Journal of Immunological Methods 310 (2006) 136 – 148

www.elsevier.com/locate/jim

Research paper
Dextramers: New generation of fluorescent MHC class I/peptide
multimers for visualization of antigen-specific CD8 + T cells
Pascal Batard a , Daniel A. Peterson a , Estelle Devêvre a , Philippe Guillaume b ,
Jean-Charles Cerottini b , Donata Rimoldi b , Daniel E. Speiser a ,
Lars Winther c , Pedro Romero a,⁎
a
Division of Clinical Onco-Immunology, Ludwig Institute for Cancer Research, Lausanne Branch, University Hospital (CHUV),
1005 Lausanne, Switzerland
b
Ludwig Institute for Cancer Research, Lausanne Branch, University of Lausanne, 1066 Epalinges, Switzerland
c
Detection Technology, Dako, DK-2600 Glostrup, Denmark
Received 7 September 2004; received in revised form 27 December 2005; accepted 6 January 2006
Available online 17 January 2006

Abstract

Direct identification as well as isolation of antigen-specific T cells became possible since the development of “tetramers” based
on avidin–fluorochrome conjugates associated with mono-biotinylated class I MHC–peptide monomeric complexes. In principle, a
series of distinct class I MHC–peptide tetramers, each labelled with a different fluorochrome, would allow to simultaneously
enumerate as many unique antigen-specific CD8+ T cells. Practically, however, only phycoerythrin and allophycocyanin
conjugated tetramers have been generally available, imposing serious constraints for multiple labeling. To overcome this limitation,
we have developed dextramers which are multimers based on a dextran backbone bearing multiple fluorescein and streptavidin
moieties. Here we demonstrate the functionality and optimization of these new probes on human CD8+ T cell clones with four
independent antigen specificities. Their applications to the analysis of relatively low frequency antigen-specific T cells in peripheral
blood, as well as their use in fluorescence microscopy, are demonstrated. The data show that dextramers produce a stronger signal
than their fluoresceinated tetramer counterparts. Thus, these could become the reagents of choice as the antigen-specific T cell
labeling transitions from basic research to clinical application.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Multimer; Dextran; Antigen-specific CTL; Flow cytometry

1. Introduction
Abbreviations: MHC, major histocompatibility complex; CTL,
cytolytic T lymphocyte; PE, phycoerythrin; APC, allophycocyanin; Accurate identification and enumeration in ex vivo
ECD, phycoerythrin-texas red; EBV, Epstein Barr virus; CMV, tissue samples of cytolytic T lymphocytes (CTL) specific
cytomegalovirus; FITC, fluorescein isothyocyanate; PBMC, periph- for tumor and viral antigens are of utmost importance for
eral blood mononucleated cells; MFI, mean fluorescence intensity. the understanding of cell mediated adaptive immunity.
⁎ Corresponding author. Division of Clinical Onco-Immunology,
Among multiple approaches, fluorescent probes called
Ludwig Institute for Cancer Research, Hôpital Orthopédique, Aile est,
Niveau 5, Avenue Pierre Decker, 4, 1005 Lausanne, Switzerland. Tel.: “tetramers” offer the most direct one (Altman et al.,
+41 21 314 02 12; fax: +41 21 314 74 77. 1996). Their use in combination with conjugated
E-mail address: pedro.romero@isrec.unil.ch (P. Romero). antibodies directed against cell surface antigens such as
0022-1759/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jim.2006.01.006
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 137

CD45RA, CCR7, CD27, CD28 and others allows to monomers, are as efficient as conventional PE- or APC-
distinguish the functional stage of differentiation of conjugated tetramers for the labeling of specific CTL.
antigen-specific T cells (Pittet et al., 1999; Pittet et al., The optimal concentration of dextran, the effect of the
2001; Appay et al., 2002; Speiser et al., 2002; Klenerman monomer-to-dextran ratio and the kinetics of binding
et al., 2002; Rufer et al., 2003; Wolfl et al., 2004). Such were carefully analyzed on a panel of human antigen-
detailed analyses rely on the use of multicolor flow specific CTL clones (EBV, CMV, Influenza and mela-
cytometry. Key to the multiparameter approach to noma antigen reactive). Further we show that these
quantitative and qualitative assessment of CD8+ T cell reagents are sensitive enough for use in primary human
responses is the ability to combine as many tetramers and peripheral blood mononuclear cells (PBMC) where the
antibodies conjugated to fluorochromes as possible. The frequency of antigen-specific CTL is limiting. We also
tetramers result from the association of one fluorescent found that Alexa 532-conjugated dextramers could be
avidin and four biotinylated class I MHC molecules used in microscopy.
preloaded with a given antigenic peptide (monomer)
(Altman et al., 1996). Because of the requirement of high 2. Materials and methods
sensitivity, the preparation of tetramers has until now
been restricted to avidin conjugates containing the 2.1. Cells
fluorochromes phycoerythrin (PE) or allophycocyanin
(APC). Indeed, these particular phycobiliproteins display Peripheral blood mononuclear cells (PBMC) from
a strong fluorescence due to their exceptionally high healthy donors were separated by centrifugation over
molar absorption coefficient and high quantum yield. Ficoll-Paque (Amersham Pharmacia Biotech, Uppsala,
However, this prevented the use of PE and APC for Sweden), washed three times and labeled.
simultaneous detection of poorly expressed antigens in HLA-A2 restricted CTL clones specific for peptides
multicolor analyses. Likewise, this limited the number of derived from the matrix protein of influenza virus
antigen specificities that could be detected simultaneous- (FLU), Melan-A/MART-1 melanoma antigen (ELA),
ly when using tetramers prepared with various avidin– Epstein-Barr virus BMLF1 protein (EBV) or cytomeg-
fluorochrome conjugates. Indeed, this limitation is alovirus pp65 (CMV) were obtained and maintained in
highlighted in the analysis of precious samples, like culture as described elsewhere (Pittet et al., 2001).
tumor infiltrating lymphocytes from resected tumors
(Mortarini et al., 2003; Zippelius et al., 2004). 2.2. Antibodies, tetramers and dextran
We have now developed fluorescent multimers based
on dextran backbones that we have named dextramers. Antibodies specific for human CD8 (SK1, HIT8a and
Dextrans are polymers of glucose mainly associated RPA-T8) and for human CD69 (L78) were obtained
through a 1–6 linkage and 1–3 linkage at the branched from BD Biosciences (San Jose, CA). They were con-
sites. Their length is variable, ranging from tens to jugated with either PE, APC or APC⁎Cy7 (PharRed).
hundreds kDa. Various fluorochromes and proteins such Streptavidin conjugated with ECD was obtained from
as streptavidin or horseradish peroxidase can be co- Beckman/Coulter (Miami, FL).
valently coupled to dextrans (Yamashita et al., 1997). Tetramers were prepared as described elsewhere
While a protein, antibody, streptavidin or avidin can be (Altman et al., 1996; Ogg and McMichael, 1998). The
typically conjugated with 4 to 5 fluoresceins, a 270 kDa sequences for FLU, ELA, EBV and CMV peptides
dextran can be cross linked with 20 fluorescein and 10 presented to specific T cells in association with the
streptavidin moieties. Thus, such a fluoresceinated dex- HLA-A2 molecule were respectively GILGFVFTL,
tran provides a strong fluorescent signal upon excitation ELAGIGILTV, GLCTLVAML and NLVPMVATV.
and it can bind 10 times more biotinylated class I MHC- Four different fluoresceinated dextran polymers
peptide monomers than conventional tetramers. The covalently coupled to streptavidin were supplied from
increased sensitivity of various dextran conjugates for Dako Cytomation (Table 1). To prepare the different
analysis of antigen expression has been established dextramers, dextrans (0.5–1.5mg/ml) and HLA-A2-
mainly for microscopy (Chilosi et al., 1994; Tsutsumi antigenic peptide monomers (1–3mg/ml) were diluted
et al., 1995; Sabattini et al., 1998; Richter et al., 1999; in phosphate buffered saline (PBS), 0.2% bovine serum
Turner et al., 1999) and only, in rare cases, for flow albumine (BSA) and 0.02% azide (PBS/BSA/Azide).
cytometry (Siiman et al., 1999). Generally, when preparing dextrans saturated with mo-
Here, we demonstrate that dextrans conjugated with nomers, they were mixed together by stepwise addition
FITC and streptavidin, then loaded with MHC/peptide of monomer into dextran (one fifth of the total amount
138 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

Table 1
Characteristics of the fluoresceinated dextrans conjugated with streptavidin used in this study
Name Units/dextrana Molecular weight (kDa) Maximal monomers per dextranb
FITC Streptavidin Dextran Streptavidin Total
Dex70F 6.6 3.6 70 60 290 14
Dex150F 10 6.5 150 60 550 26
Dex270F 20 10.5 270 60 920 42
Dex500F 29 17 500 60 1550 68
a
The number of streptavidins and fluoresceins per dextran chain was determined after the conjugation by UV measurements of the HPLC purified
polymeric conjugates and the unconjugated components.
b
The maximum number of monomers per dextran is based on the assumption that each streptavidin has four binding sites. This theoretical
assumption may be oversimplified, as the binding sites do not have equal binding strength with biotin, and steric hindrance may not allow for loading
of four monomers per streptavidin.

of monomers at the time). Additions were done every show that significant labeling was obtained for all four
5min at room temperature. The addition of dextran dextrans saturated with HLA-A2/ELA monomers. In
solution into monomer solution was also done in a addition, the intensity of the specific labeling increased
similar fashion when dextran was saturated with with the molecular weight of dextrans. The intensity of
monomers. Table 1 shows that the number of fluorescein unspecific labeling also correlated positively with their
and streptavidin units/dextran varies from 6.6 to 29 and molecular weight, resulting in net similar sensitivity
from 3.6 to 17, respectively, depending on the dextran's (ratio of labelings with specific peptide versus irrelevant
molecular weight. peptide) for all saturated dextramers. Based on these
results, we decided to use without distinction depending
2.3. Flow cytometry on the available material either the 270 or the 500kDa
fluoresceinated dextran because the intensity of labeling
Cells (0.25 to 1 × 105) were resuspended in 50 μl of
PBS/BSA/Azide. Dextramers or tetramers were added 1000
for 20 to 60min at room temperature. Cells were washed
once in the same buffer and analyzed, immediately or
after an additional labeling with anti-CD8 antibody, in
either a FACSVantage SE or a FACScan (BD Bios-
ciences). Data acquisition and analysis were performed 100
using CellQuest software (BD Biosciences). The results
MFI

were generally expressed as mean fluorescence intensity

(MFI).
10
2.4. Fluorescence microscopy

Cells were examined with an Axioskop microscope

(Zeiss, Germany) equipped with a 100 W, HBO lamp
and pictures were taken with a CCD camera from 1
Photonic Science (Robertsbridge, England). 0 10 20 30 40
Concentration (µg/ml)
3. Results
Fig. 1. Comparitive specific T cell labeling efficiency of dextramers
built with dextrans of different molecular weights. Fluoresceinated
3.1. Analysis of antigen-specific CTL by flow cytometry dextrans of different molecular weights (70, 150, 270, 500kDa) were
with dextramers and effect of dextran molecular weight saturated with HLA-A2/ELA (dark symbols) or HLA-A2/FLU (open
symbols) monomers. A CTL clone specific for a Melan-A-derived
We first compared the ability of the four fluorescei- peptide was incubated with increasing concentrations of these
nated dextrans with different molecular weights (70, dextramers. The 70, 150, 270 and 500kDa dextramers are represented
by squares, diamonds, circles and triangles, respectively. The hori-
150, 270 and 500 kDa) loaded with HLA-A2/ELA mo- zontal bar indicates unlabeled cells and the + and x symbols indicate
nomers to bind to specific CTL. The results obtained the intensity of labeling observed with specific (ELA) and irrelevant
using a ELA-specific clone are illustrated in Fig. 1. They (FLU) PE-conjugated tetramers.
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 139

obtained with both molecules was equivalent. A FLU three independently derived FLU-specific CTL clones
specific CTL clone was also labeled with the same series with the fluoresceinated Melan-A dextramer showed
of dextrans loaded with HLA-A2/FLU monomers and backgrounds of 4.4, 7.7 and 8.4 MFI (not shown).
comparable results were obtained, albeit with a higher
unspecific labeling (data not shown). This higher un- 3.2. Optimization of dextran concentration
specific background is a particularity of the FLU clone
used for comparing the series of dextramers presented in To establish the optimal dextramer dose, a fluor-
this paper. Indeed, additional stainings performed on esceinated, 500kDa dextran (Dex500F) (60 μg/ml) was

A ELA clone
D FLU clone
1000 1000

100 100

MFI
MFI

10 10

1 1
0 5 10 05 10
Concentration (µg/ml) Concentration (µg/ml)

B E
100

100
80

80
40 60

40 60
Counts

Counts
20

20
0

100 101 102 103 104 100 101 102 103 104
Dextramer (FITC) Dextramer (FITC)

C F
100
0 10 20 30 40 50 60 70 80

80
40 60
Counts
Counts

20
0

100 101 102 103 104 100 101 102 103 104
Tetramer (PE) Tetramer (PE)

Fig. 2. Titration of fluoresceinated, 500kDa dextramers on CTL clones specific for Influenza- and Melan-A-derived peptides. The fluoresceinated,
500kDa dextran was saturated with biotinylated HLA-A2 monomers loaded either with an Influenza-derived peptide (FLU, open circles) or with a
melanoma-derived peptide (ELA, full circles). They were titrated on human CTL clones specific for the ELA peptide (A), and for the FLU peptide (D).
The cloned T cells were labeled at room temperature for 60min, washed once and analyzed. The clones were also labeled separately with conventional
PE-conjugated tetramers loaded with the FLU peptide (open squares) or the ELA peptide (full squares). The results are expressed as mean fluorescent
intensity (MFI). Typical flow cytometry analyses of the ELA-specific CTL clone (B and C) and the FLU-specific CTL clone (E and F) are illustrated by
overlaying the histograms for unlabeled cells (dashed lines), cells labeled with irrelevant peptide (grey) and specific peptide (dark), labelled with either
dextramers (B and E) or conventional tetramers (C and F).
140 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

1000
incubated with fluoresceinated 270 kDa dextran
(Dex270F) (20μg/ml) mixed with different concentra-
tions of HLA-A2monomers preloaded either with FLU or
ELA peptides. Fig. 3 shows that while specific labeling
100 attained saturation at about 5μg/ml of monomers, the
unspecific labeling increased steeply at concentrations of
MFI

monomers ranging from 15 to 40μg/ml. Similar results

were obtained with dextran of higher molecular weight
10
(500kDa) or with ELA-specific clones, albeit with less
unspecific labeling. These data suggest that good
sensitivity is obtained with ratios of 4 to 20monomers
per dextran, with the optimum at about 10monomers per
dextran. For higher ratios, the unspecific labeling
1 increased and for lower ratios the specific labeling
0 15 30 45
decreased. Having defined the optimal monomer-to-
Monomer concentration (µg/ml)
dextran ratio, we tested the effect of different multi-
Fig. 3. Titration of HLA-A2/peptide monomers loaded onto a
merization conditions on the signal intensity on CTL
fluoresceinated dextran. Fluoresceinated dextran (270kDa) was pre- cells. Series of Melan-A dextramers made of 5μg/ml
mixed with different concentrations of HLA-A2/FLU or HLA-A2/ monomers or 20μg/ml were prepared under the four
ELA monomers. A FLU-specific CTL clone was incubated with these following multimerization conditions: 25min at room
dextramers at fixed concentration of dextran (20 μg/ml). The titrations temperature, 25min at 4°C, 10min at room temperature
of HLA-A2/FLU and HLA-A2/ELA monomers are represented by
open and full circles, respectively. The FLU-specific CTL clone was and 10min at 4°C. The molecules were then tested on two
also labeled with a conventional tetramer presenting FLU peptide irrelevant CTL clones, FLU- and EBV-specific, and on an
(open square) or ELA peptide (full square). ELA-specific one. The level of specific dextramer
staining was similar for all the preparations and the
saturated with HLA-A2 monomers (90 μg/ml) preloaded non-specific staining did not change when the tempera-
with peptides derived from either an influenza virus ture or the time of multimerization decreased (data not
protein or from the melanoma antigen Melan-A. The shown).
polymers were titrated, at room temperature for 60 min,
on a human CTL clone specific for either the Melan-A-
104

derived peptide ELAGIGILTV (“ELA”, Fig. 2A) or the

Addition of dextramers
influenza matrix-derived peptide GILGFVFTL (“FLU”,
Fig. 2D). Specific labeling was detected on both clones,
103

with saturation reached at dextramer concentrations

ranging between 2.5 to 10μg/ml. Fig. 2B and E show
typical results obtained with dextramers on the ELA and
FITC

FLU clones, respectively, with unlabeled cells (dashed

102

line), irrelevant peptide (grey) or specific peptide

(black).The labeling was comparable to that obtained
with conventional PE-conjugated tetramer for the ELA
101

clone (Fig. 2A–C) and three times lower for the FLU
clone (Fig. 2D–F). Higher unspecific labeling was
detected on the FLU clone when the Melan-A dextramer
was used.
100

0 15 30 45
3.3. Optimization of the monomer-to-dextran ratio and Time (min)
of the multimerization conditions
Fig. 4. Kinetics of labeling with dextramers. A FLU-specific CTL
Since high molecular weight dextrans can be associ- clone was continuously analyzed by flow cytometry at room
temperature. After few minutes of analysis, a fluoresceinated dextran
ated with more HLA-A2/peptide monomers than con- (500kDa) unsaturated with HLA-A2/FLU monomers (8 monomers per
ventional tetramers, we next defined the optimal amount dextran) was added at 7μg/ml and the acquisition of events continued
of monomers per dextran. A FLU-specific CTL clone was for the indicated time (X-axis).
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 141

3.4. Kinetics of dextramer labeling of antigen-specific HLA-A2/FLU monomers was added. Fig. 4 shows that
CTL and cell activation labeling reached saturation after 15 to 20min. Similar
results were obtained with an ELA-specific CTL clone or
To define the optimal time of labeling with dextra- by analyzing the samples after different time-points of
mers, a FLU clone was analyzed continuously by flow dextramer labeling followed by washing (data not
cytometry at room temperature. After few minutes, shown). To test whether dextramers may trigger T cell
Dex500F premixed with an optimal concentration of activation during the course of staining, we labeled three

Fig. 5. Pattern of dextramer labeling on different antigen-specific CTL clones. Four human CTL clones specific for FLU, ELA, CMVor EBV peptides
were analyzed for either autofluorescence or after labeling with fluorescent dextran (270kDa) unsaturated with HLA-A2/FLU, HLA-A2/ELA, HLA-
A2/CMV or HLA-A2/EBV monomers (10 monomers per dextran). Mean fluorescence intensities are given for unlabeled cells (top histogram row)
and dextramer labeled cells.
142 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

CTL clones specific for the FLU, EBV and ELA antigens tested on the same three clones but without dextramer. As
with the Melan-A ELA dextramer during either 30min or expected, specific T cell activation by Melan-A dextra-
4 h and at either 4°C, room temperature or 37°C. The mers was apparent only after 4 h of labeling at 37°C. This
appearance of the CD69 cell surface antigen was used as activation is defined by a specific increase of the CD69
an indicator of T cell activation. In the case of the MFI value of the specific clone vs the irrelevant ones and
4 h labeling, the fluorochrome conjugated anti-CD8 and the clones which didn't receive any dextramers. In
anti-CD69 antibodies were added after 3 h and 30min contrast, 4 h labeling at room temperature or at 4 °C did
incubation and in the case of the 30min labeling they not result in an detectable increase of the CD69 MFI
were added at t = 0, together with the dextramers. As value of the specific clone (data not shown). Therefore,
negative controls the same labeling conditions were 20 to 30min labeling at room temperature would be

Fig. 6. Analysis of HLA-A2+ PBMC with fluoresceinated dextramers. PBMC from two HLA-A2+ donors were labeled with either fluoresceinated
dextran (500kDa) unsaturated with HLA-A2/CMV monomers (8 monomers per dextran, A and C) or conventional tetramer presenting CMV peptide
(B and D). They were further labeled with an APC-conjugated antibody specific for CD8 and analyzed by flow cytometry. E shows the correlation
between the frequency of tetramer+ and dextramer+ cells in CD8+ cells obtained from different HLA-A2+ donors and different specificities (FLU,
CMV and EBV). Linear regression for these data (y = 1.0103 × − 10− 5, R2 = 0.9989) is also represented.
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 143

suitable for a wide panel of studies including the ones to FLU, ELA, CMV or EBV peptides. Fig. 5 shows an
assess the activation status of dextramer labeled cells. example of labeling with Dex270F presenting FLU,
ELA, CMV or EBV peptides (monomer-to-dextran ratio
3.5. Analysis of CTL clones with different antigen of 10). The labeling was highly specific for each CTL
specificities clone and unspecific labeling was very low (≤ 1 /
10,000), as demonstrated by labeling with dextramers
To further test their specificity, the labeling with dex- presenting irrelevant peptide. These results were con-
tramers was tested on four human CTL clones previously firmed with other clones (not shown). Similar results
isolated with conventional PE tetramers specific for were also obtained with Dex500F dextramers. These data

Fig. 7. Simultaneous analysis of four different antigen-specific CTL clones by five color flow cytometry. Four different CTL clones specific for
Influenza-, Melan-A-, CMV- and EBV-derived peptides were mixed and incubated with dextramerFITC, tetramerPE, tetramerECD and tetramerAPC
presenting the FLU, ELA, CMV and EBV peptides, respectively. After incubation, cells were labeled with an APC⁎Cy7-conjugated anti-CD8
antibody, washed and analyzed on a FACSVantage. A–H display the dextramer labeling on mixed CTL clones in histogram form (A–D) or as a dot-
plot (E–H). I–N demonstrate the dextramer/tetramer labeling on CD8 gated lymphocytes.
144 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

further confirm that specificity is well preserved when analyzed PBMCs from HLA-A2+ individuals. Fig. 6
the optimal monomer-to-dextran ratio is maintained. shows typical analyses of two PBMCs labeled with
Dex500F presenting HLA-A2/CMV-derived peptide
3.6. Analysis of PBMCs with dextramers and and CD8 antibody (Fig. 6A and C) as well as the
comparison with conventional tetramers analysis with conventional tetramer conjugated with PE
(Fig. 6B and D). In these examples, the intensity of
To probe the ability of dextramers to detect low dextramer labeling was similar to the intensity obtained
numbers of antigen-specific T cells in human tissues, we by labeling with PE-conjugated tetramers, relative to the

Fig. 8. Simultaneous analysis of four different antigen-specific CTL populations in PBMC by five color flow cytometry. Two PBMCs were mixed to
obtain the four different specificities at detectable levels (N0.01%). Cells were incubated with dextramerFITC, tetramerPE, tetramerECD and tetramerAPC
presenting the FLU, ELA, CMV and EBV peptides, respectively. After incubation, cells were labeled with an APC⁎Cy7-conjugated anti-CD8
antibody, washed and analyzed on a FACSVantage. A–H display the dextramer/tetramer labeling on mixed CTL clones in histogram form (A–D) or
as a dot-plot (E–H). I–N demonstrate the dextramer/tetramer labeling on CD8 gated lymphocytes.
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 145

background, but, in some cases, the intensity obtained APC one. ELA labeling shows a broad distribution as is
with these conventional tetramers was brighter. Excel- usual in PBMCs (Pittet et al., 1999). In the case of CMV
lent correlation between the percentage of tetramer + specificity, CD8+ CMV+ T cells were clearly detectable
cells and dextramer+ CD8+ T cells (Fig. 6E; linear with ECD-conjugated tetramers but the background was
regression: slope = 1.0103, intercept = − 10 − 5 , R 2 = high, as previously observed with clones. In addition,
0.9989, n = 20) was obtained by analyzing CD8+ T this labeling required ten times more tetramers in
cells of multiple HLA-A2+ donors with dextramers for
different specificities (FLU, CMV, EBV).
A ELA clone
The detection limit with fluoresceinated dextramers

100
was also evaluated. PBMCs from HLA-A2− donors
were analyzed with dextramers presenting different

80
autofluorescence
specificities (FLU, CMV, EBV). The mean, the standard

60
Counts
irrelevant peptide
deviation, the minimal and the maximal percentages of
dextramer+ cells in HLA⁎A2− CD8+ T cells were re-

40
specific peptide
spectively 0.0042%, 0.0034%, 0.0000% and 0.0108%.

20
The detection limit of dextramers was thus considered to
be about 1 for 10,000 cells, which is comparable to the

0
100 101 102 103 104
detection limit of conventional tetramers. FL2

3.7. Simultaneous analysis of four different B CMV clone

antigen-specific CTL populations by five color 120 150
flow cytometry

Next, we evaluated the combination of fluorescei-

90
Counts

nated dextramer with conventional PE and APC tet-

60

ramers to enumerate simultaneously three different

antigen-specific CTL populations. In addition to these
30

probes, we added a tetramer produced with an ECD-

0

conjugated streptavidin in order to test the feasibility of 100 101 102 103 104
a protocol to analyze simultaneously four different spe- FL2
C
cificities. FLU, ELA, CMV and EBV specificities were
detected respectively in the FITC (525 nm, dextran), PE
(575 nm), ECD (610 nm) and APC (675 nm) channels.
Additionally, an APC ⁎ Cy7 conjugated antibody
(780 nm) was used to define the CD8+ T cell population.
We first analyzed a cell suspension containing high
percentages (N 1%) of different antigen-specific CTL
clones prepared by mixing cultures of individual clones.
Fig. 7 shows that all specificities were well detected with
this protocol with only few aggregates (less than 2% for
each specificity). Each antigen-specific population was
clearly discriminated from the others, even in the ECD
channel where higher background was observed.
Then we analyzed PBMCs with lower percentages
(b1%) of antigen-specific CTLs. PBMCs containing
simultaneously significant amounts (N1 / 10,000) of
FLU-, CMV-, EBV- and ELA-specific CTLs are scarce. Fig. 9. Fluorescence microscope analysis with Alexa 532 dextramers
We thus used a mixture of 2 to 3 PBMCs to evaluate this conjugates. Two clones specific for Melan-A- (A) and CMV- (B)
protocol. Fig. 8 shows that four different antigen- derived peptides were either unlabeled (dashed line) or labeled with
irrelevant (grey) or specific peptide (black) dextramers and analyzed
specific CTL populations can be simultaneously by flow cytometry. C shows a picture of CMV-specific T cells labeled
detected in PBMCs. FLU-specific CTLs were adequate- with dextramers presenting CMV peptide and observed by fluorescent
ly detected in the FITC channel, as well as EBV in the microscopy.
146 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

comparison with clone analysis (data not shown). This fluorescent fluorochromes, such as fluorescein, would
could be explained by very high unspecific binding of open the possibility to either analyze simultaneously
ECD tetramers on CD8− populations, as illustrated in multiple CTL specificities or keep the PE and APC
Fig. 8. channels for the detection of poorly expressed antigens.
The increase of sensitivity in the fluorescein channel can
3.8. Microscopic analysis of antigen-specific CTL with be obtained by increasing the amount of fluorochromes
fluorescent dextramers on the specific probe. Alternatively, this hurdle could be
overcome by choosing brighter fluorochromes. Here,
Antigen-specific staining of CTL in tissues is a we investigated the former possibility by replacing one
second and more challenging application of this fluorescent (PE or APC) streptavidin conjugate associ-
technology. Dextramer staining in tissue sections has ated with four monomers (conventional tetramer) by a
been reported (Andersen et al., 2001) and dextran de- polyglucosidic backbone (dextran) associated with
rived reagents are widely used in clinical immunohis- multiple streptavidins (up to 17) and multiple fluor-
tochemistry (Tsutsumi et al., 1995; Richter et al., 1999). esceins (up to 29).
Preliminary observations of CTL clones labeled with Our experiments demonstrate that T cell labeling
fluoresceinated dextramers confirmed that specific label- with these dextramers is bright enough to allow
ing was detectable under microscope (data not shown). detection of antigen-specific T cells. The fluorescence
However, the photobleaching effect of FITC was too of fluorescein per se is weaker than that of PE or APC.
strong thus hampering appropriate examination of the However, our data show that conjugation of many
samples for more than a few minutes. We thus used fluoresceins to one dextran molecule rendered the dex-
a 270 kDa dextramer conjugated with Alexa 532, tramer as sensitive a tool as conventional PE- or APC-
which has better photostability than fluorescein. Two conjugated tetramers. Titration experiments also
CTL clones, specific for ELA and CMV peptides, were showed that the unspecific labeling increased when
labeled with either HLA-A2/ELA or HLA-A2/CMV dextrans are saturated with HLA-A2/peptide mono-
dextramers. The cells were first analyzed with a standard mers. In addition, the higher the molecular weight of
flow cytometer equipped with an argon laser to check the dextran is, the higher is the binding capacity for mo-
labeling. Although this laser is not optimal for Alexa nomers and the higher is the unspecific labeling. This is
532, specific labeling was at least ten fold higher than reminiscent of results obtained with the bright and large
autofluorescence or unspecific labeling (Fig. 9A and B). molecular weight fluorochrome called “PBXL-3L”
Some unspecific labeling was detected on the CMV (Telford et al., 2001, unpublished results). Nonetheless,
clone (Fig. 9B) but not on the ELA clone (Fig. 9A). The an optimal ratio of monomers per dextran was found
cells were then examined by fluorescence microscopy. (about 10 HLA-A2/peptide monomers per dextran).
Specific labeling was strong and photostable enough for Decreasing this ratio to values lower than 4 strongly
long (N one minute) examination. The unspecific label- decreased the labeling, while higher ratios resulted in
ing on CMV was nearly undetectable (data not shown). increased unspecific labeling. The incubation of cells at
Fig. 9C shows pictures of the CMV clone labeled with 37°C for 15 min followed by washing could also
the Alexa 532-conjugated HLA-A2/CMV dextramers. decrease preferentially the unspecific labeling (data
They show a membrane distribution of fluorescence with not shown). The replacement of wild type MHC class I
a punctate pattern. This cellular distribution is very monomer with mutants having reduced or abolished
similar to that observed with conventional tetramers ability to interact with CD8 (Choi et al., 2003; Pittet
(data not shown). et al., 2003) may also contribute to reduce the non-
specific labeling with multimers when using high
4. Discussion molecular weight dextran saturated with monomers.
Finally, under the described optimal conditions, dex-
Monitoring of T cell mediated immune responses tramers made of 270 and 500 kDa dextrans appeared
critically relies on the direct identification and charac- to be sensitive and specific probes as demonstrated
terization of antigen-specific T cells. Multicolor flow by the analysis of different clones and different
cytometry is currently a powerful approach to T cell specificities.
monitoring in both experimental animal models and Titration experiments showed an optimal concentra-
immunotherapy in humans. Until now, PE and APC tion of 10 μg/ml of dextramer. This is very close to the
have been the only fluorochromes used for the detection optimal concentration of conventional tetramers (about
of specific CTLs with tetramers. The use of other less 2μg/ml assuming a m.w. of about 400kDa), when
 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148 147

taking into account the increase in size due to the deposited on slide. The application of dextran con-
dextran polymer (total molecular weight of about jugates to tissue sections has been illustrated in one
2000 kDa for multimer prepared with 500kDa dextran). instance using Survivin derived dextramers (Andersen
The kinetics of binding demonstrated that saturation et al., 2001). However, the potential of dextramers
is reached after 15–20min incubation at room temper- to identify and enumerate antigen-specific CD8+ T
ature. This is also close to the behaviour of conventional cells in tissue sections remains to be carefully explored.
tetramers. Taken together, these data suggest that dex- Further, the availability of SA-HRP derived dextramers
tramers label T cells in a manner similar to conventional may enable combinations with tyramide-based ampli-
tetramers. If we assume a valency of four for the so- fication systems in both conventional and fluorescent
called “tetramers”, this may suggest that only few immunohistochemistry.
monomers out of the ten available contribute to the In conclusion, fluorescent, dextran-based multimers
labeling with the dextramer reagents. This is highly offer a flexible, user friendly and potent platform to
probable because not all monomers of a dextran con- extend the possibilities of both flow cytometry and
jugate are simultaneously accessible to the outer cell fluorescence-based microcopy for the monitoring of T
membrane. Nonetheless, the precise amount of HLA- cell mediated immune responses.
A2/peptide monomers engaged in the binding remains
unclear. First, as PE and APC tetramers may form Acknowledgements
unpredictable non-covalent aggregates (Guillaume
et al., 2003), this may practically result in a higher We thank Nathalie Rufer for the access to her large
valency value than the theoretical four monomers per bank of useful sorted CTLs and total PBMCs, and Linda
conjugate. Second, the preparation of dextran conjugate D. Thejl, Séverine Reynard, Andrée Porret and Céline
may also result in different conjugate products in terms Baroffio for their technical help.
of molecular weight and number of HLA-A2/peptide
monomers per conjugate. Selective binding of con- References
jugates with more than ten monomers to cells may thus
occur. Altman, J.D., et al., 1996. Phenotypic analysis of antigen-specific T
We demonstrate the first reported ex vivo enumer- lymphocytes. Science 274 (5284), 94.
ation of four different CTL antigen specificities in Andersen, M.H., et al., 2001. Spontaneous cytotoxic T-cell responses
simultaneous multicolor flow cytometry. The value of against survivin-derived MHC class I-restricted T-cell epitopes in
situ as well as ex vivo in cancer patients. Cancer Res. 61 (16),
this application may continue to grow as more antigen/ 5964.
MHC combinations are identified and move into clinical Appay, V., et al., 2002. Memory CD8+ T cells vary in differentiation
analysis. This protocol however requires some refine- phenotype in different persistent virus infections. Nat. Med. 8 (4),
ment and careful choice of reagent combinations. First, 379.
some specificities (e.g. ELA) may be characterized by Chilosi, M., et al., 1994. A rapid immunostaining method for frozen
sections. Biotech. Histochem. 69 (4), 235.
broad distribution of labeling and conventional, PE- Choi, E.M., et al., 2003. High avidity antigen-specific CTL identified
conjugated tetramers could be preferentially used in by CD8-independent tetramer staining. J. Immunol. 171 (10),
these cases. Second, conventional ECD-conjugated 5116.
tetramers gave high background and the discrimination Guillaume, P., et al., 2003. Soluble major histocompatibility complex-
peptide octamers with impaired CD8 binding selectively induce
between positive and negative populations was poor.
Fas-dependent apoptosis. J. Biol. Chem. 278 (7), 4500.
Multimers produced with ECD-conjugated dextran Klenerman, P., Cerundolo, V., Dunbar, P.R., 2002. Tracking T cells
might however improve this sensitivity, as suggested with tetramers: new tales from new tools. Nat. Rev. Immunol. 2
by the comparison of antibody directly labeled with (4), 263.
ECD and antibody conjugated with ECD-labeled Mortarini, R., et al., 2003. Lack of terminally differentiated tumor-
dextran (Siiman et al., 1999). specific CD8+ T cells at tumor site in spite of antitumor immunity
to self-antigens in human metastatic melanoma. Cancer Res. 63
Finally, it is noteworthy that dextran conjugated with (10), 2535.
Alexa 532 is sensitive and photostable enough for use Ogg, G.S., McMichael, A.J., 1998. HLA–peptide tetrameric com-
in microscopic analysis. Alexa 488 conjugates are plexes. Curr. Opin. Immunol. 10 (4), 393.
probably as efficient for such applications since the Pittet, M.J., et al., 1999. High frequencies of naive Melan-A/MART-1-
fluorochrome has similar excitation and emission specific CD8(+) T cells in a large proportion of human
histocompatibility leukocyte antigen (HLA)-A2 individuals. J.
spectra as FITC but higher photostability. In this Exp. Med. 190 (5), 705.
study, microscopic analyses were performed on cell Pittet, M.J., et al., 2001. Ex vivo IFN-gamma secretion by circulating
suspensions labeled with dextran multimers and then CD8 T lymphocytes: implications of a novel approach for T cell
148 P. Batard et al. / Journal of Immunological Methods 310 (2006) 136–148

monitoring in infectious and malignant diseases. J. Immunol. 166 Telford, W.G., et al., 2001. Cyanobacterial stabilized phycobilisomes
(12), 7634. as fluorochromes for extracellular antigen detection by flow
Pittet, M.J., et al., 2003. Alpha 3 domain mutants of peptide/MHC cytometry. J. Immunol. Methods 254 (1–2), 13.
class I multimers allow the selective isolation of high avidity Tsutsumi, Y., Serizawa, A., Kawai, K., 1995. Enhanced polymer one-
tumor-reactive CD8 T cells. J. Immunol. 171 (4), 1844. step staining (EPOS) for proliferating cell nuclear antigen (PCNA)
Richter, T., et al., 1999. Protocol for ultrarapid immunostaining of and Ki-67 antigen: application to intra-operative frozen diagnosis.
frozen sections. J. Clin. Pathol. 52 (6), 461. Pathol. Int. 45 (2), 108.
Rufer, N., et al., 2003. Ex vivo characterization of human CD8+ T Turner, H.E., et al., 1999. The enhanced peroxidase one step method
subsets with distinct replicative history and partial effector increases sensitivity for detection of Ki-67 in pituitary tumours.
functions. Blood 102 (5), 1779. J. Clin. Pathol. 52 (8), 624.
Sabattini, E., et al., 1998. The EnVision++ system: a new immunohis- Wolfl, M., et al., 2004. Quantitation of MHC tetramer-positive cells
tochemical method for diagnostics and research. Critical compar- from whole blood: evaluation of a single-platform, six-parameter
ison with the APAAP, ChemMate, CSA, LABC, and SABC flow cytometric method. Cytometry A 57 (2), 120.
techniques. J. Clin. Pathol. 51 (7), 506. Yamashita, K., et al., 1997. A multimeric form of soluble recombinant
Siiman, O., et al., 1999. Fluorescent neoglycoproteins: antibody– sheep LFA-3 (CD58) inhibits human T-cell proliferation. Immu-
aminodextran–phycobiliprotein conjugates. Bioconjug. Chem. 10 nology 92 (1), 39.
(6), 1090. Zippelius, A., et al., 2004. Effector function of human tumor-specific
Speiser, D.E., et al., 2002. In vivo activation of melanoma-specific CD8 T cells in melanoma lesions: a state of local functional
CD8(+) T cells by endogenous tumor antigen and peptide vaccines. tolerance. Cancer Res. 64 (8), 2865.
A comparison to virus-specific T cells. Eur. J. Immunol. 32 (3),
731.