In Vivo Persistence of Codominant Human CD8?T Cell
Clonotypes Is Not Limited by Replicative Senescence or
Laurent Derre ´,2* Marc Bruyninx,2*†Petra Baumgaertner,* Estelle Devevre,*
Patricia Corthesy,†Ce ´dric Touvrey,* Yolanda D. Mahnke,‡Hanspeter Pircher,§
Verena Voelter,¶Pedro Romero,* Daniel E. Speiser,2* and Nathalie Rufer2,3†
T cell responses to viral epitopes are often composed of a small number of codominant clonotypes. In this study, we show that
tumor Ag-specific T cells can behave similarly. In a melanoma patient with a long lasting HLA-A2/NY-ESO-1-specific T cell
response, reaching 10% of circulating CD8 T cells, we identified nine codominant clonotypes characterized by individual TCRs.
These clonotypes made up almost the entire pool of highly differentiated effector cells, but only a fraction of the small pool of less
differentiated “memory” cells, suggesting that the latter serve to maintain effector cells. The different clonotypes displayed full
effector function and expressed TCRs with similar functional avidity. Nevertheless, some clonotypes increased, whereas others
declined in numbers over the observation period of 6 years. One clonotype disappeared from circulating blood, but without
preceding critical telomere shortening. In turn, clonotypes with increasing frequency had accelerated telomere shortening, cor-
relating with strong in vivo proliferation. Interestingly, the final prevalence of the different T cell clonotypes in circulation was
anticipated in a metastatic lymph node withdrawn 2 years earlier, suggesting in vivo clonotype selection driven by metastases.
Together, these data provide novel insight in long term in vivo persistence of T cell clonotypes associated with continued cell
turnover but not replicative senescence or functional alteration. The Journal of Immunology, 2007, 179: 2368–2379.
Recognition of such cells relies on the specific interaction of
clonally distributed TCRs on effector lymphocytes with Ag-de-
rived peptides bound to self-MHC molecules on infected or ma-
lignant cells (1). The introduction of fluorescent multimers of
MHC/peptide that bind stably to specific TCR on the surface of T
cells has enabled detailed TCR repertoire analysis of Ag-specific T
cells isolated ex vivo (2, 3).
The CD8?T cell responses to primary virus infections are char-
acterized by large expansions of activated T cell clones bearing
particular TCRs (4). In humans, highly restricted TCR usage has
been described in several viral systems, including influenza (5, 6),
ytolytic CD8?T lymphocytes play an important role in
adaptive immunity and are responsible for specific rec-
ognition and elimination of infected or transformed cells.
EBV (7, 8), CMV (9), and HIV-1 (10–12). Thus, the limited TCR
diversity seems to be a conserved feature of CD8?T cell responses
to viral infection. Our group recently identified a naturally primed
T cell clone that dominated the human CD8?T cell response to the
Melan-A/MART-1 tumor Ag (13). Taken together, our data and
those reported by others (reviewed in Ref. 14) indicate that similar
to T cell responses against immunodominant viral Ags, selection
and amplification of tumor-specific T cell clones occurs in cancer
Long-term persistence of clonally restricted CD8?T cell ex-
pansions has been observed in chronic viral infections such as
EBV (15, 16), HSV (17), or HIV (11, 18, 19). In line with these
studies, individual T cell clonotypes expressing high avidity TCRs
to cognate tumor Ag were detected in melanoma patients with
favorable disease outcome (20), as well as in a patient vaccinated
repeatedly with Melan-A26–35peptide mixed with IFA and CpG
7909 over extended periods of time (13). Despite major progress in
the analysis of Ag-specific T cells, however, in vivo TCR reper-
toire evolutions in response to chronic antigenic exposure, e.g., in
viral infection (EBV, HIV) or in the tumor-bearing state, remain
largely unexplored. Furthermore, most of these studies provide
limited information concerning the factors controlling turnover
and persistence of particular T cell clonotypes. These questions are
fundamental to our understanding of protective immunity and have
important implications for vaccine design and development.
In the present study, we explored a natural tumor-specific im-
mune response against the cancer testis Ag NY-ESO-1, subse-
quently boosted by peptide vaccination in a melanoma patient. To
study and define the molecular evolution of epitope-specific CD8?
T lymphocytes, we used a novel ex vivo molecular-based approach
at the single cell level (13). We identified nine distinct and codomi-
nant T cell clonotypes bearing BV1, BV8, or BV13 TCRs. Over a
period of several years, we observed changes in frequencies of
*Division of Clinical Onco-Immunology, Ludwig Institute for Cancer Research, Uni-
versity Hospital of Lausanne, Lausanne, Switzerland;†Swiss Institute for Experimen-
tal Cancer Research, Epalinges, Switzerland;‡Vaccine Research Center, National
Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda,
MD 20892;§Institute for Medical Microbiology and Hygiene, Department of Immu-
nology, University of Freiburg, Freiburg, Germany; and¶Multidisciplinary Oncology
Center, Centre Hospitalier Universitaire Vaudois (CHUV), Lausanne, Switzerland
Received for publication March 26, 2007. Accepted for publication May 30, 2007.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1This study was sponsored and supported by the Swiss National Center of Compe-
tence in Research (NCCR) Molecular Oncology, the Ludwig Institute for Cancer
Research, the Cancer Research Institute, NY, the Swiss Cancer League/Oncosuisse
Grant 01323-02-2003, and the Swiss National Science Foundation Grants 3200B0-
107693 and 3100A0-105929.
2L.D., M.B., D.S., and N.R. contributed equally to this work.
3Address correspondence and reprint requests to Dr. Nathalie Rufer, Swiss Institute
for Experimental Cancer Research, 155 ch. des Boveresses, Epalinges, Switzerland.
E-mail address: Nathalie.Rufer@isrec.ch
Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00
The Journal of Immunology
clonotypes with an expansion of BV13 T cells, and decrease and
even disappearance of BV8 subpopulations. Clonotypic fluctua-
tions were concomitant to vaccination with NY-ESO-1 analog
peptide. We quantified differentiation, proliferative potential, and
ability to produce effector mediators and cytokines. Because our
analysis was performed on individual T cell clonotypes, it pro-
vides, for the first time, a detailed insight in factors associated with
persistence and survival of tumor-specific T cell subpopulations.
Materials and Methods
Patient LAU 50, HLA-A*0201 positive, was diagnosed at the age of 62
years with primary skin melanoma of the right leg in January 1992.
Breslow tumor thickness was 4.3 mm. In the following year he developed
multiple skin metastases in the right leg and was treated with isolated limb
perfusion with melphalan, TNF-?, and IFN-?. Five inguinal lymph nodes
were removed and found tumor free. Two years later he developed a single
inguinal lymph node metastasis, and inguinocrural lymph node dissection
revealed that 13/14 nodes were tumor free. Eight years later (in November
2003) he had a contralateral metastasis (i.e., left inguinal), which was re-
moved in January 2004 (?3 mo before the start of vaccination). Similar to
the previous lesions, this metastasis expressed multiple tumor Ags, i.e.,
Melan-A, gp100, Tyr, Mage-1, Mage-3, Mage-10, Lage-1, NY-ESO-1,
SSX-2, and SSX-4. Starting April 2004 (D0), the patient was enrolled in
the Ludwig Institute vaccination trial LUD 01-003 (21) and received 12
monthly s.c. vaccines (until ?13 mo) composed of 3 ? 500 mg of peptide
(NY-ESO-1157–165 SLLMWITQA C165A analog, Mage-A10254–262
GLYDGMEHL and Melan-A26–35ELAGIGILTV A27L analog), emulsi-
fied in 1 ml of Montanide ISA-51, prepared altogether in one syringe as a
stable emulsion (2-ml injection volume). Peptides were provided by
ClinAlfa, MerckBiosciences, La ¨ufelfingen, Switzerland, and adjuvant IFA;
Montanide ISA-51 was provided by Seppic. Increased frequencies of
Mage-A10 Ag-specific CD8?T lymphocytes were observed in patient
LAU 50 upon vaccination, with values reaching up 0.16% of CD8?T cells
(21). Future studies involve the molecular characterization of such tumor-
specific T cells. In July 2004, one metastatic node in the right calf had
regressed and was no longer detectable by radiological imaging and
positron-emission tomography. The patient was thus in complete remission
and remained so during the study, until 18 mo later (September 2005) when
several new metastases developed in the right lower limb. All studies have
been reviewed and approved by an appropriate institutional review
A2/peptide multimers and flow cytometry immunofluorescence
PBMCs were obtained by density centrifugation using Ficoll-Hypaque
(Pharmacia Biotech) and cryopreserved in RPMI 1640 supplemented with
40% FCS and 10% DMSO (1 ? 107–4 ? 107cells per vial). Synthesis of
PE- and allophycocyanin-labeled HLA-A*0201/peptide multimers were
prepared as described previously (2, 3) with NY-ESO-1 analog peptide
SLLMWITQA. Five color stains were done with PE- or allophycocyanin-
HLA-A2/peptide multimers, FITC-conjugated anti-CD28, -CD27, -CD57,
and -programmed death (PD)-1 (BD Biosciences), PE-conjugated
anti-TCR BV8 and BV13 (Beckman Coulter) and anti-CD127 (BD Bio-
sciences), PE-Texas Red-conjugated anti-CD45RA (Beckman Coulter), al-
lophycocyanin/Cy7-conjugated anti-CD8 (BD Biosciences) reagents, and
anti-CCR7 mAb (BD Biosciences) followed by allophycocyanin-conju-
gated goat anti-rat IgG Ab (Caltag Laboratories). In brief, CD8?T lym-
phocytes 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- or allophycocyanin-labeled
multimers for 30 min at 4°C in PBS, 0.2% BSA, 50 ?M EDTA, and then
with appropriate Abs (30 min, 4°C). Intracellular content of granzyme B,
perforin, and Ki-67 was measured in CD8?T lymphocytes without pre-
vious stimulation. After staining with appropriate mAbs, cells were fixed
for 20 min at room temperature in PBS containing 1% formaldehyde, 2%
glucose, and 5 mM sodium azide. Fixation was followed by permeabili-
zation with PBS/0.1% saponin (Fluka)/0.2% BSA/50 ?M EDTA and stain-
ing with granzyme B-FITC (Ho ¨lzel Diagnostika), perforin-FITC mAbs
(Alexis) or Ki-67-FITC (BD Biosciences), both for 20 min at room tem-
perature. Cells were immediately analyzed on a BD Vantage or a LSRII
flow cytometer using CellQuest software (BD Biosciences). KLRG1 ex-
pression was determined by Alexa 488-conjugated anti-KLRG1 mAb
cDNA amplification, TCR spectratyping, sequencing, and
Five-cell aliquots were sorted with a FACSVantage SE machine directly
into wells of 96 V-bottom plates. cDNA preparation, cDNA amplification,
and PCR were performed as described (23). CDR3 size of TCR transcripts
and the sequences of oligonucleotides corresponding to the 22 variable
segments of the TCR ?-chain (based on the nomenclature proposed by
Arden et al. (24)) were analyzed as follows: in brief, 8 ?l taken from 10
individually sorted and amplified 5-cell cDNA samples were pooled to-
gether to obtain total cDNA material equivalent to ex vivo sorted 50 cells.
cDNA pools generated from circulating EM28?and EM28?NY-ESO-1-
specific T cells at different time points before and after vaccination as well
as cDNA pools from NY-ESO-1-specific T cells sorted from single-cell
suspension of a metastatic lymph node tissue (TILN)4were subjected to
individual PCR using a set of validated 5? sense fluorescent-labeled prim-
ers specific for the 22 BV subfamilies and one 3? antisense primer specific
for the corresponding C gene segment (25). 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 BV-BC PCR products were directly purified and se-
quenced (Fasteris) when single dominant PCR peaks were identified. Two
distinct sets of primers (Metabion) specific for the CDR3 region of each
identified BV1, BV8, and BV13 T cell clonotype were validated and used
for clonotyping PCR as recently described (13); 1) CDR3 clonotype for-
ward and C? reverse, and 2) BV-subfamilies (BV1, BV8, or BV13) and
CDR3 clonotype reverse. Forward and reverse clonotype primers are de-
picted as following; BV1 clonotype 1: 5?-AGCGTAACAGGGACAGGG
G-3?; rev-5?-GCCCCCTGTCCCTGTTACG-3?, BV1 clonotype 2: 5?-GTA
C-3?, BV8 clonotype 1: 5?-ACTTCTGTGCCAGCCAACAG-3?; rev-5?-A
AAGCTTCAGTACCCCCCTG-3?, BV8 clonotype 2: 5?-ACTTCTGTGC
clonotype 3: 5?-GTTTGGGGGGCAATGAGCAG-3?; rev-5?-GAACTGC
TCATTGCCCCCCA-3?, BV13 clonotype 1: 5?-GAACAGGGTTGGACG
GCTAC-3?; rev-5?-GTAGCCGTCCAACCCTGTTC-3?, BV13 clonotype
2: 5?-AGTTACGTAGGGGGGAAGG-3?; rev-5?-AGCCTTCCCCCCTAC
GTAA-3?, and BV13 clonotype 3: 5?-GACACTATAATTCACCCCTCC-
Gene expression analysis
The procedures for cDNA preparation, cDNA amplification as well as the
RT-PCR were recently described in detail (23). Primers to detect CD3,
CCR7, IFN-?, granzyme B, perforin, CD94, and TNF-? mRNA transcripts
were previously reported (26). Additional primers were used in the present
TCTTCTAGTTGCTGAGGAAACG-3?, KLRG1: 5?-CTTGAGCCCAGG
T cell cloning and culture
Multimer?CD8?T cell subsets (EM28?and EM28?) were sorted by flow
cytometry, cloned by limiting dilution, and expanded in RPMI 1640 me-
dium supplemented with 8% human serum, 150 U/ml recombinant human
IL-2 (rhIL-2; a gift from GlaxoSmithKline), 1 ?g/ml PHA (Sodiag) and
1 ? 106/ml irradiated allogeneic PBMC (3000 rad) as feeder cells.
Telomere fluorescence in situ hybridization and flow cytometry
The average length of telomere repeats at chromosome ends in individual
cells, was measured by fluorescence in situ hybridization (FISH) and flow
cytometry (flow FISH) as previously reported (27, 28). Telomere length
measurements were performed on 45 in vitro derived T cell clones sorted
from EM28?NY-ESO-1-specific T subpopulations at various time points
before and 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 stimu-
lation, 2 ? 105cells 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 clone. We estimated that all clones had, on average,
shortened their telomere lengths by ?1.5 kb through their in vitro expan-
sion round (29). Telomere fluorescence was calculated by subtracting the
mean fluorescence of the background control (no probe) from the mean
4Abbreviations used in this paper: TILN, tumor-infiltrated lymph node cells; PD,
programmed death; int, intermediate; EM, effector-memory; CM, central-memory.
2369The Journal of Immunology
fluorescence obtained from cells hybridized with the telomere probe after
calibration with FITC-labeled fluorescent beads (Quantum TM-24 Pre-
mixed; 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: bp ? MESF ? 0.495 (27).
Ex vivo lytic activity and Ag recognition was assessed as recently de-
scribed (30) with some modifications. In brief, peptide-pulsed T2 cells
(HLA-A2?/TAP?/?) were labeled with 0.1 ?M CFSE (NY-ESO-1 pep-
tide; T2-CFSElow) or with 2 ?M CFSE (irrelevant peptide; T2-CFSEhigh).
A 1/1 mixture of T2-CFSElowand T2-CFSEhighwas prepared. BV8?and
BV13?NY-ESO-1-specific CD8?T lymphocytes were sorted as described
above, and increasing numbers of sorted cells (63, 125, 250, 500, 1000, and
2000) were dispensed into a plate containing the peptide-pulsed T2 cells
(125 cells per peptide). The percentage of specific lysis was then calculated
as described by Devevre et al. (30).
IFN-? Cytospot assay
Measurement of intracellular IFN-? production was combined to multimer,
CD28 and CD45RA labeling. 1 ? 106CD8?enriched T cells (Miltenyi
Biotec) were incubated for 5 h with 1 ? 106T2 cells pulsed with 10 ?g/ml
irrelevant HIV-1 Pol476–484(ILKEPVHGV) peptide, 10 ?g/ml cognate
peptide, or 1 ?g/ml PMA/0.25 ?g/ml ionomycin, respectively. After 1 h,
10 ?g/ml brefeldin A (Sigma-Aldrich) was added. After 4 additional hours,
cells were stained with multimers and Abs, fixed, permeabilized, and in-
cubated with anti-IFN-?-FITC in PBS/0.1% saponin for 30 min at 4°C.
Cells were analyzed on a LSRII flow cytometer using CellQuest software
CFSE proliferation assay
We incubated 2 ? 106CFSE-labeled PBMCs per ml in the presence of 10
?g/ml cognate peptide in RPMI 1640 containing 10% human serum and
150 U/ml IL-2 for 5 days. After day 3, 4, and 5, we collected ?2 ? 106
cells and stained cells with multimers and Abs as described above. Cells
were analyzed on a LSRII flow cytometer using CellQuest software (BD
Dominant NY-ESO-1-specific CD8?T cell clones expand and
persist at high frequency ex vivo
Patient LAU 50 with advanced melanoma remained disease free
for a period of 9 years before developing two new metastases, one
of which was removed by surgery. Subsequently, the patient was
enrolled in a vaccination trial and received monthly vaccinations
with NY-ESO-1157–165C165A analog peptide emulsified in IFA
during 13 mo (21). During the first 3 mo of vaccination, the
remaining metastasis regressed and was no longer detectable by
radiological imaging and positron-emission tomography (see
patient’s clinical history in Materials and Methods). Using fluo-
rescent HLA-A2/peptide multimers incorporating peptide NY-
ESO-1157–165(thereafter NY-ESO-1 multimers), we identified a
high frequency (?3%) of NY-ESO-1 multimer?T cells that were
already detectable in the circulating CD8?compartment 8 mo be-
fore the relapse of disease (Fig. 1A). This population expanded up
to 10% of CD8?T cells during vaccination. Of note, the high
frequency values of NY-ESO-1-specific T cells observed in this
particular patient represent a rare exception, because in most HLA-
A2?individuals cancer testis Ag-specific T cells are not detectable
ex vivo (31). Multicolor flow cytometry analysis revealed that NY-
ESO-1-specific T lymphocytes predominantly bore a differentiated
effector-memory phenotype (EM28?; CD45RA?CCR7?CD28?),
which persisted for at least 18 mo following immunotherapy (Fig.
1B; data not shown). The majority of these cells expressed intra-
cellular granzyme B and had down-regulated CD27. In contrast, a
small but yet detectable proportion (?7%) of the effector memory
compartment contained CD28?cells (designed as EM28?).
In several viral systems, responses to a given T cell-defined Ag
often showed strong TCR selection resulting in highly restricted
TCR usage (7–12, 32). Our finding that the robust NY-ESO-1-
specific T cell response is characterized by increased frequencies
and differentiation into EM28?cells expressing multiple effector
mediators suggests a similar occurrence of oligoclonal expansions
of tumor-reactive T lymphocytes. To address this point, the clonal
composition of NY-ESO-1 multimer?EM28?and EM28?ex
vivo sorted fractions was analyzed by spectratyping (Fig. 1C),
which measures both TCR-?-chain variable segment usage (BV)
and CDR3 length heterogeneity as described recently (13). Primed
EM28?cells displayed large polyclonal TCR repertoires with a
diverse usage of the 22 different BV families as well as high vari-
ability within CDR3 size products. In sharp contrast, EM28?T
cells exhibited a restricted TCR repertoire diversity with the ma-
jority of cells using BV1, BV8, or BV13 gene segments of defined
CDR3 lengths. These dominant TCRs were conserved over the
period during which the patient received immunotherapy (0–18
mo). TCR sequencing revealed the presence of nine distinct clono-
types; three for each of the identified TCR-BV (Table I).
Evolution of the T cell clonotypic composition during disease
progression and recall responses to vaccination
To gain insight into the kinetics of each distinct BV1, BV8, and
BV13 T cell clonotype over time, we first examined the proportion
of the different clonotypes among ex vivo purified EM28?and
EM28?NY-ESO-1-specific T cell subsets. To directly assess the
presence or absence of particular clonotypes, we used a modified
RT-PCR protocol that detects specific cDNAs after global ampli-
fication of expressed mRNAs from as few as five cells (23) and
combined it with designed clonotypic primers (13). Transcript
analysis revealed the presence of TCR-BV8 and -BV13 clonotypes
1 and 2 in the majority of the EM28?and EM28?characterized
5-cell aliquots, and these represented the most abundant clono-
types (Fig. 1, D and E; data not shown). BV8-clono1 and BV8-
clono2 were dominant within blood recovered 11 mo before
vaccination. The ratio BV8:BV13 was reversed 18 mo after
immunotherapy, with an increase in the proportion of 5-cell
samples positive for BV13 TCR clonotypes. Most strikingly,
we observed the complete disappearance of BV8-clono2 TCR at
the latest time point.
Increased proportions of BV13 T cell clonotypes after
Our strategy combining ex vivo cell sorting with molecular anal-
ysis of 5-cell T lymphocytes provides the identification of specific
T cell clonotypes as well as insight in TCR repertoire evolution
over time. However, the 5-cell approach does not allow a precise
estimate of T cell frequencies. Therefore, we generated 480 T cell
clones derived by in vitro limiting dilution from circulating tumor-
specific EM28?and EM28?T cell subsets isolated at various time
points. The TCR of each T cell clone was analyzed by sequencing
and/or specific clonotypic PCR analysis. In agreement with the ex
vivo 5-cell data, we found that over 50% of the EM28?NY-ESO-
1-specific T cell clones were composed of TCR BV8 clonotypes
before the start of immunotherapy, whereas BV13 and BV1 clono-
types, represented 28.6 and 7.6% of the repertoire, respectively
(?11 mo; Fig. 2A). Following vaccination, the proportion of cells
bearing the TCR BV13 clonotypes gradually increased to 42.5%
before reaching up to ?60% at 13 and 18 mo. In sharp contrast, the
fraction of BV8-clonotypic T cell clones declined by 13 mo after
treatment (23%), and we could no longer detect the BV8-clono2 at
this time point (Table II). Similar shifts in the proportion of BV8
and BV13 clonotypes were found in the EM28?subset (Fig. 2A;
Table II). Moreover, the proportion of other and infrequent TCRs
was greater within the latter subset (43–58%) than among the
2370 DYNAMICS OF TUMOR-REACTIVE CD8?T LYMPHOCYTE RESPONSES EX VIVO
EM28?cells (5–10%). The proportions of NY-ESO-1-specific
BV8 and BV13 T cell clonotypes 1 and 2 were calculated as per-
centages of circulating CD8?T lymphocytes (Fig. 2B). A drastic
increase in frequencies of both BV13-clono1 and BV13-clono2 T
cells was observed during immunization, reaching up to 3.1 and
2.2%, respectively. In contrast, the proportion of BV8-clono1 and
BV8-clono2 T cells diminished over time, which became particu-
larly evident at 18 mo. Similar kinetics were observed when the
proportion of individual BV8 and BV13 clonotypes was adjusted
to the total counts of leukocytes (data not shown).
Finally, we also used mAbs directed against the variable re-
gion of TCR BV8 and BV13 to sort 5-cell aliquots from
BV8highand BV8?, as well as BV13high, BV13int, and BV13?
NY-ESO-1-specific EM28?T cells ex vivo (Fig. 2C). A con-
venient feature of the anti-BV13 mAb reactivity allowed us
to clearly discriminate between bright cells that exclusively
subsets. A, Percentage of multimer?cells of circulating CD8?T cells over time. Occurrence of two metastases (?3 mo) and regression of one
remaining after surgery (?3 mo) are indicated by filled circles and an opened circle, respectively. The time period of 12 monthly vaccinations is
indicated. B, For characterization of T cell differentiation, CD8?NY-ESO-1?T cells (R1 gated) were analyzed ex vivo by flow cytometry for their
cell surface expression of the tyrosine phosphatase CD45RA and the homing chemokine receptor CCR7. The proportion of CD28-, CD27-, perforin-,
and granzyme B?cells among multimer?CD45RA?CCR7?(EM gated) specific T cells was determined by immunofluorescence (?11 mo; black
histograms). The reference for CD28 and CD27 negativity, and granzyme B and perforin positivity is based on the signal obtained after gating on
bulk CD8?CD45RA?CCR7?naive T cells known to be CD28?CD27?perforin?granzyme B?cells (open histogram). C, cDNA pools (50 cells)
generated from EM28?and EM28?NY-ESO-1-specific T cells sorted from circulating CD8?T cells at different time points were amplified by PCR
using 22 BV-specific primers 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), and displayed by black squares on a grid. The columns corresponding to the
BV1, BV8, and BV13 subfamilies are labeled. The figures inserted in each grid indicate the total number of BV subfamily gene segment usage vs
the total number of all amplified BV-CDR3-BC products within each positive BV subfamily. D and E, Unique primers corresponding to the CDR3
gene segment of each identified TCR clonotype were designed and validated (see Materials and Methods). Clonotypic PCR was performed on cDNA
obtained from individually sorted 5-cell samples of NY-ESO-1-specific EM28?(D; #1–20) and EM28?(E; #21–40) CD8?T cells isolated before
(?11 mo) and after (?18 mo) vaccination (n ? 10). Samples yielding detectable clonotypic-specific signals are depicted in color?. Gene expression
patterns of global TCRs BV8 and BV13 are also shown. BV8-clono1, BV8 clonotype 1; BV8-clono2, BV8 clonotype 2; BV13-clono1, BV13
clonotype 1; BV13-clono2, BV13 clonotype 2. The asterisk (?) corresponds to samples expressing nonclonotypic BV8 or BV13-TCR gene segments.
As this was mostly observed in EM28?cells, these data confirm increased TCR repertoire heterogeneity in EM28?cells compared with the EM28?
Ex vivo analysis of T cell frequency, phenotype, TCR-BV chain repertoire and TCR-BV clonotypes of circulating NY-ESO-1-specific T
Table I. TCR-BV usage of A2/NY-ESO-1157–165specific T cell
clonotypes from patient LAU 50
BV13S1 CAS RTGLDGY
BV13S1 CAS SYVGGKAEA
BV13S6 CAS SLTGHYNSPL
aThe TCR-BV nomenclature used was according to Arden and colleagues (24).
bClonotypes recently identified by Le Gal et al. (38).
2371The Journal of Immunology
comprised the BV13-clonotype 2 and dim cells selectively con-
taining the BV13-clonotype 1 (Table III). Moreover, all BV8-
clono1 and BV8-clono2 were comprised within the BV8?cells
whereas none were detectable in the BV8?fraction of the cells.
The proportion of BV8-clono1/clono2, BV13-clono1 and BV13-
clono2 T cells represented 44, 29, and 1.2%, respectively, of total
NY-ESO-1-specific EM28?T cells (Fig. 2C). These results are in
excellent agreement with the data from the T cell clones (Fig. 2A;
Table II), indicating that our in vitro T cell cloning approach did not
introduce major biases and can thus be efficiently used to assess
clonotype frequencies among Ag-specific CD8?T cells.
Predominance of BV13 T cell clonotypes in a metastatic lymph
We next evaluated the TCR usage of NY-ESO-1-specific T cell
subsets within a metastatic lymph node, resected 3 mo before the
start of peptide vaccination (?3 mo). Most tumor-specific T cells
exhibited the effector-memory CD45RA?CCR7?phenotype, with
a dominant fraction of these cells that had down-regulated CD28
and CD27, and up-regulated granzyme B and perforin (Fig. 3A).
As their counterparts in peripheral blood samples, TCR-BV di-
versity in the multimer-specific EM28?T cell subpopulation
was performed on T cell clones in vitro generated from circulating tumor-specific EM28?(n ? 122) and EM28?(n ? 324) T cell subsets at various
time points (?11 mo to ?18 mo). The same set of clonotypic primers as described in Materials and Methods was used. Results are presented as
percentage of BV8 (red), BV13 (blue), and BV1 (green) T cell clonotypes, and of nonclonotypic (yellow) T cells in EM28?and EM28?tumor-
specific T cells, respectively (see Table II). B, Percentage of total BV8-clono1, BV8-clono2, BV13-clono1, and BV13-clono2 NY-ESO-1-specific
T cells among the circulating CD8?compartment over time. C, The proportion of BV8?and BV13?cells within EM28?and EM28?
multimer?CD45RA?CD8?T cells was determined by immunofluorescence (at time point 11 mo before the start of immunotherapy). Of note,
EM28?and EM28?T cells comprising low and high staining intensities were observed when using the anti-BV13 mAb.
Quantification of distinct TCR-BV1, -BV8, and -BV13 NY-ESO-1-specific T cell clones using clonotypic primers. A, Clonotypic PCR
Table II. Estimated proportions of TCR-BV1, -BV8 and -BV13 clonotypes among NY-ESO-1-specific T cell subsets over time and
in a metastatic lymph node
BV1 ClonotypingBV8 Clonotyping BV13 Clonotyping
Others Clono1 Clono2Clono3 Clono1Clono2 Clono3 Clono1Clono2 Clono3
aTotal number of in vitro generated T cell clones analyzed for their TCRs by sequencing and/or clonotyping.
bProportion of each distinct BV1, BV8, and BV13 clonotype in percentage. na, not applicable. TILN (?3 mo). ?11 mo to ?18 mo represents blood
samples retrieved at various time points before and after vaccination.
2372 DYNAMICS OF TUMOR-REACTIVE CD8?T LYMPHOCYTE RESPONSES EX VIVO
was limited with preferential presence of BV1-, BV8-, and
BV13-expressing cells (Fig. 3B). Again, the TCR repertoire in
the EM28low/?fraction was more diverse than in the EM28?
subset. We further investigated the proportion of each T cell
clonotype by ex vivo sorting of 5-cell samples (Fig. 3C) and
analysis of in vitro generated T cell clones (n ? 70; Fig. 3D).
All of the clonotypes identified within the circulating NY-ESO-
1-specific CD8?T cells were also found in the metastatic tis-
sue. In contrast to the data obtained from peripheral blood (?11
mo; Fig. 2A), the BV13 clones were predominant in both
EM28?and EM28?T cell subpopulations, whereas BV8 T cell
clonotypes were represented at reduced frequencies (Table II).
This dominance was particularly marked within the EM28?
compartment. Altogether, these results provide molecular evi-
dence that the prevalence for BV13 clonotypes observed within
multimer-specific CD8?T lymphocytes of a metastatic lymph
node resected 3 mo before the start of vaccination (?3 mo)
precedes the BV13? ?BV8 ratio attained about 2 years later in
the peripheral blood (?18 mo; Fig. 2A).
Rapid telomere shortening within BV13 T cell clonotypes
contrasts with stabilized telomeres in BV8 clonotypes over time
We next measured the turnover of distinct NY-ESO-1-specific T
cell clonotypes before and after immunotherapy (Fig. 4A). Because
telomeres progressively shorten as a function of cell division, telo-
mere length is a powerful indicator of the in vivo replicative his-
tory of lymphocytes (27). We observed a drastic reduction in the
mean telomere fluorescence of BV13 T cell clonotypes over time,
that corresponded to a loss of ?2.4 kb, indicating extensive in vivo
proliferation. Telomere shortening seemed coincident with immu-
notherapy, because BV13 clonotypes from the metastatic lymph
subsets from metastatic tissue. A, An inguinal metastatic lymph node (TILN) was surgically removed 3 mo before immunotherapy, and CD8?NY-
ESO-1?-specific T cells (R1 gated) were characterized ex vivo by flow cytometry for their cell surface expression of CD45RA and CCR7. The
proportion of CD28-, CD27-, granzyme B-, and perforin-positive cells among multimer?CD45RA?CCR7?(EM gated)-specific T cells was deter-
mined by immunofluorescence. The reference for CD28 and CD27 negativity, and granzyme B and perforin positivity is based on the signal obtained
after gating on bulk CD8?CD45RA?CCR7?naive T cells (dotted line). B, cDNA pools (50 cells) generated from EM28lowand EM28?NY-ESO-
1-specific T cells sorted from freshly prepared single-cell suspensions of the metastatic TILN were amplified by PCR using 22 BV-specific primers
and subjected to electrophoresis on an automated sequencer. The columns corresponding to the BV1, BV8, and BV13 subfamilies are labeled. Total
numbers of TCR-BV usage (x-axis) comprising all amplified BV-CDR3-BC size product (y-axis) are indicated in each grid. C, Clonotypic PCR was
performed on cDNA obtained from individually sorted 5-cell samples of NY-ESO-1-specific EM28?(#41–50) and EM28low(#51–60) CD8?T cells
isolated from the metastatic TILN (n ? 10). Samples yielding detectable clonotypic specific signals are depicted in color (?). Gene expression
patterns of global TCRs BV8 and BV13 are also shown. The asterisk (?) corresponds to samples expressing nonclonotypic BV8 or BV13-TCR gene
segments. D, Clonotypic PCR was performed on in vitro generated T cell clones from EM28lowtumor-specific T cells isolated from TILN (n ? 53).
Results are presented as percentage of BV8 (red), BV13 (blue), and BV1 (green) T cell clonotypes, and of nonclonotypic (yellow) T cells. The same
set of clonotypic primers as described in Materials and Methods was used.
Ex vivo analysis of T cell frequency, phenotype, TCR-BV chain repertoire, and TCR-BV clonotypes of NY-ESO-1-specific T cell
Table III. Ex vivo estimated proportions of TCR-BV8 and -BV13 clonotypes among BV8- and BV13-positive
NY-ESO-1-specific EM28?sorted 5-cell samplesa
BV1 Expression BV8 Expression BV13 Expression
aAll shown data are from PBMC collected from a lymphocytapheresis withdrawn 11 mo before vaccination. nd, not
determined. Of note, most TCR-BV1 expressing T cells were detected in the sorted BV8- and BV13-negative fraction of the
bProportion of positive TCR-BV1, -BV8, or -BV13 subfamilies per 5-cell aliquots.
cProportion of positive clonotypes per 5-cell aliquots.
2373The Journal of Immunology
node (?3 mo) displayed similar average telomere lengths com-
pared with the clones isolated from the earliest blood sample be-
fore vaccination (?11 mo). Despite its subsequent disappearance,
the BV8-clono2 displayed the brightest telomere signal, and no
telomere loss was observed within the BV8-clonotype 1 subpopu-
lation over time. Altogether, our data indicate that several tumor-
specific T cell clones may persist over extended periods of time in
vivo, likely reflecting the repetitive triggering by Ag derived from
tumor cells or vaccination. Others eventually disappear from the
blood, but this is not associated with a state of replicative
EM28?T cell clonotypes mediate efficient ex vivo killing,
produce IFN-? and retain proliferative capacity upon antigenic
The finding of distinct kinetics among NY-ESO-1-specific BV
subpopulations over time prompted us to assess their ex vivo cy-
tolytic activity using a novel flow cytometry-based cytotoxic assay
(30). As depicted in Fig. 4B, BV8?and BV13?NY-ESO-1-spe-
cific T lymphocytes efficiently and similarly killed NY-ESO-1-
peptide pulsed T2 cells. These data are in agreement with chro-
mium release assays performed with EM28?derived BV8?and
BV13?T cell clones (n ? 103), where we found similar efficiency
by the different clonotypes to recognize NY-ESO-1 expressing au-
tologous tumor cells and T2 cells labeled with titrated amounts of
NY-ESO-1 peptides (Fig. 4C; data not shown). Efficiency of target
cell lysis apparently remained stable over time, as we obtained
similar results with clones of BV8 and BV13 T cell clonotypes 1
and 2 generated from blood samples retrieved before (?11 mo) or
after (?18 mo) immunotherapy (Fig. 4C; data not shown).
We next investigated whether BV8 and BV13 T cell clonotypes
differed in the expression of molecules involved in T cell effector,
survival, or regulatory functions. A similar proportion of 5-cell
samples containing ex vivo detectable IFN-?, TNF-?, granzyme B,
cell clones derived from circulating EM28?NY-ESO-1-specific T cells isolated at different time points and from TILN. Each T cell clone was
assessed for its TCR clonotype by clonotypic PCR. As several clones were tested per time point, tissue (PBMC and TILN), and clonotype, a
representative example of each was chosen (total of 41 tested clones). The dotted line was arbitrarily set at the mean telomere signal obtained for
BV8-clonotype 1 (?11 mo) and allows the direct comparison between samples. The mean average telomere fluorescence from several analyzed T
cell clonotypes is depicted in kilobase (kb). Of note, telomere fluorescence of BV13-clono2 (?11 mo) and BV8-clono2 (?8 mo) T cell clonotypes
could not be assessed, as no (BV8-clono2) or only limited numbers (BV13-clono2) of T cell clones could be obtained due to their low frequency
within the EM28?compartment (see Table II). One experiment of two is shown. B, Ex vivo assessment of cytotoxicity using the LiveCount assay
(30). Increasing numbers of total, BV8?or BV13?NY-ESO-1-specific CD8?T cells (?11 mo) were sorted and coincubated with an equal number
of NY-ESO-1 peptide pulsed T2-CFSElowand irrelevant (ir) peptide pulsed T2-CFSEhighcells at the indicating E:T cell ratios. C, The relative TCR
avidity was compared using T2 target cells (HLA-A2?/TAP?/?) pulsed with graded concentrations of either analog NY-ESO-1157–165peptide
(SLLMWITQA; filled squares) or native NY-ESO-1157–165peptide (SLLMWITQC; unfilled diamonds). Representative examples of percentage of
specific killing by BV8- and BV13-clonotype 1 T cells before (?11 mo) and after (?18 mo) peptide vaccination are depicted (n ? 103 clones).
Similar data were obtained for BV8- and BV13-clonotype 2 T cells (data not shown).
Analysis of replicative history and ex vivo cytotoxicity. A, Analysis of telomere fluorescence was performed by flow FISH on 45 T
2374 DYNAMICS OF TUMOR-REACTIVE CD8?T LYMPHOCYTE RESPONSES EX VIVO
perforin, and C-type killer cell lectin-like receptor CD94 tran-
scripts, was found in both BV clonotypes (Fig. 5A; data not
shown). These clones displayed a highly differentiated phenotype,
because a majority of the cells expressed CD57 whereas having
down-regulated CD127 (IL-7R?), CD27, and L-selectin (CD62-L)
expression (Fig. 5B; data not shown). Moreover, BV8?and
BV13?NY-ESO-1-specific T cells expressed similar levels of
PD-1. Surprisingly, the number of samples positive for KLRG1
mRNA, another killer cell lectin-like receptor, was much higher in
both BV13-1 and BV13-2 clonotypes than in BV8 clonotypes (Fig.
5A), correlating with the analysis of KLRG1 protein expression by
FACS (Fig. 5B). Indeed, the majority of NY-ESO-1-specific
BV13?cells expressed KLRG1 (?80%), whereas only 15–20% of
BV8 clonotypes expressed the protein. Similar profiles of PD-1
and KLRG1 protein expression were observed 8 mo after immu-
notherapy (data not shown).
A significant proportion of NY-ESO-1-specific T cells produced
IFN-? after short-term antigenic challenge as well as after non-
specific PMA/ionomycin stimulation (Fig. 5C). As observed pre-
viously, most of the IFN-?-secreting cells stimulated by the cog-
nate peptide were differentiated EM28?cells. Finally, we
stimulated CFSE-labeled PBMC with IL-2 alone or IL-2 plus NY-
ESO-1 peptide to determine the proliferative potential of NY-
ESO-1-specific CD8?T cells (Fig. 5D). We found an important
fraction of EM28?and EM28?Ag-specific T cells as well as of
BV8?and BV13?Ag-specific T cells that divided in response to
antigenic stimulation. This is in line with the small proportion of
ex vivo NY-ESO-1-specific EM28?and EM28?T cell clonotypes
that expressed low, but readily detectable levels of Ki-67, indicat-
ing that cycling cells are present in both compartments (Fig. 6A).
Collectively, our data show that tumor-specific EM28?BV8 and
BV13 T cell clonotypes are composed of differentiated cells with
strong and efficient effector properties, although retaining their
Preferential expression of CD127/IL7R? but not of PD-1 by
We monitored CD27, granzyme B, perforin, CD127, PD-1, and
KLRG1 expression in combination with CD28 (Fig. 6). An im-
portant fraction of tumor-specific EM28?T cells expressed CD27
(87%) and CD127 (60–70%), whereas they expressed granzyme B
(36%), perforin (27%), and PD-1 (20–30%) at much lower levels
analysis was performed on cDNA obtained from individually sorted 5-cell samples of BV13high(BV13-clono2), BV13low(BV13-clono1), and
BV8high(mix of BV8-clono1 and BV8-clono2) EM28?NY-ESO-1-specific T cells as described in Fig. 2C. All shown data are from PBMC collected
from a lymphocytapheresis withdrawn 11 mo before vaccination (?11 mo). PCR products designed for CD3, IFN-?, TNF-?, CD127, and KLRG1
mRNA transcript analyses are depicted. Data from 10 independent 5-cell aliquots are shown; negative (?) and positive (?) controls. B, The
proportion of CD57-, PD-1-, and KLRG1-positive cells within BV8?and BV13?(multimer?CD8?CD45RA?) T cells was determined by mul-
tiparameter flow cytometry. Quadrants are set according to the internal control staining obtained from bulk multimer?CD8?T lymphocytes. Of note,
we were unable to determine the proportion of CD127?cells within BV8?and BV13?specific T cells due to the limitation of available mAb
combinations. C, IFN-? production by NY-ESO-1-specific CD8?T cells (?11 mo). Isolated CD8?T cells were stimulated with either with irrelevant
HIV-1 peptide (left), PMA/ionocycin (middle), or cognate peptide (right). Quadrants are set according to the internal control staining obtained from
whole CD8?multimer?T cells (see top panels). D, Representative dot plots (multimer?CD8?CD45RA?gated cells) of a 5-day stimulation assay
performed on CFSE-labeled PBMCs in the presence of IL-2 alone or IL-2 plus cognate peptide. At day 5, 2 ? 106cells were labeled with
NY-ESO-1-specific multimers, CD28 or TCR-BV8 or -BV13 Abs, and analyzed by flow cytometry. A representative experiment of three is shown
(C and D).
Ex vivo expression of mediators involved in T cell effector, survival and regulatory functions within clonotypes. A, Gene expression
2375The Journal of Immunology
than EM28?cells (70–90%). In contrast, the same proportion of
KLRG1?cells was observed within EM28?and EM28?compart-
ments (?50%). Altogether, our data show that unlike tumor-spe-
cific EM28?T cells that are closely related to effector-type cells,
the EM28?Ag-specific cells share functional features with mem-
ory lymphocytes. As the latter subset contains all identified tumor-
specific T cell clonotypes, it may potentially serve as a pool for
clonotypic T cells that can differentiate, expand, and mediate ef-
fector functions when required. Finally, KLRG1 expression is as-
sociated with distinct tumor-specific CD8?T cell clonotypes,
rather than with their functional differentiation status.
It is now well established that cancer patients often acquire Ag-
specific T cell responses to various targets. In a few cases it is even
possible to directly study the population of responding T cells
because of the presence of a particularly strong response. Insight
into the magnitude of such responses has been gradually gained
mainly because of the use of MHC/peptide multimers that allow
the direct identification and isolation of specific T cells. However,
a truly detailed knowledge of the ex vivo dynamics of individual
T cell clones is still lacking. Here, we present an extensive study
on the functional and proliferative potential of a dominant CD8?
T cell response directed against NY-ESO-1, a well-known tumor
Ag, on analyzing individual T cell clonotypes in tumor tissue and
peripheral blood over a prolonged period of time.
We could identify two functionally distinct populations of mul-
timer?T cells. The major population making up to 90% of the
cells displayed the hallmarks of highly differentiated and active
effector T cells (Fig. 1). Indeed, beside down-regulating lymph
node homing (CCR7 and CD62L) and costimulatory (CD28 and
CD27) receptors, these repetitively stimulated T cells (also desig-
nated as EM28?) down-regulated IL-7R? (involved in prosur-
vival/homeostatic signals delivered by IL-7) while up-regulating
NK-like receptors such as CD57 and CD94, as well as PD-1, an
inhibitory receptor. This dominant subset was mostly composed of
nine expanded T cell clonotypes incorporating variable TCR-BV
domains from only three subfamilies (BV1, BV8 and BV13). The
other relatively minor population (EM28?), representing between
5 and 10% of the NY-ESO-1-reactive CD8?T lymphocytes, was
also differentiated with features consistent with a resting memory
state (CD28?CD27?CD127?PD-1?granzymeB?perforin?) (26,
33, 34). The 9:1 ratio between these two subsets prevailed over the
entire observation period.
A remarkably and somewhat surprising finding was that all NY-
ESO-1-specific T cell clonotypes were found to be present within
the tiny memory multimer?T cell population, despite their oth-
erwise large TCR heterogeneity (Fig. 2). Thus, our results indicate
that such population serves as a reservoir for clonal expansion of
tumor-reactive dominant effector T cell responses with efficient
effector properties. This view is supported by two recent studies
reporting that murine TCR repertoires of both central-memory
(CM; CD62Lhigh) and effector-memory (EM; CD62Llow) Ag-spe-
cific CD8?T cells were largely overlapping (35, 36). Kedzierska
and coworkers (36) also showed that the memory CD62LhighT
cell repertoire was more diverse, thus preserving clonal diver-
sity, and proposed that the “best-fit” TCRs were selected from
the CM subset into the EM subset. The very limited human data
published so far indicated that T cell clonotypes can indeed be
shared by CM and EM cells (37), but more studies are necessary
to precisely describe human T cell differentiation at the clono-
Another major finding is that at least three of the dominant BV1,
BV8, and BV13 T cell clonotypes displayed long-term in vivo
persistence for up to 6 years (38). This is consistent with the ca-
pacity of the NY-ESO-1-specific EM28?and EM28?T cells to
proliferate when exposed to cognate Ag (Fig. 5). However, despite
long-term persistence, we observed a progressive shift in the pro-
portion of dominant clonotypes with an increase in BV13 T cells,
whereas BV8 populations declined over time in vivo (Fig. 2). Such
changes were coincident with repeated peptide vaccination and
eration, survival, effector, and regulatory functions within EM28?and
EM28?subsets. A, Analysis of Ki-67 expression (a marker of prolifera-
tion) within circulating BV8?or BV13?(top panels) and EM28?or
EM28?(bottom panels) specific CD8?T cells (?11 mo). The gating for
Ki-67 positive cells is based on isotype control stainings. One representa-
tive experiment of three is shown. B and C, Analysis of CD27, CD127,
PD-1, and KLRG1 cell surface expression on EM28?and EM28?
(multimer?CD8?CD45RA?) T cell subsets before (?11 mo) and after
(?8 mo) vaccination. The proportion of granzyme B- and perforin-positive
cells in EM28?and EM28?multimer?CD8?CD45RA?specific T cells
was also determined. For comparison, stainings on whole multimer?CD8?
T lymphocytes (CD8t) are depicted.
Ex vivo expression of mediators involved in T cell prolif-
2376DYNAMICS OF TUMOR-REACTIVE CD8?T LYMPHOCYTE RESPONSES EX VIVO
with a 2-fold expansion of the multimer?population. Fluctuations
in the TCR repertoire of Ag-specific CD8?T cell populations have
also been reported during primary HIV infection (39), in HIV in-
fected individuals with partial control of viremia (19), as well as in
an healthy subject during the first year of EBV infection (40).
Whether the decline in frequency of particular T cell clones is
permanent or whether it represents a temporary or random fluctu-
ation of the TCR repertoire remains still unclear, and deserves
further in-depth analyses.
Our data revealed that the BV8-clonotype 2 cell subpopulation
was undetectable at the latest time point analyzed (?18 mo),
despite displaying relatively long average telomere lengths
(Fig. 4). This indicates that the differential evolution observed
between BV8 and BV13 T cell clonotypes cannot be attributed
to replicative senescence due to the presence of critically short
telomeres (27). Moreover, we found a rapid loss of telomere
length within these clonotypes, corresponding to a 20-fold in-
creased turnover rate when compared with average telomere
shortening in total CD8?T cells associated with aging (28),
The results reported here support the view that the loss of the
BV8 clonotype 2 is likely due to dilution upon clonal expansion
of the BV13 clonotypes. However, one cannot formally exclude
that the progressive deletion of BV8 T cell clonotypes is asso-
ciated with activation-induced cell death following repetitive
triggering by Ag derived from tumor cells or vaccination. Be-
cause apoptosis is rapidly induced upon TCR triggering, this
question remains difficult to assess experimentally, because
MHC-peptide-multimers are required to study T cells in the
context of natural TCR repertoires, but multimers trigger TCRs
and thus promote apoptosis. Intriguingly, the BV13:BV8 ratio
attained by the end of the observation period in the circulating
lymphocyte compartment (?18 mo) was already present in the
NY-ESO-1-specific T cell population isolated from a tumor in-
filtrated lymph node resected 21 mo earlier, at a time predating
the instauration of therapeutic vaccination (Fig. 3). Due to the
low frequencies of circulating EM28?Ag-specific T cells, we
were unable to assess the telomere lengths of such cells. Future
work involving the careful evaluation of their replicative his-
tory combined to their cell cycle status are needed to fully elu-
cidate the role of EM28?cells in CD8?T cell differentiation,
eventually leading to the generation of differentiated EM28?
One question raised by our data concerns the biological param-
eters that may trigger the preferential selection of BV13 T cell
clonotypes over time. The avidity of the TCR for MHC/peptide
complexes is unlikely to be involved because the different BV8
and BV13 clonotypes recognized and killed autologous NY-
ESO-1 tumor cells and peptide-pulsed T2 cells with similar func-
tional avidity (Fig. 4; data not shown). In addition, immunotherapy
had no detectable impact on the functional avidity of BV8 and
BV13 T cell clonotypes, because the clones shared similar killing
efficacy, whether the cells were retrieved before or after the start of
peptide vaccination. High levels of KLRG1 expression were seen
on chronically activated EBV- and CMV-specific CD8?T lym-
phocytes, and to a lesser extent on T cells specific for influenza, a
resolved infection without a latent stage (41, 42). In line with these
results, we found that a significant proportion of NY-ESO-1-spe-
cific EM28?and EM28?T lymphocytes expressed KLRG1 (Fig.
6). Moreover, the work reported here extends recent findings (41),
that KLRG1?CD8?T cell population is heterogeneous, as it con-
tains both differentiated (EM28?) and less differentiated (EM28?)
cells. Our observations further indicate that KLRG1 expression is
associated to distinct TCR-BV clonotypes (Fig. 5), regardless of
their differentiation status. Although KLRG1 is expressed on T
cells that have undergone a large number of cell divisions (43, 44),
the function of this molecule has not been fully explored. Recently,
Gru ¨ndemann and coworkers (45) identified E-cadherin as a ligand
for murine KLRG1, and proposed that its ligation by E-cadherin in
healthy tissues may exert an inhibitory effect on primed T cells. In
addition, mouse KLRG1 also binds to N- or R-cadherin (46), but
ligand(s) for human KLGR1 have yet to be defined. Because in our
tumor model, KLRG1 expression was preferentially observed on
BV13-specific T cell clonotypes displaying increased in vivo cell
turnover, another hypothesis is that KLRG1 downstream receptor
signaling may be involved in promoting long-term survival of cells
rather than their inhibition (47). Alternatively, KLRG1 expression
is preferentially up-regulated in strongly proliferating clonotypes
and may allow their specific inhibition by ligand-expressing tis-
sues or melanoma cells.
Emerging findings suggest that the expression of PD-1 con-
tributes to the functional impairment that characterizes T cells
during chronic viral infections, because blocking the PD-1/
PD-1L pathway enhances both proliferation and effector func-
tions of “exhausted” T cells (48–51). Intriguingly, these results
do not exactly support those obtained here in which chronically
expanded tumor-reactive T cells expressing PD-1 also retained
their capacity to undergo proliferation, cytokine production,
and cytotoxic activity. Both BV8 and BV13 NY-ESO-1-specific
T cell clonotypes efficiently secreted effector mediators such as
granzyme B and perforin and killed tumor cells when tested
directly ex vivo in a LiveCount assay (Fig. 4). Moreover, an
important proportion of EM28?NY-ESO-1-specific T cells
were able to release IFN-? upon stimulation with cognate pep-
tide. Ongoing studies on blood samples from patient LAU 444,
who exhibited a persisting and dominant Melan-A-specific
CD8?T cell response (13) further emphasized the finding that
PD-1 was preferentially expressed within the differentiated
EM28?compartment, whereas IL-7R? expression was mostly
found on EM28?cells (data not shown). Collectively, our data
strongly support the notion that circulating tumor-specific
CD8?T cell clonotypes not only share phenotypic features with
that of differentiated cells but also exhibit functional character-
istics similar to those of effective CTL specific for immuno-
dominant viral Ags such as EBV or CMV (52). Whether PD-1
expression on tumor-reactive CTL may regulate such cells di-
rectly at the tumor site where melanoma cells, particularly in
the presence of IFN-?, express the PD-1L surface molecule
(53), remains unclear and deserves additional studies.
We recently reported that both the natural T cell triggering by
endogenous Ags and subsequent vaccination preferentially pro-
moted an endogenous T cell clonotype with relatively high TCR
avidity and antitumor activity (13). In the present study, we found
that all individual T cell clonotypes that were identified after the
start of immunotherapy were already present within the NY-ESO-
1-specific CD8?T cell response several months to years before
vaccination. Importantly, both EM28?and EM28?compartments
comprised the same T cell clonotypes, thus revealing a tight in-
terplay of T cells in early and differentiated stages. These data
suggest that effective therapeutic vaccination for cancer may only
be accomplished in the presence of both memory (EM28?) and
effector (EM28?) subsets of tumor-specific T cells and that mul-
tifactorial events determine the rise and fall of dominant clono-
types that contribute to dynamically sustain antitumor CD8?T
We gratefully acknowledge patient LAU 50 for active participation, and
the hospital staff for excellent collaboration. We gratefully thank
2377The Journal of Immunology
C. Barbey, J.-C. Cerottini, F. Lejeune, S. Leyvraz, D. Lie ´nard, M. Matter,
K. Muehlethaler, D. Rimoldi, and S. Salvi for collaboration and advice,
I. Luescher and P. Guillaume for multimers, Seppic for Montanide ISA-51
(IFA). We also thank the excellent technical and secretarial help of
C. Geldhof, R. Milesi, D. Minaı ¨dis, N. Montandon, and M. van Overloop.
The authors have no financial conflict of interest.
1. Zinkernagel, R. M., and P. C. Doherty. 1997. The discovery of MHC restriction.
Immunol. Today 18: 14–17.
2. Altman, J. D., P. A. Moss, P. J. Goulder, D. H. Barouch, M. G.
McHeyzer-Williams, J. I. Bell, A. J. McMichael, and M. M. Davis. 1996. Phe-
notypic analysis of antigen-specific T lymphocytes. Science 274: 94–96.
3. Romero, P., P. R. Dunbar, D. Valmori, M. Pittet, G. S. Ogg, D. Rimoldi,
J. L. Chen, D. Lienard, J. C. Cerottini, and V. Cerundolo. 1998. Ex vivo staining
of metastatic lymph nodes by class I major histocompatibility complex tetramers
reveals high numbers of antigen-experienced tumor-specific cytolytic T lympho-
cytes. J. Exp. Med. 188: 1641–1650.
4. Butz, E. A., and M. J. Bevan. 1998. Massive expansion of antigen-specific CD8?
T cells during an acute virus infection. Immunity 8: 167–175.
5. Moss, P. A., R. J. Moots, W. M. Rosenberg, S. J. Rowland-Jones, H. C. Bodmer,
A. J. McMichael, and J. I. Bell. 1991. Extensive conservation of ? and ? chains
of the human T-cell antigen receptor recognizing HLA-A2 and influenza A ma-
trix peptide. Proc. Natl. Acad. Sci. USA 88: 8987–8990.
6. Lehner, P. J., E. C. Wang, P. A. Moss, S. Williams, K. Platt, S. M. Friedman,
J. I. Bell, and L. K. Borysiewicz. 1995. Human HLA-A0201-restricted cytotoxic
T lymphocyte recognition of influenza A is dominated by T cells bearing the V
? 17 gene segment. J. Exp. Med. 181: 79–91.
7. Annels, N. E., M. F. Callan, L. Tan, and A. B. Rickinson. 2000. Changing pat-
terns of dominant TCR usage with maturation of an EBV-specific cytotoxic T cell
response. J. Immunol. 165: 4831–4841.
8. Price, D. A., J. M. Brenchley, L. E. Ruff, M. R. Betts, B. J. Hill, M. Roederer,
R. A. Koup, S. A. Migueles, E. Gostick, L. Wooldridge, et al. 2005. Avidity for
antigen shapes clonal dominance in CD8?T cell populations specific for persis-
tent DNA viruses. J. Exp. Med. 202: 1349–1361.
9. Trautmann, L., M. Rimbert, K. Echasserieau, X. Saulquin, B. Neveu, J. Dechanet,
V. Cerundolo, and M. Bonneville. 2005. Selection of T cell clones expressing
high-affinity public TCRs within human cytomegalovirus-specific CD8 T cell
responses. J. Immunol. 175: 6123–6132.
10. Pantaleo,G., J.F.Demarest, H.
J. W. Adelsberger, P. Borrow, M. S. Saag, G. M. Shaw, R. P. Sekaly, et al. 1994.
Major expansion of CD8?T cells with a predominant V ? usage during the
primary immune response to HIV. Nature 370: 463–467.
11. Kalams, S. A., R. P. Johnson, A. K. Trocha, M. J. Dynan, H. S. Ngo, R. T.
D’Aquila, J. T. Kurnick, and B. D. Walker. 1994. Longitudinal analysis of T cell
receptor (TCR) gene usage by human immunodeficiency virus 1 envelope-spe-
cific cytotoxic T lymphocyte clones reveals a limited TCR repertoire. J. Exp.
Med. 179: 1261–1271.
12. Dong, T., G. Stewart-Jones, N. Chen, P. Easterbrook, X. Xu, L. Papagno,
V. Appay, M. Weekes, C. Conlon, C. Spina, et al. 2004. HIV-specific cytotoxic
T cells from long-term survivors select a unique T cell receptor. J. Exp. Med. 200:
13. Speiser, D. E., P. Baumgaertner, C. Barbey, V. Rubio-Godoy, A. Moulin,
P. Corthesy, E. Devevre, P. Y. Dietrich, D. Rimoldi, D. Lienard, et al. 2006. A
novel approach to characterize clonality and differentiation of human melanoma-
specific T cell responses: spontaneous priming and efficient boosting by vacci-
nation. J. Immunol. 177: 1338–1348.
14. Boon, T., P. G. Coulie, B. J. Van den Eynde, and P. van der Bruggen. 2006.
Human T cell responses against melanoma. Annu. Rev. Immunol. 24: 175–208.
15. Silins, S. L., S. M. Cross, S. L. Elliott, S. J. Pye, S. R. Burrows, J. M. Burrows,
D. J. Moss, V. P. Argaet, and I. S. Misko. 1996. Development of Epstein-Barr
virus-specific memory T cell receptor clonotypes in acute infectious mononucle-
osis. J. Exp. Med. 184: 1815–1824.
16. Levitsky, V., P. O. de Campos-Lima, T. Frisan, and M. G. Masucci. 1998. The
clonal composition of a peptide-specific oligoclonal CTL repertoire selected in
response to persistent EBV infection is stable over time. J. Immunol. 161:
17. Posavad, C. M., M. L. Huang, S. Barcy, D. M. Koelle, and L. Corey. 2000. Long
term persistence of herpes simplex virus-specific CD8?CTL in persons with
frequently recurring genital herpes. J. Immunol. 165: 1146–1152.
18. Islam, S. A., C. M. Hay, K. E. Hartman, S. He, A. K. Shea, A. K. Trocha,
M. J. Dynan, N. Reshamwala, S. P. Buchbinder, N. O. Basgoz, and S. A. Kalams.
2001. Persistence of human immunodeficiency virus type 1-specific cytotoxic
T-lymphocyte clones in a subject with rapid disease progression. J. Virol. 75:
19. Meyer-Olson, D., K. W. Brady, M. T. Bartman, K. M. O’Sullivan, B. C. Simons,
J. A. Conrad, C. B. Duncan, S. Lorey, A. Siddique, R. Draenert, et al. 2006.
Fluctuations of functionally distinct CD8?T-cell clonotypes demonstrate flexi-
bility of the HIV-specific TCR repertoire. Blood 107: 2373–2383.
20. Le Gal, F. A., V. M. Widmer, V. Dutoit, V. Rubio-Godoy, J. Schrenzel,
P. R. Walker, P. J. Romero, D. Valmori, D. E. Speiser, and P. Y. Dietrich.
2007. Tissue homing and persistence of defined antigen-specific CD8?tumor-
Soudeyns, C.Graziosi, F. Denis,
reactive T-cell clones in long-term melanoma survivors. J. Invest. Dermatol.
21. Baumgaertner, P., N. Rufer, E. Devevre, L. Derre, D. Rimoldi, C. Geldhof,
V. Voelter, D. Lienard, P. Romero, and D. E. Speiser. 2006. Ex vivo detectable
human CD8 T-cell responses to cancer-testis antigens. Cancer Res. 66:
22. Marcolino, I., G. K. Przybylski, M. Koschella, C. A. Schmidt, D. Voehringer,
M. Schlesier, and H. Pircher. 2004. Frequent expression of the natural killer cell
receptor KLRG1 in human cord blood T cells: correlation with replicative his-
tory. Eur. J. Immunol. 34: 2672–2680.
23. Rufer, N., P. Reichenbach, and P. Romero. 2005. Methods for the ex vivo char-
acterization of human CD8?T subsets based on gene expression and replicative
history analysis. Methods Mol. Med. 109: 265–284.
24. Arden, B., S. P. Clark, D. Kabelitz, and T. W. Mak. 1995. Human T-cell receptor
variable gene segment families. Immunogenetics 42: 455–500.
25. Roux, E., C. Helg, F. Dumont-Girard, B. Chapuis, M. Jeannet, and E. Roosnek.
1996. Analysis of T-cell repopulation after allogeneic bone marrow transplanta-
tion: significant differences between recipients of T-cell depleted and unmanipu-
lated grafts. Blood 87: 3984–3992.
26. Rufer, N., A. Zippelius, P. Batard, M. J. Pittet, I. Kurth, P. Corthesy,
J. C. Cerottini, S. Leyvraz, E. Roosnek, M. Nabholz, and P. Romero. 2003. Ex
vivo characterization of human CD8?T subsets with distinct replicative history
and partial effector functions. Blood 102: 1779–1787.
27. Rufer, N., W. Dragowska, G. Thornbury, E. Roosnek, and P. M. Lansdorp. 1998.
Telomere length dynamics in human lymphocyte subpopulations measured by
flow cytometry. Nat. Biotechnol. 16: 743–747.
28. Rufer, N., T. H. Brummendorf, S. Kolvraa, C. Bischoff, K. Christensen,
L. Wadsworth, M. Schulzer, and P. M. Lansdorp. 1999. Telomere fluorescence
measurements in granulocytes and T lymphocyte subsets point to a high turnover
of hematopoietic stem cells and memory T cells in early childhood. J. Exp. Med.
29. Rufer, N., M. Migliaccio, J. Antonchuk, R. K. Humphries, E. Roosnek, and
P. M. Lansdorp. 2001. Transfer of the human telomerase reverse transcriptase
(TERT) gene into T lymphocytes results in extension of replicative potential.
Blood 98: 597–603.
30. Devevre, E., P. Romero, and Y. D. Mahnke. 2006. LiveCount assay: concomitant
measurement of cytolytic activity and phenotypic characterisation of CD8?T-
cells by flow cytometry. J. Immunol. Methods 311: 31–46.
31. Valmori, D., V. Dutoit, D. Lienard, D. Rimoldi, M. J. Pittet, P. Champagne,
K. Ellefsen, U. Sahin, D. Speiser, F. Lejeune, J. C. Cerottini, and P. Romero.
2000. Naturally occurring human lymphocyte antigen-A2 restricted CD8?T-cell
response to the cancer testis antigen NY-ESO-1 in melanoma patients. Cancer
Res. 60: 4499–4506.
32. Stewart-Jones, G. B., A. J. McMichael, J. I. Bell, D. I. Stuart, and E. Y. Jones.
2003. A structural basis for immunodominant human T cell receptor recognition.
Nat. Immunol. 4: 657–663.
33. Appay, V., P. R. Dunbar, M. Callan, P. Klenerman, G. M. Gillespie, L. Papagno,
G. S. Ogg, A. King, F. Lechner, C. A. Spina, et al. 2002. Memory CD8?T cells
vary in differentiation phenotype in different persistent virus infections. Nat. Med.
34. Kaech, S. M., J. T. Tan, E. J. Wherry, B. T. Konieczny, C. D. Surh, and
R. Ahmed. 2003. Selective expression of the interleukin 7 receptor identifies
effector CD8 T cells that give rise to long-lived memory cells. Nat. Immunol. 4:
35. Bouneaud, C., Z. Garcia, P. Kourilsky, and C. Pannetier. 2005. Lineage relation-
ships, homeostasis, and recall capacities of central- and effector-memory CD8 T
cells in vivo. J. Exp. Med. 201: 579–590.
36. Kedzierska, K., V. Venturi, K. Field, M. P. Davenport, S. J. Turner, and
P. C. Doherty. 2006. Early establishment of diverse T cell receptor profiles for
influenza-specific CD8?CD62Lhimemory T cells. Proc. Natl. Acad. Sci. USA
37. Baron, V., C. Bouneaud, A. Cumano, A. Lim, T. P. Arstila, P. Kourilsky,
L. Ferradini, and C. Pannetier. 2003. The repertoires of circulating human CD8?
central and effector memory T cell subsets are largely distinct. Immunity 18:
38. Le Gal, F. A., M. Ayyoub, V. Dutoit, V. Widmer, E. Jager, J. C. Cerottini,
P. Y. Dietrich, and D. Valmori. 2005. Distinct structural TCR repertoires in
naturally occurring versus vaccine-induced CD8?T-cell responses to the tumor-
specific antigen NY-ESO-1. J. Immunother. 28: 252–257.
39. Pantaleo, G., H. Soudeyns, J. F. Demarest, M. Vaccarezza, C. Graziosi,
S. Paolucci, M. Daucher, O. J. Cohen, F. Denis, W. E. Biddison, et al. 1997.
Evidence for rapid disappearance of initially expanded HIV-specific CD8?T cell
clones during primary HIV infection. Proc. Natl. Acad. Sci. USA 94: 9848–9853.
40. Cohen, G. B., S. A. Islam, M. S. Noble, C. Lau, C. Brander, M. A. Altfeld,
E. S. Rosenberg, J. E. Schmitz, T. O. Cameron, and S. A. Kalams. 2002. Clono-
type tracking of TCR repertoires during chronic virus infections. Virology 304:
41. Ibegbu, C. C., Y. X. Xu, W. Harris, D. Maggio, J. D. Miller, and A. P. Kourtis.
2005. Expression of killer cell lectin-like receptor G1 on antigen-specific human
CD8?T lymphocytes during active, latent, and resolved infection and its relation
with CD57. J. Immunol. 174: 6088–6094.
42. Thimme, R., V. Appay, M. Koschella, E. Panther, E. Roth, A. D. Hislop,
A. B. Rickinson, S. L. Rowland-Jones, H. E. Blum, and H. Pircher. 2005. In-
creased expression of the NK cell receptor KLRG1 by virus-specific CD8 T cells
during persistent antigen stimulation. J. Virol. 79: 12112–12116.
2378 DYNAMICS OF TUMOR-REACTIVE CD8?T LYMPHOCYTE RESPONSES EX VIVO
43. Voehringer, D., C. Blaser, P. Brawand, D. H. Raulet, T. Hanke, and H. Pircher. Download full-text
2001. Viral infections induce abundant numbers of senescent CD8 T cells. J. Im-
munol. 167: 4838–4843.
44. Voehringer, D., M. Koschella, and H. Pircher. 2002. Lack of proliferative ca-
pacity of human effector and memory T cells expressing killer cell lectinlike
receptor G1 (KLRG1). Blood 100: 3698–3702.
45. Grundemann, C., M. Bauer, O. Schweier, N. von Oppen, U. Lassing, P. Saudan,
K. F. Becker, K. Karp, T. Hanke, M. F. Bachmann, and H. Pircher. 2006. Cutting
edge: identification of E-cadherin as a ligand for the murine killer cell lectin-like
receptor G1. J. Immunol. 176: 1311–1315.
46. Ito, M., T. Maruyama, N. Saito, S. Koganei, K. Yamamoto, and N. Matsumoto.
2006. Killer cell lectin-like receptor G1 binds three members of the classical
cadherin family to inhibit NK cell cytotoxicity. J. Exp. Med. 203: 289–295.
47. Ugolini, S., C. Arpin, N. Anfossi, T. Walzer, A. Cambiaggi, R. Forster, M. Lipp,
R. E. Toes, C. J. Melief, J. Marvel, and E. Vivier. 2001. Involvement of inhibitory
NKRs in the survival of a subset of memory-phenotype CD8?T cells. Nat.
Immunol. 2: 430–435.
48. Barber, D. L., E. J. Wherry, D. Masopust, B. Zhu, J. P. Allison, A. H. Sharpe,
G. J. Freeman, and R. Ahmed. 2006. Restoring function in exhausted CD8 T cells
during chronic viral infection. Nature 439: 682–687.
49. Day, C. L., D. E. Kaufmann, P. Kiepiela, J. A. Brown, E. S. Moodley, S. Reddy,
E. W. Mackey, J. D. Miller, A. J. Leslie, C. DePierres, et al. 2006. PD-1 expres-
sion on HIV-specific T cells is associated with T-cell exhaustion and disease
progression. Nature 443: 350–354.
50. Trautmann, L., L. Janbazian, N. Chomont, E. A. Said, S. Gimmig, B. Bessette,
M. R. Boulassel, E. Delwart, H. Sepulveda, R. S. Balderas, et al. 2006. Upregu-
lation of PD-1 expression on HIV-specific CD8?T cells leads to reversible
immune dysfunction. Nat. Med. 12: 1198–1202.
51. Petrovas, C., J. P. Casazza, J. M. Brenchley, D. A. Price, E. Gostick, W. C.
Adams, M. L. Precopio, T. Schacker, M. Roederer, D. C. Douek, and R. A. Koup.
2006. PD-1 is a regulator of virus-specific CD8?T cell survival in HIV infection.
J. Exp. Med. 203: 2281–2292.
52. Appay, V., C. Jandus, V. Voelter, S. Reynard, S. E. Coupland, D. Rimoldi,
D. Lienard, P. Guillaume, A. M. Krieg, J. C. Cerottini, et al. 2006. New gener-
ation vaccine induces effective melanoma-specific CD8?T cells in the circulation
but not in the tumor site. J. Immunol. 177: 1670–1678.
53. Dong, H., S. E. Strome, D. R. Salomao, H. Tamura, F. Hirano, D. B. Flies,
P. C. Roche, J. Lu, G. Zhu, K. Tamada, et al. 2002. Tumor-associated B7-H1
promotes T-cell apoptosis: a potential mechanism of immune evasion. Nat. Med.
2379 The Journal of Immunology