A Novel Approach to Characterize Clonality and Differentiation of Human Melanoma-Specific T Cell Responses: Spontaneous Priming and Efficient Boosting by Vaccination

Clinical history

Patient LAU 444, HLA-A*0201 positive, was diagnosed at the age of 27 years with primary nodular skin melanoma of the right foot in January 1998. Breslow tumor thickness was 1.9 mm. In June 2000, a metastatic lymph node (LN) prompted a LN dissection (right inguinal 1N⁺/8N, right iliacal 0N⁺/16N). A sample of the metastatic LN is referred to as tumor-infiltrated lymph node cell 1 (TILN-1). Local radiotherapy was applied from July to September 2000. In May 2001, a s.c. metastasis was removed from the right thigh (tumor-infiltrating lymphocyte 2 (TIL-2)). Subsequently, after informed consent, the patient was enrolled in a vaccination trial (Ludwig 96-010; see Fig. 1⇓A) (13) and received s.c. injections of a mixture of 100 μg of Melan-A_26–35 ELAGIGILTV analog peptide and 100 μg of influenza matrix protein_58–66 GILGFVFTL peptide, emulsified in 600 μl of IFA (Montanide ISA-51). This was started in July 2001, referred to as day 0. Additional vaccines were given after 1/2/3 mo, 5/6/7/8 mo, 11/12/14/15 mo, and 19/20/21/23 mo. In October 2001 (3 mo), a metastasis in the left shoulder was removed by surgery. A metastasis at the left talus was found in April 2002 (9 mo) and treated with radiotherapy, and a s.c. metastasis at the right arm in August 2002 (13 mo) was resected by surgery. Two intestinal metastases required laparotomy in June 2003 (24 mo) and a s.c. metastasis at the right knee was removed by surgery (25 mo). The patient was enrolled in a second vaccination trial (LUD 00-018; see Fig. 1⇓A) (31) and received s.c. injections of 100 μg of Melan-A ELAGIGILTV peptide and 500 mg of CpG oligodeoxynucleotide 7909 (provided by Coley Pharmaceutical Group), emulsified in 300 μl of IFA. This was started in Sept 2003 (27 mo), followed by recall vaccinations after 28/29/30 mo; 31/32/34/35 mo; 37/38/39/40 mo, and 41/42/43/44 mo. The single persisting bone metastasis in the left talus had remained stable but ultimately progressed and was surgically removed (43 mo; TIL-3). The disease remained stable for another 5 mo, then several liver metastases arose that responded well to chemotherapy. This study has been reviewed and approved by an appropriate institutional review committee.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Ex vivo analysis of the differentiation phenotype of Melan-A-specific CD8⁺ T cells over time. A, Clinical course of patient LAU 444. Occurrence of metastases is indicated by filled circles, vaccinations by short vertical lines, and withdrawal of blood samples by long vertical lines. Metastatic tissues from which T lymphocytes were isolated are referred to as TILN-1, TIL-2, and TIL-3. PBMCs were collected from blood samples before (−13 mo) and at regular time points during peptide vaccination. B, Multimer⁺ T cells were characterized ex vivo by flow cytometry for their cell surface expression of CD45RA, CCR7, and CD28. CCR7/CD45RA phenotype in Melan-A⁺CD8⁺ T lymphocytes (R1) (middle). CD28 expression was characterized after gating on naive (R2; RA⁺CCR7⁺) and effector-memory (R3; RA⁻CCR7⁻) Melan-A-specific T cells (right). The percentage of CD28⁺ T cells is indicated. C, Percentage of Melan-A-specific T cells in CD8⁺ T cells over time. Of note, PBMCs contained detectable frequencies of multimer⁺ T cells before immunotherapy with 0.05, 0.13, and 0.14% of CD8 T cells at −13 mo, −2 mo, and day 0, respectively. D, Percentage of naive (▪) and primed EM (□) Melan-A-specific T cells in CD8⁺ T cells (log scale) over time. Note that naive Melan-A-specific T cells remained at low frequencies ranging between 0.01 and 0.06%. EM, effector memory.

A2/peptide multimers and flow cytometry immunofluorescence analysis

PBMCs were obtained by density centrifugation using Ficoll-Hypaque (Pharmacia) and cryopreserved in RPMI 1640 supplemented with 40% FCS and 10% DMSO (1 × 10⁷ to 2 × 10⁷ cells/vial). Synthesis of PE-labeled HLA-A*0201/peptide multimers were prepared as described previously (32, 33) with Melan-A analog peptide_26–35 ELAGIGILTV. Five color stains were done with PE-HLA-A2/peptide multimers, FITC-conjugated anti-CD28, -CD27, and -CD11a (BD Biosciences), PE-Texas Red-conjugated anti-CD45RA (Beckman Coulter), allophycocyanin/Cy7-conjugated anti-CD8 (BD Biosciences) reagents, and anti-CCR7 mAb (BD Biosciences) followed by allophycocyanin-conjugated goat anti-rat IgG Ab (Caltag Laboratories). Briefly, CD8⁺ T lymphocytes were positively enriched from PBMCs using anti-CD8-coated magnetic microbeads (Miltenyi Biotec), resulting in >90% CD3⁺CD8⁺ lymphocytes. Cells were first stained with PE-labeled multimers for 1 h at room temperature (RT) in PBS, 0.2% BSA, 50 μM EDTA, and then with appropriate Abs (30 min, 4°C). Intracellular content of granzyme B and perforin was measured in CD8⁺ T lymphocytes without previous stimulation. After staining with appropriate mAbs, cells were fixed for 20 min at RT in PBS containing 1% formaldehyde, 2% glucose, and 5 mM sodium azide. Fixation was followed by permeabilization with PBS/0.1% saponin (Fluka)/0.2% BSA/50 μM EDTA and staining with granzyme B-FITC (Hölzel Diagnostika) or perforin-FITC mAbs (Alexis), both for 20 min at RT. Cells were immediately analyzed into defined subpopulations on a FACSVantage SE using CellQuest software (BD Biosciences).

cDNA amplification, TCR spectratyping, and clonotyping

Five-cell aliquots were directly sorted with a FACSVantage SE machine into wells of 96-V-bottom plates. cDNA preparation, cDNA amplification, and PCR were performed as described (34). The complete procedure for CDR3 size analysis of TCR transcripts as well as the sequences of oligonucleotides corresponding to the 22 variable segments of the TCR β-chain (based on the nomenclature proposed by Arden et al. (35)) and the TCR AV2.1 gene segment used in this study have been published previously (25, 36). In brief, 8 μl taken from 10 individually sorted and amplified five-cell cDNA samples were pooled together to obtain total cDNA material equivalent to ex vivo-sorted 50 cells. cDNA pools generated from circulating naive, EM28⁺, and EM28⁻ Melan-A-specific T cells at different time points before and after vaccination as well as cDNA pools from Melan-A-specific T cells sorted from a single-cell suspension of metastatic tissues (TILN-1, TIL-2, and TIL-3) were subjected to individual PCRs using a set of validated 5′ sense fluorescent-labeled primers specific for AV2.1 and 22 V region of the β-chain (BV) subfamilies and one 3′ antisense primer specific for the corresponding C gene segment. PCR products were then run on an automated sequencer in the presence of fluorescent size markers and data analysis was performed with the Genescan analysis software (Applied Biosystems). TCR spectratyping analysis performed directly on ex vivo-sorted 50 cells gave comparable results to analysis done on pooled 50 cells (10 × 5 cells; data not shown). TCR Vβ-Cβ and Vα-Cα PCR products were directly purified and sequenced (Fasteris) when single-dominant PCR peaks only were identified. A set of primers specific for the CDR3 region of the identified BV13- and BV3-TCR clonotypes was used for clonotyping PCR (Metabion): BV13 clonotype-left (L): 5′-GTCACTCAGACCCCAAAATTC-3′; reverse (rev)-5′-TAGGATGCTGTCCCCAGTTC-3′, BV13 clonotype-right (R): 5′-GAACTGGGGACAGCATCCTA-3′; rev-5′-TGACCGTGAGCCTGGTGCCCG-3′ and BV3 clonotype: 5′-TGGAATATAGCGGGAGTGGG-3′; rev-5′-TGACCGTGAGCCTGGTGCCCG-3′.

Immunohistochemistry

For Melan-A staining, sections (4 μm) of Formalin-fixed, paraffin-embedded tissue samples were stained with the anti-Melan-A Ab A103 (37) using an avidin-biotin peroxidase system following citrate buffer (pH 6.0) Ag retrieval. Diaminobenzidine was used as chromogen. Sections were counterstained with hematoxylin. S100 staining was performed as routinely described.

T cell cloning, culture, and cytolytic activity

Multimer⁺ CD8⁺ T cell subsets (naive, EM28⁺, and EM28⁻) were sorted by flow cytometry, cloned by limiting dilution, and expanded in RPMI 1640 medium supplemented with 8% human serum, 150 U/ml recombinant human IL-2 (rIL-2; a gift from GlaxoSmithKline), 1 μg/ml PHA (Sodiag) and 1 × 10⁶/ml irradiated allogeneic PBMC (3000 rad) as feeder cells. Positive Melan-A-specific T cell clones were periodically (every 15 days) restimulated in 24-well plates with PHA, irradiated feeder cells, and rIL-2. Lytic activity and Ag recognition was assessed functionally in 4-h ⁵¹Cr-release assays using peptide-pulsed T2 cells (HLA-A2⁺/TAP^−/−), and the melanoma cell lines Me 275 (HLA-A2⁺/Melan-A⁺), NA8-MEL (HLA-A2⁺/Melan-A⁻), and Me 260 (HLA-A2⁻). The percentage of specific lysis was calculated as following: 100 × ((experimental − spontaneous release)/(total − spontaneous release)).

Telomere fluorescence in situ hybridization (FISH) and flow cytometry (flow FISH)

The average length of telomere repeats at chromosome ends in individual cells, was measured by FISH and flow FISH as previously reported (34, 38). Telomere length measurements were performed on in vitro-derived T cell clones sorted from naive and EM28⁻ Melan-A-specific T subsets at various time points following immunotherapy, allowing the recovery of sufficient numbers of cells for flow FISH analysis. All T cell clones were expanded in identical in vitro culture conditions, and after a single round of stimulation, 2 × 10⁵ cells were further processed by flow FISH. Because the average telomere fluorescence from all these clones was evaluated in the same experimental design, this allowed direct telomere comparison between each tested clones. We estimated that all clones had on average shortened their telomere lengths by ∼1.5 kb following their in vitro expansion round (39). Telomere fluorescence was calculated by subtracting the mean fluorescence of the background control (no probe) from the mean fluorescence obtained from cells hybridized with the telomere probe after calibration with FITC-labeled fluorescent beads (Quantum TM-24 Premixed; Bangs Laboratories) and conversion into molecules of equivalent soluble fluorochrome (MESF) units. The following equation was used to estimate the telomere length in base pair: base pair = MESF × 0.495 (38).

Detection of Ag-experienced Melan-A-specific T cells before immunotherapy, followed by increased frequencies along subsequent peptide vaccinations

Patient LAU 444 with advanced melanoma regularly developed multiple metastases despite repetitive vaccination with Melan-A_26–35 peptide emulsified in IFA (Fig. 1⇑A). Subsequently, the vaccines were supplemented with CpG oligodeoxynucleotides 7909 (31), and the patient remained without new metastases for a period of 16 mo. Using fluorescent HLA-A2/peptide multimers for ex vivo analysis, we quantified Melan-A-specific CD8⁺ T lymphocytes in PBMCs and characterized their T cell differentiation stage (Fig. 1⇑B). We found that the Melan-A-specific T lymphocytes predominantly bore a naive phenotype (RA⁺CCR7⁺28⁺) when analyzed at the earliest time point, 13 mo, before the start of peptide vaccination (−13 mo). Sometime before the start of immunotherapy (−2 mo and day 0), the majority of tumor-specific T cells differentiated into EM cells (RA⁻CCR7⁻). This phenotypic change lasted over >44 mo. EM Melan-A T cells were further split into two distinct subsets with a large proportion being CD28 negative (EM28⁻; mean ± SD, 89.6 ± 7.9%), thus representing the predominant phenotype among total multimer⁺ T cells (mean ± SD, 77 ± 8.7%). Although PBMCs contained low but readily detectable frequencies of Melan-A-specific T cells before peptide vaccination, peptide/IFA immunization induced a significant increase of tumor-specific T cells with frequencies as high as 1.34 ± 0.25% (Fig. 1⇑C). Vaccination with CpG 7909 in addition to Melan-A_26–35 peptide mixed with IFA, allowed further substantial frequency increases (2.9% ± 0.77%). This selective expansion of Melan-A-specific T cells was primarily due to increased frequencies of EM cells (Fig. 1⇑D).

The dominant Melan-A-specific T cell subset (EM28⁻) acquired enhanced ex vivo effector properties following serial peptide vaccinations

To directly assess expression of genes involved in effector functions, we used a modified RT-PCR protocol that detects specific cDNAs after global amplification of expressed mRNAs (34, 40). Because this method yields sufficient cDNA from as few as five cells, it allows the analysis of gene expression in small purified subsets of tumor-specific T cells ex vivo, obtained from limited amount of blood and metastatic material. We sorted naive, EM28⁺, and EM28⁻ Melan-A-specific T cell subsets from blood at different time points, and compared the presence of corresponding mRNAs in five-cell aliquots of these populations (Fig. 2⇓). Naive T cells contained no detectable granzyme B, perforin, nor IFN-γ mRNA, whereas these mRNA transcripts were found in a significant fraction of EM28-negative Melan-A-specific T cell aliquots over time (Fig. 2⇓A). This was particularly true for granzyme B mRNA expression, and correlated with the high expression level for this protein (Fig. 2⇓C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Ex vivo analysis of expression of effector mediators in Melan-A-specific CD8⁺ T cell subsets over time. A and B, Gene expression analysis was performed on sorted naive, EM28⁺, and EM28⁻ Melan-A-specific T lymphocytes from blood at regular time points using a modified RT-PCR protocol (34 ). Primers designed for CD3, CCR7, CD27, granzyme B, perforin, and IFN-γ mRNA transcript analysis are depicted. Data from six independent five-cell aliquots are shown; (−), negative; (+) positive controls. Of note, all naive T cell samples but almost none of the aliquots of EM28⁺ and EM28⁻ T cells yielded a detectable CCR7-specific product, thus correlating with CCR7 cell surface expression. C, The proportion of CD27-, granzyme B-, perforin- and CD11a-positive cells among multimer⁺-RA⁻CCR7⁻ (EM) specific T cells, was determined over time by immunofluorescence. Bulk CD8⁺RA^+/− CCR7⁻ T cells known to express effector mediators are also depicted (CD8t, total CD8; gray histogram). Note that the perforin signal was significantly lower at +16 mo (post-IFA) than at later time points (post CpG; +30 and +44 mo). Moreover, the increased proportion of CD27⁺ T cells observed after peptide/IFA/CpG vaccination was concomitant with the frequency increase of EM28⁺ T cells (Fig. 1⇑B; see +44 mo/CpG).

Recent studies have shown that CD8⁺ T lymphocytes sequentially down-regulate CCR7, CD28, and CD27 surface expression, while up-regulating expression of molecules that confer cytolytic activity (29, 30). Before (−2 mo) and early (+3 mo) after the start of immunotherapy, the proportion of EM28⁻ samples containing detectable levels of CD27 transcripts was similar to that of the naive subset (Fig. 2⇑A). However, down-regulation of CD27 expression was observed within this subset from month 9 onward, and was highly coincident with the progressive emergence of both mRNA (Fig. 2⇑A) and protein (Fig. 2⇑C) expression for perforin. Overall, our data show that EM28-negative T cells, although present before immunotherapy, acquired enhanced effector functions during serial peptide vaccinations. Similar changes in effector gene expression were observed for the Melan-A-specific EM28-positive T cell subset after subsequent CpG 7909 coinjection (Fig. 2⇑B).

Oligoclonally expanded cells were selectively identified in the differentiated EM28-negative Melan-A multimer⁺ CD8⁺ T cell subset

In several viral systems, responses to a given T cell Ag often showed strong biases in TCR selection, resulting in highly restricted TCR usage (5, 6, 7, 8, 9, 10, 11). As we observed a dominant Melan-A-specific T cell response composed of functionally differentiated cells with potent effector properties (Figs. 1⇑ and 2⇑), we next asked the question of whether similarly to antiviral responses this subpopulation would be composed of oligoclonally expanded T cells. Using the ex vivo-amplified cDNA material, we compared the diversity of the T cell repertoires from sorted naive, EM28⁺, and EM28⁻ Melan-A-specific T cell subsets before and during peptide vaccination. We identified TCRs with different CDR3 lengths of the BV using spectratyping (36, 41). Naive and primed EM28⁺ Melan-A-specific T cells revealed large polyclonal TCR repertoires (Fig. 3⇓A), with a diverse usage of the 22 different BV families as well as high variability within CDR3 size products. In contrast, restricted TCR diversity with large expansions of distinct BV families was found among the dominant EM28-negative Melan-A-specific T cell subset. We identified precisely within the latter subset, a TCR-BV13 of defined CDR3 length, which was present in all of the samples analyzed (169 bp; Fig. 3⇓B). Strikingly, TCR sequencing revealed that all EM28⁻ T cells bearing that particular TCR-BV13 subfamily shared the same CDR3 loop sequence (SELGTASYEQ) as well as the same joining gene element (BJ2.7), hereafter called BV13-ELGTASY clonotype (Table I⇓). Moreover, this molecular-based strategy allowed us to identify a second dominant T cell clonotype expressing the TCR-BV3-FWNIAGVGEQ-BJ2.7 gene segments, which was, however, undetectable in three of seven blood samples analyzed (Fig. 3⇓A, Table I⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Ex vivo longitudinal analysis of TCR BV chain repertoire among sorted Melan-A-specific CD8⁺ T cell subsets in blood and metastatic tissues. A, cDNA pools (50 cells) generated from naive, EM28⁺, and EM28⁻ Melan-A-specific T cells sorted from circulating CD8⁺ T cells at different time points were amplified by PCR using 22 BV-specific primers (36 ) and subjected to electrophoresis on an automated sequencer. Each BV subfamily (x-axis) was analyzed for the presence of amplified BV-CDR3-BC products of defined CDR3 size (y-axis). These are displayed by black squares (▪) on a grid. The columns corresponding to the BV3 and BV13 subfamilies are depicted. The figures inserted in each grid indicate total number of BV subfamily gene segment usage vs (the total number of all amplified BV-CDR3-BC products within each positive BV subfamily). B, Distribution of PCR products by CDR3 size for BV13 subfamily in naive, EM28⁺, and EM28⁻ subsets over time. C, TCR spectratypes performed on Melan-A-specific CD8⁺ T cells (50-cell cDNA pools) sorted from freshly prepared single-cell suspensions of metastatic tissues (TILN-1, TIL-2, and TIL-3). Total numbers of TCR-BV usage comprising all amplified BV-CDR3-BC size products are indicated in each grid. D, Specific PCR amplification patterns for the BV13 subfamily in metastatic tissues (TILN-1, TIL-2, and TIL-3). Spectratype from a normal naive TCR-BV13 repertoire follows a Gaussian distribution of eight or more peaks (C⁺). Note the recurrent presence of the BV13 subfamily with amplified BV-CDR3-BC size product, 169 bp, among EM28⁻ T cells in both blood and metastatic tissues (see arrows).

These data were in sharp contrast with BV13 sequences obtained from naive or EM28⁺ Melan-A-specific T subsets, which were, beside a single exception, highly heterogeneous in terms of BJ usage, CDR3 length, and sequence (Table I⇑). Recent studies described a preferential usage of the V region element of the TCR α-chain (AV2.1) in circulating Melan-A-specific naive and experienced T cells (21, 25). Using primers specific for TCR-AV2.1 followed by sequencing, we again found a single dominant TCR α-chain composed of the AV2.1-MNYGGSQGNL-AJ42 gene segments, present before immunotherapy (−2 mo) and persistent ex vivo over 44 mo of serial peptide injections (Table I⇑).

Collectively, our strategy to focus on distinct T cell subsets allowed identifying two dominant T cell clonotypes (BV13-ELGTASY and BV3-FWNIAGVG). In particular, we observed a highly restricted TCR repertoire in the most differentiated T cell compartment (EM28 negative). Such TCR-restricted features are usually observed in antiviral responses. As we found oligoclonal T cell expansions, our data strongly support the notion that similar principles may also apply for antitumor responses.

In vivo long-term persistence and efficient homing to metastases of the dominant Melan-A-specific BV13-ELGTASY T cell clonotype

Although tumor-Ag-specific CD8 T cell responses such as those directed against the Melan-A/MART-1 Ag often develop in cancer patients, they rarely result in tumor eradication. Recently, we described that while circulating Melan-A-specific CD8⁺ T cells exhibit characteristics in common with the effector type of T cells, tumor-specific cells that reside in metastatic lymph nodes and in soft/visceral metastasis are functionally attenuated as compared with circulating T cells (12). To get insight into the discrepancy between successful T cell boosting and clinical outcome, we investigated the clonal composition within metastatic tissues, using the same strategy combining cell sorting with molecular analysis of single T lymphocytes. Undoubtedly, the BV13-ELGTASY T cell clone was not only confined to peripheral blood, but also present in two metastatic nonlymphoid tissues (TIL-2 and TIL-3) recovered −2 mo before and +43 mo after immunotherapy, respectively. In contrast, the secondly identified T cell clonotype, BV3-FWNIAGVG, was neither detectable in blood before vaccination nor in metastatic tissues (Fig. 3⇑, C and D; Table I⇑).

We further characterized whether the BV13-ELGTASY T cell clone was enriched directly in an inguinal lymph node metastasis, surgically resected as early as −13 mo before immunotherapy. Strikingly, this particular clonotype was already present in this metastatic lymph node (TILN-1; Fig. 3⇑, C and D; Table I⇑), while not detectable among the Melan-A-specific T cells from peripheral blood taken at the same early time-point (−13mo; Fig. 3⇑A). Together, these data revealed that the BV13-ELGTASY clonotype arose at least one year before vaccination and displayed efficient homing to metastases.

Ex vivo quantification of prevalent Melan-A-specific TCR clonotypes in blood and metastatic tissues

We next assessed the frequencies of both TCR-BV13 and BV3 clonotypes in peripheral blood and metastatic lesions as well as changes in their frequencies during serial peptide injections. None of the aliquots of naive cells and only very few of EM28⁺ Melan-A-specific T cells gave detectable TCR BV-clonotype-specific products (Fig. 4⇓C, Table II⇓), using designed clonotypic primers (Fig. 4⇓A). It remains unknown whether this observation is the result of small contaminations from the dominant EM28⁻ subset during cell sorting or reflects true biological presence of the BV13-ELGTASY clonotype, at low frequencies, in the EM28⁺ subset. In contrast, transcript analysis from circulating EM28⁻ T cells revealed the presence of the BV13-ELGTASY clonotype in most of five-cell aliquots from this subset (Fig. 4⇓B). Interestingly, this clonotype was already dominant within blood and a metastatic nonlymphoid tissue (TIL-2) recovered 2 mo before vaccination. In line with TCR-spectratyping analysis (Fig. 3⇑C), BV13-clonotypic frequencies were nevertheless significantly reduced in the metastatic lymph node (TILN-1; 4+/30 samples) and in the resected bone metastasis at 43 mo (TIL-3; 7+/18 samples) (Table II⇓). In addition, PCR analysis using BV3-clonotype-specific primers showed that the presence of this clonotype was restricted to blood samples following immunotherapy (Fig. 4⇓B) and when present, revealed significantly lower frequencies (Table II⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Quantification of particular Melan-A-specific T cell clones using clonotypic primers. A, Unique primers corresponding to the CDR3 gene sequence of each identified TCR clonotypes were designed and validated. Two distinct sets of primers were used for BV13-ELGTASY clonotype (BV13 clono-L and BV13 clono-R). The expected size for each primer set is indicated. Clonotypic PCR was performed on cDNA obtained either from individually sorted five-cell samples (B) or pooled 50 cells (C). B, Data from six independent five-cell aliquots generated from circulating EM28⁻ T cells over time and metastatic tissues are shown. C, Characterization of TCR BV13 and BV3 clonotypes among naive and EM28⁺ T cell subsets over time; (−), negative; (+) positive controls. Analysis using primers designed for the TCR AV2.1 clonotype sequence was performed on the same samples, and gave similar patterns of expression than for BV13 clonotype (data not shown). D, Percentage of ELGTASY-specific T cells in CD8⁺ T cells over time.

Altogether, our molecular-based approach not only provided the identification of specific T cell clonotypes ex vivo, but also allowed a direct estimate of their frequencies within blood and metastatic tissues. Indeed, considering the extended analysis of five-cell samples (n = 122; Table II⇑) and the highly restricted TCR repertoire diversity found within the EM28⁻ subset (Fig. 3⇑A), we estimated that ∼90% of EM28-negative Melan-A-specific T lymphocytes bore the dominant BV13-clonotypic TCR. Given that the Melan-A-specific compartment was largely composed by EM28⁻ T cells (Fig. 1⇑), the proportion of this particular clonotype among peripheral CD8⁺ T lymphocytes was further extrapolated (Fig. 4⇑D). Thus, the dramatic increase in frequencies of total multimer⁺ T cells observed following serial immunizations (Fig. 1⇑C) was largely due to the strong expansion of this monoclonal subpopulation, reaching up to 2.4% of the circulating CD8 pool. Finally, a major finding was that serial tumor peptide-based vaccinations were successful in boosting a pre-existing dominant clonotypic response (BV13-ELGTASY).

Loss of Melan-A expression by the tumor cells present within metastatic tissues during disease progression

Despite the strong immunological response induced by the clone BV13-ELGTASY and its efficient homing to metastases, patient LAU 444 exhibited stabilization of disease for some time but eventually developed progressive disease (Fig. 1⇑A). To investigate whether melanoma cells may have escaped from CTL recognition, expression of the Melan-A Ag by tumor cells was assessed in the metastatic tissues removed during the course of disease (Fig. 5⇓, Table III⇓). Ninety percent of tumor cells were positively stained for Melan-A in the metastatic LN (−13 mo) and 60–80% strongly expressed Melan-A in metastases (s.c. and gastrointestinal (GI) tract) resected at later time points (−2 mo, +24 mo). The bone metastasis in the left thalus, which remained stable over a period of 16 mo (Fig. 1⇑A), ultimately progressed and had 40% of tumor cells that expressed the Ag (bone, +43 mo; Table III⇓). This lesion showed for the first time large tumor areas that were completely Melan-A negative (Fig. 5⇓E). The patient remained stable for five additional months then liver metastases arose in which one biopsy showed complete loss of Melan-A expression (+48 mo; Table III⇓). In addition, immunohistochemistry (IHC) revealed a much lower degree of lymphocyte infiltration in the latter metastasis when compared with tumors removed at earlier time points during disease progression (data not shown). Similar results were obtained when Melan-A mRNA expression was analyzed by PCR (Table III⇓). Finally, the progressive loss of Melan-A expression by tumor cells was timely concomitant to that observed for the other melanocyte differentiation Ags tyrosinase and gp-100. In contrast, the cancer-testis Ag MAGE-3 remained expressed, despite reduced expression at the latest time point (Table III⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Melan-A expression in paraffin-embedded sections of resected metastases during the course of disease progression. IHC was performed on a LN metastasis (A, −13 mo), a s.c. metastasis (B, −2 mo), two intestinal metastases (C, jejenum, and D, ileum; +24 mo) and a bone metastasis (E and F, +43mo). A–E, Staining with anti-Melan-A Ab; F, S100 staining of a section adjacent to E to confirm presence of melanoma cells. Selected tumor areas are depicted, with reddish/brown color indicating positive Melan-A staining. A–C, Areas of tumor with nearly 100% of cells expressing Melan-A. D, A highly heterogeneous tumor area while E shows a completely negative nodule. The overall proportion of Melan-A-positive tumor cells present within the entire tissue section was, from A to E, 90, 70, 80, 60 and 40%, respectively (see Table III⇓). B, C, E, and F, ×100; A and D, ×200.

Efficient tumor cell recognition and killing by Melan-A-specific T cells bearing the BV13-ELGTASY TCR

For functional analysis, we generated over 200 T cell clones in vitro by limiting dilution from circulating Melan-A-specific naive (n = 22), EM28⁺ (n = 64), and EM28⁻ (n = 132) derived T cell subsets at various time-points. Almost all clones from EM28⁻ origin comprised the same dominant BV13-ELGTASY gene segments (>97%), whereas this VDJ sequence was found in <5% of naive and EM28⁺ T cell clones. T cell clones bearing the BV13 clonotype also coexpressed the recurrent AV2.1 clonotypic gene segment. These results are in excellent agreement with the ex vivo five-cell frequency data (Fig. 4⇑) confirming that the tumor-specific EM28 negative compartment is largely composed of a single long-term persistent T cell clone.

Twelve representative T cell clones were further tested in a functional chromium release assay to assess their ability to recognize Melan-A-expressing tumors (Fig. 6⇓A). Most of the EM28-negative-derived clones efficiently killed the Melan-A-expressing tumor cell line Me 275 as well as the patient autologous melanoma cell line, T444C, in contrast to naive-derived clones (Fig. 6⇓, A and B). Remarkably, most EM28⁻ T cell clones required 10² to 10³ times less peptide to achieve similar lysis than naive-derived T cells (Fig. 6⇓, C and D), demonstrating high functional avidity for the naturally processed A2/Melan-A peptides (EAAGIGILTV (EAA) and AAGIGILTV (AA) peptides) (42). Indeed, aside a single exception all EM28⁻ T clones recognized ELAGIGILTV (ELA) and EAA peptides with similar high avidity (IC₅₀ ranging between 10⁻¹² and 10⁻¹⁰ M; median, 3 × 10⁻¹¹ M). In contrast, naive derived clones killed T2 cells with an IC₅₀ ranging widely between 10⁻¹⁰ and 10⁻⁶ M (median, 4 × 10⁻⁹ M for ELA and 8 × 10⁻⁸ M for EAA peptide, respectively). Similar differences in functional avidity between naive and EM28⁻ T cell clones were observed when the Melan-A_27–35 (AA) peptide was used (Fig. 6⇓, C and D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Fine specificity of Melan-A Ag recognition and tumor cell killing. Naive and EM28⁻ derived Melan-A-specific T cell clones were generated from blood samples corresponding to +9, +16, and +32 mo after vaccination through flow cytometry sorting and limiting dilution. Ag-specific lytic activity was assessed in a functional 4-h chromium release assay. Upper panels, Naive-derived T cell clones; lower panels, EM28⁻-derived T cell clones. A and B, Specific lysis of melanoma cell lines Me 275 (A2⁺/Melan-A⁺; circles), NA8 (A2⁺/Melan-A⁻; squares), Me 260 (A2⁻; triangles) and the patient autologous melanoma cell line T444C (A2⁺/Melan-A⁺; diamonds) in the presence (filled symbols) or absence (open symbols) of synthetic Melan-A analog peptide (ELA). Results with a representative CTL clone are depicted (n = 12 clones). C, The relative TCR avidity was compared using T2 target cells (HLA-A2⁺/TAP^−/−) pulsed with graded concentrations of either analog Melan-A_26–35 peptide (ELAGIGILTV) or Melan-A_26–35 (EAAGIGILTV) and Melan-A_27–35 (AAGIGILTV) peptides, which mimic the naturally processed peptides (42 ). Representative examples are depicted. Melan-A analog (ELA; squares), Melan-A natural decamer (EAA; circles), Melan-A natural nonamer (AA; diamonds), and influenza matrix protein GILGFVFTL (triangles). D, Complete set of data (n = 8 clones) representing the peptide concentration (either ELA, EAA, or AA) used to achieve 50% of maximal lysis. Each data point represents the result from an individual clone.

Rapid telomere shortening within the dominant Melan-A-specific EM28-negative subset compared with the naive cells

To measure the turnover of the dominant Melan-A-specific EM28⁻ T cell subset in peripheral blood, the loss of telomeres was studied at different time points following vaccination. Because telomeres progressively shorten as a function of cell division (43), telomere length is a powerful indicator of the replicative in vivo history of lymphocytes (38, 44, 45). The naive T cell subset displayed the brightest telomere fluorescence at both time points (+9 and +16 mo) with a mean value corresponding to an average telomere lengths of 7.5 and 7.9 kb, respectively (Fig. 7⇓A). In contrast the EM28-negative T cell clones showed significantly reduced telomere fluorescence corresponding to a loss of ∼1.5 kb, indicating extensive proliferation in vivo. In line with these observations, the mean telomere fluorescence in those cells rapidly declined during serial peptide vaccination with an estimated telomere rate loss of >500 bp/year (Fig. 7⇓B). This corresponds to a 10-fold increased rate when compared with the previously reported telomere shortening in naive and memory CD8⁺ T lymphocytes with ageing (45). Most likely, the rapid and sustained loss of telomere repeats in the monoclonal EM28-negative subset reflects proliferation due to triggering by Ag derived from tumor and vaccination.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Analysis of telomere fluorescence in naive- and EM28⁻-derived Melan-A-specific T cell clones by flow FISH. A, Analysis of telomere fluorescence was performed on 10 in vitro-generated T cell clones derived from either naive or EM28⁻ Melan-A-specific T subset at various time points following serial peptide vaccination. The average telomere fluorescence (mean ± SD) for each subset and time point was converted to kilobase as described in Materials and Methods. B, Progressive loss of telomere fluorescence in EM28⁻-derived T cell clones along vaccination.