Conceived and designed the experiments: YG. Performed the experiments: VR EG PL FF. Analyzed the data: VR FJ. Wrote the paper: FJ YG.
The authors have declared that no competing interests exist.
We previously demonstrated that the matrix metalloproteinase-2 (MMP-2) contained an antigenic peptide recognized by a CD8 T cell clone in the HLA-A*0201 context. The presentation of this peptide on class I molecules by human melanoma cells required a cross-presentation mechanism. Surprisingly, the classical endogenous processing pathway did not process this MMP-2 epitope.
By PCR directed mutagenesis we showed that disruption of a single disulfide bond induced MMP-2 epitope presentation. By Pulse-Chase experiment, we demonstrated that disulfide bonds stabilized MMP-2 and impeded its degradation. Finally, using drugs, we documented that mutated MMP-2 epitope presentation used the proteasome and retrotranslocation complex.
These data appear crucial to us since they established the existence of a new inhibitory mechanism for the generation of a T cell epitope. In spite of MMP-2 classified as a self-antigen, the fact that cross-presentation is the only way to present this MMP-2 epitope underlines the importance to target this type of antigen in immunotherapy protocols.
Over the last 20 years, several human tumor antigens recognized by autologous cytolytic T lymphocytes (CTL) have been characterized. Four classes of tumor antigens can be distinguished: cancer testis antigens
We recently described the recognition, on about half HLA-A2 melanoma cell lines, of a new HLA-A2-restricted tumor antigen derived from Matrix Metalloproteinase 2 (MMP-2) by a CTL clone (M134.12) derived from Tumor Infiltrating Lymphocytes (TILs) of a melanoma patient
MMP-2 belongs to the gelatin-binding MMPs structural group. This enzyme contains an amino-terminal signal sequence (Pre) that directs it to the secretory pathway via the endoplasmic reticulum (ER) and a propeptide (Pro) that maintains it as inactive zymogen
In this report, we provide evidence that disulfide bonds contained in the MMP-2 protein prevent endogenous presentation of the appropriate HLA-A2-restricted epitope. Furthermore, we demonstrate that wild type MMP-2 is less sensitive to degradation than mutated forms obtained by disulfide bond deletion. Altogether these results identify a novel mechanism underlying selective tumor cell recognition through cross-presentation of a post-translationally modified ubiquitous protein.
We previously described the recognition by a TIL-derived clone of a MMP-2 derived HLA-A2-restricted peptide cross-presented by αvβ3+ tumor cells. This clone, however failed to react against αvβ3-negative HLA-A2+ cells expressing MMP-2, thus indicating inefficient presentation of the MMP-2560–568 epitope by the endogenous pathway. To address the underlying mechanisms of these observations, we made progressive deletions of the MMP-2 cDNA from the 5′ to the 3′ end by PCR. COS-7 cells co-transfected with plasmids coding for truncated variants of MMP-2 and HLA-A*0201 were then tested for their ability to stimulate TNFα release from the relevant CTL clone (M134.12). The minimally deleted construct thus obtained, corresponding to the cDNA with a 1 to 96 aa deletion (Δ96MMP-2) restored expression of the MMP-2 epitope (
COS-7 cells were cotransfected with HLA-A*0201 plasmid and with (A) plasmids coding for deleted MMP-2 or (B) plasmids coding for mutated MMP-2. 48 h later, M134.12 CTL clones were added to transfected COS-7 cells (E/T ratio 1∶3) and the TNF response was tested after 6 h on Wehi-13 cells. Standard deviations were obtained from duplicates. cDNA NA134-A corresponding to the C-terminal part of MMP-2 contains MMP-2560-568 epitope and was used as positive control. Transfection efficiency was controlled with GFP transfected COS-7 cells. PS corresponding to the signal sequence (pre), PD corresponding to the prodomaine (pro) and PEX corresponding to the hemopexine domaine of the MMP-2. Data are representative of at least two independent experiments. Error bars indicate standard deviations of duplicates.
To address these two hypotheses, we constructed by PCR two deleted forms of MMP-2 cDNA. One form harbored a signal peptide deletion (Δ Pre) and the other, a deletion of the region encoding the propeptide (Δ Pro). Both cDNAs (Δ Pre and Δ Pro) led to effective MMP-2 epitope presentation by HLA-A*0201-transfected COS-7 cells and subsequent CTL clone activation (
To address the role of disulfide bond formation in modulation of MMP-2 endogenous presentation more directly, we generated mutants in which the disulfide bond contained in the prodomain was disrupted. To this end we substituted cystein 60 and/or cystein 65 by an alanine. All mutated MMP-2 constructs (MMP-2 C60A or MMP-2 C65A or MMP-2 C60A/C65A) allowed efficient endogenous expression of the MMP-2 antigen when cotransfected with HLA-A2.1 plasmids into COS-7 cells (
In order to determine whether the endogenous presentation of MMP-2 by tumors cells is similarly affected by the removal of disulfide bonds, we used HLA-A2 tumor cell lines that fail to cross-present MMP-2 due to their lack of αvβ3 integrin
Melanoma cell line M117 and non small cell lung carcinoma line 1355 were transfected with plasmids coding for cystein deleted MMP-2. 48 h later, M134.12 CTL clones were added to tumor cells (E/T ratio 1∶3) and the TNF response was tested after 6 h on wehi-13 cells. Standard deviations were obtained from duplicates. cDNA NA134-A corresponding to the C-terminal part of MMP-2, contains MMP-2560–568 epitope and was used as positive control. Transfection efficiency was controlled with GFP transfected tumor cells. Data are representative of at least two independent experiments. Error bars indicate standard deviations of duplicates. p<0.005 was considered significant.
To determine whether disulfide bond deletion could affect MMP-2 protein secretion, we transfected MMP-2 cDNA with mutated Cys codon into COS-7 cells. 48 h later, culture supernatants were harvested and MMP-2 secretion was assessed by detecting the enzymatic activity of MMP-2 by zymography (
COS-7 cells transfected with indicated plasmids were cultured for 48 h without FCS. Supernatants were collected and equal amounts of protein were loaded on SDS-PAGE. MMP-2 release was analyzed by (A) gelatin zymographie and (B) western blotting.
The fact that disruption of a single disulfide bond within MMP-2 was sufficient to induce endogenous presentation of the MMP-2 epitope, suggests that decreased conformational stability of MMP-2 due to impaired disulfide bond formation lead to enhanced endogenous degradation and subsequent MMP-2560–568 epitope presentation. To test this hypothesis, we sought to determine whether disulfide bond integrity affects MMP-2 stability. Cells were metabolically labeled with [35S] methionine/cystein for 15 mn, followed by a 24 h chase. As shown in
COS-7 cells transfected with indicated plasmids were pulse-labeled with [35S] methionine/cystein for 15 min and chased for 0–24 h. MMP-2 immunoprecipitates were separated by SDS-PAGE and analyzed by autoradiography (A). Data were plotted to indicate the residual protein remaining where the amount of this protein at 0 h time point was calculated to represent 100% of total MMP-2 in each case (B). Data are representative of at least two independent experiments.
Misfolded secretory proteins undergo retrograde translocation from the ER and are then hydrolyzed by the ER-associated degradation pathway (ERAD)
In this current work, we investigated the mechanisms underlying the lack of MMP-2 epitope (560–568) presentation via the classical endogenous pathway by MMP-2 cDNA construct transfection into COS-7 cells and into tumor cells. In this pathway, antigenic peptides derived from endogenous proteins intersect the MHC class I biosynthetic pathway in the ER, where MHC class I heavy chain and β2-microglobulin are synthesized. These peptides are transported from the cytosol to the ER via the transporter associated with antigen processing (TAP) complex and loaded on nascent MHC class I molecule and β2-microglobulin complexes
In 1–96, pre or pro deleted MMP-2, we eliminated a quite high quantity of amino acids compared to mutated MMP-2, where only one cysteine was replaced by an alanine. This probably induces a higher deleted proteins misfolding compared to mutated proteins. This can explain why deleted proteins have significantly better endogenous presentation of MMP-2 antigen than cysteine mutants. However, in this paper we showed that elimination of only one disulfide bond in the MMP-2 protein is sufficient to induce endogenous presentation of the MMP-2560–568 epitope. Further more, we showed that the wild type MMP-2 protein is less degraded than the mutated forms. We analyzed these proteins by circular dichroism (CD). The results showed that the spectra of MMP-2 mutants are found to be practically identical to the spectrum of wild-type MMP-2 (data not shown). So, the CD analysis confirms the protein folding and shows that the mutations do not induced significant changes in the secondary structure of mutated MMP-2. However, this absence of changes in secondary structures does not exclude other conformational changes in the whole MMP-2 protein. Indeed, based on MMP-2 crystal structure (Morgunova et al., 1999) we learnt that disulfide bond C60–C65 is located in the loop between the helix H1 and H2 of the prodomain. This prodomain maintained MMP-2 as a zymogen and this loop is cleaved by MMP-14 after MMP-2 secretion given active form. Moreover disulfide bond C233–C247 is located in the first Fibronectin type II domain and this domain is necessary for substrate binding. So these two disulfide bonds seem to be located in important regions for wild type MMP-2. These observations and our experiments probably mean that, for wild type MMP-2, quality control of these two regions is essential, so that none, or few, misfolded enzymes are naturally produced and consequently no, or few, MMP-2 epitope presentation occur. Altogether these results suggest that MMP-2 folding has an inhibitory role on the endogenous degradation of MMP-2 by the proteasome and its subsequent immunogenicity.
Disulfide bond formation in the endoplasmic reticulum of eukaryotic cells is catalyzed by the protein disulfide isomerase (PDI) and other members of the thioredoxin family
In a similar way, the structural complexity of a protein and especially the presence of disulfide bonds, seems to be critical for MHC class II presentation. Indeed disulfide bond reduction of proteins facilitates processing and presentation to antigen-specific CD4 T cells
By preventing the formation of at least one disulfide bond, we demonstrated that the redox status of MMP-2 significantly increased its degradation and resulted in processing of the specific T cell epitope by the endogenous pathway. Furthermore it did not prevent MMP-2 secretion. Based on these findings, we propose that disruption of one disulfide bond is detected by the quality control mechanism present in the ER
According to this hypothesis, we propose in
Cross-presentation (1): Newly synthesized wild-type MMP-2 acquire disulfide bonds in the endoplasmic reticulum (ER) before joining the secretory pathway. In the extracellular environment, physiologic activation of the pro-MMP-2 induce the cleavage of the propeptide domain which contains a disulfide bridge (C60-C65: unique to the MMP-2). MMP-2 active form then interact with the integrine αvβ3 and is internalized in clathrin-coated vesicle. Finally MMP-2 is transported to the cytosol, in an unknown mechanism, and degraded by the proteasome. Peptides generated can reach the endogenous pathway (peptides are transported in the ER through TAP, bind to HLA-A*0201 and transported to the cell surface). Endogenous presentation (2): Mutated MMP-2 lacking a disulfide bond can't join the secretory pathway and is retrotranslocated via Sec61. In the cytosol, mutated MMP-2 is degraded by the proteasome and resulting peptides are loaded on MHC class I molecules.
MMP-2 is a self-antigen secreted by many cells and involved, through the degradation of the extracellular matrix and basal membranes, in multiple physiological processes and in tumor progression. Nevertheless, the MMP-2 epitope was found to be cross-presented exclusively by melanoma cells and not presented by the endogenous pathway
The CTL clone M134.12, melanoma cell line M117 were established in our laboratory (Godefroy et al 2005). M134.12 was obtained from Tumor Infiltrating Lymphocytes (TIL) of melanoma patient M134 by limiting dilution and selected for it specificity against the antigen MMP-2 in an HLA-A*0201 context. Non-small cell lung cancer cell line (1355) (established by H. Oie) was gift from C. Saï (UMR 892 INSERM/Université de Nantes, France). Mouse fibrosarcoma WEHI 164 clone 13 (established by Rollinghoff and N.L. Warner) and COS-7 cells (established by Y. Gluzman) were gifts from T. Boon (Ludwig Institute for Cancer Research, Brussels, Belgium).
Tumor cell lines and COS-7 cell were maintained in RPMI-1640 and DMEM 1 g glucose/L respectively, supplemented with penicillin-streptomycin (100 U/ml and 10 µg/ml respectively), L-Glutamine (2 mM) and 10% fetal calf serum (all from Sigma-Aldrich). The CTL clone was maintained in RPMI-1640 supplemented with penicillin-streptomycin, L-Glutamine, IL-2 (150 U/ml) and 8% pooled human serum (pHS).
Written consents were obtained from all patients and healthy donors. The local ethics committee “Comité de Protection des Personnes Ouest IV- Nantes” and the “Agence française de sécurité sanitaire des produits de santé” approved all these studies.
NA-134A corresponding to the end of the MMP-2 sequence and containing the epitope was obtained from a cDNA library of melanoma cell line M134. Truncated MMP-2 Δ96MMP-2 was obtained by PCR. ΔpreMMP-2 and ΔproMMP-2 were obtained by deletion by PCR of the nucleotide sequence coding for the signal peptide (Δpre) or the propeptide (Δpro). Point mutations were introduced into the cDNA coding for MMP-2 using directional mutagenesis: site-directed mutagenesis was performed using primers with nucleotide substitutions in order to change cystein in alanine at amino acid 60 and/or 65 and 233.
Each of the cDNA constructs was digested with EcoRI and XbaI, inserted into pcDNA3.1 vector (Invitrogen) and plasmids were electroporated into a bacterial strain (Escherichia coli TOP 10 F').
We used the DEAE-dextran-chloroquine method as described (Brichard et al. 1993, Coolie et al. 1994).
Briefly: 1,5.10 4/well COS-7 cells were plated in flat-bottom 96 well plates. After 24 h cells were cotransfected with 125 ng of plasmids containing the cDNA coding for the different constructs of MMP-2 and with 125 ng of plasmids containing the cDNA coding for HLA-A*0201. Each transfection condition was performed in duplicate. After 48 h, cells were tested in a CTL stimulation assay. Transfection efficiency was controlled by transfected COS-7 cells with plasmid coding for the GFP and measured by flow cytometry analysis.
1.106 COS-7 cells were transfected with 5 µg of plasmid (coding for mutated MMP-2) according to the manufacturer's instructions.
For pulse-chase experiments, 7.106 COS-7 transfected with the same plasmid were pooled in 100 mm dishes and cultured 48 h at 37°C before metabolic labeling.
For zymography and Western blotting assays, 1.106 transfected COS-7 were plated in a 6-well plate. 1 day later, cells were washed and cultured for 48 h in medium without FCS.
2.104/well tumor cell lines were plated in flat-bottom 96 wells plates 24 h before transfection. Cells were washed with OPTI-MEM medium (Invitrogen) and were transfected using lipofectamine and Plus reagent (Invitrogen) and 100 ng of plasmid containing the cDNA coding for the mutated MMP-2, in OPTI-MEM. All transfection conditions were performed in duplicate. After 48 h, cells were tested in a CTL stimulation assay. Transfection efficiency was controlled by transfected tumor cells with plasmid coding for the GFP and measured by flow cytometry analysis.
COS-7 cells were transiently cotransfected with plasmids coding for mutated MMP-2 and for HLA-A*0201 using DEAE-dextran-chloroquine method. 24 h later, COS-7 cells were treated by addition to the culture medium of 0,38 µM proteasome inhibitor MG-132 (Calbiochem) or 3 µg/ml sec61 inhibitor Exotoxin A (Sigma-Aldrich). After overnight incubation, cells were washed, counted and tested in CTL stimulation assay.
Transfected cells were tested for their capability to present a MMP-2 derived HLA-A*0201 and consequently to stimulate the production of Tumor Necrosis Factor (TNF) by the specific CTL M134.12.
1.104 CTL were added to 3.104 stimulator cells (transfected tumoral or COS-7 cells) for 6 h at 37°C. Then, culture supernatants were collected and TNF release by CTL was measured by testing TNF cytotoxicity on WEHI 164 clone 13 cells in a MTT colorimetric assay (Espevik et al, 1986).
Secretion of MMP-2 was evaluated by gelatin zymography, as previously described (Barille et al. 1999).
Supernatants were obtained from transfected COS-7 cells (Amaxa) cultured in DMEM 1 g glucose/L without FCS during 48 h and total proteins concentrations were measured (BC Assay, Interchim). 22,5 µg of total protein were mixed with sample buffer without reducing agent and proteins were separated by SDS-PAGE (7,5% acrylamide gels containing 0,2% gelatin). After electrophoresis, SDS was removed from the gel by an incubation in 2,5% triton X-100 for 1 h at room temperature. Gels were then incubated in a buffer containing 50 mmol/L Tris-HCl, 5 mmol/L CaCl2, pH 7,6 for 20 h at 37°C and stained with Coomassie blue R250 for 30 min. Proteolytic activity of MMP-2 was evidenced as a clear band against the blue background of stained gelatin.
Supernatants were obtained from transfected COS-7 cells (Amaxa) cultured in DMEM 1 g glucose/L without FCS during 48 h and total proteins concentrations were measured (BC Assay, Interchim). 50 µg of total protein were mixed with sample buffer containing reducing agent and proteins were separated by SDS-PAGE (7,5% acrylamide), electrotransferred to polyvinylidene difluoride (PVDF) membrane and blocked 1 h in 1% Western Blocking Reagent (Roche Diagnostic) TBS 1X. Blots were probed overnight at 4°C with a rabbit polyclonal antibody anti-MMP-2 hinge region (Biomol, Tebu-bio SAS, France), washed in TBS 1X, 0,1% Tween-20 and probed with HRP-conjugated anti-rabbit, anti-mouse antibody (Roche). The signal was detected by enhanced chemiluminescence detection (Perbio).
48 h after transfection using the Cell Line Nucleofector™ kit V (Amaxa), 7.106 transfected COS-7 were incubated at 37°C for 1 h30 in cystein and methionine-free DMEM 4,5 g glucose/L (Invitrogen). Cells were then pulsed for 15 min with 350 µCi of [35S] methionine/cystein (GE Healthcare), plus Brefeldin A (BFA) 10 µg/ml (Sigma-Aldrich) in 5 ml of DMEM 4,5 g glucose/L without cystein and methionine. Cells were then recovered and redistributed: 1.106 cells in 2 ml of DMEM containing FCS plus BFA, and chased from 0 to 24 h. After chase, cells were pelleted, resuspended in lysis buffer (10 mM Tris-HCl pH 7,6, 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, 2 µg/ml aprotinin, 1% Triton X-100) for 45 min on ice and centrifugated at 10000 g for 30 min at 4°C. Protein concentration in supernatants was measured using bicinchoninic acid (BCA Protein Assay, Pierce, Rockford, IL). 70 µg of lysates were precleared with protein G Agarose (Perbio) and then immunoprecipitated overnight at 4°C with a rabbit polyclonal antibody anti-MMP-2 (H-76, Santa-Cruz, Tebu-bio SAS, France) and protein G agarose. Beads were pelleted, washed and boiled in SDS sample buffer. Proteins in immunoprecipitated samples were separated by SDS-PAGE (7,5% acrylamide) and analyzed by autoradiography.
Statistical analysis was done with InStat 2.01. Data were analyzed using ANOVA test. p<0.005 was considered significant.
We thank M. Bonneville and S. Amigorena for critically reviewing the manuscript.
We thank J. Ménager for proteins purification.