Abstract
The signal peptide peptidase (SPP) is an intramembrane cleaving aspartyl protease involved in release of leader peptide remnants from the endoplasmic reticulum membrane, hence its name. We now found a new activity of SPP that mediates liberation of C-terminal peptides. In our search for novel proteolytic enzymes involved in MHC class I (MHC-I) presentation, we found that SPP generates the C-terminal peptide-epitope of a ceramide synthase. The display of this immunogenic peptide–MHC-I complex at the cell surface was independent of conventional processing components like proteasome and peptide transporter TAP. Absence of TAP activity even increased the MHC-I presentation of this Ag. Mutagenesis studies revealed the crucial role of the C-terminal location of the epitope and “helix-breaking” residues in the transmembrane region just upstream of the peptide, indicating that SPP directly liberated the minimal 9-mer peptide. Moreover, silencing of SPP and its family member SPPL2a led to a general reduction of surface peptide–MHC-I complexes, underlining the involvement of these enzymes in Ag processing and presentation.
This article is featured in In This Issue, p.3971
Introduction
All nucleated cells of our body display a representative selection of the cellular proteome in MHC class I (MHC-I) molecules at their surfaces. This wide array of peptides serves as ligands for CD8+ T cells for cellular immunity. The process of peptide presentation starts with the proteolysis of aged or misfolded proteins, a process mainly executed by the multicatalytic action of proteasomes. The generated peptide sequences are translocated from the cytosol into the endoplasmic reticulum (ER) by the TAP peptide transporters, where they are assembled with different MHC-I molecules with help from members of the peptide loading complex, including calreticulin, tapasin, and ERp57. Stable peptide–MHC-I complexes are routed to the cell surface. This proteasome–TAP pathway is considered as the conventional processing route used to generate and shuttle the majority of peptides (1–3). Additional players, like tripeptidyl peptidase II and nardilysin, are responsible to further shorten proteasomal intermediates or, sometimes, to directly generate minimal peptide sequences that fit in MHC-I molecules (4, 5).
Although these central elements of the conventional Ag processing pathway are important for the mainstream of peptides, substantial MHC-I Ag presentation is still detectable on cells in which the function of key components is compromised. For example, deficiency in the peptide transporter TAP, which can be regarded as a bottleneck in the pathway, results in a decreased surface MHC-I, but the residual peptide repertoire is sufficient to generate a diverse CD8+ T cell subset in mice (6–8) and humans (9, 10). Importantly, this T cell repertoire is capable of controlling common viral infections, and from some TAP-deficient patients, TCRαβ+ CD8+ T cell clones were isolated that recognized TAP-independent viral Ags (11, 12). Therefore, TAP-independent processing pathways may sufficiently compensate for the loss of the conventional route to select a functional CD8+ T cell repertoire in these patients.
This reasoning is in line with our recent findings on a novel category of T cell epitopes presented by cancer cells with TAP defects. We showed that deficiency in the expression of the peptide transporter TAP, a feature frequently found in human cancers, leads to the presentation of a broad repertoire of immunogenic peptides by MHC-I (13–17). Biochemical characterization of peptide repertoires from TAP-deficient cells and their direct proficient counterparts indeed revealed unique complementing pools of peptides (14, 18, 19).
Two pathways for TAP-independent peptide loading are described thus far (20). One of these is active in the secretory route of the Golgi and is mediated by the protein convertase furin (21–23). The other concerns N-terminal leader sequences, which are liberated after arrival in the ER by the signal peptidase (SP). Cleavage by SP leaves small transmembrane leader peptide remnants in the ER membrane, which are removed by the intramembrane-cleaving signal peptide peptidase (SPP). The ER-directing parts of these peptides end up in the lumen of the ER and can be loaded unto MHC-I molecules (24, 25). An immunogenic human tumor Ag from the leader sequence of calcitonin is released by this means, and TAP is dispensable for its presentation on cancer cell lines (26, 27).
In this study, we reveal the processing pathway of a C-terminal peptide from the ceramide synthase Trh4 (also known as CerS5). This protein resides in ER membranes and docks its C-terminal epitope-containing tail into the bilayer. Intramembrane proteolysis by SPP just in front of the epitope resulted in liberation and TAP-independent loading on Db molecules. This represents a novel role for SPP, acting independently from the SP. Our data indicate that SPP is able to directly liberate minimal peptide-epitopes from the C terminus of proteins by intramembrane proteolysis and underlines the involvement of this enzyme in Ag processing.
Materials and Methods
Cell lines
The mouse tumor cell lines RMA, RMA-S (TAP2-deficient), MCA (TAP1-deficient), MC38, B78H1 (TAP2-deficient), and human HeLa and 293T have been described previously (14, 17, 28). Where indicated, cell lines were transfected with H-2Db or Trh4 genes (accession no.: gene BC043059 [http://www.ncbi.nlm.nih.gov/nuccore], protein UniProtKB Q9D6K9-2 [http://www.uniprot.org/uniprot]) via retroviral transduction using the LZRS vector containing GFP behind an internal ribosome entry site (16). The mouse TAP genes were introduced into MCA and B78H1 cells by gene transfer. MC38.UL49.5 cells contain the UL49.5 gene from the Bovine Herpes Virus-1, which blocks mouse TAP activity (16, 29).
Generation and culture of CD8+ T cell clones used in this study were previously described: Db-restricted CTL clone B5 specific for the TAP-independent Trh4 peptide (MCLRMTAVM), Kb-restricted CTL clone mi3 specific for a yet unknown TEIPP peptide, and the Db-restricted CTL clone 1 specific for the TAP-dependent MuLV gag-leader peptide (CCLCLTVFL) (16, 17, 30).
All cells were cultured in complete IMDM (Invitrogen, Carlsbad, CA) containing 8% heat-inactivated FCS, 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies, Rockville, MD), 2 mM l-glutamine (Invitrogen), and 30 μmol/L 2-ME (Merck, NJ) at 37°C in humidified air with 5% CO2.
Protease inhibitor experiments
Cells were resuspended in X-vivo medium (Lonza) supplemented with 0.5% of BSA and incubated with protease inhibitors for 1 h at 37°C under continuous mixing. Cells were incubated for 2–4 min with mild acid citrate/phosphate buffer (pH 3.1) at room temperature to disrupt MHC-I–peptide complexes (28). Cells were washed twice and incubated with protease inhibitors for additional 6 h under the conditions mentioned earlier and then treated with brefeldin A. Inhibitor-treated cells were then incubated with the indicated T cell clones for 18 h in the presence of brefeldin A and intracellular stained for IFN-γ production (28). Chemical inhibitors used in these assays are epoxomicin (1 μM; Sigma-Aldrich), 3,4-dichloroisocoumarin (5 μM; Sigma-Aldrich), decanoyl-RVKR-CMK (10 μM; Merck), calpeptin (30 μM; Merck), calpain inhibitor IV (13 μM; Merck), calpastatin peptide (10 μM; Merck), PD151746 (50 μM; Santa Cruz Biotechnology), PD150606 (50 μM; Santa Cruz Biotechnology), butabindide oxalate (400 μM; TOCRIS Bioscience), captopril (100 μM; Sigma-Aldrich), leupeptin (100 μM; Merck), (z-LL)2-ketone (5 μM; Merck), 1,10-phenanthroline monohydrate (400 μM; Sigma-Aldrich), aprotinin (10 μM; Merck), DAPT (10 μM; Sigma-Aldrich), bestatin hydrochloride (100 μM; Sigma-Aldrich), z-vad-fmk (10 μM; Merck), l-leucinethiol (60 μM; Sigma-Aldrich), and DL-DTT (5 mM; Sigma-Aldrich). None of the inhibitors was toxic at the applied concentrations, as determined by trypan blue counting and Annexin V staining.
T cell activation assays and flow cytometry
Intracellular cytokine staining of IFN-γ in T cell clones was performed for the protease inhibitor screen and expressed as percentage positive cells (31). IFN-γ secretion by T cell clones was measured by ELISA as previously published (28). Data shown represent mean values obtained from triplicate test wells, and error bars represent SD of these values. Flow cytometry analysis of surface expression of Db and Kb molecules was performed using directly labeled anti-Db b
Quantitative PCR analysis
Total RNA isolation was performed using RNeasy Mini Kit (Qiagen, Gaithersburg, MD). A total of 500 ng purified total RNA was used to synthesize cDNA using High Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA). Quantitative PCR (qPCR) on two different transcripts of Trh4 was done using unique forward primers. Forward primer for the Trh4-epitope containing transcript was 5′-GCAGACCCCTTACTGGAAGCTGC-3′ and for the short transcript 5′-TACATCACTGCGGTCATC-3′. Common reverse primer was 5′-CTGCGGTCATCCTTAGACACCTTTCC-3′. The primers used to detect the expression of mouse transcripts from SPP, SPPLs, and calpains were designed by us using the Beacon designer software and are listed in Supplemental Table I (calpain family) and Table I (SPP family). All primer sets were validated to be specific for the cognate genes by sequencing of the amplified PCR products. For the PCR, SensiMix SYBR No-ROX kit from GC Biotech Bioline (Alphen aan den Rijn, The Netherlands) was used in a C1000 Thermal Cycler (Bio-Rad, Hercules, CA), and results were analyzed using Bio-Rad CFX manager software.
Confocal microscopy
Human HeLa cells or mouse MCA cells were plated in 12-well plates on glass coverslips and transfected the next day with the indicated Trh4 gene constructs using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. After 1 d, cells were stained with mAbs: cells were fixed with methanol for 30 min at 4°C, washed with TBS with blocking reagent from Boehringer Mannheim (TBSB), and incubated for 1.5 h at room temperature with the first Ab diluted in 500 μl TBSB. The Abs used were anti-V5 (R960-25; Invitrogen), anti-Giantin (ab24586; Abcam), and anti-BIP (ab21685; Abcam). Cells were then washed again with TBSB and incubated with second Abs: anti-rabbit Alexa 488 (a11008; Invitrogen) and anti-mouse Alexa 594 (a11005; Invitrogen) for 1.5 h at room temperature. Finally, cells were washed with TBSB and the coverslips were put on microscope glass slides with 5 μl ProLong Gold antifade reagent (Invitrogen) containing Hoechst (Hoechst 33342; Enzo). The stained cells were imaged with a TCS SP5 fluorescent confocal microscope (Leica, Germany) using LAS AF software (Leica). Stained samples were observed at room temperature. All images were acquired with a 63× glycerol immersion objective lens NA 1.4 (Leica). Analysis of colocalization was performed with LAS AF software.
Short hairpin RNA treatment
MISSION short hairpin RNA (shRNA) constructs specific for mouse SPP, SPPL2a, SPPL2b, SPPL3, and calpain 2 were obtained from Sigma-Aldrich, and lentiviral encapsulation was reached in HEK.293T cells via transfections with calcium phosphate (Promega) of pCMV-VSVG, pMDLg-pRRE, pRSV-REV, and pLKO plasmid with puromycin resistance gene and an shRNA sequence or empty vector. RMA-S cells were incubated with lentivirus particles encoding shRNA and particles encoding GFP in a 10:1 ratio for 20 h at 37°C. Then the cells were washed and resuspended in complete medium supplemented with 1 μg/ml puromycin and cultured for 4 d. GFP-expressing cells were sorted and used in experiments.
Trh4 gene constructs
Length variants of Trh4 (deletions and extensions) were performed on the cloned cDNA (accession no. BC043059) in pcDNA3.1 plasmid. C-terminal extensions were created by removing the stop codon of Trh4 in the pcDNA3.1/V5-His vector (Invitrogen, Life Technologies), resulting in the in-frame encoding V5 Ab epitope. The deletion variants were based on an N-terminally tagged V5 construct in which the Ab site was introduced after nucleotide number 96, which was supposed to be the leader sequence. We introduced a NOT-I restriction site at this place by creating two complementary PCR products, resulting in two altered amino acids at position 32–33 (from DL to AA). Then the 14-aa-long V5 sequence was introduced at this NOT-I site using long synthetic oligonucleotides including NOT-I overhangs. Orientation and sequence of this insert was validated by sequencing. This “N-terminal” tagged Trh4 construct was then used to create several deletion variants using PCR primer sets. All PCR amplifications were performed with Accuprime kit (Invitrogen).
Site-directed mutagenesis was done on the N-terminal V5 construct in pcDNA3.1 using USB Change-IT Multiple Mutation Site Directed Mutagenesis technology (Affymetrix). Generated clones were sequenced to validate the introduced changes.
Western blot
Cells were lysed in 1% Nonidet P-40 lysis buffer (50 mM Tris-HCl, 150 mM NaCl [pH 8.0], 1% Nonidet P-40, 1 mM leupeptin, 1 mM 4-[2-aminoethyl]-benzene-sulfonyl fluoride hydrochloride [Sigma-Aldrich, Zwijndrecht, The Netherlands]). The samples were mixed 1:1 with sample buffer (0.1 M Tris [pH 8.0], 4% SDS, 20% glycerol, 10% 2-ME, 0.05% bromophenol blue) and heated up to 95°C for 1 min. The samples were loaded in precast gels 4–15% (Bio-Rad) using the Mini-PROTEAN Tetra 4-gel vertical electrophoresis system (Bio-Rad). The BenchMark Pre-Stained Protein Ladder (Invitrogen) was included. Gels ran for 25 min at 250 V and blotted in Trans-Blot Turbo Transfer System (Bio-Rad) according to the manufacturer’s recommendations. After blotting, the membrane was incubated overnight at 4°C in 5% low-fat milk/TBST (10 mM Tris-base, 150 mM NaCl, 0.05% Tween 20, pH 7.4). The next day, the blots were stained with anti-V5 mAb 1:2500 (Invitrogen) in 1% low-fat milk/TBST for 2 h at room temperature. After the staining, the membrane was washed three times for 5 min in TBST. Next, the membrane was incubated with HRP-labeled goat anti-mouse Ig Ab (1:2000) in 1% low-fat milk/TBST for 2 h at room temperature, followed by three times washing in TBST and three times in TBS. Visualization of the blot was performed using ECL Plus Western blotting detection reagents (GE Healthcare) on films.
Results
The TAP-independent processing pathway of the Trh4-derived peptide is common and conserved
The Trh4-derived peptide MCLRMTAVM is presented by the MHC-I molecule H-2Db on TAP-deficient tumor cells (17). We were interested in the nature of its TAP-independent processing pathway that leads to the generation and surface presentation of the Trh4 peptide in Db and, therefore, analyzed the generality of this Ag-processing pathway in different mouse and human tumor cell lines (Fig. 1). Four mouse tumor cell lines of different origin and containing a TAP defect (a lymphoma, a fibrosarcoma, a colon carcinoma, and a melanoma) were analyzed for recognition by a previously established MCLRMTAVM-specific CD8+ T cell clone. These T cells efficiently recognized all TAP-deficient tumors, irrespective of their tissue origin (Fig. 1A). Importantly, the different mechanisms underlying the TAP impediment in these tumors, being TAP2 mutation (in RMA-S), TAP1 gene loss (in MCA), expression of a small viral inhibitor (in MC38), or TAP2 gene loss (in B78H1), all led to efficient presentation of the Trh4-derived peptide, indicating that presentation of the Trh4 peptide is related to TAP deficiency as such and not to a specific molecular mechanism. The fact that TAP+ counterpart tumors presented much lower Trh4 peptide at their cell surface was not surprising, because we previously showed that restoration of TAP function decreases Trh4/Db complexes at the cell surface, most likely because of competition (28). Therefore, the Trh4 peptide presentation is TAP independent and operates in different tumor types.
The Trh4 processing pathway is conserved and broadly active. (A) Presentation of Trh4-derived peptide by MHC-I was evaluated on lymphoma (RMA), fibrosarcoma (MCA), colon carcinoma (MC38), and melanoma cells (B78H1.Db). Activation of Trh4-specific T cells is enhanced by the absence on peptide transporter TAP. The black bar represents a culture with 20,000 target cells and the grey bars represent cultures with five-fold serial dilutions of the target cells. (B) mRNA expression of Trh4 and a natural Trh4 splice variant was determined by qPCR. The splice variant does not encode the Th4 peptide sequence because of a frameshift. One of three comparable experiments is shown. (C) Human 293T and HeLa cells were transfected with mouse Trh4 and H-2Db genes, and used as targets. The black bar represents a culture with 20,000 target cells and the grey bars represent cultures with five-fold serial dilutions of the target cells. Mean and SD of triplicates are shown from one of three experiments.
The degree of CTL recognition of the tumor cell lines varied and we wondered whether these differences could be explained by Trh4 expression levels. Trh4 transcripts were quantified by qPCR and indeed revealed higher expression in the lymphoma RMA-S compared with the solid tumors (Fig. 1B). Expression of a splice variant of Trh4, which does not contain the Db-binding peptide, showed less variety in the tumor panel. We concluded that surface display of Trh4/Db complexes is determined by TAP function, Trh4 protein levels, and possibly other factors like the protease responsible for peptide liberation.
We also wanted to know whether human cells were equipped with the TAP-independent processing pathway. The mouse Trh4 gene was introduced into the human cell lines HeLa and HEK293T, together with the mouse MHC-I gene H-2Db that presents this peptide. The Trh4-derived peptide was efficiently presented at the cell surface of both human lines, indicating that the involved processing pathway was indeed also active in human cells (Fig. 1C). Although these human cells have intact TAP function and thereby impede Trh4 peptide presentation, the high expression levels of Trh4 protein, from strong heterologous promoters, overcome the competition of peptides from the proteasome–TAP pathway, as illustrated by efficient CTL recognition of TAP+ RMA lymphoma cells in which Trh4 is expressed 800 times higher than the endogenous level (28). Together, these data show that the TAP-independent processing pathway that generates the Trh4 peptide is a common and conserved mechanism operational in mouse and human tumor cell lines of different histological origins.
Trh4 protein is localized in the membrane of the ER
To unravel the processing pathway of the Trh4-derived peptide, we next determined the cellular localization the Trh4 protein. The Trh4 protein (also known as CerS5) belongs to a family of ceramide synthases (CerS1 to CerS6) and exists in two isoforms as a result of alternative splicing of the transcript (17). The shorter protein variant (UniProt Q9D6K9-2) contains the Db-presented peptide MCLRMTAVM at the complete end of its C terminus. The longer protein (UniProt Q9D6K9-1) lacks this peptide sequence because of a frameshift at the last intron–exon boundary (17). Although the longer isoform was described to be a multitransmembrane-spanning protein in the ER (32), no data were available concerning the Trh4 variant comprising the peptide-epitope. Algorithm programs predicting topology of transmembrane proteins anticipated that the short Trh4 protein would transgress membranes seven times with the C-terminal aa 373–384 located within the lipid bilayer, directed toward the lumen (Fig. 2A). The V5 Ab epitope was introduced at two distinct sites of Trh4 gene constructs (C terminus after the epitope and N terminus at aa 32), and subcellular localization studies were performed using confocal microscopy. We observed a strong and clear colocalization with ER markers, but not with the Golgi with both constructs (Fig. 2B for N-terminal tag and Supplemental Fig. 1A for C-terminal tag). ER colocalization was also observed in TAP− cells (Supplemental Fig. 1B). An intermediate degree of colocalization with mitochondria was found (Fig. 2B). It is known that the CerS1, CerS2, and CerS6 ceramide synthases can indeed be detected in mitochondria (33); but, alternatively, this mitochondria colocalization might merely reflect ER membranes that are in close proximity of mitochondria. These data indicate that Trh4 is a multimembrane-spanning protein in ER membranes, and that its C-terminal tail, comprising the epitope, docks into this membrane directing toward the ER lumen.
The C-terminal tail of Trh4 is localized within the ER membrane. (A) Prediction of the Trh4 topology in membranes according to algorithm software (http://www.predictprotein.org). The C-terminal peptide-epitope is indicated in red. (B) Confocal microscopy imaging of V5-tag at the N terminus of Trh4 protein, in red color, in combination with staining for ER (BiP), Golgi (giantin), or mitochondria (mitofilin), in green color, respectively, using fluorescently labeled Abs. Original magnification ×1000. Mean percentages of colocalization were calculated on 10 images of at least three independent experiments.
Inhibitors of intramembranous aspartyl proteases prevent peptide liberation
Next, we investigated which family of proteases was responsible for the proteolytic cleavage of Trh4 resulting in liberation of the C-terminal peptide. An array of inhibitors blocking hydrolysis by different classes of proteases were applied in a recovery assay in which TAP-deficient RMA-S cells were briefly acid stripped to remove cell-surface MHC-I and allow for re-emergence of Trh4/Db peptide complexes. Complete recovery of Trh4/Db was reached within 6 h, as determined with a peptide-specific T cell clone (Fig. 3A). This highlighted a continuous and efficient supply of Trh4 peptide from this alternative processing pathway. Two enzymes were likely candidates, because they were described to result in TAP-independent peptide loading: SP and furin (20, 21). However, the specific inhibitors DCI and Dec-RVKR-CMK, respectively, failed to impair the generation of Trh4/Db complexes (Fig. 3B), indicating that Trh4 processing is mediated by a novel mechanism. In line with our previous data, inhibition of the proteasome rather increased Trh4/Db complexes (Fig. 3B), suggesting that the proteasome somehow acts as a competitor for the novel processing mechanism (28).
The Db-presentation of the Trh4-derived peptide is generated by intramembrane-cleaving aspartyl proteases. (A) Recovery kinetics of Trh4/Db complexes on TAP-deficient RMA-S cells after brief stripping with mild acid buffer. Cells were tested with Trh4-specific T cell clone in intracellular IFN-γ staining. (B) Recovery efficiency of Trh4/Db complexes was tested in the presence of an array of protease inhibitors. Inhibitors of proteasome and enzymes known to be involved in TAP-independent processing pathways, (C) inhibitors of calpain family, and (D) all other tested inhibitors. (E) TAP-deficient fibrosarcoma MCA cells were tested with inhibitors for aspartyl proteases. (F) Concentration range of the functional inhibitor (z-LL)2-ketone and (G) DAPT on RMA-S cells. Trh4-specific T cells (black bars) were used and control T cells (white bars) that recognize an independent Kb-presented peptide on RMA-S. All panels show compiled means and SDs from at least three independent experiments, setting the recovered cells at 100%.
Next, we observed that the common calpain inhibitor calpeptin partly decreased Trh4 presentation at relatively high concentrations (>20 μM; Fig. 3C, Supplemental Fig. 2A). The main targets for calpeptin are calpain 1 and calpain 2. We determined the expression of these two calpain members in RMA-S cells and observed that calpain 2 was expressed, but calpain 1 was not detectable (Supplemental Fig. 2B). Therefore, we silenced the expression of calpain 2, but this did not lead to reduced Trh4/Db peptide complexes at the cell surface (Supplemental Fig. 2C, 2D). Furthermore, more selective inhibitors for the calpain family did not affect the CTL response, collectively suggesting that the marginal blocking effect by calpeptin was based on cross-reactivity to other proteases (Fig. 3C).
We then continued the screen with several other classes of chemical inhibitors targeting serine-, cysteine-, aspartyl-, and metallo-proteases and aminopeptidases (Fig. 3D). The results revealed two inhibitors for aspartyl proteases, (z-LL)2-ketone and DAPT, that strongly decreased presentation of the Trh4-derived peptide. Importantly, none of the other eight protease inhibitors influenced the Trh4/Db presentation, suggesting that the involved aspartyl protease was necessary and sufficient. The inhibitory effect was also observed when we tested fibrosarcoma cells, indicating that the inhibition was not cell-type restricted (Fig. 3E). Titration of the two active compounds demonstrated a 50% decrease at concentrations in the lower range in this cellular assay, 300 nM for (z-LL)2-ketone and 5 μM for DAPT (Fig. 3F, 3G). These concentrations were clearly not toxic for the target cells, and moreover, recognition of treated RMA-S cells by control T cell clones specific for another peptide epitope was unaltered, excluding a general defect in the MHC-I presentation capacity by (z-LL)2-ketone and DAPT (Fig. 3F, 3G). These two compounds were known to target aspartic acid proteases that are located within membranes, particularly members of the SPP family (34, 35). Interestingly, calpain inhibitors were shown to block the action of SPP at higher concentrations, corroborating our findings in the screen with calpeptin (Fig. 3C) (36). Together, these results let us conclude that the liberation of the C-terminal Trh4 peptide within the ER membrane is part of a novel processing pathway for TAP-independent peptides and is distinct from the previously described routes.
SPP is responsible for the processing of the C-terminal peptide
The SPP family is also known as intramembrane-cleaving proteases of the aspartyl type, and this family is involved in many biological processes, including cell proliferation, differentiation, and immunity (34, 37, 38). SPP and SPPL3 were reported to be localized in ER membranes, forming catalytic cavities in which hydrolysis can take place within membranes (39). We first analyzed the expression of the family members at mRNA level in the tumor cell lines active in Trh4 peptide processing (Fig. 4A, Table I). All family members were expressed to varying degree in this cell panel, except SPPL2c, which was hardly detectable. High levels of SPP were measured in all cells. Differences in expression levels in the tumor cell panel prompted us to profile transcripts of these genes in normal mouse tissues, especially because this information is lacking in the current literature (Supplemental Fig. 2E). RNA extracts from whole mouse organs were analyzed for these intramembrane aspartic peptidases, and a general low expression for SPPL2b and SPPL2c was found. Similar expression profiles were seen for SPP, SPPL2a, and SPPL3 in that thymus and brain contained low transcript numbers and lung was high (Supplemental Fig. 2E).
The intramembrane-cleaving SPP is involved in Trh4 processing. (A) mRNA expression of SPP, SPPL2a, SPPL2b, SPPL2c, and SPPL3 genes was determined by qPCR in tumor cell panel. (B) Silencing of SPP family members in RMA-S cells was reached by lentiviral transfer of shRNA constructs. Empty shRNA vector served as control. Efficiency and specificity of downregulated levels is shown as measured by qPCR. (C and D) Cell panel with silenced single SPP family members was tested for recognition by Trh4-specific T cells (C) or control T cells (D) reactive to the independent Kb-presented peptide mi3 by RMA-S. Mean and SD from one of three experiments is shown. Difference between SPP and control vector is statistically significant (p < 0.001, two-way ANOVA test). Similar data were obtained in the fibrosarcoma cell line (Supplemental Fig. 2F, 2G).
Next, we wanted to determine which family member was responsible for the liberation of the C-terminal Trh4 peptide in tumor cells and silenced each gene individually using shRNA constructs. SPPL2c was not included in this analysis because it was not expressed in the tumor cells (Fig. 4A). Efficiency of gene silencing was ∼50–60% for all genes and, importantly, the shRNA constructs were specific in that the other members were not targeted (Fig. 4B). We then used this cell panel to determine effects on recognition by Trh4-specific CTL. Downregulation of SPP clearly decreased the recognition by these CTL (Fig. 4C). Similar decrease of Trh4 peptide presentation was observed in TAP− MCA fibrosarcoma cells (Supplemental Fig. 2F, 2G). Silencing of the other SPP family members resulted in a marginal variation of CTL recognition (Fig. 4C). The presentation of the Trh4 peptide was not completely blocked by SPP silencing, most likely because of a substantial residual expression of the protease. The residual CTL recognition could furthermore be blocked by (z-LL)2-ketone, supporting this notion. Importantly, an independent other peptide-epitope presented by RMA-S was not affected by SPP silencing (Fig. 4D). These data and the fact that SPP is expressed in all mouse cells that naturally display Trh4/Db complexes and the ER localization of Trh4 substantiate the conclusion that the SPP enzyme is crucial for the processing of the C-terminal peptide-epitope of Trh4 in a TAP-independent way.
The transmembrane region directly upstream of the Trh4 peptide is crucial for Trh4 processing
To study the underlying mechanism of Trh4 peptide processing in detail and to determine critical regions for the proteolytic process, we performed extensive mutagenesis analysis of Trh4. First, some larger segments at the N and C termini of the protein were deleted. Removal of the first 32 aa did not diminish the processing efficiency of the peptide-epitope (Fig. 5A), whereas we hypothesized that this region functioned as a leader sequence routing the protein to the ER, as indeed was predicted by the online SignalP 4.1 Server (http://www.cbs.dtu.dk). Confocal microscopy analysis, however, confirmed that this construct was still localized in the ER (Supplemental Fig. 1C). We then removed segments located near the C terminus, starting with the sequence corresponding to aa 340–374, which is a cytosolic loop located between the two last transmembrane domains (see Fig. 2A). We did a progressive deletion in this construct comprising the upstream aa 315–330, which includes most of the second last transmembrane domain. However, none of these large deletions decreased Trh4 peptide processing (Fig. 5A). To exclude the possibility that these altered constructs delivered defective and thus proteasome-targeted proteins, we confirmed that the Trh4 peptide presentation from these mutant constructs was still (z-LL)2-ketone and DAPT dependent (Fig. 5B). Then 9 aa just upstream of the C-terminal peptide were removed. This resulted in a >60% impairment of CTL recognition (Fig. 5A), indicating that this intramembrane stretch, just upstream of the peptide-epitope, was critical for SPP cleavage. Furthermore, the C terminus was extended by cloning small V5-containing tags of 14 and 47 aa. Although these proteins were properly expressed, as tested by Western blot (Supplemental Fig. 3A), and were localized in the ER (Supplemental Fig. 1A), they did not result in any Trh4 peptide presentation (Fig. 5A), implying that the C-terminal location of the peptide in the Trh4 protein is crucial for its liberation. Failure to process the Trh4 peptide when it was not at the complete C terminus might be related to the absence of carboxyl-peptidases in the ER and suggests that the SPP cleavage products do not travel through the cytosol. Furthermore, we also tested whether the C-terminal region would function as a leader sequence for the Trh4 protein, but removal of the last 20 aa (aa 368–387) did not change its subcellular localization (Supplemental Fig. 1D). We therefore concluded that Trh4 peptide is directly liberated by SPP from the mature protein after which the peptide is available in the ER for loading into MHC-I molecules.
The intramembrane region directly upstream of the Trh4 peptide is crucial for processing. (A and B) Trh4 constructs with deletions or extensions were tested for liberation and MHC-I presentation of the C-terminal peptide-epitope (aa 379–387). (A) DNA plasmid constructs were transiently transfected into HeLa cells together with the mouse Db gene. Extent of IFN-γ release by T cells was quantified and scored as “+++” (wild-type construct, 100%), “++” (50–90%), “+” (10–50%) and “−” (<10%). (B) Processing from mutant constructs was still sensitive for the protease inhibitors (z-LL)2-ketone and DAPT. All constructs were tested at least four times. (C and D) Site-directed mutagenesis directly upstream of the C-terminal peptide-epitope of the Trh4 gene. (C) The amino acid substitutions and the position of the amino acids in the protein are indicated. Degree of T cell recognition was quantified based on area under the curve values of line graphs as shown in (D). Compiled means and SDs of at least three independent experiments are shown in Supplemental Fig. 3B.
The importance of the direct upstream amino acids of the Trh4 peptide suggested that the SPP cleavage site was located within this last membrane region (aa 374–387). Single amino acid substitutions were introduced using site-directed mutagenesis to determine the precise requirements for cleavage by SPP. Several publications demonstrated the strong preference of SPP for substrates with helix-destabilizing residues in their transmembrane domain (40–42). Amino acids such as asparagine, serine, and cysteine disturb a perfect α-helical conformation of the transmembrane domain and are therefore referred to as “helix-bending” or “helix-breaking” residues. Three of such amino acids are present in the Trh4 membrane region: two serines (aa 374 and 375) and one cysteine (aa 378; Fig. 5C). First, the two serines were replaced by isoleucines, which strongly contribute to a perfect α-helical structure. In contrast with our expectations, these mutations did not affect Trh4 presentation at all (Fig. 5C, 5D; SDM #2). Second, the cysteine residue (aa 378) in front of the epitope was substituted for a glycine (SDM #3), which still has some helix-breaking capacity, or an isoleucine (SDM #4), which removes the helix-breaking nature. These alterations had a very strong impact on peptide liberation, and the C378I construct nearly completely failed to produce the Trh4 peptide. Third, we combined the cysteine mutation C378G with the serine mutations and surprisingly found that introduction of two isoleucines improved the Trh4 processing (Fig. 5C, 5D; SDM #5), indicating that the more upstream serine residues were actually counterproductive for optimal SPP cleavage and support the importance of the cysteine residue directly upstream of the epitope. Fourth, we progressively substituted the complete stretch of amino acids in this region also using alanines, which are considered to promote helix destabilization (43). Collectively, these constructs displayed a clear gradual decrease in peptide presentation (Fig. 5C, Supplemental Fig. 3B). Together, these data point to the importance of the cysteine directly upfront of the epitope, most likely marking the actual SPP cleavage site, and suggest that nearby flanking residues indirectly influence this process via conformational change in the intramembrane α-helix. In line with this hypothesis is the finding that presentation of the Trh4 peptide does not depend on the N-terminal trimming peptidase ERAAP (Fig. 3D), suggesting that the minimal 9-mer epitope is directly generated by SPP cleavage without further need for C- or N-terminal trimming. The C-terminal liberation of this CTL peptide-epitope by the aspartyl protease SPP within the ER membrane exemplifies a novel way by which TAP-deficient cells are still able to present a diverse Ag repertoire.
Contribution of SPP protease family to MHC-I peptide repertoire of TAP− cells
We further examined the contribution of the SPP family of proteases to the overall MHC-I presentation of TAP-deficient cells. We analyzed the surface display of the two alleles of this mouse strain, Db and Kb, on RMA-S cells in which individual SPP members were silenced by shRNAs. SPP and SPPL2a silencing resulted in reduction of Kb molecules to 77 and 65% of control, respectively (Fig. 6). The presentation of Db molecules on these cells and MHC-I surface display on RMA-S cells in which SPPL2b and SPPL3 were silenced was not affected (Fig. 6). These data underline the involvement of intramembrane cleaving proteases in Ag processing, although precise underlying mechanisms remain to be elucidated.
Silencing of SPP and SPPL2a leads to decrease in surface MHC-I. MHC-I surface display of RMA-S cells expressing shRNA constructs for SPP, SPPL2a, SPPL2b, and SPPL3 was measured by flow cytometry (Kb and Db molecules separately). Percentages in the upper right corners reflect the residual MHC-I expression compared with control cells (“vector”), based on mean fluorescence intensity. Data are representative of three independent experiments.
Discussion
Studies with TAP-deficient cells showed that peptides can still gain access to MHC-I molecules, although with much lower efficiency. This peptide repertoire is different from that of processing intact cells, indicating that alternative processing pathways partly compensate for a defective conventional route based on proteasome and TAP (14, 18, 19). The processing mechanisms that lead to presentation of TAP-independent peptides are poorly characterized. In this study, we describe a new mechanism that leads to liberation of a C-terminal peptide-epitope from the ER membrane-spanning Trh4 protein. We found that the intramembrane-cleaving enzyme SPP was responsible for the cleavage event at the direct upstream region of the T cell epitope within the lipid bilayer. We speculate that SPP activity in the ER membrane is sufficient to liberate the minimal 9-mer peptide and release of this peptide into the ER lumen. Additional N-terminal trimming of the cleavage product is unlikely, because inhibitors of amino-peptidases did not prevent Trh4 peptide presentation and the cysteine just upfront of the 9-mer epitope appeared to be critical in the SPP cleavage process (see Figs. 3, 5C). Moreover, direct release of the liberated peptide into the ER lumen is very likely, because of the type II transmembrane orientation of the Trh4 protein tail (see Fig. 2A) and the fact that the epitope is located at the very C-terminal end of the protein. The incompetence of C-terminally extended versions of Trh4 to present the peptide (see Fig. 5A) corroborate the idea that the peptide is directly released in the ER lumen where it needs a free C terminus for loading into MHC-I molecules. The exact peptide loading mechanism of the Trh4 membrane peptide, however, remains to be determined.
The involvement of SPP in the C-terminal processing of a transmembrane protein initially surprised us. The activity of SPP was thus far largely related to the removal of small trunks of signal-sequence–derived transmembrane peptides from the ER membrane (34, 37, 42). The processing of leaders from nascent proteins requires a first cut by SP, after which the remaining transmembrane stretch is cut in two parts by SPP. However, the liberation of the Trh4 peptide did not require SP activity, because its inhibitor DCI did not prevent the presentation of the epitope. Furthermore, the Trh4-peptide is not part of a signal sequence, and removal of the N- or C-terminal domain did not route the Trh4 protein to a different subcellular localization (see Supplemental Fig. 1C, 1D). Nevertheless, the involvement of SPP with Trh4 cleavage fits with the strong preference for type II membrane segments and with the “helix-breaking” residues that are often found in SPP target proteins (25, 41, 42). Therefore, it seems that the role of SPP is more extended than the release of transmembrane leader sequences from the ER membrane. Interestingly, it has been suggested that SPP is physically associated with misfolded membrane proteins (44, 45), and as such might function as a chaperone to dispose membrane aggregates.
The Trh4-derived peptide is an example of TAP independently processed peptides. This particular SPP-mediated processing pathway is operational in TAP− and TAP+ cells (see Fig. 1), and also in tumor and healthy cells (28). Therefore, the TAP-independent pathway described in this article seems to be part of the “normal activity” of cells and does not represent a disease-related feature. Over the years, some other TAP-independent pathways have been characterized. They comprise peptides generated in the secretory and vesicular compartments by the protein convertase furin (22, 23). Second, leader sequence–derived peptides are frequently TAP independently processed (24). Peptides that arrive in the ER via this way overcome the need of TAP transport. Interestingly, TAP-independent processing was observed earlier using several C-terminally tagged constructs of ER-targeted proteins, feeding the speculation of a common “C-terminal” processing rule (46, 47). We hypothesize that SPP might be a central player for this category of TAP-independent Ag processing.
The family of intramembrane-cleaving aspartyl proteases (SPP family and presenilins) have important biological roles in signal transduction, cell differentiation, and immunity (25, 48). For instance, SPPL2a and SPPL2b were shown to be responsible for the release of the TNF-α intracellular domain, leading to expression of the proinflammatory cytokine IL-12 in dendritic cells (38), and also for the release of the intracellular domain of fasL in T cells (49). These proteases are mainly located in the vesicular pathway and at the cell surface. Recently, SPPL2a was reported to be involved in the proteolysis of the invariant chain (CD74), and thereby critical for the development and survival of B cells and DCs (50–52), maybe explaining the effect we found on MHC-I expression (Fig. 6). We now extend on the involvement of SPP family members in immunity by showing their role in alternative pathways of MHC-I Ag processing, thereby enabling T cell immune responses.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We acknowledge Dr. Pieter van der Pol, Nicole Schlagwein, and Joop Wiegant (Leiden University Medical Center) for help in immunofluorescent confocal microscopy. Dr. Isaac Donkor (University of Tennessee, Memphis, TN) kindly provided specific calpain inhibitors, and Dr. Ramon Arens (Leiden University Medical Center) is acknowledged for critical reading of the manuscript.
Footnotes
This work was supported by the Portuguese Foundation for Science and Technology (Ministério da Ciência, Tecnologia e Ensino Superior), Portugal (Grant SFRH/BD/33539/2008 to C.C.O.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- ER
- endoplasmic reticulum
- MHC-I
- MHC class I
- qPCR
- quantitative PCR
- shRNA
- short hairpin RNA
- SP
- signal peptidase
- SPP
- signal peptide peptidase
- TBSB
- TBS with blocking reagent from Boehringer Mannheim.
- Received June 10, 2013.
- Accepted August 14, 2013.
- Copyright © 2013 by The American Association of Immunologists, Inc.