Fine Specificity of High Molecular Weight-Melanoma-Associated Antigen-Specific Cytotoxic T Lymphocytes Elicited by Anti-Idiotypic Monoclonal Antibodies in Patients with Melanoma

The treatment of malignant melanoma with chemotherapy and radiotherapy has shown little progress in the past 20 years (1). As a result, alternative therapeutic strategies are being developed and applied. In recent years, active specific immunotherapy has been emphasized because immunological events appear to play a role in the pathogenesis and in the clinical course of melanoma. A broad spectrum of immunogens has been and is being used in clinical trials, which have been performed or are in progress. They range from cultured human melanoma cells or their lysates (2, 3) to well defined moieties. The latter include melanoma-associated antigen (MAA) -derived peptides with HLA class I antigen binding motifs, which are targets of CTLs (4, 5), gangliosides (GD2; 6), and anti-idiotypic (anti-id) monoclonal antibodies (mAb), which mimic antigenic determinants expressed on MAA (7, 8, 9, 10). Anti-id mAbs, which mimic human GD2 ganglioside, and high molecular weight (HMW)-melanoma-associated antigen have been used as immunogens in clinical trials (7, 8, 10). A humoral and/or a cellular immune response to the nominal antigen has been elicited in a variable percentage of patients with melanoma immunized with anti-id mAbs, which mimic the HMW-MAA (7, 8, 9, 11, 12, 13, 14). This immune response has been found to be associated with regression of metastases in a few patients (11) and with disease-free interval and/or survival prolongation (7, 8, 12, 13). The mechanism underlying the latter association in the case of HMW-MAA has been suggested to be an inhibition of its role in melanoma cell biology because this antigen has been shown to influence adhesion, migration, and proliferation of melanoma cells by acting as a receptor for components of the extracellular matrix and various circulating molecules. Furthermore, anti-HMW-MAA antibodies have been shown to inhibit melanoma cell spreading on basement membrane synthesized by endothelial cells (15, 16, 17, 18).

There is only scant knowledge about CTL responses elicited by anti-id mAbs that mimic tumor-associated antigen (TAA). Furthermore, whether anti-id mAbs contain epitopes recognized by T cells and whether these epitopes are homologous with the amino acid sequences in the nominal TAA remains to be determined.

The basis of structural mimicry between an antigen (Ag) and the corresponding anti-id antibody is unclear. One possibility is represented by the structural basis of the interactions of a TAA and corresponding anti-id antibody with the same area of the Ag combining site of the anti-TAA antibody. These interactions are likely to involve side chains of amino acids, which can form strong H-bonds (involving OH groups) and even stronger electrostatic interactions between positively/negatively charged terminal groups of the side chains and reversely charged groups in Ag-binding regions of antibodies and of T-cell receptors (TCR).

Peptides derived from TAA, including the HMW-MAA, may be presented by HLA class I molecules, but they are all weak inducers of CTLs (19). Peptides from xenogeneic proteins in the vaccine can enhance the existent CTL responses to tumor cells. Alternatively, if peptides presented by melanoma cells are tolerogenic (20), then anti-id mAb peptides in the vaccine can induce CTLs, which can cross-recognize melanoma cells. CTL precursors of unknown specificity are present during induction of the id/anti-id cascade (21). Little attention has been paid to such CTLs, because the T-helper (Th)2 environment suitable for induction of humoral anti-anti-id responses was not considered to be favorable to effector CTL development (22).

The question is whether this pool of potential effectors, when activated, can mediate an antitumor effector response. To address this question we used as a model ex vivo isolated lymphocytes from high-risk melanoma patients treated with MELIMMUNE (14). The latter is a combination of the anti-id mAb MEL-2 and MF11–30, which mimic the antigenic determinants of the HMW-MAA defined by mAbs MEM136 and 225.28, respectively (23, 24). MELIMMUNE was shown to induce HLA-A2-restricted CTLs, which lyse melanoma cells expressing both HLA-A2 Ag and HMW-MAA. However, the epitope(s) recognized by those CTLs was not characterized.

In the present study we found that the amino acid sequence 3–11 of the heavy chain variable region (VH) of the anti-id mAb MF 11–30 is partially homologous with that of HMW-MAA core protein (76–84). We designed a variant (v)HMW-MAA (76–84) peptide to align the shifted Tyr⁸¹Ser⁸² in HMW-MAA sequence with Tyr⁹ Ser¹⁰ in VH (3–11) sequence. In vitro stimulation with peptides VH (3–11) and vHMW-MAA (76–84) but not with wild-type (wt)HMW-MAA (76–84) peptide generated HMW-MAA-specific CTLs from peripheral blood mononuclear cells (PBMC) from HLA-A2⁺ patients who had been immunized with MELIMMUNE. Melanoma cell lysis by VH (3–11) and vHMW-MAA peptide-elicited CTLs was inhibited by VH (3–11) peptide indicating structural similarities between these epitopes. These results suggest an evolutionary mechanism of tumor escape from immunosurveillance, which can be counteracted by CTL antigenic diversity for epitopes generated from VH anti-id antibodies during induction of a humoral response.

Of the 10 studied patients, 5 are free of disease 5–7 years after vaccination with MELIMMUNE, and 5 are currently being followed in the melanoma outpatient clinic at M.D. Anderson Cancer Center. All of the patients had disease resected a minimum of 2 months before study and had an Eastern Cooperative Oncology Group performance status of 0. None were currently receiving and/or had received chemotherapy and/or biological therapy within 2 months from study.

The anti-HLA-A2.1 mAbs MA2.1 and BB7.2 (25, 26) were purified from the corresponding hybridoma supernatants by ammonium sulfate precipitation followed by column chromatography. Both anti-HLA-A2 mAb secreting hybridomas were obtained from American Type Culture Collection (Rockville, MD). FITC-conjugated isotype-specific antimouse IgG and all of the other antibodies used in the present study were purchased from BD PharMingen (La Jolla, CA.)

The hybridoma secreting the anti-id mAb MF11–30 was developed as described (23). It was cultured in medium RPMI 1640 supplemented with 10% heat-inactivated FCS (Sigma Chemical Co., St. Louis, MO). The melanoma cell line A375SM (HMW-MAA⁺/HLA-A2⁺; donated by Dr. Michael Rosenblum, M. D. Anderson Cancer Center, Houston, TX) and the lymphoblastoid T2 cells (donated by Dr. Peter Creswell, Yale University, New Haven, CT) were grown in DMEM containing 10% heat-inactivated fetal bovine serum (Sigma Chemical Co.), 5 mm l-glutamine, 5 mm nonessential amino acids, 5 mm sodium pyruvate, and 0.1% gentamicin (Life Technologies Inc., Grand Island, NY). PBMCs were obtained from patients with resected melanoma (stages I-IV American Joint Committee on Cancer) using Ficoll gradients (Sigma Chemical Co).

Peptide I-100 (LLVLLYSKL) is from the anti-id mAb MF11–30 V_H residues 3–11. Peptide I-101 (LLQLYSGRL) corresponds to the wild-type amino acid sequence of HMW-MAA residues 76–84. Peptide I-102 (LLQLGYSGRL) has the amino acid sequence of the HMW-MAA stretch 76–84 modified by inclusion of Gly at position 80. Peptide F119 (AAGIGILTV) corresponds to the CTL epitope from MAA MART-1 (amino acids 27–35). Peptide E71, HER-2 (828–836) corresponds to an inactive HLA-A2 epitope from HER-2/neu protein. All of the peptides were synthesized in the M.D. Anderson Cancer Center Peptide Core Laboratory and purified by high-performance liquid chromatography. Their identity was determined by amino acid analysis. The purity of these peptides was >95%.

mAb V_H and V_L sequences were determined and analyzed as described previously (27).

Patient PBMCs were sequentially incubated with mAb BB7.2 and with FITC-conjugated antimouse IgG xenoantibodies as described (24). The number of fluorescence-positive cells and the fluorescence intensity were determined using an EPICS-V Profile Analyzer (Coulter Corporation, Hialeah, FL; Ref. 28).

PBMCs were cultured in tetra-triplicate in 96-well flat-bottomed microtiter plates (Costar Corp., Cambridge, MA) at the concentration of 2 × 10⁵ cells/well in 100 μl of complete RPMI 1640 containing l-glutamine and 10% fetal bovine serum. Peptides I-100 and I-101 were added at 5, 25, and 50 μg/ml in serum-free RPMI 1640 up to a total volume of 200 μl. PBMCs were cultured for 6 days at 37°C in a 5% CO₂ atmosphere before addition of tritiated thymidine at a concentration of 1 μCi/well. Results are expressed as cpm of tritiated thymidine incorporated and as a stimulation index (26). The latter is the average cpm of lymphocytes cultured with peptide divided by the average cpm of lymphocytes cultured without peptide. A proliferative response was scored as positive when mean cpm ± SD values for cultures that were incubated with peptides differed significantly from those for cultures that were incubated without peptides (P < 0.05, Student’s t test). Peptide-specific T cells were propagated by adding recombinant interleukin (IL)2 (90 IU/ml; Biosource International, Camarillo, CA) to PBMCs that had been cultured for 96 h at 37°C with peptides. After 5 additional days of culture at 37°C cells were washed, replated, and restimulated at a 1:1 stimulator to responder ratio with irradiated (30 Gy) autologous PBMC stimulators and peptides at concentrations of 5, 25, and 50 μg/ml. IL-2 was added after 96 h, and cells were cultured for an additional 5 days at 37°C, after which they were restimulated for an additional 10 days at 37°C. Lymphocytes stimulated over three cycles were collected and used as effectors in a cytotoxicity (CTL) assay.

Lymphocytes collected after three rounds of peptide stimulation were tested for their ability to lyse ⁵¹Cr-labeled targets as described previously (29, 30). To prove HLA-A2 restriction in CTL recognition, targets were incubated with 50 μg/ml of anti-HLA-A2 mAb MA-2.1. Cold target inhibition of tumor cell lysis experiments was performed as described (30).

Supernatants were collected at 24, 48, and 72 h after primary and secondary stimulation with 5, 25, and 50 μg/ml of peptides and stored frozen at −20°C. IFN-γ and IL-10 levels were measured by a double sandwich ELISA using the corresponding kits supplied by Biosource International. Assays were calibrated with human recombinant IFN-γ and IL-10 to detect each cytokine in a range of 15–1000 pg/ml.

The V_H and V_L nucleotide and deduced amino acid sequences of anti-id mAb MF11–30 are shown in Table 1. The V_H gene segment of the mAb MF11–30 shows high homology with the Q52 V_H gene family. The D gene segment is part of the DSP2 mini-gene family. The J_H gene segment is derived from the J_H3 family. The V_L gene segment shows homology with subgroup III of the V_K gene family. The J_K gene segment is 100% homologous with the J_K2 mini-gene.

The amino acid sequences of the mAb MF11–30 V_H and V_L were compared with the published amino acid sequence of the HMW-MAA core protein (31) using the program BLOSUM62 from the ExPASy website.⁶ Partial (between 20% and 40%) homology between these two proteins was observed only on short stretches of 10–17 amino acids. However, these areas of homology are unlikely to contain HLA-A2 binding peptides, because HLA-A2 binding motifs were not identified in these areas of homology.

We then searched the amino acid sequence of mAb MF11–30 VH and VL for HLA-A2 binding motifs, using the program by Taylor et al. available online.⁷ The sequence VH (3–11) LLVLLYSKL in the framework region 1 met the criteria of an HLA-A2 binding peptide, i.e., hydrophobic aliphatic residues at positions 2 and 9, as well as hydrophobic aliphatic residue at position 1 and aromatic residue at position 6. This peptide had the highest binding affinity for HLA-A2 Ag of all VH peptides (Table 2). Analysis of the VH amino acid sequence with proteasome digestion programs (32, 33) indicated that this peptide is protected from proteasome digestion (data not shown). A search in the databases using the FASTA.3 program identified only few partially homologous sequences with the VH (3–11) in immunoglobulins and other proteins (data not shown) suggesting that this peptide sequence was “idiotypic” for mAb MF11–30.

The VH3 (3–11) sequence was used for homology search in HMW-MAA. We found a partially homologous peptide (five of nine residues) in the HMW-MAA sequence 76–84. The first four amino acids of 76–84, LLQL, were similar to the first four amino acids from the VH (3–11) sequence. The change V→Q is neither a significant structural change nor a change in charge but results in a change in polarity. The last five amino acids, YSGRL, are similar with the last four amino acids of VH (3–11) sequence, YSKL, with the differences that Tyr⁵ replaces Leu⁵, and Gly is intercalated between Ser⁷ and Arg⁸. The HMW-MAA (76–84) peptide has lower binding affinity to HLA-A2 Ag than VH (3–11) peptide (Table 1). It also appears to be less protected from proteasome destruction than VH (3–11) peptide. On the basis of the nucleotide sequence, the Leu⁵→Tyr⁵ change is not the result of a point mutation but rather of the deletion of the base triplet encoding Leu. Thus, the position of the group Tyr⁶ Ser⁷ in VH (3–10) was shifted to Tyr⁵Ser⁶ in HMW-MAA. Tyr and Ser should enhance Ag binding to TCR, because both contain free OH groups, which are most likely to form H bonds with the TCR. Gly⁷ is intercalated before the COOH-terminal groups Arg⁸Leu⁹ in HMW-MAA and Lys⁸ Leu⁹ in VH (3–11), respectively. On the basis of the nucleotide sequence, the change Lys→Arg is the result of a point mutation.

Because the shift of Tyr-Ser group may affect TCR recognition and Ag immunogenicity, Gly, which lacks side chain and allows flexible positioning, was introduced at position 5 of HMW-MAA (76–84) to align the Tyr and Ser in both VH (3–11) and HMW-MAA peptides. The resulting sequence was LLQLGYSGRL. In this decapeptide, the amino acid positions 1, 2, 4, 6, and 7, and the COOH-terminal amino acids 9 and 10 are aligned with VH (3–11) peptide. Because this decapeptide contains the same COOH-terminal anchors as the nonapeptide LLQLYSGRL, the same TCR is likely to interact with both peptides, although with different affinities. VH3–11 and variant HMW-MAA (vHMW) peptides have the identical amino acids at positions 6 and 7. Their homology is seven of nine residues (77%). The binding affinities to HLA-A2 Ag of both wild-type (wt) and variant (v) HMW-MAA peptides are similar. The corresponding peptides were designated as I-100, VH: LLVLLYSKL; as I-101, wtHMW-MAA: LLQLYSGRL; and as I-102, vHMW-MAA: LLQLGYSGRL (Table 2).

We tested first whether patients immunized with MELIMMUNE had T cells that recognized peptides from the anti-id mAb used as an immunogen as well as from the HMW-MAA. Thus, we determined whether PBMCs isolated from immunized patients responded to peptides I-100 and I-101 better than PBMCs from nonimmunized patients. In addition, we tested whether PBMCs contained T cells that recognized wtHMW-MAA and vHMW-MAA peptides. In preliminary experiments we observed that peptide I-101 did not stimulate the proliferation as well as IFN-γ and IL-10 secretion by PBMCs from HLA-A2⁺ patient 6 and from four HLA-A2⁻ patients, because these parameters were similar to those found in cultures of autologous PBMCs incubated without peptide. At restimulation with peptide I-101, the PBMCs of patient 6 became apoptotic, suggesting that the wtHMW-MAA was tolerogenic by deletional mechanisms (data not shown). Therefore, we focused our subsequent investigations on the analysis of VH3–11 and vHMW-MAA peptides, which appeared to be stimulatory.

As shown in Table 3 when incubated with I-100 and I-102 peptides for up to 6 days (1° stimulation; STIM), PBMCS from patients 1, 2, 4, and 6 proliferated to a markedly higher extent than autologous PBMCS incubated with no peptide (NP). The extent of proliferation induced by I-100 and I-102 peptides was similar.

Lymphocytes from the HLA-A2⁺ patients 1, 2, and 6 who had been immunized with MELIMMUNE (Table 3) responded with significantly stronger proliferation to both I-100 and I-102 peptides and had significantly higher stimulation index (2.7 ± 0.7) than HLA-A2⁻ patients 3, 5, 9, and 10 who had not been immunized (1.2 ± 0.4; P < 0.001). Thus, induction of proliferation by I-100 and I-102 peptides was associated with HLA-A2 Ag expression. These results also indicate that immunization activated a large population of T cells capable of responding to both VH3–11 and vHMW-MAA peptides. The mitogenic effect of VH3–11 and vHMW-MAA peptides was specific, because irrelevant HER-2/neu-specific peptide, E71, was not mitogenic for lymphocytes from HLA-A2⁺ immunized patients 1, 2, and 6 (NP = 353 ± 84 cpm versus E71 = 411 ± 190 cpm; P = 0.65) and HLA-A 2⁻ patient 8 (NP = 560 ± 22 versus E71 = 685 ± 103; P = 0.06). Like the E71 peptide, the MART-1 peptide F119 was not mitogenic for PBMCs from patients 1 (330 ± 36 cpm; stimulation index = 0.7) and 10 (215 ± 26 cpm; stimulation index = 1.1).

Interaction of T cells with Ag can induce a few cycles of division resulting in Ag-primed uncommitted T cells or can induce T cells to differentiate and gain effector function. The effector function is best characterized by the ability of Ag-stimulated T cells to secrete Th1 cytokines and to mediate lysis of specific targets. To determine whether T cells from immunized patients acquired cytokine effector function, IFN-γ and IL-10 were measured in supernatants harvested from PBMC cultures stimulated with I-100 and I-102 peptides. IL-10 was assayed instead of IL-4 as a representative Th2 cytokine, because the PBMCs used as responders were from patients with melanoma. Therefore, they might contain regulatory cells. Patient 6 had supernatants saved from two separate experiments. This afforded validation of conclusions in replicated analyses performed at separated time points. Because after priming, all of the stimulated PBMCs were expanded in IL-2, we determined the effect of culturing PBMCs in IL-2 on IFN-γ induction. Representative individual patient examples of IFN-γ and IL-10 induction are shown in Fig. 1. Both IFN-γ and IL-10 levels from peptide-stimulated PBMCs from HLA-A2⁺ patient 6 and HLA-A2⁻ patient 7 were low after 1° STIM with peptide I-100 (Fig. 1, A and B).

To verify the significance of cytokine levels induced by I-100 and I-102 peptides, the cytokine responses induced by the dominant MART-1 epitope, represented by peptide F119, were determined in parallel. IFN-γ levels produced by PBMCs from HLA-A2⁺ patient 6 stimulated with each of the peptides I-100, I-102, and F119 were similar (P > 0.05; Fig. 1,C). IL-10 levels after restimulation with control F119 peptide were significantly lower than those obtained with peptide I-102 (F119 = 58 ± 11 pg/ml versus I-102 = 212 ± 24 pg/ml; P < 0.025) but were not significantly different from those obtained with peptide I-100 (86 ± 11 pg/ml; P > 0.05; Fig. 1 D).

IFN-γ and IL-10 responses from all of the patient PBMCs are shown in Fig. 2. Peak responses varied from patient to patient but, as a rule, responses were highest when measured after a 48- or a 72-h incubation with peptide (25 μg/ml). Cytokine secretion by PBMC from patients 8 and 9 could not be tested because additional lymphocytes could not be obtained from these two patients. After in vitro stimulation, there were no significant differences in mean pg/ml of IFN-γ ± SD secreted after stimulation with I-100 peptide versus stimulation with I-102 peptide (Fig. 2,A). The IFN-γ:IL-10 ratios at priming with both peptides were <1.0, although they were lower with I-100 peptide than with I-102 peptide. The IFN-γ:IL-10 ratios changed at restimulation to >3.0. IFN-γ levels were significantly (P < 0.05) higher after 2° STIM than after 1° STIM, whereas IL-10 levels did not differ (Fig. 2,B). After restimulation, IFN-γ levels in patients 2 and 6 increased by >1 log. The average IFN-γ levels after 2° STIM were significantly higher (616.1 + 86.3) for PBMCs from HLA-A2⁺ patients than those for PBMCs from HLA-A2− patients (59.5 ± 24.4; P < 0.007; Fig. 2 C). A similar analysis for IL-10 did not detect significant differences (pg/ml IL-10: HLA-A2⁺ = 155.7 ± 36.3 versus HLA-A2⁻ = 61.1 ± 25; P = 0.06). The IFN-γ:IL-10 ratios for HLA-A2+ patients were 4-fold higher (3.96) than those for HLA-A2⁻ patients (0.97) indicating a trend toward Th1 responses in HLA-A2⁺ patients. Therefore, immunization with anti-id mAb MF11–30 recruited and activated T cells to acquire the potential of IFN-γ production. Demonstration of activation of effector cytokines required either priming and restimulation in vitro with peptide or culture of PBMCs in IL-2 before stimulation with peptide.

To determine the effects of the immunization with anti-id mAb on activation of melanoma cell-lytic CTLs, we compared the responses of immunized HLA-A2⁺ patients 1 and 6 with that of nonimmunized HLA-A2⁺ patient 4. As specificity and restriction control we analyzed the responses of nonimmunized HLA-A2⁻ patients 3 and 5. All of the CTLs were tested for their ability to lyse T2 cells pulsed with the stimulating peptide and A375sm melanoma cells. I-100-CTL and I-102-CTL from the HLA-A2⁺ patients 1 and 6 lysed I-100 pulsed T2 cells (T2-I-100) to a significantly greater extent than control targets T2-NP, i.e., T2 cells that had not been pulsed with peptide (Fig. 3, A and B). CTLs induced by the HMW-MAA peptide I-102 also recognized the VH (3–11) peptide I-100 (data not shown). The HLA-A2⁺ patient 4 I-100-CTL did not lyse T2-I-100, yet autologous I-102-CTL lysed T2-I-102 (Fig. 3 C). These results indicate that immunization with MELIMMUNE activated CTL precursors to both the VH3–11 peptide and to the HMW-MAA peptide, whereas in vitro stimulation with the same peptides of HLA-A2⁺ PBMCs from the nonimmunized patient activated only the existent CTL precursors for HMW-MAA.

PBMCs from the HLA-A2⁻ patients 3 and 5 that had not been immunized with MELIMMUNE but were only stimulated in vitro with VH3–11 and HMW-MAA peptides did not develop CTLs recognizing the stimulating peptide. As shown in Fig. 3, D and E, there was no specific lysis of T2-I-102 targets by CTL from patient 3 stimulated repetitively with I-102 peptide compared with targets incubated without peptide (T2-NP). Lymphocytes from patient 5 were also unable to specifically lyse T2-I-100 and T2-I-102 targets (Fig. 3 E). This indicates that CTL activation by I-100 and I-102 peptides was restricted by HLA-2 Ag.

To determine whether in vitro stimulation with peptides generated CTLs that recognized epitopes on HMW-MAA^+/HLA-A2⁺ melanoma cells, PBMCs from immunized HLA-A2⁺ patients 1 and 6 were stimulated with I-100 and I-102 peptides and then tested for their ability to lyse A375SM melanoma cells. As shown in Fig. 4,A, the I-100-stimulated PBMCs and I-102-stimulated PBMCs of patient 1 effectively lysed A375SM cells, demonstrating that CTLs were present in peptide-stimulated PBMCs. Therefore, these effectors were designated as I-100 CTL and I-102 CTL, respectively. Lysis was effectively blocked by “cold” T2-I-100. This result demonstrated both the specificity of CTLs for the VH3–11 peptide and expression of a structurally similar peptide-HLA-A2 epitope on melanoma cells. The results confirmed that the significant degree of homology in the sequence of epitopes formed by I-100 and I-102 peptides (>67%) resulted in CTLs cross-recognizing the HMW-MAA epitope presented by tumor cells. In addition, lysis was inhibited by HLA-A2-specific mAb MA2.1 by >50%, confirming that this molecule is the restricting element for CTL (Fig. 4 B).

Similar results were obtained with CTLs from patient 6. Fig. 4 C shows the kinetics of lysis of A375SM cells by I-100-CTL and I-102-CTL. Effectors used as controls were generated from autologous PBMCs by stimulation in the same experiment and under the same experimental conditions with T2 cells, which had been pulsed with HMW-MAA peptide or incubated with IL-2 as a growth factor. The latter were designated as NP-CTL. NP-CTLs were used as effectors to determine whether memory CTLs induced by the vaccine were present in this patient and up-regulated their cytolytic function after stimulation with cytokines and HLA-A2 Ag, which was not presenting cognate Ag. The results show that NP-CTLs (“memory-like”) specifically recognized an epitope on tumor cells, which is structurally similar to the one formed by T2-I-100. However, their lytic activity was weak and did not increase over time. In contrast, I-100-CTL recognized the same epitope with faster kinetics in the first 5 h than I-102-CTL. Their kinetics of lysis slowed over time. I-102-CTL recognized the same epitope on tumor cells with apparently slower kinetics than I-100-CTL, during the first 5 h, but the kinetics of lysis became somewhat faster over time compared with that of I-100-CTL (59% increase from 5 h to 20 h). We could not determine the exact number of effectors in each population bearing specific TCRs for each peptide, because the necessary reagents were not available. Nevertheless, the differences in kinetics of lysis between I-100-CTL and I-102-CTL suggest small quantitative differences in their functional avidity for tumor cells.

To the best of our knowledge, this is the first study to identify a peptide from an anti-id mAb that induces CTLs recognizing HMW-MAA⁺ melanoma cells in a HLA-A2-restricted fashion. PBMCs used to generate CTLs were obtained from patients with melanoma who had been immunized with anti-id mAb MEL-2 and MF11–30. The two anti-id mAb mimic two distinct determinants of HMW-MAA. The peptide VH3–11 derived from anti-id mAb MF11–30 induced proliferation and cytokine production by T cells and induced HMW-MAA-specific CTLs. The CTLs elicited by VH3–11 peptide were specific for a peptide structurally similar to that expressed on tumor cells because tumor cell lysis was inhibited by T2 cells pulsed with I-100 peptide. Lysis was restricted by HLA-A2 Ag because it required HLA-A2 Ag matching of effector and target cells and was markedly inhibited by anti-HLA-A2 mAb MA2.1. Induction of CTL by anti-id mAb MF11–30 derived peptide and by the variant HMW-MAA peptide demonstrated the existence in the immunized patients of populations of T cells bearing TCRs, which can be activated by the vaccine and which cross-recognize wtHMW-MAA epitopes. wtHMW-MAA (76–84) peptide, I-101, did not stimulate proliferation or IFN-γ secretion in PBMCs. Peptide I-101 might delete or inactivate CTLs induced by peptide I-100.

A significant increase in IFN-γ secretion compared with IL-10 was observed at restimulation with peptides I-100 and I-102 suggesting that both peptides have the potential to induce a Th1 polarizing environment. Their Th1 potential may be even stronger, because culture of patient lymphocytes in IL-2 before peptide stimulation resulted in >10-fold higher levels of IFN-γ than of IL-10. This finding raises the possibility that γ-chain cytokines such as IL-2, if administered after vaccination, may synergize with immunogens such as anti-id mAb in inducing inflammatory cytokines.

Our results are novel with regard to induction of adoptive CTL immunity to tumor cells by immunization with anti-id mAb. A target CTL epitope from HMW-MAA on tumor cells and the homologous epitope on the anti-id mAb used as an immunogen have been identified for the first time. Cellular responses elicited by immunization with anti-id antibodies, which mimic TAA, have been reported in preclinical models (34, 35, 36). Anti-id-specific T cells have been proposed to be involved in tumor regression, but their specificity and effector mechanisms have not been identified (37). In patients with malignant diseases, Durrant et al. (38) described tumor cell-specific killing by lymphocytes derived from lymph nodes adjacent to tumors and from PBMCs from three patients with rectal cancer after immunization with the human anti-id mAb 105AD7. The target epitopes on the Ag and the anti-id mAb have not been described yet. Furthermore, Saha et al. (36) reported that murine dendritic cells pulsed with an anti-id mAb induce Ag-specific protective antitumor immunity. In agreement with our results, the same authors (39) found that peptides from the VH and VL of the anti-id mAb 3H1, which mimics carcinoembryonic antigen, induced in vitro proliferation of PBMCs from patients with colon cancer who had been immunized with this anti-id mAb. Similarly, anti-id mAb 11D10 induced T-cell proliferation in PBMCs from breast cancer patients immunized with this antibody (40). Analysis of the subsets of T cells stimulated revealed that CD4⁺ cells that secrete IFN-γ were predominant. The carcinoembryonic antigen peptides used by Chatterjee et al. (39) were mostly 12–14 amino acids long and induced strong HLA class II Ag-restricted responses involving CD4⁺ as well as CD8⁺ T cells. Proliferative T-cell responses to anti-id mAb GA733 were also observed in colon cancer patients (41) indicating the existence of a large T-cell repertoire capable of response to various anti-id mAbs.

Several mechanisms may be envisioned for the generation of HMW-MAA-specific CTLs after immunizations with anti-id mAb. First, peptides with HLA class I Ag binding motifs from anti-id mAb VH or VL when bound to particular HLA-A, B, or C molecules may form complexes that structurally resemble HMW-MAA epitopes expressed by the same HLA class I specificities. The VH peptides are more immunogenic than the HMW-MAA peptides because they can induce CTLs, which recognize the wt epitope on tumor cells. This possibility is supported by the recently reported ability of myeloma immunoglobulins to elicit in an autologous system CTL effectors capable of lysing the corresponding myeloma secreting tumor cells (42). An alternative, although not exclusive, possibility is that anti-anti-id antibodies generated by immunization with anti-id antibodies, such as the one used in this study, bind HMW-MAA present in plasma. The complexes taken up by dendritic cells induce HMW-MAA-specific CTLs. This possibility is supported by the high immunogenicity of Ag-antibody complexes, which has been shown in a number of antigenic systems (43). An additional possibility that needs future investigations is the genetic association between expression of HLA-A2 Ag and induction of CTLs by HLA-A2- presented peptides derived from anti-id mAb and HMW-MAA in patients with melanoma (44).

The significance of the sequence shift of Tyr⁶ Ser⁷ in the HMW-MAA sequence compared with the VH3–11 sequence deserves additional investigation. The shift is induced by deletion of Leu⁵ in HMW-MAA and is then “repaired” by introduction of Gly⁷. Thus, the last two amino acids in both VH3–11 and HMW-MAA 76–84 sequences are similar. In CTL epitopes, amino acids in the central area, position 4–7 lift from the HLA-A2 plane, are more likely to contact the TCR CDR3 and activate its signaling (45). Is the Leu⁵ codon deletion an evolutionary response to CTL pressure to protect HMW-MAA-expressing cells? Additional studies are required to address this question.

In summary, the peptides we have identified may be useful alone or in combination with other MAA-derived peptides in the development of immunogens for immunotherapy of melanoma. The HMW-MAA is homogeneously and stably expressed in >80% of primary and metastatic melanoma lesions (46). Induction of anti-HMW-MAA antibodies by immunization with anti-id antibodies, with consequent expression of CTL-activating peptides, followed by proinflammatory cytokines such as IL-12 and T-cell growth promoting γ-chain cytokines, may provide a novel approach to cancer therapy.

We thank Samantha J. Caldwell, Charlene DeMont, and Thomas D. Spence for manuscript preparation.