Melanomas with concordant loss of multiple melanocytic differentiation proteins: immune escape that may be overcome by targeting unique or undefined antigens.

Melanoma-reactive HLA-A x 0201-restricted cytotoxic T lymphocyte (CTL) lines generated in vitro lyse autologous and HLA-matched allogeneic melanoma cells and recognize multiple shared peptide antigens from tyrosinase, MART-1, and Pme117/gp100. However, a subset of melanomas fail to be lysed by these T cells. In the present report, four different HLA-A x 0201+ melanoma cell lines not lysed by melanoma-reactive allogeneic CTL have been evaluated in detail. All four are deficient in expression of the melanocytic differentiation proteins (MDP) tyrosinase, Pme117/gp100, gp75/ trp-1, and MART-1/Melan-A. This concordant loss of multiple MDP explains their resistance to lysis by melanoma-reactive allogeneic CTL and confirms that a subset of melanomas may be resistant to tumor vaccines directed against multiple MDP-derived epitopes. All four melanoma lines expressed normal levels of HLAA x 0201, and all were susceptible to lysis by xenoreactive-peptide-dependent HLA-A x 0201-specific CTL clones, indicating that none had identifiable defects in antigen-processing pathways. Despite the lack of shared MDP-derived antigens, one of these MDP-negative melanomas, DM331, stimulated an effective autologous CTL response in vitro, which was restricted to autologous tumor reactivity. MHC-associated peptides isolated by immunoaffinity chromatography from HLA-A1 and HLA-A2 molecules of DM331 tumor cells included at least three peptide epitopes recognized by DM331 CTL and restricted by HLA-A1 or by HLA-A x 0201. Recognition of these CTL epitopes cannot be explained by defined, shared melanoma antigens; instead, unique or undefined antigens must be responsible for the autologous-cell-specific anti-melanoma response. These findings suggest that immunotherapy directed against shared melanoma antigens should be supplemented with immunotherapy directed against unique antigens or other undefined antigens, especially in patients whose tumors do not express MDP.

The class I antigen-processing pathway for the membrane protein tyrosinase involves translation in the endoplasmic reticulum and processing in the cytosol.

Formation of major histocompatibility complex class I-associated peptides from membrane proteins has not been thoroughly investigated. We examined the processing of an HLA-A*0201-associated epitope, YMDGTMSQV, that is derived from the membrane protein tyrosinase by posttranslational conversion of the sequence YMNGTMSQV. Only YMDGTMSQV and not YMNGTMSQV was presented by HLA-A*0201 on cells expressing full-length tyrosinase, although both peptides have similar affinities for HLA-A*0201 and are transported by TAP. In contrast, translation of YMNGTMSQV in the cytosol, as a minigene or a larger fragment of tyrosinase, led to the presentation of the unconverted YMNGTMSQV. This was not due to overexpression leading to saturation of the processing/conversion machinery, since presentation of the converted peptide, YMDGTMSQV, was low or undetectable. Thus, presentation of unconverted peptide was associated with translation in the cytosol, suggesting that processing of the full-length tyrosinase occurs after translation in the endoplasmic reticulum. Nevertheless, presentation of YMDGTMSQV in cells expressing full-length tyrosinase was TAP (transporter associated with antigen processing) and proteasome dependent. After inhibition of proteasome activity, tyrosinase species could be detected in the cytosol. We propose that processing of tyrosinase involves translation in the endoplasmic reticulum, export of full-length tyrosinase to the cytosol, and retransport of converted peptides by TAP for association with HLA-A*0201.

Definition of a human T cell epitope from influenza A non-structural protein 1 using HLA-A2.1 transgenic mice.

Previous results from this laboratory demonstrated that the dominant influenza A epitope recognized by HLA-A2.1-restricted cytotoxic T lymphocytes (CTL) from HLA-A2.1 transgenic mice was the matrix protein 1 (M1) peptide epitope that is immunodominant in human CTL responses. However, analysis of a large number of CTL lines revealed a subset of influenza A/PR/8/34-specific murine CTL that recognized an HLA-A2.1-restricted epitope distinct from M1. Using recombinant vaccinia viruses encoding different influenza gene segments, the epitope recognized by these CTL was shown to be derived from A/PR/8 non-structural protein 1 (NS1). Because these CTL did not recognize targets infected with the A/Alaska/6/77 strain of influenza, candidate peptide epitopes were synthesized based on sequences that included an HLA-A2.1-specific binding motif, and that differed between A/PR/8 and A/Alaska. All of these CTL recognized a nonamer and a decamer peptide which contained a common eight amino acid sequence and two distinct sets of binding motif residues. However, the nonamer peptide was able to sensitize CTL for half-maximal lysis at 80- to 2500-fold lower doses than either the octamer or decamer. The homologous peptide derived from A/Alaska NS1 contained conservative amino acid changes at positions 4 and 8, and was not recognized at any tested concentration, although it bound with higher affinity to HLA-A2.1 than the peptide from A/PR/8. The A/PR/8 NS1 nonamer epitope was also recognized by human influenza A-specific CTL derived from two individuals. These results substantiate the general utility of HLA class I transgenic mice for the identification of human CTL epitopes for other pathogens.

Characteristics of endogenous peptides eluted from the class I MHC molecule HLA-B7 determined by mass spectrometry and computer modeling.

Microcapillary HPLC electrospray ionization tandem mass spectrometry was used to sequence 15 peptides eluted from HLA-B7. Sequence alignment implicated four peptide positions in specific interactions with the class I molecule, and their importance was confirmed using synthetic peptides. Because no crystal structure for HLA-B7 was available, computer-assisted modeling was used to understand novel aspects of peptide binding specificity and to accurately predict the effect of defined changes in peptide structure. The results demonstrate that mass-spectrometric sequencing coupled with computer-assisted modeling can be used in the absence of a crystal structure to make accurate predictions concerning requirements for peptide binding to class I molecules. These techniques may be valuable to predict or engineer T cell epitopes.