To detect viral infections and tumors, CD8+ T lymphocytes monitor cells for the presence of antigenic peptides bound to MHC class I molecules. The majority of MHC class I-presented peptides are generated from the cleavage of cellular and viral proteins by the ubiquitin-proteasome pathway. Many of the oligopeptides produced by this process are too long to stably bind to MHC class I molecules and require further trimming for presentation. Leucine aminopeptidase (LAP) is an IFN-inducible cytosolic aminopeptidase that can trim precursor peptides to mature epitopes and has been thought to play an important role in Ag presentation. To examine the role of LAP in generating MHC class I peptides in vivo, we generated LAP-deficient mice and LAP-deficient cell lines. These mutant mice and cells are viable and grow normally. The trimming of peptides in LAP-deficient cells is not reduced under basal conditions or after stimulation with IFN. Similarly, there is no reduction in presentation of peptides from precursor or full-length Ag constructs or in the overall supply of peptides from cellular proteins to MHC class I molecules even after stimulation with IFN. After viral infection, LAP-deficient mice generate normal CTL responses to seven epitopes from three different viruses. These data demonstrate that LAP is not an essential enzyme for generating most MHC class I-presented peptides and reveal redundancy in the function of cellular aminopeptidases.
Ctytotoxic T lymphocytes recognize virus-infected or otherwise abnormal cells through TCR interactions with MHC class I molecules (MHC class I). MHC class I is expressed on most nucleated cells, and consists of a genetically polymorphic H chain, an L chain (β2-microglobulin), and a small peptide, usually 8–10 aa long. This peptide provides specificity for recognition by the TCR and is produced by proteolysis of cellular proteins, predominately by the proteasome (1, 2).
Purified proteasomes, when incubated with full-length proteins such as chicken OVA, casein, or insulin-like growth factor, generate many peptides, ranging in length from three to 22 aa (3, 4, 5). Only ∼15% of the peptides produced by digestion of these proteins are of optimal binding length for MHC class I ligands, but an additional 15–25% of the peptides are longer than 10 residues and could therefore be further processed to produce MHC class I-binding peptides (3, 4, 6). In vivo, proteasomes also generate longer peptides that are trimmed and presented (7, 8, 9, 10). In fact, it was recently suggested that most peptides produced by proteasomes in vivo are longer than 15 residues (11). In any case, the majority of peptides generated by the proteasome are destroyed by cytosolic peptidases before they encounter the TAP transporter (11, 12).
Treating cells with proteasome inhibitors almost completely prevents presentation of peptides not only from full-length protein precursors, but also from peptide precursors that are extended even by one amino acid at the C terminus (13, 14). Therefore, the proteasome appears to be required for generation of the C-terminal amino acid of most MHC class I-binding peptides. However, peptides derived from precursors that are extended by 25 aa at the N terminus can be efficiently presented on MHC class I even in the presence of proteasome inhibitors (14), but only if the N terminus is unblocked (13), implicating aminopeptidases in this form of processing.
Some precursor peptides are trimmed in the endoplasmic reticulum (ER)3 by the aminopeptidase ERAP1 after transport to the ER by TAP (7, 9, 10). In cells lacking ERAP1, N-extended precursors that are transported into the ER fail to be trimmed and presented (10). However, in the absence of ERAP1, there is still trimming and presentation of a significant fraction of precursor peptides by aminopeptidases in the cytosol (10). There are several aminopeptidases in the cytosol that can potentially trim peptides.
One of the first aminopeptidases to be implicated in Ag processing was leucine aminopeptidase (LAP), which was identified as a major activity in cytosolic extracts from HeLa cells that was capable of trimming N-extended precursors of SIINFEKL, an antigenic peptide from OVA, to the mature presented epitope (15). Another indication that this enzyme might play an important role in MHC class I peptide processing is that transcription of the LAP gene is strongly up-regulated by IFN-γ, a cytokine that, among its many proinflammatory effects, up-regulates many of the essential components of Ag presentation (15). Moreover, overexpressing LAP in cells led to more rapid trimming of peptides (11, 16). For these reasons it was thought that LAP was an important peptidase in generating MHC class I-presented peptides.
In addition to LAP, mammalian cells contain many other peptidases in the cytosol, and several of these have been proposed to play a role in Ag presentation in certain situations (1). Puromycin-sensitive aminopeptidase (PSA) and bleomycin hydrolase (BH) are cytosolic aminopeptidases that were shown to trim a vesicular stomatic virus (VSV) precursor peptide in cell extracts (17). Treatment of cells with inhibitors that block PSA and BH activity, among other things, reduced the presentation of the VSV epitope (17). Tripeptidyl peptidase II has exo- and endopeptidase activities and may play a unique role in the initial trimming of very long peptides (>16 residues) in the cytosol (8). A major unresolved question is whether the various aminopeptidases have unique roles in protein catabolism and Ag presentation or whether there is significant functional redundancy.
In addition to generating mature MHC class I-binding peptides from precursors, peptidases have the potential to destroy epitopes by trimming them to a size that is too short to bind MHC class I. ERAP1 and the endopeptidase thimet oligopeptidase (TOP) have been shown to destroy peptides in cells and limit Ag presentation (9, 10, 18). In addition, PSA (19) and LAP (11, 20) have been shown to have the potential to destroy antigenic peptides, although the extent to which they limit Ag presentation in vivo, if any, is not known.
In this study we demonstrate that mice lacking LAP and human cells in which LAP levels are reduced with small interfering RNA (siRNA) do not exhibit any detectable differences in Ag presentation. These results demonstrate that LAP does not play an essential role in the generation or destruction of many antigenic peptides and indicate that there is functional redundancy among the cytosolic aminopeptidases.
Materials and Methods
Plasmids, primers, and PCR
The plasmids used to make the LAP-KD and control cell lines were generated by designing short-hairpin RNA (shRNA) inserts using RNAiOligoRetriever software and were based on the National Center for Biotechnology Information sequence for human LAP, NM_015907. Synthetic oligos (IDT) were inserted into BseRI and BamHI digested pSHAG-1 plasmid (21) (gift of Dr. G. Hannon, CSH Laboratories, Cold Spring Harbor, NY). The oligos used were: for LAP, GAATGATCCCATTGCCTGTTCCTCAATCgaagcttgGGTTGAGGAATAGGCAGTGGGATCATTCTTCtttttt and gatcaaaaaaGAAGAATGATCCCACTGCCTATTCCTCAACCcaagcttcGATTGAGGAACAGGCAATGGGATCATTCcg; and for the control cell line, TTCTGACACCAAGAAGGAGATGACACAGgaagcttgCTGTGTCGTCTTCTTCTTGGTGTCAGGATAGtttttt, and gatcaaaaaaCTATCCTGACACCAAGAAGAAGACGACACAGcaagcttcCTGTGTCATCTCCTTCTTGGTGTCAGAAcg. Kanamycin-resistant bacterial clones were selected, and the plasmids were sequenced. pSHAG+shRNA plasmids were digested with Pvu NruI and EcoRV, removing the CMV promoter.
Primers used for identification of shRNA knockdown (KD) clones and for real-time PCR were 5′-GCACGCCAATTGATGGAG-3′ and 5′-GGTCTGATATGGACCTCG-3′. β-Actin was used as a control for normalization in real-time PCR assays. The sequences of the primers used were 5′-CGAGGCCCAGAGCAAGAGAG-3′ and 5′-CGGTTGGCCTTAGGGTTCAG-3′ (22).
To express mature SIINFEKL with no N-terminal residues, we constructed pUG-SIINFEKL. This plasmid consists of ubiquitin with SIINFEKL fused to the C terminus; C-terminal ubiquitin hydrolases efficiently release peptides thus fused to ubiquitin (23). An internal ribosome entry site downstream of the ubiquitin-SIINFEKL fusion was followed by GFP.
mRNA was isolated from 106
To generate the upstream and downstream homologous sequences for ligation to the 5′ and 3′ ends of the knockout reporter gene, two sets of primers were used. For the upstream sequence, the upstream forward primer was 5′-AGACCCTAGAAAGGACGACGG-3, and the upstream reverse primer was 5′-GGCCCTGTGACTGGCTACTC-3′. For the downstream sequence, the downstream forward primer was 5′-TGGTGCCATCTTTCTCAGGAC-3′, and the downstream reverse primer was 5′-GTGGTCACCTTGGTCTGCAAG-3′. For screening of embryonic stem (ES) cells that contained a knockout allele, TaqMan primers were used. The forward sequence was 5′-AGGATTGTCCCAAAGCCTGCTACGCT3′, and the reverse sequence was 5′-TGGTGTTCAGTGATGGAGGTCTAGCATGCA-3′. Each primer sequence spanned the point of recombination between the reporter gene and the endogenous genomic sequence.
A three-primer PCR protocol was used to screen for LAP-deficient mice. LAP-WT(R1) 5′-CAGATATGGCTGATTCTAGC-3′ lies downstream of the knockout (KO) insert in the genomic sequence and therefore amplifies both the wild-type (WT) and KO alleles. LAP-KO(F4) 5′-GCCTGAAGAACGAGATCAGC-3′ lies within the KO allele and amplifies only the KO allele. LAP-WT(F1) 5′-GCACACTTAGACATAGCAG-3′ lies within the WT allele and amplifies only the WT allele.
Generation of LAP−/− mice
The mouse LAP gene (gene identification no. 66988) was deleted using VelociGene technology (24). Briefly, a large targeting vector (BACvec) was constructed by bacterial homologous recombination in which the 19.2-kb LAP gene was replaced by a lacZ-neo cassette. An 129 × C57BL/6 F1 ES cell line was electroporated with the BACvec and selected for G418 resistance. Drug-resistant clones were screened for loss of one copy of the LAP gene by quantitative PCR using probes at either end of the deletion. Two independent targeted ES cell clones were microinjected into C57BL/6 blastocysts, which were implanted in C57BL/6 females to generate chimeras. Chimeras were bred back to BL/6 to generate F1 heterozygote mice. Both ES cell clones produced healthy-appearing knockout mice in proper Mendelian proportions. A line derived from one of these clones was used for subsequent experiments.
Poly I:C (poly I:C) treatment of mice
Virus infection of mice
Mice were injected i.p. with 5 × 104 PFU/mouse of lymphocytic choriomeningitis virus (LCMV) Armstrong (a gift from Dr. R. Welsh, University of Massachusetts Medical School, Worcester, MA) or with 5 × 106 PFU/mouse of recombinant vaccinia (Vac; provided by Drs. J. Yewdell and J. Bennink, National Institutes of Health, Bethesda, MD), containing chicken OVA (25). Mice were infected i.v. with 5 × 106 PFU/mouse of VSV (a gift from Dr. R. Welsh). Nine days (LCMV) or 7 days (Vac-OVA and VSV) after infection, splenocytes were harvested and incubated for 5 h with the appropriate peptide (5 μM for LCMV and Vac-OVA; 2 μM for VSV) or with anti-CD3ε (BD Biosciences) in the presence of GolgiPlug (BD Biosciences) and rIL-2 (BD Biosciences). Cells were then stained for CD8, CD44, and intracellular levels of IFN-γ using commercial Abs (BD Biosciences) and were analyzed by flow cytometry.
Construction of stable shRNA cells
HeLa-Kb cells (10
Cells and tissue culture
The shRNA KD cell lines were incubated at 37°C and 10% CO2b. Mouse embryonic fibroblasts (MEFs) were generated from 12- to 14-day embryos and cultured at 37°C in 10% CO2 with DMEM/20% FCS. Cells were cultured in flasks, and transfections were performed in six-well plates (Corning-Costar). During transfection periods, cells were cultured in DMEM/10%FCS.
For incubations in IFN-γ, MEFs or shRNA KD cells were transfected first with the indicated construct. Then 8 h later, the transfection medium was removed, and DMEM, FCS, and IFN-γ were added to each well. The IFN-γ concentrations used were 250 U/ml recombinant human IFN-γ for HeLa cells (Biogen) or 50 U/ml recombinant murine IFN-γ for MEFs (BD Biosciences). The cells were incubated in IFN-γ until they were analyzed.
Abs, immunoblotting, and flow cytometry
The mAb 25.D1.16 (anti-Kb+SIINFEKL) (26), AF6-88.5 (anti-Kb) (27), Y3 (anti-Kb) (28), M1/42 (anti-H2) (29), or H36.4.5 (anti-influenza hemagglutinin; gift from W. Gerhard, The Wistar Institute, University of Pennsylvania, Philadelphia, PA) were used as primary Abs in staining HeLa cells and MEFs for flow cytometry. After incubation in one of the primary Abs, the cells were washed with PBS and stained with donkey anti-mouse (or donkey anti-rat) F(ab′)2 conjugated to Cy5 (Jackson ImmunoResearch Laboratories). For staining cells isolated from spleen, AF6-88.5 (H-2Kb) and KH95 (H-2Db) Abs conjugated to a fluorophore were used according to the manufacturer’s directions (BD Biosciences). The cells were then analyzed by flow cytometry on a FACSCalibur apparatus (BD Biosciences) with FlowJo software (TreeStar).
For immunoblotting HeLa cells, cells were lysed in a 1-ml Dounce homogenizer (Kontes) on ice in Dounce buffer (10 mM Tris-HCl and 0.5 mM MgCl2, pH 7.6). Complete EDTA-free protease inhibitor mixture minitablets (Roche) were added according to the manufacturer’s instructions. After 50 strokes in the Dounce homogenizer, tonicity buffer was added (10 mM Tris-HCl, 0.5 mM MgCl2, 0.6 M NaCl, and complete mini-tablets) to the lysate to stabilize the nuclei. After a 5-min spin in a minicentrifuge at 4°C at full speed, the resulting supernatant was centrifuged at 4°C in an ultracentrifuge (Beckman Coulter) with a 70.1Ti rotor (Beckman Coulter) at 35,000 rpm for 1 h. 3× SDS sample buffer and DTT (New England Biolabs) was added to the resulting supernatant, and the samples were heated to 95°C for 5 min.
For immunoblotting MEFs, cells were lysed by adding Nonidet P-40 lysis buffer (1% Nonidet P-40, 300 mM NaCl, and 50 mM NaH2PO4), with complete EDTA-free protease inhibitor mixture minitablets added (Roche) according to the manufacturer’s instructions. After a 5-min incubation on ice, they were spun in a minicentrifuge at 4°C at full speed for 5 min. 3× SDS/DTT buffer was added to the resulting supernatant, and the samples were heated to 95°C for 5 min.
HeLa cell equivalents (1.5 × 105 or 3 × 105 for MEFs) were run on a 12% SDS gel, followed by protein transfer to a nitrocellulose membrane (Schleicher & Schuell). After transfer, the membrane was rotated overnight in PBS, 5% milk, and 0.2% Tween 20 to block. After 18 h, the membrane was stained with rabbit anti-LAP polyclonal Ab (gift from Dr. A. Goldberg, Harvard Medical School, Boston, MA) diluted 1/10,000 in PBS, 0.2% Tween 20, and 0.02% NaN3. After 2 h, the blot was washed three times with PBS and 0.2% Tween 20 for 30 min. The blot was then stained with HRP-conjugated goat anti-rabbit diluted 1/50,000. The blot was developed with ECL (Pierce).
Peptide injection and measurement
Peptides were injected into cells, and their half-lives were measured as previously described (11). The synthesis and sequences of the internally quenched peptides have been previously described in detail (8). Analysis of peptide degradation rates was performed as previously described (11).
Small interfering RNA transfection of shRNA-stable cells
LAP-KD or shRNA control cells were transfected with siRNA specific for ERAP1 or with a control siRNA (directed against murine TOP in a region that differs from human TOP) as previously described (10).
Generation of LAP-deficient mice
LAP-deficient mice were generated using VelociGene (Regeneron Pharmaceuticals) technology (24). Homologous recombination resulted in the loss of all exons and introns of the LAP genomic sequence (Fig. 1⇓), beginning 6 bp downstream of the LAP start codon and ending 624 bp after the LAP stop codon. In total, ∼19.2 kb was deleted. The presence of the neogene and subsequent loss of the entire LAP gene were confirmed by PCR (Fig. 1⇓B). LAP−/− mice expressed no LAP protein in any of the organs tested, even after IFN stimulation (Fig. 1⇓C).
Lymphocyte ratios and MHC class I levels in LAP−/− mice are normal
Because MHC class I surface expression is dependent on peptide supply, changes in peptide supply can be detected by measuring surface MHC class I levels. Elimination of LAP could, in principle, reduce the peptide supply to MHC class I (if LAP predominately trims peptides longer than nine to 10 aa) or increase peptide supply (if LAP preferentially trims peptides nine to10 aa in length that could otherwise bind to MHC class I). We therefore evaluated the expression of MHC class I on cells in the LAP-deficient animals. The levels of H-2Db and H-2Kb from LAP−/− mice were similar to those on splenocytes from WT C57BL/6 mice (Fig. 2⇓). The small reduction in Kb expression on LAP−/− cells in this experiment was not statistically significant and was not observed in other experiments. LAP is normally induced by IFN (15, 30), and in WT mice, LAP protein levels increased after i.p. injection of the type I IFN-inducer, poly I:C. However, even under these conditions, surface levels of H-2Db and H-2Kb were not significantly different in WT and LAP−/− mice (Fig. 2⇓).
Mice lacking important components of the MHC class I Ag presentation pathway sometimes have reduced levels of CD8+ T lymphocytes due to reduced positive selection in the thymus (31, 33). Therefore, we examined the levels of CD8+ T cells in LAP-deficient mice. Parenchymal, mesenteric, and maxillary lymph nodes and spleens were harvested from C57BL/6 and LAP−/− mice, and their CD4+ and CD8+ T cells were enumerated by flow cytometry. CD4+B220− cells and CD8+B220− cells were present in similar ratios and numbers in both strains, suggesting that T cell generation and maturation are grossly normal in LAP−/− mice (data not shown).
LAP−/− MEFs present peptide as well as WT MEFs
Several independent lines of MEFs were generated from WT and LAP−/− embryos. As with lymphocytes, no significant differences in surface H-2Db, H-2Kb, or total MHC class I levels were detected either under normal conditions or after stimulation with IFN-γ (Fig. 3⇓). Using real-time PCR, we detected no increase in BH, PSA, or ERAP1 expression (data not shown). Therefore, these other aminopeptidases were not up-regulated and compensating for a lack of LAP.
Cleavage of full-length OVA by the proteasome in vitro generates SIINFEKL as well as SIINFEKL extended at the N terminus by up to 12 residues (3). Similarly, N-extended precursors of SIINFEKL are generated from full-length OVA constructs and require trimming for presentation in vivo (10). To examine the role of LAP in this process, we transfected WT or LAP−/− MEFs with plasmids expressing full-length OVA and GFP, so that transfected cells could be identified by GFP expression. WT and LAP−/− MEFs (gated for comparable GFP expression) generated similar amounts of SIINFEKL-Kb complexes (Fig. 4⇓A), as detected by flow cytometry using the H-2Kb-SIINFEKL-specific Ab 25.D1.16 (26).
To more thoroughly examine the dependence of the processing of N-extended peptides on aminopeptidases, we expressed in these MEFs N-extended forms of SIINFEKL from minigenes. In this situation, all peptides are produced in the cytosol as precursors that require trimming for presentation (13, 14). SIINFEKL generation from precursors extended by one (the initiating methionine) or five residues was identical in WT and LAP−/− MEFs (Fig. 4⇑A). Because these 9- to 13-residue antigenic precursors can be transported into the ER and trimmed by ERAP1 (10), we also examined the presentation of a 33-residue precursor that is too long to be efficiently transported into the ER by TAP (34). This construct, which has a 25-residue N-terminal extension, was trimmed and presented similarly by both WT and LAP−/− MEFs (Fig. 4⇑A). Therefore, under constitutive conditions, LAP is not essential for the trimming of N-extended SIINFEKL precursors.
Aminopeptidases can also destroy antigenic peptides by trimming them below the size required for stable binding to MHC class I molecules. To test whether LAP might be destroying some presentable peptides, MEFs were transfected with a construct encoding a ubiquitin-SIINFEKL fusion protein. When expressed in cells, the N-terminal ubiquitin is cleaved, thereby producing the mature epitope SIINFEKL. If LAP were to trim SIINFEKL, the resulting products would not bind stably to H-2Kb. However, LAP−/− and WT cells equivalently presented SIINFEKL generated from the ubiquitin fusion construct (Fig. 4⇑A). Therefore, LAP does not limit the amount of SIINFEKL available for presentation.
IFN treatment of cells alters the composition and enzymatic activities of the proteasome, resulting in relatively more N-extended precursors of antigenic epitopes (3) as well as increased levels of LAP (15). However, even though IFN treatment increased overall presentation, the presentation of SIINFEKL from each precursor was the same in WT and LAP−/− MEFs (Fig. 4⇑B), suggesting that even at induced levels, LAP is not essential in the generation of some MHC class I peptides.
Presentation of viral epitopes is not altered in LAP−/− mice
To test Ag processing in vivo, we examined the CTL responses to several viruses. WT and LAP−/− mice were infected with VSV, LCMV, or a recombinant Vac-OVA. At the peak of each infection, splenic lymphocytes were isolated, stimulated in vitro with the appropriate antigenic peptide, and stained for intracellular IFN-γ levels. The frequency of CTL (Fig. 5⇓) and their level of production of IFN-γ (not shown) compared with the VSV peptide (Fig. 5⇓A), to four antigenic LCMV peptides (Fig. 5⇓B), and to two recombinant Vac-OVA peptides (Fig. 5⇓C) in LAP−/− mice were equal to those of WT mice, suggesting that neither the quantity nor the quality of the CTL response in vivo was dependent on LAP.
LAP mRNA expression and protein levels are reduced in shRNA stable transfectants of HeLa cells
The finding that elimination of LAP does not alter Ag presentation in mouse cells in vivo or in vitro was unexpected because LAP has been shown to process N-extended SIINFEKL peptides in cell lysates (15). Because the previous studies used HeLa cell extracts, it seemed possible that LAP might play a more important role in human cells or in particular cell types. To investigate this issue, HeLa-Kb cells (10) were stably transfected with a plasmid encoding a shRNA targeting LAP or encoding a control shRNA that does not recognize human sequences. Clone LAP-KD showed a reduction in LAP of 90–95% by real-time PCR (Fig. 6⇓A) and Western blot (Fig. 6⇓B) compared with the control cell line. In contrast, real-time PCR showed no significant change over time in the expression of BH or PSA between these cell lines (data not shown).
LAP is inducible by IFN-γ in HeLa cells (15). Therefore, we examined the silencing of LAP in LAP-KD cells treated with IFN-γ. After 48 h of IFN-γ treatment, LAP mRNA was increased ∼10-fold in the control cell line (Fig. 6⇑A), but LAP mRNA levels in the LAP-KD cell line, although increased, were still 90–95% lower than in the control (IFN-stimulated) cell line. Analysis of protein levels by semiquantitative immunoblotting gave similar results (Fig. 6⇑B). LAP-KD constitutively had 5–10% as much LAP as the control cell line. Although treatment with IFN-γ increased the levels of LAP protein ∼10-fold in both the LAP-KD and control cell lines, LAP levels in LAP-KD were ∼10-fold lower than in control cells treated with IFN-γ.
Reduction of LAP has no effect on surface presentation of H-2Kb-SIINFEKL in HeLa cells
Because LAP was originally described as an enzyme in HeLa cell extracts that trimmed N-terminally extended precursors of the OVA peptide SIINFEKL, we tested such peptides in LAP-KD cells. Therefore, LAP-KD cells and control cells were transiently transfected with plasmids encoding OVA or N-extended SIINFEKL precursors, as described above. For each of the constructs, the generation of H-2Kb-SIINFEKL was the same in control and LAP-KD cells (Fig. 7⇓A). There was also no difference between the control and LAP-KD cells when they were treated with IFN-γ (Fig. 7⇓B). Therefore, LAP is not required for generation of MHC class I peptides in HeLa cells.
N-terminally extended peptides may be transported to the ER by TAP, where they can be trimmed to mature epitopes by ERAP1. To determine whether ERAP1 was masking an effect of LAP deficiency, we examined the phenotype of cells in which both LAP and ERAP1 were silenced. LAP-KD and control cells were treated with an siRNA targeting ERAP1 or with control siRNA (10) and then transiently transfected with plasmids encoding N-terminally extended SIINFEKL precursors, as described above. As previously reported (10), presentation of SIINFEKL from N-extended precursors was reduced in HeLa cells treated with ERAP1 siRNA (Fig. 8⇓). In comparison, SIINFEKL presentation was not further reduced in LAP and ERAP1 double-KD cells. Presentation of the SIINFEKL precursor was also not different in ERAP1-deficient vs LAP- and ERAP1-deficient cells after stimulation with IFN-γ for 24 h after transfection (Fig. 8⇓).
Reduction of LAP has no effect on the rate of peptide trimming in cells
The quantitation of Ag presentation is an indirect measure of the generation of antigenic peptides. It is possible that a contribution of LAP to this process could be missed if the trimming of peptides was not a rate-limiting step in the pathway. We therefore directly measured the rate of peptide degradation in control and LAP-KD cells by microinjection of peptides containing fluorescein and quencher adducts (8, 11). These substrates generate a fluorescent signal when aminopeptidases cleave and thereby separate the residues containing the fluorophore and quencher moieties. Peptides with different amino acids in the P1 position were microinjected into cells, and fluorescence resulting from trimming the peptides was measured.
Although there was some variation in the rate of trimming of different peptide sequences, there was no statistically significant difference in the half-life of any of the tested peptides between the two cell lines (Fig. 9⇓A). Treating the cells with IFN-γ before microinjection increased LAP expression (Fig. 9⇓B), but did not alter the rate of processing of the peptides (Fig. 9⇓, A and C). These data demonstrate that LAP activity does not influence the rate of peptide trimming in the cytosol.
In this study, we show 1) that LAP is not essential for viability or normal development of mice, and 2) that LAP does not play an indispensable role in generating peptides presented by MHC class I under constitutive conditions or after stimulation with IFN. These results indicate that there is considerable functional redundancy among aminopeptidases within cells.
There is considerable evidence that aminopeptidases are important in trimming antigenic precursors for presentation. Proteasomes have been shown in vitro (3, 35) and in vivo (10) to generate many N-extended precursors. Other experiments have shown that if such N-extended peptide precursors are expressed or injected into cells, they are trimmed and presented by MHC class I on the cell surface (11, 14). Processing of such peptides can occur in the presence of proteasome inhibitors (14, 36, 37), indicating that other peptidases in the cell can cleave N-terminal residues. Aminopeptidases can trim N-extended precursors that have a free N terminus, but not if the N terminus is blocked (11, 13). Finally, eliminating or blocking aminopeptidases in vivo reduces Ag presentation (10, 17, 38). Together, these data suggest that N-extended precursors are generated during protein degradation and then trimmed by aminopeptidases in vivo.
N-extended peptides that are preceded by an ER-localizing signal sequence are efficiently trimmed to mature epitopes in TAP-deficient cells, demonstrating that aminopeptidases localized in the ER can generate peptides for presentation (14, 39, 40). In cells lacking ERAP1, N-extended peptides targeted to the ER are not presented (9, 10) indicating that ERAP1 plays an essential, nonredundant role in trimming in the ER. However, several lines of evidence indicate that N-extended peptides can also be trimmed in the cytosol before their transport into the ER. First, cytosolic N-extended SIINFEKL precursors expressed in ERAP1-deficient cells are trimmed and presented (albeit in reduced amounts), whereas when the same constructs are targeted to the ER, SIINFEKL is not presented (10). Second, peptides preceded by N-extensions of up to 25 aa are too long to be efficiently transported by TAP, but when expressed in the cytosol, they are still trimmed and presented by MHC class I in the absence of proteasome activity (14). Third, cytosolic extracts devoid of ER contaminants can trim N-extended peptides (15, 17). Fourth, treatment of cells with inhibitors of cytosolic aminopeptidases reduces the presentation of some peptides (8, 17). A major unresolved issue in the field is the identity and extent of contribution of the cytosolic aminopeptidases that participate in the generation of MHC class I-presented peptides.
Tripeptidyl peptidase II is a cytosolic tripeptidyl peptidase that has been shown to play an important role in trimming peptides that are longer than ∼16 residues (8). However, it remains unclear which cytosolic aminopeptidase(s) trims shorter peptides of 15 aa or less. It has also been unclear how much redundancy exists among the cytosolic aminopeptidases.
LAP is a cytosolic aminopeptidase that was thought to be a major contributor to the MHC class I peptide pool for a number of reasons. First, it was shown to be a major peptidase in cytosolic extracts that generated SIINFEKL from N-extended precursors. Second, transcription of LAP is inducible by both type I and type II IFNs. Because IFNs enhance MHC class I presentation and up-regulate most components of this pathway, the IFN induction of LAP suggested that it might have a particularly important role during inflammation. Third, overexpression of LAP results in a shorter half-life of peptides in living cells (11). Given these findings, our present results are very surprising.
We have found no detectable defect in the ability of LAP-deficient mice to generate or present MHC class I peptides, even after treatment with the type I IFN inducer poly I:C. In these experiments we have measured the presentation of a number of peptides to CTLs, including the antigenic model epitope SIINFEKL, which is a well-characterized substrate of LAP. No qualitative or quantitative difference was detected in the priming of CTL responses to any of these epitopes in LAP-deficient and WT mice. In addition, we have shown in MEFs and HeLa cells that LAP was not required for the trimming of peptide precursors from various N-extended SIINFEKL constructs, even in the presence of IFN-γ.
One possibility is that in the absence of LAP, peptides that would normally be trimmed into the cytosol are transported into the ER (either in their original form or as an intermediate precursor that has been partially trimmed by other cytosolic aminopeptidases). ERAP in the ER can rapidly process N-extended SIINFEKL precursors to SIINFEKL (10), and it was possible that even if LAP deficiency led to an increased amount of N-extended precursors entering the ER, that ERAP1 activity could mask this effect. However, although reducing ERAP1 dramatically reduced the presentation of SIINFEKL generated from these peptides, there was no further reduction seen when LAP was also knocked down (Fig. 8⇑). Therefore, ERAP1 is not masking a contribution of LAP to trimming precursor peptides.
Although we examined a number of different peptides in this study, it is still possible that a subset of peptides exists for which LAP is required for presentation. LAP has been shown to hydrolyze different amino acid residues (from amino acid-7-amino-4-methylcoumarin substrates) at different rates, so LAP may potentially play a more dominant role in trimming certain sequences. However, when peptides with different N-terminal residues were injected into LAP-deficient HeLa cells, no difference in peptide half-life was detected even after stimulation with IFN-γ. Moreover, LAP-deficient mouse cells express normal levels of MHC class I, indicating that the overall supply of peptides to MHC class I is not changed. Therefore, if there are LAP-dependent epitopes, they must constitute a minority of peptides.
Together, these findings demonstrate that LAP does not play an essential role in producing the majority of peptides for MHC class I presentation. The fact that SIINFEKL was presented from N-extended precursors, even in LAP and ERAP1 double-deficient cells indicates that there are other cytosolic aminopeptidases that can substitute for LAP. Although it is possible that other aminopeptidases could be up-regulated to compensate for the absence of LAP, we have not found any change in BH, PSA, and ERAP1 expression that is correlated with the loss of LAP, in MEFs or of BH or PSA in LAP-KD cells (data not shown). In any case, our data indicate that there must be considerable redundancy in the trimming function of cytosolic aminopeptidases in vivo.
It is possible that the contribution of LAP to peptide supply is not apparent because the trimming step is not rate limiting. This would explain why peptides with one, two, or three extra residues on the N terminus are presented with similar kinetics to one another and to the mature epitope itself, at least in some systems (14) (data not shown). Under these conditions, other aminopeptidases would be sufficient to trim antigenic precursors to mature epitopes.
Surprisingly, we also found no evidence that LAP destroys mature epitopes, thereby limiting their presentation. There was no increase in peptide supply in the absence of LAP. It is possible that this is because LAP destroys as many peptides as it produces; however, our data argue against such a possibility. First, LAP-deficient cells show no defect in generating and presenting the mature SIINFEKL epitope from the ubiquitin-SIINFEKL fusion construct, even though trimming of just one residue would generate IINFEKL, which is too short to bind H-2Kb (Fig. 4⇑). Second, the hydrolysis rates of individually injected peptides are not different in LAP-KD cells. For these reasons, we believe that LAP does not play a major role in the destruction of antigenic peptides under physiological conditions. However, it should be noted that when LAP is expressed at supraphysiological levels, peptides are hydrolyzed more rapidly in cells (11).
The redundancy of aminopeptidases is likely to be useful to the cell in some way. Redundancy would ensure optimal recycling of peptides into amino acids, which would conserve energy for the cell. It would also prevent the build-up of proteasomal products, which could interfere with protein-protein interactions or be toxic in other ways. Perhaps it is for these reasons that LAP-deficient mice are viable and demonstrate a normal phenotype.
However the question remains: if LAP is not essential for Ag presentation then why is it inducible by IFN? It is known that IFN stimulation up-regulates many components of the Ag-processing pathway, including the immunoproteasome. Immunoproteasomes can generate distinct MHC class I-presented peptides that are not produced by the constitutive proteasome (41, 42, 43, 44) In addition, the immunoproteasome has been shown to generate longer peptides than the constitutive proteasome for at least one Ag (3), and there are indirect data suggesting that this might be generally true (10). It is possible that in some cell types the immunoproteasome generates antigenic precursors that are more efficiently processed by IFN-inducible aminopeptidases such as LAP. Alternatively, it is possible that LAP is more essential in a pathway not involved in Ag presentation, such as the breakdown of some other component involved in an immune response, or in the breakdown of viral peptides during infection. The LAP-deficient mice and cells should be useful in future studies to help answer these questions. In addition, further investigations into the roles of other cytosolic aminopeptidases will help to define the redundancy and/or unique roles of these various peptidases compared with those of LAP.
We thank Dr. Raymond Welsh for providing viruses and peptides.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by a National Institutes of Health grant (to K.L.R.) and National Institutes of Health Training Grant AI07349 (to C.F.T.). Core resources supported by Diabetes Endocrinology Research Grant DK42520 were also used.
↵2 Address correspondence and reprint requests to Dr. Kenneth L. Rock, Department of Pathology, University of Massachusetts Medical Center, Room S2-109, 55 Lake Avenue North, Worcester, MA 01655-0125. E-mail address:
↵3 Abbreviations used in this paper: ER, endoplasmic reticulum; BH, bleomycin hydrolase; ERAP, ER aminopeptidase; ES, embryonic stem; KD, knockdown; KO, knockout; LAP, leucine aminopeptidase; LCMV, lymphocytic choriomeningitis virus; MEF, mouse embryonic fibroblasts; poly I:C, polyinositic-polycytidylic acid; PSA, puromycin-sensitive aminopeptidase; shRNA, short-hairpin RNA; siRNA, small interfering RNA; TOP, thimet oligopeptidase; Vac, vaccinia; VSV, vesicular stomatic virus; WT, wild type.
- Received May 23, 2005.
- Accepted September 8, 2005.
- Copyright © 2005 by The American Association of Immunologists