surface expression of the H-2Dd and H-2Ld molecules when compared with B8, B-LMP2, or B8 mock-transfected control cells (Fig. 2). No difference in cell surface expression was found for the H-2Kd molecules (data not shown). That this is not a clonal effect is demonstrated by the finding that identical data were obtained with different B8-LMP2 T1A cell lines. These results suggest that the elimination of the two active sites restricts the overall quality of peptide generated, thus possibly limiting the supply of peptide and, in consequence, negatively affecting MHC class I molecule assembly and cell surface expression.
To test this hypothesis we took advantage of the observation that MHC class I molecules can reach the cell's surface without prior peptide binding when cells are incubated at 27°C and that empty MHC class I molecules can be stabilized by binding of externally added peptides (24, 25). B8-LMP2 T1A cells were therefore loaded either with a synthetic 9-mer peptide that binds to H-2Ld or peptides extracted from B8 cells. As expected from its binding characteristics, the synthetic 9-mer peptide restores the level of H-2Ld, but not that of H-2Dd on B8-LMP2 T1A cells (Fig. 3,C). Furthermore, peptides extracted from nontransfected B8 cells were able to stabilize the levels of both MHC alleles on the surface of B8-LMP2 T1A cells (Fig. 3 D). These data demonstrate that it is indeed the lack of peptides that is responsible for reduced MHC class I expression on the surface of B8-LMP2 T1A cells. In support of this, pulse chase experiments and immunoprecipitation of H-2Dd and H-2Ld molecules showed that these molecules are equally well expressed in all cell lines analyzed (data not shown). It may be argued that reduced temperatures can increase MHC expression independent of peptide supply. However, under the same experimental conditions, the number of Kd molecules does not increase at the cell surface at 27°C, even when peptides extracted from B8 cells are externally loaded.The elimination of active sites in the 20S proteasome complex therefore decreases the general amount of peptides available for binding to MHC class I H-2Dd and H-2Ld molecules. In contrast, the level of H-2Kd molecules is not reduced. Interestingly, all three haplotypes possess similar preferences for the COOH-terminal anchor residue but differ with regard to the residue preference at position 2 of the epitope. This indicates that the functional importance of the activity of the δ/LMP2 varies depending on the type of peptide products that have to be generated for binding to a given MHC class I haplotype. In addition, despite the fact that the peptide hydrolyzing activities of the different active sites of the 20S proteasome are overlapping as deduced from in vitro data obtained with short fluorogenic peptide substrates (27), the δ/LMP2 active sites exert a specific cleavage property that is responsible for the in vivo generation of a specific peptide quality from natural protein substrates.
Enhanced Specific Cytolysis of Cells Expressing the LMP2 T1A Protein.
So far the experiments showed that the functional elimination of two defined active sites in the 20S proteasome complex affects the quality of peptide generation and results in the downregulation of certain, but not all, MHC class I molecules. To determine how far the active site mutation influences the presentation of a specific peptide, we analyzed the different transfectant B8 cell lines with regard to their ability to present an immunodominant 9-mer peptide of the murine CMV (MCMV) pp89 to a H-2Ld–specific T cell line in a cytotoxicity assay. As shown in Fig. 4, B8 cells and B8 control–transfectant cells were lysed to the same extent, whereas B8-LMP2 cells were slightly less susceptible to lysis. In contrast, three- to fivefold less pp89/H-2Ld–specific cytotoxic T cells were required to lyse the B8-LMP2 T1A cells.
To exclude the possibility that increased pp89 expression in B8-LMP2 T1A cells was responsible for the observed effect, we investigated the expression of pp89 by Northern blot analysis since pp89 is quite stable at the posttranslational level (22). As shown in Fig. 4 B, no significant differences in the expression of pp89 can be detected in the investigated cell lines. Thus, despite the elimination of two active sites, sufficient pp89 antigen is generated to allow an increase in peptide-specific MHC class I presentation. Although in vitro experiments do not necessarily reflect the in vivo situation, it is interesting to note that in vitro digestions experiments of the pp89 25-mer synthetic polypeptide harboring the 9-mer epitope (15) show that the improved MHC class I presentation may be due to altered proteasomal processing properties. Although δ/LMP2 proteasomes have the tendency to destroy the epitope, mutant LMP2 T1A proteasomes do not use the internal cleavage site and thus seem to preserve the epitope (Ruppert, T., unpublished observations). In consequence, the increased maximum of lysis observed may be due to an increase in specific peptide supply. Considering that the overall H-2Ld surface expression is reduced in B8-LMP2 T1A cells, these experiments represent the first example that the specific elimination of a proteasomal active site, in this case δ/LMP2, may be beneficial for presentation of certain class I epitopes, despite reduced MHC class I expression.
This work was supported in part by the Deutsche Forschungsgemeinschaft grant Kl 427 9-2 and by the European Community grant BIO 4-CT97-0505 to P.M. Kloetzel and U.H. Koszinowski.
Abbreviations used in this paper
The present address of Marcus Groettrup is Kantonsspital St. Gallen, Laborforschungsabteilung, 9007 St. Gallen, Switzerland.
Address correspondence to P.-M. Kloetzel, Zentrum für Experimentelle Medizin, Institut für Biochemie– Charité, Humboldt Universität zu Berlin, Monbijoustrabe 2, 10117 Berlin, Germany. Phone: 030-2802-6382; Fax: 030-2802-6608; E-mail: firstname.lastname@example.org