The immediate early protein ICP47 of the Herpes simplex virus is known to block the human transporter associated with antigen processing (TAP), thereby creating a TAP-deficient phenotype in any human cell transfected with the corresponding cDNA. Exploiting this inhibitory activity, we constructed a selection of human cell lines each co-expressing one of the cDNAs of human leukocyte antigen (HLA) class I alleles HLA-A*1101, A24, A*3101, A*6601, B8 and B*1516, and the cDNA encoding the ICP47 molecule. The cell lines generated showed diminished HLA class I surface expression and the inhibition of the TAP function was confirmed in peptide translocation assays. The addition of specific exogenous peptide ligands restored the expression of the corresponding HLA class I molecules. Thus, the ICP47 transfectants provide us with a tool to closely examine peptide-HLA class I interactions, to confirm HLA class I ligand motifs and to test peptides predicted to bind.

Two siblings with a peptide TAP deficiency were recently described. Despite poor cell surface expression of HLA class I molecules, these patients were not unusually susceptible to viral infections. The majority of the cell surface-expressed class I molecules were HLA-B products as assessed by cytofluorometry and biochemical analysis. Analysis of two peptides eluted from the class I molecules expressed by TAP-deficient EBV B lymphoblastoid cell lines indicated that both were derived from cytosolic proteins and presented by HLA-B molecules. Peripheral alphabeta CD8+ T cells were present and their TCR repertoire was polyclonal. Most of the alphabeta CD8+ T cell clones studied (21 of 22) were nonreactive against cells expressing normal levels of the same HLA alleles as those of the TAP-deficient patients. However, it was possible to isolate one cytotoxic CD8+ alphabeta T cell clone recognizing the EBV protein LMP2 presented by HLA-B molecules on TAP-deficient cells. These observations suggest that in the TAP-deficient patients, CD8+ alphabeta T cells could mature and be recruited in immune responses to mediate HLA class I-restricted cytotoxic defense against viral infections. They also strengthen the physiologic importance of a TAP-independent processing pathway of the LMP2 protein, which was previously shown to contain several other TAP-independent epitopes.

The endoplasmic reticulum (ER)-resident stress protein gp96 induces protective immunity and specific cytotoxic T lymphocyte (CTL) responses against antigens expressed in those cells it has been isolated from. This ability is based on peptides associated with gp96. Because gp96 is located inside the ER, our experiments address the question whether or not the repertoire of peptides associated with gp96 is influenced by the transporter associated with antigen processing (TAP). For this purpose, gp96 was isolated from cells with and without a TAP defect and used for immunization of mice. We found that for some antigens the association of peptides with gp96 required functional TAP molecules, whereas the association of peptides from other antigens was TAP independent. In the case of a TAP-dependent association of peptides with gp96, our results prove that peptide binding by gp96 in vivo occurs inside the ER and is not an artifact induced by cell lysis during the gp96 purification. The finding that some antigens can also associate with gp96 in the absence of functional TAP molecules indicates that the repertoire of peptides bound by gp96 truly reflects the entire repertoire of peptides present inside the ER and not only those peptides transported by TAP. These results, together with the earlier finding that the gp96 peptide repertoire is independent of the major histocompatibility complex molecules expressed by the cell gp96 is isolated from, give the theoretical foundation for the ability of gp96 to induce CTL responses against all kinds of intracellular antigens.

The substrate specificity of the non-lysosomal aspartic protease cathepsin E from three different species has been studied using the method of automated N-terminal sequencing and a newly developed method for C-terminal sequencing of peptides and peptide pools. The combination of N-terminal and C-terminal sequencing of peptide pools is a fast and easy method to identify and compare the substrate specificity of endopeptidases. Our analysis shows a conserved hydrolytic specificity between human, mouse and bovine cathepsin E, with only small differences in fine specificity. Furthermore, our results confirm and extend the rules governing the interactions of the substrate with the amino acid (aa) side chains of the various pockets within the enzyme's active cleft. We found that the positions flanking the scissile peptide bond P1-P1' are occupied exclusively by hydrophobic aa with both aliphatic or aromatic side chains; Val and Ile, however, are not allowed in the S1 binding site. The S2 and S2' subsites accept hydrophilic aa. Additional requirements concerning the S3' to S5' subsites were also revealed. Finally, the sequences of single peptides generated by cathepsin E from the three different species can be easily aligned to the determined cleavage motif, showing the reliability of our pool sequencing methods.

Degradation of protein antigens by cellular proteases is a crucial step in the initiation of a T-cell-mediated immune response. But still little is known about the enzymes responsible for the processing of antigens, including their specificity. In this paper, we show that the combination of automated N-terminal sequencing with a newly developed method for C-terminal sequencing of peptide pools generated by the aspartic proteases cathepsins D and E is a fast and easy method to obtain detailed information of the substrate specificity of these endopeptidases. Using a 15-residue synthetic peptide library and a native protein as substrates, we confirm and extend the knowledge about the cleavage motif of cathepsin E where positions P1 and P1' of the substrate must be occupied exclusively by hydrophobic amino acids with aromatic or aliphatic side chains. However, Val and Ile residues are not allowed at position P1. Position P2' accepts a broad range of amino acids, including charged and polar ones. Additional requirements concerning the substrate positions P3' and P4' were also defined by pool sequencing. Furthermore, pool sequencing analysis of melittin digests with the aspartic proteases cathepsin D and E provided evidence that both enzymes share the same cleavage motif, identical to the one derived from the peptide library and the native protein. Therefore, pool sequencing analysis is a valuable and fast tool to determine the substrate specificity of any endopeptidase.

The endoplasmic reticulum (ER)-resident stress protein gp96 induces a major histocompatibility complex class I-restricted cytotoxic T lymphocyte (CTL) response against antigens present in the cells from which it has been prepared. In this study, photoreactive peptides were translocated into the ER by the transporter associated with antigen processing (TAP). These peptides can be cross-linked specifically to gp96. Thus, we provide the first evidence that gp96 binds TAP-translocated peptides which have been implicated in the induction of specific CTL responses after immunization with gp96 (Srivastava, P. K. et al., Immunogenetics 1994. 39: 93).

Immunization of mice with gp96 induces CTL with specificity for proteins that are expressed in the cells from which gp96 was isolated (Arnold et al., J. Exp. Med. 1995. 182: 885, Udono et al., Proc. Natl. Acad. Sci. USA 1994. 91: 3077). Recently, it has been shown that gp96 from cells transfected with vesicular stomatitis virus (VSV) nucleocapsid protein as well as gp96 loaded in vitro with peptides containing an epitope of this protein are taken up by phagocytic cells which obtain thereby the capacity for stimulating VSV‐specific cytotoxic T lymphocytes (Suto and Srivastava, Science 1995. 269: 1585). The immunization experiments together with the peptide transfer from gp96/peptide complexes to major histocompatibility complex (MHC) class I molecules of phagocytic cells are consistent with the hypothesis that the endoplasmic reticulum‐resident protein gp96 plays a crucial role in the antigen presentation of a cell (Srivastava et al., Immunogenetics 1994. 39: 93). To examine the involvement of gp96 in class I‐restricted antigen presentation, we reduced gp96 RNA and protein levels by transfecting P13.1 cells with a vector containing part of gp96 cDNA in antisense orientation to the promoter. We found that antisense clones expressing strongly reduced levels of gp96 mRNA and gp96 protein show normal levels of MHC class I molecules on the cell surface and are recognized by T cells to the same extent as wild‐type cells. Thus, our results show that normal levels of gp96 expression in a cell are not limiting for class I‐restricted antigen presentation.

The major histocompatibility complex-encoded transporter associated with antigen processing (TAP) is required for the efficient presentation of cytosolic antigens to class I-restricted T cells. TAP is thought to be formed by the interaction of two gene products, termed TAP1 and TAP2. We find that TAPs consisting either of human subunits, or mouse TAP1 and human TAP2, facilitate the presentation of numerous defined viral peptides to mouse class I-restricted T cells. As human and mouse TAP2 and TAP1 differ in 23 and 28% of their residues, respectively, this indicates that TAP1 and TAP2 can form a functional complex with partners considerably different from those they coevolved with. Moreover, these findings indicate that widely disparate TAPs facilitate delivery of the same peptides to class I molecules. These findings suggest that TAP polymorphism does not greatly influence the types of peptides presented to the immune system.

The finding that two subunits of the proteasome, LMP2 and LMP7, are encoded in the major histocompatibility complex (MHC) has linked the proteasome which represents a major extralysosomal proteolytic system to the processing of intracellular antigens. Here we describe a second form of the human LMP7 cDNA, LMP7-E2, which has been identified during the characterization of novel genes in the MHC. The analysis of the genome organization of LMP7 revealed that LMP7-E1 and LMP7-E2 arise by alternative exon usage. Using specific antibodies against LMP2 and LMP7, we show that they are co-expressed with class I MHC molecules as well as a putative peptide transporter. The polypeptides encoded by LMP7 and LMP2 undergo proteolytic processing when incorporated into proteasomes, and the LMP7 precursor is derived mainly from LMP7-E2. Furthermore, our data suggest that LMP7 and LMP2 are mutually dependent for their incorporation into the proteasomal complex.

Transporter in Ag processing-1 (TAP-1, previously called PSF-1 or Ring-4) is an MHC-encoded gene product that is required for efficient association of intracellular peptide Ag with nascent HLA class I H chain and beta 2-microglobulin, thereby permitting assembly and normal surface expression of the class I molecules. TAP-1 is thought to function as a component of a transmembrane pump, that transports cytoplasmically-derived peptides into the lumen of the endoplasmic reticulum where class I molecules assemble. Synthesis and expression of HLA class I molecules is increased in human endothelial cells by IFN-beta, IFN-gamma, and TNF. We report these same cytokines increase TAP-1 expression. As with class I, TAP-1 is also synergistically increased by combinations of TNF with IFN. Interestingly, cytokine-induced increases in TAP-1 mRNA are markedly more rapid than increases in class I mRNA. This rapid increase in TAP-1 mRNA is reflected in a rapid increase in TAP-1 protein. These results demonstrate that TAP-1 synthesis and class I synthesis are regulated in parallel. The rapidity of the cytokine response of TAP-1 compared to class I further suggests that the constitutive level of TAP-1 expression in endothelial cells is not sufficient to support inducible increases in class I expression.

Antigen processing provides major histocompatibility complex (MHC) class I molecules with short peptides, which they selectively bind and present to cytotoxic T lymphocytes. The proteolytic system generating these peptides in the cytosol is unidentified, but their delivery into the endoplasmic reticulum is mediated by the TAP1-TAP2 transporter encoded in the MHC class II region. Closely linked to TAP1 and TAP2 are genes for the LMP2 and LMP7 proteins, which resemble components of proteasomes, proteolytic complexes known to degrade cytosolic proteins. This association has led to the common assumption that proteasomes function in this immunological pathway (discussed in ref. 15). We now show that the expression of stably assembled class I molecules and apparently normal peptide processing can be completely restored in the absence of LMP2 and LMP7 in the human lymphoblastoid cell line mutant 721.174 (refs 16, 17). The identity of LMP7 is directly confirmed by reconstitution of a proteasomal subunit after gene transfer. These results therefore dispute the hypothetical involvement of proteasomes in antigen processing, although a more subtle effect of LMP2 and LMP7 cannot be ruled out.

Major histocompatibility complex (MHC) class I molecules export peptides to the cell surface for surveillance by cytotoxic T lymphocytes(1-3). Intracellular peptide binding is critical for the proper assembly and transport of class I molecules(4-6). This mechanism is impaired as a result of a non-functional peptide supply factor gene (PSF) In several human mutant cell lines with genomic lesions In the MHC. We have now Identified PSF in the MHC class II region by deletion mapping in mutants and chromosome-walking. PSF is homologous to mammalian and bacterial ATP-dependent transport proteins, suggesting that It operates In the Intracellular transport of peptides. The