A Tumor-Infiltrating Lymphocyte from a Melanoma Metastasis with Decreased Expression of Melanoma Differentiation Antigens Recognizes MAGE-12

Cell lines

The bulk melanoma cell line 624-MEL (HLA-A*0201/0301, B*1402/0702, Cw*0702/0802) was generated by limiting dilution from a metastasis. In this study, two clones from 624-MEL were used that are characterized by a similar pattern of MA and HLA allele expression with the exception of HLA-A*0201. One clone (624.38-MEL) maintains expression of this allele, while the other (624.28-MEL) does not because of aberrant pre-mRNA splicing of the HLA-A*0201 transcript (32). A375-MEL and SK23-MEL were purchased from the American Type Culture Collection (Manassas, VA). 397-MEL, 537-MEL, 697-MEL, 888-MEL, 938-MEL, 1102-MEL, 1123-MEL, 1280-MEL, 1359-MEL, and 1495-MEL are archival cell lines derived from surgically removed melanoma metastases and have been previously characterized extensively with regard to HLA and MA expression (33). F001-MEL, F002-MEL, and F010-MEL cell lines consist of early passage (<5 passages) melanoma cell lines derived from fine needle aspirate (FNA) of melanoma metastases from different patients (34). B lymphoblastoid cells (F001-EBV, 583-EBV) were derived from patients’ PBMC transformed with exogenous EBV virus (35). T2 cells (American Type Culture Collection) were used to test CTL specificity toward HLA-A*0201-restricted epitopes (36). 293-HEK (American Type Culture Collection) is a line of easily transfectable primary human embryonal kidney cells transformed with human adenovirus type 5 DNA. All cell lines were maintained in complete medium (CM) consisting of RPMI 1640 (Biofluids, Rockville, MD) 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 FBS (Biofluids). 293-HEK cells were maintained in DMEM (Biofluids) supplemented as described for CM except 7.5% FBS. Normal human epithelial melanocytes (NHEM) generated from human foreskin were donated by Dr. Mehenard Herlyn (Wistar Institute, Philadelphia, PA) and expanded as described (37) in melanocyte growth medium (Clonetics, San Diego, CA).

Peptides

Peptides were produced by solid-phase synthesis techniques and solubilized in sterile water or DMSO (Sigma, St. Louis, MO) according to their biochemical characteristics. Peptide identity was confirmed by mass spectral analysis. Relevant peptide sequences are shown throughout the text.

Expansion and characterization of F001-TIL from FNA

Using a 23-gauge needle, cells were aspirated from an axillary metastasis of patient F001. The material obtained from the FNA was immediately suspended in Iscove’s Medium (IM) (Biofluids) prepared as the CM with the exception of 10% heat inactivated human AB serum (Gemini Bioproducts, Calabasas, CA). Bulk TIL cultures were initiated in a Costar 24-well plate (Corning, Corning, NY) at a concentration of 5 × 10⁶ cells/well in 2 ml of IM supplemented with 6000 IU/ml hrIL-2 (Chiron, Emeryville CA). Twenty separate cultures were attempted. Two melanoma cell lines were expanded in different wells by plating in 24-well plates material from the same FNA in CM without hrIL-2. Preliminary characterization of two distinct cell lines demonstrated that they were practically identical in MA and HLA molecule expression. Therefore, for the remainder of the manuscript they will be referred to as F001-MEL. MA recognition by F001-TIL was assessed by IFN-γ release assay: 1 × 10⁵ effector cells were 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 harvested for analysis by ELISA (Endogen, Cambridge, MA). IFN-γ is reported as pg/ml secreted by 5 × 10⁴ effector cells in 24 h, and values double the background and >100 pg/ml were considered positive.

FACS staining of F001-MEL and F001-TIL

Cell-surface expression of HLA and MA by FACS staining has been previously described (38). In particular the following mAbs were used for evaluation of expression of HLA-class I alleles: KS-1 for HLA-A*0201 (33), GAP-A3 for HLA-A*0301 (39), and SFR8-B6 (40) recognizing the public epitope HLA-Bw6 common to HLA-B*0801 and -B*0702 present in F001 genotype. Abs used to characterize F001-TIL included: anti-human CD8-FITC, CD4-PE, TCR-FITC, CD45RA-FITC, CD45RO-PE, CD95L, CD11b (MAC-1)-PE, CD44(pgp-1)-FITC, CD152(CTLA-4)-PE (PharMingen, San Diego, CA), CD56-PE, CD28-FITC, CD62L-PE, CD154-PE, (Becton Dickinson, San Jose, CA), and FITC-anti-human CD49a, CD49b, CD49d, CD49e, CD49f (Serotec, Raleigh, NC). Primary staining with the purified Ab CD95L was followed by secondary staining with FITC-goat anti-mouse IgG.

Cloning of HLA-B*0702 and -Cw*0702

The HLA-B*0702 and Cw*0702 alleles were isolated from total RNA from F001-EBV-B cells using the RNeasy kit (Qiagen, Chatsworth, CA). RNA (4 μg) was transcribed to first-strand cDNA using the SuperScript Preamplification System (Life Technologies, Grand Island, NY). The following HLA-B locus-specific primers were designed for amplification of HLA-B*0702 cDNA: 5′UT-B-1, 5′-TCTCCTCAGACGCCGAGATGC-3′ and 3′B-3 ut-1, 5′-GTCCCTCACAAGACAGCTGTCTCAGG-3′. The following HLA-Cw locus-specific primers were designed for amplification of HLA-Cw*0702 cDNA: 5′UT-C-2, 5′-TCTCCCCAGAGGCCGAGATGC-3′ and 3′C-ex8–1, 5′-GTCCCACACAGGCAGCTGTCTCAGG-3′. The final reaction mix contained 3 μl cDNA, 4 μ l PFU turbo DNA polymerase (Stratagene) 10 μl of 15 pmol/μl of the 5′ primer, and 10 μl of the 3′ primer, 0.8 μl of 10 μ M dNTPs, 4 μl 10× buffer. PCR analysis were conducted in a Perkin-Elmer Thermal Cycler 9600 (Branchburg, NJ) using the following parameters: 1 cycle of 94°C for 1 min, 30 cycles of 94°C for 45 s, 62°C for 45 s, 72°C for 4 min, and 1 cycle of 72°C. The HLA-B*0702 cDNA was then cloned in vector pCR2.1 (Invitrogen, Carlsbad, CA) using the Original TA Cloning Kit (Invitrogen), and the construct was confirmed to contain the HLA-B*0702 gene by sequencing. HLA-B*0702 was then inserted into the eukaryotic retroviral expression vector pCRV3 (kindly provided by Dr. R. F. Wang, Bethesda, MD), and the correct insertion was confirmed by sequencing after digestion of pCRV3-B*0702 with BglII plus HinDIII (Life Technologies). Transfection of relevant tumor cell lines (553-MEL and 1495-MEL) was performed by plating 2.5 × 10⁵cell/well in six-well plates in 3 ml of CM. After 24 h, the CM was aspirated and 1 ml of retroviral supernatant diluted 1:2 or 1:3 in CM was added to each well and incubated overnight. On the following day, the HLA-B*0702-transduced tumor cells were harvested and resuspended at 1 × 10⁶/ml for FACS analysis and cytokine release assay.

HLA-Cw*0702 cDNA was cloned in vector pT7blue (Novagen, Madison, WI) using the Original TA Cloning Kit (Invitrogen). The pT7blue was confirmed to contain the HLA-Cw*0702 gene by sequencing. The HLA-Cw*0702 gene was subcloned into the episomal vector pCI-NEO (Promega, Madison, WI) by digestion with XbaI/SmaI (Promega). The XbaI/SmaI/Cw*0702 insert, in pCI-NEO, was electrophoresed on a 1% gel, and gel was extracted using the QIAquick gel extraction kit (Qiagen). The XbaI/SmaI/Cw*0702 insert was ligated to pCI-NEO using T4 ready-to-go ligation mix from Pharmacia (Piscataway, NJ). Proper orientation of the insert into the two constructs was confirmed by sequencing. An unexpected ATG codon in the episomal vector pCI-NEO-Cw*0702 preceding the natural initiation codon of the Cw*0702 gene was corrected by site-directed mutagenesis (ATG→CTG). The vector was then used to transfect Cw*0702-negative tumor cell lines (537-MEL and 1495-MEL). Transfection was conducted as follows. First, 100 μl of lipofectamine (2 mg/ml stock; Life Technologies) mix consisting of 40 μg of lipofectamine per ml of DMEM supplemented with 5 ml/L glutamine was added to each well of a 96-well flat-bottom Costar plate. Next, 1 μl of 200 ng/μl plasmid DNA (pCI-NEO- Cw*0702) was added to each well, and the plate incubated for 20 min at room temperature. After incubation, 100 μl of 1 × 10⁶/ml cells were added to the well, mixed with lipofectamine solution, and incubated overnight. On the following day, the supernatant was removed, and effector cells were added at a concentration of 5 × 10⁴/200 μl CM and incubated for 24 h. The supernatant was tested for the release of IFN-γ.

Construction of F001-MEL cDNA library

A cDNA library was constructed from F001-MEL. Total RNA was isolated using RNeasy (Qiagen) with a total recovery of 2.5 mg RNA from 6 × 10⁷ tumor cells. A total of 16 μg mRNA were then isolated using the polyATtract mRNA isolation system (Promega). cDNA synthesis with directional random priming was performed using Orient Express cDNA System (Novagen). After EcoRI/HindIII linker ligation and enzyme digestion, the cDNA was size fractionated by column chromatography using the Life Technologies cDNA size fractionation column. Two groups of nine fractions each were selected on the basis of the fraction volume and the total volume of the eluate and werw pooled into fraction A (larger fragments) and fraction B (smaller fragments). Fraction A and fraction B cDNA fragments with EcoRI/HindIII linkers were subsequently cloned into to the cloning vector V-mut (Vical, San Diego, CA) digested with EcoRI and HindIII. Then, 100 ng of vector V-mut was ligated to 1 or 2 μl of insert (fraction A or fraction B) using T4 ligase and incubated overnight at 16°C. The ligation reactions were then diluted 1:5 and used to transform Escherichia coli cells. Next, 20 μl E. coli DH10B-competent cells (Invitrogen) were transferred to a 1-mm sterile cuvette (model 610; BTX Genetronix, San Diego, CA) on ice, and 1 μl of the diluted ligation reaction mixture was added to the cells. Immediately after addition of the ligation mixture, the cells were electroporated with 1.5 kV pulsing. After electroporation, E. coli cells were transferred to a Falcon tube (model # 2059) containing 1 ml of SOC medium (Biofluids, Rockville, MD) and incubated on a shaker at 37°C for 1 h. After incubation, 10, 100, and 200 μl of the transformation reactions were plated on Luria-Bertani Agar plates supplemented with 30 μg/ml kanamycin and incubated overnight. Growth of colonies was used to establish the titer of transformed E. coli and to assess the average size of cDNA inserts in fraction A and B. Accurate titers were obtained in plates inoculated with 10 μl of transformation reaction. The average size of cDNA fragments from 20 randomly chosen colonies demonstrated that fraction B cDNA inserts ranged in size from 0.5 to 1.5 kilobases. Because the size of the known MA is in this range, fraction B ligated to V-mut was chosen to screen F001-TIL reactivity. Subsequently, E. coli DH10B cells transformed with fraction B were diluted to 100 cells/1.2 ml of Super Broth (Advanced Biotechnologies, Columbia, MD) supplemented with 30 μg/ml kanamycin and inoculated into 96-deep well blocks (Edge Biosystems, Gaithersburg, MD) at 1.2 ml/well. Pools of 100 bacteria per well were incubated at 37°C shaking at 300 rpm for 48 h. After incubation, plasmid DNA was isolated from each pool using the Qiaprep 96 Turbo Miniprep kit (Qiagen). Plasmid DNA of each pool was eluted in 150 μl buffer EB. On average, 25–30 μg of plasmid DNA were isolated from each pool.

Screening of F001-MEL cDNA library by transfection of 293-HEK

Each cDNA pool was used to transfect 293-HEK cells (HLA-A*0301, B*0702, Cw*0702). To enhance the expression of Cw*0702, 293-HEK cells were incubated for 48 h with 1000 IU/ml IFN-α before transfection. Then, 100 μl of serum- and antibiotic-free DMEM (s/a-DMEM) supplemented with glutamine and containing 40 μg/ml lipofectamine (Life Technologies) were added to each well of a 96-well flat-bottom plate. Next, 1 μl of each cDNA library-pool (corresponding to 200–300 ng plasmid DNA) was then added. The DNA-lipofectamine mix was incubated for 20 min at room temperature. After incubation, 1 × 10⁵ 293-HEK cells were added to each well in 100 μl s/a-DMEM. Plates were then centrifuged at 500 rpm for 5 min and incubated for 24 h at 37°C. The following day, plates were centrifuged again at 500 rpm for 5 min. The supernatant was aspirated and discarded. A total of 4 × 10⁴ TIL in 200 μl of IM were added in each well, and the plates were incubated for another 24 h at 37°C. After incubation, plates were centrifuged at 1000 rpm for 5 min, and the supernatant were harvested for detection of IFN-γ.

Subcloning of cDNA pools and identification of target gene

A consistently reactive cDNA pool containing an estimated 100 gene transcripts was subcloned to identify the gene coding the MA recognized by F001-TIL. This pool was plated on Luria-Bertani Agar plate supplemented with 30 μg/ml kanamycin, and 400 single colonies were picked and tested by transfection of 293-HEK cells as described for the screening of the cDNA pools. Positive clones were subcloned at higher dilution to exclude possible contamination of plasmid DNA. Two subclones were identified that stimulated F001-TIL upon transfection of 293-HEK. These pools were restriction enzyme digested with HindIII and EcoRI and sequenced by automated fluorescent method using an ABI 277 DNA analyzer (Applied Biosystems, Foster City, CA). The DNA sequence obtained was then compared with reported sequences using the BlastSearch program of GenBank (National Library of Medicine, National Institutes of Health, Bethesda, MD).

Quantitative assessment of MAGE 1–12 mRNA expression in cell lines

Quantitative assessment of MA mRNA expression was evaluated with the ABI Prism 7700 Sequence Detection System (Perkin-Elmer, Norwalk, CT) using uniplex quantitative real-time PCR (qRT-PCR) (41). Each tube contained a Taqman probe that targeted a single gene of interest. Each probe consisted of an oligonucleotide with a 5′ reporter and a downstream 3′ quencher dye. Reporter/quencher dyes were analyzed by dual analysis based upon the different emission wavelength maxima. Uniplex qRT-PCR was performed using different reaction tubes for the target gene of interest and for the endogenous reference housekeeping transcript. A parallel tube contained the probe specific for the endogenous reference housekeeping gene β-actin. Probes were labeled with a reporter dye, 6-carboxyfluorescein (6-FAM, λ_max = 518 nm) quenched by 6-carboxytetramethylrhodamine (TAMRA). Cycling of cDNA involved denaturation at 95°C for 15 s, annealing/extension at 60°C for 1 min for a total of 40 cycles. Absolute measurement of mRNA copy number was performed with a standard curve for each gene of interest and for β-actin mRNA, dividing the test gene amount by the housekeeping gene amount. The final value represented the absolute number of mRNA copies per 10⁵ copies of β-actin mRNA. The sets of primers and labeled probes used for qRT-PCR are presented elsewhere (42) with the exception of MAGE-12, for which the following primers and probes were designed: MAGE-12, 6FAM-TGGTCCGCATCGGCCACTTGTAC; 5′-TGGCATCGAGGTGGTGG and 5′-CCCAGGCAGGTGACAAGG.

Routine and high-resolution HLA typing

Patients’ and cell lines’ HLA type was screened using genomic DNA amplification by sequence-specific primer PCR as previously described (43). To confirm the identity of relevant alleles, cDNA from cell lines was sequenced after locus-specific cDNA amplification. Evaluation of surface expression of HLA alleles was achieved by FACS analysis when mAb specific for a given allele were available. Surface expression of HLA-Cw alleles was demonstrated by complement-mediated cytotoxicity using prefabricated trays (One Lamba, Canoga Park, CA).

Development and characterization of a TIL (F001-TIL)/tumor (F001-MEL) pair from a melanoma metastasis

A patient with metastatic melanoma (F001) had undergone vaccination with gp100:209–217(210M) peptide in IFA in combination with systemic IL-12. A pretreatment FNA had demonstrated high expression of gp100/PMel17. Following transient reduction in size, the metastasis grew rapidly and failed to respond to further treatment including systemic high-dose hrIL-2. At the time of rapid growth of the metastasis, the tumor cells had totally lost expression of gp100/PMel17 according to immunocytochemical staining of material from a second FNA obtained at that point (38). In vitro expansion of material from the FNA in the presence of hrIL-2 yielded 20 separate bulk TIL cultures. None of the cultures recognized gp100:209–217 (target of the vaccine), perhaps explained by the loss of gp100/PMel17 by the tumor. However, 12 TIL cultures could recognize the autologous tumor F001-MEL. TCR Vβ-based molecular typing by PCR demonstrated that the 12 tumor-specific TIL cultures used predominantly Vβ7s2 (38). Other Vβ7s2 rearrangements were excluded by directed heteroduplex analysis (38). Subcloning of the Vβ7s2 amplification product and sequencing identified a unique Vβ7s2-TTY-Jβ2.1 rearrangement for all cultures. Thus, one of the cultures (culture 9), chosen as representative and used for identification of the epitopic determinant, will be referred to as F001-TIL. F001-TIL consisted of CD8⁺, CD45 R0^high/RA^low T cells. This culture was also CD44-, CD56-, CD49b-, and CD49d-positive. F001-MEL has been partially characterized previously (38). The neoplastic origin of this line was demonstrated by electron microscopy and karyotyping. F001-MEL did not express HLA class II but expressed high levels of HLA class I molecules. In particular, HLA-A*0201 and -A*0301, HLA-B*0702 and -B*0801 were highly expressed without pretreatment with IFN-γ (data not shown). Also, detectable levels of HLA-Cw*0701 and/or -Cw*0702 could be demonstrated by complement-mediated cytotoxicity without pretreatment with IFN-γ. Sequencing of F001-MEL cDNA confirmed expression and excluded mutations in all six HLA class I alleles. gp100/PMel17 expression was lost in F001-MEl, while other MDA such as MART-1/Melan A and tyrosinase were expressed at low levels both at the protein and mRNA level (data not shown). Quantitative RT-PCR confirmed no or low levels of expression of MDA (gp100/PMel17, MART-1/MelanA, tyrosinase, and TRP-1) and relatively high expression of TSA (MAGE-1, MAGE-3, MAGE-12, and NY-ESO-1) in comparison with other cell lines (Table I⇓).

Identification of HLA-Cw*0702 as the restriction element associated with the F001-TIL epitope

Functional characterization demonstrated that F001-TIL did not recognize gp100:209–217(210 M) and gp100:209–217 targeted by the vaccination or other HLA-A*0201-associated epitopes. However, F001-TIL could recognize the autologous tumor (F001-MEL) and other melanoma cell lines matched for HLA-B*0702 and -Cw*0702 but not -Cw*0701 (differing from HLA-Cw*0702 by two amino acids) (Fig. 1⇓). No cell lines could be identified that expressed singularly either HLA-B*0702 or -Cw*0702, presumably because these alleles are strongly linked in the ancestral haplotype HLA-B7, -Cw7 (44). Because F001-TIL could not recognize autologous or HLA-matched EBV lines, it was concluded that a new MA epitope had been identified in association with either HLA-B*0702 or -Cw*0702.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

IFN-γ release by F001-TIL cocultured with different melanoma cell lines plus the autologous EBV-B cell line (F001-EBV) and a breast cancer cell line (MCF7-BRE). The legend on the right panel details the HLA class I phenotype of the cell lines. Underlined alleles for HLA-B and -Cw loci point at matches with patient F001. All HLA-B*07** and HLA-Cw*07** typings were confirmed by sequencing of cDNA obtained from each cell line expressing such specificities.

To test whether the epitope recognized by F001-TIL was presented in association with HLA-B*0702, melanoma cell lines mismatched for this allele and expressing HLA alleles that did not contain the HLA-Bw6 public specificity were transduced with pCRV3-B*0702. FACS analysis was used to assure the presence of the HLA-B*0702 allele on tumor cells after transduction using SRF8-B6 mAb specific for HLA-Bw6. 553-MEL cells were used for transfection of this allele because this cell line HLA phenotype is -A2, 23; -B 44 (Bw4⁺/Bw6⁻), and -Cw3, 4. HLA-Bw6-expressing, pCRV3-B*0702-transduced 553-MEL cells were recognized by an archival HLA-B*0702-restricted CTL clone recognizing a HLA-B*0702-associated gp100/PMel17 epitope but were not recognized by F001-TIL (Fig. 2⇓). Surface expression of the HLA-Cw*0702 on transfected tumor cells was confirmed by complement-mediated cytotoxicity. Transfection of 537-MEL, a HLA-Cw*0702-negative, MAGE-12-expressing melanoma cell lines (537-MEL and 1495-MEL) with pCI-neo-Cw*0702, consistently restored recognition by F001-TIL (Table II⇓). Similar results were obtained using another melanoma cell line (1495-MEL, data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

IFN-γ release by F001-TIL and by the gp100-specific HLA-B*0702-restricted M5-CTL cocultured with 553-MEL cell lines transfected with pCRV3-EGFP or pCRV3-B*0702. Expression of pCRV3-EGFP was confirmed by fluorescence microscopy, while expression of HLA-B*0702 was confirmed by FACS analysis using SRF8-B6 mAb.

The MA recognized by F001-TIL is encoded by a one-residue variant of a polymorphic MAGE-12 gene

A cDNA library was constructed from F001-MEL cDNA and cloned into the vector V-mut. In the process of selecting target cells for transfection, it was coincidentally noted by sequence-based typing of 293-HEK cDNA that the HLA phenotype of these cells was naturally HLA-A*0301, B*0702, and Cw*0702. However, complement-mediated cytotoxicity suggested minimal surface expression of HLA-Cw*0702 compared with other cell lines. However, expression of HLA-Cw*0702 could be restored by culturing 293-HEK with IFN-γ or IFN-α for 48 h. Thus, IFN-α (not cross-reacting with IFN-γ in ELISA)-treated 293-HEK cells were used as target for transfection of F001-MEL cDNA library. Efficiency of transfection was based on the expression of green fluorescent protein (GFP) by 293-HEK cells transfected with V-mut-GFP (data not shown). Twelve hundred pools containing an estimated 100 genes each were generated. A consistently positive F001-TIL-reactive pool was in this fashion identified. Subcloning of this pool identified two pure subclones that induced IFN-γ release by F001-TIL upon transfection of 293-HEK cells. Restriction enzyme digestion with HinD-III and EcoR-I identified an ∼800-bp band. Minipreps from both subclones were sequenced in both directions, and in both cases an insert of 780 bp was identified homologous to the published sequence of MAGE-12 except for one amino acid (aa 187 D→A) (Fig. 3⇓). Sequencing of genomic DNA from F001-EBV-B with MAGE-12-specific primers (18) confirmed the aa 187 D→A difference from the previously published sequence, suggesting a novel MAGE-12 allele.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Nucleotide and deduced amino acid sequences of MAGE-12 derived from analysis of pool-20 cDNA from F001-MEL. This sequence is aligned with the MAGE-12 sequence (L18877) accessible through GenBank. Numbering of nucleotides and amino acids is arbitrarily presented from the first codon and amino acid. The epitope recognized by F001-TIL is underlined (aa 170–178). The variant residue and related nucleotide are also underlined in position 187. The variant was confirmed by sequencing of genomic DNA obtained from F001-EBV-B cells.

MAGE-12:170–178 is the epitope recognized by F001-TIL

Mapping of the F001-TIL epitope was performed by synthesis of 25 peptides from the MAGE-12 sequence based on the HLA-Cw*0702 binding motif (45). This library was screened by culturing F001-EBV-B cells with 10- and 1-μM concentrations of each peptide (Table III⇓). In this fashion, four candidate peptides could be identified that could stimulate IFN-γ release by F001-TIL at 1-μM concentration. Titration of these candidate epitopes identified one (MAGE-12:170–178; VRIGHLYIL) that could stimulate IFN-γ release by F001-TIL at 10- to 100-pM concentration (Fig. 4⇓ and 5). Alignment of this peptide with the peptide sequences of other MAGE proteins commonly expressed by tumor cells revealed limited homology (18). The closest sequences were those from MAGE-2 and MAGE-3 (two and three amino acid differences respectively). We then tested whether F001-TIL specificity was limited to MAGE-12:170–178 or included other MAGE proteins also expressed by F001-MEL at levels probably compatible with CTL recognition (46). Peptides deduced from aligned MAGE sequences homologous to MAGE-12:170–178 were synthesized; however, exogenous pulsing of F001-EBV with limiting dilutions of these peptides demonstrated that F001-TIL recognized exclusively MAGE-12:170–178 (Fig. 5⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

IFN-γ release by F001-TIL cocultured with F001-EBV-B cells exogenously pulsed with decreasing concentrations of peptide as indicated in the dose-response plot. The peptides from the MAGE-12 sequence are identifiable from the starting amino acid position (see Table III⇓). In the panel on the right, F001-TIL was exposed to control cell lines including autologous tumor (Au Tu) F001-MEL, 624.38-MEL (HLA-Cw*0702⁺), 888-MEL (HLA-Cw*0702⁻), autologous F001-EBV-B cells (Au EBV), and IFN-α-treated 293-HEK cells transduced or not with MAGE-12 cDNA.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

IFN-γ release by F001-TIL cocultured with F001-EBV-B cells exogenously pulsed with decreasing concentrations of peptide as indicated in the dose-response plot. Homologous peptide sequences to MAGE12:170–178 were tested for cross-reactivity with F001-TIL.

Expression of MAGE-12 by melanoma cell lines

The frequency of expression of MAGE-1 through 12 transcripts was tested in 17 melanoma cell lines (data not shown). Only one cell line (1123-MEL) did not express MAGE-12. This finding explains why all HLA-Cw*0702-expressing melanoma cell lines cocultured with F001-TIL induced IFN-γ release. This analysis confirms previous reports suggesting that MAGE-12 is among the most frequently expressed MAGE proteins in cell lines derived from melanoma metastases (18). Quantitative RT-PCR demonstrated that the level of MAGE-12 mRNA expression was comparably close among most cell lines (Table IV⇓), suggesting that the expansion of F001-TIL was not due to unusually high expression of this gene.