The Cytosolic Endopeptidase, Thimet Oligopeptidase, Destroys Antigenic Peptides and Limits the Extent of MHC Class I Antigen Presentation

Immunity, Vol. 18, 429–440, March, 2003, Copyright 2003 by Cell Press

The Cytosolic Endopeptidase, Thimet Oligopeptidase, Destroys Antigenic Peptides and Limits the Extent of MHC Class I Antigen Presentation Ian A. York,1 Annie X.Y. Mo,1,4 Kristen Lemerise,1,5 Wanyong Zeng,1,6 Yuelei Shen,1 Carmela R. Abraham,3 Tomo Saric,2,7 Alfred L. Goldberg,2 and Kenneth L. Rock1,* 1 Department of Pathology University of Massachusetts Medical School Worcester, Massachusetts 01655 2 Department of Cell Biology Harvard Medical School Boston, Massachusetts 02155 3 Department of Biochemistry and Medicine Boston University School of Medicine Boston, Massachusetts 02118

Summary Most antigenic peptides presented on MHC class I molecules are generated by proteasomes during protein breakdown. It is unknown whether these peptides are protected from destruction by cytosolic peptidases. In cytosolic extracts, most antigenic peptides are degraded by the metalloendopeptidase, thimet oligopeptidase (TOP). We therefore examined whether TOP destroys antigenic peptides in vivo. When TOP was overexpressed in cells, class I presentation of antigenic peptides was reduced. In contrast, TOP overexpression didn’t reduce presentation of peptides generated in the endoplasmic reticulum or endosomes. Conversely, preventing TOP expression with siRNA enhanced presentation of antigenic peptides. TOP therefore plays an important role in vivo in degrading peptides released by proteasomes and is a significant factor limiting the extent of antigen presentation. Introduction Cellular proteins are continuously degraded by proteasomes into oligopeptides ranging in size from about 3 to 24 residues (Kisselev et al., 1999; Rock and Goldberg, 1999; Rock et al., 2002). Most of these oligopeptides are rapidly hydrolyzed by cytosolic peptidases to amino acids (Goldberg et al., 2002), which are then re-utilized in protein synthesis or energy production. However, in mammalian cells, a small fraction of these peptides are transported into the endoplasmic reticulum (ER), where, if they have the correct motif and are 8–10 residues long, they become tightly bound by specific MHC class I molecules (Rock and Goldberg, 1999). These peptideMHC complexes are then transported to the plasma *Correspondence: [email protected] 4 Present address: Antigenics Inc., Woburn, Massachusetts 01801. 5 Present address: Viacell Inc., Boston, Massachusetts 02116. 6 Present address: Dana-Farber Cancer Institute, Boston, Massachusetts 02115. 7 Present address: ATABIS GmbH, Joseph-Stelzmann Str. 50, 50931 Cologne, Germany.

membrane for display to T cells. This process allows the immune system to monitor for the presence of cells bearing foreign peptides—for example, ones derived from viral or mutated proteins. It is unclear whether antigenic peptides (or longer precursors) released from proteasomes are protected from hydrolysis by cytosolic peptidases. Some antigenic peptides or larger precursors have been found in association with heat shock proteins (Srivastava et al., 1998) or in other high molecular weight complexes (Paz et al., 1999). It has been proposed that such interactions protect these peptides from destruction by cytosolic enzymes until they are translocated into the ER. It has also been suggested that proteasomes discharge their products directly into the TAP transporter (Hirsch and Ploegh, 2000). Such a mechanism would also protect peptides from hydrolysis in the cytosol. However, there is presently no evidence for either mechanism, each of which has been proposed to account for the high efficiency of peptide presentation that has been reported for certain peptides (Villanueva et al., 1994). Saric et al. (2001) found that a variety of antigenic peptides were rapidly degraded when added to cytosolic extracts. Although cell extracts contained many exo- and endopeptidases capable of degrading oligopeptides, one endopeptidase, thimet oligopeptidase (TOP, EC3.4.25.15), was particularly active in degrading most of the antigenic peptides tested (Saric et al., 2001). TOP is a cytosolic metallopeptidase that hydrolyzes oligopeptides (typically of 6–17 residues) but not larger polypeptides or proteins (Barrett et al., 1995). A variety of functions for TOP have been proposed that involve its attacking substrates in the extracellular space, e.g., degrading neurotransmitters or amyloid ␤ peptide (Vincent et al., 1997; Yamin et al., 1999). Recent findings indicate that in cell extracts, TOP is primarily responsible for further degradation of most larger peptides produced by the proteasome (T.S. et al., unpublished data). However, it is unknown whether TOP can actually destroy antigenic peptides or their longer precursors in intact cells or whether those peptides that are presented on MHC class I molecules are protected from this enzyme and other cytosolic peptidases. The present studies were undertaken to test whether antigenic peptides released by proteasomes in the cytosol are susceptible to destruction by cytosolic peptidases, especially TOP. Results Effect of TOP Overexpression on the Presentation of Peptides Expressed in the Cytosol from Minigenes To test whether TOP can influence antigen presentation, we initially studied the effects of transiently overexpressing this peptidase in vivo on the presentation of SIINFEKL expressed in the cytoplasm from a synthetic minigene. In this situation, the peptide is produced by ribosomes, rather than proteasomes, and therefore we could test if an antigenic peptide in the cytosol is susceptible to hydrolysis by TOP. In one set of experiments,

Immunity 430

Figure 1. TOP Overexpression Inhibits MHC Class I Presentation of Cytosolic SIINFEKL (A) The activities of TOP, aminopeptidases (AP), prolyl oligopeptidase (POP), or trypeptidyl peptidase II (TPPII) were measured as described in Saric et al., 2001, in extracts from COS-Kb cells transfected for 48 hr with pTracer or pTracer-TOP or E36-Kb or DAP-Kb infected for 3 hr with vaccinia virus TOP (VV-TOP) or vaccinia virus-␤gal (VV-␤gal). The data are expressed as the fold-increase in activity in TOP-transfected cells compared to control cells. The coefficient of variation in all groups was ⬍10%. (B) COS-Kb cells were cotransfected with a SIINFEKL minigene in pCDNA3 (or empty pCDNA3 [No Ova]) and a pTracer CMV plasmid encoding GFP (Vector), GFP ⫹ human TOP (TOP), or GFP ⫹ the control protein ORP150 (ORP150), and after 48 hr stained for SIINFEKL ⫹ Kb complexes with 25.D1.16 or an isotype control (Bkg). For clarity the background fluorescence of only one group is shown because it was essentially identical for all others (mean fluorescence intensity [MFI] ⫾ SD for the background on all groups ⫽ 14.6 ⫹/⫺ 1.6). MFI were Bkg, 15.8; No OVA,16.8; TOP, 214; Vector, 875; and ORP150, 922 fluorescent units. (C) E36-Kb APCs were infected with vaccinia-T7 polymerase and either vaccinia-TOP or vaccinia-␤gal and then transfected with a SIINFEKL minigene under the control of a T7 promoter in a pBluescript plasmid. After 1.25 hr the cells were fixed with 1% paraformaldehyde and the indicated number incubated with the SIINFEKL ⫹ Kb specific T cell hybrid, RF33.70. IL-2 production by RF33.70 is displayed. (D) Similar to (B) except E36-Db cells were used instead of E36-Kb, vaccinia-T7 polymerase was omitted, and after 1.5 hr cells were superinfected for 1.5 hr with a vaccinia recombinant expressing ASNENMETM (a gift from J. Yewdell, NIH) instead of being transfected with a SIINFEKL minigene, and the 12.3 hybridoma (Deckhut et al., 1993) (gift from D. Woodland, Trudeau Institute, NY) was used instead of RF33.70. (E) Similar to (B) except that a KVVRFDKL minigene was used instead of SIINFEKL, the 1G8 hybridoma (Cole et al., 1994) (gift from J. McCluskey, Flinders Medical Center) was used instead of RF33.70, and the transfection time was 2.5 hr.

expression vectors encoding SIINFEKL and TOP plus green fluorescent protein (GFP) (which allowed transfected cells to be identified) or control plasmids (GFP alone; or an irrelevant protein, ORP150, plus GFP) were transiently cotransfected into H-2Kb-expressing COS7 cells (COS-Kb). The presentation of SIINFEKL on H-2Kb molecules was then detected by staining with a SIINFEKL⫹Kb-specific monoclonal antibody (25.D1.16) (Porgador et al., 1997) and quantified by flow cytometry. In the TOP-transfected COS-Kb cells, TOP activity was increased 16-fold (Figure 1A), and the generation of SIINFEKL-Kb complexes was markedly reduced to levels below those detected in cells expressing control proteins GFP or ORP150 (Figure 1B). We sought to confirm and extend these results in other APCs. Since most APCs (unlike COS7 cells) do not replicate plasmids episomally, we constructed a recombinant vaccinia virus expressing TOP and used the vaccinia T7-polymerase system to express antigens from the transfected plasmids. E36 and L cells (DAP) were chosen because they are readily infected with vac-

cinia virus and present antigens well to T cell hybridomas ((Michalek et al., 1993; our unpublished data). H-2Kb-expressing E36 cells (E36-Kb) were infected with vaccinia-TOP and vaccinia-T7 and transfected with a plasmid containing the SIINFEKL minigene under the control of a T7 promoter. The presence of SIINFEKL on H-2Kb molecules was assayed by measuring the production of IL-2 from a specific T cell hybridoma. In cells infected with the TOP-encoding vaccinia vector the enzymatic activity of TOP was increased about 4- to 5-fold (Figure 1A). The presentation of SIINFEKL on H-2Kb was markedly inhibited in the cells overexpressing TOP compared to cells infected with a vaccinia control (expressing lacZ) (Figure 1C). Similar results were obtained in experiments performed with DAP cells (see Figure 4C below). Therefore, TOP in at least three different cell types from different species can limit the presentation of SIINFEKL expressed in the cytoplasm. We next investigated whether TOP overexpression would affect the presentation of other antigenic peptides. Overexpression of TOP in intact cells also inhib-

Endopeptidase Limits Antigen Presentation 431

Figure 2. TOP Overexpression Does Not Affect Cell Viability, Protein Synthesis, or Expression of HA (A) E36-Kb were infected with vaccinia encoding ␤gal (VV-␤gal), TOP (VV-TOP), or uninfected (none), or COS-Kb cells were untransfected (none), or transfected with pTracerCMV (GFP), or pTracerCMV-TOP (TOP) as described in Figure 1, and viability was determined by exclusion of trypan blue dye. Data are shown as the mean viability ⫾ SEM for multiple experiments. (B) COS-Kb cells were transfected, or E36-Kb cells were infected, as described in Figure 1, and incorporation of 35S-methionine into protein was measured as described in the Experimental Procedures. Data are displayed as the mean percent inhibition ⫾ SEM for multiple experiments of cells expressing TOP compared to a control protein (␤gal or ORP150). (C) COS-Kb cells were transfected with pTracerCMV expressing TOP ⫹ GFP (TOP), ORP150 ⫹ GFP (ORP150), or GFP alone (Vector), together with pJFE14 expressing influenza HA. After 48 hr the cells were stained with anti-HA or an irrelevant antibody. The Bkg traces are cells stained with an irrelevant antibody. In this same experiment, TOP inhibited the presentation of SIINFEKL (data not shown).

ited the presentation of the influenza-derived peptide ASNENMETM on H-2Db class I molecules (Figure 1D), as well as the presentation on H-2Kb of a subdominant epitope from ovalbumin, KVVRFDKL (Figure 1E). Thus TOP can reduce MHC class I presentation of at least three antigenic peptides expressed from minigenes. TOP-overexpressing cells showed no reduction in viability, as judged by their exclusion of trypan blue (Figure 2A) and their normal rate of protein synthesis, as assessed by incorporation of 35S-methionine/cysteine into proteins (Figure 2B). Furthermore, when TOP was overexpressed with influenza hemagglutinin (HA), the amount of HA expressed on the cell surface was not decreased below control levels (Figure 2C). Therefore, TOP overexpression is not toxic and can markedly decrease antigen presentation without interfering with plasmid transfection, expression of transfected genes, protein synthesis, protein trafficking, or exocytosis. Effect of TOP Overexpression on the Presentation of Peptides Generated from Full-Length Proteins Most MHC class I-presented peptides or their precursors are generated by proteasomes during breakdown of intracellular proteins (Cascio et al., 2001), and it is possible that proteasome-derived peptides are delivered directly to the peptide transporter TAP on the ER membrane (Hirsch and Ploegh, 2000), that components of the proteasome complex protect antigenic peptides from cytosolic peptidases, or that the proteasome-generated peptides might be preferentially bound by chaperones or other proteins that target them to the TAP transporter. We therefore examined whether the presentation of SIINFEKL and KVVRFDKL, which were generated by proteasomes from whole ovalbumin, were affected by the expression of TOP. TOP overexpression markedly inhibited the presentation of both peptides from ovalbumin in E36-Kb cells (Figures 3A and 3B) and also reduced the presentation of ASNENMETM from fulllength influenza nucleoprotein (Figure 3C), which also requires proteasomes (Mo et al., 1999). When TOP was

coexpressed in COS-Kb cells with a plasmid that expresses both ovalbumin and GFP or with SIINFEKL minigene and GFP (data not shown), TOP overexpression reduced the generation of SIINFEKL-Kb complexes (Figure 3D) but did not decrease the expression of GFP (Figure 3E), further demonstrating that TOP overexpression specifically affected MHC antigen presentation rather than transfection efficiency or protein expression. Thus, cytosolic TOP can inhibit the presentation of multiple antigenic peptides that are generated by proteasomes from full-length proteins. We have also compared the effects of overexpressing TOP to those of overexpressing another cytosolic peptidase. In cytosolic extracts, puromycin-sensitive aminopeptidase (PSA) was found to destroy certain antigenic peptides (though not SIINFEKL or several of its N-extended precursors) (Saric et al., 2001), and PSA was reported to trim certain N-extended precursors to the proper size for presentation (Stoltze et al., 2000). COS-Kb cells were transfected with ovalbumin together with a vector encoding PSA and GFP, and SIINFEKL-Kb complexes were then quantified using the 25.D1.16 antibody. PSA transfection resulted in increased expression of PSA mRNA (data not shown) but did not reduce the presentation of SIINFEKL-Kb complexes on the cell surface (Figure 3F), which is consistent with findings in cytosolic extracts (Saric et al., 2001). Similar results were obtained when PSA was cotransfected with the SIINFEKL minigene instead of full-length ovalbumin (Figure 3G). Therefore, overexpression of any cytosolic peptidase does not inevitably reduce antigen expression, but instead the effects depend on the particular enzyme (and perhaps the antigenic peptide). Effect of TOP on the Presentation of Peptides in the ER TOP is a cytosolic enzyme and is not detected in the ER even after transfection (Saric et al., 2002; data not shown). To test if the inhibition of antigen presentation was in fact due to rapid hydrolysis of antigenic peptides

Immunity 432

Figure 3. TOP Transfection Inhibits MHC Class I Presentation of Peptides from Full-Length Antigen Constructs (A) Similar to Figure 1C except cells were transfected with pCDNA3 encoding a full-length ovalbumin cDNA instead of the SIINFEKL minigene. These data come from the same experiment as in Figure 1C. Although it has previously been reported that higher numbers of peptide-MHC complexes are generated from minigenes than whole protein antigens (Anton et al., 1997), we have consistently failed to observe this phenomenon in our system, possibly because of the longer times of antigen expression or different APCs. (B) Similar to Figure 1E except cells were transfected for 2 hr with full-length ovalbumin cDNA (5 ␮g) instead of SIINFEKL. These data come from the same experiment as in Figure 1E. (C) Similar to Figure 1D except cells were coinfected for 6 hr with vaccinia recombinants encoding TOP or ␤gal and full-length influenza nucleoprotein (gift from J. Yewdell, NIH) instead of ASNENMETM. (D) Similar to Figure 1B except that ovalbumin was expressed instead of SIINFEKL, and the plasmid and inserts were reversed (ovalbumin was expressed from pTracerCMV, and TOP, ORP150, or no insert was expressed from pcDNA). Expressing ovalbumin from pcDNA, and TOP, ORP150, or no insert from pTracer, gave essentially identical effects on SIINFEKL-H-2Kb generation. MFI were Bkg, 3.84; Vector, 90.3; ORP150, 107; and TOP, 28.2. (E) The same experiment as Figure 3D, showing the GFP fluorescence of the total population. With pTracerCMV, about half the transfected cells express high levels of GFP. Transfection efficiency, about 30%, was low in this experiment. The curves for TOP, ORP150, and Vector are superimposed on one another. (F) COS-Kb cells were cotransfected with full-length ovalbumin in pCDNA3, and a pTracer CMV plasmid encoding GFP (Vector), or GFP ⫹ human puromycin-sensitive aminopeptidase (PSA), and after 48 hr stained for SIINFEKL⫹Kb complexes with 25.D1.16 and analyzed by flow cytometry. In this experiment pTracerCMV-TOP was also tested and inhibited antigen presentation by approximately 80%, as expected (data not shown). MFI were Bkg, 6.2; PSA, 527; and Vector, 405 fluorescent units. (G) Similar to Figure 3F except SIINFEKL minigene was expressed from the pcDNA3 vector. MFI were Bkg, 5.61; Vector, 233; PSA, 454 fluorescent units.

in the cytosol, we examined the effect of TOP transfection on the presentation of an antigenic peptide that was delivered directly into the ER, thus avoiding exposure to cytosolic enzymes. A SIINFEKL minigene was expressed in COS-Kb cells with an N-terminal signal sequence, which causes the SIINFEKL peptide to be cotranslationally

transported into the ER (Anderson et al., 1991; Craiu et al., 1997). The leader sequence then is removed by the signal sequence peptidase acting by itself, or together with the ER-resident aminopeptidase ERAP1 (Saric et al., 2002; Serwold et al., 2002; York et al., 2002). Overexpression of TOP did not inhibit the presentation of ER-

Endopeptidase Limits Antigen Presentation 433

Figure 4. Overexpression of TOP Does Not Reduce the Presentation of Peptides Generated in the ER or in Endosomes (A) Similar to Figure 1B except cells were transfected with a SIINFEKL minigene with an N-terminal signal sequence instead of the unmodified SIINFEKL minigene. These data come from the same experiment as in Figure 1B. MFI were Bkg, 15.1; No OVA, 17; Vector, 1337; ORP150, 1395; and TOP ⫹ ER-targeted SIINFEKL, 1301 fluorescent units. (B) I-Ad ⫹ p41 invariant chain-expressing DAP L cells were infected with vaccinia-TOP (TOP) or vaccinia ␤gal (␤gal), incubated with the indicated concentrations of ovalbumin for 4 hr, and then fixed with paraformaldehyde. The presence of ovalbumin ⫹ I-Ad complexes was detected by measuring the production of IL-2 from the DO11.10 hybridoma (Shimonkevitz et al., 1983). (C) Similar to Figure 1B except DAP-Kb cells were used instead of E36-Kb and the incubation with antigen was 4 hr instead of 1.5 hr.

targeted SIINFEKL (Figure 4A), even though it markedly reduced the presentation of SIINFEKL generated in the cytoplasm (Figures 1B and 1C). Thus, TOP transfection specifically inhibits the presentation of cytoplasmically derived peptides and does not nonspecifically affect other steps needed for class I antigen presentation, such as MHC class I synthesis, assembly, or exocytosis (see also Figures 4A, 5B, and 5E, described below). Effect of TOP on the Presentation on MHC Class II Molecules of Peptides Generated in Endosomal Compartments We also tested the effect of TOP on MHC class II antigen presentation. In this pathway, peptides are generated by proteolysis in the endocytic compartment (Germain and Margulies, 1993), which does not contain TOP. L cells (DAP) expressing the MHC class II molecule I-Ad, and the p41 isoform of invariant chain (Peterson and Miller, 1992), were both infected with vaccinia recombinants expressing TOP or ␤-galactosidase and incubated with different concentrations of ovalbumin added to the medium. It has previously been shown that transfected L cells internalize exogenous ovalbumin, degrade it in endocytic compartments, and present ovalbumin-derived peptides on the transfected class II molecules (Diment, 1990; A. Sant, personal communication). Overexpression of TOP had little or no effect on the presentation to a specific T cell hybridoma of a class II-presented peptides generated from ovalbumin (Figure 4B), even though in L cells it strongly inhibited the presentation on MHC class I molecules of SIINFEKL from full-length ovalbumin or a minigene (Figure 4C). Effect of Overexpressing TOP on the Levels of Surface MHC Class I Molecules Newly synthesized MHC class I molecules are retained in the ER until they bind peptides (York and Rock, 1996).

Consequently, conditions that limit the supply of peptides to MHC class I molecules (e.g., inhibition of the proteasome [Rock et al., 1994] or of the TAP transporter [Townsend and Trowsdale, 1993])) reduce MHC appearance on the cell surface. If TOP is destroying large numbers of antigenic peptides, there should be a reduction in MHC class I molecules on the cell surface. We therefore analyzed the effect of TOP transfection on the total expression of MHC class I molecules on the cell surface. We cotransfected into COS7 cells cDNAs for an MHC class I molecule (either H-2Kb or HLA-A3) together with TOP or a control vector, and after 48 hr measured the expression of the MHC class I molecule on the transfected cells. Overexpression of TOP caused a marked reduction in the levels of H-2Kb (Figure 5A) and HLA-A3 (Figure 5C) molecules on the cell surface. To confirm that this decrease in surface MHC class I molecules was due to a reduction in the supply of antigenic peptides, we coexpressed in these cells the ER-targeted SIINFEKL construct used in Figure 4A, which bypasses TOP in the cytoplasm. In this situation, in which peptide supply to the SIINFEKL binding H-2Kb is restored, TOP overexpression no longer reduces the levels of H-2Kb on the cell surface (Figure 5B), whereas the levels of HLA-A3 (which does not bind SIINFEKL) remain low (Figure 5D). These results indicate that most MHC-presented peptides are susceptible to digestion by cytoplasmic peptidases and again confirm that high intracellular levels of TOP do not cause any general defects in MHC class I assembly or transport. In these experiments, MHC class I expression is reduced on the cell surface presumably because TOP limits the supply of peptides needed to stabilize MHC class I molecules and allow their movement from the ER to the cell surface. To further test this point, we investigated whether TOP overexpression reduced the

Immunity 434

Figure 5. TOP Inhibits MHC Class I Assembly and Transport (A) COS7 cells were cotransfected with H-2Kb in pcDNA3 and a pTracerCMV plasmid encoding GFP (Vector), OPR150 ⫹ GFP (ORP150), or TOP ⫹ GFP (TOP). The H-2Kb transfection was omitted in the background group (Bkg). After 48 hr the cells were stained with anti-H-2Kb (Y3) and analyzed as described in Figure 1. MFI were Bkg, 22.7; Vector, 1922; ORP150, 1948; and TOP, 708 fluorescent units. (B) Same as (A) except cells were also transfected with an ER-targeted signal sequence-SIINFEKL minigene. MFI were Bkg, 25.7; Vector, 1922; and TOP, 2151 fluorescent units. (C) Similar to (A) except cells were transfected with pJFE14-HLA-A3 instead of H-2Kb and were stained with anti-HLA-A3 instead of anti-H2Kb. MFI were Bkg, 25.6; TOP, 89.9; ORP150, 211; and Vector, 237 fluorescent units. (D) Same as (C) except cells were also transfected with a signal sequence-SIINFEKL minigene. MFI were Bkg, 25.6; TOP, 93.4; and vector, 244 fluorescent units. All panels are from the same experiment. (E) DAP-Kb cells infected with vaccinia-TOP (TOP), vaccinia-␤gal (␤gal), or uninfected (none) were labeled with 35S-methionine and their MHC class I molecules were immunoprecipitated from detergent lysates with an antibody that detects all free heavy chain (leftmost three lanes), or with antibodies that detect only the indicated assembled MHC class I complexes (remaining lanes). (F) The autoradiographs shown in (E) were scanned and densitometric analysis was performed.

assembly of stable MHC class I molecules. DAP-Kb cells infected with recombinant vaccinia virus expressing TOP were also found to have a marked reduction in their surface expression of MHC class I molecules (data not shown). These cells were infected with vaccinia virus expressing TOP, or with vaccinia virus expressing ␤-galactosidase, and were then labeled with 35S-cysteine and methionine. The extent of assembly of the labeled MHC class I molecules was investigated by immunoprecipitation using antibodies specific for assembled MHC class I molecules or the component heavy and light chains.

Compared to cells infected with control vaccinia virus, TOP overexpression minimally reduced the synthesis of MHC class I heavy chains (Figure 5E, leftmost three lanes, and Figure 5F) or light chains (data not shown), but it did cause a marked reduction in assembled class I H-2Dk, H-2Kk, and H-2Kb heterodimers (Figure 5E and 5F). The extent of the reduction in assembly varied for different alleles, ranging from approximately 35% of control assembly for H-2Dk to about 55% of control for H-2Kk. In other words, TOP overexpression (without affecting the synthesis of MHC class I heavy or light

Endopeptidase Limits Antigen Presentation 435

chains) reduces the assembly of stable MHC class I molecules, presumably by causing the destruction of many antigenic peptides in the cytosol. The difference in the magnitude of the effect between alleles may reflect differing susceptibility of the peptides that bind to the particular alleles. Effect of Preventing TOP Expression on MHC Class I Presentation These results indicate that increasing the cellular content of TOP by destroying antigenic peptides in the cytosol reduces MHC class I presentation. In HeLa cell extracts, the endogenous TOP is the peptidase primarily responsible for the rapid destruction of many synthetic antigenic peptides (Saric et al., 2001), and 9–17-residue proteasome products (T.S. et al., unpublished data). To determine whether this enzyme, at the levels found normally in cells, also hydrolyzes antigenic peptides and thus limits antigen presentation in vivo, we tested in Hela-Kb cells whether selective elimination of TOP with small interfering RNA (siRNA) oligonucleotides (Elbashir et al., 2001) would increase antigen presentation. As a control, we used siRNA directed against mouse TOP whose sequence in this region is different from that of human TOP. Transfection of this control siRNA had no effect on expression of human TOP. However, when human TOP siRNA was transfected into HeLa-Kb cells, TOP mRNA was reduced by 85%–90% compared to control siRNA-treated cells (Figure 6A) and TOP protein was reduced at least by 90% (the limit of sensitivity of the antibody), while several other assayed mRNAs (e.g., ␤-actin and ERAP1 [data not shown]) and proteins (such as calreticulin [Figure 6B]) were not affected. The transfection efficiency of siRNA was estimated at 95% (data not shown), so that even if TOP was completely eliminated from transfected cells, 5% of control TOP levels would still be present in protein or RNA preparations. HeLa-Kb cells were initially transfected with control or TOP siRNA and 3 days later were transfected with a plasmid coexpressing a SIINFEKL minigene and GFP. Quantification of SIINFEKL-Kb complexes on transfected cells revealed that reducing TOP expression enhanced the generation of these complexes nearly 2-fold (Figure 6C). Similarly, the presentation of SIINFEKL was approximately doubled in TOP siRNA-treated cells transfected with full-length ovalbumin (Figure 6D). These results, which are opposite to those seen when TOP expression is increased (Figure 1), indicate that under physiological conditions TOP normally limits the presentation of SIINFEKL. Inhibition of TOP did not affect the expression of transfected influenza hemagglutinin (Figure 6E), showing that the effect on SIINFEKL presentation was not due to general effects on protein expression. Effect of Preventing TOP Expression on MHC Class I Molecule Assembly and Surface Expression If TOP at physiological levels constantly destroys antigenic peptides or their longer precursors, then eliminating TOP should increase the supply of antigenic peptides, which should in turn result in increased assembly and surface expression of MHC class I molecules. To test this hypothesis, we treated HeLa-Kb cells with con-

trol or TOP siRNA and examined the synthesis and assembly of MHC class I molecules in a pulse-chase experiment. Although the siRNA did not affect the synthesis of MHC class I heavy chain (Figure 7A, upper panel), TOP siRNA increased the assembly of stable complexes approximately 2-fold (Figure 7A, lower panel, and Figure 7B). Elimination of TOP from HeLa-Kb cells with siRNA increased the levels on the cell surface of mouse H-2Kb (Figure 7C) by about 80%, and of human HLA-A, B, and C molecules (Figure 7D) about 2-fold, as measured by flow cytometry. This increase in surface expression correlated well with the 2-fold increase in MHC class I assembly measured in immunoprecipitations (Figures 7A and 7B). Again, this effect was precisely opposite to that observed when TOP was overexpressed in cells (Figure 5). Therefore, under physiological conditions TOP must destroy many antigenic peptides and generally limit antigen presentation. Discussion Together, the present findings demonstrate that TOP limits MHC class I antigen presentation by destroying peptides in the cytoplasm. When overexpressed, TOP inhibited only the presentation of peptides generated in the cytosol and did not reduce the MHC class I presentation of a peptide delivered directly into the ER or the presentation on MHC class II molecules of a peptide generated in endocytic compartments. Also, TOP overexpression did not reduce overall protein synthesis, the production of MHC heavy or light chains, or the appearance on the cell surface of an unrelated membrane protein, influenza hemagglutinin. Therefore, TOP’s effects are highly selective and are not due to some nonspecific or toxic effect on the cells. Increasing the cellular content of TOP inhibited the presentation of three well-defined antigenic peptides. It also reduced generally the assembly and surface expression of many class I molecules, and this was due to a reduction in the supply of antigenic peptides because it was reversed when a MHC class I binding peptide was targeted by a signal sequence into the ER. It is very unlikely that TOP reduces peptide supply to the ER by inhibiting the function of the TAP transporter directly because pure TOP only hydrolyzes oligopeptides and not proteins (Barrett et al., 1995). Therefore, the inhibition of antigen presentation when TOP levels rise is almost certainly due to accelerated degradation of proteasome products in the cytosol, which is consistent with the finding that TOP rapidly degrades many peptides in cell extracts (Saric et al., 2001). Moreover, rapid destruction of antigenic peptides by TOP must occur normally in cells, since eliminating TOP with siRNA, or by introducing selective inhibitors of TOP enzyme (data not shown), enhanced the assembly and transport to the cell surface of multiple MHC class I molecules, as well as the presentation of a specific antigenic peptide. Elimination of TOP with siRNA did not affect cell growth or viability (data not shown), endogenous proteins were synthesized at normal rates (e.g., MHC class I heavy chain: Figure 7A, upper panel), and expression of a transfected protein (influenza hemagglutinin: Figure 6E)

Immunity 436

Figure 6. Loss of TOP Expression Enhances the Presentation of SIINFEKL on H-2Kb (A) Four days after treatment with siRNA directed against human TOP (hTOP1; hTOP1 ⫹ 3) or with control siRNA (mTOP1: directed against mouse TOP in a region not homologous to human TOP), mRNA was extracted from HeLa-Kb cells and cDNA was prepared. Real-time PCR for human TOP, and for human ␤-actin, was performed on the cDNA and on plasmids containing the appropriate genes for standardization. The copy number of human TOP was normalized to ␤-actin expression. Shown is the average of three experiments ⫹/⫺ standard deviation. (B) Hela-Kb cells were treated with siRNA as in (A) and detergent lysates were prepared. 3-fold serial dilutions of the lysates were immunoblotted with antibodies specific for TOP (upper panel) or calreticulin (lower panel). (C) HeLa-Kb cells were transfected with the siRNA hTOP1, or with mTOP1 as a control. Three days later, the cells were transfected with pTracerSR␣ expressing the minigene SIINFEKL as well as GFP. After an additional day the cells were stained for SIINFEKL ⫹ Kb complexes (25.D1.16), and GFP-positive cells were analyzed by flow cytometry as described in Figure 1. No OVA: cells transfected with pTracerSR␣ vector only and stained with 25.D1.16. Background fluorescence (based on staining with an irrelevant antibody) was very similar to that of the No OVA group. MFI were Bkg, 12.2; TOP siRNA, 3826; and Ctrl siRNA, 2167 fluorescent units. (D) Similar to (A) except that cells were transfected with pTracerSR␣ expressing ovalbumin. MFI were Bkg, 38.8; hTOP, 2549, and Ctrl siRNA, 1647 fluorescent units. (E) HeLa-Kb cells were transfected with the siRNA hTOP1, or with mTOP1 as a control. Two days later, the cells were transfected with pTracerCMV (Bkg), or with pTracerCMV expressing influenza hemagglutinin. After an additional 2 days the cells were stained for hemagglutinin and analyzed by flow cytometry. MFI were Bkg, 6.01; Ctrl siRNA, 5981; and hTOP siRNA, 5848 fluorescent units.

was unaffected by TOP inhibition. Although TOP normally limits the availability of most peptides for antigen presentation, other still-unidentified cytosolic endopeptidases and aminopeptidases can hydrolyze antigenic peptides in cell extracts (Saric et al., 2001) (T.S. et al., unpublished data) and presumably in vivo, and may also help limit the efficiency of antigen presentation. Thus, antigenic peptides (and their longer precursors) appear to resemble the great majority of peptides released from proteasomes in that they are subject to further rapid degradation by cytosolic peptidases.

Although TOP has been known and characterized enzymologically for many years and is present in virtually all mammalian cells (Carvalho and Camargo, 1981; Orlowski et al., 1983; Camargo et al., 1997), its physiological role(s) was unknown. Our data and related biochemical studies on cell extracts (Saric et al., 2001) (T.S. et al., unpublished data) clearly demonstrate that this endopeptidase is responsible for hydrolyzing oligopeptides in the cytosol of cells. In the final steps in the pathway for degradation of proteins, TOP appears to cleave proteasome products of the sizes of antigenic

Endopeptidase Limits Antigen Presentation 437

Figure 7. Loss of TOP Expression Enhances Assembly and Surface Expression of Class I Molecules (A) HeLa-Kb cells were transfected with the siRNAs hTOP1 and hTOP3, or with mTOP1 as a control. Three days later, the cells were radiolabeled for 10 min and chased for 0, 15, or 45 min before lysis and immunoprecipitated with antibodies specific for unfolded heavy chain, or HLA-A, B, and C assembled with ␤2-m, as described in Experimental Procedures. (B) The autoradiographs from (A) were scanned and analyzed by densitometry. (C) Hela-Kb cells were transfected with the siRNAs hTOP1 and hTOP3, or with mTOP1 as a control. Four days later, the cells were stained with anti-H-2Kb (Y3) and phycoerythrin-anti-Ig antibodies and analyzed by flow cytometry as described in Figure 1. Bkg, background fluorescence, based on staining with an irrelevant antibody. MFI were Bkg, 4; TOP siRNA, 189; and Ctrl siRNA, 102 fluorescent units. (C) Similar to (B) except that cells were stained with anti-HLA-A, B, C (PA2.6). MFI were Bkg, 4; TOP siRNA, 413; and Ctrl siRNA, 205 fluorescent units.

peptides or their precursors (9–17 residues) to pieces, most of which are rapidly hydrolyzed by exopeptidases to amino acids. An important implication of our findings is that the extent to which any particular peptide is presented on MHC class I molecules is a function both of its rate of production and cytosolic destruction. A similar interplay between peptide generation and destruction occurs in the endocytic compartment for MHC class II presentation (Vidard et al., 1991). Therefore, blocking TOP or mutating sequences of antigens so that they become resistant to degradation by peptidases might be a useful approach for enhancing antigen presentation and immune responses to vaccines. On the other hand, it is also possible that overproduction of TOP or other cytosolic peptidases may be a mechanism by which some cancers or viruses evade immune destruction, or may even represent a novel approach for suppressing immune response in transplanted cells or organs. These findings argue that antigenic peptides and their precursors are not protected by molecular chaperones from intracellular degradation in the cytosol or that if such protective mechanisms do exist, they are not very effective. How then do any antigenic peptides escape destruction? In fact most probably don’t and are destroyed before binding to MHC molecules. Thus after release from proteasomes, antigenic precursors confront a kinetic competition between being cleaved by

cytosolic peptidases, such as TOP, and uptake into the ER. It remains possible that there is some protection by cytosolic binding proteins (e.g., heat shock proteins) (Srivastava et al., 1998). If so TOP, by destroying the free peptide, may cause dissociation of bound peptides and then further hydrolysis. The present findings and our related biochemical studies (Saric et al., 2001) clearly do not support the earlier report that TOP strongly binds antigenic peptides but does not efficiently degrade them (Portaro et al., 1999). On this basis, it was even proposed that TOP actually functions to facilitate peptide delivery to TAP (Portaro et al., 1999; Silva et al., 1999). The reasons for these discrepant earlier observations are still not clear (for a critical discussion, see Saric et al., 2001). While the destruction of antigenic peptides by TOP and presumably other cytosolic peptidases can reduce MHC class I presentation and thus may appear to be deleterious, this degradative process is probably an unavoidable consequence of the immune system utilizing a phylogenetically older catabolic pathway whose principal function in all cells is to break down proteins into amino acids. On the other hand, for efficient generation of antigenic peptides, it may also be important that proteasome products are free in the cytosol, since some longer precursors to antigenic peptides have to be trimmed by aminopeptidases to the MHC class I-presented epitopes (Beninga et al., 1998; Craiu et al., 1997; Mo et al., 1999; Stoltze et al., 2000). Therefore, to the

Immunity 438

extent that this N-terminal trimming occurs in the cytoplasm, peptides need to be accessible to peptidases. In addition, this rapid destruction of antigenic peptides by cytosolic peptidases may be an important factor in the regulation of immune responses. Under normal conditions most cells express very low levels of MHC class I, presumably because high constitutive expression of these molecules might increase the danger of autoimmunity. In the absence of infection, peptide destruction will be one of the factors that helps to limit MHC class I antigen presentation. However, in the presence of inflammatory mediators, especially interferon-␥, MHC class I presentation is enhanced, in part by inducing new proteasome subunits and the proteasome activator, PA28 (REG). The immunoproteasomes and complexes containing PA28 tend to produce more N-extended versions of antigenic peptides than “non-immune” proteasomes (Cascio et al., 2002). Because these precursor peptides are present in greater amounts and tend to be degraded more slowly by TOP and other cytosolic peptidases than the mature epitopes, they are more likely to reach the ER and after trimming be presented on MHC class I molecules. (Goldberg et al., 2002; Saric et al., 2001). Experimental Procedures Cells COS-Kb cells, HeLa-Kb, DAP-Kb, E36-Kb, and E36-Db cells were previously described (Michalek et al., 1993; Craiu et al., 1997; Mo et al., 1999; York et al., 2002). DAP (L) cells expressing I-Ad and invariant chain were a gift from Jim Miller (University of Rochester). Transfections and Infections COS7 or COS-Kb cells were transiently cotransfected with pTracerCMV (Invitrogen) or pTracerSR␣ (pTracerSV40 [Invitrogen] with an SR␣ promoter [Takebe et al., 1988]) and pCDNA3 (Invitrogen) or pJFE14 (Elliott et al., 1990) (1–2 ␮g of each plasmid) expressing enzymes, antigens, or MHC class I molecules, using FuGene6 (Roche, Indianapolis, IN) according to the manufacturer’s directions, and incubated for 24–48 hr. E36-Kb, E36-Db, or DAP-Kb APCs were infected with either vaccinia-TOP or, as a control, vaccinia-␤gal (MOI 10) alone or in some experiments together with vaccinia-T7 polymerase (MOI 10). Under these conditions, 99 ⫾ 0.28% of the cells were infected with the vaccinia vectors. After 1.5–2 hr, antigen was introduced by infection with a vaccinia recombinant. Alternatively, in the cells infected with vaccinia-T7, antigen was introduced by transfection with an antigen cDNA under the control of a T7 promoter in a pBluescript plasmid (Stratagene, La Jolla, CA) (1 ␮g) using Lipofectin liposomes (Invitrogen) as previously described (Craiu et al., 1997); in this system the level of antigen expressed in replicate groups transfected with a pBluescript antigen plasmid is almost identical (SD for the fluorescence intensity varying by less than 5% of the mean) and is proportional to the amount of plasmid transfected (Craiu et al., 1997). siRNA RNA oligonucleotides hTOP1 (directed to positions 645-663 of human TOP [accession number XM_055686]), hTOP3 (directed to positions 819-837 of human TOP), and mTOP1 (directed to positions 673-691 of mouseTOP [GenBank Accession XM_122062]) were purchased from Dharmacon (Lafeyette, CO), and annealed as described (Elbashir et al., 2001). Hela-Kb cells were transfected as described (York et al., 2002). In some experiments, hTOP1 and hTOP3 oligonucleotides were cotransfected. Immunofluorescence, Immunoprecipitation, and Western Blotting Transfected cells were stained with monoclonal antibodies as previously described (York et al., 2002). Metabolic labeling, immuno-

precipitation, and Western blotting were performed as previously described (York et al., 2002). Viability and Protein Synthesis Assays Cell viability was determined by trypan blue exclusion. Protein synthesis was measured by the incorporation of 35S-methionine into TCA insoluble fractions as previously described (Rock et al., 1994). Antigen Presentation Assays Transfected and/or infected E36-Kb, E36-Db, or DAP-Kb APCs were fixed with 1% paraformaldehyde and incubated with antigen-specific T cell hybridomas as described (Craiu et al., 1997). After 18 hr, the production of IL-2 from the hybridomas was measured using CTLL2 cells as described (Craiu et al., 1997). These APCs were chosen because they present well to the murine T-T hybridomas and are highly infectable with vaccinia virus. SIINFEKL⫹Kb complexes were quantified on COS-Kb and Hela-Kb cells by immunofluorescence using a specific mAb (25.D1.16) (Porgador et al., 1997) because these xenogenic cells present poorly to our murine T-T hybridomas but worked well in this staining assay. Antibodies The monoclonal antibody 15.5.5 recognizes ␤2-microglobulin (␤2m)-associated H-2Dk (Ozato et al., 1980); 16.1.1N recognizes ␤2m-associated H-2Kk (Ozato et al., 1980); and Y3 recognizes ␤2-massociated H-2Kb (Jones and Janeway, 1981). A rabbit antiserum that recognizes unfolded mouse heavy chain was kindly provided by Stan Nathenson (Albert Einstein College of Medicine). The monoclonal antibody W6/32 (Parham et al., 1979) recognizes HLA-A, B, and C alleles only in association with ␤2-m. HC10 (Stam et al., 1986) recognizes unfolded HLA-B and C. Anti-TOP antiserum and the antiTOP monoclonal antibody IVD6 have been previously described (Conn et al., 1996). Real-Time PCR mRNA was extracted from cells with an RNeasy kit (Qiagen, Valencia, CA). cDNA was prepared using SuperScript II (Invitrogen). Realtime PCR using Taq (Invitrogen) and an iCycler (BioRad, Hercules, CA) was performed using primers directed against human TOP (ACTTCCCCCTCCTGAAGAAA and CCTTGAGGATAGCGCAGTTC) or ␤-actin (CGAGGCCCAGAGCAAGAGAG and CGGTTGGCCTT AGGGTTCAG). Enzyme Assays Enzymatic activities in cell extracts were measured as described (Saric et al., 2001). The substrates used were, for TOP, Mcc-ProLeu-Gly-Pro-D-Lys-Dnp-OH; for AP, Leu-Amc; for POP, Z-Gly-ProAmc; and for TPP-II, Ala-Ala-Phe-Amc. Acknowledgments This work was supported by grants from the NIH to K.L.R. and A.L.G. A.X.Y.M was supported by a training grant from the NIH. We thank Dr. Shih-Chung Chang (Harvard Medical School) for performing assays measuring TOP activity, and Drs. Lou Hersh and Michael Thompson (University of Kentucky) for PSA cDNA. We are grateful to Sarah Trombley for expert assistance in preparation of this manuscript and to Jennifer Keyes and Janice Favreau for technical assistance. Received: July 12, 2001 Revised: December 18, 2002 References Anderson, K., Cresswell, P., Gammon, M., Hermes, J., Williamson, A., and Zweerink, H. (1991). Endogenously synthesized peptide with an endoplasmic reticulum signal sequence sensitizes antigen processing mutant cells to class I-restricted cell-mediated lysis. J. Exp. Med. 174, 489–492. Anton, L., Yewdell, J., and Bennink, J. (1997). MHC class I-associated peptides produced from endogenous gene products with vastly different efficiencies. J. Immunol. 158, 2535–2542.

Endopeptidase Limits Antigen Presentation 439

Barrett, A.J., Brown, M.A., Dando, P.M., Knight, C.G., McKie, N., Rawlings, N.D., and Serizawa, A. (1995). Thimet oligopeptidase and oligopeptidase M or neurolysin. Methods Enzymol. 248, 529–556. Beninga, J., Rock, K.L., and Goldberg, A.L. (1998). Interferon-gamma can stimulate post-proteasomal trimming of the N terminus of an antigenic peptide by inducing leucine aminopeptidase. J. Biol. Chem. 273, 18734–18742. Buchler, M., Tisljar, U., and Wolf, D.H. (1994). Proteinase yscD (oligopeptidase yscD). Structure, function and relationship of the yeast enzyme with mammalian thimet oligopeptidase (metalloendopeptidase, EP 24.15). Eur. J. Biochem. 219, 627–639. Camargo, A.C., Gomes, M.D., Reichl, A.P., Ferro, E.S., Jacchieri, S., Hirata, I.Y., and Juliano, L. (1997). Structural features that make oligopeptides susceptible substrates for hydrolysis by recombinant thimet oligopeptidase. Biochem. J. 324, 517–522. Carvalho, K.M., and Camargo, A.C.M. (1981). Purification of rabbit brain endooligopeptidases and preparation of anti-enzyme antibodies. Biochemistry 20, 7082–7088. Cascio, P., Hilton, C., Kisselev, A., Rock, K., and Goldberg, A. (2001). 26S proteasomes and immunoproteasomes produce mainly N-extended versions of a MHC class I presented peptide. EMBO J. 20, 2357–2366. Cascio, P., Call, M., Petre, B., Walz, T., and Goldberg, A.L. (2002). Properties of the hybrid form of the 26S proteasome containing both 19S and PA28 complexes. EMBO J. 21, 2636–2645. Cole, G.A., Hogg, T.L., and Woodland, D.L. (1994). The MHC class I-restricted T cell response to Sendai virus infection in C57BL/6 mice: a single immunodominant epitope elicits an extremely diverse repertoire of T cells. Int. Immunol. 6, 1767–1775. Conn, K.J., Pietropaolo, M., Ju, S.T., and Abraham, C.R. (1996). Monoclonal antibodies against the human metalloprotease EC 3.4.24.15 label neurofibrillary tangles in Alzheimer’s disease brain. J. Neurochem. 66, 2011–2018. Craiu, A., Akopian, T., Goldberg, A., and Rock, K.L. (1997). Two distinct proteolytic processes in the generation of an MHC class I-presented peptide. Proc. Natl. Acad. Sci. USA 94, 10850–10855. Deckhut, A.M., Allan, W., McMickle, A., Eichelberger, M., Blackman, M.A., Doherty, P.C., and Woodland, D.L. (1993). Prominent usage of V beta 8.3 T cells in the H-2Db-restricted response to an influenza A virus nucleoprotein epitope. J. Immunol. 151, 2658–2666. Diment, S. (1990). Different roles for thiol and aspartyl proteases in antigen presentation of ovalbumin. J. Immunol. 145, 417–422.

(1999). Distinct proteolytic processes generate the C and N termini of MHC class I-binding peptides. J. Immunol. 163, 5851–5859. Orlowski, M., Michaud, C., and Chu, T.G. (1983). A soluble metalloendopeptidase from rat brain. Purification of the enzyme and determination of specificity with synthetic and natural peptides. Eur. J. Biochem. 135, 81–88. Ozato, K., Mayer, N., and Sachs, D.H. (1980). Hybridoma cell lines secreting monoclonal antibodies to mouse H-2 and Ia antigens. J. Immunol. 124, 533–540. Parham, P., Barnstable, C.J., and Bodmer, W.F. (1979). Use of a monoclonal antibody (W6/32) in structural studies of HLA-A,B,C, antigens. J. Immunol. 123, 342–349. Paz, P., Brouwenstijn, N., Perry, R., and Shastri, N. (1999). Discrete proteolytic intermediates in the MHC class I antigen processing pathway and MHC I-dependent peptide trimming in the ER. Immunity 11, 241–251. Peterson, M., and Miller, J. (1992). Antigen presentation enhanced by the alternatively spliced invariant chain gene product p41. Nature 357, 596–598. Porgador, A., Yewdell, J., Deng, Y., Bennink, J., and Germain, R. (1997). Localization, quantitation, and in situ detection of specific peptide-MHC class I complexes using a monoclonal antibody. Immunity 6, 715–726. Portaro, F.C., Gomes, M.D., Cabrera, A., Fernandes, B.L., Silva, C.L., Ferro, E.S., Juliano, L., and de Camargo, A.C. (1999). Thimet oligopeptidase and the stability of MHC class I epitopes in macrophage cytosol. Biochem. Biophys. Res. Commun. 255, 596–601. Rock, K.L., and Goldberg, A.L. (1999). Degradation of cell proteins and the generation of MHC class I- presented peptides. Annu. Rev. Immunol. 17, 739–779. Rock, K.L., Gramm, C., Rothstein, L., Clark, K., Stein, R., Dick, L., Hwang, D., and Goldberg, A.L. (1994). 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. Rock, K.L., York, I.A., Saric, T., and Goldberg, A.L. (2002). Protein degradation and the generation of MHC class I-presented peptides. Adv. Immunol. 80, 1–70. Saric, T., Beninga, J., Graef, C.I., Akopian, T.N., Rock, K.L., and Goldberg, A.L. (2001). Major histocompatibility complex class I-presented antigenic peptides are degraded in cytosolic extracts primarily by thimet oligopeptidase. J. Biol. Chem. 276, 36474–36481.

Elbashir, S.M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K., and Tuschl, T. (2001). Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411, 494–498.

Saric, T., Chang, S.C., Hattori, A., York, I.A., Markant, S., Rock, K.L., Tsujimoto, M., and Goldberg, A.L. (2002). An IFN-gamma-induced aminopeptidase in the ER, ERAP1, trims precursors to MHC class I-presented peptides. Nat. Immunol. 3, 1169–1176.

Elliott, J.F., Albrecht, G.R., Gilladoga, A., Handunnetti, S.M., Neequaye, J., Lallinger, G., Minjas, J.N., and Howard, R.J. (1990). Genes for Plasmodium falciparum surface antigens cloned by expression in COS cells. Proc. Natl. Acad. Sci. USA 87, 6363–6367.

Serwold, T., Gonzalez, F., Kim, J., Jacob, R., and Shastri, N. (2002). ERAAP customizes peptides for MHC class I molecules in the endoplasmic reticulum. Nature 419, 480–483.

Germain, R.N., and Margulies, D.H. (1993). The biochemistry and cell biology of antigen processing and presentation. Annu. Rev. Immunol. 11, 403–450. Goldberg, A.L., Cascio, P., Saric, T., and Rock, K.L. (2002). The importance of the proteasome and subsequent proteolytic steps in the generation of antigenic peptides. Mol. Immunol. 39, 147–164. Hirsch, C., and Ploegh, H.L. (2000). Intracellular targeting of the proteasome. Trends Cell Biol. 10, 268–272. Jones, B., and Janeway, C.A.J. (1981). Cooperative interaction of B lymphocytes with antigen-specific helper T lymphocytes is MHC restricted. Nature 292, 547–549. Kisselev, A.F., Akopian, T.N., Woo, K.M., and Goldberg, A.L. (1999). The sizes of peptides generated from protein by mammalian 26 and 20 S proteasomes. Implications for understanding the degradative mechanism and antigen presentation. J. Biol. Chem. 274, 3363– 3371. Michalek, M., Grant, E., Gramm, C., Goldberg, A., and Rock, K. (1993). A role for the ubiquitin-dependent proteolytic pathway in MHC class I-restricted antigen presentation. Nature 363, 552–554. Mo, X.Y., Cascio, P., Lemerise, K., Goldberg, A.L., and Rock, K.

Shimonkevitz, R., Kappler, J., Marrack, P., and Grey, H. (1983). Antigen recognition by H-2 restricted T cells. Cell-free antigen processing. J. Exp. Med. 158, 303–316. Silva, C.L., Portaro, F.C., Bonato, V.L., de Camargo, A.C., and Ferro, E.S. (1999). Thimet oligopeptidase (EC 3.4.24.15), a novel protein on the route of MHC class I antigen presentation. Biochem. Biophys. Res. Commun. 255, 591–595. Srivastava, P.K., Menoret, A., Basu, S., Binder, R.J., and McQuade, K.L. (1998). Heat shock proteins come of age: primitive functions acquire new roles in an adaptive world. Immunity 8, 657–665. Stam, N.J., Spits, H., and Ploegh, H.L. (1986). Monoclonal antibodies raised against denatured HLA-B locus heavy chains permit biochemical characterization of certain HLA-C locus products. J. Immunol. 137, 2299–2306. Stoltze, L., Schirel, M., Schwarz, G., Schro¨ter, C., Thompson, M., Hersh, L., Kalbacher, H., Stevaovic, S., Rammensee, H.G., and Schild, H. (2000). Two new proteases in the MHC class I processing pathway. Nat. Immunol. 1, 413–418. Takebe, Y., Seiki, M., Fujisawa, J., Hoy, P., Yokota, K., Arai, K., Yoshida, M., and Arai, N. (1988). SR alpha promoter: an efficient and versatile mammalian cDNA expression system composed of

Immunity 440

the simian virus 40 early promoter and the R-U5 segment of human T-cell leukemia virus type 1 long terminal repeat. Mol. Cell. Biol. 8, 466–472. Townsend, A., and Trowsdale, J. (1993). The transporters associated with antigen presentation. Semin. Cell Biol. 4, 53–61. Vidard, L., Rock, K., and Benacerraf, B. (1991). The generation of immunogenic peptides can be selectively increased or decreased by proteolytic enzyme inhibitors. J. Immunol. 147, 1786–1791. Villanueva, M.S., Fischer, P., Feen, K., and Pamer, E.G. (1994). Efficiency of MHC class I antigen processing: a quantitative analysis. Immunity 1, 479–489. Vincent, B., Jiracek, J., Noble, F., Loog, M., Roques, B., Dive, V., Vincent, J.P., and Checler, F. (1997). Contribution of endopeptidase 3.4.24.15 to central neurotensin inactivation. Eur. J. Pharmacol. 334, 49–53. Yamin, R., Malgeri, E.G., Sloane, J.A., McGraw, W.T., and Abraham, C.R. (1999). Metalloendopeptidase EC 3.4.24.15 is necessary for Alzheimer’s amyloid-beta peptide degradation. J. Biol. Chem. 274, 18777–18784. York, I.A., Chang, S.C., Saric, T., Keys, J.A., Favreau, J.M., Goldberg, A.L., and Rock, K.L. (2002). The ER aminopeptidase ERAP1 enhances or limits antigen presentation by trimming epitopes to 8–9 residues. Nat. Immunol. 3, 1177–1184. York, I., and Rock, K.L. (1996). Antigen processing and presentation by the class I major histocompatibility complex. Annu. Rev. Immunol. 14, 369–396.