First Comparative Delineation of the T Cell ReceptorRepertoire in Primary and Multiple Subsequent/CoexistingMetastatic Melanoma Sites

Robert Strohal, Christine Brna, Ulrike Mossbacher,* Gottfried Fischer,† Hubert Pehamberger,* and Georg StinglDivision of Immunology, Allergy and Infectious Diseases and *Division of General Dermatology, Department of Dermatology, †Department of Blood GroupSerology, University of Vienna Medical School, Vienna, Austria

At present, very little is known about the types andheterogeneity of T cell responses and immunodominantepitopes of melanoma-associated antigens at coexistingsites of primary melanoma and metastatic lesions. Toaddress this issue, we compared the T cell receptor(TCR) gene usage, complemetary-determining region 3diversity, and melanoma-associated antigens expressionpatterns of primary and metastatic melanoma specimensfrom three patients with partially homologous HLAclass-1 types. Results obtained showed an overallpredominance of a very limited number of TCRVregions with AV13 and BV14 being most frequentlyoverexpressed. Sequencing of the dominating TCRtranscripts confirmed the restricted usage of certainTCR specificities and, in two of the three patients,

Several authors have shown that CD31 T cell receptor(TCR) α/β-bearing T cells that represent the vast majorityof tumor-infiltrating lymphocytes (TIL) within primary andmetastatic melanoma, can be propagated in vitro by growthfactors such as IL-2 and then exhibit specific HLA class-1-

restricted lytic activity for autologous tumor targets (Rosenberg et al,1986; Muul et al, 1987; Itoh et al, 1988). By using tumor-specificcytotoxic T lymphocytes (CTL) derived from either autologous peri-pheral blood lymphocytes (PBL) or TIL, a variety of melanoma-associated antigens (MAA), together with their HLA-binding motifs,could be identified (Van den Eynde and Brichard, 1995). These includetumor-specific antigens (Mage, Bage, Gage) and melanocyte-specificdifferentiation antigens (MART-1/Melan-A, tyrosinase, Pmel17/gp100, gp75). Sequencing of TCR moieties in CTL clones recognizinga given MAA in the context of the appropriate HLA class-I molecules,demonstrated that differences in the fine specificities of these CTLclones correlate with the selective usage of particular TCRVJ genes(Sensi et al, 1993, 1995; Van der Bruggen et al, 1994; Boel et al, 1995;Maeurer et al, 1995; Zarour et al, 1996). Semiquantitative polymerasechain reaction (PCR) analyses of TCR transcripts in TIL of human

Manuscript received June 17, 1998; revised August 26, 1998; accepted forpublication September 18, 1998.

Reprint requests to: Dr. Robert Strohal, Department of Dermatology,Division of Immunology, Allergy & Infectious Diseases, University of ViennaMedical School, Wahringer Gurtel 18-20, A-1090 Vienna, Austria.

Abbreviations: CDR3, complemetary-determining region; CTL, cytotoxicT-lymphocytes; MAA, melanoma-associated antigens; NM, nodular melanoma;TIL, tumor-infiltrating lymphocytes.

0022-202X/98/$10.50 · Copyright © 1998 by The Society for Investigative Dermatology, Inc.

1085

identified several identical TCR clonotypes at morethan one metastatic site. Nevertheless, we failed todetect TCR transcripts that were common to all tumordeposits in a given patient and, within the majority ofcoexisting metastases, tumor-infiltrating lymphocytespreferentially used individual site-specifically expandedTCR β-chain VJ segment combinations. This occurrenceof individual responses simultaneously executed at andinfluenced in their specificity by the different sites oftumor growth, has important implications for the typeof strategies chosen in the development of efficaciousvaccines for patients with metastatic melanoma. Keywords: melanoma-antigens/melanoma immunology/sequence-homology-amino-acid/T lymphocytes. J Invest Dermatol111:1085–1091, 1998

melanoma specimens, showed that their repertoire is often skewedtowards 1–3 predominantly expressed TCRAV or TCRBV regionsindicative of clonal/oligoclonal expansion and/or recruitment of tumor-reactive T cell populations at lesional sites (Sensi et al, 1995; Strohalet al, 1994a). Because the vast majority of published data were generatedfrom solitary tumor lesions only, the question still remains whetherthe same TCR transcripts dominating the TCR repertoire in singlemelanoma specimens (primary melanoma, metastases) are also preferen-tially used within subsequent and/or coexisting melanoma metastasesat different organ sites. In order to address this issue, we comparativelydelineated TCRAV/BV gene usage, complemetary-determining region3 (CDR3) diversity, and MAA expression patterns (MAGE-1, MAGE-2, MAGE-3, MAGE-4, MAGE-6, BAGE, GAGE, MART-1/Melan-A, gp100) of one primary melanoma and multiple coexisting metastasesfrom three different HLA-typed patients. By doing so, we attemptedto gather information not only about the diversity and nature of theanti-melanoma immune response but also about the type, structure,and heterogeneity of the immunodominant epitopes in situ.

MATERIALS AND METHODS

Study population, tumor specimens and normal controls We obtained12 different autopsy specimens of tumor tissue from three female patients (1622,1464, 1214) with metastatic melanoma. These included (i) the primary melanomaof patient 1622 and coexisting lymph node, liver, lung, brain, and skinmetastases; (ii) liver and lung metastases of patient 1464; and (iii) lung, spleen,kidney, and thyroid gland metastases of patient 1214. In addition to surgery ofthe primary tumor, two patients (1464, 1214) had received one or morecombination therapies with dacarbazine and fotemustine. Patient 1622 under-went palliative surgery only. A representative portion of each specimen wasprocessed for histopathology that classified the primary tumor of patient 1622

1086 STROHAL ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

as nodular melanoma (NM) and confirmed the malignant melanocytic natureof the remaining metastatic lesions. The rest of each specimen was snap-frozenin liquid nitrogen and stored at –70°C until further use.

For control purposes, we isolated mononuclear cells from the heparinizedblood (PBMC) of three healthy female donors by Ficoll-Hypaque (Pharmacia,Uppsala, Sweden) density gradient centrifugation. Cells recovered from theinterface were resuspended in guanidinium isothiocyanate at a density of 5 3 106

cells per ml and stored at –70°C.

HLA typing HLA class-I typing of each patient was performed from normaland affected tissues (patient 1622, liver and skin metastases; patient 1214, normalliver; patient 1464, liver metastasis) using a cDNA-based multistep PCR andsequencing protocol. The procedure, including all PCR and sequencing primers,has been described previously (Fae et al, 1996).

Purification of RNA and cDNA synthesis Total RNA was extracted fromthe frozen cell and tissue samples by the guanidinium isothiocyanate-cesiumchloride procedure (Glisin et al, 1974). First-strand cDNA was synthesized from5 µg RNA in a final volume of 40 µl containing 8 µl 5 3 reverse MoMLVreverse transcriptase reaction buffer (GIBCO/BRL, Gaithersburg, MD), 10 µldNTP at 10 mM each, 2 µl of a 200 µM solution of oligo dT(15), 1 µl RNAsinat 40 U per µl (Promega, Madison, WI), and 5 µl of MoMLV reversetranscriptase at 200 U per µl (GIBCO/BRL). After 10 min at room temperature,the mixture was incubated at 42°C for 1 h, heated to 95°C for 5 min, chilledon ice, and stored at –20°C until further use.

Amplification of cDNA by PCR and quality control To determine theMAA (MAGE-1, MAGE-2, MAGE-3, MAGE-4, MAGE-6, BAGE, GAGE,MART-1/Melan-A, gp100) and TCRAV and TCRBV expression patterns,cDNA from the various samples were subjected to PCR amplification.

For the MAA-specific PCR, one of 25 of the cDNA product was incubatedwith 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl2, all four dNTP(each at 200 µM), 2.5 U AmpliTaq DNA polymerase (Perkin-Elmer Cetus,Emeryville, CA), and 0.4 µM of each sense and anti-sense primer in a finalvolume of 50 µl. Oligonucleotide primers were either devised from knowncDNA sequences (Adema et al, 1994; Kawakami et al, 1994) (gp100, 59-TTACTGACCAGGTGCCTTTCT-39 sense, 59-TGTAGCCTCTGAGTT-GACAT-39 anti-sense; MART-1/Melan-A, 59-CACTTCATCTATGGTTAC-CCC-39 sense, 59-TGAATAAGGTGGTGGTGACTG-39 anti-sense) or takenfrom the literature (Brasseur et al, 1992; De Plaen et al, 1994; De Smet et al,1994; Boel et al, 1995; Van den Eynde et al, 1995; MAGE-1, MAGE-2, MAGE-3, MAGE-4, MAGE-6, BAGE, GAGE – VDE18/VDE24), commerciallysynthesized and high performance liquid chromatography-purified (Pharmacia).Amplification was performed on a DNA thermal cycler (model 480, Perkin-Elmer). Before the first cycle, the reaction mixture was heated at 94°C for4 min and after the last cycle the incubation was extended for another 15 minat 72°C. The PCR cycle profile was as follows: denaturation, 1 min at 94°C;annealing of primers, 2 min at 56°C (GAGE), 60°C (gp100, MART-1/Melan-A), 62°C (BAGE), 67°C (MAGE-2), 68°C (MAGE-4, MAGE-6), 72°C(MAGE-3), or 1 min at 72°C (MAGE-1); primer extension, 2 min at72°C (gp100, MART-1, MAGE-1, MAGE-2, MAGE-3, MAGE-4, MAGE-6, GAGE) or 73°C (BAGE). After 30 cycles, 10 µl of the reaction mixture wasfractionated on a 1.2% agarose gel and stained with ethidium bromide(4 µg per ml).

To study the TCRV repertoire, a semiquantitative PCR was performed usingsense primers complementary to each of the 22 TCRAV and 24 BV familysequences together with one TCRAC- and TCRBC-specific anti-sense primer,respectively. The TCRBC primer was constructed to detect and amplify bothTCR β-chain constant regions. All primer sets were purchased from ClontechLaboratories (TCR Typing Amplimer Kit, Palo Alto, CA) and the PCRprocedure was carried out according to protocols provided by the manufacturer.Briefly, a final 50 µl reaction mixture containing 2 µl cDNA, 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl2, 200 µM dNTP, 2.5 U AmpliTaqDNA polymerase, and 0.5 µM of the respective TCR primers was subjectedto 30 amplification cycles, each consisting of 1 min at 95°C, 1 min at 55°C,and 1 min at 72°C followed by a final 15 min extension at 72°C.

To control for the occurrence of amplification artefacts by contaminatingDNA, cDNA from EBV-transformed B cells or samples without cDNA wereused as negative controls and checked by hybridization with the appropriateTCR or MAA probes. These experiments yielded negative results. Moreover,a 20-cycle PCR assay with primers specific for β-actin (β-Actin Control RT-PCR Amplimer Set, Clontech Laboratories) was initially carried out to ensurethat each specimen contained intact RNA.

Southern blot analysis and quantitation of TCRAV and TCRBV geneexpression A 10 µl aliquot of each PCR product was separated by electro-phoresis on 1.2% agarose gels at 30 V overnight and blotted onto nylon

membranes (Nytran-NY13N, Schleicher und Schuell, Dassel, Germany). Blot-ting, radioactive 39 end-labeling of the oligonucleotide probes with (α32P)dATP (Amersham, Bucks, U.K.) and terminal deoxynucleotidyl transferase(Boehringer, Vienna, Austria), prehybridization, and hybridization were carriedout according to standard protocols (Southern, 1975; Elbe et al, 1992).Oligonucleotides used for hybridization either matched with the central partof the respective MAA genes (gp100, 59-CTGACCAGGTGTAGTACCCACA-ACTTCTGT-39; MART-1/Melan-A, 59-CCGATGATCAAACCCTTCTT-GTGGGCATCT-39; MAGE-1, 59-AGTGCAGACTCCTCTGCTCAAGA-GACATGA-39; MAGE-2, MAGE-3, and MAGE-6, 59-AGTGCTGACT-CCTCTGCTCAAGAGGCATGA-39; MAGE-4, 59-AGTGCTGACTCTT-CTGCTCAGAAGACATGA-39; BAGE, 59-CCACAACCTCAGAAGATG-AAGCACAGAGCT-39; GAGE, 59-TTGAGTTGCTGGTTCCCCTTCTT-CAGGTGT-39) or corresponded to 59 TCR α/β constant region sequences(Cα, 59-Gtacacggcagggtcagggttctggatat-39; Cβ, 59-CTTTtgggtgtgggagatctctg-CTTCTGA-39). Blots were air-dried and exposed to Kodak X-Omat/AR X-ray films (Kodak AG., Vienna, Austria) for 4–8 h at –70°C in the presence ofan intensifying screen. In order to detect very weak signals, certain blots withamplified MAA cDNA were re-exposed for up to 96 h.

For quantitation of TCRAV/BV expression, the respective autoradiographswere scanned on a Bio IMAGE System (BIO IMAGE, Ann Arbor, MI)computing densitometer. To generate area volumes from linear autoradiographicscans, individual bands were digitized and integrated using the VISAGE 4.6 Msoftware package provided by the manufacturer. Because the relative frequencyof each TCRAV/BV gene segment was expressed as a percentage of the sumof all TCRAV/BV signals detected, the individual volume of the particularTCRAV/BV band was divided by the total volume of all integrated bands andthe results obtained were normalized to 100.

According to the literature (Weidmann et al, 1993; Thor Straten et al, 1994;Sensi and Parmiani, 1995), overexpression of a certain TCRAV/BV gene familywas defined as TCRV ratios ù 1, whereby the individual TCRAV/BV ratiowas calculated as follows: relative percentage of TCRV expression in melanomaspecimens/mean 1 2 SD of relative percentage of TCRV expression in PBLof three healthy donors.

Cloning of PCR-amplified TCR β-chain transcripts and sequencingreactions We then established a PCR protocol that enabled us to amplifythe dominating BV14 and BV13 gene segment transcripts within one series ofPCR experiments. For this purpose, we lowered the annealing temperatureand designed a new BV primer (59-ACATGTCCTGGTATCGACAAG-39),which represents a BV13S1-specific nucleotide sequence differing from BV14S1and all other members of the amplified BV13 subfamilies in only one residue(59end, position 2 or 7, respectively). PCR products were ligated into theT/A vector PCR II (TA cloning kit, Invitrogen, San Diego, CA) and used totransform INVαF9 Escherichia coli strains. White colonies were screened forpositive ligation using Eco-RI digests and electrophoresis of the purifiedDNA plasmids.

Following the instructions provided by the company, plasmid DNA-con-taining inserts of appropriate length were sequenced by a modified dideoxychain termination procedure using the PRISM Ready Reaction DyeDeoxyTerminator Cycle Sequencing Kit (Perkin-Elmer Cetus) and run on a DNAsequencer (model 373, Applied Biosystems, Weiterstadt, Germany). All TCRsequences were compared with GenBank entries (EMBL, Heidelberg, Germany)using the on-line software provided by the European Bioinformatics Institute(EBI, Cambridge, U.K. at http://www.ebi.ac.uk) and classified according tofamily designations defined by Arden et al (1995). We have adopted the newTCR nomenclature proposed by the International Union of ImmunologicalSocieties (WHO-IUIS TCR nomenclature, 1995).

RESULTS

HLA typing of patients 1622, 1464, and 1214 RNA was isolatedfrom normal and affected tissues (Table I) of the three patients andsubjected to reverse transcription using HLA-A, -B, and -C locus-specific primers. The cDNA preparations obtained were separatelyamplified by PCR and the HLA genotypes determined by nucleotidesequencing of the PCR products. In patient 1464 the typing did notallow discrimination between the HLA-A*1101 and the *1102 allele,and between the HLA-B*35 subtypes B*3501, *03, *04, and *06. Allother alleles were typed at high resolution. Whereas in patient 1622all loci appeared heterozygous, only one allele could be detected atthe HLA-B and -C loci of patient 1464 and at the HLA-A locus ofpatient 1214. This could be due either to homozygosity or tononexpression of one allele.

Although each of the patients has a unique pattern of HLA class Ialleles, they share certain similarities: all three have alleles encoding

VOL. 111, NO. 6 DECEMBER 1998 TCR REPERTOIRE IN METASTATIC MELANOMA 1087

Table I. HLA class I genotypes and melanoma specimens of patients analyzed

Patients HLA-A HLA-B HLA-C Melanoma specimens

1622 *2402, *2605 *3503, *27052 *0401, *0202 NM,a LN,b liver, skin, lung, brain1464 *0301, *11 *35 *0401 lung, liver1214 *2402 3501, *5101 *0401, *02022 lung, spleen, kidney, thyroid gland

aNodular malignant melanoma.bLymph node metastasis.

Table II. Comparative analysis of MART-1/Melan-A, gp100, Mage-1, Mage-2, Mage-3, Mage-4, Mage-6, Bage, and GagemRNA expression in 12 melanoma specimens from three different patients

MART-1Patients/specimen Site /Melan-A gp100 Mage-1 Mage-2 Mage-3 Mage-4 Mage-6 Bage Gage

1622primary melanoma sacral region 1 1 – – – – – – –lymph node metastasis right groin 1 1 – – – – – – –organ metastases liver 1 1 – – – – – – –

skin 1 1 – – – – – – –lung 1 1 – – – – – – –brain 1 1 – – – – – – –

1214organ metastases lung 1 1 1 1 1 – 1 – 1

spleen 1 1 1 1 1 – 1 – 1kidney 1 1 1 1 1 – 1 – 1thyroid gland 1 1 1 1 1 – 1 – 1

1464organ metastases lung 1 1 – 1 1 – 1 – –

liver 1 1 – 1 1 – 1 – –

Table III. Comparison of BVJC and CDR3 region amino acid sequences from this study with reported TCR β-chain transcriptseither detected in melanoma-associated TIL, or expressed by CTL with defined MAA specificitya

Tumor site of No. ofComparison with Patient occurrence occurrences TCRB Vb CDR3 Jb

gp100-specific CTL clone – – – V14S1D2 J2S7C2 C A S S L G S S Y.. E Q Y F G P(Zarour et al, 1996) 1622 skin 1 V14S1D2 J2S7C2 – – – – – S P G –.. – – – – – –

1622 lung 1 V14S1D2 J2S7C2 – – – – Y S G G –.. – – – – – –1622 skin 1 V14S1D2 J2S7C2 – – – – A R D T G V W – – – – – –1464 lung 1 V14S1D2 J2S7C2 – – – – G R D T G G W – – – – – –1464 lung 18 V14S1D2 J2S7C2 –– – – A R D T G A L – – – – – –1214 lung 1 V14S1D2 J2S7C2 –– – – P L – R G P Y – – – – – –

Mart-1 specific CTL clone 2A13 – – – V14S1D2 J2S1C2 C A S S. R T V G G P N E Q F F G P(Sensi et al, 1995) 1214 lung 10 V14S1D1 J2S1C2 –– – – S – – S – – S R – – – – – –

TIL-1/regressing melanoma – – – V13S1D1 J1S1C1 C A S S W G G D T E A F F G Q(Mackensen et al, 1993) 1214 lung & kidney 1 V13S1D1 J1S1C1 –– – –. – V N – – – – –– –

TIL-6/regressing melanoma – – – V13S1D1 J1S1C1 C A S S E R R N T E A F F G Q(Mackensen et al, 1993) 1214 lung & kidney 1 V13S1D1 J1S1C1 –– – – G V. – – – – – – – –

aDashes within the sequence indicate residue identity and dots represent estimated gaps introduced for proper alignment

the HLA-B35 and the HLA-Cw4 specificities. Although the HLA-Cw4 specificities are identical, HLA-B35 alleles differed in the subtype.

Assessment and comparative analysis of the lesions’ MAAprofile RNA was purified from all lesional tissues, PCR amplifiedand subjected to Southern blot hybridization using the appropriateMAA primers and probes. Results obtained showed that all specimenstranscribed MART-1/Melan-A and gp100 (Tables II and III), whereasBAGE and MAGE-4 were consistently absent (Table II). MAGE-1,MAGE-2, MAGE-3, MAGE-6, and GAGE mRNA were unevenlydistributed. GAGE and MAGE-1 were expressed in metastatic lesionsof patient 1214 only; MAGE-2, MAGE-3, and MAGE-6 mRNAcould be amplified from tumor specimens of patients 1214 and 1464but not of patient 1622. In fact, MART-1/Melan-A and gp100 werethe only MAA transcripts detected in patient 1622’s primary melanomaand in her metastatic lesions. Given the limitations of quantitating therelative amounts of MAA mRNA expression by means of conventionalPCR, cDNA amplification with MART-1/Melan-A and gp100-specific primers revealed quite homogeneous results in all specimenstested, whereas, in the case of MAGE, GAGE, and BAGE, some

differences in the specimen-specific amount of transcribed mRNAwere seen between the different samples.

Assessment and comparative quantitation of the lesions’ TCRVregion profile Using a semiquantitative PCR/hybridization proced-ure with validated TCRAV/BV-specific primers, we comparativelydelineated the relative frequency of TCRV gene expression withineach of the 12 tumor lesions and correlated the relative amounts ofTCRV family transcripts with the range of TCRBV gene usage innonactivated PBL from three healthy donors. According to TCRstudies by Weidmann et al (1993) and Thor-Straten et al (1994), TCRVfamilies were considered to be overexpressed when their individuallesional value exceeded or at least equalled the mean value 1 2 SD ofthe same TCRV transcripts in healthy controls. As shown in Tables IVand V, TIL of our patients exhibited a substantial, albeit limited,heterogeneity of TCRV specificities [number of overexpressedTCRAV/BV families: 10 of 15 (1622); four of six (1464); five of 10(1214)]. Certain TCRV regions (e.g., AV8, AV17, BV2, BV20)predominated within only one, others (AV13, AV18, BV13, BV14)within the majority of metastases. When TCRV family transcripts were

1088 STROHAL ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table IV. Relative amounts of TCRAV gene family transcripts in allogeneic PBL and multiple melanoma lesions of threedifferent patients

Patient 1622 Patient 1464 Patient 1214

PBL a

TCRAV cut-off (SD) NM lymph node liver skin lung brain liver lung lung spleen kidney thyroid gland

1 7 (0.78) 6.1 12.5 b 14.1 0.3 5.8 0 4.8 3.1 3.3 5.1 4 1.72 15.5 (3.2) 12 21.1 16.9 9.3 25.1 32.2 11.9 8.8 15.4 11.8 12.7 93 9.9 (1.8) 8.8 12.2 11.5 12 3.4 12.2 9.4 7.2 9.3 7.4 7.5 10.74 7.7 (2.4) 1.3 0.7 0 0 0 0 0 1.3 0 0.7 0.5 05 10.3 (2.3) 7.2 3 1.8 0 1 0 3.2 2.4 3 3.9 4.4 0.96 15.8 (3.9) 6.6 3.3 0 0.6 1 9.3 8.2 8.3 4.6 3.6 6.1 0.27 6.4 (2.4) 3.9 1.5 2 2.9 1 0 1.1 1.9 1.6 3.2 1.5 0.88 16.2 (4.3) 10.2 10 11.3 0 3.6 10.3 8.3 7.7 15.4 10.4 7.5 18.29 7.5 (1.5) 2.4 2.1 0 0 0 2.6 2.4 0.9 0 1.1 0.5 0.1

10 9.2 (3.6) 0 0 0 1.2 0 0 0 0 0 0.8 0.2 011 10.4 (4.0) 0 0 0 0 0 0 0 0 0 0.2 0.2 012 7 (2.6) 4 4.4 0 1.6 0 8.7 2.4 5.9 2 0.8 0.6 0.913 2.2 (0.7) 6 0.9 6.2 0 8 0 9.4 12.7 2.2 3.5 4.1 0.414 16.6 (5.1) 8.8 5.9 5.1 12.3 1 0 4 8.8 10.4 10.3 6 11.715 10.8 (3.2) 6.2 0.7 4.9 23.1 2.1 4.7 8.1 5.9 10.7 8 12.1 17.816 1.8 (0.6) 0 0.5 0 0.8 0 0 2.4 0.7 0 0.5 0 017 6.5 (1.1) 0 3.6 9.4 3 0.8 0.6 1.8 0.5 3 4.1 2.8 2.618 8 (2.2) 2.2 1.4 9.1 1 17.3 5.1 8.1 9.5 5.9 6 9.5 12.119 7.7 (2.0) 1.6 1.2 2.3 9.9 1.3 2.1 3.8 6.5 1.1 3 4.6 0.620 3.7 (0.1) 1.7 1.7 0 0.6 5.6 8.8 3.9 2.5 2.3 2.4 3.3 021 13.3 (3.1) 3.6 4.9 0 9.3 4.4 2.9 2.8 3.6 4.8 4.6 4.7 5.922 19.8 (5.2) 6.9 8 4.2 9.8 17.9 0 3.8 1.8 5.1 8.6 7 6.5

aAllogeneic PBL were generated from 3 healthy donors.bUnderlined values represent TCRAV ratios (TCRAV melanoma/mean 1 2SD TCRAV PBL) ù 1. According to the literature (Sensi and Parmiani. 1995; Weidmann et al. 1993;

Thor Straten et al. 1994). These TCRAV families are considered to be overexpressed within melanoma specimens.

Table V. Relative amounts of TCRBV gene family transcripts in allogeneic PBL and multiple melanoma lesions of threedifferent patients

Patient 1622 Patient 1464 Patient 1214

PBL a

TCRBV cut-off (SD) NM lymph node liver skin lung brain liver lung lung spleen kidney thyroid gland

1 13.5 (3.8) 3.5 5.9 0 0 1.9 12.5 7.9 15.8 b 2.7 6.6 1.8 1.02 13.6 (2.2) 8.8 4.0 15.6 2.3 9.4 4.6 3.5 3.2 1.0 5.6 10.1 4.83 21.1 (6.3) 4.8 16.9 9.9 21.0 27.6 2.2 10.4 0.2 0.8 5.1 5.6 25.64 14.0 (2.6) 5.1 9.0 2.9 3.6 17.7 0 1.8 1.0 2.6 12.5 9.1 3.45 10.4 (0.3) 4.5 0 0 0 0.1 13.9 0.4 2.5 0.9 5.6 5.3 4.26 21.1 (2.7) 8.7 25.7 3.7 6.4 1.4 19.4 18.0 19.1 6.5 8.0 17.6 5.17 14.3 (4.1) 14.0 25.9 4.6 23.3 15.1 18.3 23.69 21.1 7.8 12.8 10.5 88 9.2 (2.2) 8.0 0.6 4.0 0.9 0.4 0 16.1 18.1 1.8 4.2 1.5 09 6.6 (2.3) 0.3 1.7 3.8 0.2 0.2 0 0 0 4.8 2.2 0.4 0

10 1.2 (0.5) 0 0 0.1 0.3 0 0 1.7 1.5 0 0.9 0 011 0.2 (0.06) 0 0 0 0 0 0 0 0 0 0.7 0.7 2.212 3.1 (1.1) 0.9 0 5.5 0.2 0.2 1.8 7.3 3.5 2.0 0 0 013 17.8 (3.3) 27.7 2.4 6.4 28.6 19.4 6.9 5.7 4.1 28.8 20.2 13.0 31.714 3.8 (0.2) 8.2 3.8 4.2 6.0 5.2 0 3.3 9.5 14.6 0.8 6.8 7.415 2.3 (0.5) 0 0 1.2 0 0 0 0 0.2 4.2 0.6 0.7 016 3.7 (0.8) 0 0 10.0 0 0 0 0 0 0 1.8 0.4 017 5.3 (1.3) 0.6 0 0.2 0 0 5.7 0 0 2.5 2.2 2.4 0.318 5.5 (1.6) 0.1 0 0.5 0 0 13.1 0 0 7.4 4.0 0.5 019 3.6 (1.0) 4.2 4.0 26.8 2.1 1.1 0 0 0 3.3 6.5 3.5 0.420 3.8 (1.0) 0.3 0 0.7 5.1 0.3 0 0 0 1.9 1.0 0 021 0.6 (0.2) 0 0 0 0 0 0 0 0 0 0 0 0.122 1.0 (0.3) 0 0 0 0 0 1.5 0 0 2.1 0.8 0 0.123 0.3 (0.1) 0 0 0 0 0 0 0 0.2 2.2 8.0 2.2 024 2.8 (0.8) 0.1 0 0 0 0 0 0 0 1.9 2.2 7.9 5.6

aAllogeneic PBL were generated from three healthy donors.bUnderlined values represent TCRBV ratios (TCRBV melanoma/mean 12 SD TCRBV PBL) ù 1. According to the literature (Weidmann et al. 1993; Thor Straten et al. 1994; Sensi

and Parmiani. 1995), these TCRBV families are considered to be overexpressed within melanoma specimens.

grouped according to their percentage of overexpression within allspecimens analyzed, the overall predominance of only a few TCR α/β specificities became apparent. As seen in Table VI, two differentTCRAV and three TCRBV gene segment families were particularlyabundant with TCRAV13 and BV14 being the specificities most oftenidentified. In this respect, it is interesting to note that these two

specificities are also part of the indigenous skin-associated TCRrepertoire (Dunn et al, 1993; Strohal et al, 1994a) and often predominatecutaneous melanoma lesions (Strohal et al, 1994a).

Cloning and sequencing of the TCRBV transcripts dominatingthe TCR repertoire in all melanoma specimens We determined

VOL. 111, NO. 6 DECEMBER 1998 TCR REPERTOIRE IN METASTATIC MELANOMA 1089

Table VI. Percentages of overexpressed TCRV gene segments in TIL of three melanoma patients

Overexpressed Patient 1622 Patient 1464 Patient 1214 totalTCRV (n 5 6)a (n 5 2) (n 5 4) (n 5 12)transcripts % of occurrencesb % of occurrences % of occurrences % (no.) of occurrences

TCRAV2 66 – – ,503 66 – ,50 ,50

13c 50 100 75 66 (8)18 ,50 100 50 50

TCRBV7 66 100 – 508 – 100 – ,50

10 – 100 – ,5012 ,50 100 – ,5013 50 – 75 5014 83 50 75 75 (9)23 – – 75 ,50

aNumber of tumor specimens analyzed.bOnly those TCRV families that are overexpressed in more than 50% of melanoma tissues from at least one patient. are listed.cBolded TCRV families represent the overall predominating TCR moieties.

Table VII. List of TCR β-chain sequences with identical CDR3 regions occurring more than twice in melanoma lesions ofpatient 1622. 1464 and 1214

TCRB No. of occurrences/Total no. of sequences per site (%)

V JC CDR3 Primary melanoma lymph node liver skin lung brain spleen kidney thyroid gland

162213S1 2S1C1 TGGQGAWNEQF 7/7 (100) – – – – –

2S7C2 HRTSARKEQY – – – 1/6 (16) – 6/6 (100)13S2 1S5C1 NGGDVRPQH – – 6/12 (50) – – –

2S1C2 SPPTSGENEQF – – 5/12 (42) – 2/5 (40) –1464

14S1 1S2C1 SLSGASYGYT 6/21 (29) –2S7C SARDTGAWEQY – 18/21 (86)

12S2 2S3C2 KAGGSSTDTQY 5/21 (24) –3S1 1S2C1 SAGLNAGSVGYT 6/21 (29) –

121414S1 1S2C1 SQSGTLGYT 5/27 (19) – – –

2S1C2 SSRTSGGSRNEQF 10/27 (37) – – –2S3C2 LAGTDTQY 3/27 (11) – – –

13S1 1S6C1 ASLDRGNSPLH – – 1/4 (25) 2/15 (13)2S1C2 SPRGGYNEQF 1/27 (4) 1/6 (16) – 3/15 (20)

15S1 2S5C2 TSDRGAETQY 1/27 (4) – – 4/15 (27)

nucleic acid sequence diversity of junctional and CDR3 (definedaccording to Chothia et al, 1988) from overexpressed TCR transcriptsto confirm semiquantitative PCR results and identify clonally expandedTCR specificities. For this purpose, we chose a protocol that enabledus to predominately amplify the overexpressed TCRBV14 and BV13gene segment transcripts within one series of PCR experiments.Probing all the 12 tumor specimens, we found that 131 (95%) of theTCR transcripts analyzed (n 5 138) showed fully coding in-framerearrangements suggesting the functional expression of these transcripts.One hundred and nine (83%) of these 131 TCR β-chain transcriptscontained the dominating BV13 and BV14 gene segment families. Theremaining 22 TCR sequences showed BV regions of subgroup II(Schiffer et al, 1986). One hundred and four of 109 BV14/BV13-containing transcripts were found to belong to only three differentsubfamilies (BV14S1, BV13S1, BV13S2), from which 91% wereassociated with the TCRBJ regions 1S2, 2S1, and 2S7.

The complete sequence analysis of 131 in frame TCR β-chaintranscripts showed that 14 cDNA clones occurred more than twice inthe 12 tumor lesions of the three patients analyzed (Table VII). Thenumber of occurrences per patient and site varied considerably. Withinpatient 1622 (Table VII) four distinct cDNA clones accounted for100% (seven of seven; BV13S1 J2S1), 100% (six of six; BV13S1 J2S7),50% (six of 12; BV13S2 J1S5), and 40% (two of five; BV13S2 J2S1)of the functionally rearranged TCR β-chain sequences within the

primary melanoma, brain, liver, and lung metastases, respectively.Lymph node and skin metastases did not show clonally expanded TCRβ-chain transcripts. To our surprise, the clonally expanded BV13S1 J2S1cDNA transcript within patient 1622’s primary tumor could not befound in any of her five metastases, although two other TCR cDNAclones were simultaneously expressed in liver and brain metastases,respectively. Clone 13S2 J2S1, which had been amplified from thelung metastasis, also comprised 42% (five of 12) of the cloned andsequenced TCR β-chains from the liver, and the brain-associated cloneBV13S1 J2S7 was expressed once within the skin metastasis ofthis patient.

The analysis of patient 1464’s lung and liver metastases yieldedsimilar results (Table VII). Clone BV14S1 J2S7 accounted for 86%(18 of 21) of the in-frame β-chain transcripts within the lung specimen,but was not detected in the liver. In contrast, the liver metastasis ofthe same patient harboured three different cDNA clones at frequenciesof 29% (six of 21; BV14S1 J1S2 and BV3S1 J1S2) and 24% (five of21; BV12S2 J2S3).

In patient 1214, we again did not find a cDNA clone ubiquitouslyexpressed in all four metastases. Yet, clones BV13S1 J1S6 andBV15S1 J2S5 co-occurred at two different sites (kidney/thyroid glandand lung/thyroid gland) and clone BV13S1 J2S1 was simultaneouslyisolated from three tumor deposits (lung/spleen/thyroid gland). Theirindividual frequency of expression within each metastasis ranged from

1090 STROHAL ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

4% (BV13S1 J2S1, lung, one of 27) up to 27% (BV15S1 J2S5, thyroidgland, four of 15). As seen in patient 1464, a lung-specific β-chaintranscript (BV14S1 J2S1) represented the most abundantly expressedcDNA clone (10 of 27, 37%) of patient 1214’s metastases.

Taken together, the TCR β-chain repertoire of the patients’ tumorspecimens consists of certain TCR clonotypes simultaneously occurringat multiple sites, as well as of locally expanded TCR transcriptsexhibiting site-specific differences in the composition of their BVJ rep-ertoire.

Comparison of TCR β-chain transcripts from the literaturewith TCR sequences generated in this study We then performedamino acid sequence similarity searches by comparing all TCR β-chains from the EMBL database with those TCR transcripts generatedin this study. From a total of 131 in frame TCR β-chain sequences,eight different cDNA clones showed BVJC gene segment combinationsidentical to TCR β-chain transcripts found on either in vivo clonallyexpanded melanoma-specific TIL (Mackensen et al, 1993), or on CTLwith defined MAA specificities (Table III). To our surprise, thesecorresponding TCR transcripts also displayed CDR3 region similaritiesof up to 90%. One of these eight clonotypes occurred simultaneouslyin lung and kidney metastases. The other seven were found in eitherskin or lung lesions. Table III shows that two of eight TCR specificitieswere clonally expanded, dominating the TCR repertoire of a singlelung metastasis in patients 1464 and 1214 at very high frequencies.

DISCUSSION

Prevailing opinion holds that the TCR repertoire in TIL of singlemelanoma specimens is often skewed to a small number of TCRVregions (Mackensen et al, 1993; Strohal et al, 1994a; Thor Straten et al,1994; Salvi et al, 1995), but there exist only few comparative dataconcerning the TCR repertoire in primary melanoma and subsequent/coexisting melanoma metastases at different organ sites. SemiquantitativeTCRV analysis of TIL in multiple coexisting melanoma specimensrevealed the overall predominance of a restricted repertoire of TCRVregions, with AV13 and BV14 being by far the most frequentlyoverexpressed specificities. By cloning and sequencing of the predomin-ating TCR β-chain transcripts, we found that the majority were biasedto only three BV and BJ gene segment subfamilies, indicating an evenmore restricted usage of certain TCR specificities. In addition, asubstantial number of tumor specimens harbored clonally expandedTCR transcripts dominating the individual tumor sites at very highfrequencies. Support for the assumption that T cell populationsexpressing the clonally expanded TCR specificities represent tumor-reactive effector cells rather than innocent bystander lymphocytes,comes from a series of studies by Mackensen et al (1993, 1994) andFerradini et al (1993). By comparing TCR phenotypes in a regressingmelanoma and CTL reactivity of T cell lines generated from the verysame lesion, they were able to show that those TCR clonotypes thatpredominated the in situ TCR repertoire at the tumor site also displayedspecific HLA class 1-restricted cytotoxic activity against autologousmelanoma cells in vitro. A synoptic view of these results and the datapresented in this manuscript supports the concept of a specific tumor-triggered immune response leading to the in vivo overexpression of alimited number of TCR specificities with tumor-directed effectorfunction.

Concerning the antigenic epitopes recognized, patient-related differ-ences in the MAA expression patterns and the corresponding HLA-Aphenotypes could not be attributed to the preferential use of any TCRβ-chain transcripts; however, by comparing TCR sequences from theliterature with cDNA clones herein delineated, we found that β-chainsreportedly involved in the recognition of gp100 and MART-1/Melan-A show BVJC segment combinations with substantial CDR3 regionhomology identical or, at least, very similar to those of certain T cellclonotypes isolated from our patients’ gp100- and MART-1/Melan-A-expressing tumor specimens. Although we are aware that the precisespecificity of TCR moieties can only be determined by functional CTLstudies, it is certainly noteworthy that a TCR clonotype simultaneouslyoccurring in lung and kidney metastases of patient 1214 (HLA-B5101/3501) exhibits, with the exception of only two CDR3 region residues,

the same TCR β-chain expressed on two related HLA-B24-restrictedmelanoma-reactive CTL (TIL-1 and TIL-6, Mackensen et al, 1993,1994). Similarly, a clonally expanded BV14 transcript from the MART-1/Melan-A-expressing lung metastasis of the same HLA-A2-negativepatient (HLA-A2402, B5101/3501, C0401/02022) showed high struc-tural homology with TCR β specificities on HLA-A2-restricted CTLreportedly targeting MART-1/Melan-A (Sensi et al, 1995). It isremarkable that the immunogenic peptide AAGIGILTV on the MART-1/Melan-A antigen that is recognized by almost all HLA-A2-restrictedMART-1/Melan-A-reactive TIL (Parker et al, 1994) has a considerablyhigher affinity to our patient’s HLA-B5101 moieties than HLA-A2(computerized HLA Peptide Binding Prediction Model at http://bimas.dcrt.nih.gov:80/molbio/hla_bind, BioInformatics & MolecularAnalysis Section, NIH, Washington DC, Bouwer et al, 1994; Strohalet al, unpublished observation).

By comparatively analysing TRC gene usage, CDR3 region diversity,and MAA expression pattern in disseminated melanoma, we sought toglean information about the mode of generation as well as the diversityof anti-melanoma host responses. Our finding that in two of ourpatients a total of five different TCR clonotypes simultaneouslyoccurred at more than one anatomical site provides strong evidencefor the ability of the tumor to recruit unifocally activated T cell clonesof identical specificity to different organ sites. Support for this conceptof a monocentrically originated T cell response with resultant dissemina-tion of homogeneous effector cell populations also comes from severalrecent studies by Cole et al (1997) showing the same TCR specificitiesin matched MART-1/Melan-A peptide-stimulated PBL and TIL, byHishi et al (1997), who isolated similar tumor-reactive CTL clonotypesfrom two separate cutaneous metastases (abdomen, shoulder) of a givenpatient, and by Sensi et al (1997) identifying BV14 TCR cDNA cloneswith identical TCR β-chain sequences in four different cutaneousmetastases of a vaccine-treated patient.

Despite the existence of such identical T cell clones at different sites,we failed to detect a TCR clonotype that was expressed in all coexistingmetastases of a given patient.

Instead, the majority of TCR transcripts dominating the individualtumor sites at very high frequencies presented with site-specificdifferences in the composition of their CDR3 region and, lessfrequently, also in their BVJ repertoire. It thus appears that multifocallygenerated and locally expanded TCR cDNA clones are also part ofthe anti-tumor immunity in our patients’ specimens. Notably, theseexperimental observations are in keeping with our previous findingthat TIL within cutaneous melanoma lesions often express the cutaneousleukocyte antigen, whereas lymphocytes infiltrating extracutaneoustumor deposits do not display the skin-homing receptor (Strohal et al,1994b). As functional data are lacking, no definitive decision can bemade whether this combination of highly conserved BVJ gene segmentcombinations in the context of heterogeneous CDR3 regions withincoexisting melanoma sites is indicative of the occurrence of functionallyunrelated TCR specificities, which were generated by randommutations in rapidly growing melanoma cells, or, alternatively, resultsfrom the recognition of related epitopes of a restricted number ofMAA. Support for the latter assumption comes from a manuscript byVisseren et al (1997), who found that in vitro priming of PBL fromhealthy donors with the tyrosinase 369–377 peptide can induce peptide-specific CTL that, despite a restricted usage of AV and BV genesegments, exhibit a considerable diversity in their CDR3 regionsresulting in a distinct lytic fine specificity of each clonotype.

A word of caution is warranted at this point, however, in that theTCR and MAA profiles described in this report are not necessarilyrepresentative of the entire disease spectrum of the patients studied.Because we were analyzing autopsy specimens from patients withmetastatic melanoma, we must assume that the immune responsesstudied were largely nonprotective. Thus, the observation that certaindisseminated TCR clonotypes were not present in all tumor depositsof a given patient can also be explained by the delivery of anergizing/tolerizing signals to tumor-reactive lymphocytes, possibly resulting inthe deletion of such clonotypes (Schwartz, 1992; Enk et al, 1997). Asfar as MAA expression is concerned, immune escape mechanisms cancertainly contribute to an only limited number of specificities displayed

VOL. 111, NO. 6 DECEMBER 1998 TCR REPERTOIRE IN METASTATIC MELANOMA 1091

on a given melanoma cell nest/nodule (Browning and Bodmer, 1992;Cromme et al, 1994).

Notwithstanding these considerations, we postulate that the T cell-mediated host defense in metastasizing melanoma comprises multiplesimultaneously executed immune responses that are also influenced bythe different sites of tumor growth. Conceivably, this situation hasimportant implications for the development of more effective immuno-therapeutic regimens in melanoma patients with advanced disease. Inthe case of adoptive immunotherapies, the readministration to thepatient of in vitro expanded TCRV region-defined TIL isolated frommultiple tumor sites, should be advantageous to the infusion of TILof unifocal origin. Following this reasoning, one must also assume thatindividual melanoma patients should benefit most from peptide-basedvaccines containing a broad array of tumor cell-associated specificitiesand/or cancer cell vaccines generated from MAA- and TCR-typedtumor deposits of different sites.

We are indebted to Dr. M. Sensi for helpful comments, Dr. van der Bruggen forsupporting us with TCR sequences on MAGE- and BAGE-specific CTL as well ascDNA clones encoding the MAGE, BAGE, and GAGE MAA families, and toL. Stingl for reviewing the manuscript. This work was supported by grant 5562 fromthe Federal Reserve Bank of Austria, by a grant from Bender Inc., Vienna, Austria,and by the Commission of Cancer Research, University of Vienna Medical School.

REFERENCES

Adema GJ, de Boer AJ, Vogel AM, Loenen WA, Figdor CG: Molecular characterizationof the melanocyte lineage-specific antigen gp100. J Biol Chem 269:20126–20133, 1994

Arden B, Clark SP, Kabelitz D, Mak TW: Human T-cell receptor variable gene segmentfamilies. Immunogenetics 42:455–500, 1995

Boel P, Wildmann C, Renauld JC, Coulie P, Boon T, van der Bruggen P: BAGE: a newgene encoding an antigen recognized on human melanomas by cytolytic Tlymphocytes. Immunity 2:167–175, 1995

Bouwer HGA, Fischer-Lindahl K, Baldridge JR, Wagner CR, Barry RA, Hinrichs DJ:An H2-TMHC class Ib. molecule presents Listeria monocytogenes-derived antigen toimmune CD81 cytotoxic T cells. J Immunol 152:5352–5360, 1994

Brasseur F, Marchand M, Vanwijck R, Herin M, Lethe B, Chomez P, Boon T: Humangene MAGE-1, which codes for a tumor-rejection antigen, is expressed by somebreast tumors. Int J Cancer 52:839–841, 1992

Browning MJ, Bodmer WF: MHC antigens and cancer: implications for T-cell surveillance.Curr Opin Immunol 4:613–618, 1992

Chothia C, Boswell DR, Lesk AM: The outline structure of the T-cell α/β receptor.EMBO J 7:3745–3755, 1988

Cole DJ, Wilson MC, Rivoltini L, Custer M, Nishimura MI: T-cell receptor repertoirein matched MART-1 peptide-stimulated peripheral blood lymphocytes and tumor-infiltrating lymphocytes. Cancer Res 57:5320–5327, 1997

Cromme FV, Airey J, Heemels MT, et al: Loss of transporter protein, encoded by theTAP-1 gene, is highly correlated with loss of HLA expression in cervical carcinomas.J Exp Med 179:335–340, 1994

De Plaen E, Arden K, Traversari C, et al: Structure, chromosomal localization, andexpression of 12 genes of the MAGE family. Immunogenetics 40:360–369, 1994

De Smet C, Lurquin C, van der Bruggen P, de Plaen E, Brasseur F, Boon T: Sequenceand expression pattern of the human MAGE2 gene. Immunogenetics 39:121–129, 1994

Dunn DA, Gadenne AS, Simha S, Lerner EA, Bigby M, Bleicher PA: T-cell receptor Vβexpression in normal human skin. Proc Natl Acad Sci USA 90:1267–1271, 1993

Elbe A, Kilgus O, Strohal R, Payer E, Schreiber S, Stingl G: Fetal skin – a site of DETCdevelopment. J Immunol 149:1694–1701, 1992

Enk AH, Jonuleit H, Saloga J, Knob J: Dendritic cells as mediators of tumor-inducedtolerance in metastatic melanoma. Int J Can 73:309–316, 1997

Fae I, Petrasek M, Broer E, Mayr WR, Fischer GF: Nucleotide sequencing analysis ofHLA-C alleles. In: Charron D (eds). Genetic Diversity of HLA. Functional and MedicalImplications, Paris: EDK, 1996, pp 1–3

Ferradini L, Mackensen A, Genevee C, Bosq J, Duvillard P, Avril MF, Hercend T: Analysisof the T cell receptor variability in tumor-infiltrating lymphocytes from a humanregressing melanoma. J Clin Invest 91:1183–1190, 1993

Glisin V, Crkvenjakov R, Byus C: Ribonucleic acid isolated by caesium chloridcentrifugation. Biochemistry 13:2633–2637, 1974

Hishi M, Andrews D, Boyle LA, Wong JT, Pandolfi F, van den Elsen PJ, Kurnick JT: In-vivo accumulation of the same anti-melanoma T cell clone in two different metastaticsites. Proc Natl Acad Sci USA 94:1378–1383, 1997

Itoh K, Platsoucas CD, Balch CM: Autologous tumor-specific T lymphocytes in theinfiltrate of human metastatic melanomas. Activation by interleukin 2 and autologoustumor cells, and involvement of the T cell receptor. J Exp Med 168:1419–1441, 1988

Kawakami Y, Eliyahu S, Delgado CH, et al: Cloning of the gene coding for a sharedhuman melanoma antigen recognized by autologous T cells infiltrating into tumors.Proc Natl Acad Sci USA 91:3515–3519, 1994

Mackensen A, Ferradini L, Carcelain G, Triebel F, Faure F, Viel S, Hercend T: Evidencefor in situ amplification of cytotoxic T-lymphocytes with antitumor activity in ahuman regressive melanoma. Cancer Res 53:3569–3573, 1993

Mackensen A, Carcelain G, Viel S, et al: Direct evidence to support the immunosurveillanceconcept in a human regressing melanoma. J Clin Invest 93:1397–1402, 1994

Maeurer MJ, Martin DM, Storkus WJ, Lotze MTTCR: usage in CTLs recognizingmelanoma/melanocyte antigens. Immunol Today 16:603–604, 1995

Muul LM, Spiess PJ, Director EP, Rosenberg SA: Identification of specific cytolyticimmune responses against autologous tumor in humans bearing malignant melanoma.J Immunol 138:989–995, 1987

Parker KC, Bednarek MA, Coligan JE: Scheme for ranking potential HLA-A2 bindingpeptides based on independent binding of individual peptide side-chains. J Immunol152:163–167, 1994

Rosenberg SA, Spiess P, Lafreniere RA: new approach to the adoptive immunotherapy ofcancer with tumor-infiltrating lymphocytes. Science 233:1318–1321, 1986

Salvi S, Segalla F, Rao S, et al: Overexpression of the T-cell receptor β-chain variableregion TCRBV14 in HLA-A2-matched primary human melanomas. Cancer Res55:3374–3379, 1995

Schiffer M, Wu TT, Kabat EA: Subgroups of variable region genes of beta chains of T-cell receptors for antigen. Proc Natl Acad Sci USA 83:4461–4463, 1986

Schwartz RH: Costimulation of T lymphocytes: the role of CD28, CTLA-4, and B7/BB1 in interleukin-2 production and immunotherapy. Cell 71:1065–1068, 1992

Sensi M, Parmiani G: Analysis of TCR usage in human tumors: a new tool for assessingtumor-specific immune responses. Immunol Today 16:588–595, 1995

Sensi M, Salvi S, Castelli C, et al: T cell receptor (TCR) structure of autologous melanoma-reactive cytotoxic T lymphocyte (CTL) clones: tumor- infiltrating lymphocytesoverexpress in vivo the TCR β-chain sequence used by an HLA-A2-restricted andmelanocyte-lineage-specific CTL clone. J Exp Med 178:1231–1246, 1993

Sensi M, Traversari C, Radrizzani M, et al: Cytotoxic T-lymphocyte clones from differentpatients display limited T-cell-receptor variable-region gene usage in HLA-A2-restricted recognition of the melanoma antigen Melan-A/MART-1. Proc Natl AcadSci USA 92:5674–5678, 1995

Sensi M, Garina C, Maccalli C, et al: Clonal expansion of T lymphocytes in humanmelanoma metastases after treatment with a hapten-modified autologous vaccine.J Clin Invest 99:707–717, 1997

Southern EM: Detection of specific sequences among DNA fragments separated by gelelectrophoresis. J Mol Biol 98:503–517, 1975

Strohal R, Paucz L, Pehamberger H, Stingl G: T-cell receptor repertoire of lymphocytesinfiltrating cutaneous melanoma is predominated by Vα specificities present in T-cells of normal human skin. Cancer Res 54:4734–4739, 1994a

Strohal R, Marberger K, Pehamberger H, Stingl G: Immunohistological analysis of anti-melanoma host responses. Arch Derm Res 287:28–35, 1994b

Thor Straten P, Scholler J, Jensen KH, Zeuthen J: Preferential usage of T-cell receptor α/β variable regions among tumor-infiltrating lymphocytes in primary human malignantmelanomas. Int J Cancer 56:78–86, 1994

Van den Eynde B, Brichard VG: New tumor antigens recognized by T cells. Curr OpinImmunol 7:674–681, 1995

Van den Eynde B, Peeters O, de Backer O, Gaugler B, Lucas S, Boon T: A new familyof genes coding for an antigen recognized by autologous cytolytic T lymphocyteson a human melanoma. J Exp Med 182:689–698, 1995

Van der Bruggen P, Szikora JP, Boel P, Wildmann C, Somville M, Sensi M, Boon T:Autologous cytolytic T lymphocytes recognize a MAGE-1 nonapeptide onmelanomas expressing HLA-Cw*1601. Eur J Immunol 24:2134–2140, 1994

Visseren MJ, van der Burg SH, Hawes GE, van der Voort EIH, van den Elsen PJ, MeliefCJM: Affinity specificity T-cell receptor diversity of melanoma-specific CTLgenerated in vitro against a single tyrosinase epitope. Int J Cancer 72:1122–1128, 1997

Weidmann E, Elder EM, Trucco M, Lotze MT, Whiteside TL: Usage of T-cell receptorVβ-chain genes in fresh and cultured tumor-infiltrating lymphocytes from humanmelanoma. Int J Cancer 54:383–390, 1993

WHO-IUIS: Nomenclature Sub-Committee on TCR Designation: Nomenclature for T-cell receptor (TCR) gene segments of the immune system. Immunogenetics 42:451–453, 1995

Zarour H, de Smet C, Lehmann F, et al: The majority of autologous cytolytic T lymphocyteclones derived from peripheral blood lymphocytes of a melanoma patient recognizean antigenic peptide derived from Pmel17/gp100. J Invest Dermatol 107:63–67, 1996