Nature Immunology
3, 1169 - 1176 (2002)
Published online: 18 November 2002; | doi:10.1038/ni859
An IFN- −induced aminopeptidase in the ER, ERAP1, trims precursors to MHC class I−presented peptidesTomo Saric1, 4, Shih-Chung Chang1, Akira Hattori2, Ian A. York3, Shirley Markant1, Kenneth L. Rock3, Masafumi Tsujimoto2
& Alfred L. Goldberg11 Department of Cell Biology, Harvard Medical School, Boston, MA 02115, USA. 2 Laboratory of Cellular Biochemistry, RIKEN, 2-1 Hirosawa, Wako, Saitama, 351-0198 Japan. 3 Department of Pathology, University of Massachusetts Medical School, Worcester, MA 06155, USA. 4 Present address: ATABIS GmbH, Joseph-Stelzmann Str. 50, 50931 Cologne, Germany.
Correspondence should be addressed to Alfred L. Goldberg alfred_goldberg@hms.harvard.eduPrecursors to major histocompatibility complex (MHC) class I−presented peptides with extra NH2-terminal residues can be efficiently trimmed to mature epitopes in the endoplasmic reticulum (ER). Here, we purified from liver microsomes a lumenal, soluble aminopeptidase that removes NH2-terminal residues from many antigenic precursors. It was identified as a metallopeptidase named "adipocyte-derived leucine" or "puromycin-insensitive leucine-specific" aminopeptidase. However, because we localized it to the ER, we propose it be renamed ER−aminopeptidase 1 (ERAP1). ERAP1 is inhibited by agents that block precursor trimming in ER vesicles and although it trimmed NH2-extended precursors, it spared presented peptides of 8 amino acid and less. Like other proteins involved in antigen presentation, ERAP1 is induced by interferon- . When overexpressed in vivo, we found that ERAP1 stimulates the processing and presentation of an antigenic precursor in the ER.The ability of the immune system to recognize and eliminate virally infected and cancer cells depends upon the cells' capacity to present on its surface major histocompatibility complex (MHC) class I molecules peptide fragments derived from intracellular proteins. Most of these antigenic peptides are derived from peptides generated in the cytosol or nucleus during protein degradation by the ubiquitin-proteasome pathway1,
2,
3,
4. The 26S proteasome degrades proteins to peptides of  2−25 residues long5,
6. Nearly all these peptides are hydrolyzed quickly to amino acids by cytosolic peptidases6,
7. However, in higher vertebrates, a small fraction escapes complete degradation and is transported into the endoplasmic reticulum (ER), where it binds to MHC class I molecules and is exported to the cell surface8.
To fit into the groove in MHC class I molecules, these peptides must be 8−11 residues long. Approximately 70% of proteasome products are too short to do so5, about 15% are of appropriate length and 15% are too long, but could function in antigen presentation if trimmed by exopeptidases5,
6,
12. Studies with inhibitors have shown that proteasomes generate the correct COOH terminus for the MHC class I−presented peptide9,
10 and sometimes also the correct NH2 terminus11,
12. However, proteasomes—and especially the alternative forms induced by interferon- (IFN-) termed "immunoproteasomes"12—primarily generate longer precursors of MHC class I epitopes that are extended at their NH2 termini by one or more residues12,
13,
14,
15. The conversion of these NH2-extended precursors to mature epitopes is not catalyzed by proteasomes9 but by aminopeptidases, as this process can be blocked by derivatization of the peptide's NH2-terminal group10.
This trimming of precursor peptides to mature epitopes can occur in the cytosol or the ER. Three aminopeptidases have been implicated in this cytosolic process: leucine aminopeptidase (LAP)16, puromycin-sensitive aminopeptidase and bleomycin hydrolase17. One of these cytosolic enzymes, LAP, is induced by IFN-16, which stimulates antigen presentation, and thus LAP is likely to play a special role in processing antigenic precursors in vivo. IFN- also stimulates the induction of other components of this pathway; these components include MHC class I, the peptide transporter (TAP) and three proteasome  -subunits, whose incorporation in place of the constitutive subunits alters the cleavage specificity so that more peptides are generated with the correct COOH-terminal residues for MHC class I binding12,
19,
20,
21. In addition, these changes in specificity influence the NH2 termini of the proteasomal products and increase specifically the yield of NH2-extended precursors to antigenic peptides12.
Aminopeptidase(s) in the ER also can contribute to the generation of mature MHC class I epitopes from these larger precursors. In fact, certain MHC class I epitopes appear to be transported into the ER primarily as NH2-extended peptides. This is probably because such precursors have higher affinities for TAP than the mature epitopes22,
23 and perhaps because these longer forms are destroyed more slowly by peptidases in the cytosol6,
24. A variety of studies have demonstrated proteolytic processing of precursor peptides transported into the ER in vivo and in isolated microsomes. For example, when NH2-extended versions of antigenic peptides fused to an ER-targeting sequence were expressed in cells, the mature epitopes were presented on surface MHC class I molecules9,
25,
26,
27 after cleavage of the signal sequences by signal peptidase and removal of the additional NH2-terminal residues by aminopeptidase(s) in the ER. Studies with purified microsomes have demonstrated that this trimming of TAP-translocated precursors occurs in the lumen of the ER, involves metallopeptidase(s) that can be inhibited by o-phenanthroline, leucinethiol and leucine-chloromethylketone (L-Cmk) and can efficiently remove from the NH2 termini all flanking residues except proline14,
28,
29,
30.
However, the ER aminopeptidase(s) that catalyzes this critical step has not yet been identified. Aside from signal peptidase31, no known peptidase has been demonstrated convincingly in the ER32. The proposal33 that MHC class I might function as the trimming enzyme has been disproved26. More recently, the ER-lumenal chaperone Grp94 (also known as gp96) has been reported as having aminopeptidase activity34; however, this claim is now in question35.
We undertook these studies to identify and characterize the ER-aminopeptidase involved in antigen presentation. Using traditional biochemical approaches, we isolated from the ER lumen an enzyme that is capable of trimming various NH2-extended antigenic peptides and showed that it corresponds to an aminopeptidase called adipocyte-derived leucine aminopeptidase (A-LAP)36,
37,
38 or puromycin-insensitive leucyl-specific aminopeptidase (PILS-AP)39. We show here that this enzyme, like many components of the antigen-presentation pathway, is induced by IFN-. In addition, its inhibitor sensitivity and preference for longer peptide substrates accounts for the processing activity reported in the ER. We found that when overexpressed, this enzyme enhanced processing in the ER of an NH2-extended ovalbumin-derived precursor, LEQLESIINFEKL, and presentation of the mature epitope SIINFEKL. Because the present names for this enzyme are inaccurate and misleading, we propose that it be renamed endoplasmic reticulum aminopeptidase 1 (ERAP1).
Results
Aminopeptidases are localized in the ER lumen
Peptide trimming must occur primarily inside the ER, as the processing in isolated microsomes of NH2-extended precursors to MHC class I epitopes requires their transport by TAP and does not involve proteins on the outer face of the ER14,
28,
29,
30. To determine whether the responsible aminopeptidases are associated with the membrane fraction or are soluble in the ER lumen, rough ER vesicles purified from rat liver were fractionated with digitonin (Fig. 1a). The soluble fraction contained high amounts of the lumenal marker protein disulfide isomerase (PDI) and was free of ER membranes, as shown by the complete absence of the membrane marker TRAP (translocon-associated protein ) (Fig. 1b). Virtually all of the aminopeptidase activity recovered (94% of the activity against leucine-7-amino-4-methylcoumarin (L-Amc), 98% of the activity against R-Amc- and 86% of the activity against AAF-Amc) were recovered in the ER lumenal fraction. We also analyzed by reverse-phase high-performance liquid chromatography (RP-HPLC) trimming of ESIINFEKL, the NH2-extended precursor to the ovalbumin-derived H-2Kb−binding epitope SIINFEKL. This process, which is catalyzed by liver vesicles30, occurred predominantly (89%) in the lumenal fraction of our preparation. No endoproteolytic activity was present, but single amino acids were trimmed sequentially from the NH2 terminus of ESIINFEKL (Fig. 1c), QLESIINFEKL and several other NH2-extended precursors to antigenic peptides (see below). The inability of the membrane fraction to hydrolyze these substrates was not due to the substrate's limited access into membrane vesicles because all reactions were carried out with 0.2% digitonin (which did not itself inhibit the aminopeptidase activities). Thus, the enzyme responsible for antigen processing was found predominantly in the lumenal fraction.
 | | |
Purification and identification of the key aminopeptidase
To identify the enzymes responsible for trimming ESIINFEKL, we carefully fractionated ER-lumenal proteins on a UNO Q-1 anion exchange column. At least four peaks of aminopeptidase activity were identified upon high-resolution anion exchange chromatography of ER lumenal proteins (Fig. 2a). Weak ESIINFEKL-trimming and major L-Amc− and R-Amc−degrading activities were detected in the flow-through fractions in peak I (Fig. 2a), but the major trimming activity coeluted specifically with L-Amc−degrading aminopeptidase in peak IIA (Fig. 2a). One additional R-Amc−degrading peak was found in peak IIB (Fig. 2a). We were unable to detect Grp94 in peak fractions, which had been suggested to function as the ER-trimming aminopeptidase34.
| | |
To identify the responsible enzyme, we performed SDS−polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the fractions active in the trimming of the antigenic precursors (Fig. 2a). Three Coomassie blue−stained protein bands appeared to coelute with the ESIINFEKL-trimming activity (Fig. 2b). These candidate proteins were excised, digested with trypsin and the peptide sequences were obtained by online RP-HPLC and tandem mass spectrometrical (MS-MS) analysis. Among the sequences obtained, 11 tryptic fragments derived from a 106-kD protein matched to sequences in a zinc-dependent metallopeptidase that has been described36,
39. This protein has been assigned the names A-LAP and PILS-AP, which are misleading in light of its localization and biochemical properties (see below)36,
39. Therefore, we suggest that it be renamed the ER aminopeptidase 1 (ERAP1). This enzyme coeluted with the ESIINFEKL-trimming activity in chromatographic fractions, as demonstrated by immunoblotting with antiserum raised against the recombinant human protein (Fig. 2c). Size-exclusion chromatography of the ER lumen revealed one peak of ESIINFEKL trimming, which coeluted with the ERAP1 immunoreactivity (Web Fig. 1 online). The polypeptide sequence predicts a molecular weight of 106 kD without glycosylation. The molecular mass of this trimming peak was 150 kD, which agrees with that reported for this enzyme37.
Subcellular localization of ERAP1
Purification of ERAP1 from microsomes suggested that this enzyme is localized to the ER. However, published reports have concluded that this enzyme is cytosolic36, secreted37 or associated with undefined intracellular vesicles39. To localize the endogenous enzyme, HeLa S3 cells were analyzed immunocytochemically with affinity-purified ERAP1 antiserum and antibodies raised against the ER-retention signal sequence KDEL. With confocal microscopy, the enzyme was located in vesicular structures in the cytosol of HeLa S3 cells together with the KDEL immunoreactivity (Fig. 3). This colocalization—which was consistent with the presence of a signal sequence at the enzyme's NH2 terminus39—indicated that ERAP1, although it lacks a KDEL sequence36,
39, is found specifically in the ER lumen in intact cells together with the KDEL-containing proteins.
Trimming of precursors of antigenic peptides by ERAP1
Initial work on this enzyme concluded that it hydrolyzed L-Amc rapidly, M-Amc slowly and did not hydrolyze other amino acid−Amc substrates36,
39. Such narrow substrate specificity is not consistent with a role in processing the diverse sequences preceding MHC class I epitopes. More recently, however, this enzyme was reported to remove a much broader set of residues from the NH2 termini of peptide hormones37. To test whether ERAP1 can trim antigen precursors in a similar way to ER vesicles (Fig. 1c), we incubated the recombinant enzyme with NH2-extended versions of SIINFEKL, ESIINFEKL and QLESIINFEKL, and precursors of the Sendai virus−derived epitope FAPGNYPAL, EFAPGNYPAL and HGEFAPGNYPAL. RP-HPLC analysis of the products revealed that the pure enzyme efficiently removed diverse NH2-terminal residues from these and other NH2-extended antigen precursors to generate the final epitopes (Fig. 4 and Web Fig. 2 online); this result contrasted with its strong specificity for leucines in dipeptides, which we confirmed with various fluorogenic dipeptides (Web Table 1 online).
| | |
This apparent preference for longer peptides was investigated further because it could be particularly important in the processing of NH2-extended precursors to the MHC class I epitopes, most of which are 8−10 residues long. Further analysis of these data revealed that, upon incubation with pure ERAP1, the 11-residue precursor QLESIINFEKL disappeared rapidly (as shown by RP-HPLC at different times). Concomitantly, there was a transient buildup of the 10-residue peptide and a later, transient accumulation of the 9-residue peptide ESIINFEKL. With time, the amount of mature 8-residue peptide steadily increased, and the trimming process almost ceased when most of the 11-residue peptide was converted to SIINFEKL (Fig. 4a). This lack of further trimming was not caused by enzyme inactivation because, if additional substrate was provided, ERAP1 remained active. These kinetics indicated sequential (nonprocessive) removal of the NH2-terminal residues to yield the mature epitope and suggested preferential degradation of the longer precursors, as was directly demonstrated by incubating ERAP1 with equal concentrations of these peptides. Both QLESIINFEKL and ESIINFEKL were digested much faster than the 8-residue peptide or the 7-residue peptide IINFEKL (Fig. 4b). Preferential digestion of peptides longer than 9 residues was also seen with HGEFAPGNYPAL (Web Fig. 2 online) and other antigenic precursors. This inability to digest rapidly peptides of 8 residues or shorter suggested that ERAP1 is specially adapted to function in antigen processing.
Effect of inhibitors on trimming activities
The processing of antigenic precursors in microsomes is inhibited by leucinethiol14, 1,10-phenanthroline28,
30 and L-Cmk29. These inhibitors, but not puromycin, AAF-Cmk or F-Cmk, completely blocked the conversion of ESIINFEKL to SIINFEKL by both the ER lumen and pure ERAP1 (Table 1). The general aminopeptidase inhibitor bestatin also blocked trimming by the pure enzyme completely, but reduced this process in the lumenal fraction by 49% (Table 1); this was presumably because of the other aminopeptidases present in this fraction. Inhibitors of proteasomes (lactacystin) and of serine (phenylmethyl sulfonyl fluoride), cysteine (E64) and aspartic (pepstatin) peptidases had no consistent effects on the lumenal extract or pure enzyme36,
39. Because of these similar susceptibilities to inhibitors and similar substrate specificities, ERAP1 appears responsible for the trimming activity in microsomes and in the ER lumen.
| | Table 1. Effect of various protease inhibitors on ESIINFEKL trimming by the ER-lumenal fraction and recombinant ERAP1 | | |  |
 Full Table |
| |
Effect of ERAP1 depletion on trimming in the ER
To test whether ERAP1 is in fact responsible for most of this activity in the ER, we studied the effects of loss of this enzyme on different aminopeptidase activities in the lumenal fraction. After immunodepletion with an antiserum against human ERAP1, immunoblot analysis confirmed almost complete removal of this protein without any loss of Grp94 (Fig. 5a). As expected, loss of ERAP1 did not affect the hydrolysis of R-Amc, which was cleaved by a distinct enzyme (Fig. 2a). After removal of ERAP1, L-Amc hydrolysis decreased by only 19%, in accordance with the finding that the ER lumen contained at least three other proteins that are active against L-Amc (Fig. 2a). In contrast, immunodepletion of ERAP1 decreased cleavage of the NH2-terminal residue of ESIINFEKL by 75% (Fig. 5b), of QLESIINFEKL by 54% (Fig. 5c) and of the NH2-extended Sendai virus peptide HGEFAPGNYPAL by 40% (data not shown). Thus, ERAP1 is a key player in the trimming of certain precursors to antigenic peptides in the ER of normal cells.
| | |
Effect of IFN- on ERAP1 expression and activity
IFN- induces many components of the MHC class I pathway, including LAP, the major ESIINFEKL-trimming activity in the cytosol16. To test whether ERAP1 is likely to contribute to antigen processing in vivo, we tested whether it is also induced by IFN-. After treatment of HeLa S3 cells with this cytokine, the expression of ERAP1 protein and mRNA was determined. This protein (but not Grp94) was induced several fold in a concentration-dependent manner by IFN- (Fig. 6a); induction was even stronger in U937 (Fig. 6b) and SW620 cells (data not shown). In addition, differential centrifugation of the HeLa cell extracts after IFN- treatment showed that most of the ERAP1 immunoreactivity (in contrast to that of LAP) was in the microsomal fraction (Fig. 6c). This induction appears to have resulted from increased transcription, as ERAP1 mRNA also increased after IFN- treatment (Fig. 6d). In contrast, IFN- did not affect mRNA abundance for other aminopeptidases or thimet oligopeptidase24 (data not shown).
| | |
To examine the biochemical consequences of this increase in ERAP1 content, we analyzed ESIINFEKL-trimming in the microsomal lumen from IFN-−treated and untreated HeLa S3 cells. IFN- enhanced the conversion of ESIINFEKL to SIINFEKL, and this increase was completely eliminated by immunodepletion of ERAP1 (Fig. 7). Thus, ERAP1 represented the major trimming activity in the ER lumen after IFN- treatment and appeared to be the only ESIINFEKL-trimming activity induced by IFN- (Fig. 7).
ERAP1 enhances antigen presentation in vivo
These findings suggested that ERAP plays an important role in the processing of antigenic peptides in vivo. To test whether an increase in ERAP1 content, as occurs upon IFN- treatment, can actually enhance the generation of antigenic peptides from NH2-extended precursors in the ER of intact cells, we used as a model substrate LEQLESIINFEKL targeted to the ER with a signal sequence. This construct is trimmed efficiently in the ER to the presented peptide9. COS cells stably transfected with H-2Kb were cotransfected with plasmids expressing LEQLESIINFEKL and either ERAP1 or an empty vector. Two days later, cell-surface expression of SIINFEKL−H-2Kb complexes was measured with the specific antibody 25.D1.16. When ERAP1 was overexpressed, the generation of SIINFEKL−H-2Kb complexes from this precursor was increased approximately threefold (Fig. 8a). This increase in SIINFEKL presentation required the aminopeptidase activity of ERAP1, as no such stimulation was seen upon transfection of an inactive form—an E354A mutant that was generated by site-directed mutagenesis and should lack the catalytic Zn2+—(Fig. 8b). To rule out nonspecific effects on ER function, we cotransfected ERAP1 or the vector alone with a plasmid expressing influenza hemagglutinin (HA). The expression and proper folding of this surface protein was not greatly affected by ERAP1 overexpression (Fig. 8c). These data demonstrated that trimming of antigenic precursors in the ER was rate-limiting for antigen presentation in vivo, and increasing ERAP1 expression by transfection or IFN- treatment enhanced this process.
| | |
Discussion
There is growing evidence that the production by proteasomes (or signal peptidase) of antigenic peptides with one or more additional NH2-terminal residue is an important source of MHC class I epitopes7. For ovalbumin, the one model antigen that has been extensively studied, NH2-extended precursors of SIINFEKL are released preferentially by proteasomes, and especially by immunoproteasomes12, and appear to be more stable in cytosolic extracts6,
24. Trimming of these precursors in vivo is highly efficient and can be catalyzed by aminopeptidases in both the cytosol and ER14,
22,
27,
28,
29,
30,
32,
40. In the ER, ERAP1 appears to be the major aminopeptidase that is active in trimming many antigenic precursors. This enzyme displays a high homology (45−49% identity) to other enzymes in the M1 family of zinc-metallopeptidases, all of which contain a catalytic HEXXH(X)18E Zn-binding motif41,
42 that is essential for the function of ERAP1 in antigen processing.
The function of this enzyme was unclear and roles in blood pressure regulation43 and angiogenesis44 have been suggested. We have demonstrated here that an important function of this metallopeptidase is in the processing of NH2-extended precursors of MHC class I epitopes in the ER. Not only is it in the correct cellular compartment for such a role, but it accounts for most of the trimming of SIINFEKL precursors in microsomal extracts, especially after IFN- treatment. Also, ERAP1 is sensitive to the same protease inhibitors that block precursor processing in ER vesicles14,
28,
29,
30 and it is expressed in all tissues tested36,
39. Such a broad distribution is consistent with a role in MHC class I presentation, which occurs in all nucleated cells.
The subcellular localization of this enzyme has been controversial36,
37,
39. We have established here that its normal localization is in the ER lumen because it was purified from rough microsomes, was recovered primarily in this fraction and was colocalized by immunocytochemistry with the KDEL-retention signal, a specific marker of the ER. The previously reported localization of A-LAP to the cytosol was probably caused by disruption of the ER during cell homogenization36. In addition, this protein contains a predicted 20-residue signal sequence at its NH2 terminus39. Because ERAP1 lacks a KDEL sequence on its COOH terminus, its retention in the ER lumen is probably mediated through an association with a KDEL-containing protein or an ER-membrane component, such as the TAP-MHC-tapasin complex.
For certain antigenic precursors, ERAP1 is the primary processing enzyme in the ER. We found no evidence to support the proposal that the ER chaperone Grp94 could serve this role34, which had also been questioned by others35. Grp94 did not coelute with any aminopeptidase peak, and the inhibitor sensitivity reported for its proposed aminopeptidase activity34 does not correlate with that found for ESIINFEKL-trimming by the ER lumen or ERAP130. Although immunodepletion of ERAP1 from the ER lumen reduced the trimming of SIINFEKL precursors, additional soluble aminopeptidases were detected in the microsomal extracts that might play a role in the processing of other antigen precursors. However, it is presently unclear whether these enzymes are in fact ER-resident proteins.
We found in the ER only soluble aminopeptidases and found no evidence of endoproteolytic or carboxypeptidase activity against extended antigenic peptides. The cytosol also lacks carboxypeptidase activities9,
10,
16 but does contain several endopeptidases, which—in concert with aminopeptidases—catalyze rapid digestion of the majority of proteasome products to amino acids (unpublished data). These endopeptidases, especially thimet oligopeptidase, can destroy MHC class I antigenic peptides or their precursors in the cytosol6,
24 and thus limit the rate of antigen presentation in vivo45.
An unusual feature of this enzyme, which suggests a role in antigen processing, is that it removes sequentially a wide range of NH2-terminal amino acids from NH2-extended antigenic peptides, but not from peptides of 8 residues or shorter. In fact, our initial identification of the ESIINFEKL-trimming activity with the enzyme termed A-LAP and PILS-AP, ERAP1, seemed problematic based on its reported specificities. When first characterized against fluorogenic aminopeptidase substrates, it was active only against L-Amc36,
39, which led to the inaccurate name "leucyl-specific aminopeptidase". However, it was later shown that it acts on peptide hormones with diverse NH2 termini37, and we found that it can cleave all NH2-terminal residues tested from antigenic precursors. This inability of ERAP1 to cleave these same residues from dipeptides reflects its preference for substrates longer than 8 residues. For example, ERAP1 converted the 11-residue precursor to SIINFEKL, which resisted further hydrolysis. These features are unprecedented for aminopeptidases and even suggest that ERAP1's substrate-binding site may somehow resemble that in MHC class I molecules.
We found that efficient trimming of antigenic precursors by ERAP1 to 8-residue peptides45 thus occurred in the absence of ER membranes, and peptide processing in the ER does not require MHC class I or TAP28,
30. It is possible that tight binding of the mature epitopes to MHC complexes in vivo may prevent further cleavages by aminopeptidases because in microsomes lacking the appropriate MHC class I, antigenic peptides tend to be degraded rapidly13,
28,
29,
30. ERAP1 may also play a role in this degradative process, as shown by its ability to trim the mature epitope FAPGNYPAL at similar rates to its NH2-extended precursors. Thus, in addition to processing antigenic precursors, under certain conditions, ERAP1 can destroy certain epitopes, as shown by York et al. in this issue45.
Further evidence that ERAP1 plays a key role in antigen presentation in vivo was provided by the finding that it is induced by IFN-, which leads to increased processing of antigenic precursors in the ER lumen. In addition to ERAP1, IFN- induces LAP—the enzyme most active in trimming ESIINFEKL in the cytosol16—and stimulates antigen presentation by increasing the expression of many other components of the MHC class I pathway, including immunoproteasomes, which generate preferentially NH2-extended versions of SIINFEKL during ovalbumin degradation12. IFN- also induces the PA28 complex and the formation of hybrid, PA28−20S-19S, proteasome complexes, which generate a greater diversity of peptide products48 and may also produce longer forms of antigenic precursors12,
46,
47. These NH2-extended precursors tend to be less susceptible to cytosolic peptidases24 and to be transported efficiently by TAP22,
23. Therefore, the IFN-−induced changes in proteasome function and TAP (by favoring the generation of longer precursors) and the induction of the major ER and cytosolic aminopeptidases should have synergistic effects in increasing the yield of antigenic peptides. Presumably the inducibility of these enzymes helps maintain antigen presentation at low levels normally, except during inflammatory states, when IFN- production rises.
We have presented here a variety of circumstantial evidence that indicates a major role for ERAP1 in MHC class I antigen processing in vivo. However, definitive evidence for a key role in antigen processing in vivo was obtained by overexpressing ERAP1 in cultured cells, which mimicked the response of cells to IFN- and stimulated the presentation of an ER-targeted precursor. Thus, ERAP1 can catalyze a rate-limiting step in antigen presentation, but its actual importance in the processing of any particular antigenic peptide must depend on the amino acid sequence released by proteasomes, its affinity for the TAP transporter and its susceptibility to hydrolysis by cytosolic peptidases. The findings we have presented here show that ERAP1 can catalyze a rate-limiting step in antigen processing; for systematic studies on ERAP1's importance in vivo under different physiological conditions, see the article by York et al. in this issue45.
Methods
Reagents.
Peptides were synthesized by Research Genetics (Huntsville, AL), New England Peptide, Inc. (Fitchburg, MA) or at a core facility at UMass Medical School and were stored in dimethyl sulfoxide at -20 °C. The peptide ESIINFEKL was >95% pure by MS analysis and eluted as a single peak when a trifluoroacetic acid (TFA)−acetonitrile mobile phase was used in RP-HPLC. The purity of other peptides was at least 80% by HPLC. The fluorogenic substrates and chloromethylketone inhibitors were obtained from Bachem (Basel, Switzerland) and all others were from Sigma (St. Louis, MO). Digitonin (5% solution) and the TRAP antiserum were provided by T. Rapoport (Harvard Medical School) and the PDI-antiserum by H. Ploegh (Harvard Medical School). Recombinant human A-LAP and the antisera against A-LAP were as described37. The monoclonal antibodies to Grp94 and ER-retention signal KDEL were from StressGen Bioreagents (Victoria, Canada).
Cell lines.
All cells were from ATCC (Manassas, VA). HeLa S3 cells were grown at 37 °C in Dulbecco's modified Eagle's medium (Irving Scientific, Santa Ana, CA), the SW620 cells were in Leibovitz's L-15 medium (Invitrogen Corporation, Carlsbad, CA) and the U937 cells were in RPMI-1640 medium (ICN Biochemicals, Aurora, OI). All media were supplemented with 10% fetal calf serum and antibiotics.
Purification of the rough ER.
The ER was isolated from rat liver by discontinuous density gradient centrifugation49. Livers were removed from freshly killed rats (200−350 g) and homogenized in four volumes (w/v) of buffer (50 mM HEPES-KOH at pH 7.6, 50 mM K acetate, 5 mM MgCl2, 1 mM DTT and 250 mM sucrose) in a Potter-Elvehjem glass-Teflon type tissue grinder. All procedures were done on ice or at 4 °C. The homogenate was centrifuged at 1,500g for 10 min and then at 20,000g for 10 min. The lipid layer was discarded and the rough ER in the supernatant was pelleted by ultracentrifugation through a cushion of 1.3 M sucrose in the homogenization buffer for 2.5 h at a maximum of 240,000g. To reduce contamination by cytosolic proteins, pellets were resuspended and recentrifuged at a maximum of 240,000g for 60 min. Final pellets were resuspended in 50 mM HEPES-KOH at pH 7.6 containing 1 mM DTT and 250 mM sucrose and were frozen in liquid nitrogen and stored in aliquots at -80 °C. Protein concentration in the purified ER was 10−25 mg/ml. The concentration of ER proteins was also determined by UV-absorbance at 280 nm, where 1 Eq corresponded to a vesicle concentration in 1  l of solution adjusted to 50 U/ml at A280. Contamination of the purified ER by cytosolic proteins was negligible, as assessed by assay of the cytosolic enzymes, thimet oligopeptidase and prolyl oligopeptidase24.
Extraction of the ER-lumenal fraction.
Separation of the soluble (lumenal) and membrane fractions was done as described50. The purified ER vesicles were permeabilized at a concentration of 0.5 Eq/l in 20 mM Tris-HCl buffer at pH 7.6, containing 5 mM Mg acetate, 2 mM DTT, 50 mM NaCl, 12% glycerol and 0.2% digitonin. After 15−30 min on ice, samples were centrifuged at 300,000g for 40 min at 4 °C in a TLA 120.2 rotor. The supernatants containing lumenal proteins were combined, and the pelleted membrane components resuspended in the starting volume. All samples were snap frozen in liquid nitrogen and stored at -80 °C.
Anion-exchange chromatography.
Proteins extracted from the ER lumen (14 mg) were loaded on the UNO Q-1 column (BioRad Laboratories, Hercules, CA) equilibrated in 20 mM Tris-HCl buffer adjusted to pH 7.4 at room temperature, containing 2 mM Mg acetate and 12% glycerol. Bound proteins were eluted at a flow rate of 1 ml/min with a 30 ml 0−0.4 M NaCl gradient. Fractions (0.5 ml) were collected, and aliquots were analyzed for aminopeptidase activities or by SDS-PAGE (4−12% NuPAGE Bis-Tris gradient gel, Invitrogen). Gels were stained with Coomassie blue and the protein bands that coeluted with the ESIINFEKL-trimming activity were selected for online RP-HPLC fractionation and tandem mass spectrometrical analysis and sequencing (LC−MS-MS). Mass spectrometric analysis was performed by the Taplin Spectrometry Facility (Harvard Medical School).
Aminopeptidase assays.
Aminopeptidase activities were monitored with L-Amc, R-Amc or various NH2-extended antigenic peptides as substrates. Samples were incubated at 37 °C with fluorogenic substrates (100 M) or synthetic peptides (150 nmol/ml) in 50 l of buffer containing 50 mM Tris-HCl at pH 7.8 and 0.5 g/l protease-free bovine serum albumin (Sigma). Hydrolysis of fluorogenic substrates was measured at excitation and emission wavelengths of 380 and 460 nm, respectively, in a continuous assay on a microplate fluorescence reader (FLUOstar Galaxy, BMG Labtechnologies GmbH, Offenburg, Germany).
The trimming of synthetic peptides was analyzed by RP-HPLC. The reactions were terminated by adding an equal volume of 20% trichloroacetic acid or 0.6% TFA. After 15 min on ice, the precipitated protein was removed by centrifugation. The peptide-containing supernatant was loaded on a 4.6  50 mm TARGA 3 m C18 column (Higgins Analytical, Inc., Mountain View, CA) in 10 mM sodium phosphate buffer at pH 6.8 containing 7% acetonitrile or in 0.06% TFA−7% acetonitrile. Elution was done with a linear 7−31.5% acetonitrile gradient. The amount of a peptide trimmed was calculated by the integration of peptide peaks. The identity of peaks was determined by comparing their retention times to those of pure synthetic peptides.
Immunodepletion of ERAP1.
Protein G Plus−Protein A Agarose (30 l of settled resin, Oncogene, San Diego, CA) was incubated with 4 l of rabbit preimmune serum or anti−human A-LAP serum in 300 l of 50 mM HEPES at pH 7.6 containing 140 mM NaCl and 10 mM KCl (immunoprecipitation-buffer). After 60 min at room temperature, resins were washed, and 200 g of the ER-lumenal proteins were added in a final volume of 300 l/resin. After mixing for 2 h at 4 °C, resins were pelleted and the supernatants transferred to new tubes. Resins were washed with the immunoprecipitation buffer, and the proteins bound to resins were eluted by boiling in SDS-PAGE−loading buffer. All samples were stored at -80 °C.
Immunocytochemistry.
HeLa S3 cells grown on coverslips were rinsed in PBS and fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. After washing with PBS, cells were permeabilized for 5 min in PBS containing 0.3% Triton X-100 (T-PBS). Coverslips were blocked for 1 h with T-PBS containing 3% bovine serum albumin (blocking buffer), then incubated with 5 g/ml of affinity-purified anti A-LAP and anti-KDEL in blocking buffer for 1.5 h. After washing with T-PBS, cells were incubated for 1 h with 0.5 g/ml of Alexa Fluor 488−labeled anti−rabbit IgG and Alexa Fluor 568−labeled anti−mouse IgG (Molecular Probes, Eugene, OR) in blocking buffer, mounted on microscope slides with PermaFluor Aqueous Mounting Medium (Immunon, Pittsburgh, PA), and viewed with a Leica TCS NT laser scanning microscope (Leica, Wetzlar, Germany.
IFN- treatment and immunoblot analysis.
HeLa S3 and U937 cells were treated with IFN- (PeproTech, Rocky Hill, NJ and Biogen, Cambridge, MA). After treatment, cells were washed with PBS, and HeLa cells were scraped into a buffer (50 mM HEPES-KOH at pH 7.6, 50 mM K acetate, 5 mM Mg acetate, 1 mM DTT, 0.5 mM EDTA and 250 mM sucrose). After gentle homogenization in a Potter-Elvehjem glass-Teflon tissue grinder, the homogenate was centrifuged at 1,500g for 5 min at 4 °C. The microsomes were separated from the cytosolic fraction by spinning the 1,500g supernatant at 10,000g for 5 min. The U937 cells were lysed in 50 mM Tris-HCl at pH 7.4, 150 mM NaCl, 1% NP-40, 0.25% Na deoxycholate, 0.1% SDS and 1mM EDTA. HeLa lysate (20 g) and the cytosolic or ER fractions (10 g) were separated by SDS-PAGE. Immunoblot analysis was done as described37.
Transfection of ERAP1 and antigen-presentation assay.
Cos-Kb (COS7 cells stably transfected with H-2Kb) cells were cultured in RPMI-1640 medium and supplemented with 10% fetal calf serum and G418 in the presence of 5% CO2. Human ERAP1 cDNA36,
37 was subcloned into pTracerCMV (Invitrogen). A minigene encoding LEQLESIINFEKL9 targeted to the ER with a signal sequence was subcloned into pcDNA3.1 (Invitrogen), and an influenza A/PR8/34 HA cDNA was subcloned into pcDNA1Amp (Invitrogen). COS-Kb cells were cotransfected with the use of FuGene6 with pTracerERAP1 (or a control plasmid) and a pCDNA plasmid expressing ERAP1, HA or no antigen, according to the manufacturer's directions (Roche, Indianapolis, IN). After 24−48 h, surface SIINFEKL-MHC complexes were detected with the antibody 25.D1.1651, which recognizes SIINFEKL in combination with H-2Kb. Influenza HA was detected with the monoclonal antibody H36.4.5 (provided by W. Gerhard, University of Pennsylvania). Surface fluorescence of transfected, green fluorescent protein−positive (GFP+) cells was quantified by flow cytometry.
Site-directed mutagenesis.
Human ERAP1 cDNA was subcloned into pcDNA6/myc-His (Invitrogen). Glu354 within the active site of ERAP1 was mutated to alanine by the PCR overlap primer method (QuikChange XL Site-Directed Mutagenesis Kit, Stratagene, La Jolla, CA) with the 5' primer-1 CATCACAATGACTGTGGCCCATGCACTAGCCC and the 3' primer-2 CCCAAACCACTGGTGGGCTAGTGCATGGGCCA to create the point mutation (GAA to GCA). The resulting plasmid was used for in vivo transfection. Italics denote changed codon.
Note added in proof.
While this article was in press, similar results were reported in Serwold, T. et al., ERAAP customizes peptides for MHC class I molecules in the endoplasmic reticulum. Nature
419, 480−483 (2002).
Note: Supplementary information is available on the Nature Immunology website.
Received 3 July 2002; Accepted 18 October 2002; Published online: 18 November 2002.
REFERENCES
-
Rock, K.L. et al. Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules. Cell 78, 761–771 (1994). | PubMed | ISI | ChemPort |
-
Harding, C.V. et al. Novel dipeptide aldehydes are proteasome inhibitors and block the MHC-I antigen-processing pathway. J. Immunol. 155, 1767–1775 (1995). | PubMed | ISI | ChemPort |
-
Cerundolo, V. et al. The proteasome-specific inhibitor lactacystin blocks presentation of cytotoxic T lymphocyte epitopes in human and murine cells. Eur. J. Immunol. 27, 336–341 (1997). | PubMed | ISI | ChemPort |
-
Craiu, A. et al. Lactacystin and clasto-lactacystin -lactone modify multiple proteasome -subunits and inhibit intracellular protein degradation and major histocompatibility complex class I antigen presentation. J. Biol. Chem. 272, 13437–13445 (1997). | Article | PubMed | ISI | ChemPort |
-
Kisselev, A.F., Akopian, T.N., Woo, K.M. & Goldberg, A.L. The sizes of peptides generated from protein by mammalian 26 and 20S proteasomes. Implications for understanding the degradative mechanism and antigen presentation. J. Biol. Chem. 274, 3363–3371 (1999). | Article | PubMed | ISI | ChemPort |
-
Goldberg, A.L., Cascio, P., Saric, T. & Rock, K.L. The importance of the proteasome and subsequent proteolytic steps in the generation of antigenic peptides. Mol. Immunol. 39, 147–164 (2002). | Article | PubMed | ISI | ChemPort |
-
Botbol, V. & Scornik, O.A. Peptide intermediates in the degradation of cellular proteins. Progress Biol. Res. 180, 573–583 (1985). | ChemPort |
-
Rock, K.L. & Goldberg, A.L. Degradation of cell proteins and the generation of MHC class I-presented peptides. Annu. Rev. Immunol. 17, 739–779 (1999). | Article | PubMed | ISI | ChemPort |
-
Craiu, A., Akopian, T., Goldberg, A. & Rock, K.L. Two distinct proteolytic processes in the generation of a major histocompatibility complex class I-presented peptide. Proc. Natl. Acad. Sci. USA 94, 10850–10855 (1997). | Article | PubMed | ChemPort |
-
Mo, X.Y., Cascio, P., Lemerise, K., Goldberg, A.L. & Rock, K. Distinct proteolytic processes generate the C and N termini of MHC class I-binding peptides. J. Immunol. 163, 5851–5859 (1999). | PubMed | ISI | ChemPort |
-
Lucchiari-Hartz, M. et al. Cytotoxic T lymphocyte epitopes of HIV-1 Nef: Generation of multiple definitive major histocompatibility complex class I ligands by proteasomes. J. Exp. Med. 191, 239–252 (2000). | Article | PubMed | ISI | ChemPort |
-
Cascio, P., Hilton, C., Kisselev, A.F., Rock, K.L. & Goldberg, A.L. 26S proteasomes and immunoproteasomes produce mainly NH2-extended versions of an antigenic peptide. EMBO J. 20, 2357–2366 (2001). | Article | PubMed | ISI | ChemPort |
-
Paz, P., Brouwenstijn, N., Perry, R. & Shastri, N. Discrete proteolytic intermediates in the MHC
|
