Peptide vaccination against tumor Ags can induce powerful systemic CTL responses. However, in the majority of patients, no tumor regression is noted. To study this discrepancy, we analyzed CTL reactivity in a melanoma patient (F001) vaccinated with g209-2M peptide, a single residue variant of gp100209–217. G209/g209-2M-reactive CTL were identified in post- but not pre-vaccination PBL. Limiting dilution analysis identified one predominant CTL clone (C1-35), with TCR Vβ6s2, recognizing g209/HLA-A*0201-expressing targets. Additionally, two autologous melanoma lines (F001TU-3 and -4) and 20 separate tumor-infiltrating lymphocyte cultures were generated from a fine needle aspirate of a metastatic lesion progressing after initial response to vaccination. Both F001TU did not express gp100 and were not recognized by C1-35. Loss of gp100 by F001TU correlated with a marked reduction of gp100 expression in the same metastatic lesion compared with prevaccination. Thus, ineffectiveness of C1-35 and tumor progression could be best explained by loss of target Ag expression. Interestingly, 12 of 20 tumor-infiltrating lymphocyte cultures recognized F001TU, but none demonstrated g209/g209-2M reactivity, suggesting a functional dissociation between systemic and local immune response. This study suggests that vaccination effects must be analyzed in the target tissue, rather than in the systemic circulation alone.

The identification of melanoma-associated Ags (MAA)2 and their respective CTL epitopes has raised interest in peptide-based vaccination approaches (1). Among MAA, MART-1/Melan-A and gp100/Pmel 17 (for which the respective CTL epitopes, MART-127–35 (2) and gp100209–217 (g209) (3), have been identified) have been noted to effectively induce CTL reactivity in vitro (4, 5, 6). In addition, clinical studies have shown that vaccination with MART-127–35 and g209 can powerfully enhance specific CTL reactivity in PBMC (7, 8, 9). However, the systemic CTL response to the vaccine most of the time does not correspond to clinical regression.

Although the clinical response remains the ultimate therapeutic goal, it is a parameter of little value for the identification of the reason for the most common therapeutic failures. Evaluation of systemic CTL reactivity is generally equated to the level of immune competence toward a certain epitope, and as a consequence is used for the assessment of the effects of a vaccination protocol (10). Assessment of competence toward an immunogen, while yielding an accurate view of the systemic immune response to a vaccine, may not provide sufficient information regarding target/host interactions at the site in which they are likely to occur. In fact, clinical response, including complete responses, has been reported in the context of MAGE-3 peptide vaccination without stimulation of detectable CTL activity at the systemic level (Thierry Boon, personal communication).

The development of peptide-based vaccination protocols for the immunotherapy of melanoma has given us the unique opportunity of comparing systemic T cell responses to a vaccine with localization and status of activation of the same T cells in the target organ.

We therefore wanted to establish a strategy suited for the analysis of CTL response to vaccination at the tumor site. Utilizing functional assays and TCR β-chain analysis, we studied the immune response of a melanoma patient after four cycles of vaccination with g209-2M peptide, a single residue variant of gp100209–217 identified as one of the immunodominant HLA-A*0201-restricted CTL epitopes of gp100 (3, 8, 11). An in-depth analysis of T cell reactivity was undertaken in the peripheral circulation and at the tumor site, which revealed a functional dissociation between local and systemic immune response during anti-melanoma vaccination.

The melanoma cell lines 624.38 (HLA-A*0201/0301, B*1402/−, Cw*0702/0802) and 624.28 (HLA-A*0301/−, B*1402/−, Cw*0702/0802) were generated by limiting dilution from a metastatic lesion (12). The cell lines 888-MEL (HLA-A*01/2402, B*52/55, Cw*0102/1201) and 1102-MEL (HLA-A*0201/24, B*55/62, Cw*03/−) were derived from other metastatic melanoma lesions. SK23 MEL (HLA-A*0101/0201, B*0702/0801, Cw*0702/0702) and A375 MEL (HLA-A*01/0201, B*17/−, Cw*06/−) melanoma cell lines were purchased from American Type Culture Collection (ATCC, Rockville, MD). All cell lines were maintained in complete medium (CM) consisting of RPMI 1640 (Biofluids, Rockville, MD) supplemented with 10 mM HEPES buffer, 100 U/ml penicillin-streptomycin (Biofluids), 10 μg/ml Ciprofloxacin (Bayer, West Haven, CT), 0.03% l-glutamine (Biofluids), 0.5 mg/ml amphotericin B (Biofluids), and 10% heat-inactivated human AB serum (Gemini Bioproducts, Calabasas, CA). T2 (ATCC), a cell line defective of endogenous processing and expressing HLA-A*0201 (13), was used to test CTL specificity toward HLA-A*0201-restricted epitopes. 1520 TIL was expanded with IL-2 from a metastatic lesion of an HLA-A*0201 melanoma patient not previously exposed to vaccination. 1520 TIL naturally recognizes g209.

All peptides were produced by solid-phase synthesis technique. The same gp100209–217 (g209) (ITDQVPFSV; Chiron Mimotopes Peptide Systems, San Diego, CA), gp100209–217(2M) (g209-2M) (IMDQVPFSV; Chiron Mimotopes), and MART-127–35 (AAGIGILTV; Peptide Technologies, Gaithersburg, MD) produced for clinical use were used for the in vitro sensitization assays. The residue 2 (T to M) substitution in g209-2M enhances binding to HLA-A*0201, efficiency of T cell induction in vitro (11), and was, for these reasons, preferred to the wild type for vaccination.

HLA class I and II were determined on PBL or tumor cell lines using sequence-specific primer PCR (14). PCR was also used for molecular subtyping of HLA-A2 (15). When necessary, the identity of some HLA alleles was determined conclusively by sequencing of cDNA.

Cell surface expression of HLA and other surface Ags (CD8, CD4) was determined by flow cytometry. Intracellular staining for the detection of MAA was performed by fixing cells in 200 μl of acetone for 10 min at room temperature before staining with the primary Ab (16). The following mAbs were used: W6/32 (Sera Labs, Westbury, NY) specific for a monomorphic determinant of the HLA class I heavy chain (17); IVA-12 (ATCC) for HLA class II; KS-I (18) for HLA-A2, (FITC) anti-human CD8, and (FITC) anti-human CD4 (PharMingen, San Diego, CA); and anti-MART-1/Melan-A murine IgG2b (M2-7C10) (16, 19) and anti-Pmel17/gp100 mAbHMB45 (Enzo Diagnostics, Farmingdale, NY). Cytospin preparations of sequentially obtained FNA material were fixed in acetone and stained with the same mAbs used for the FACS analysis, with the exception of HMB45 (Biogenex, San Ramon, CA). For secondary staining, biotinylated goat anti-mouse IgG (Kirkegaard & Perry Laboratories, Gaithersburg, MD) was used, followed by avidin-biotin-peroxidase (Vectasin Elite Kit; Vector Laboratories, Burlingame, CA) (16).

Autologous DC utilized for in vitro sensitization of CD8+ T cells were prepared as previously described (20). PBMC were separated from blood by centrifugation on a Ficoll-Hypaque gradient, incubated for 3 h at 37°C, and, after removal of nonadherent cells, cultured for 5 days in CM with 1000 IU/ml IL-4 (PeproTech, Rocky Hill, NJ) and 1000 IU/ml GM-CSF (PeproTech). On day 5, detached DC were harvested and used for stimulation either by peptide pulsing (1 μg/ml g209/DC or g209-2M/DC) or by infection with recombinant vaccinia virus (rVV-MART/DC) (Therion Biologics, Cambridge, MA). CD8+ cells were isolated from PBMC by positive separation (Dynabeads; Dynal, Lake Success, NY). A total of 4 × 106 CD8+ cells/well was cultured in a 24-well plate with 1 × 106 DC. After 24 h and every 2 days thereafter, 300 IU/ml IL-2 was added to the cultures. After 7 days, the cultures were restimulated with rVV-MART-1/DC, g209/DC, or g209-2M/DC and maintained in IL-2 for another week. On days 7 and 14, the cultures were tested for Ag recognition.

After the second testing for MAA recognition (day 14 of culture), CTL cultures were plated at 100, 10, and 1 cell/well ratio in 96-well round-bottom plates with 5 × 104 irradiated (50 Gy) donor PBMC and 1 × 104 irradiated (100 Gy) EBV-B cells in 200 μl CM supplemented with 30 ng/ml OKT3. After 1 day and every 2 to 3 days thereafter, 300 IU/ml IL-2 was added for 14 to 21 days. Clones were then tested for MAA reactivity. MAA-specific clones were restimulated and expanded in T25 flasks (Costar, Cambridge, MA) with 2.5 × 107 donor PBMC and 5 × 106 1088-EBV-B cells in 25 ml CM with 30 ng/ml OKT3 and IL-2. After in vitro expansion, the cultures were retested for specificity and analyzed for clonality by TCR β PCR, directed heteroduplex analysis (DHDA), and sequencing.

Using a 23-gauge needle, cells were aspirated from a metastatic lesion of patient F001 and plated immediately in CM. For TIL expansion, total cells were counted and plated in 24-well plates (of 4 × 106 cells/well) in the presence of 6000 IU/ml IL-2. After 2 wk, the cultures were further expanded in T25 flasks. For expansion of tumor cells, culture conditions were identical to TIL cultures, with the exception of IL-2. The F001TU autologous cell lines were totally HLA class I and II matched to the phenotype of patient F001 and had electron microscopy and karyotyping characteristics consistent with malignant melanoma.

IFN-γ release assay.

A total of 1 × 105 effector cells was plated with 5 × 104 stimulator cells in 96-well round-bottom plates in 200 μl CM. After 24-h incubation at 37°C, the plates were centrifuged and the supernatant was harvested for analysis by ELISA (Endogen, Cambridge, MA). IFN-γ is reported as pg/ml IFN-γ secreted by 5 × 104 effector cells in 24 h.

Calcein-AM fluorescent cytotoxicity assay.

A total of 106 target cells/well was incubated with 15 μl Calcein-AM (Molecular Probes, Eugene, OR) for fluorescent labeling. After 30 min, all targets were washed three times in CM and plated in triplicate in 96-well flat-bottom plates at 3000 target cells/100 μl. Effector cells were harvested and added to the target cells at E:T ratios of 10:1, 2.5:1, and 0.625:1 in 100 μl CM. The plates were centrifuged at 500 rpm for 3 min. After 3 h at 37°C, 5 μl of FluoroQuench (One Lambda, Canoga Park, CA) was added to each well to extinguish background fluorescence. The plates were centrifuged, incubated for an additional 60 min, and then scanned on a FluorImager 595 (Molecular Dynamics, Sunnyvale, CA). Fluorescence was quantified using ImageQuant software (Molecular Dynamics). Lysis was calculated using the following formula: (1 − [experimental fluorescence − background fluorescence]/[target only fluorescence − background fluorescence]) × 100.

For RNA isolation, cells were either taken directly from culture or, if frozen, after overnight culture to allow recovering of physiologic cell metabolism. RNeasy mini or midi kit (Qiagen, Santa Clarita, CA) was used for all RNA isolations. The RNA was eluted with water and stored at −70°C. For cDNA synthesis, about 1 μg of total RNA was transcribed with the SuperScript preamplification system (Life Technologies, Gaithersburg, MD) using the oligo(dT) primer. cDNA was eluted and stored at −25°C.

A set of 35 primers was selected to amplify 45 functional Vβ. Each primer mix was composed of 10 × PCR buffer, 1.5 mM MgCl2, 200 μM dNTP, 1.25 U AmpliTaq Gold, 0.5 μl cDNA, 0.5 μM Vβ primer, 0.5 μM TC-1 constant region primer (AYACCAGTGTGGCCTTTT), and water up to 20 μl final reaction volume. A total of 10 μl of light mineral oil covered the reaction mixture, and PCR was run using the following protocol: initial activation of the enzyme at 94°C for 9 min; 10 high-stringency cycles of 94°C for 30-s denaturation, 65°C for 1-min annealing, and 72°C for 1-min elongation; 20 low-stringency cycles of 94°C for 30 s, 60°C for 1 min, and 72°C for 1 min; and final extension at 72°C for 10 min. The following 35 primers were used for the Vβ region: TV2 CGAGTTTCTGGTTTCCTTTT for Vβ22s1; TV3-1 ATTTCTGAAGATAATGTTTAGC for Vβ9s1; TV4-1 GAAAGCTAAGAAGCCACCG for Vβ7s1; TV4-2/3 TACAAGCAAAGTGCTAAGAAGC for Vβ7s2, 7s3; TV5-1 GCCTTCAGTTCCTCTTTGA for Vβ5s1; TV5-4/5/6 GGCCCCAGTTTATCTTTC for Vβ5s2, 5s3, 5s6; TV5-8 TCCAGTTCCTCCTTTGGTATG for Vβ5s4; TV6-1 TGGGACTGAGGCTGATTT for Vβ13s3; TV6-2/3/5 GGCTGAGGCTGATTCATTAC for Vβ13s1, 13s2; TV6-4 GGCTAAGGCTCATCCATTAT for Vβ13s5; TV6-6 GGCTGAAGCTGATTTATTAT for Vβ13s6; TV7-2/3 GAGTTTTTAATTTACTTCCAAGGCA for Vβ6s1, 6s5; TV7-6/7/9 CCCAGAGTTTCTGACTTACTTC for Vβ6s3, 6s4, 6s6; TV7-8 GGCCAGAGTTTCTGACTTATT for Vβ6s2; TV9 CCTCCAGTTCCTCATTCAG for Vβ1s1; TV10-1/3 GGCTGAGGCTGATCCATTAC for Vβ12s1, 12s2; TV10-2 CATGGGCTGAGGCTGATCTA for Vβ12s3; TV11-1 GAGCTTCTGGTTCAATTTCA for Vβ21s1; TV11-2 CCAAAGCTTCTGATTCAGTT for Vβ21s3; TV11-3 GAGCTTCTGATTCGATATGAGA for Vβ21s2; TV12-3/4 GGACTGGAGTTGCTCATTT for Vβ8s1, 8s2; TV12-5 CAGACAGACCATGATGCAA for Vβ8s3; TV13 CCCAGTTCCTCATTTCGTT for Vβ23s1; TV14 TCGACGTGTTATGGGAAA for Vβ16s1; TV15 CAAAGCTGCTGTTCCACTACTA for Vβ24s1; TV16 GGTCCTGAAAAACGAGTTCAAG for Vβ25s1; TV18 GGTCTGAAATTCATGGTTTATCT for Vβ18s1; TV19 GACAGGACCCAGGGCAAG for Vβ17s1; TV20-1 ATGCTGATGGCAACTTCCA for Vβ2s1; TV24-1 CCTACGGTTGATCTATTACTCCTT for Vβ15s1; TV25-1 CTACACCTCATCCACTATTCCTA for Vβ11s1; TV27 GGGCTTAAGGCAGATCTACT for Vβ14s1; TV28 GGGCTACGGCTGATCTATTTC for Vβ3s1; TV29-1 CACTGATCGCAACTGCAA for Vβ4s1; and TV30 CCTCCAGCTGCTCTTCTA for Vβ20s1.

After PCR, 6 μl of the product and 3 μl of bromphenol blue-loading buffer were mixed and run on a 1% agarose gel for 45 min at 150 V. The gel was stained with Vistra Green (Amersham Life Science, Arlington Heights, IL) 1/10,000 dilution in 1× TBE for 50 min and analyzed on a FluorImager 595.

Vβ-specific DHDA was established by a modification of a previously described technique (21). From a donor CD8+ cell pool, a TCR with Vβ6s2 was amplified using TV7-8 and TC-1. The PCR product was cloned into pCR2.1 plasmid (TA-cloning kit; Invitrogen, San Diego, CA) and sequenced to ensure the correct Vβ. From this reference sequence and from C1-35, probes were generated using 6-carboxyfluorescein-labeled TC-1 primer. PCR was performed to obtain the fragments of the samples and the probes for DHDA. To chelate the Mg, 0.2 × vol of 25 mM EDTA was added to the probe. The heteroduplex generation was done with 2 μl of the sample PCR product and 2 μl of the probe by denaturating at 96°C for 5 min, then cooling rapidly to 50°C, and renaturating at this temperature for 45 min. The samples were mixed with 2 μl of loading buffer/Prism Genescan-500 TAMRA size marker (ABI Perkin-Elmer, Foster City, CA) and loaded on a native gel (4% bis-acrylamide:acrylamide 1:19). The gel was run in an ABI Sequencer 377 at 50°C for 2 h. The result was analyzed using the GeneScan software (ABI Perkin-Elmer).

The TCR β-chain of C1-35 was amplified with TC-1 and TV7-8 using Pfu polymerase, and the product was cloned into the TA vector. Sequences from both directions were obtained from five bacterial clones using the ABI PRISM Dye Primer kit and the ABI PRISM 377 DNA Sequencer. The sequences were analyzed with the ABI software, Sequence Analysis and Sequence Navigator. The Vβ6s2 PCR products of the other 16 cultures that were analyzed for clonal identity were sequenced directly from PCR products using ABI PRISM Dye Terminator kit. The sequences of the H.3-1 and F001TIL-9 were also obtained directly from PCR products using TV12-3/4 and TV4-2/3, respectively.

Clone-specific TCR β-chain primers were derived from the sequence of the CDR3 region of C1-35 (C35-2 CAT CGCCCCGCTCCCCCCAG) and F001TIL-9 (T9-2 AAGAACTGCTCATTGTAGTAAGTA). The direct PCR used the primers TV7-8 and C35-2 for the amplification of the C1-35 TCR β, and TV4-2/3 and T9-2 for the amplification of F001TIL-9 TCR β. The reaction mixture was composed of 10× PCR buffer, 1.5 mM MgCl2, 200 μM dNTP, 1.25 U AmpliTaq Gold, 0.5 μl cDNA, 1 μl each primer (10 μM), and water up to 20 μl. A total of 10 μl of light mineral oil covered the reaction mixtures, and PCR was run as described for TCR β PCR. A total of 6 μl of the PCR product and 3 μl of bromphenol blue-loading buffer were mixed and run on a 1% agarose gel for 45 min at 150 V. The gel was stained and analyzed as described for TCR β PCR. To control for the relative amount of T cells, a Cα control was run with the primers TAC-F1 (ATATCCAGAACCCTGACCTGC) and TCA-R1 (GCTTTTCTCGACCAGCTTGACATC). For the nested PCR, an amplification using the external primer RTV7-2 (ATCACACAGGRGCTGGAGT) for C1-35 and RTV4-2 (ATGGAAACGGGAGTTACG) for F001TIL-9 was performed before the clone-specific amplification. After the first amplification, the PCR product was diluted 1/5 with water, and from the dilution 1 μl used for the second amplification. The second amplification was done as described above for the direct PCR.

A single lesion from a patient with metastatic melanoma (F001) undergoing vaccination with g209-2M peptide was followed. The protocol was approved by Institutional Review Board of National Cancer Institute. The patient’s HLA class I phenotype determined by sequencing was A*0201/0301, B*0702/0801, Cw*0701/0702. After two vaccinations with g209-2M in combination with IL-12, an initial reduction of the tumor mass was observed on physical examination and radiographic evaluation. PBMC were collected before and after vaccination. Within 1 mo, the tumor mass became unresponsive to further treatment and progressed in size. At this point, a FNA of the mass was performed for analysis of tumor cells and TIL.

Cultures of CD8+ cells from pre- and postvaccination PBMC were generated using autologous DC either infected with rVV-MAA or loaded with 1 μg/ml g209 or g209-2M. When tested for MAA recognition after 2 wk in culture, no reactivity to g209 or g209-2M could be detected in prevaccination cultures, whereas postvaccination cultures demonstrated a strong sensitization against the natural as well as the modified gp100 epitope (Fig. 1). Although the patient had not been exposed to exogenous administration of MART-1, MART-1 reactivity could be equally observed in pre- and postvaccination cultures. This is not uncommon, as naturally occurring MART-1-specific reactivity can be readily detected in the peripheral circulation of HLA-A*0201-expressing melanoma patients (4).

FIGURE 1.

MAA-specific reactivity of CTL bulk cultures from pre- and postvaccination PBMC. After 2 wk, CTL bulk cultures were tested for IFN-γ release in response to T2 alone (gray bars) or T2 pulsed with MART-127–35 (open bars), g209 (hatched bars), or g209-2M (filled bars). MART-1 effectors, CTL culture stimulated with rVV-MART-1-infected autologous DC; g209 effectors, CTL culture stimulated with g209-pulsed autologous DC; g209-2M effectors, CTL culture stimulated with g209-2M-pulsed autologous DC. These results are representative of three independent experiments.

FIGURE 1.

MAA-specific reactivity of CTL bulk cultures from pre- and postvaccination PBMC. After 2 wk, CTL bulk cultures were tested for IFN-γ release in response to T2 alone (gray bars) or T2 pulsed with MART-127–35 (open bars), g209 (hatched bars), or g209-2M (filled bars). MART-1 effectors, CTL culture stimulated with rVV-MART-1-infected autologous DC; g209 effectors, CTL culture stimulated with g209-pulsed autologous DC; g209-2M effectors, CTL culture stimulated with g209-2M-pulsed autologous DC. These results are representative of three independent experiments.

Close modal

CTL clones were raised by limiting dilution from postvaccination CTL cultures induced in vitro. Cloning efficiency was about 100, 30, and 10% for the 100, 10, and 1 cell/well plates, respectively. Proliferating wells were tested for recognition of g209/g209-2M. From all g209/g209-2M-reactive 10 and 1 cell/well cultures, 25 clones were randomly selected for further expansion. After 2 wk of expansion, 18 of 25 cultures induced with g209-2M-pulsed DC maintained their specific reactivity. To assess the clonality of these 18 cultures, TCR β PCR was performed for a general overview: 17 of the 18 cultures showed a common strong band in Vβ6s2 (Fig. 2,A). Among these, all 12 cultures expanded from the 10 cell/well plates demonstrated additional weaker bands specific for other Vβ families, whereas no other bands were observed in the five cultures from 1 cell/well plates, suggesting the purity of these clones. To determine whether the Vβ6s2 bands represented the same TCR or originated from a different TCR utilizing the same Vβ, DHDA was performed with the Vβ6s2 PCR products from the 17 samples (Fig. 2,B). When an irrelevant TCR β with Vβ6s2 from a healthy donor was utilized as probe, all 17 samples showed a single heteroduplex band that migrated with identical delay relative to the homoduplex band. This indicated that the samples contained only one TCR and that the mismatches between the sample TCR β and the probe were similar, if not identical, among the clones. To validate the identity of these clones, one of the sample TCR β was labeled and used as probe against all other samples. With this probe, only a homoduplex was detected in all samples. These data strongly suggested identity of TCR β among the clones. The identity of the 17 TCR β-chains was verified additionally by sequencing each of them (Fig. 3). Thus, a predominant g209/g209-2M-specific clone expanded by the g209-2M vaccine was identified. For functional studies, one clone (C1-35) was selected as representative of the g209/g209-2M-reactive CTL population. The 18th clone (C10-80) included in the TCR β PCR analysis showed only a faint Vβ6s2 band and two strong bands in other Vβ.

FIGURE 2.

Analysis of clonality by TCR β PCR and DHDA of 18 cultures. A, TCR β PCR for 18 g209-reactive CTL clones generated by limiting dilution of g209-2M-induced postvaccination CTL cultures from patient F001. Each circle represents a strong band; weak bands are not included. Cultures with two strong bands are indicated with superscript numbers; circles with the same number belong to the same culture. FB CTL are limiting dilution cultures raised similarly to F001 CTL from PBL of a melanoma patient who had never received Ag-specific vaccination. In this case, the TCR β usage of CTL specific for the same epitope was highly polyclonal. B, DHDA was done with the predominant TCR Vβ6s2 PCR products of 17 cultures (only 14 of 17 shown) using an unrelated TCR Vβ6s2 with a different CDR3 (I) and C1-35 (II). I, Lanes 2–15, 14 samples + labeled probe; lane 16, unlabeled probe + labeled probe; lane 17, labeled probe alone. II, Lanes 18–31, 14 samples + labeled C1-35; lane 32, unlabeled probe + labeled C1-35; lane 33, labeled C1-35 alone. The two panels of electrophorograms below the gel display results of computer analyses for individual lanes. In the left panel, lanes 15, 16, 31, and 32 of the gel above are shown. The right panel exemplifies how different clones presenting identical bands in the TCR β PCR can be distinguished by DHDA. Three of four FB CTL clones with Vβ13s3 (see A) were analyzed. The heteroduplexes of two (middle and bottom) migrated identically, but one (top) migrated significantly faster, proving this clone to carry a different TCR β.

FIGURE 2.

Analysis of clonality by TCR β PCR and DHDA of 18 cultures. A, TCR β PCR for 18 g209-reactive CTL clones generated by limiting dilution of g209-2M-induced postvaccination CTL cultures from patient F001. Each circle represents a strong band; weak bands are not included. Cultures with two strong bands are indicated with superscript numbers; circles with the same number belong to the same culture. FB CTL are limiting dilution cultures raised similarly to F001 CTL from PBL of a melanoma patient who had never received Ag-specific vaccination. In this case, the TCR β usage of CTL specific for the same epitope was highly polyclonal. B, DHDA was done with the predominant TCR Vβ6s2 PCR products of 17 cultures (only 14 of 17 shown) using an unrelated TCR Vβ6s2 with a different CDR3 (I) and C1-35 (II). I, Lanes 2–15, 14 samples + labeled probe; lane 16, unlabeled probe + labeled probe; lane 17, labeled probe alone. II, Lanes 18–31, 14 samples + labeled C1-35; lane 32, unlabeled probe + labeled C1-35; lane 33, labeled C1-35 alone. The two panels of electrophorograms below the gel display results of computer analyses for individual lanes. In the left panel, lanes 15, 16, 31, and 32 of the gel above are shown. The right panel exemplifies how different clones presenting identical bands in the TCR β PCR can be distinguished by DHDA. Three of four FB CTL clones with Vβ13s3 (see A) were analyzed. The heteroduplexes of two (middle and bottom) migrated identically, but one (top) migrated significantly faster, proving this clone to carry a different TCR β.

Close modal
FIGURE 3.

Sequences of TCR β-chains around the CDR3. These sequence data are available from GenBank: F001 C1-35 under accession number AF037565, 1520 H.3–1 under AF037566, and F001 TIL-9 under accession number AF037567.

FIGURE 3.

Sequences of TCR β-chains around the CDR3. These sequence data are available from GenBank: F001 C1-35 under accession number AF037565, 1520 H.3–1 under AF037566, and F001 TIL-9 under accession number AF037567.

Close modal

C1-35 was able to lyse HLA-A*0201, gp100-expressing melanoma cell lines, including 624.38 MEL matched at three alleles (HLAA*0201/0301, -B*1402, -Cw*0702/0802), 526 MEL and 1102 MEL matched only at HLA-A*0201, but could not lyse the totally mismatched 888 MEL. Furthermore, the HLA-A*0201-expressing, gp100-negative F001TU-3 autologous cell line was insensitive to lysis by C1-35 (Fig. 4,A). Upon stimulation with other HLA-A*0201-matched melanoma cell lines, C1-35 was noted to secrete IFN-γ only in response to some targets characterized by high expression of HLA-A*0201 and gp100 (Fig. 4,B). As the ability of C1-35 to release IFN-γ when stimulated with HLA-A*0201-matched tumor lines could be dependent upon the avidity of TCR/epitope interactions (22), epitope density requirements for IFN-γ release were compared between C1-35 and a clone (H.3-1) from 1520 TIL. This TIL was derived from a metastatic lesion of a different patient not previously exposed to epitope-specific vaccination, and is characterized by high avidity interactions with gp100/HLA-A*0201-expressing melanoma cells. Loading T2 cells with decremental g209 or g209-2M doses, the epitope density needed to activate IFN-γ release by the two clones was compared. Although both were sensitive to stimulation, a 50- and a 100-fold lower concentration of g209 and g209-2M, respectively, were found to stimulate comparable amounts of IFN-γ release by H.3-1 compared with C1-35 (Fig. 4 C). Apparently, the g209-2M vaccination had induced and expanded in vivo a CTL clone with an avidity for its target that was lower than the one observed in a naturally occurring g209-specific TIL clone. As a consequence, C1-35 could recognize some, although not all, gp100-expressing, HLA-A*0201-matched melanoma cell lines recognized by H.3-1.

FIGURE 4.

Recognition of tumors by C1-35 and F001TIL-9 and functional avidity analysis. A, Calcein-AM cytotoxicity assay of C1-35 compared with H.3-1, a clone from a naturally occurring TIL culture (1520 TIL) recognizing g209/g209-2M. Lysis of Calcein-AM-labeled targets was measured at the E:T ratios of 10:1, 2.5:1, and 0.6:1. Targets included T2 + MART-127–35 (□), T2 + g209 (▪); the HLA-A*0201+/gp100+ 1102 MEL (•), 526 MEL (♦), 624.38 MEL (▴) lines; the HLA-A*0201−/gp100+ 888 MEL (▵) and the autologous gp100− F001TU-3 (○). B, IFN-γ release assay of C1-35 and TIL-9 from patient F001 compared with 1520 CTL and 1520 TIL. Targets included the HLA-Cw*0702-matched lines 624.38 MEL, 624.28 MEL, SK23 MEL, and autologous F001; the HLA-A*0201-matched, but not Cw*0702-matched, 1102 MEL and A 375 MEL; and the totally unmatched 888 MEL. C, IFN-γ release by C1-35 (▪) and H.3-1 (□) after incubation with T2 cells pulsed with 10-fold dilutions of MAA peptides ranging from 10 to 0.0001 μg/ml. Results are representative of three independent experiments.

FIGURE 4.

Recognition of tumors by C1-35 and F001TIL-9 and functional avidity analysis. A, Calcein-AM cytotoxicity assay of C1-35 compared with H.3-1, a clone from a naturally occurring TIL culture (1520 TIL) recognizing g209/g209-2M. Lysis of Calcein-AM-labeled targets was measured at the E:T ratios of 10:1, 2.5:1, and 0.6:1. Targets included T2 + MART-127–35 (□), T2 + g209 (▪); the HLA-A*0201+/gp100+ 1102 MEL (•), 526 MEL (♦), 624.38 MEL (▴) lines; the HLA-A*0201−/gp100+ 888 MEL (▵) and the autologous gp100− F001TU-3 (○). B, IFN-γ release assay of C1-35 and TIL-9 from patient F001 compared with 1520 CTL and 1520 TIL. Targets included the HLA-Cw*0702-matched lines 624.38 MEL, 624.28 MEL, SK23 MEL, and autologous F001; the HLA-A*0201-matched, but not Cw*0702-matched, 1102 MEL and A 375 MEL; and the totally unmatched 888 MEL. C, IFN-γ release by C1-35 (▪) and H.3-1 (□) after incubation with T2 cells pulsed with 10-fold dilutions of MAA peptides ranging from 10 to 0.0001 μg/ml. Results are representative of three independent experiments.

Close modal

Progressive tumor growth could have been due to lack of CTL localization at the tumor site, despite the presence of g209-specific CTL in the peripheral blood. To examine molecularly and functionally whether C1-35 was detectable at the time of progression, we performed a FNA of the growing metastasis at the moment of clinical progression. From the FNA material, expansion of TIL (IL-2, 6000 IU/ml added to CM) and autologous tumor was attempted. Twenty separate TIL cultures (F001TIL-1 to -20) and two tumor cell lines (F001TU-3 and -4) were generated.

To characterize the TCR repertoire in the PBMC, CTL cultures, and TIL, TCR β PCR with 35 Vβ-specific primers for 45 functional Vβ was performed (Fig. 5). CD8+ T cells from pre- and postvaccination PBMC showed a broad usage of Vβ with bands in 23 of 35 reactions. Postvaccination g209 CTL cultures showed little variation compared with the PBMC with losses of bands in Vβ8s3 (lane 7), Vβ13s3 (lane 29), Vβ13s5 (lane 30), and Vβ1s1 (lane 35), and relatively denser bands in Vβ21s3 (lane 4), Vβ15s1 (lane 16), and Vβ6s3/6s4/6s6 (lane 33). Postvaccination g209-2M CTL cultures presented a denser band corresponding to the Vβ6s2 (lane 34) chain utilized by the majority of the CTL clones expanded from the bulk culture, suggesting that C1-35 TCR expansion was not an artifact related to the cloning conditions. C1-35, representative of all other related clones, showed a dominant band corresponding to Vβ6s2 (lane 34). Faint bands for Vβ13s1/13s2 (lane 9), Vβ17s1 (lane 14), Vβ14s1 (lane 18), and Vβ3s1 (lane 19) were not detected with ethidium bromide staining, but only with the more sensitive Vistra Green staining, and were regarded as trace contamination. C10-80, the only clone of 18 that did not present a dominant Vβ6s2 (only a weak band in lane 34), was found to have predominantly Vβ8s1/8s2 (lane 6) and Vβ22s1 (lane 15). The FNA, which represents the local TCR β repertoire of the lesion, displayed bands for Vβ13s1/13s2 (lane 9), Vβ17s1 (lane 14), Vβ14s1 (lane 18), Vβ4s1 (lane 20), Vβ7s1 (lane 24), Vβ7s2/7s3 (lane 25), and Vβ6s2 (lane 34). Due to low amount of RNA from the FNA material, the intensity of the bands was much weaker, and it cannot be ruled out that other Vβs were missed for technical reasons. When F001TIL-9 was tested, it appeared to be almost pure, with only a faint band for Vβ6s2 (lane 34), besides the dominant band for Vβ7s2/7s3 (lane 25). Taken together, the monitoring of the TCR β repertoire of different original and in vitro culture samples allowed a general overview of the changes and clonality status. For example, based on TCR β PCR alone, cultures such as C10-80 could be identified as not clonal. This method, however, could not provide any information about the composition of a band, nor could it tell whether the same clone was responsible for corresponding bands in different samples. For the former limitation, DHDA and sequencing were applied as described above, and for the latter, clone-specific PCR analysis was performed.

FIGURE 5.

TCR β usage of PBMC, CTL cultures, TIL, and FNA. M, 100-bp ladder; A, β-actin control; lane 1, TV10-1/3 (Vβ12s1, 12s2); lane 2, TV10-2 (Vβ12s3); lane 3, TV11-1 (Vβ21s1); lane 4, TV11-2 (Vβ21s3); lane 5, TV11-3 (Vβ21s2); lane 6, TV12-3/4 (Vβ8s1, 8s2); lane 7, TV12-5 (Vβ8s3); lane 8, TV13 (Vβ23s1); lane 9, TV6-2/3/5 (Vβ13s1, 13s2); lane 10, TV14 (Vβ16s1); lane 11, TV15 (Vβ24s1); lane 12, TV16 (Vβ25s1); lane 13, TV18 (Vβ18s1); lane 14, TV19 (Vβ17s1); lane 15, TV2 (Vβ22s1); lane 16, TV24-1 (Vβ15s1); lane 17, TV25-1 (Vβ11s1); lane 18, TV27 (Vβ14s1); lane 19, TV28 (Vβ3s1); lane 20, TV29-1 (Vβ4s1); lane 21, TV20-1 (Vβ2s1); lane 22, TV3-1 (Vβ9s1); lane 23, TV30 (Vβ20s1); lane 24, TV4-1 (Vβ7s1); lane 25, TV4-2/3 (Vβ7s2, 7s3); lane 26, TV5-1 (Vβ5s1); lane 27, TV5-4/5/6 (Vβ5s2, 5s3, 5s6); lane 28, TV5-8 (Vβ5s4); lane 29, TV6-1 (Vβ13s3); lane 30, TV6-4 (Vβ13s5); lane 31, TV6-6 (Vβ13s6); lane 32, TV7-2/3 (Vβ6s1, 6s5); lane 33, TV7-6/7/9 (Vβ6s3, 6s4, 6s6); lane 34, TV7-8 (Vβ6s2); lane 35, TV9 (Vβ1s1).

FIGURE 5.

TCR β usage of PBMC, CTL cultures, TIL, and FNA. M, 100-bp ladder; A, β-actin control; lane 1, TV10-1/3 (Vβ12s1, 12s2); lane 2, TV10-2 (Vβ12s3); lane 3, TV11-1 (Vβ21s1); lane 4, TV11-2 (Vβ21s3); lane 5, TV11-3 (Vβ21s2); lane 6, TV12-3/4 (Vβ8s1, 8s2); lane 7, TV12-5 (Vβ8s3); lane 8, TV13 (Vβ23s1); lane 9, TV6-2/3/5 (Vβ13s1, 13s2); lane 10, TV14 (Vβ16s1); lane 11, TV15 (Vβ24s1); lane 12, TV16 (Vβ25s1); lane 13, TV18 (Vβ18s1); lane 14, TV19 (Vβ17s1); lane 15, TV2 (Vβ22s1); lane 16, TV24-1 (Vβ15s1); lane 17, TV25-1 (Vβ11s1); lane 18, TV27 (Vβ14s1); lane 19, TV28 (Vβ3s1); lane 20, TV29-1 (Vβ4s1); lane 21, TV20-1 (Vβ2s1); lane 22, TV3-1 (Vβ9s1); lane 23, TV30 (Vβ20s1); lane 24, TV4-1 (Vβ7s1); lane 25, TV4-2/3 (Vβ7s2, 7s3); lane 26, TV5-1 (Vβ5s1); lane 27, TV5-4/5/6 (Vβ5s2, 5s3, 5s6); lane 28, TV5-8 (Vβ5s4); lane 29, TV6-1 (Vβ13s3); lane 30, TV6-4 (Vβ13s5); lane 31, TV6-6 (Vβ13s6); lane 32, TV7-2/3 (Vβ6s1, 6s5); lane 33, TV7-6/7/9 (Vβ6s3, 6s4, 6s6); lane 34, TV7-8 (Vβ6s2); lane 35, TV9 (Vβ1s1).

Close modal

To assess the presence of clone C1-35 TCR in the different samples, clone-specific PCR was performed using a primer for the CDR3 region of the C1-35 TCR β-chain. The direct PCR clearly detected the clone in the g209-2M bulk culture from which the clone was derived. In addition, weak bands were visible in the postvaccination PBL, g209/g209-2M cultures, FNA, and F001TIL (Fig. 6). To corroborate this result, a nested PCR consisting of a Vβ-specific first amplification and a clone-specific second amplification was performed. The results of the nested PCR confirmed the bands in the postvaccination samples, while all prevaccination samples remained negative. By clone-specific analysis, F001TIL-9 could be detected only in TIL cultures, but not in the FNA from which it was expanded. This sensitivity limitation could not be overcome even by nested PCR (Fig. 6). Several unsuccessful attempts to tailor the PCR conditions to this specific reaction suggested that F001TIL-9 was present in minor proportion in vivo, although was readily sensitive to the proliferative stimulus provided in vitro by high dose IL-2. A panel of C region α-chain (C-α) amplifications was run along with the direct amplification to make semiquantitative assessments of the PCR results. Considering the low signal intensity for the C-α and the stronger signal for clone C1-35 TCR β-chain in FNA compared with the postvaccination CD8+ PBL preparations, the technically inevitable contamination of the FNA with peripheral blood is unlikely to solely account for the C1-35 TCR β signal detected in the FNA. These data suggested that lack of localization of C1-35 at tumor site could not explain the regained tumor growth.

FIGURE 6.

Clone-specific analysis by CDR3-specific PCR. C1-35 direct, clone-specific PCR with CDR3 primer specific for C1-35 TCR β; C1-35 nested, nested PCR for C1-35 with first amplification using the constant primer and external Vβ primer and second amplification identical to the direct PCR; TIL-9 direct, clone-specific PCR with CDR3 primer specific for TIL-9 TCR β; TIL-9 nested, nested PCR for TIL-9 with first amplification using constant primer and external Vβ primer and second amplification identical to the direct PCR. Cα control, amplification of Cα fragment to assess the amount of T lymphocytes in the samples. RNA samples were obtained from purified CD8+ preparations from PBMC obtained before (CD8-Pre) and after vaccination (CD8-Post), CTL cultures induced from the same CD8+ preparations with g209-pulsed DC (g209-Pre and g209-Post, respectively), or g209-2M-pulsed DC (g209-2M-Pre and g209-2M-Post, respectively). RNA was also prepared from C1-35 and F001TIL-9 CTL (as controls) from the postvaccination FNA from which the reagents described in this study were obtained (FNA), and from 1520 PBL as a negative control PBL from a different melanoma patient.

FIGURE 6.

Clone-specific analysis by CDR3-specific PCR. C1-35 direct, clone-specific PCR with CDR3 primer specific for C1-35 TCR β; C1-35 nested, nested PCR for C1-35 with first amplification using the constant primer and external Vβ primer and second amplification identical to the direct PCR; TIL-9 direct, clone-specific PCR with CDR3 primer specific for TIL-9 TCR β; TIL-9 nested, nested PCR for TIL-9 with first amplification using constant primer and external Vβ primer and second amplification identical to the direct PCR. Cα control, amplification of Cα fragment to assess the amount of T lymphocytes in the samples. RNA samples were obtained from purified CD8+ preparations from PBMC obtained before (CD8-Pre) and after vaccination (CD8-Post), CTL cultures induced from the same CD8+ preparations with g209-pulsed DC (g209-Pre and g209-Post, respectively), or g209-2M-pulsed DC (g209-2M-Pre and g209-2M-Post, respectively). RNA was also prepared from C1-35 and F001TIL-9 CTL (as controls) from the postvaccination FNA from which the reagents described in this study were obtained (FNA), and from 1520 PBL as a negative control PBL from a different melanoma patient.

Close modal

From the FNA material, two autologous tumor cell lines (F001TU-3 and F001TU-4) and 20 separate TIL cultures were generated. The TIL cultures were >95% CD8+ and were tested for recognition of g209 and HLA-matched tumors (Fig. 7). Surprisingly, none of the 20 TIL cultures recognized T2 targets pulsed with either g209 or g209-2M. TIL were also unable to recognize other known HLA-A*0201-associated MAA epitopes, including MART-127–35 (2, 3, 23, 24, 25). However, 12 of 20 cultures could recognize autologous as well as other HLA-B*0702- and HLA-Cw*0702-matched melanoma targets, but not the autologous EBV-B nor other matched nonmelanoma cells. To exclude the possibility of a transplantation Ag reaction, HLA typing of TIL was performed and found to be identical to the patient’s (A*0201/0301, B*0702/0801, Cw*0701/0702). Attempts to expand TIL populations from the FNA with OKT-3 and feeder cells (as used for cloning of CTL) also failed to generate g209/g209-2M-reactive TIL (data not shown). Thus, C1-35 and F001TIL-9 demonstrated a functional dissociation in epitope specificity between reagents obtained from the peripheral circulation and those obtained from the tumor (Fig. 4 B).

FIGURE 7.

Recognition of peptide-pulsed targets and tumors by 20 TIL cultures developed from a FNA. Twenty TIL cultures developed from a FNA obtained from a progressing lesion post-g209-2M vaccination were tested for IFN-γ release in response to MART-127–35 or g209-pulsed T2 cells and in response to a HLA-matched melanoma 624.38 MEL (matched at HLA-A*0201,0301 and -Cw*0702). The melanoma cell line 888 (888 MEL) represented a totally HLA class I- and class II-mismatched negative control.

FIGURE 7.

Recognition of peptide-pulsed targets and tumors by 20 TIL cultures developed from a FNA. Twenty TIL cultures developed from a FNA obtained from a progressing lesion post-g209-2M vaccination were tested for IFN-γ release in response to MART-127–35 or g209-pulsed T2 cells and in response to a HLA-matched melanoma 624.38 MEL (matched at HLA-A*0201,0301 and -Cw*0702). The melanoma cell line 888 (888 MEL) represented a totally HLA class I- and class II-mismatched negative control.

Close modal

As noted in Figure 4,B, C1-35, 1520 TIL, and its high avidity clone H.3-1 failed to recognize F001TU-3. These findings suggested loss of either HLA-A*0201 or gp100 expression by the autologous tumor. FACS analysis of F001TU-3 and -4 demonstrated loss of gp100 expression and retention of expression of HLA-A*0201 (Fig. 8,A). Sequencing of F001TU-3 cDNA ruled out mutations of the HLA-A*0201 heavy chain. Because of the possibility that F001TU-3 and -4 had originated from rare cells in the FNA material not representative of the tumor, cytospin preparations from the original FNA were analyzed by immunocytochemistry and compared with FNA material obtained from the same lesion at various time points. This analysis confirmed a marked decrease in gp100-expressing cells postvaccination (Fig. 8,B), while the expression of HLA-A*0201 remained unchanged. Indeed, while gp100 was detectable in >75% of tumor cells in the prevaccination FNA, less than 5% of cells expressed gp100 postvaccination (Table I). MART-1 expression was not affected by the vaccination, although analysis of F001TU-3 and -4 revealed decreased expression of this MAA. Expression of HLA-A*0201 was similar in all FNA analyzed. Although C1-35 and H.3-1 could not naturally recognize F001TU-3 and -4 (Fig. 4, A and B), exogenous loading of peptide on F001TU and other HLA-A*0201 melanomas not recognized by C1-35 could stimulate IFN-γ release (Fig. 9). These data suggest that the poor recognition of autologous tumor by C1-35 and H.3-1 was due to inadequate epitope density on the cell surface rather than abnormalities of the HLA-A2 heavy chain or killer inhibitory receptor-HLA interactions (26). Thus, tumor escape from peptide vaccination was associated with severely decreased expression of target Ag by the tumor, which led to proliferation of a cell population not recognizable not only by the intermediate avidity CTL elicited by the vaccination, but also by high affinity CTL effectors.

FIGURE 8.

A, Analysis of HLA-A2 and MAA expression by F001TU-3 and F001TU-4 melanomas. F001TU-3 and F001TU-4 were analyzed by FACS for the expression of HLA-A2, MART-1, and gp100 Ags. Tumor cells were stained with mAbs specific for the HLA-A2 (KS1 (18)), MART-1 (M2-7C10 mAb (19)), and gp100 (HMB-45). The EBV-B cell line 888 and the melanoma cell line SK23 were used as negative and positive controls for the expression of MAA and HLA-A2, respectively. B, Immunocytochemical analysis of HLA-A2 and MAA expression in FNA obtained pre- and postvaccination with 209-2M peptide. Cytospin preparations of FNA material obtained from a metastasis pre-g209-2M vaccination (A, C, E) and from the same progressing lesion postvaccination (B, D, F). Cytospins were stained with HMB-45 mAb (Biogenex) for detection of gp100 (A, B), M2-7C10 mAb (19) for MART-1 (C, D), and KS-1 (18) for HLA-A2 (E, F). Positive cells are indicated by the brown chromogen 3,3′-diaminobenzidine. All cells were counterstained with hematoxylin (blue).

FIGURE 8.

A, Analysis of HLA-A2 and MAA expression by F001TU-3 and F001TU-4 melanomas. F001TU-3 and F001TU-4 were analyzed by FACS for the expression of HLA-A2, MART-1, and gp100 Ags. Tumor cells were stained with mAbs specific for the HLA-A2 (KS1 (18)), MART-1 (M2-7C10 mAb (19)), and gp100 (HMB-45). The EBV-B cell line 888 and the melanoma cell line SK23 were used as negative and positive controls for the expression of MAA and HLA-A2, respectively. B, Immunocytochemical analysis of HLA-A2 and MAA expression in FNA obtained pre- and postvaccination with 209-2M peptide. Cytospin preparations of FNA material obtained from a metastasis pre-g209-2M vaccination (A, C, E) and from the same progressing lesion postvaccination (B, D, F). Cytospins were stained with HMB-45 mAb (Biogenex) for detection of gp100 (A, B), M2-7C10 mAb (19) for MART-1 (C, D), and KS-1 (18) for HLA-A2 (E, F). Positive cells are indicated by the brown chromogen 3,3′-diaminobenzidine. All cells were counterstained with hematoxylin (blue).

Close modal
Table I.

Immunocytochemistry of FNA material for MAA and HLA-A2 expression

Time of FNADateExpression of
gp100aMART-1aHLA-A*0201
Prevaccination 1/30/1997 50–75 25–50 Positive 
Postvaccination 5/28/1997 <5 50–75 Positive 
 6/18/1997 <5 50–75 Positive 
 7/11/1997 <5 50–75 Positive 
Time of FNADateExpression of
gp100aMART-1aHLA-A*0201
Prevaccination 1/30/1997 50–75 25–50 Positive 
Postvaccination 5/28/1997 <5 50–75 Positive 
 6/18/1997 <5 50–75 Positive 
 7/11/1997 <5 50–75 Positive 
a

Data are % of staining cells.

FIGURE 9.

Recognition of naturally processed g209-2M by F001 and 1520 CTL. F001 CTL and F001C1-35 were tested by IFN-γ release assay for recognition of the autologous F001TU-3 and F001TU-4; the HLA-A*0201-matched melanomas 624.38 MEL, A375 MEL, SK23 MEL. 624.38 MEL, and SK23 MEL were selected as cell lines with intermediate expression of HLA-A*0201 and gp100. These lines can be killed in cytotoxicity assays, but do not stimulate IFN-γ release by C1-35 (see Fig. 4 B). The 209-2 M 1520 CTL, 1520 TIL, and the 1520 TIL clones H.3-1 and H.3-6 obtained from a different melanoma patient. 1520 TIL were expanded from a metastasis before vaccination, while the 1520 CTL cultures were induced in vitro by stimulation of PBMC obtained from the same patient post-209-2 M vaccination. T2 cells pulsed with gp209 were also included, as they can efficiently present the relevant peptide and stimulate IFN-γ release by peptide-specific reactive CTL. The melanoma cell line 888 (888 MEL) represented the HLA-A-mismatched negative control (HLA-A1, 24). Filled bars represent IFN-γ release in response to stimulator cells pulsed with g209 (1 μg/ml). Open bars indicate IFN-γ release in response to nonpulsed stimulator cells. Results are representative of three independent experiments.

FIGURE 9.

Recognition of naturally processed g209-2M by F001 and 1520 CTL. F001 CTL and F001C1-35 were tested by IFN-γ release assay for recognition of the autologous F001TU-3 and F001TU-4; the HLA-A*0201-matched melanomas 624.38 MEL, A375 MEL, SK23 MEL. 624.38 MEL, and SK23 MEL were selected as cell lines with intermediate expression of HLA-A*0201 and gp100. These lines can be killed in cytotoxicity assays, but do not stimulate IFN-γ release by C1-35 (see Fig. 4 B). The 209-2 M 1520 CTL, 1520 TIL, and the 1520 TIL clones H.3-1 and H.3-6 obtained from a different melanoma patient. 1520 TIL were expanded from a metastasis before vaccination, while the 1520 CTL cultures were induced in vitro by stimulation of PBMC obtained from the same patient post-209-2 M vaccination. T2 cells pulsed with gp209 were also included, as they can efficiently present the relevant peptide and stimulate IFN-γ release by peptide-specific reactive CTL. The melanoma cell line 888 (888 MEL) represented the HLA-A-mismatched negative control (HLA-A1, 24). Filled bars represent IFN-γ release in response to stimulator cells pulsed with g209 (1 μg/ml). Open bars indicate IFN-γ release in response to nonpulsed stimulator cells. Results are representative of three independent experiments.

Close modal

With the identification of CTL epitopes believed to be clinically relevant, peptide-based vaccinations have been developed. The rationale is to provide proper stimulatory conditions that could adequately turn on an immune system otherwise insensitive to Ags expressed by tumor cells (27). Murine models predict that tumors do not induce spontaneously nor maintain an activated CTL response even after activation of protective T cells by vaccines, and repetitive immunizations are necessary for prolonged effects (28). The primary physiologic effect of vaccination is a systemic enhancement of responsive T cells. The desired secondary effect, however, is the localization of activated CTL at the tumor site to perform their therapeutic functions. We have shown that the s.c. administration of MAA epitopes in IFA induces MAA-specific CTL detectable in the peripheral blood that can recognize naturally processed epitopes on the surface of tumor cells (7, 8). However, by itself, peptide administration rarely leads to tumor regression. Clinical response after vaccination is generally attributed to the ability of activated CTL to localize at tumor site and kill tumor cells. However, the exact mechanism responsible for the more common therapeutic failures is unclear. Analysis of tumor/host interactions at the site of disease may provide such information. Excisional biopsy of tumors and subsequent expansion of TIL/tumor pairs is a useful tool for the analysis of tumor/host interactions at a given point (29). Although useful, this strategy has not yielded conclusive information for three reasons. First, homogeneity among tumors must be assumed to take the excised lesion as representative of other metastases left in vivo for clinical correlation. However, synchronous metastases are often heterogeneous in expression of MAA and HLA (30). Second, the natural course of the tumor cannot be determined after its removal: only a retrospective correlation can be performed between clinical parameters and characteristics of the reagents obtained from the biopsy. Finally, the removal of the tumor excludes comparative studies of the same lesion at different points in time in relation to the natural progression of the disease or in response to immune pressure.

To overcome the limitations posed by excisional biopsy, we suggest following metastases by serial FNA, which allows the evaluation of tumors at various points (16). By following the same lesion serially, heterogeneity among tumors can be avoided as a confounding factor. The ability to expand TIL and autologous tumor from the FNA permits the analysis of CTL localization and function at tumor site. This strategy was tested on a melanoma patient with a metastasis of particular interest: the mass had shrunk after vaccination with g209-2M, suggesting effectiveness of treatment. However, after the initial shrinkage, the lesion became insensitive to further g209-2M vaccines.

The identification of g209/g209-2M-reactive CTL from postvaccination PBMC, in concordance with the lack of C1-35 detection by PCR in prevaccination samples, was an indication of successful systemic induction and expansion of vaccine-specific CTL. The presence of g209-reactive CTL at that time point suggested that tumor progression was not the result of functional deletion of tumor-specific CTL by the vaccine, as suggested by some murine models (31). The predominance of a single clone after vaccination was of interest and contrasted with the capacity observed in naturally occurring CTL, recognizing immunodominant epitopes such as MART-127–35 (32) or EBNA4416–424 (33) to maintain a broad TCR repertoire. In fact, we observed, in a second melanoma patient not previously exposed to MAA-specific vaccination, a widely polyclonal, MART-127–35-specific CTL population characterized by broad Vβ usage.

Several explanations are plausible for the predominance of C1-35, including an artifact of in vitro culture (34), a consequence of the vaccination procedure, or a direct effect of the high affinity interactions between g209-2M and HLA-A*0201 (11). In a functional peptide dilution assay, C1-35 demonstrated a lower avidity for its target than the naturally occurring H.3-6 TIL. Alexander-Miller et al. proposed that high dose peptide stimulation could inhibit high avidity CTL while maximally stimulating low avidity CTL (35). Since g209-2M is characterized by high affinity for HLA-A*0201 (11), it is possible that the epitope density reached in vivo as a result of dose and route of administration led to extinction of high avidity CTL in this patient and stimulated the expansion of a low avidity clone, C1-35. This finding, if confirmed, will have important implications for the administration of vaccines, and studies of the avidity of the predominant CTL after vaccination could become necessary to find optimal dose ranges for peptide-based vaccines. It could be postulated that the vaccination led to the in vivo loss of highly reactive g209-specific CTL, and such loss could explain the progression of tumor after initial response. Furthermore, the repeated in vitro sensitizations with g209-2M could have skewed the original proportion of g209-specific CTL present in postvaccination PBMC. One cannot rule out that we may have missed other g209-specific CTL, possibly with higher avidity. To minimize this bias, we reduced the time in bulk culture to 14 days before cloning. Direct cloning from peripheral blood would be the preferred method with minimal in vitro bias. However, multiple attempts to directly clone PBMC resulted in poor efficiency and lack of generation of reactive CTL.

Although C1-35 demonstrated lower avidity for its target compared with a naturally occurring TIL, it could kill with high efficiency all HLA-A*0201-matched, gp100-expressing melanoma targets and could release IFN-γ in response to stimulation with several melanoma cell lines characterized by combined high expression of HLA-A2 and gp100. Gervois et al. have shown that the epitope density requirements necessary to stimulate target cell kill by CTL are 10- to 10,000-fold less than that required for induction of IL-2 and IFN-γ release by the same CTL (22). Tumor cells that could be efficiently lysed by MAA-specific CTL could not stimulate IFN-γ and/or IL-2 release unless exogenously supplemented with the appropriate peptide. In this study, we confirm this finding and we postulate that C1-35 could have been responsible for the initial response of the metastasis by killing melanoma cells, while the stimulatory requirements for its expansion and activation were provided by the systemic administration of g209-2M in IFA. At the time of tumor progression, although vaccine-induced stimulation was ongoing, most tumor cells became resistant to lysis by losing expression of target Ag. Comparison of FNA material obtained before and after vaccination revealed that the tumor had drastically decreased the fraction of cells expressing the gp100 over time. We hypothesize that this loss was due to specific killing of gp100-positive cells by CTL expanded by the vaccine, including C1-35. C1-35 appeared to still be present in the tumor mass at the time of progression (perhaps in response to few remaining gp100-expressing tumor cells or the effects of the still ongoing vaccination). However, its presence correlated with an inactive status, as none of 20 TIL bulk cultures expanded from the postvaccination FNA reacted to g209/g209-2M. This might indicate a dormant state of C1-35 in the tumor secondary to the decreased expression of gp100. The detection of TIL that recognize tumor cells, but not g209-2M or g209, suggests that gp100 loss by the tumor was counteracted, at that time point, by induction of CTL with another specificity, and underscores the dynamic and interactive nature of the immune response at the tumor site.

In this study, several observations could be made that would have escaped detection by monitoring the systemic immune response alone. First, CTL activated by the vaccine, although capable of recognizing and killing other gp100-expressing melanomas, were unable to recognize autologous tumor from a progressing metastatic lesion. The finding could be best explained by lost (or severely decreased) expression of the target MAA at that point in time. Second, despite the lack of target Ag by the tumor, C1-35 was found to localize at tumor site by molecular methods. The presence at tumor site, however, was associated with a dormant state, as this CTL could not be expanded by general proliferative stimuli consisting of IL-2 or OKT-3. Third, a new TIL emerged that could recognize an unidentified MAA in association with a restriction element unrelated to the vaccination (HLA-Cw*0702). Furthermore, the use of FNA allowed for serial sampling of the same metastasis without interference with its clinical course throughout treatment and afterward. This permitted a direct correlation between functional studies and therapeutic outcome.

This study illustrates the necessity of analyzing target tissue/host interactions at the site in which they are likely to occur. Such information may complement data obtained with the analysis of the systemic effects of vaccination and might enhance the understanding of the complex mechanisms underlying the success and failure of vaccination.

We are grateful to Dr. William E. Biddison for critical review of the manuscript and helpful discussions. We thank Theresa Arthur for excellent technical assistance.

2

Abbreviations used in this paper: MAA, melanoma-associated antigen; CDR, complementarity-determining region; CM, complete medium; DHDA, direct heteroduplex analysis; FNA, fine needle aspirate; rVV, recombinant vaccinia virus; TIL, tumor-infiltrating lymphocyte.

1
Rosenberg, S. A..
1997
. Cancer vaccines based on the identification of genes encoding cancer regression antigens.
Immunol. Today
18
:
175
2
Kawakami, Y., S. Eliyahu, K. Sakaguchi, P. F. Robbins, L. Rivoltini, J. R. Yannelli, E. Appella, S. A. Rosenberg.
1994
. Identification of the immunodominant peptides of the MART-1 human melanoma antigen recognized by the majority of HLA-A2-restricted tumor infiltrating lymphocytes.
J. Exp. Med.
180
:
347
3
Kawakami, Y., S. Eliyahu, C. Jennings, K. Sakaguchi, X. Kang, S. Southwood, P. F. Robbins, A. Sette, E. Appella, S. A. Rosenberg.
1995
. Recognition of multiple epitopes in the human melanoma antigen gp100 by tumor-infiltrating T lymphocytes associated with in vivo tumor regression.
J. Immunol.
154
:
3961
4
Rivoltini, L., Y. Kawakami, K. Sakaguchi, S. Southwood, A. Sette, P. F. Robbins, F. M. Marincola, M. Salgaller, J. R. Yannelli, E. Appella, S. A. Rosenberg.
1995
. Induction of tumor reactive CTL from peripheral blood and tumor infiltrating lymphocytes of melanoma patients by in vitro stimulation with an immunodominant peptide of the human melanoma antigen MART-1.
J. Immunol.
154
:
2257
5
Marincola, F. M., L. Rivoltini, M. L. Salgaller, M. Player, S. A. Rosenberg.
1996
. Differential anti-MART-1/MelanA CTL activity in peripheral blood of HLA-A2 melanoma patients in comparison to healthy donors: evidence for in vivo priming by tumor cells.
J. Immunother.
19
:
266
6
Salgaller, M. L., A. Afshar, F. M. Marincola, L. Rivoltini, Y. Kawakami, S. A. Rosenberg.
1995
. Recognition of multiple epitopes in the human melanoma antigen gp100 by peripheral blood lymphocytes stimulated in vitro with synthetic peptides.
Cancer Res.
55
:
4972
7
Cormier, J. N., M. L. Salgaller, T. Prevette, K. C. Barracchini, L. Rivoltini, N. P. Restifo, S. A. Rosenberg, F. M. Marincola.
1997
. Enhancement of cellular immunity in melanoma patients immunized with a peptide from MART-1/Melan A.
Cancer J. Sci. Am.
3
:
37
8
Salgaller, M. L., F. M. Marincola, J. N. Cormier, S. A. Rosenberg.
1996
. Immunization against epitopes in the human melanoma antigen gp100 following patient immunization with synthetic peptides.
Cancer Res.
56
:
4749
9
Rosenberg, S. A., J. C. Yang, D. Schwartzentruber, P. Hwu, F. M. Marincola, S. L. Topalian, N. P. Restifo, E. Dufour, L. Schwartzberg, P. Spiess, J. Wunderlich, M. R. Parkhurst, Y. Kawakami, C. Seipp, J. H. Einhorn, D. White.
1998
. Immunologic and therapeutic evaluation of a synthetic tumor associated peptide vaccine for the treatment of patients with metastatic melanoma.
Nat. Med.
4
:
321
10
Arden, B., S. P. Clark, D. Kabelitz, T. W. Mak.
1995
. Mouse T-cell receptor variable gene segment families.
Immunogenetics
42
:
501
11
Parkhurst, M. R., M. L. Salgaller, S. Southwood, P. F. Robbins, A. Sette, S. A. Rosenberg, Y. Kawakami.
1996
. Improved induction of melanoma reactive CTL with peptides from the melanoma antigen gp100 modified at HLA-A*0201 binding residues.
J. Immunol.
157
:
2539
12
Rivoltini, L., K. C. Baracchini, V. Viggiano, Y. Kawakami, A. Smith, A. Mixon, N. P. Restifo, S. L. Topalian, T. B. Simonis, S. A. Rosenberg, F. M. Marincola.
1995
. Quantitative correlation between HLA class I allele expression and recognition of melanoma cells by antigen specific cytotoxic T lymphocytes.
Cancer Res.
55
:
3149
13
Wei, M. L., P. Cresswell.
1992
. HLA-A2 molecules in an antigen-processing mutant cell contain signal sequence-derived peptides.
Nature
356
:
443
14
Bunce, M., C. M. O’Neill, M. C. Barnardo, P. Krausa, M. J. Browning, P. J. Morris, K. I. Welsh.
1995
. Phototyping: comprehensive DNA typing for HLA-A, B, C, DRB1, DRB3, DRB4, DRB5 & DQB1 by PCR with 144 primer mixes utilizing sequence-specific primers (PCR-SSP).
Tissue Antigens
46
:
355
15
P., Krausa, M. Brywka, D. Savage, K. M. Hui, M. Bunce, J. L. Ngai, D. L. Teo, Y. W. Ong, D. Barouch, C. E. Allsop, et al
1995
. Genetic polymorphism within HLA-A*02: significant allelic variation revealed in different populations.
Tissue Antigens
45
:
223
16
Marincola, F. M., Y. M. Hijazi, P. Fetsch, M. L. Salgaller, L. Rivoltini, J. Cormier, T. B. Simonis, P. H. Duray, M. Herlyn, Y. Kawakami, S. A. Rosenberg.
1996
. Analysis of expression of the melanoma associated antigens MART-1 and gp100 in metastatic melanoma cell lines and in in situ lesions.
J. Immunother.
19
:
192
17
Brodsky, F. M., P. Parham.
1982
. Monomorphic anti-HLA-A,B,C monoclonal antibodies detecting molecular subunits and combinatorial determinants.
J. Immunol.
128
:
129
18
Tsujisaki, M., K. Sakaguchi, M. Igarashi, P. Richiardi, F. Perosa, S. Ferrone.
1988
. Fine specificity and idiotype diversity of the murine anti-HLA-A2, A28 monoclonal antibodies CR11-351 and KS1.
Transplantation
45
:
632
19
Kawakami, Y., J. K. Battles, T. Kobayashi, W. Ennis, X. Wang, J. P. Tupesis, F. M. Marincola, P. F. Robbins, V. J. Hearing, M. A. Gonda, S. A. Rosenberg.
1997
. Production of recombinant MART-1 proteins and specific antiMART-1 polyclonal and monoclonal antibodies: use in the characterization of the human melanoma antigen MART-1.
J. Immunol. Methods
202
:
13
20
Kim, C. J., T. Prevette, J. Cormier, W. Overwijk, M. Roden, N. P. Restifo, S. A. Rosenberg, F. M. Marincola.
1997
. Dendritic cells infected with poxviruses encoding MART-1/MelanA sensitize T lymphocytes in vitro.
J. Immunother.
20
:
276
21
Zimmerman, P. A., M. N. Carrington, T. B. Nutman.
1993
. Exploiting structural differences among heteroduplex molecules to simplify genotyping the DQA1 and DQB1 alleles in human lymphocyte typing.
Nucleic Acids Res.
21
:
4541
22
Gervois, N., Y. Guilloux, E. Diez, F. Jotereau.
1996
. Suboptimal activation of melanoma infiltrating lymphocytes (TIL) due to low avidity of TCR/MHC-tumor peptide interactions.
J. Exp. Med.
183
:
2403
23
Wolfel, T., A. Van Pel, V. Brichard, J. Schneider, B. Seliger, K. H. Meyer zum Buschenfelde, T. Boon.
1994
. Two tyrosinase nonapeptides recognized on HLA-A2 melanomas by autologous cytolytic T lymphocytes.
Eur. J. Immunol.
24
:
759
24
Van der Bruggen, P., J. Bastin, T. Gajewski, P. G. Coulie, P. Boel, C. De Smet, C. Traversari, A. Townsend, T. Boon.
1994
. A peptide encoded by human gene MAGE-3 and presented by HLA-A2 induces cytolytic T lymphocytes that recognize tumor cells expressing MAGE-3.
Eur. J. Immunol.
24
:
3038
25
Guilloux, Y., S. Lucas, V. G. Brichard, A. Van Pel, C. Viret, E. De Plaen, F. Brasseur, B. Lethe, F. Jotereau, T. Boon.
1996
. A peptide recognized by human cytolytic T lymphocytes on HLA-A2 melanomas is encoded by an intron sequence of the N-acetylglucosaminyltransferase V gene.
J. Exp. Med.
183
:
1173
26
Ikeda, H., B. Lethe, F. Lehmann, N. van Baren, J. F. Baurain, C. De Smet, H. Chambost, M. Vitale, A. Moretta, T. Boon, P. G. Coulie.
1997
. Characterization of an antigen that is recognized on a melanoma showing partial HLA loss by CTL expressing an NK inhibitory receptor.
Immunity
6
:
199
27
Fuchs, E. J., P. Matzinger.
1996
. Is cancer dangerous to the immune system?.
Semin. Immunol.
8
:
271
28
Speiser, D. E., R. Miranda, A. Zakarian, M. F. Bachmann, K. McKall-Faienza, B. Odermatt, D. Hanahan, R. M. Zinkernagel, P. S. Ohashi.
1997
. Self antigens expressed by solid tumors do not efficiently stimulate naive or activated T cells: implications for immunotherapy.
J. Exp. Med.
186
:
645
29
Pandolfi, F., L. A. Boyle, L. Trentin, J. T. Kurnick, K. J. Isselbacher, S. Gattoni-Celli.
1991
. Expression of HLA-A2 antigen in human melanoma cell lines and its role in T-cell recognition.
Cancer Res.
51
:
3164
30
Cormier, J. N., Y. M. Hijazi, A. Abati, P. Fetsch, M. Bettinotti, S. M. Steinberg, S. A. Rosenberg, F. M. Marincola.
1998
. Heterogeneous expression of melanoma-associated antigens (MAA) and HLA-A2 in metastatic melanoma in vivo.
Int. J. Cancer
75
:
517
31
Toes, R. E., R. Offringa, R. J. Blom, C. J. Melief, W. M. Kast.
1996
. Peptide vaccination can lead to enhanced tumor growth through specific T-cell tolerance induction.
Proc. Natl. Acad. Sci. USA
93
:
7855
32
Cole, D. J., D. P. Weil, P. Shamamian, L. Rivoltini, Y. Kawakami, S. Topalian, C. Jennings, S. Eliyahu, S. A. Rosenberg, M. I. Nishimura.
1994
. Identification of MART-1-specific T-cell receptors: T cells utilizing distinct T-cell receptor variable and joining regions recognize the same tumor epitope. [Published erratum appears in 1994 Cancer Res 54:6014.].
Cancer Res.
54
:
5265
33
Campos-Lima, P. O., V. Levitsky, M. P. Imreh, R. Gavioli, M. G. Masucci.
1997
. Epitope-dependent selection of highly restricted or diverse T cell receptor repertoires in response to persistent infection by Epstein-Barr virus.
J. Exp. Med.
186
:
83
34
Dufour, E., G. Carcelain, C. Gaudin, C. Flament, M. F. Avril, F. Faure.
1997
. Diversity of the cytotoxic melanoma-specific immune response: some CTL clones recognize autologous fresh tumor cells and not tumor cell lines.
J. Immunol.
158
:
3787
35
Alexander-Miller, M. A., G. R. Leggatt, A. Sarin, J. A. Berzofsky.
1996
. Role of antigen, CD8, and cytotoxic T lymphocyte (CTL) avidity in high dose antigen induction of apoptosis of effector CTL.
J. Exp. Med.
184
:
485