Development of a Murine Mutant Ras CD8⁺ CTL Peptide Epitope Variant That Possesses Enhanced MHC Class I Binding and Immunogenic Properties

The ras p21 proto-oncogenes encode a family of intracellular GTP binding proteins involved in signal transduction that are thought to be important in regulating cellular processes such as proliferation and differentiation (1, 2). While a broad spectrum of human cancers have been identified that contain point mutations in ras proto-oncogenes at codons 12, 13, and 61, a predominant subset of adenocarcinomas of the pancreas, colon, and lung are frequently found mutated at codon 12, in which the normal glycine is replaced with valine, aspartic acid, or cysteine (3). Biochemically, such point mutations alter the intrinsic GTPase activity of Ras p21, ultimately contributing to cellular transformation. Immunologically, oncoproteins such as those encoded by the ras oncogenes may present neo-epitopes for the recognition and induction of tumor-specific, cell-mediated immune responses.

CD8⁺ CTL have been implicated as an important cellular component involved in the recognition and eradication of tumor cells in both murine and human systems (4, 5). The heterodimeric transmembrane TCR complex expressed by CD8⁺ CTL recognize antigenic peptides, typically 8 to 11 amino acids in length, bound to class I MHC molecules presented at the extracellular surface of APC or tumor cells (6, 7).

We recently reported on the identification of a T cell epitope, Ras_4–12(V12), capable of inducing a CD8⁺ CTL response restricted by the H-2K^d allele by immunizing BALB/c mice with a synthetic peptide in IFA (8). Whereas the normal Gly at codon 12 (position 9 of the nonamer) did not constitute an anchor residue, the introduction of Val at the C-terminus of this nonamer rendered the peptide an H-2K^d binding peptide. A CTL line generated by Ras_4–12(V12) peptide immunization was shown to lyse syngeneic A20 tumor cell targets transduced retrovirally to express endogenous point-mutated Ras epitopes. In addition, we identified a CD4⁺ T cell epitope comprising the Ras sequence 5 to 17(V12) that was immunogenic in BALB/c mice (9). Class II-restricted CD4⁺ T cell lines were derived from peptide immunization, which also lysed A20 (Ia^d) tumor cells expressing the Val¹² mutated ras oncogene. These studies demonstrated that functional T cell responses were inducible against multiple, overlapping mutant Ras peptide epitopes, which may have implications for peptide-based active immunotherapies for tumors expressing the mutated ras oncogene.

While the mutation of Gly to Val at codon 12 creates a C-terminal anchor residue for binding to H-2K^d, this peptide binds weakly to the class I molecule and elicits a weak in vivo primary CD8⁺ CTL response from immunized mice. Thus, the relative degree of immunogenicity in vivo may have correlated with the binding characteristics of the peptide and/or the availability of a limited precursor CTL population. Since other amino acids, such as Leu or Ile, have been reported to serve as dominant C-terminal anchors and bind to H-2K^d with higher activity (10), we hypothesized that the replacement of Val¹² with such a residue might strengthen peptide binding to H-2K^d, thus rendering a more stable immunogenic complex. Since our earlier work supported the hypothesis that the Val¹² amino acid substitution created a C-terminal anchor residue, this suggested that altering this anchor position would be unlikely to affect TCR recognition. Thus, the anti-Ras_4–12(V12) CTL response probably reflected TCR recognition of a previously unseen peptide/MHC complex. In this study, we examined this hypothesis in further detail and demonstrated that such a ras oncogene-derived peptide variant could be defined that displays enhanced MHC class I binding activity and in vitro and in vivo immunogenicity without compromising Ag specificity and TCR recognition.

Female BALB/c mice (H-2^d) were obtained from Taconic Farms (Germantown, NY) and were at least 8 wk old when immunizations began.

The peptides used in this study reflected the Ras sequence 4 to 12 (YKLVVVGAG) with substitutions at codon 12 from Gly to Val, Ile, or Leu. Ras_4–12(V12) and Ras_4–12(L12) peptides were purchased (>95% pure) from Multiple Peptide Systems, Inc. (San Diego, CA). Other experimental peptides, including the Ala- and Gly-substituted nonamer peptides, were chemically synthesized in our laboratory on an Applied Biosystems 432A (Foster City, CA) peptide synthesizer by F-moc chemistry. Peptides were purified and analyzed by reverse phase HPLC using a C₁₈ column (>90% pure). All peptides were dissolved in distilled water at 2 mg/ml, filter-sterilized, and stored in aliquots at −80°C. BALB/c mice were injected s.c. near the base of the tail with three injections of peptide separated by 2 wk (100 μg of peptide/injection in 100 μl; final volume adjusted with sterile PBS). Before injection, peptides were admixed with a modified stable formulation of Detox-PC adjuvant (provided by RIBI ImmunoChem Research, Inc., Hamilton, MT) as previously described (11).

Three to four weeks after the third injection of peptide, a single cell suspension of pooled splenocytes was resuspended to 25 × 10⁶/T-25 flask in 10 ml of culture medium consisting of RPMI 1640 supplemented with 15 mM HEPES (pH 7.4), 2 mM l-glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin (above reagents from Life Technologies Co., Gaithersburg, MD), 10% heat-inactivated FBS (HyClone Laboratories, Logan, UT), and 50 μM 2-ME (Sigma Chemical Co., St. Louis, MO). The peptide concentration used during the first in vitro stimulation was 25 μg/ml; this was subsequently reduced gradually over a period of 4 wk to 1 μg/ml. T cell cultures were restimulated weekly in 24-well plates (2 × 10⁵/well; Costar Corp., Cambridge, MA) containing the appropriate Ras peptide, irradiated (2000 rad) syngeneic BALB/c splenocytes (5 × 10⁶/well), plus IL-2. Recombinant human IL-2 (Cetus Corp., Emeryville, CA) was added to 10 U/ml to the culture beginning at the second in vitro stimulation. T cell cultures were tested for cytolytic activity 4 to 6 days after Ag stimulation. After the fourth in vitro stimulation, CD8⁺ T cells were further enriched by negative selection by removal of residual CD4⁺ T cells. Briefly, T cells were preincubated with rat (IgG2a) anti-mouse CD4 mAb for 90 min at 4°C (GK 1.5 hybridoma, American Type Culture Collection, Rockville, MD). After this incubation, lymphocytes were washed and incubated on mouse anti-rat IgG Ab-coated flasks (Applied Immune Sciences, Inc., Santa Clara, CA) as described by the manufacturer. The nonadherent CD8⁺-enriched T cell fraction was collected and restimulated as described above. Cell surface phenotype was evaluated by flow cytometry (FACScan, Becton Dickinson Corp., Mountain View, CA) following immunostaining with FITC-conjugated mAb (PharMingen, Inc., San Diego, CA).

The target cell lines used in CTL assays were P815 (H-2^d), a mastocytoma of DBA/2 origin that expresses MHC class I, but not class II, molecules, and A20 (H-2^d), a B cell lymphoma of BALB/c origin that expresses both MHC class I and class II molecules. Both cell lines were obtained from American Type Culture Collection. A20 cells expressing the point-mutated human K-ras oncogene (A20-ras) endogenously were produced by retroviral transduction as described previously (9).

CTL activity was examined in a standard 4-h chromium release assay. Briefly, target cells (2–3 × 10⁶) were radiolabeled with 250 μCi of Na₂[⁵¹Cr]O₄ (Amersham Corp., Arlington Heights, IL) in Opti-MEM (Life Technologies Co., Gaithersburg, MD) at 37°C for 90 min, then washed. Viable effector T cells were recovered from culture by density centrifugation over a Ficoll-Hypaque gradient. Effector and target cells were coincubated in 96-well, U-bottom plates, either at graded E:T cell ratios in the presence or the absence of peptide or at a constant E:T cell ratio with graded peptide concentrations. Plates were centrifuged at 100 × g for 2 min to initiate contact between cells, then incubated at 37°C for 4 h. In mAb blocking experiments, purified mAb (10 μg/ml final concentration of Abs from PharMingen: anti-H-2K^d (clone SF1-1.1), anti-H-2L^d (clone 28-14-8), and anti-H-2D^d (clone 34-2-12); or 25 μg/ml final concentration of anti-TCR Vβ9 (clone MR10-2)) were preincubated for 30 min at 4°C with either target cells (for anti-MHC mAb) or effector cells (for anti-TCR mAb) before their addition to the assay. Similarly, anti-CD4 mAb (GK1.5 hybridoma supernatant; 10% final concentration, v/v) or anti-CD8 mAb (2.43 hybridoma supernatant; 10% final concentration, v/v) was preincubated with T cells for 30 min at 4°C before their addition to the assay. After the 4-h incubation, plates were centrifuged at 400 × g for 5 min, and supernatants were harvested using a Supernatant Collection System (Skatron Co., Sterling, VA). Radioactivity was quantitated in a gamma counter (Packard Instrument Co., Downers Grove, IL), and the percent specific ⁵¹Cr release was calculated as the mean ± SEM of triplicate wells according to the formula: percent specific lysis = ((experimental cpm − spontaneous cpm)/(maximum cpm − spontaneous cpm)) × 100. Maximum ⁵¹Cr release was obtained by adding Triton X-100 to target cells (0.2% final concentration). Spontaneous ⁵¹Cr release was obtained from target cells incubated in the absence of T cells, but in the presence or the absence of peptides and/or mAb.

The ability of various Ras peptides to bind to MHC class I H-2K^d was evaluated in a functional bioassay that measures specific inhibition of cytolytic activity by a positive control CTL response, as described previously (8). Briefly, a CTL line specific for the influenza nucleoprotein immunodominant peptide, NP_147–155 (sequence TYQRTRALV), was established and used as an H-2K^d-restricted biologic readout. Competitor peptides were preincubated for 30 min at 37°C at various concentrations with ⁵¹Cr-labeled P815 target cells in a 96-well, round-bottom plate. NP_147–155 peptide (final concentration, 0.3 ng/ml) and anti-NP_147–155 CTL (E:T ratio, 3:1) were then added. Plates were incubated at 37°C for an additional 4 h, then harvested and counted as described above. Controls included incubation of targets with peptides containing a model H-2K^d anchor motif (sequence AYAAAAAAL) or a model H-2L^d anchor motif (sequence APAAAAAAL), incubation of T cells with the highest concentration of competitor peptides in the absence of NP_147–155 peptide, incubation of T cells with NP_147–155 peptide in the absence of competitor peptides, and the spontaneous release of ⁵¹Cr in the presence of the highest dose of each peptide.

CTL lines were cultured in flat-bottom 96-well plates at a cell density of 2 × 10⁴ cells/well with 5 × 10⁵ irradiated (2000 rad) syngeneic BALB/c splenocytes/well containing Ras nonamer peptides at 10 or 2 μg/ml in the absence of exogenous IL-2. Viable cells were enumerated by trypan blue dye exclusion on days 2 to 6 following in vitro stimulation. Additionally, proliferation was measured after 48 h of incubation at 37°C by adding [³H]thymidine (1 μCi/well; Amersham Corp.) to the wells 18 h before harvesting. Cells were harvested and counted by liquid scintillation spectroscopy (Wallac, Inc., Gaithersburg, MD).

BALB/c mice were immunized three times with 100 μg of Ras peptides or influenza NP_147–155 peptide in Detox-PC adjuvant as described above. Three weeks after the third immunization, splenic T cells from individual mice were isolated by nylon wool purification. As an additional positive control, a MLR (H-2^d anti-H-2^b) was prepared and cultured in a parallel fashion. Lymphocytes were cultured in 96-well plates at graded cell densities (2.4 × 10⁵/well starting density, with fourfold dilutions for a total of four dilutions) with 5 × 10⁵ irradiated BALB/c splenocytes/well (or irradiated C57Bl/6 splenocytes for the MLR), 5 U/ml IL-2, and specific Ras peptides or NP_147–155 peptide at 20 μg/ml. On day 7, medium containing any residual free peptide was aspirated from the wells, and the cells were resuspended in fresh medium and divided equally into two new parallel 96-well plates. One plate received fresh irradiated BALB/c APC, while the second plate received fresh irradiated APC and specific Ras peptide or NP_147–155 peptide to 20 μg/ml. Neither plate received IL-2. After 48-h incubation, supernatants were harvested and analyzed for IFN-γ production by ELISA for murine IFN-γ as previously described (9). Cultures were scored positive based on the sensitivity of each assay for a mouse IFN-γ standard. Well-to-well differences in OD₄₉₅ between cultures that received APC and Ag for the assay and cultures that received only APC were calculated, and the average well-to-well difference in OD₄₉₅ was calculated for each plated cell density. This method selected as positive those cultures that expressed IFN-γ in an Ag-dependent manner. The data presented are compiled from individual mice analyzed in parallel (Ras_4–12(L12), Ras_4–12(V12), NP_147–155, and the mixed lymphocyte reaction) from three identical, independent experiments.

The expression of specific Vα and Vβ chains of the TCR was assessed by conversion of RNA prepared from T cells into cDNA by reverse transcriptase followed by PCR amplification using a panel of 19 Vα chain-specific and 20 Vβ chain-specific oligodeoxyribonucleotides (12). Approximately 5 × 10⁷ T cells were purified by Ficoll-Hypaque gradient density centrifugation and were used for total RNA isolation using the Triazol reagent (Life Technologies, Gaithersburg, MD). Synthesis of cDNA was performed essentially as suggested by the manufacturers of murine leukemia virus reverse transcriptase (Perkin-Elmer/Cetus Corp., Emeryville, CA) using oligo(dT) to prime the cDNA synthesis from polyadenylated mRNA. Approximately 20 ng of total RNA was included in each 20-μl reaction tube. cDNA synthesis was performed at 42°C for 15 min followed by heat denaturation of the enzyme at 95°C for 15 min. Identification of individual Vα and Vβ TCR chains was performed in a subsequent PCR using oligodeoxyribonucleotides derived from target sequences from specific mouse Vα or Vβ genes paired with Vα common or Vβ common primers (12). Amplification of target sequences by Taq DNA polymerase in 25 cycles of thermal cycling was achieved by 94°C for 20 s, 55°C for 20 s, and 72°C for 10 s. All target sequence products were between 200 and 600 bp. PCR was performed in a Perkin-Elmer/Cetus thermal cycler 9600. All reaction products were analyzed by 4% agarose gel electrophoresis, and positives were scored by visualization under an UV light source following ethidium bromide staining. Intensely stained bands were assigned three plus signs, and weakly stained bands were assigned one plus sign. Lanes with no detectable bands or nonspecific bands of DNA were assigned a negative sign.

Figure 1 A depicts a functional H-2K^d binding assay, as determined by the ability of different competitor peptides to inhibit a positive control H-2K^d-restricted CD8⁺ CTL response. The Ras_4–12 nonamer peptide variants containing Leu or Ile at the C-terminus competed for H-2K^d binding 25 to 50 times better than the mutated Ras_4–12(V12) peptide based on the peptide concentrations required to inhibit half the maximum activity of the control CTL. By contrast, the wild-type Ras_4–12(G12) peptide did not compete, since it lacked a putative C-terminal anchor for binding H-2K^d. To demonstrate the peptide binding specificity of this assay, a positive control peptide containing the H-2K^d consensus binding motif (poly-Ala-K^d) was shown to inhibit the CTL activity at each concentration tested, whereas a negative control peptide that lacked the H-2K^d consensus anchor motif (poly-Ala-L^d) did not bind to H-2K^d as measured by this competition assay. These results demonstrated that replacement of Val in a weak MHC class I binding peptide with Leu or Ile as the more preferred and dominant C-terminal binding anchor substantially improved H-2K^d binding.

To evaluate the potential usefulness of mutant Ras peptide variants, we first examined whether the introduction of Leu or Ile at position 9 altered TCR recognition, as measured by cytotoxicity. To that end, a previously established anti-Ras_4–12(V12)-derived CTL line (8) was tested for its ability to lyse target cells incubated with the Ras peptide variants. Figure 1 B shows that P815 target cells incubated with Ras_4–12(L12) or Ras_4–12(I12) peptide variants were each approximately 500 times more sensitive to lysis by the CTL line, based on the half-maximal activity, compared with target cells incubated with the mutated Ras_4–12(V12) peptide. Target cells incubated with higher concentrations of the wild-type Ras_4–12(G12) peptide were not sensitive to lysis with the CTL line. These results demonstrated that the Ras peptide variants could sensitize target cells at a lower concentration and, importantly, that substitution of the C-terminal Val anchor residue with Leu or Ile did not interfere with TCR recognition by the anti-Ras_4–12(V12)-derived CTL line.

In addition to cytotoxicity as a measure of TCR recognition, we examined whether the Ras_4–12(L12) peptide variant could replace the Ras_4–12(V12) peptide to stimulate the in vitro proliferation of the anti-Ras_4–12(V12)-derived CTL line. Since previous experiments (Fig. 1) revealed comparable functional properties of both Ras peptide variants, we arbitrarily chose Ras_4–12(L12) as the model peptide variant for comparison to the mutant Ras epitope in this and subsequent experiments. The CTL line was stimulated with the different Ras peptides in the absence of exogenous IL-2, and cell growth was determined by viable cell counts from 2 to 6 days after Ag stimulation. As shown in Figure 2,A, anti-Ras_4–12(V12)-derived CTL proliferated at similar rates when stimulated with a higher concentration of either Ras peptide (10 μg/ml), whereas a fivefold decrease in peptide concentration (2 μg/ml) resulted in poor T cell proliferation in cultures containing the Ras_4–12(V12) peptide (Fig. 2,A). In contrast, T cells proliferated similarly when stimulated with Ras_4–12(L12) peptide variant at either higher or lower concentrations. Similar proliferation patterns for each Ras peptide at higher or lower concentrations were observed by [³H]thymidine uptake experiments, and it was found that this CTL line did not proliferate when stimulated with the highest concentration of the wild-type Ras_4–12(G12) peptide (Fig. 2 B). These experiments demonstrated the superiority of the Ras_4–12(L12) peptide variant over that of the mutant Ras_4–12(V12) peptide to stimulate the in vitro growth and expansion of anti-Ras_4–12(V12)-derived CTL. Furthermore, the anti-Ras_4–12(V12)-derived CTL incubated with the Ras_4–12(L12) peptide variant for at least two in vitro stimulation cycles maintained specific cytotoxicity against targets incubated with the Ras_4–12(V12) peptide (not shown).

Since the in vitro data suggested that higher MHC binding of the Ras peptide variant correlated with immunogenicity, we tested its ability to induce a greater in vivo T cell response, since the peptide:MHC:β₂m ternary complex would presumably be a more stable immunogen for presentation to precursor T cell populations. To this end, splenic T cells obtained from mice immunized with a fixed concentration of Ras_4–12(V12) peptide or the Ras_4–12(L12) peptide variant (each in the same adjuvant) were compared for their abilities to produce IFN-γ in vitro as a biologic readout of effector function. T cells were stimulated in vitro with the immunizing peptide and IL-2 for 7 days at the same graded cell densities, then washed and restimulated in the presence or the absence of the immunizing peptide. After 2 days of incubation, culture supernatants were tested for the production of IFN-γ by ELISA in an Ag-dependent fashion. Since similar rates of proliferation were observed by anti-Ras_4–12 CTL lines stimulated in vitro by either Ras_4–12(V12) or Ras_4–12(L12) peptide at 10 μg/ml (Fig. 2), the concentration of Ras_4–12 peptides in this analysis was doubled (20 μg/ml) to ensure that this in vitro response would better reflect any biologic in vivo differences. Data compiled from three independent determinations showed, at each cell density tested, quantitative increases in both the number of positive T cell cultures and the mean levels of γ-IFN produced in vitro by T cells from mice immunized with the Ras_4–12(L12) peptide variant compared with the Ras_4–12(V12) peptide (Table I). The results demonstrated the in vivo superiority of the Ras_4–12(L12) peptide variant to induce an anti-Ras_4–12 T cell response compared with the Ras_4–12(V12) peptide, using IFN-γ production as a measurement of effector function. Mice immunized with the influenza nucleoprotein immunodominant CTL peptide epitope (NP_147–155) as well as an alloreactive culture (H-2^d anti-H-2^b) were included as positive controls for this analysis. It is notable that the Ras_4–12(L12) peptide variant induced a less than twofold difference in the number of positive T cell cultures compared with the immunodominant NP_147–155 peptide in these experiments.

By demonstrating that the Ras_4–12(L12) peptide variant could replace the mutant Ras peptide and stimulate enhanced T cell responses both in vivo and in vitro, we next generated a Ras_4–12(L12)-derived CTL line for comparison to the Ras_4–12(V12)-derived CTL line described previously (8). BALB/c mice were injected with Ras_4–12(L12) peptide in adjuvant, and a cytotoxicity assay was performed on bulk culture splenocytes after one in vitro stimulation with the Ras_4–12(L12) peptide (Fig. 3,A). Efficient CTL activity was observed against targets incubated with the Ras_4–12(V12) peptide, whereas targets incubated with the Ras_4–12(G12) peptide or without peptide were not sensitive to lysis. Importantly, this result showed that the immune specificity for the mutant Ras_4–12(V12) peptide was not compromised by T cells induced in vivo by the Ras_4–12(L12) peptide variant, and the T cell response did not cross-react with cells expressing the wild-type Ras gene. In a second independent experiment, an anti-Ras_4–12(L12) CTL line was established from mice immunized with Ras_4–12(L12) peptide and was tested for cytolytic activity against P815 target cells incubated with different Ras_4–12 peptides. Figure 3,B shows that the half-maximal lytic activity observed with Ras_4–12(L12 or I12) peptides was achieved at approximately 600-fold less peptide compared with that observed for targets incubated with Ras_4–12(V12) peptide, similar to the sensitivity observed with the anti-Ras_4–12(V12)-derived CTL line (see Fig. 1 B). This CTL line also did not lyse target cells incubated with higher concentrations of wild-type Ras_4–12(G12) peptide.

Also, in a manner parallel to that used with the anti-Ras_4–12(V12)-derived CTL line, both peptides were compared for their ability to stimulate the in vitro proliferation of the antiRas_4–12(L12)-derived CTL line. As with the anti-Ras_4–12(V12)-derived CTL, the anti-Ras_4–12(L12)-derived CTL displayed greater proliferation when stimulated with the Ras peptide variant, as measured by this assay (Fig. 2 C). Higher concentrations of either peptide (10 μg/ml) stimulated cell growth in the absence of exogenous IL-2; however, only the Ras_4–12(L12) peptide variant stimulated proliferation at a fivefold lower concentration.

Both anti-Ras_4–12 CTL lines were compared and shown to be restricted by H-2K^d, since a mAb specific for this MHC class I molecule inhibited cytolytic activity (Fig. 4). Abs reactive with H-2L^d or H-2D^d molecules on P815 targets did not affect the lytic activity of either CTL line. Furthermore, mAb reactive with the CD8 molecule abolished lytic activity, while mAb reactive with CD4 had little effect (Fig. 4). These results confirm our earlier results with the anti-Ras_4–12(V12)-derived CTL line and extend them to show a similar requirement for CD8 and H-2K^d for cytotoxic activity by the anti-Ras_4–12(L12)-derived CTL line.

An important objective of this study was to demonstrate whether an anti-Ras_4–12(L12)-derived CTL activity could recognize foreign Ag processed and presented endogenously by tumor cells expressing the point-mutated ras (Val¹²) oncogene. Figure 5 demonstrated the ability of both anti-Ras CTL lines to lyse A20-ras, a syngeneic A20 tumor cell line transduced by a retroviral vector encoding the ras oncogene harboring the Val mutation at codon 12 (8). Neither CTL line could lyse the vector-only transduced tumor cell line, A20, demonstrating that the cytolytic activity was directed to an epitope derived from the point-mutated ras oncogene. The maximal cytolytic sensitivity of the A20-ras target cells was revealed by adding exogenous Ras_4–12(V12) peptide to the assay (Fig. 5). This suggested that the point-mutated ras CTL epitope was expressed on the surface of the tumor cells in a limited amount, since approximately one-half of the maximal cytolytic activity was induced in the absence of exogenous Ras_4–12(V12) peptide. Thus, immunization of mice with the Ras peptide variant led to the production of a class I-restricted, CD8⁺ CTL capable of recognizing and lysing tumor cells harboring the ras Val12 point mutation, but not cells expressing wild-type Ras protein.

Since the above experiments demonstrated similar phenotypic and functional properties of both anti-Ras_4–12 CTL lines, this suggested that these independently derived CTL may express similar TCR subunits. To test that possibility, we examined Vα and Vβ chain expression patterns. By reverse transcriptase-PCR analysis using oligodeoxynucleotides for 19 specific Vα chains and 20 specific Vβ chains to amplify the Vα and Vβ chain cDNAs, we observed that both CTL lines expressed predominantly the Vα1 and Vβ9 subunits in their TCR heterodimer (Table II). Flow cytometry, using commercially available mAb specific for the anti-TCR Vβ9 subunit, also revealed the predominant Vβ9 protein expression pattern of these CTL lines (>90% Vβ9 positive; not shown). Reverse transcriptase-PCR analysis of a third anti-Ras_4–12(V12) CTL line also revealed similar Vα1 and Vβ9 TCR subunit predominance (J. A. Bristol, unpublished observation).

The functional role for Vβ9 in cytotoxicity was examined using mAb directed against that molecule. Figure 6 shows that preincubation of both anti-Ras_4–12 CTL lines with purified mAb against the murine Vβ9 subunit inhibited killing, confirming their phenotypic expression of Vβ9 and its functional role in cytotoxicity. Similar inhibition was observed with an mAb directed against a nonpolymorphic epitope of the TCR αβ-chain heterodimer. The observed inhibition was specific for Vβ9, since preincubation of CTL lines with mAb directed against irrelevant TCR did not alter the cytotoxicity against target cells presenting the Ras_4–12(V12) peptide (<10% inhibition for each mAb against Vβ2, Vβ8, and Vβ10).

Single, sequential, Ala-substituted peptides, from position 5 to position 11 of the Ras_4–12(V12) peptide sequence, were synthesized to compare both CTL lines for potential similarities or differences in Ag recognition. Since the naturally occurring amino acid at position 11 is Ala, this residue was replaced by Gly. The anchor residues Tyr⁴ and Val¹² were left intact. After demonstrating that each peptide bound similarly to H-2K^d by the functional competition binding assay described above (not shown), the ability of each peptide to sensitize target cells for lysis by either Ras_4–12 CTL line was assessed (Fig. 7). The results showed that the amino acid contact sites of both CTL lines included residues 6 through 10, since substitutions at these positions abolished nearly all cytotoxic activity. Amino acid substitution of either Lys⁵ with Ala or Ala¹¹ with Gly had no apparent effect on T cell recognition of the Ras peptide and thus did not appear to participate in T cell recognition. The result of this comparison of the two CTL lines further suggested that the TCR expressed by each anti-Ras_4–12 CTL line was functionally similar.

Ag recognized by CTL are typically eight to 11 amino acids in length and contain at least two dominant anchor positions. The anchor residues are allele specific, in that certain amino acids within the peptide reflect binding potential to specific MHC class I molecules (10, 13). We recently identified a mutant Ras nonamer peptide reflecting the Gly to Val mutation at codon 12 that elicited an H-2K^d-restricted CD8⁺ CTL response in BALB/c mice (H-2^d haplotype) (8). The ras mutation encoding Val at codon 12 created a C-terminal anchor residue that enabled binding, albeit weakly, to H-2K^d as determined in a functional bioassay. In this report we modified the mutant Ras peptide to contain Leu in place of Val at position 9 of the peptide, which has been described as the dominant and more preferred C-terminus anchor residue for peptide/H-2K^d interactions (10).

The Ras_4–12(L12) peptide variant substantially enhanced binding to H-2K^d compared with the Ras_4–12(V12) peptide (Fig. 1) and effectively induced a H-2K^d-restricted, CD8⁺ CTL response that recognized the mutant Ras_4–12(V12) CTL epitope (Figs. 3, 4, 5, and 7). Additionally, compared with Ras_4–12(V12) peptide, the Ras peptide variant was more potent as an in vitro immunogen to stimulate proliferation and functional CTL activity of anti-Ras_4–12 CTL lines ( Figs. 1–3). Likewise, the Ras peptide variant was a more potent immunogen in vivo, at least quantitatively, and induced a greater T cell response compared with the mutant Ras_4–12(V12) peptide (Table I) as determined by Ag-dependent IFN-γ production at multiple T cell dilutions. Since T cell cultures from mice immunized with the Ras_4–12(L12) peptide variant resulted in a greater percentage of positive wells and produced greater mean levels of IFN-γ at each cell density tested, this suggested at least two possibilities: 1) a quantitative effect, which simply may be proportional to an increased number of T cells producing similar levels of IFN-γ; and/or 2) a qualitative effect, which may reflect enhanced IFN-γ production, as a measurement of effector function, on a per cell basis. We did not, however, observe any apparent functional qualitative differences between the established anti-Ras_4–12(V12)-derived CTL line and the anti-Ras_4–12(L12)-derived CTL line as determined by their sensitivity to exogenous peptide in cytolytic activities (Fig. 1,B vs Fig. 3,B) and proliferation (Fig. 2) or by their cytolytic potential to lyse A20 cells expressing endogenous levels of Ras containing the Val¹² point mutation (Fig. 5). However, it is possible that qualitative differences between T cell populations induced by either peptide may have existed during early stimulation cycles but were lost due to culture conditions over the course of multiple (>10) in vitro stimulations. Although many in vitro parameters may contribute to the generation of CTL, one example is the Ag concentration, which may have influenced the selection and/or proliferation of specific T cell clones. Consistent with that possibility, Alexander-Miller et al. (14) reported the generation of various CTL reactive against the peptide I10, derived from the immunodominant CTL epitope of HIV-1 gp160, by varying the Ag concentration in vitro. They generated high avidity CTL using a very low dose of I10 peptide (100 pM) and generated low avidity CTL using a high dose of I10 peptide (100 μM), as measured by the ability of these CTL to lyse target cells pulsed with varying concentrations of I10 peptide. In addition, they reported that two higher avidity CTL lines expressed only a small percentage of the predominant Vβ8 chain, whereas the low avidity CTL expressed higher percentages of the Vβ8 chain, implying that other CTL reactive toward the I10 peptide exist that express different TCR. Nonetheless, we demonstrated that a surrogate design of a CTL epitope peptide markedly enhanced the immunogenicity of a tumor-specific Ag and elicited essentially similar CTL compared with CTL elicited by the Ras_4–12(V12) epitope, as shown by TCR analysis (Table II) and the identification of TCR contact residues in the Ras_4–12(V12) peptide (Fig. 7) without cross-reaction against self (or proto-ras forms).

The novelty of the study reported here importantly illustrates the conceptual design of an altered peptide epitope (denoted X) that can induce more potent immune responses compared with the relevant peptide epitope (denoted Y) found in tumors, yet this immune response does not react with peptide epitopes found in normal tissues (denoted Z). Previous studies on the design of altered peptides have shown only that altered peptide X can elicit a more potent response than the relevant peptide Y. Consistent with this idea, our study demonstrated the creation of a novel potent CTL epitope derived from a weak CTL epitope contained in point-mutated ras oncogenes. To our knowledge, this is the first report demonstrating the rational design of a tumor-specific CTL epitope peptide with the ability to recognize not only the altered peptide ligand used as the immunogen, but also the naturally mutated Ras Val¹² epitope. Moreover, the CTL generated does not recognize the wild-type Ras Gly¹² peptide, since the Gly¹² peptide does not bind to H-2K^d, suggesting that any immune response directed toward mutations at ras codon 12 would be exquisitely specific for the mutation and would be unlikely to cross-react with cells expressing normal ras in an autoimmune fashion. It is noteworthy that similar in vitro results were observed when the Ras_4–12(I12) peptide variant was compared with the Ras_4–12(L12) peptide, suggesting that either of the codominant residues, Leu or Ile, at the C-terminal anchor position could enhance H-2K^d binding similarly.

The engineering of altered peptide ligands has been shown by others to increase immune responses in several models unrelated to tumor-specific Ag. For example, Lipford et al. showed a correlation between the MHC binding potential and immunogenicity with peptides derived from the model Ag OVA (15) or from human papilloma virus protein, E6 (16). In both reports, specific amino acid substitutions of CTL epitope peptides that enhanced binding to class I molecules also enhanced the ability of such peptide variants to induce an in vivo CTL response that recognized the native peptide Ag. Similarly, in a human system describing the optimization of HLA-A2 binding peptides, Parkhurst et al. (17) designed peptides derived from the melanoma-associated Ag, gp100. Anchor residue-substituted peptides that enhanced class I binding correlated with an increased ability of these peptide variants to induce CTL from PBL of patients with melanoma; the CTL generated recognized target cells presenting the wild-type peptide from the gp100 Ag. Others increased the immunogenicity of peptides that bind moderately to MHC class II by designing hybrid peptides containing a Th cell epitope of hemagglutinin fused between the aggretopes of an influenza A helper epitope peptide (18). The resulting peptide greatly enhanced the cellular and humoral responses in mice immunized with the fusion peptide. Another approach by Tourdot et al. (19) designed chimeric peptides replacing low affinity MHC contact residues from subdominant CTL epitopes with high affinity MHC (D^b) binding motifs from an immunodominant CTL epitope (NP₃₆₆). The resulting chimeras not only efficiently induced Ag-specific CTL populations in immunized B6 mice that reacted with the corresponding low D^b affinity nonimmunogenic peptide on pulsed targets, but also cleared viral infection in vivo. Our data are in agreement with these reports and extend them by showing that this approach may prove useful toward increasing the immunogenicity of weak MHC binding epitopes from point-mutated tumor Ag, such as Ras or p53.

Since the point mutation at codon 12 generates a neo-epitope that can bind to MHC class I molecules (H-2K^d shown here and in 8 , it is worth considering what the TCR recognizes as foreign. Data presented here show that residues 6 to 10 (corresponding to positions 3–7 of the nonamer peptide) are recognized by the TCR expressed by anti-Ras_4–12 CTL lines (Fig. 7). Thus, a self sequence can become a nonself sequence after the introduction of an appropriate class I binding anchor residue. It is notable that an artificial or surrogate CTL peptide epitope [Ras_4–12(L12)] used to induce a similar immune response against the biologically relevant CTL epitope [Ras_4–12(V12)] expressed a similar TCR heterodimer and not an aberrant TCR. Interestingly, we recorded a similar Vαβ chain usage from a third anti-Ras_4–12(V12) CTL line (our unpublished observations), suggesting that this particular TCR heterodimer may be genetically restricted to the Vα1β9 heterodimer in the BALB/c strain (Table II). However, our data do not exclude the possibility that other Vαβ combinations of TCR in the BALB/c repertoire exist that are reactive toward the Ras_4–12 sequence. In fact, we observed some minor Vαβ chains by reverse transcriptase-PCR analysis (Table II; Vα4, Vα8, Vβ2, and Vβ6) that may represent TCR reactive toward Ras_4–12(V12), but the relevance of these individual TCR chains is unclear since mAb against the Vβ9 chain inhibited most of the CTL activity (Fig. 7) from both CTL lines. Nevertheless, there may be other CTL clones reactive toward Ras_4–12(V12) expressing different TCR that were lost due to particular culture conditions as discussed above. The genetic restriction of TCR Vα1β9 chain usage by anti-Ras_4–12 CTL is supported by similar reports of TCR gene expression in response to specific Ag (20, 21, 22). For example, Lehner et al. (20) demonstrated in 21 HLA-A2⁺ patients exposed to influenza A virus that the predominant TCR Vβ chain expressed, which was directed toward the immunodominant matrix peptide, M_58–66, was Vβ17. However, such genetic restriction of a specific TCR to a CTL epitope is not absolute. Cole et al. (23) described a widely diverse repertoire of TCR gene usage in response to the immunodominant NP_324–332 CTL epitope peptide from Sendai virus. It is presently not clear which characteristic of a particular CTL epitope results in the expression of either a predominant TCR or a diverse TCR that recognizes the epitope or whether the differences relate to the various culture conditions.

Together, our data represent the development of an altered peptide ligand of a previously identified point-mutated Ras CTL epitope that rendered a more immunogenic CTL epitope without compromising the specificity of the anti-mutant Ras immune response. The work extends altered peptide ligand design for the induction and expansion of CTL reactive toward tumor-specific target Ag. This approach may be clinically useful for active specific immunotherapy of tumors harboring point-mutated ras oncogenes as peptide vaccines or for passive immunotherapy by rapid ex vivo expansion of CTL from vaccinated patients for adoptive cellular immunotherapy. Mutated Ras peptide epitopes have been described that can bind to HLA class I molecules and induce cellular immune responses in humans either in vitro (24, 25) or in vivo, albeit weakly (26). Thus, the potential exists for the rational design of altered peptide ligands with enhanced MHC binding potential that could induce greater quantitative and/or qualitative immune responses in patients with tumors harboring point-mutated ras oncogenes without compromising the exquisite immune specificity against the naturally expressed target Ag.

We thank Drs. Sam Zaremba and Margaret Schott for synthesizing several peptides used in this study, Dr. Elena Barzaga for a critical review of the manuscript, and Mr. Elwood McDuffie for excellent technical support of this work. We also thank RIBI ImmunoChem Research for their generous supply of Detox-PC adjuvant.