Efficient presentation of exogenous antigen by liver endothelial cells to CD8+ T cells results in antigen-specific T-cell tolerance.

Myeloid antigen-presenting cells (APC) are known to cross-present exogenous antigen on major histocompatibility class I molecules to CD8+ T cells and thereby induce protective immunity against infecting microorganisms. Here we report that liver sinusoidal endothelial cells (LSEC) are organ-resident, non-myeloid APC capable of cross-presenting soluble exogenous antigen to CD8+ T cells. Though LSEC employ similar molecular mechanisms for cross-presentation as dendritic cells, the outcome of cross-presentation by LSEC is CD8+ T cell tolerance rather than immunity. As uptake of circulating antigens into LSEC occurs efficiently in vivo, it is likely that cross-presentation by LSEC contributes to CD8+ T cell tolerance observed in situations where soluble antigen is present in the circulation.

CC chemokine receptor (CCR)4 and the CCR10 ligand cutaneous T cell-attracting chemokine (CTACK) in lymphocyte trafficking to inflamed skin.

The chemokine thymus and activation-regulated chemokine (TARC; CCL17) is displayed by cutaneous (but not intestinal) venules, and is thought to trigger vascular arrest of circulating skin homing memory T cells, which uniformly express the TARC receptor CC chemokine receptor (CCR)4. Cutaneous T cell-attracting chemokine (CTACK; CCL27), expressed by skin keratinocytes, also attracts cutaneous memory T cells, and is hypothesized to assist in lymphocyte recruitment to skin as well. Here we show that chronic cutaneous inflammation induces CD4 T cells expressing E-selectin binding activity (a marker of skin homing memory cells) in draining lymph node, and that these E-selectin ligand+ T cells migrate efficiently to TARC and to CTACK. In 24 h in vivo homing assays, stimulated lymph node T cells from wild-type mice or, surprisingly, from CCR4-deficient donors migrate efficiently to inflamed skin; and an inhibitory anti-CTACK antibody has no effect on wild-type lymphocyte recruitment. However, inhibition with anti-CTACK monoclonal antibody abrogates skin recruitment of CCR4-deficient T cells. We conclude that CTACK and CCR4 can both support homing of T cells to skin, and that either one or the other is required for lymphocyte recruitment in cutaneous delayed type hypersensitivity.

Enhanced phosphorylation of p53 by ATM in response to DNA damage.

The ATM protein, encoded by the gene responsible for the human genetic disorder ataxia telangiectasia (A-T), regulates several cellular responses to DNA breaks. ATM shares a phosphoinositide 3-kinase-related domain with several proteins, some of them protein kinases. A wortmannin-sensitive protein kinase activity was associated with endogenous or recombinant ATM and was abolished by structural ATM mutations. In vitro substrates included the translation repressor PHAS-I and the p53 protein. ATM phosphorylated p53 in vitro on a single residue, serine-15, which is phosphorylated in vivo in response to DNA damage. This activity was markedly enhanced within minutes after treatment of cells with a radiomimetic drug; the total amount of ATM remained unchanged. Various damage-induced responses may be activated by enhancement of the protein kinase activity of ATM.

Butyrylcholinesterase-positive cells of the developing chicken retina that are non-cholinergic and GABA-positive.

Butyrylcholinesterase (BChE) is closely related to acetylcholinesterase (AChE), but its function in nervous system development or physiology is unclear. Here, the distribution of BChE was investigated by immunohistochemical methods in the developing chick retina. Using a specific anti-BChE antibody, we detected immunoreactivity associated with different cell types in two nuclear layers and in plexiform layers of the retina. At embryonic day 10 (E10), a transient BChE staining is detected in the inner plexiform layer (IPL) and in radial cells, the latter possibly representing Müller glia. At E12, a subpopulation of amacrine cells appeared, followed by cells in the middle and outer half of the inner nuclear layer. These cells at locations of amacrine, bipolar and horizontal cells represented the predominant three cell types persisting until hatching. The BChE+ amacrine cells were studied in more detail. Their distribution was not significantly different in the central and peripheral retina. Double labelling experiments revealed that BChE+ amacrine cells did not express choline acetyltransferase (ChAT), and, thus, are non-cholinergic. Only a minority of them coexpressed AChE. On the other hand, the majority of them colocalized with anti-GABA immunoreactivity. Taken together, these data support a hitherto unsuspected role of BChE in non-cholinergic cells, possibly in conjunction with GABA.

Affinity purification of ubiquitin-protein ligase on immobilized protein substrates. Evidence for the existence of separate NH2-terminal binding sites on a single enzyme.

Previous studies have indicated the existence of separate binding sites of ubiquitin-protein ligase, E3, specific for basic (Type I) or bulky hydrophobic (Type II) NH2-terminal amino acid residues of proteins. Another class (Type III) of protein substrates appeared to interact with E3 at regions other than the NH2 terminus (Reiss, Y., Kaim, D., and Hershko, A. (1988) J. Biol. Chem. 263, 2693-2698). In the present study we have used affinity chromatography on immobilized protein substrates to examine the question of whether the different binding sites belong to one E3 enzyme, or to different E3 species. Another objective was to develop a procedure for the extensive purification of E3. When a crude extract of reticulocytes is applied to Type I or Type II protein substrates linked to Sepharose, E3 becomes strongly bound to the affinity columns and is not eluted with salt at high concentration. However, the enzyme can be specifically eluted by a dipeptide that has an NH2-terminal residue similar to that of matrix-bound protein substrate. A 350-fold purification is obtained in this single step. Preparations of E3 purified on either Type I or Type II protein substrate affinity columns act on both types of protein substrates, indicating that the separate binding sites for basic and hydrophobic NH2-terminal residues belong to one enzyme. Another species of E3 that acts strongly on some Type III protein substrates does not bind to Type I or Type II protein substrate affinity columns.

T cell interaction with ICAM-1-deficient endothelium in vitro: transendothelial migration of different T cell populations is mediated by endothelial ICAM-1 and ICAM-2.

The trafficking of T lymphocytes is carefully regulated by adhesive interactions with the vascular endothelium. Depending on their maturation and activation stage, T lymphocytes exhibit distinctive patterns of homing and recirculation, which is at least partly due to the selective expression of cell adhesion molecules (CAM) on the T cell surface. In order to define whether the differential usage of CAM during the steps of transendothelial migration is involved in organ-specific recirculation of different T cell subsets we compared the interaction of three different T cell populations with mouse endothelioma cell lines in vitro. Using a novel approach, where we directly compared T cell interaction with ICAM-1-deficient endothelium to wild-type endothelium, we recently demonstrated that endothelial ICAM-1 and ICAM-2 play a key role in mediating the transendothelial migration of CD4(+) memory T cells. Here we show that endothelial ICAM-1 and ICAM-2 are equally required for the transendothelial migration of other T cell populations such as thymocytes and T lymphoma cells, which differ from CD4(+) memory T cells in their maturation and activation stage, as well as in their surface expression of adhesion molecules. Our data therefore demonstrate that transendothelial migration of different T cell populations is mediated by the same endothelial CAM, i.e. ICAM-1 and ICAM-2, and thus subset-specific interaction of T cells with endothelial cells must be regulated prior to transendothelial migration.

Specificity of binding of NH2-terminal residue of proteins to ubiquitin-protein ligase. Use of amino acid derivatives to characterize specific binding sites.

Previous studies have indicated that at least part of the selection of proteins for degradation takes place at a binding site on ubiquitin-protein ligase, to which the protein substrate is bound prior to ligation to ubiquitin. It was also shown that proteins with free NH2-terminal alpha-NH2 groups bind better to this site than proteins with blocked NH2 termini (Hershko, A., Heller, H., Eytan, E., and Reiss, Y. (1986) J. Biol. Chem. 261, 11992-11999). In the present study, we used simple derivatives of amino acids, such as methyl esters, hydroxamates, or dipeptides, to examine the question of whether the protein binding site of the ligase is able to distinguish between different NH2-terminal residues of proteins. Based on specific patterns of inhibition of the binding to ligase by these derivatives, three types of protein substrates could be distinguished. Type I substrates are proteins that have a basic NH2-terminal residue (such as ribonuclease and lysozyme); these are specifically inhibited by derivatives of the 3 basic amino acids (His, Arg, and Lys) with respect to degradation, ligation to ubiquitin, and binding to ligase. Type II substrates (such as beta-lactoglobulin or pepsinogen, that have a Leu residue at the NH2 terminus) are not affected by the above compounds, but are specifically inhibited by derivatives of bulky hydrophobic amino acids (Leu, Trp, Phe, and Tyr). In these cases, the amino acid derivatives apparently act as specific inhibitors of the binding of the NH2-terminal residue of proteins, as indicated by the following observations: (a) derivatives in which the alpha-NH2 group is blocked were inactive and (b) in dipeptides, the inhibitory amino acid residue had to be at the NH2-terminal position. An additional class (Type III) of substrates comprises proteins that have neither basic nor bulky hydrophobic NH2-terminal amino acid residues; the binding of these proteins is not inhibited by homologous amino acid derivatives that have NH2-terminal residues similar to that of the protein. It is concluded that Type I and Type II proteins bind to distinct and separate subsites of the ligase, specific for basic or bulky hydrophobic NH2-terminal residues, respectively. On the other hand, Type III proteins apparently predominantly interact with the ligase at regions of the protein molecule other than the NH2-terminal residue.

Binding sites of ubiquitin-protein ligase. Binding of ubiquitin-protein conjugates and of ubiquitin-carrier protein.

It was found previously that the enzyme ubiquitin-protein ligase (E3) contains specific protein substrate binding sites that are responsible for the selection of proteins for degradation by the ubiquitin system. In the present study, we have tried to gain more insight into the mode of action of E3 by the characterization of other binding sites of this enzyme. Following the ligation of ubiquitin to 125I-lysozyme, the conjugates produced are very tightly bound to E3, as indicated by size analysis on glycerol density gradient centrifugation. The strong binding of ubiquitin-protein conjugates to the enzyme may account for the apparently processive addition of multiple molecules of ubiquitin to the protein substrate. Both the protein substrate moiety and the ubiquitin moiety participate in the interaction of ubiquitin-protein conjugates with E3, as indicated by competition with specific agents and by the comparison of the binding of ubiquitin-conjugated protein to that of free protein. In addition to the binding of its substrates and products, E3 also appears to interact with some of the enzymes with which it acts in concert. When E3 is incubated with the ubiquitin-carrier protein E2, a complex is formed between the two enzymes as analyzed on glycerol gradients. The formation of an E2.E3 complex may facilitate the transfer of activated ubiquitin from E2 to the protein substrate bound to the ligase.

The protein substrate binding site of the ubiquitin-protein ligase system.

In order to gain insight into the mechanisms that determine the selectivity of the ubiquitin proteolytic pathway, the protein substrate binding site of the ubiquitin-protein ligase system was identified and examined. Previous studies had shown that the ligase system consists of three components: a ubiquitin-activating enzyme (E1), ubiquitin-carrier protein (E2), and a third enzyme, E3, the mode of action of which has not been defined. E3 from rabbit reticulocytes was further purified by a combination of affinity chromatography, hydrophobic chromatography, and gel filtration procedures. A 180-kDa protein was identified as the subunit of E3. Two independent methods indicate that E3 has the protein binding site of the ubiquitin ligase system. These are the chemical cross-linking of 125I-labeled proteins to the E3 subunit and the functional conversion of enzyme-bound labeled proteins to ubiquitin conjugates in pulse-chase experiments. The trapping of E3-bound protein for labeled product formation was allowed by the slow dissociation of E3 X protein complex. The specificity of binding of different proteins to E3, examined by both methods, showed a direct correlation with their susceptibility to degradation by the ubiquitin system. Proteins with free alpha-NH2 groups, which are good substrates, bind better to E3 than corresponding proteins with blocked NH2 termini, which are not substrates. Oxidation of methionine residues to sulfoxide derivatives greatly increases the susceptibility of some proteins to ligation with ubiquitin, with a corresponding increase in their binding to E3. However, a protein derivative which was subjected to both amino group modification and oxidation binds strongly to the enzyme, even though it cannot be ligated to ubiquitin. It thus seems that the substrate binding site of E3 participates in determining the specificity of proteins that enter the ubiquitin pathway of protein degradation.

Divalent cation and prenyl pyrophosphate specificities of the protein farnesyltransferase from rat brain, a zinc metalloenzyme.

The separate catalytic roles of Zn2+ and Mg2+ and the specificity of the prenyl pyrophosphate-binding site of the rat brain protein farnesyltransferase were explored using a purified enzyme preparation. The binding of p21Hras to the enzyme was abolished by dialysis against EDTA and restored by addition of ZnCl2, as demonstrated by chemical cross-linking. The binding of the other substrate, farnesyl pyrophosphate, was independent of divalent cations, as demonstrated by gel filtration. Transfer of the enzyme-bound farnesyl group to the bound p21Hras required Mg2+. Geranylgeranyl pyrophosphate bound to the prenyl pyrophosphate-binding site with an affinity equal to that of farnesyl pyrophosphate, but the geranylgeranyl group was not transferred efficiently to p21Hras. It also was not transferred to a modified p21Hras containing COOH-terminal leucine, a protein that was shown previously to be a good substrate for a rat brain geranylgeranyltransferase. We conclude that the protein farnesyltransferase is a metalloenzyme that most likely contains Zn2+ at the peptide-binding site. It thus resembles certain metallopeptidases, including carboxypeptidase A and the angiotensin-converting enzyme. Strategies previously developed to screen for inhibitors of those enzymes may aid in the search for inhibitors of the protein farnesyltransferase.

Immunosurveillance modelled in vitro: naive and memory T cells spontaneously migrate across unstimulated microvascular endothelium.

As a model for T cell immigration into non-lymphoid tissue we set up an in vitro assay that would allow us to investigate the phenotype of T lymphocytes from peripheral lymph nodes (PLN), mesenteric lymph nodes (MLN) or peripheral blood (PBL) of mice, which were able to spontaneously migrate across unstimulated microvascular endothelium. The transendothelial migrating T cell population was enriched for T lymphocytes expressing a "recently activated/memory' phenotype: LFA-1/CD44/ICAM-1high, but also contained CD45RBhigh and LFA-1low T cells, which in the case of MLN T cells were phenotyped as CD4+ and thus characterized as naive T cells. Transmigrated T cells could be further distinguished from their original populations and from each other by their distinct but heterogeneous expression patterns for L-selectin, alpha 4 beta 7-integrin and PECAM-1. This observation suggests the presence of phenotypically different migratory T cells among MLN, PLN and PBL. Additional studies provided evidence that the capacity to migrate across unstimulated microvascular endothelium was a characteristic of a T cell population that could phenotypically be differentiated from activated T cells. The endothelial cells were found to play an active role in selecting the traversing T cell population, as they controlled the number and phenotype of spontaneously transmigrating T cells. Our studies suggest that the capacity to transmigrate across unstimulated microvascular endothelium and hence to immigrate into non-lymphoid tissue is owned by a phenotypically heterogeneous T cell population, which is enriched for memory T cells but not devoid of naive T cells.

T cell interaction with ICAM-1-deficient endothelium in vitro: essential role for ICAM-1 and ICAM-2 in transendothelial migration of T cells.

Transendothelial migration is a crucial step in the complex process of lymphocyte extravasation during lymphocyte homing, immunosurveillance and inflammation. However, little is known about the precise role of cell adhesion molecules (CAM) involved in this particular event. To define the CAM involved in T cell adhesion versus transendothelial migration, we have previously established an in vitro transendothelial migration system using mouse T cells and mouse endothelioma cells. We demonstrate here that, using ICAM-1-deficient endothelioma cells derived from ICAM-1 mutant mice, transendothelial migration of T cells was inhibited to a much greater extent when compared to migration across wild-type cells treated with a blocking anti-ICAM-1 monoclonal antibody. This unexpected result was confirmed by a rescue experiment using retroviral transfer of wild-type ICAM-1 into ICAM-1-deficient endothelial cells. Additional experiments showed that, in the absence of functional ICAM-1, only ICAM-2 was involved in transendothelial migration, but not PECAM-1, VCAM-1, or E-selectin. Taking this novel approach, we show that ICAM-1 and ICAM-2 are essential for transendothelial migration of T cells.

Inhibition of purified p21ras farnesyl:protein transferase by Cys-AAX tetrapeptides.

We report the identification, purification, and characterization of a farnesyl:protein transferase that transfers the farnesyl moiety from farnesyl pyrophosphate to a cysteine in p21ras proteins. The enzyme was purified approximately 60,000-fold from rat brain cytosol through use of a chromatography step based on the enzyme's ability to bind to a hexapeptide containing the consensus sequence (Cys-AAX) for farnesylation. The purified enzyme migrated on gel filtration chromatography with an apparent molecular weight of 70,000-100,000. High resolution SDS-polyacrylamide gels showed two closely spaced approximately 50 kd protein bands in the final preparation. The enzyme was inhibited competitively by peptides as short as 4 residues that contained the Cys-AAX motif. These peptides acted as alternative substrates that competed with p21H-ras for farnesylation. Effective peptides included the COOH-terminal sequences of all known p21ras proteins as well as those of lamin A and B.

Nonfarnesylated tetrapeptide inhibitors of protein farnesyltransferase.

The protein farnesyltransferase from rat brain was previously shown to be inhibited competitively by tetrapeptides that conform to the consensus Cys-A1-A2-X, where A1 and A2 are aliphatic amino acids and X is methionine, serine, or phenylalanine. In the current studies we use a thin layer chromatography assay to show that most of these tetrapeptides are themselves farnesylated by the purified enzyme. Two classes of tetrapeptides are not farnesylated and therefore act as true inhibitors: 1) those that contain an aromatic residue at the A2 position and 2) those that contain penicillamine (beta,beta-dimethylcysteine) in place of cysteine. The most potent of these pure inhibitors was Cys-Val-Phe-Met, which inhibited farnesyltransferase activity by 50% at less than 0.1 microM. These data indicate that the inclusion of bulky aromatic or methyl residues in a tetrapeptide can abolish prenyl group transfer without blocking binding to the enzyme. This information should be useful in the design of peptides or peptidomimetics that inhibit farnesylation and thus block the action of p21ras proteins in animal cells.

Sequence requirement for peptide recognition by rat brain p21ras protein farnesyltransferase.

We tested 42 tetrapeptides for their ability to bind to the rat brain p21ras protein farnesyltransferase as estimated by their ability to compete with p21Ha-ras in a farnesyltransfer assay. Peptides with the highest affinity had the structure Cys-A1-A2-X, where positions A1 and A2 are occupied by aliphatic amino acids and position X is occupied by a COOH-terminal methionine, serine, or phenylalanine. Charged residues reduced affinity slightly at the A1 position and much more drastically at the A2 and X positions. Effective inhibitors included tetrapeptides corresponding to the COOH termini of all animal cell proteins known to be farnesylated. In contrast, the tetrapeptide Cys-Ala-Ile-Leu (CAIL), which corresponds to the COOH termini of several neural guanine nucleotide binding (G) protein gamma subunits, did not compete in the farnesyl-transfer assay. Inasmuch as several of these proteins are geranylgeranylated, the data suggest that the two isoprenes (farnesyl and geranylgeranyl) are transferred by different enzymes. A biotinylated heptapeptide corresponding to the COOH terminus of p21Ki-rasB was farnesylated, suggesting that at least some of the peptides serve as substrates for the transferase. The data are consistent with a model in which a hydrophobic pocket in the protein farnesyltransferase recognizes tetrapeptides through interactions with the cysteine and the last two amino acids.

Protein farnesyltransferase and geranylgeranyltransferase share a common alpha subunit.

Mammalian farnesyltransferase, which attaches a 15 carbon isoprenoid, farnesyl, to a cysteine in p21ras proteins, contains two subunits, alpha and beta. The beta subunit is known to bind p21ras proteins. We show here that the alpha subunit is shared with another prenyltransferase that attaches 20 carbon geranylgeranyl to Ras-related proteins. Farnesyltransferase and geranylgeranyltransferase have similar molecular weights on gel filtration, but are separated by ion exchange chromatography. Both enzymes are precipitated and immunoblotted by multiple antibodies directed against the alpha subunit of farnesyltransferase. The two transferases have different specificities for the protein acceptor; farnesyltransferase prefers methionine or serine at the COOH-terminus and geranylgeranyltransferase prefers leucine. The current data indicate that both prenyltransferases are heterodimers that share a common alpha subunit with different beta subunits.

Production of a specific major histocompatibility complex class I-restricted epitope by ubiquitin-dependent degradation of modified ovalbumin in lymphocyte lysate.

Peptide epitopes presented through class I major histocompatability complex (MHC class I) on the cell surface, are generated by proteolytic processing of protein-antigens in the cytoplasm. The length and amino acid sequence determine whether a given peptide can fit into the peptide binding groove of class I heavy chain molecules and subsequently be presented to the immune system. The mode of action of the processing pathway is therefore of great interest. To study the processing mechanism of MHC class I-restricted intracellular antigens, we reconstituted the proteolytic processing of a model antigen in a cell-free system. Incubation of oxidized and urea-treated OVA in lymphocyte lysate resulted in partial degradation of the antigen. Degradation of the antigen depended on the presence of ATP. Addition of methylated ubiquitin abolished the reaction which was then restored by addition of an excess of native ubiquitin, indicating that the breakdown of the antigen in lymphocyte lysate is mediated by the ubiquitin proteolytic system. Upon incubation of modified OVA in lymphocyte lysate, a specific antigenic peptide was generated. The peptide was recognized by cytotoxic T lymphocytes directed against OVA-derived, H-2Kb-restricted peptide (SIINFEKL), and by a monoclonal antibody that recognizes cell-bound Kb-SIINFEKL complexes. Formation of the peptide epitope depended on the presence of ATP and ubiquitin. These results indicate that proteolytic processing of modified OVA is carried out by the ubiquitin-mediated degradation system. The experimental system described provides a tool to analyze the molecular mechanisms underlying the generation of specific, MHC class I-restricted peptide epitopes.

Extracellular and asymmetric forms of acetylcholinesterase are expressed on cholinergic and noncholinergic terminal neuropil of the developing chick retina.

Only two out of four major acetylcholinesterase (AChE) subbands in the inner plexiform layer (IPL) of vertebrate retinae correspond to sites of cholinergic synaptic transmission, as has been shown by the co-distribution of AChE and choline acetyltransferase (ChAT) staining. The function and molecular identity of AChE in non-cholinergic subbands is unknown. We have used immunocytochemical methods to compare the development of asymmetric or extracellularly localized AChE with that of total AChE and ChAT in embryonic and adult chicken retinae. After injection of the AChE-specific monoclonal antibody 3D10 into the vitreous body of live embryos, a method that labels only extracellular AChE, five subbands in the IPL were labelled, whereas cell somata or their radial processes remained unstained. In contrast, the entire cell including processes was immunoreactive, when the 3D10 antibody was applied to permeabilized cryosections, suggesting that in cell bodies the enzyme is exclusively localized intracellularly. Compared with total AChE, detection of asymmetric AChE with the monoclonal antibody 6B6 was delayed, first being seen in cells of the inner nuclear layer and finally appearing on all subbands, reflecting more closely the course of synaptogenesis. Thus, extracellular and asymmetric forms of AChE are predominantly found on the terminal arbor neuropil of both cholinergic and non-cholinergic IPL subbands. These data show a differential distribution of extra- and intracellular AChE and suggest novel roles for the AChE in non-cholinergic IPL subbands.

A role for a novel luminal endoplasmic reticulum aminopeptidase in final trimming of 26 S proteasome-generated major histocompatability complex class I antigenic peptides.

Peptides presented to cytotoxic T lymphocytes by the class I major histocompatability complex are 8-11 residues long. Although proteasomal activity generates the precise C termini of antigenic epitopes, the mechanism(s) involved in generation of the precise N termini is largely unknown. To investigate the mechanism of N-terminal peptide processing, we used a cell-free system in which two recombinant ornithine decarboxylase (ODC) constructs, one expressing the native H2-K(b)-restricted ovalbumin (ova)-derived epitope SIINFEKL (ODC-ova) and the other expressing the extended epitope LESIINFEKL (ODC-LEova), were targeted to degradation by 26 S proteasomes followed by import into microsomes. We found that the cleavage specificity of the 26 S proteasome was influenced by the N-terminal flanking amino acids leading to significantly different yields of the final epitope SIINFEKL. Following incubation in the presence of purified 26 S proteasome, ODC-LEova generated largely ESIINFEKL that was efficiently converted to the final epitope SIINFEKL following translocation into microsomes. The conversion of ESIINFEKL to SIINFEKL was strictly dependent on the presence of H2-K(b) and was completely inhibited by the metalloaminopeptidase inhibitor 1,10-phenanthroline. Importantly, the converting activity was resistant to a stringent salt/EDTA wash of the microsomes and was only apparent when transport of TAP, the transporter associated with antigen processing, was facilitated. These results strongly suggest a crucial role for a luminal endoplasmic reticulum-resident metalloaminopeptidase in the N-terminal trimming of major histocompatability complex class I-associated peptides.