Mutations in the alpha 2 helix of HLA-A2 affect presentation but do not inhibit binding of influenza virus matrix peptide.

Previous studies have suggested that MHC class I molecules bind and present peptides to CTL in a manner that is analogous to the presentation of peptides by class II molecules to Th. Crystallographic studies of HLA-A2 have led to the assignment of a putative peptide binding site that is bordered by two alpha helices consisting of residues 50-84 and 138-180. In this study, we have investigated whether residues in the alpha 2 helix are involved in the binding and/or presentation of a peptide to CTL. We have generated CTL to type A influenza virus by stimulation of human PBL with a synthetic peptide from the influenza A virus matrix protein (M1 residues 57-68) in the presence of rIL-2. Such HLA-A2.1-restricted influenza virus-immune CTL do not recognize infected HLA-A2.3+ targets. A2.1 and A2.3 differ by three amino acids in the alpha 2 domain: Ala vs. Thr at position 149, Val vs. Glu at position 152, and Leu vs. Trp at position 156. Site-directed mutants of the A2.1 gene that encode A2 molecules that resemble A2.3 at positions 149, 152, and 156 have been constructed, transfected into human cells, and assayed for their ability to present the M1 peptide. The results demonstrate that most, but not all, A2.1-restricted M1-peptide-specific CTL fail to recognize M1 peptide-exposed transfectants with certain single amino acid substitutions at positions 152 and 156. In contrast, M1 peptide-exposed transfectants that express A2 molecules with an Ala----Thr substitution at position 149 were recognized by all CTL tested, but they exhibited an apparent difference in the kinetics of peptide binding. These results indicate that amino acid substitutions at positions 152 and 156 of the putative peptide binding site of the A2 molecule can affect presentation without eliminating binding, and indicate that the failure to recognize complexes between the peptide and the mutant A2 molecules is due to different TCR specificities and not to the failure to bind the peptide.

The peptide recognized by HLA-A68.2-restricted, squamous cell carcinoma of the lung-specific cytotoxic T lymphocytes is derived from a mutated elongation factor 2 gene.

The identification of naturally processed tumor peptides that can stimulate a tumor-specific, CTL response is crucial to the development of a vaccine-based, immunotherapeutic approach to cancer treatment. One type of cancer in which a tumor-specific, CTL response has been observed is squamous cell carcinoma of the lung. In the system investigated here, the tumor-specific CTLs are HLA-A68.2 restricted. Immunoaffinity chromatography was used to isolate the HLA-A68.2 molecules from the tumor cell line, and peptide was eluted with acid from the HLA-A68.2 molecules and subjected to three rounds of separation by reversed phase-high performance liquid chromatography (RP-HPLC). To determine which fractions contained the peptide recognized by the tumor-specific CTLs, an aliquot of each RP-HPLC fraction was added to the autologous, B-lymphoblastoid cell line, and the cells were then tested as targets for tumor-specific CTLs. After the third round of RP-HPLC, mass spectrometry was used to sequence individual peptide candidates, and a peptide with a m/z of 497 was identified as the active peptide. Collision-activated dissociation of m/z 497 allowed identification of the peptide sequence as ETVSEQSNV. With the exception of a single amino acid difference (glutamic acid versus glutamine as the sixth position in the peptide), this peptide is identical to residues 581 to 589 of elongation factor 2. The PCR was used to amplify the elongation factor 2 gene in both the tumor cells and the autologous B cell line, and DNA sequencing of the products revealed the presence of a heterozygous mutation in the tumor cells that accounts for the difference between the two peptide sequences. Although a similar analysis did not reveal the presence of the mutation in three additional lung cell carcinomas, this does not rule out the possibility that a survey of a larger population of tumor cells would reveal the presence of the mutation at a low frequency. These results demonstrate the utility of this approach for identifying tumor-specific antigens that are the targets of a CTL response.

Both major and minor peptide-binding pockets in HLA-A2 influence the presentation of influenza virus matrix peptide to cytotoxic T lymphocytes.

Most of the polymorphic residues in class I MHC molecules are concentrated in the alpha 1- and alpha 2-domains with their side chains pointing towards the antigen peptide site. Previous crystal structure analysis revealed six pockets inside the peptide-binding groove and the "extra" electron density in some of the pockets indicated that the pockets are involved in direct peptide binding. In order to investigate the functional role of individual positions from each pocket in antigen presentation, 37 HLA-A2 variants with single amino acid substitution in the peptide-binding groove were generated and used to analyse the specificity of influenza A virus matrix peptide-specific, HLA-A2-restricted CTL. The ability to present peptide by each variant was studied in detail by peptide titration, cold target inhibition, time course and limiting dilution analysis. The direct effect on peptide binding by these substitutions was determined by cell surface class I MHC molecule reconstitution analysis. The results demonstrated that each of the six peptide binding pockets plays a role in T cell recognition. Substitutions introduced into pocket F had less effect on CTL recognition than substitutions introduced in other pockets. With the exception of Tyr substitution for Phe9, single amino acid substitutions in the peptide-binding groove had only minor effects on peptide binding. Therefore, the impact of the substitutions in altering the epitopes recognized by CTL seems to be mediated through an alteration in the conformation of the bound peptide.

Residues outside of the HLA-A2 peptide-binding groove can abrogate or enhance recognition of influenza virus matrix peptide pulsed cells by cytotoxic T lymphocytes.

An examination of the crystal structure of HLA-A2.1 reveals two classes of residues on the class I MHC molecule that could affect CTL recognition: (1) those predicted to interact with the TCR directly; and (2) those that interact with bound peptides. To examine the role of individual TCR contacting residues, as well as residues not predicted to interact with bound peptide or the TCR, a panel of 28 HLA-A2 variants that differ from each other by a single amino acid substitution in either the alpha 1- or alpha 2-domain was utilized. Peptide titration, time course and cold target inhibition analysis of these targets showed that only the substitution of position 62 in the alpha 1-domain had a significant effect on recognition of the MHC-peptide complex by influenza matrix protein M1 (57-68) peptide-specific, HLA-A2.1-restricted CTL. In contrast, substitutions at positions 154, 162 and 163 in the alpha 2-domain abolished recognition by the same CTL. Additionally, substitutions at position 138 in the alpha 2-domain and positions 107 and 127 on the loops connecting the beta-strand in the alpha 2-domain were recognized in a more efficient, heteroclitic fashion. Overall, there was no direct correlation between the level of peptide binding to the variants and the level of T cell recognition of the variants. These results indicate that residues in the alpha 2-domain may be more important than residues in the alpha 1-domain in controlling TCR binding to the class I MHC molecule and suggest that the "footprint" of the TCR may be more extensive than previously predicted and encompass a broad region that extends beyond the alpha 2-helix. These findings also imply that the class I MHC molecule may exist in a "tipped" orientation on the cell surface during T cell recognition.

Identification of class I MHC regions which bind to the adenovirus E3-19k protein.

The adenovirus early region 3 glycoprotein E3-19k binds to and inhibits expression of class I major histocompatibility complex (MHC)-encoded molecules, which may help infected cells evade immune recognition. The role of specific regions of the class I MHC molecule in the interaction with E3-19k was evaluated using a series of HLA-A2.1-, HLA-A2 variant-, and HLA-B7-expressing cell lines. The monoclonal antibody (mAb) W6/32, which recognizes a monomorphic epitope on class I MHC molecules, readily co-immunoprecipitated E3-19k with HLA-A2.1 and 14 different HLA-A2 variant molecules that differ from HLA-A2.1 by single amino acid substitutions. Thus, no single residue tested in the regions of the class I MHC molecule that bind peptide or the T-cell receptor controls the binding to E3-19k. Additional immunoprecipitations performed with mAbs directed against well-defined epitopes on the surface of HLA-A2.1 revealed a dichotomy in the ability of the mAbs to co-immunoprecipitate HLA-A2.1 and E3-19k. The mAbs LGIII-220 (directed against the C-terminal end of the alpha 1-helix), CR11-351 (directed against the N-terminal end of the alpha 2-helix), and PA2.1 (directed against the middle of the alpha 2-helix and an underlying beta-loop) readily co-precipitated HLA-A2.1 and E3-19k. In contrast, mAbs MA2.1 (directed against the N-terminal end of the alpha 1-helix and the C-terminal end of the alpha 2-helix) and HO-2 (directed against the N-terminal end of the alpha 1-helix) did not co-precipitate E3-19k with HLA-A2.1. Similarly, mAb MB40.2 (directed against residues 169-182 of HLA-B7) also did not co-precipitate E3-19k with HLA-B7. These studies lead to the conclusion that the N-terminal end of the alpha 1-helix and the C-terminal end of the alpha 2-helix play an important role in dictating the ability of the E3-19k protein to bind to the class I MHC molecule.

A software package for the calculation and statistical analysis of 51Cr-release assays.

A software package has been developed to calculate, store, and statistically analyze the results of 51Cr-release assays. It is written in Applesoft BASIC, is 'user friendly', and contains error trapping routines to catch most common mistakes. The greeting program contains a master menu from which one of the following programs may be called up: (1) 51Cr-release calculation, (2) statistical analysis, or (3) file manager. These programs give the user the capability to rapidly reduce cpm data to % specific release and statistically analyze the results using randomized design analysis of variance (ANOVA) and Turkey's honestly significant difference (HSD) test.

Localization and characterization of serologic epitopes on HLA-A2.

A panel of cells expressing 68 different mutant HLA-A2 genes was generated by site-directed mutagenesis and DNA-mediated gene transfer in order to define the regions of class I MHC molecules that contribute to the epitopes recognized by mAb. Each of the variant HLA-A2 molecules differed from HLA-A2.1 by a single amino acid substitution. The substitutions were located in both the alpha-helices and beta-strands of the alpha 1 and alpha 2 domains, and included residues that are highly polymorphic and that are conserved. All but five of the variant HLA-A2 molecules were expressed at levels that ranged from approximately 25%-100% the levels found for HLA-A2.1. The remaining five variants had no detectable expression and all involved substitutions at highly conserved residues. Eleven mAbs with specificities that ranged from highly HLA-A2 specific to monomorphic were analyzed for their ability to bind the variant HLA-A2 molecules. The results demonstrate that the binding of five of 11 mAbs could be mapped to the alpha 1 and alpha 2 domains. MA2.1 was the only antibody mapped to the alpha 1 domain. CR11-351 and A2,A28M1 recognized an overlapping epitope at the amino terminal end of the alpha 2-helix, and PA2.1 and BB7.2 recognized an overlapping epitope that includes the carboxy terminus of the alpha 2-helix and a turn on one of the underlying beta-strands. These results demonstrate that positions located on the surface of the molecule, but not within the peptide-binding cleft of the molecule, are important in serological specificities.

Nonspecific cytotoxic cells in fish (Ictalurus punctatus). IV. Target cell binding and recycling capacity.

The morphology of nonspecific cytotoxic cells (NCC) was identified. NCC were purified by target cell conjugate formation and density gradient separation. NCC are monocyte-like. They have reniform nuclei and a low nucleus/cytoplasm ratio. Cytoplasmic granules were not seen after giemsa staining. Scanning electron microscopy demonstrated moderate surface villi and target cell attachment occurred via long membraneous filament-like surface projections extending to the target cell membranes. Transmission electron microscopy of effector:target cell conjugates revealed membrane contact areas without fusion or fragmentation. The nucleus of the NCC had accentuated peripheral chromation and a prominent Golgi apparatus; the cytoplasm contained osmiophilic granules. Michaelis-Menten and Lineweaver-Burk transformation of target cell binding revealed a Vmax of 11-15,000 and a Km of 40,000. The percentage of NCC bound to target cells was 16-18%. Results of these studies were combined with the conjugate experiment to obtain an estimated percentage of active NCC (5-7%). A maximum recycling capacity of .16-.30 indicated that once attachment by NCC to the target cell occurred (and a lethal signal delivered by an effector cell), either the NCC did not recycle or a long lag period was required to restore its cytotoxic capability.

Epitope fine specificity of human anti-HLA-A2 antibodies. Identification of four epitopes including a haptenlike epitope on HLA-A2 at lysine 127.

Anti-HLA-A2 CREG antibodies were purified from seven individuals by affinity chromatography. The binding of the purified antibodies to single or multiple amino acid variants of HLA-A2.1 was measured with an inhibition RIA. Substitutions at 10 amino acid residues in the polymorphic alpha 1 and alpha 2 domains were important for human antibody binding; eight of these have previously been shown to be important in the binding of murine anti-HLA-A2 CREG antibodies. Unlike any previously reported murine mAbs, the binding of antibodies from two individuals was eliminated by a substitution at the HLA-A2, -24, -28 shared loop amino acid residue lysine 127. Conversely, when the asparagine at residue 127 on the non-cross-reactive HLA-A3 was replaced with lysine, antibody binding was completely restored. The results further suggest that both lambda- and kappa-containing human antibodies that bind to this region may recognize lysine 127 as a haptenlike epitope. Anti-HLA-A2 antibodies that recognized a conformational epitope defined by changes at glycine 62 in the alpha 1 domain were predominated by lambda light chains whereas those that recognize an epitope defined by a loop residue at tryptophan 107 in the alpha 2 domain were predominated by kappa light chains. The data are consistent with a model of restricted epitope recognition of HLA-A2 by human B cells that is similar to, but distinct from, epitope recognition by mouse B-cell hybridomas, and may help to explain the phenomenon of public or cross-reactive idiotypes in the HLA system.

Nonspecific cytotoxic cells in fish (Ictalurus punctatus). III. Biophysical and biochemical properties affecting cytolysis.

Nonspecific cytotoxic cells (NCC) from the catfish (Ictalurus punctatus) may comprise a population of cells that are responsible for cellular immunity in the fish. NCC kill a wide variety of transformed target cells, and previous studies have indicated that NCC share properties with mammalian natural killer cells. In the present study, many biophysical and biochemical properties of NCC were defined. NCC were nylon wool nonadherent and adherent. NCC activity was also enriched in plastic nonadherent cells. NCC were nonphagocytic (for carbonyl iron), and they did not bind to Sephadex G-10. Characterization of NCC by density gradient centrifugation indicated that they comprise a relatively homogenous population of cytolytic cells that band at 45.5% Percoll. Moderate to high doses (500-2500 R) of X-irradiation produced a stimulatory effect on NCC lysis of labeled target cells. Additional studies indicated that a soluble suppressor protein in catfish serum (CFS) regulated NCC activity. This S. aureus protein A binding component isolated from CFS suppressed NCC activity. Analysis by SDS-PAGE indicated that the soluble regulatory protein had properties similar to immunoglobulin. These data indicate that NCC share some biophysical properties with mammalian natural killer cells. In addition, NCC appear to be under partial cell regulation by a radiation sensitive suppressor cell and also by a soluble regulator serum immunoglobulin component.

Clonal analysis of the cytotoxic T-lymphocyte response to reovirus.

Reovirus has previously been classified into three serotypes based on hemagglutination inhibition assays. In the present study, the specificity of cytotoxic T lymphocytes (CTL) generated in reovirus type 3-infected C3H/HeN mice was investigated at the population and clonal levels. Short-term CTL lines generated in response to reovirus type 3 preferentially lysed cells infected with reovirus type 1 or 3 and, in some instances, type 2-infected cells as well. Eleven CTL clones established from the lines demonstrated two unique patterns of recognition. A single clone was exquisitely specific for reovirus type 3-infected cells and did not cross-react on reovirus type 1- or 2-infected cells. Ten of the clones recognized reovirus type 1- and type 3-infected cells. These clones had low levels of cross-reactivity on reovirus type 2-infected cells that was revealed only at high effector:target cell ratios. Precursor frequency analysis further revealed that the majority of the CTL generated against reovirus type 3 could cross-react on both reovirus type 1- and type 2-infected cells. Some CTL could be detected that had a more restricted pattern of recognition and recognized reovirus type 3-infected cells exclusively or recognized reovirus type 3-infected cells and either reovirus type 1- or type 2-infected cells. These results indicate that a minimum of four epitopes are recognized by reovirus-specific CTL and that the response is dominated by CTL that recognize an epitope common to all three serotypes of reovirus.

The reovirus nonstructural protein sigma1NS is recognized by murine cytotoxic T lymphocytes.

The cytotoxic T-lymphocyte (CTL) response in reovirus-infected C3H mice was investigated by using reovirus-vaccinia virus recombinants. Results of cytotoxicity assays indicated that the nonstructural protein sigma1NS elicited a significant CTL response. Experiments with sigma1NS-specific CTL lines showed that both strain-specific and cross-reactive epitopes exist in the sigma1NS protein.

A class I MHC-restricted recall response to a viral peptide is highly polyclonal despite stringent CDR3 selection: implications for establishing memory T cell repertoires in "real-world" conditions.

In this study, we analyze the recall response to influenza A matrix peptide M1(58-66) restricted by HLA-A2 in one individual and find a strict CDR3 selection as well as a high degree of polyclonality. The TCR beta-chain repertoire of memory T cells specific for this Ag system has been shown previously to be constrained by the use of the BV17 family and the I/sRS(A)/S amino acid motif in the CDR3 region. Our sequence analysis of BV17 TCR from a CTL line showed the repertoire to be highly polyclonal, as 95 distinct CDR3 sequences (clonotypes) were identified expressing this CDR3 motif. The clonotype frequencies showed a power law distribution with an extensive low-frequency tail. The clonotypes present in the high-frequency component of the distribution could be measured directly in the PBMC. This measurement showed that the relative frequencies of these clonotypes before stimulation were similar to their frequencies after culturing. Analysis of short-term cultures showed that the responding clonotypes have a similar ability to proliferate, which is independent of TCR beta-chain CDR3 sequence or precursor frequency. These data indicate that the memory T cell repertoire is composed of a surprisingly diverse set of T cell clonotypes with a limited potential for expansion. We propose that the high-frequency component represents T cells that have existed the longest. In keeping with this hypothesis, these clonotypes were measured over a 2-year period, during which their precursor frequency did not change.

Effects of tumor status on the regulation of natural killer cell activity by tumor-associated fetal antigens.

In the present study we examined the effects of a tumor-associated fetal antigen (TAFA-II) on the activity of natural killer (NK) cells isolated from human cancer patients. TAFA-II suppressed the NK cell response of some patients, and the level of suppression appeared to be independent of tumor type or stage of cancer therapy. No significant correlations were found between lymphocyte, neutrophil, monocyte, or eosinophil populations and TAFA-induced suppression of NK cell activity. TAFA-II effects were also not attributable to Ia+ cells or to OKT3, OKT4, or OKT8 positive cells. This work confirmed results obtained in the rat model, in which suppression appeared to be directly mediated on the NK cell.

Suppression of polyclonal, tumor cell and alloantigen-induced proliferation: identification of cyclooxygenase pathway dependent and independent mechanisms.

Polyclonal T cell activation, syngeneic tumor cell and alloantigen-induced proliferative responses were studied to determine if the regulation of these responses in normal and tumor-bearing NBR rats is mediated through products of the cyclooxygenase pathway and prostaglandin E2 (PGE2) in particular. Young rats and tumor-bearing rats have previously been shown to produce poor proliferative responses to PHA, Con A and syngeneic methylcholanthrene (MCA)-induced fibrosarcoma cells. The poor responses to PHA and Con A are mediated by PGE2 in unfractionated ( UNF ) and nylon wool adherent (ADH) cells. The same relationship was also established in the mixed leukocyte tumor cell (MLTC) response to MCA tumor cells although it appears to be of only minor significance as the enhancement following indomethacin (IND) treatment is still a relatively poor response. Indomethacin generally had no effect on the proliferative responses of tumor-bearing animals indicating that the suppression was not mediated through the cyclooxygenase pathway. We have also extended a previous observation in which UNF cells were found to be unresponsive to alloantigen stimulation. This suppression does not appear to be mediated through cyclooxygenase products as IND treatment does not enhance the UNF response although it does enhance the ADH response. These data indicate that a complex network of cyclooxygenase dependent and independent regulation exists in normal and tumor-bearing NBR rats.

Cytotoxic T lymphocyte-defined epitope differences between HLA-A2.1 and HLA-A2.2 map to two distinct regions of the molecule.

Hemi-exon shuffling and site-directed mutagenesis have been used to determine which amino acid differences between HLA-A2.1 and HLA-A2.2 alter the CTL-defined epitopes on these two molecules. Two genes were constructed that encode novel molecules in which the effect of amino acid differences at residues 9, 43, and 95, or at residue 156 could be separately evaluated. Using both human and murine CTL that were specific for either HLA-A2.1 or HLA-A2.2, four types of epitopes were identified: 1) epitopes that were insensitive to substitutions at either residues 9, 43, and 95, or residue 156 but were lost when all four positions were changed; 2) epitopes that were dependent on the residues 9, 43, 95, but not residue 156; 3) epitopes that were dependent on residue 156, but not amino acid residues 9, 43, and 95; and 4) epitopes that were dependent on residues 9, 43, and 95, as well as amino acid residue 156. Overall, there was a roughly equal distribution of clones recognizing each of these types of epitopes. Additional molecules were constructed by hemi-exon shuffling between the HLA-A2.2 and HLA-A2.3 genes, and by site-directed mutagenesis, to analyze the epitopes recognized by two HLA-A2.2/A2.1 cross-reactive murine CTL that do not recognize HLA-A2.3. Although the epitopes recognized by these CTL were unaffected by changes occurring at residues 9, 43, and 95, or at residues 149, 152, and 156 alone, simultaneous changes in both of these regions acted in concert to destroy the epitopes. Both of the CTL recognized epitopes that were lost when substitutions were made at residues 9, 43, 95, 149, and 152. The epitope recognized by one of the CTL was also destroyed by the substitution of residues 9, 43, 95, 152, and 156. Overall, these results indicate that residues 9, 43, and 95, as well as residues in the alpha-helical region of the molecule, are all capable of contributing to the definition of the epitopes recognized by HLA-A2.1- and HLA-A2.2-specific CTL. They further indicate that some epitopes can be mapped to a particular region of the molecule, whereas other epitopes are formed through a complex interaction of residues in distant regions of the molecule.

Identification by site-directed mutagenesis of amino acid residues contributing to serologic and CTL-defined epitope differences between HLA-A2.1 and HLA-A2.3.

Site-directed mutagenesis of HLA-A2.1 has been used to identify the amino acid substitutions in HLA-A2.3 that are responsible for the lack of recognition of the latter molecule by the HLA-A2/A28 specific antibody, CR11-351, and by HLA-A2.1 specific CTL. Three genes were constructed that encoded HLA-A2 derivatives containing one of the amino acids known to occur in HLA-A2.3: Thr for Ala149, Glu for Val152, and Trp for Leu156. Three additional genes were constructed that encoded the different possible combinations of two amino acid substitutions at these residues. Finally, a gene encoding all three substitutions and equivalent to HLA-A2.3 was constructed. These genes were transfected into the class I negative, human cell line Hmy2.C1R. Analysis of this panel of cells revealed that recognition by the antibody CR11-351 was completely lost when Thr was substituted for Ala149, whereas substitutions at amino acids 152 and 156, either singly or in combination, had no effect on the binding of this antibody. The epitopes recognized by the allogeneic and xenogeneic HLA-A2.1 specific CTL clones used in this study were all affected by either one or two amino acid substitutions. Of those epitopes sensitive to single amino acid changes, none were affected by the substitution of Thr for Ala149, whereas all of them were affected by at least one of the substitutions of Glu for Val 152 or Trp for Leu156. Overall, amino acid residue 152 exerted a stronger effect on the epitopes recognized by HLA-A2.1 specific CTL than did residue 156. Of those epitopes affected only by multiple amino acid substitutions, double substitutions at residues 149 and 152 or at 152 and 156 resulted in a loss of recognition, whereas a mutant with substitutions at residues 149 and 156 was recognized normally. This reemphasizes the importance of residue 152 and indicates that residue 149 can affect epitope formation in conjunction with another amino acid substitution. These results are discussed in the context of current models for the recognition of alloantigens and in light of the recently published three-dimensional structure of the HLA-A2.1 molecule.

A panel of unique HLA-A2 mutant molecules define epitopes recognized by HLA-A2-specific antibodies and cytotoxic T lymphocytes.

HLA-A2.1 and HLA-A2.3, which differ from one another at residues 149, 152, and 156, can be distinguished by the mAb CR11-351 and many allogeneic and xenogeneic CTL. Site-directed mutagenesis was used to incorporate several different amino acid substitutions at each of these positions in HLA-A2.1 to evaluate their relative importance to serologic and CTL-defined epitopes. Recognition by mAb CR11-351 was completely lost when Thr but not Pro was substituted for Ala149. A model to explain this result based on the 3-dimensional structure of HLA-A2.1 is presented. In screening eight other mAb, only the substitutions of Pro for Val152 or Gly for Leu156 led to the loss of mAb binding. Because other non-conservative substitutions at these same positions had no effect, these results suggest that the loss of serologic epitopes is in many cases due to a more indirect effect on molecular conformation. Specificity analysis using 28 HLA-A2.1-specific alloreactive and xenoreactive CTL clones showed 19 distinct patterns of recognition. The epitopes recognized by alloreactive CTL clones demonstrated a pronounced effect by all substitutions at residue 152, including the very conservation substitution of Ala for Val. Overall, the most disruptive substitution at amino acid residue 152 was Pro, followed by Glu, Gln, and then Ala. In contrast, substitutions at 156 had little or no effect on allogeneic CTL recognition, and most clones tolerated either Gly, Ser, or Trp at this position. Similar results were seen using a panel of murine HLA-A2.1-specific CTL clones, except that substitutions at position 156 had a greater effect. The most disruptive substitution was Trp, followed by Ser and then Gly. In addition, when assessed on the entire panel of CTL, the effects of Glu and Gln substitutions at position 152 demonstrated that the introduction of a charge difference is no more disruptive than a comparable change in side chain structure that does not alter charge. Taken together, these results indicate that the effect of amino acid replacements at positions 152 and 156 on CTL-defined epitopes depends strongly on the nature of the substitution. Thus, considerable caution must be exercised in evaluating the significance of particular positions on the basis of single mutations. Nonetheless, the more extensive analysis conducted here indicates that there are differences among residues in the class I Ag "binding pocket," with residue 152 playing a relatively more important role in formation of allogeneic CTL-defined epitopes than residue 156.

The ability of cytotoxic T cells to recognize HLA-A2.1 or HLA-B7 antigens expressed on murine cells correlates with their epitope specificity.

Two groups of human and murine cytotoxic T lymphocyte (CTL) clones specific for human leukocyte antigen (HLA)-A2 or -B7 can be distinguished based on their ability to kill murine transfectants expressing these molecules. The clones which do not recognize murine transfectants exhibited greatly reduced conjugate formation with these cells, indicating that the inability to lyse these cells occurs in recognition and binding. No systematic differences in inhibitory titer between the two types of CTL clones were seen with anti-CD8 (Lyt-2), anti-LFA-1, or monoclonal antibodies against HLA class I molecules. However, blocking with anti-HLA class I monoclonal antibodies suggested that different CTL clones recognized spatially separate epitopes on HLA-A2 and -B7. In addition, a correlation between the inability to recognize murine transfectants and fine specificity was seen. Eight of nine clones which did not lyse murine transfectants also failed to recognize human cells expressing HLA-A2.2 or -A2.3. In contrast only 5 of 12 clones which lysed transfectants failed to recognize the variant molecules. Analogous data were obtained with human CTL clones raised against HLA-A2.1. These findings suggest that CTL clones that do not lyse murine cells expressing appropriate antigens recognize epitopes that have been altered or lost as a consequence of expression on the murine cell surface. It is suggested that the loss of HLA-associated epitopes on the murine cell surface may be due to differences between mouse and human cells in the processing or presentation of class I-associated peptides.