Abstract
Tripeptidyl peptidase II (TPPII) is an oligopeptidase forming giant complexes in the cytosol that have high exo-, but also, endoproteolytic activity. Immunohistochemically, the complexes appear as distinct foci in the cytosol. In part controversial biochemical and functional studies have suggested that TPPII contributes, on the one hand, positively to Ag processing by generating epitope carboxyl termini or by trimming epitope precursors, and, on the other, negatively by destroying potentially antigenic peptides. To clarify which of these roles is predominant, we generated and analyzed TPPII-deficient mice. Cell surface levels of MHC class I peptide complexes tended to be increased on most cell types of these mice. Although presentation of three individual epitopes derived from lymphocytic choriomeningitis virus was not elevated on TPPII−/− cells, that of the immunodominant OVA epitope SIINFEKL was significantly enhanced. Consistent with this, degradation of a synthetic peptide corresponding to the OVA epitope and of another corresponding to a precursor thereof, both being proteasomally generated OVA fragments, was delayed in TPPII-deficient cytosolic extracts. In addition, dendritic cell cross-presentation of phagocytosed OVA and of OVA internalized as an immune complex was increased to about the same level as direct presentation of the Ag. The data suggest a moderate, predominantly destructive role of TPPII in class I Ag processing, in line with our finding that TPPII is not induced by IFN-γ, which up-regulates numerous, predominantly constructive components of the Ag processing and presentation machinery.
Cells present samples of their protein content to CTLs via the proteolytic generation of peptides in the cytosol, which after transport to the endoplasmic reticulum (ER)4 are loaded onto MHC class I molecules, which then migrate to the cell surface. The displayed peptides are 8–10 aa long. The TAP transporter in the ER membrane prefers peptides of 8–15 aa, implying that both peptides corresponding to epitopes and epitope precursors are translocated into the ER, where final trimming of epitope precursors can occur (1, 2, 3).
Cytosolic endo- and aminopeptidases and ER-resident aminopeptidases, but not carboxypeptidases, operate in the classical class I Ag processing pathway (4). Proteasomes are in many cases involved in the initial processing steps. They preferentially generate peptides of 3–20 aa with a hydrophobic or charged C-terminal (Ct) (4) aa. Their cleavage preferences thus perfectly match the preferences of human TAP and of the vast majority of MHC class I alleles for the Ct peptide position. Thus, in many cases, proteasomes produce the epitope as well as epitope precursors, as has repeatedly been shown in in vitro digestion experiments (1, 3, 5).
Of the downstream peptidases, the ER-aminopeptidase associated with Ag processing or endoplasmic reticulum-associated aminopeptidase (ERAP)1, has been investigated most intensively. ERAP1 appears to be the only ER-associated trimming peptidase in mice. Although the knockout (KO) of ERAP1 significantly reduces cell surface MHC class I expression, only a subset of epitope precursor peptides is trimmed down to final epitope size by ERAP1, whereas other epitope or epitope precursor peptides are destroyed or unaffected (6, 7, 8, 9). ERAP2, which is exclusively expressed in humans, forms in part complexes with ERAP1, and both aminopeptidases are required to efficiently trim certain epitope precursor sequences (10).
Of the many cytosolic peptidases, leucine aminopeptidase (LAP), thimet oligopeptidase (TOP), puromycin sensitive aminopeptidase (PSA), bleomycin hydrolase, and tripeptidyl peptidase II (TPPII) have been implicated in class I Ag processing (4, 11). The endo-oligopeptidase TOP prefers peptides of 9–16 aa and preferentially destroys potential epitopes and epitope precursors (12). In vitro digestion and protease inhibitor experiments suggested that the aminopeptidases LAP, bleomycin hydrolase, and PSA significantly contribute to cytosolic epitope precursor trimming, but studies performed in KO mice could not confirm this (13, 14).
With a molecular mass of ∼6 MDa, TPPII forms the largest protease complex of eukaryotic cells. These complexes are composed of 40 identical subunits with a molecular mass of 140 kDa, arranged in two twisted strands (15). Degradation of short and long (up to 40 aa) oligopeptides but not proteins has been demonstrated with purified TPPII complexes in vitro. The complexes have a highly active aminotripeptidase, and some endopeptidase activity. TPPII’s endopeptidase preferentially cleaves after lysine and arginine in long substrates (16, 17, 18). Peptides with Ct lysine residues are not efficiently produced by proteasomes, but are preferred by some HLA class I alleles such as HLA-A3 and HLA-A11, whose ligands may therefore be produced by alternative proteases (19). Purified TPPII precisely excises an HLA-A3/A11-restricted HIV-Nef epitope with a Ct lysine from its flanking sequences and TPPII knockdown significantly reduces recognition of the epitope by CTLs (17). TPPII is thus the first and so far only nonproteasomal protease to be shown to generate the C terminus of a CTL epitope. TPPII seems also to be required for generation of the influenza NP147–155 epitope as suggested by siRNA and protease inhibitor experiments using the broadly specific serine protease inhibitor alanine-alanine-phenylalanine-chloromethyl ketone (AAF-CMK) and the more specific inhibitor butabindide, although this is controversial (20, 21).
In case of the RU131–42 epitope derived from a tumor Ag, TPPII seems to trim epitope precursors, as suggested by the inhibition of CTL recognition by AAF-CMK and analysis of peptide digests with cytosolic fractions (22). A microinjection study demonstrated that, within living cells, TPPII is the only peptidase that attacks peptides longer than 15 aa and that it does so by amino- as well as endopeptidase activity. This, together with Ag presentation studies in the presence of AAF-CMK and butabindide, suggested that long epitope precursors are frequent intermediates in class I Ag processing and that their trimming by TPPII is a crucial processing step (18). However, this conclusion was contradicted by a recent study (23) that did not report any negative effect of TPPII siRNA knockdown on MHC class I-peptide assembly. Along this line, in another recent study, AAF-CMK, which blocks TPPII activity strongly though not completely, did not negatively affect presentation of several CTL epitopes. Conversely, presentation of one of the epitopes was slightly reduced upon TPPII overexpression, suggesting that TPPII destroyed this epitope or its precursors (24). Similar evidence for epitope destruction was also recently suggested (25) for a tyrosinase epitope whose presentation by melanoma cells was enhanced upon TPPII siRNA knockdown. Thus, contradictory results regarding the role of TPPII in Ag presentation have been obtained using siRNA and protease inhibitors.
To resolve this controversy, we have generated TPPII KO mice. These mice allow for the first time the analysis of the effects of specific and complete abrogation of TPPII expression on MHC class I Ag processing and presentation including the analysis of primary cells, in vivo CTL responses, and cross-presentation by dendritic cells (DCs).
Materials and Methods
Mice
Generation and general description of TPPII KO mice will be described elsewhere (J. Huai, E. Firat, A. Nil, D. Million, S. Gaedicke, B. Kanzler, M. Freudenberg, P. van Endert, G. Kohler, H. Pahl, et al., submitted for publication). Use of all mice was with approval of the animal care committee of the Regierungspräsidium Freiburg.
Immunofluorescence staining and Western blot
For immunofluorescence staining, COS-7 cells were fixed with ice cold methanol and then permeabilized with ice cold acetone at –20°C. After incubation in blocking solution (2% BSA and 5% goat serum in PBS), cells were incubated with peptide-specific anti-TPPII Abs produced in rabbits (MPI-IB) followed by incubation with anti-rabbit Cy3-labeled secondary Ab (Sigma-Aldrich) or Alexa Fluor 546-labeled Ab (Invitrogen). Proteasomes were stained with mAb MCP21 (BIOMOL) and an Alexa Fluor 488-labeled secondary anti-mouse Ab (Invitrogen). Cells were analyzed using a fluorescence microscope (Zeiss) equipped with a Hamamatsu Dual Mode Cooled CCD camera (C4880). For Western blot analyses of TPPII expression, bone marrow (BM)-DCs were used untreated or after a 48 h incubation with LPS (10 μg/ml; Sigma-Adrich) or IFN-γ (200 U/ml; eBioscience). Cell lysates were separated by SDS-PAGE. The blots were probed with rabbit peptide-specific anti-TPPII Abs and goat polyclonal anti-actin Abs and developed by ECL (Amersham Biosciences).
Cloning of EGFP-TPPII
The murine TPPII c-DNA was amplified from genomic DNA by long-range PCR. After subcloning into the Bluescript vector (Stratagene), the insert was sequenced. EGFP was fused to the N terminus of TPPII and the EGFP-TPPII fusion protein was cloned into pcDNA3 (Invitrogen) by standard cloning procedures. COS-7 cells were transfected using DEAE dextran and analyzed after 36 h by fluorescence microscopy.
Separation of cell extracts on gel filtration columns
High molecular mass proteins in cytosolic extracts of COS cells were separated on Superose 6 gel filtration columns. TPPII activity in eluting fractions was measured with the substrate alanine-alanine-phenylalanine-4-aminomethylcoumarin.
Generation of BM-DCs and macrophages
To produce BM-DCs, BM cells were seeded in petri dishes at 5 × 105/ml in RPMI 1640 supplemented with 20 ng/ml rGM-CSF (Cell Concepts). For maturation, DCs were incubated at day 6 for 48 h with LPS (10 μg/ml; Sigma-Aldrich). To generate macrophages, BM cells were cultured in DMEM supplemented with 30% l-929 conditioned medium (26). Macrophages were used at day 7.
Flow cytometry
Mouse spleen and BM cells were incubated with anti-FcR Ab (clone 2.4G2) and then stained using the following Abs: anti-CD3, anti-CD19, anti-CD11c, anti-F4/80, anti-H-2Kb, anti-H-2Db, and anti-H-2Ab (all from BD Biosciences). Cells were analyzed on a FACScan (Beckman Coulter).
Stability of cell surface MHC class I-peptide complexes
Splenocytes were cultured in plates coated with anti-CD3 (2 μg/ml; eBioscience) in the presence of IL-2 (40 U/ml; eBioscience) for 4 days. Thereafter, dead cells were removed by centrifugation in Ficoll (PAN), and living cells were incubated with 10 μg/ml brefeldin A (Sigma-Aldrich). After different times, cells were stained with conformation-dependent anti-H-2Kb or anti-H-2Db Abs (from BD Biosciences) and analyzed by flow cytometry. Data are plotted as the relative amount of MHC class I expression with the 0 time value indicated as 100%.
Lymphocytic choriomeningitis virus (LCMV) infection and detection of LCMV-specific CTL responses
Mice were infected i.v. with 200 PFU LCMV-WE and 8 days later LCMV-specific CTL responses were analyzed by measuring the expansion of CD8+ T cells after staining with anti-CD8-Abs, as well as by intracellular cytokine staining using splenocytes ex vivo. For intracellular IFN-γ staining, splenocytes were pulsed with synthetic peptides at a final concentration of 1 μM. After 1 h, brefeldin A (Sigma-Aldrich) was added at a final concentration of 10 μg/ml and the incubation continued for 4 h. For flow cytometry analysis, cells were then stained with anti-CD8, fixed with 1% paraformaldehyde (Sigma-Aldrich) and stained with anti-IFN-γ (eBioscience) in PBS with 0.5% saponin (Sigma-Aldrich) at 4°C overnight.
Determination of virus titers
LCMV titers in infected organs were determined using a focus forming assay as described (27). Titers were expressed as PFU per organ.
In vitro Ag presentation assays
Immature BM-DCs generated as described above, were isolated with anti-CD11c beads (Miltenyi Biotec) and infected with LCMV strain WE at an MOI of 0.01 for 2 d. During this time, the DCs were also matured with LPS (10 μg/ml, Sigma-Aldrich). For vaccinia virus (VV) infection, isolated BM-DCs were first matured with LPS (10 μg/ml) for 24 h and then infected for 1 h with 10 MOI of VV-OVA or VV-SIINFEKL (both VV constructs were provided by J. Yewdell, National Institutes of Health Institute of Allergy and Infectious Diseases, Bethesda, MD), washed and incubated overnight. LCMV-infected cells were then incubated at different ratios with 105 LCMV-specific T cells overnight. VV-infected cells were first fixed with 1% formaldeyde for 1 min at RT, neutralized by washing in PBS with 0.2 M glycine and then incubated at different ratios with 105 OT1 T cells overnight. T cell stimulation was determined by measuring secreted IFN-γ by ELISA. LCMV epitope-specific T cells were generated by weekly peptide stimulation of splenocytes of mice infected 4 wk before with a low dose of LCMV.
In vitro digests of synthetic peptides
Cytosolic extracts immunodepleted of proteasomes or TPPII were incubated with synthetic peptides corresponding to the immunodominant (ID) OVA epitope SIINFEKL or the precursor QLESIINFEKL (purchased from Sigma-Genosys) at 37°C. At the indicated time points, digests were separated on a Sephasil C18 SC2.1/10 column (Pharmacia) for SIINFEKL or a Multospher 120 RP 18 AQ-5 column (CS-Chromatographie) for QLESIINFEKL. Eluents used were: eluant A - 0.1% (v/v) TFA (Sigma-Aldrich) and eluant B - 80% (v/v) acetonitrile (Mallinckrodt Baker)/H2O (0.081% TFA). Peptides were identified by MALDI-TOF mass spectrometry (model G2025A; Hewlett-Packard) and Edman degradation on a Hewlett-Packard instrument (G1000A).
In vitro cross-presentation
To study cross-presentation of soluble OVA, BM-DCs were incubated for 6 h with graded amounts of ultracentrifuged OVA, fixed with 0.002% glutaraldehyde for 1 min at RT, neutralized in PBS with 0.2 M glycine, and then added to 105 OT1 T cells overnight. For cross-presentation of immunocomplexes, soluble OVA at 10 μg/ml was incubated for 1 h with anti-OVA rabbit polyclonal IgG (Sigma-Aldrich) and then the immune complexes were added to BM-DCs. Eight hours later BM-DCs were fixed and incubated with 105 OT1 T cells for 24 h. For cross-presentation using OVA-coated latex beads, BM-DCs were incubated for 8 h with graded amounts of OVA-coated latex beads, then fixed and added to 105 OT1 T cells for 24 h. T cell activation in the cross-presentation experiments was determined in an IL-2 ELISA.
Statistical analyses
Student’s two-tailed t test was used to analyze the significance between experimental groups and relevant controls.
Results
Focal cytosolic distribution of TPPII and lack of inducibility by IFN-γ
Knowledge on the cytosolic distribution of TPPII could help to better understand its role in Ag processing. To visualize the distribution of endogenously expressed TPPII within the cytosol, we stained COS cells with antiserum to TPPII and analyzed them by immunofluorescence microscopy. An uneven distribution was observed and the fluorescence label appeared to be concentrated in discret ‘foci’ scattered over the whole cytoplasm (Fig. 1⇓, A and B). Similar patterns were observed when COS cells were transfected with an EGFP-TPPII fusion protein which was incorporated into the higher molecular mass complexes (Fig. 1⇓, C and D). In contrast, in COS cells expressing only EGFP, the fluorescence assumed a diffuse distribution. The fluorescing foci in anti-TPPII-stained or EGFP-TPPII-expressing cells may correspond to individual TPPII complexes.
Cytosolic distribution of TPPII and lack of inducibility by IFN-γ. A, Immunofluorescence staining of endogenous TPPII with anti-TPPII-antibody in COS-7 cells showing its concentration in foci scattered throughout the cytoplasm. B, Double fluorescence of TPPII and proteasomes showing cytoplasmic localization of TPPII but exclusion from the nucleus. C, Cytosolic distribution of EGFP-tagged TPPII in COS cells transfected with EGFP-TPPII fusion protein in unfixed (upper left) or fixed cells (lower left). EGFP-expressing cells are shown as control (upper right). Bar, 10 μm. D, Separation of cytosolic extracts from EGFP-TPPII-transfected or control cells on Superose 6 gel filtration columns. TPPII activity in eluting fractions was measured with the substrate alanine-alanine-phenylalanine-4-aminomethylcoumarin. Western blot detection of TPPII and EGFP protein shows incorporation of the fusion protein into the higher molecular mass complexes. E, Western blot analysis of TPPII expression in immature BM-DCs, and in BM-DCs matured with LPS or IFN-γ.
So far it has not been reported whether TPPII expression is induced by the inflammatory cytokine IFN-γ, which up-regulates many key components of the class I pathway to improve Ag presentation. As shown in Fig. 1⇑E, expression of TPPII in DCs was neither induced by IFN-γ nor by LPS. Similar results were obtained with primary fibroblasts (data not shown).
Impact of TPPII deficiency on class I cell surface expression
To evaluate the role of TPPII in Ag processing and presentation, we generated KO mice with a ubiquitous tppII deletion, in which TPPII protein expression is completely abolished in all cell types. Generation of these mice by Cre/loxP and FLP/FRT technology and their general description will be published elsewhere (J. Huai, E. Firat, A. Nil, D. Million, S. Gaedicke, B. Kanzler, M. Freudenberg, P. van Endert, G. Kohler, H. Pahl, et al., submitted for publication).
Surface expression of MHC class I peptide complexes is an indirect measure of the peptide supply in the class I pathway but can also be influenced on the transcriptional level. To evaluate the effects of TPPII deficiency on MHC class I cell surface expression, we first stained splenocytes from wild-type (WT) and mutant animals with class I-specific Abs ex vivo. We found a significant increase in the surface expression of H-2Kb and H-2Db molecules on all subpopulations examined including T cells and B cells (Fig. 2⇓, A and B), DCs, macrophages, NK cells, and granulocytes (data not shown). Levels of H-2Kb and H-2Db were increased by a factor of 1.2 to 2.4. However, class II MHC cell surface levels were also increased and class I and II levels decreased when the splenocytes were cultured for a few days in vitro. Whereas MHC class II reached WT levels, MHC class I on average stayed slightly increased on cultured T and B cells (1- to 1.4-fold; Fig. 2⇓B). Together with an increased activation of NF-κB (data not shown; J. Huai, E. Firat, A. Nil, D. Million, S. Gaedicke, B. Kanzler, M. Freudenberg, P. van Endert, G. Kohler, H. Pahl, et al., submitted for publication), this suggested that increased ex vivo MHC class I expression was to a large extent caused by inflammation. To confirm whether MHC class I peptide complexes are on average slightly increased on TPPII KO cells in the absence of inflammatory stimuli, we cultured DCs and macrophages from BM precursors in vitro. As seen in Fig. 2⇓A (right) and Fig. 2⇓B (bottom), whereas MHC class II cell surface levels were essentially normal, MHC class I levels tended to be slightly increased on immature DCs and macrophages cultured in vitro. This indicates that TPPII may predominantly destroy peptides, i.e., cleave them to sizes smaller than required to bind to MHC class I (<8–10 aa). Fibroblast lines from the skin of KO and WT mice were also analyzed. However, variability in MHC class I expression between different KO and WT lines, both with and without treatment with IFN-γ, precluded any definitive conclusions.
MHC class I surface expression. A, Ex vivo splenic T and B cells as well as in vitro generated BM macrophages and DCs from TPPII KO and WT mice were stained with Abs against H-2Kb and H-2D,b and H-2Ab was done as control. B, Summary of MHC cell surface staining data. Mean factor of increase of MHC expression on T- and B cells ex vivo and after in vitro culture for a few days with either anti-CD3 or LPS, and on DCs and macrophages cultured from BM precursors in vitro. Results indicate the mean ± SD of 14 pairs of mice for cells analyzed ex vivo and for four independent cultures of BM macrophages and DCs. Significant MHC up-regulation is indicated by an asterisk where p < 0.05.
Next, we determined whether TPPII deficiency is associated with alterations in the stability of cell surface MHC class I peptide complexes, which reflects the affinity and therefore quality of the peptides presented. To measure the stability of MHC class I peptide complexes, we treated T cell blasts with brefeldin A to prevent transport of newly assembled class I complexes to the cell surface and determined the decay of H-2Kb and H-2Db molecules with conformation-dependent Abs by flow cytometry. We observed no alteration in the decay of both surface Kb and Db molecules from TPPII−/− cells as compared with TPPII+/+ cells (Fig. 3⇓), indicating that the peptides produced in TPPII−/− cells have approximately the same affinity as those produced in WT cells.
Stability of cell surface MHC class I peptide complexes. Splenic T blasts were incubated with brefeldin A (BFA) for the indicated time intervals and then stained with conformation-dependent anti-H-2Kb and anti-H-2Db Abs followed by flow cytometry. Shown are the results for 4 WT (closed lines and closed symbols) and 4 TPPII KO mice (dashed lines and open symbols).
Presentation of LCMV epitopes and anti-LCMV CTL responses
To assess the impact of TPPII deficiency on Ag presentation in more detail, we studied the recognition of the three different ID Db-restricted LCMV epitopes gp33–41, NP396–404, and gp276–286 upon infection of DCs with LCMV. Readout was the activation of epitope specific CTLs by the infected DCs as measured by secretion of IFN-γ. As seen in Fig. 4⇓A, WT and KO DCs activated LCMV-epitope specific CTLs to a similar extent, suggesting that TPPII has no significant influence on the generation of these three epitopes, although abundant expression of viral Ag in LCMV-infected cells might preclude detection of minor differences in Ag presentation.
CD8+ T cell response to LCMV infection and presentation of three ID LCMV-derived epitopes to class I-restricted CTLs. A, Presentation of LCMV epitopes by LCMV-infected DCs. TPPII KO and WT DCs matured with LPS and infected with LCMV for 48 h were incubated with gp33–41, NP396–404, and gp276–286-specific CTLs overnight. Activation of the CTLs was determined by measuring IFN-γ secretion by ELISA. Data shown are one representative experiment of three. B, TPPII KO and WT mice were infected i.v. with LCMV, and 8 days later splenocytes were analyzed for the expansion of CD8+ T cells by staining with anti-CD8 Ab. Results are the mean ± SD of 12 pairs of mice. C, Numbers of epitope-specific CTLs were determined by intracellular IFN-γ staining after a 5-h restimulation of the splenocytes with the respective peptides in vitro. D, LCMV viral titer on day 4 of infection.
CTL responses can also reflect alterations in Ag presentation. We therefore determined the influence of TPPII deficiency on the generation of anti-LCMV CTL responses in vivo. In WT mice, infection with LCMV strongly activates the immune system and causes massive expansion of CD8+ CTLs (28). As seen in Fig. 4⇑, B and C, although statistically not significant, KO mice mounted slightly reduced CTL responses to LCMV. This was found for total CD8+ T cells (Fig. 4⇑B) as well as for individual epitope-specific CTLs (Fig. 4⇑C). The reduction in CTL numbers after infection was similar to or slightly higher than the reduction in CD8+ T cell numbers in lymphoid organs of uninfected KO mice (Fig. 4⇑B). It may thus reflect the increased apoptosis propensity of TCR-stimulated TPPII-deficient T lymphocytes (data not shown; J. Huai, E. Firat, A. Nil, D. Million, S. Gaedicke, B. Kanzler, M. Freudenberg, P. van Endert, G. Kohler, H. Pahl, et al., submitted for publication), rather than alterations in LCMV epitope presentation levels on APC. Another possible explanation for reduced CTL responses could be a reduction in virus infection by pre-existing inflammation. To exclude this, we determined LCMV titers at d4 of infection (i.e., at the peak of infection where CTL responses are not yet detectable). As seen in Fig. 4⇑D, LCMV titers were similar in spleens of WT and KO mice.
Direct presentation of nonviral Ag
To address whether nonviral epitopes were affected by TPPII, we first studied direct presentation of the ID OVA epitope SIINFEKL upon infection of DCs with VV expressing full-length OVA or VV expressing SIINFEKL as a minigene. As seen in Fig. 5⇓A, KO DCs stimulated SIINFEKL-specific CTLs significantly better than WT DCs. This was true for DCs expressing full-length OVA protein as well as for DCs expressing the SIINFEKL peptide. This suggests that TPPII is involved in the destruction of the mature epitope and perhaps of epitope precursors in the cytosol in vivo. To corroborate this, we digested synthetic peptides corresponding to the mature SIINFEKL epitope or the precursor QLESIINFEKL with cytosol preparations from cells depleted or not of either proteasomes or TPPII. Both SIINFEKL and QLESIINFEKL are abundant proteasomal cleavage products which are produced when full-length OVA or long OVA fragments are digested with purified proteasomes in vitro (29, 30, 31). As seen in Fig. 5⇓B (upper panel), destruction of the SIINFEKL peptide and appearance of the TPPII product NFEKL were significantly delayed in cytosol preparations lacking TPPII, whereas there was no difference for SIINFEKL degradation between proteasome-depleted cytosolic extracts and nondepleted controls. Similar results were obtained for QLESIINFEKL (Fig. 5⇓B, lower panel) and when protease inhibitors were used to block either proteasomes or TPPII (data not shown). These data are also in agreement with our observation that purified TPPII rapidly degrades SIINFEKL and QLESIINFEKL in vitro, whereas purified proteasomes do not attack these short peptides (data not shown). Degradation of SIINFEKL precursors other than QLESIINFEKL might also be delayed in the absence of TPPII because TPPII presumably degrades a large variety of peptides due to its broad cleavage specificity. TPPII can actually cleave any peptide bond except before or after proline residues (32).
Increased presentation of the OVA-derived SIINFEKL epitope on TPPII KO DCs to SIINFEKL-specific CTLs and in vitro degradation of the SIINFEKL and the QLESIINFEKL peptides with cytosolic extracts. A, Mature TPPII KO and WT DCs were infected with VV-OVA or VV-SIINFEKL and incubated with OT1 T cells overnight. IFN-γ secretion of activated T cells was measured by ELISA. Data shown are for one representative experiment of three. Statistically significant data are indicated by an asterisk where p < 0.05. B, In vitro degradation of synthetic SIINFEKL peptide (upper panel) and QLESIINFEKL peptide (lower panel) in proteasome- or TPPII-depleted cytosolic extracts. The substrates and the TPPII-degradation products NFEKL and SIINFEKL are highlighted in bold. “Ghost peaks” containing no peptide are marked with an asterisk (∗).
Cross-presentation
Cross-presentation of exogenous Ags via MHC class I molecules is a pathway specific for DCs. It is relevant for induction of CTL responses to viruses that do not directly infect DCs or to tumor Ags as well as for transplant rejection and the maintenance of self-tolerance. Cross-presentation requires that peptides derived from internalized Ags gain access to MHC class I molecules. Although several models with regard to the cell biology of cross-presentation have been proposed, these remain controversial, and cross-presentation both of phagocytosed and soluble Ags, which may involve distinct pathways, is as yet poorly understood (33, 34, 35).
To elucidate whether TPPII plays a role in cross-presentation, we tested the capacity of splenic DCs from mutant animals to cross-present OVA Ags in three different formulations to OVA257–264-specific T cells. OVA protein was provided either adsorbed to latex beads, complexed to anti-OVA-Abs or as soluble protein. The results given in Fig. 6⇓ revealed significantly improved cross-presentation of phagocytosed OVA (Fig. 6⇓A) and of endocytosed OVA/Ig complexes in mutant mice (Fig. 6⇓B), whereas cross-presentation of soluble OVA was only slightly increased, if at all (Fig. 6⇓C).
Cross-presentation of OVA in different antigenic formulations. A, Cross-presentation of particulate OVA. BM-DCs were incubated with graded amounts of OVA-coated latex beads. Eight hours later, cells were fixed and OT1 T cells were added. B, Cross-presentation of OVA/anti-OVA-complexes. Soluble OVA (10 μg/ml) was preincubated with anti-OVA polyclonal rabbit IgG. The immune complexes were incubated with BM-DCs for 8 h. DCs were then fixed and OT1 T cells were added. C, Cross-presentation of soluble OVA. BM-DCs pulsed for 6 h with graded amounts of soluble OVA were fixed and incubated overnight with OT1 T cells. In A, B, and C, T cell activation was determined by IL-2 ELISA. Data shown are for one representative experiment of five. Statistically significant data are indicated by an asterisk (∗) where p < 0.05.
Discussion
The tendency toward slightly enhanced cell surface MHC class I expression levels on a variety of cell types of TPPII KO mice, and the improved direct and cross-presentation of OVA by DCs together with the delayed degradation of the proteasomally produced SIINFEKL and QLESIINFEKL peptides in TPPII-depleted cytosolic extracts suggest that TPPII is predominantly involved in the destruction of potential CTL epitopes and epitope precursors. The lack of influence on presentation of the three LCMV epitopes tested may either be due to abundant Ag expression, which could obscure the detection of slight differences in Ag presentation, or may indicate that TPPII influences only selected epitopes or epitope precursors.
So far there are indications for three different roles of TPPII in Ag processing. First, there is a role in the generation of the C terminus of CTL epitopes (17). However, the proportion of epitope C-termini produced by TPPII is unclear so far. Given the comparably restricted specificity of TPPII’s endopeptidase activity for cleavages after basic residues (16, 17, 18), TPPII may mainly be important for generating ligands for the subgroup of MHC alleles preferring ligands with such Ct. Moreover, endoproteolytic TPPII cleavages have been found rather far away from the N terminus in relatively long oligopeptides and it is unclear whether many long peptides are produced in cytosolic proteolysis. Proteasomes, which are most important for cytosolic protein disposal, seem to predominantly produce shorter peptides (<20 aa) as suggested by in vitro digestion experiments (1, 5).
Secondly, evidence for a role in trimming of N terminus extended CTL epitope precursors has been described. This was first shown for relatively short precursors of an RU1 epitope (22). However, data have been presented suggesting that TPPII may be particularly important for, or even the only cytosolic peptidase capable of, trimming very long epitope precursors. The preference for long oligopeptides in vivo, which was first shown by microinjection experiments (18), has recently been confirmed (23) by the use of minigenes coding for SIINFEKL-precursors and knockdown of TPPII by siRNA. However, a central role of TPPII-mediated trimming of long epitope precursors in class I Ag processing, as originally proposed, appears unlikely, because proteasomes preferentially produce shorter peptides, and MHC class I-peptide assembly (23) and cell surface expression (this study) both slightly increase upon TPPII siRNA knockdown or genetic TPPII KO, respectively. A further argument against a major role of TPPII in epitope precursor trimming is that TPPII completely destroys most peptides it handles, as shown in in vitro digestion experiments (Ref. 23 and data not shown). Other cytosolic peptidases probably also have only a limited role in precursor trimming because they either prefer peptides shorther than eight amino acids or completely destroy the peptides they handle (36, 37). Consistent with these results, recent Ag processing studies suggest that ERAP1 is the most important trimming peptidase and that cytosolic peptidases play a minor role in epitope precursor trimming (see Introduction).
Third, there is evidence that TPPII plays a role in the destruction of epitopes or epitope precursors. Recently, slightly enhanced presentation was found for a tyrosinase epitope upon TPPII mRNA knockdown (25) and for the LCMVgp276–286 epitope by using AAF-CMK (24). Consistent with these results, we found better presentation of the ID OVA epitope on DCs expressing full-length OVA or the minimal ID epitope (Fig. 5⇑A) and delayed degradation of epitope and precursor peptides in TPPII-deficient cytosol preparations (Fig. 5⇑B). These data suggest that TPPII can reduce the half-life of relatively short peptides in vitro as well as in vivo. Whether TPPII destroys only a subset of potential epitopes or epitope precursors with specific sequence features remaining to be defined, is not clear so far. Altogether, TPPII’s contribution to epitope and precursor destruction seems to be only moderate given the slight to moderate increase in MHC class I-peptide assembly upon TPPII siRNA knockdown (23) and in MHC class I expression on KO cells (this study; Fig. 2⇑). Given the large number of cytosolic peptidases, this is not surprising. For the above mentioned study on the tyrosinase tumor Ag, siRNA knockdown experiments have demonstrated that besides TPPII, TOP, PSA, and LAP also contribute to the destruction of the epitope or its precursors (25). For TOP, siRNA and overexpression experiments have suggested a major role in the destruction of potential epitopes including SIINFEKL and of epitope precursors (12). Also in agreement with a predominantly destructive rather than productive role in Ag processing is our finding that TPPII is not induced by IFN-γ (Fig. 1⇑E). Although our study is generally in agreement with experiments by others showing enhanced MHC class I peptide assembly (23) and enhanced presentation of selected epitopes (24, 25) when TPPII activity is reduced, there are also some slight discrepancies to previously published results. These regard presentation of the SIINFEKL (23, 24) and the LCMVgp276–286 (24) epitopes in cell lines treated with TPPII siRNA (23) or AAF-CMK (24). These slight apparent discrepancies might be due to the different cell types and methods of TPPII inhibition or ablation used and the generally rather modest effects of TPPII on Ag presentation.
The role of TPPII in cross-presentation had not been investigated previously. We find that TPPII deletion increases the efficiency of SIINFEKL presentation following internalization of OVA, to an extent resembling that observed in analysis of endogenous OVA presentation. This was true both for phagocytosed OVA and for OVA internalized as immune complex, demonstrating that the two Ag forms are processed in the same proteolytic pathway which we found to implicate the proteasome (data not shown). The somewhat smaller effect of TPPII deficiency on cross-presentation of soluble OVA may be related to a larger contribution of a proteasome-indepedent pathway to presentation of this form of Ag. Thus, the effects of TPPII deletion on cross-presentation are indistinguishable from those on direct presentation, providing further evidence that Ag breakdown in these two pathways involves at least partly overlapping protease sets.
In summary, our study using TPPII KO mice suggests a moderate, predominantly negative or destructive role of TPPII in class I Ag processing. However, this does not exclude the possibility that TPPII may contribute to production of some epitopes, as suggested by previous papers. TPPII thus resembles proteasomes and ERAP1, which both can play productive as well as destructive roles, though, in contrast to TPPII, with a predominance of a productive contribution.
Acknowledgments
We thank Prof. Hermann Frommhold for his support and Dr. Randy Cassada for critical reading of the manuscript. We also gratefully acknowledge the gift of recombinant VV from Dr. Jon Yewdell.
Disclosures
The authors have no financial conflict of interest.
Footnotes
1 This work was supported by the Deutsche Forschungsgemeinschaft (NI 368/4–2), the Clotten Foundation, and a grant from the Forschungskommission of the University of Freiburg Medical Faculty (NIE346/04).
↵2 E.F. and J.H. contributed equally to this work.
↵3 Address correspondence and reprint requests to Dr. Gabriele Niedermann, Clinic for Radiotherapy, University Hospital of D-79106 Freiburg, Freiburg, Germany. E-mail address: gabriele.niedermann{at}uniklinik-freiburg.de
↵4 Abbreviations used in this paper: ER, endoplasmic reticulum; Ct, C-terminal; ERAP, endoplasmic reticulum-associated aminopeptidase; LAP, leucine aminopeptidase; TOP, thimet oligopeptidase; PSA, puromycin sensitive aminopeptidase; TPPII, tripeptidyl peptidase II; KO, knockout; AAF-CMK alanine-alanine-phenylalanine-chloromethyl ketone; DCs, dendritic cells; BM, bone marrow; LCMV, lymphocytic choriomeningitis virus; VV, vaccinia virus; ID, immunodominant; WT, wild type.
- Received August 7, 2007.
- Accepted October 15, 2007.
- Copyright © 2007 by The American Association of Immunologists
References
In this issue
