sensitive and viable identification of antigen-specific cd8+ t cells
TRANSCRIPT
www.elsevier.com/locate/jim
Journal of Immunological Methods 281 (2003) 65–78
Sensitive and viable identification of antigen-specific CD8+ T cells
by a flow cytometric assay for degranulation
Michael R. Betts*, Jason M. Brenchley, David A. Price, Stephen C. De Rosa,Daniel C. Douek, Mario Roederer, Richard A. Koup
Laboratory of Immunology, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health,
40 Convent Drive, Bethesda, MD 20892, USA
Received 22 January 2003; received in revised form 25 June 2003; accepted 7 July 2003
Abstract
Flow cytometric detection of antigen-specific CD8+ T cells has previously been limited to MHC-class I tetramer staining or
intracellular cytokine production, neither of whichmeasure the cytolytic potential of these cells. Here we present a novel technique
to enumerate antigen-specific CD8+ T cells using a marker expressed on the cell surface following activation induced
degranulation, a necessary precursor of cytolysis. This assay measures the exposure of CD107a and b, present in the membrane of
cytotoxic granules, onto the cell surface as a result of degranulation. Acquisition of cell surface CD107a and b is associated with
loss of intracellular perforin and is inhibited by colchicine, indicating that exposure of CD107a and b to the cell surface is
dependent on degranulation. CD107a and b are expressed on the cell surface of CD8+ T cells following activation with cognate
peptide, concordant with production of intracellular IFNg. Finally, CD107-expressing CD8+ T cells are shown to mediate
cytolytic activity in an antigen-specific manner. Measurement of CD107a and b expression can also be combined withMHC-class
I tetramer labeling and intracellular cytokine staining to provide a more complete assessment of the functionality of CD8 +Tcells
expressing cognate T cell receptors (TCR).
D 2003 Elsevier B.V. All rights reserved.
Keywords: Degranulation; T lymphocyte; Intracellular cytokine; CD107a; CD107b
0022-1759/$ - see front matter D 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0022-1759(03)00265-5
Abbreviations: APC, allophycocyanin; CFSE, carboxyfluorescein diacetate succinimidyl ester; CMTMR, chloromethyl-benzoyl-amino-
tetramethyl-rhodamine; CMV, cytomegalovirus; CTL, cytotoxic T lymphocyte; ELISpot, enzyme-linked immunospot; FCS, fetal calf serum;
FITC, fluorescein isothiocynate; ICS, intracellular cytokine staining; HIV, human immunodeficiency virus; HLA, human leukocyte antigen;
IFNg, interferon-gamma; LAMP, lysosomal associated membrane protein; MHC, major histocompatibility complex; MIP1h, macrophage
inflammatory protein-1-beta; NK, natural killer cell; PBMC, peripheral blood mononuclear cells; PE, phycoerythrin; PerCP, peridinin chlorophyll
protein; PHA, phytohemagglutinin; qPCR, quantitative polymerase chain reaction; SEB, staphylococcus enterotoxin-B; TNFa, tumor necrosis
factor alpha.
* Corresponding author. Tel.: +1-301-594-8612; fax: +1-301-480-2779.
E-mail address: [email protected] (M.R. Betts).
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7866
1. Introduction
After MHC-mediated recognition of cognate pep-
tide, CD8+ cytotoxic T lymphocytes exhibit two gen-
eral effector functions: production of soluble factors,
including cytokines and chemokines, and target cell
killing. Recent technological advances in multiparam-
eter flow cytometry have enabled the enumeration of
antigen-specific CD8+ T cells (MHC-class I tetramer
staining) and a direct assessment of their ability to
produce cytokine (intracellular cytokine staining
(ICS)) (Altman et al., 1996; Kern et al., 1998; Appay
et al., 2000; Betts et al., 2000). These techniques have
advantages over other methods, such as ELISpot anal-
ysis, in that they allow for precise phenotypic charac-
terization of the T cell populations of interest. CD8+ T
cell-mediated target cell killing, however, has histori-
cally been assessed by the standard chromium (51Cr)
release assay (Brunner et al., 1968), or, more recently,
by methods that monitor the release of fluorescent dyes
from target cells (Sheehy et al., 2001; Liu et al., 2002).
These techniques are cumbersome, semi-quantitative,
and potentially insensitive. Importantly, none of these
methods directly examine the CD8+ T cells that medi-
ate killing; rather, they examine the death of target
cells, essentially the aftermath of CD8+ T cell effector
function.
Cytotoxic CD8+ T lymphocytes (CTL) mediate the
killing of target cells via two major pathways, a
granule-dependent (perforin/granzyme) and indepen-
dent (ligand–ligand induced cell death, e.g. fas-fasL)
mechanism (reviewed in Trapani and Smyth, 2002).
The granule-dependent pathway does not require de
novo synthesis of proteins by the effector CD8+ T cell,
which instead utilize pre-formed lytic granules located
within the cytoplasm (Trapani and Smyth, 2002). The
lytic granules are membrane-bound secretory lyso-
somes that contain a dense core composed of various
proteins, including perforin and granzymes (Peters et
al., 1991). The core is surrounded by a lipid bilayer
containing lysosomal associated membrane glycopro-
teins (LAMPs), including CD107a (LAMP-1),
CD107b (LAMP-2), and CD63 (LAMP-3) (Peters et
al., 1991). These proteins are not normally found on the
surface of Tcells, although they have been observed on
the cell surface of PHA-activated lymphocytes
(CD107a and b), and ionomycin treated CD4+ and
CD8+ CTL clones (CD63) (Kannan et al., 1996; Bossi
and Griffiths, 1999). Although CD107a and b expres-
sion on the cell surface of peripheral blood cells has
been shown to enhance lymphocyte vascular adhesion
(Kannan et al., 1996), the function of these proteins, if
any, on the surface of activated T cells remains to be
determined. The presence of CD107a and b in the
cytotoxic granular membrane has been proposed to
protect against the leakage of contents from the granule
itself by coating the interior of the membrane (Peters et
al., 1991). CD107a, CD107b, and CD63 are constitu-
tively expressed on the surface of activated platelets,
and are found ubiquitously in lysosomal and endosomal
membranes of numerous other cell types (reviewed in
Fukuda, 1991).
Degranulation of activated CD8+ T cells occurs
rapidly after TCR stimulation, as a result of the polar-
ized mobilization of microtubules that transport the
lytic granules towards the immunological synapse
formed between the CTL and the target (reviewed in
Barry and Bleackley, 2002). Once the granules reach
the plasma membrane of the CTL, the membranes fuse
(Peters et al., 1991), releasing the granule contents into
the immunological synapse, ultimately resulting in the
death of the target cell. Degranulation is a requisite
process of perforin-granzymemediated killing, and is a
critical step required for immediate lytic function
mediated by responding antigen-specific CD8+ T cells.
To date, there are no flow cytometric assays to monitor
this process directly. Monitoring the presence or ab-
sence of perforin and granzymes A and B gives no
indication as to the ability of T cells to degranulate.
Furthermore, identifying a cell as lacking perforin does
not mean it had perforin prior to stimulation. Finally,
CD8+ T cells specific for certain viral antigens have
low baseline levels of perforin (Appay et al., 2000,
2002; Sandberg et al., 2001), causing difficulty in
measuring a further loss after antigen stimulation.
We developed a novel assay that directly measures
degranulation in primary responding antigen-specific
CD8+ T cells by multi-parameter flow cytometry. This
assay measures the cumulative exposure of granular
membrane proteins (CD107a and b) on the cell surface
of responding antigen-specific T cells, providing a
positive marker of degranulation. Significant expres-
sion of cell surface CD107a and b can be observed as
early as 30 min following stimulation of primary CD8+
T cells, and reaches maximum by 4 h. As expected,
inhibition of degranulation dramatically reduces the
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 67
acquisition of cell surface CD107a and b. This assay
can be combined with existing methods that assess
cytokine production in responding antigen-specific
CD8+ T cells directly ex vivo, thus providing simulta-
neous assessment of two critical CD8+ T cell effector
functions.
2. Materials and methods
2.1. Patient samples
Peripheral blood mononuclear cells (PBMC) were
obtained from 11 anonymous healthy donors at the
National Institutes of Health Department of Transfu-
sion Medicine. PBMC from two HIV-1 infected
patients were obtained from both the Amelia Court
HIV Clinic at the University of Texas Southwestern
Medical Center and the National Institutes of Health
OP-8 and OP-11 clinics. These donors signed informed
consent required by the institutions’ Review Boards.
2.2. Antibodies and reagents
The following anti-human monoclonal antibody
reagents were obtained from BD Pharmingen, San
Diego, CA (purified antibodies to some cell-surface
markers were conjugated to various fluorochomes in
our laboratory (Roederer), as indicated by an asterisk
following the fluorochrome used): anti-CD28, anti-
CD49d, anti-IFN-g [fluorescein isothicyanate (FITC),
allophycocyanin (APC)], anti-CD3 [phycoerythrin
(PE), peridinin chlorophyll protein (PerCP)], anti-
CD4 (FITC), anti-CD8 (PerCP), anti-CD16 (FITC),
anti-CD20 (FITC), anti-CD107a (FITC, PE*, APC*),
anti-CD107b (FITC, PE*, APC*), anti-CD63 (PE),
and anti-perforin (PE). Anti-granzyme B conjugated
to PE was obtained from Caltag, Burlingame, CA.
CMV-A2 tetramers were obtained from Beckman
Coulter Immunomics, San Diego, CA.
2.3. Cell stimulation
Fresh PBMC were isolated using Hypaque-Ficoll
(Pharmacia, Uppsala, Sweden) density centrifugation.
In some instances PBMC were frozen (90% fetal calf
serum (FCS)/10% DMSO) at � 140 jC until use. 106
PBMC were incubated with 1 Ag/ml each of anti-
CD28 and anti-CD49d and 2 Ag/ml of appropriate
peptide (when used) in a 1-ml volume. In some experi-
ments, staphylococcus enterotoxin B (SEB, 1 Ag/ml,
Sigma, St. Louis, MO) or anti-CD3 (clone HIT3a, BD
Pharmingen, 5 Ag/ml) was used to activate the cells.
Conjugated antibodies to the granular membrane pro-
teins CD107a and CD107b were added to the cells
prior to stimulation, unless otherwise noted. In every
experiment a negative control (anti-CD28/CD49d)
was included to control for spontaneous production
of cytokine and/or expression of CD107a/b. The
cultures were incubated for 1 h at 37 jC in a 5%
CO2 incubator, followed by an additional 4–5 h in the
presence of the secretion inhibitor monensin (BD
Pharmingen) or Brefeldin A (Sigma). In some experi-
ments, colchicine (Sigma) was added to inhibit granule
release.
2.4. Immunofluorescent staining/analysis
Immediately following stimulation, PBMC were
washed once, and surface stained with directly con-
jugated antibodies. The cells were washed and then
fix/permeabilized using 750 Al of fixation/permeab-
lization solution, consisting of FacsLyse (Becton
Dickinson Immunocytometry Systems, San Jose,
CA) diluted to a 2� concentration in dH2O and
0.05% Tween-20 (Sigma). After permeabilization,
the cells were washed twice, and stained with
directly conjugated antibodies specific for intracellu-
lar markers. The cells were washed a final time and
resuspended in 1% paraformaldehyde (Electron Mi-
croscopy Systems, Fort Washington, PA) in PBS.
Six-parameter flow cytometric analysis was per-
formed using a FACSCalibur flow cytometer (Becton
Dickinson Immunocytometry Systems). List mode
data files were analyzed using FlowJo software (Tree
Star, San Carlos, CA). In all cases at least 100,000
live events were collected for analysis.
2.5. Direct ex vivo cytotoxicity assay
Autologous B cells were isolated from PBMC by
positive selection using magnetic Microbeads coated
with anti-CD19 monoclonal antibody according to the
manufacturer’s instructions (MACS, Miltenyi Biotec,
Germany). Purified B cells were then labeled with
either carboxyfluorescein diacetate succinimidyl ester
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7868
(CFSE) or chloromethyl-benzoyl-amino-tetramethyl-
rhodamine (CMTMR) (Molecular Probes, Eugene,
OR). For CFSE staining, 5� 106 cells/ml were incu-
bated with 0.25 AM CFSE in PBS at 37 jC for 7 min,
then washed three times in RPMI-1640 (BioWhittaker,
Walkersville, MD)/20% FCS. For CMTMR staining,
2� 106 cells/ml were incubated in pre-warmed RPMI/
10% FCS (R10) supplemented with 5 AM CMTMR at
37 jC for 30 min, washed twice in pre-warmed R10,
and then incubated in R10 alone for 1 h prior to a final
wash. Labeled cells were protected from light during
all subsequent procedures. Cells labeled with CFSE
were either pulsed for 90 min with 200 nM CMV pp65
peptide 495–503 (NLVPMVATV) and then washed
three times in R10, or mock-pulsed in parallel; all cells
labeled with CMTMR were mock-pulsed. Cytotoxic-
ity assays used 200,000 each of CFSE and CMTMR-
labeled B cells in FACS tubes; PBMC were added in
R10 to give a range of effector to target (E/T) ratios.
Assays were incubated at 37 jC with 5% CO2 with the
tubes placed at an angle. The elimination of peptide-
pulsed CFSE-labeled cells relative to the unpulsed
CMTMR-labeled internal negative control served as
a measure of specific cytotoxicity. Parallel assays with
mock-pulsed CFSE-labeled cells were set up for each
E/T ratio to control for non-specific toxic effects of
CFSE itself.
3. Results
3.1. CD107a and b are expressed on the surface of
CD8+ T cells that degranulate in response to SEB
In order to identify degranulation by activated
CD8+ T cells, we examined the expression of
Fig. 1. Characterization of CD107a and b staining. (A) PBMC from a nor
antibodies to CD107a and b FITC as shown in the presence of Brefeldin A
(where appropriate). Events shown are gated for CD3 and CD8. (B) A co
monensin (solid line) on the fluorescence of FITC CD107a and b in SEB
FITC anti-CD107a and b for 6 h in the presence of either inhibitor, then sta
and CD8. Values shown represent the mean fluorescence of the indicated p
Cells from one donor were SEB-stimulated, pre-stained with anti-CD107a
for 6 h, followed by staining of CD3 and CD8 molecules. Events shown ar
on SEB stimulated cells. Cells from the same donor used in (C) were surfac
described in (C) and stained with CD3 FITC and CD8 PerCP. Events
degranulation in SEB stimulated PBMC. Cells were stimulated, stained, an
concentrations of colchicine, as depicted on the figure.
CD107a and b on the cell surface of CD8+ T cells
following activation with SEB. PBMC from a normal
donor were stimulated with SEB and incubated for 6
h in the presence of the secretion inhibitor Brefeldin
A. After stimulation, the cells were stained with
antibodies to T cell markers (CD3 and CD8) and a
mixture of FITC-labeled antibodies to CD107a and b.
A total of 8.4% of CD8+ T cells expressed surface
CD107 after stimulation (Fig. 1A, left panel), indicat-
ing that degranulation had likely occurred within the
responding CD8+ T cells. While this initial result
appeared promising, the frequency of the responding
population was not comparable to that observed by
intracellular cytokine staining for IFNg (approximate-
ly 12%, data not shown). We reasoned that it was
likely that cell surface expression of CD107 on T cells
may be transient. Previous studies have shown that
CD107a and b are targeted primarily to lysosomal/
endosomal membranes, and that any CD107a and b
externalized to the cell surface is rapidly retrieved via
the endocytic pathway (Fukuda, 1991). This sug-
gested that as the cells degranulate, they would
become positive for cell surface CD107 for a brief
period of time before those proteins were internalized.
Therefore, we included the antibodies to CD107a and
b for the duration of the stimulation, rather than just
staining post-stimulation. Any transient surface ex-
pression of CD107 would lead to antibody binding
and either surface retention or uptake of the CD107/
antibody complex. In either case, even transient sur-
face expression of CD107 would lead to fluorescent
labeling of that cell. This modification enhanced the
detection of responding cells, such that 12% of the
CD8+ T cells were observed to express surface
CD107a and b in response to SEB (Fig. 1A, center
panel). This protocol did not result in an increase of
mal donor were stimulated with SEB and incubated with or without
for 6 h, then stained with CD3, CD8, and CD107a and b antibodies
mparison of the differential effects of Brefeldin A (dashed line) and
stimulated CD8+ T cells. Cells were stimulated in the presence of
ined with CD3 and CD8 antibodies. Events shown are gated for CD3
opulation (C) Comparison of different CD107 antibody conjugates.
and b FITC, PE, or APC, and incubated in the presence of monensin
e gated for CD3 and CD8. (D) Coordinate staining of CD107a and b
e stained with CD107a APC and CD107b PE, and then stimulated as
shown are gated for CD3 and CD8. (E) Effect of colchicine on
d analyzed as described in (C) in the presence or absence of varying
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 69
background staining for CD107a and b (Fig. 1A, right
panel).
Our initial experiments utilized antibodies to
CD107a and b conjugated to FITC, the fluorescence
of which is sensitive to acidic pH (Roederer et al.,
1987). Because CD107 is internalized from the cell
surface following degranulation, likely into an acidic
endosomal or lysosomal compartment, the fluores-
cence of FITC would be quenched. Therefore we
examined the differential effects of Brefeldin A and
monensin on the mean fluorescence of the responding
cells (Fig. 1B). While both Brefeldin A and monensin
serve as inhibitors of secretion, monensin also neu-
tralizes the pH within endosomes and lysosomes
(Mollenhauer et al., 1990). The mean fluorescence
of CD107a and b-FITC in activated cells incubated
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7870
with monensin (mean fluorescence = 78.3) was sub-
stantially higher than cells incubated with Brefeldin A
(mean fluorescence = 41.8)(Fig. 1B). This result sug-
gests that a substantial proportion of the FITC-labeled
CD107a and b antibodies were internalized into an
endosomal or lysosomal compartment, further sup-
porting the need to label these proteins during stim-
ulation and use monensin to achieve optimal detection
of responding CD8+ T cells.
In order to further optimize the staining of CD107,
we conjugated the CD107a and b antibodies to PE and
APC. The PE and APC conjugates were substantially
brighter than the FITC conjugate, without greatly
affecting the fluorescence of the non-responding pop-
ulation (Fig. 1C). One observation of note is that the
PE-CD107 antibody conjugate appears to have a
higher background in non-stimulated cells than the
FITC and APC conjugates (Fig. 1C, upper row),
perhaps related to the hydrophobicity of PE. We
therefore utilized either the FITC or APC CD107
conjugates in our remaining experiments. In general,
background expression of CD107a and b in unstimu-
lated CD8+ T cells varied between 0.05% and 0.5%.
The background expression of CD107a and b can be a
result of several factors, including the presence of
dead/apoptotic cells, platelets, granulocytes, mono-
cytes, or B cells within the cultures. In particular,
removal of monocytes and B cells through the use of
a dump channel can dramatically reduce the back-
ground CD107 labeling observed (data not shown).
We next examined the coordinate expression of
CD107a and b on the cell surface of SEB-activated
CD8+ T cells. Both CD107a and b are coordinately
expressed on the majority of the responding cells,
although the CD107a signal appears to be stronger in
most cells than CD107b with these reagents (Fig. 1D).
CD107a and b are likely differentially regulated within
all cell types, as they are encoded on different chro-
mosomes, and expressed at different copy numbers
within the cell (Fukuda, 1991). It remains to be deter-
mined if these two proteins are differentially expressed
in cytotoxic granules within individual CD8+ Tcells or
subpopulations of CD8+ T cells. Therefore, to ensure
greatest sensitivity, we chose to use amixture of the two
antibodies in future experiments.
In order to ensure that CD107a and b expression on
the cell surface occurred as a result of degranulation,
we examined the effect of the microtubule inhibitor
colchicine on the expression of CD107a and b.
Colchicine has a potent effect on the expression of
cell surface CD107a and b following activation of
CD8+ T cells with SEB (Fig. 1E). This result, along
with the observation that CD107a and b are expressed
on the cell surface in the presence of the secretion
inhibitors Brefeldin A and monensin, indicates that
CD107a and b are indeed expressed on the cell
surface as a result of degranulation.
3.2. Cell surface expression of CD107a and b is
associated with the loss of intracellular perforin
Perforin is released from cytotoxic granules during
the degranulation process; therefore, acquisition of
cell surface CD107a and b on CD8+ T cells after
stimulation should be associated with a loss of intra-
cellular perforin. To address this, we stimulated
PBMC with anti-CD3, and performed a time course
to compare the levels of intracellular perforin with cell
surface CD107a and b. As shown in Fig. 2, CD107a
and b was rapidly expressed after stimulation, con-
comitant with a loss of perforin expression. After a 5-
h stimulation, nearly every perforin expressing cell
had degranulated, indicating that acquisition of cell
surface CD107a and b occurs as a result of degranu-
lation, rather than de novo production.
3.3. Comparison of CD107a and b expression with
CD63 expression following degranulation
CD63 (LAMP-3) can also be found within the
membrane of cytotoxic granules (Peters et al., 1991),
and has been previously used as a positive marker
associated with the deposition of fas ligand on the cell
surface during degranulation in ionomycin-stimulated
CD4+ and CD8+ T cell clones (Bossi and Griffiths,
1999). We examined the expression of CD63 in
comparison with CD107a and b in CD8+ T cells that
degranulate in response to SEB (Fig. 3). While a
portion of the responding CD8+ T cells expressed
CD63, CD107a, and CD107b, the majority of respond-
ing cells were CD63 low or negative. Furthermore, the
CD63 background was substantially higher than that
observed for CD107a and b. These results show that
CD107 expression on ex vivo activated CD8+ T cells
provides a more accurate assessment of degranulation
than does CD63 expression.
Fig. 2. Acquisition of cell surface CD107 is correlated with a loss of intracellular perforin. PBMC were stimulated with anti-CD3 (5 Ag/ml), and
incubated for up to 5 h in the presence of anti-CD107a and b FITC and monensin. Baseline perforin expression was approximately 20% (data
not shown). At the time-points designated on the figure, aliquots were removed, washed, and permeabilized, followed by staining for perforin,
CD3, and CD8. Events shown are gated for live CD3+ CD8+ T cells.
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 71
3.4. Cytotoxic-granule containing lymphocytes
express higher levels of CD107a and b than
non-granule containing lymphocytes
While all lymphocytes express CD107a and b on
endosomal and lysosomal membranes, not all lympho-
cytes have cytotoxic granules that contain perforin.
Such granules are expressed by CD8+ Tcells, NK cells,
and a subset of CD4+ T cells. We therefore examined
the relationship between intracellular expression of
CD107 and perforin expression. Nearly all lympho-
cytes express detectable CD107a and b when staining
for these molecules following permeabilization (data
not shown). Cells that contain perforin, however,
express even higher levels of CD107a and b. For
example, NK cells, which typically express high levels
Fig. 3. Comparison of CD63 expression with CD107a and b expression. PBMC were stimulated with SEB in the presence of antibodies to CD63
PE and CD107a and b FITC and monensin, incubated for 6 h, and then stained with CD3 and CD8 antibodies. Events shown are gated CD3+
CD8+ T cells.
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7872
of perforin, have higher CD107a and b expression than
B cells or CD4+ T cells. Interestingly, although only a
subset of CD8+ T cells is perforin+, the CD8+ T cell
population as a whole expresses more CD107a and b
than B cells or CD4+ T cells. This suggests that CD8+
T cells have a higher granular content, similar to NK
cells. To support this conclusion, we also compared
granzyme B, another granular component expressed by
a higher percentage of CD8+ Tcells than perforin, with
CD107a and b content. CD8+ T cells and NK cells that
express granzyme B also have high levels of CD107a
and b (data not shown).
3.5. CD107a and b are expressed on the cell surface
of activated peptide-specific CD8+ T cells
The standard for examining antigen-specific CD8+
T cell effector responses by flow cytometry is mea-
surement of intracellular cytokine production, typical-
ly IFNg, after stimulation with cognate peptide. We
therefore compared production of IFNg in response to
peptide stimulation with acquisition of cell surface
CD107a and b (Fig. 4A–D). PBMC isolated from
four different patients were stimulated with peptides
derived from either CMVor HIV, stained for CD107a
and b, then incubated for 5 h in the presence of
monensin. After stimulation, the cells were permeabi-
lized and stained for intracellular IFNg. As can be seen
in Fig. 4, nearly all of the CD8+ T cells that produce
IFNg in response to specific peptide also express
CD107a and b. This indicates that acquisition of cell
surface CD107a and b occurs in an antigen specific
manner, and that nearly all CD8+ T cells which
produce cytokine in response to cognate antigen
degranulate. As shown in Fig. 4C, a mixture of
peptides can also be used to stimulate CD8+ T cells
to degranulate, suggesting that measurement of cell
surface CD107a and b could be used in peptide
response mapping procedures. Interestingly, in some
patients a population of CD107a+ and b+ cells that did
not produce IFNg could be observed (Fig. 4B and C).
This suggests that measuring CD8+ T cell responses
by IFNc production alone may underestimate the total
response.
3.6. Kinetics of CD107a and b expression as
compared to IFNg production
Having established that cell surface expression of
CD107a and b occurs in an antigen specific manner,
we examined in more detail the rate at which cells
became surface positive for CD107a and b after
stimulation. Intracellular expression of IFNg has
previously been shown to plateau between 5 and 6
h, thus we previously had stimulated the cells for this
length of time to compare CD107 and IFNg expres-
sion. It is known, however, that degranulation occurs
very rapidly following triggering of the T cell recep-
tor complex. We therefore expected to detect acqui-
sition of cell surface CD107a and b rapidly follo-
Fig. 4. CD107a and b is expressed by ex vivo activated antigen-specific CD8+ T cells. PBMC from four different donors were stimulated with
specific peptides in the presence of anti-CD28/49d, anti-CD107a and b FITC or APC, and monensin for 6 h. Peptides used include HLA-A2
restricted CMV pp65 NLVPMVATV (A, B), overlapping HIV-Gag derived 15-mers (15-mer peptides overlapping by 11 amino acids, C), and
HLA-B57 restricted HIV Gag KAFSPEVIPMF (D). All peptides were used at a final concentration of 2 Ag/ml. Events shown are gated CD3+
CD8+ T cells.
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 73
wing stimulation. Expression of cell surface CD107
can be detected as early as 30 min to 1 h follow-
ing stimulation with anti-CD3 (Fig. 2) or specific
peptide (Fig. 5), and peaks between 4 and 5 h. IFNg,
which requires de novo synthesis following activa-
tion, also peaked at 4–5 h post stimulation, although
no IFNg production was detected at 1 h post-stimu-
lation. Thus, optimal expression of both cell surface
CD107a and b and IFNg occurs between 4 and 6 h
post-stimulation.
3.7. CD107a and b can be used to measure
degranulation in activated MHC-class I
tetramer+cells
The functionality of tetramer-binding CD8+ T cells
can be examined by staining for IFNg production after
the stimulation of tetramer stained cells with cognate
peptide. However, since cytokine production alone
does not provide a full functional assessment of the
tetramer binding cells, we examined the ability of
tetramer stained cells to degranulate, an example of
which is shown in Fig. 6. Approximately 0.8% of
CD8+ T cells in this individual are capable of binding
to the CMV-A2 MHC-class I tetramer (Fig. 6, left
panel). After stimulation with peptide, only 25% of the
tetramer + cells produced IFNg (Fig. 6, center panel),
while a substantially higher proportion (>50%) of the
same tetramer + cells degranulated following stimula-
tion (Fig. 6, right panel). Similar results were observed
in four additional individuals (data not shown). These
data demonstrate that examination of cytokine produc-
tion alone may not provide sufficient information
regarding the full functionality of tetramer + cells.
3.8. CD107a and b expression directly correlates with
cytotoxic activity
While measurement of the ability of CD8+ T cells
to degranulate provides an indication of cytotoxic
potential, it still does not prove that degranulating
cells are capable of killing targets. To address this
question more directly, we examined the cytotoxic
activity of a CMV-specific CD8+ T cell population
known to degranulate in response to an HLA-A2
restricted CMV-derived peptide (NLVPMVATV,
Fig. 5. Time course of CD107a and b expression as compared to IFNg production. PBMC were stimulated with CMV-A2 NLVPMVATV
peptide (2 Ag/ml) for up to 6 h in the presence of anti-CD28/49d, anti-CD107a and b FITC and monensin. At 1, 2, 3, 4 and 6 h post stimulation,
aliquots were removed, washed, permeabilized and stained for intracellular IFNg, CD8, and CD3. The frequency of responding CD8+ cells was
determined at each time point and the % maximal response was calculated for each time point. (x, CD3+ CD8+ IFNg+ cells; n, CD3+ CD8+
CD107a and b+ cells; D, CD3+ CD8+ CD107a and b+ IFNg+ cells).
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7874
CMV-A2). PBMC were isolated from an individual
with a high CMV-specific CD8+ T cell frequency (Fig.
7A, left panel). Approximately 10–15% of the circu-
lating CD8+ T cells in this individual are specific for
the CMV-A2 peptide as measured by direct tetramer
binding, and 80% of the tetramer + cells degranulate in
response to the CMV-A2 peptide (Fig. 7A, right
panel). We then examined the cytotoxic ability of these
cells directly by using a flow cytometry-based killing
assay, as described in the Materials and methods.
CFSE labeled cells, depicted in green, are selectively
Fig. 6. CD107a and b are expressed by functional MHC-class I tetramer
complexes, then activated with CMV-A2 peptide (NLVPMVATV) in the pr
5 h. Following stimulation, the cells were washed, permeabilized, and sta
killed by the CD8+ T cells only when CMV-A2
peptide loaded, as demonstrated by a shift into the
dead cell population determined by the forward/side
scatter profiles (Fig. 7B). The selective loss of peptide
loaded CFSE labeled B cells in the presence of CD8+
T cells can also be directly observed by comparing the
frequency of CMTMR and CFSE labeled cells within
the live cell gate (Fig. 7C). Taken together these results
indicate that antigen-specific CD8+ T cells that degra-
nulate, as measured by CD107 expression, mediate
cytotoxic activity.
+ cells. PBMC were stained with CMV-A2 MHC-class I tetrameric
esence of anti-CD28/49d, anti-CD107a and b APC and monensin for
ined for CD8 and IFNg. All events shown are gated CD8+ cells.
Fig. 7. Correlation between CD107a and b expression and cytotoxic activity. (A) PBMC were stained with a CMV-A2 MHC-class I tetramer,
and anti-CD107a and b APC and treated as follows: Left panel: stained with anti-CD3 and anti-CD8, and analyzed without further incubation.
Right panel: stimulated with cognate peptide (NLVPMVATV), incubated for 5 h in the presence of monensin, washed, then stained with anti-
CD3 and anti-CD8. All events shown are live-gated on CD3 and CD8. (B and C) Multiple parameter overlay plots showing the distribution of
CFSE and CMTMR-labeled autologous B cells within the forward/side scatter profile (B, green events = CFSE, red events = CMTMR) or live
gate (as depicted in B) only (C). In the example shown, 2� 106 PBMC were incubated with 200,000 target cells for 60 h. Left panel: neither
population of fluorescent dye-labeled cells was pulsed with cognate peptide. Right panel: the target cells labeled with CFSE only were pre-
pulsed with cognate peptide (NLVPMVATV) at 200 nM.
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 75
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7876
4. Discussion
We present here a novel assay that measures de-
granulation in response to antigen-specific stimulation,
an essential CD8+ T cell effector function. We dem-
onstrate that the granular membrane proteins CD107a
and b are expressed on the cell surface of activated
CD8+ T cells due to degranulation. Expression of cell
surface CD107a and b on CD8+ T cells often occurs in
concert with production of IFNg in response to stim-
ulation, but both functions do not necessarily occur in
all antigen-specific CD8+ T cells. Additionally, we
show that the same CD8+ T cells which degranulate
are capable of cytotoxic activity.
Numerous methods exist to examine both the phys-
ical presence and functionality of antigen-specific
CD8+ T cells. The physical presence of antigen-
specific CD8+ T cells can be monitored with MHC-
class I tetramers or qPCR clonotype analysis (Altman
et al., 1996; Douek et al., 2002). Neither of these
techniques, however, provides any indication as to the
functionality of the cells detected. Functionality can be
assessed using both direct (intracellular cytokine stain-
ing and CFSE-based proliferation assays) and indirect
(e.g. 51Cr release assays, flow-based killing assays and3H proliferation assays) methods (Kern et al., 1998;
Sheehy et al., 2001; Brenchley et al., 2002; Liu et al.,
2002). The direct methods as a whole, however,
provide no information regarding cytotoxic ability,
and conversely the indirect methods provide no indi-
cation as to the identity of the effector cells. The assay
we describe here provides a link between the direct and
indirect methods of CD8+ T cell effector analysis by
enabling precise phenotypic and functional character-
ization of responding CD8+ T cells through flow
cytometry using a marker that is only expressed during
degranulation, the initial event that takes place during
target cell lysis.
Although the CD107 assay does not directly mea-
sure target cell lysis, it does provide an indication of
the cytotoxic potential of the responding CD8+ T
cells. Current methods to assess CD8+ T cell-mediat-
ed target cell killing, including the standard chromium
release and flow-based killing assays do not identify
or quantify effector cells, instead measuring their
downstream effect on target cells. Thus, it has not
been possible to characterize directly the full func-
tional capacity and phenotype of the responding
CD8+ T cells. Our CD107 assay can provide an
assessment of the capacity, frequency and phenotype
of CD8+ T cells that kill in conditions similar to those
used in a standard 51Cr release assay.
More recently, intracellular cytokine staining has
provided a wealth of information regarding the
phenotype and functional status of antigen-specific
CD8+ T cells. The question, however, has remained
whether CD8+ T cells that produce cytokine after
stimulation are cytotoxic. By measuring degranula-
tion in the same cells, it is apparent that the majority
of the CD8+ T cells that respond to antigen by
producing cytokine also degranulate. This suggests,
therefore, that cytokine producing CD8+ T cells that
degranulate should be capable of killing targets,
provided they express the necessary granular comp-
onents to do so.
Interestingly, we observe in some patients that a
substantial population of responding CD8+ T cells
degranulate but do not produce IFNg. It is well
documented that many tetramer + cells do not produce
detectable levels of IFNg after direct ex vivo stimula-
tion with cognate peptide (Goepfert et al., 2000;
Shankar et al., 2000). Therefore, functional heteroge-
neity exists within the population of responding CD8+
Tcells. Thus, in assessing the frequency of the CD8+ T
cell response to any particular antigen, one should also
include measurement of degranulation, lest the re-
sponse frequency be underestimated.
Current data suggests that CD8+ T cell populations
specific for a single antigen can produce multiple
cytokines, but that heterogeneity exists as to the
cytokines produced by individual cells within that
population (data not shown). Importantly, there does
not appear to be a significant population of CD8+ T
cells that produce IFNg without degranulating. Al-
though it remains to be determined if the non-IFNg
producing CD107+ cells produce other cytokines in
response to stimulation, preliminary results indicate
that CD107 is expressed on the surface of MIP1h+ and
TNFa +CD8+ Tcells (data not shown). Thus, CD107a
and b expression after stimulation provides a more
complete assessment of the total frequency of respond-
ing CD8+ T cells than does monitoring production of
any one cytokine alone.
An important methodological aspect of the CD107
assay is that fixation or permeabilization of the
responding cells is not necessary, thus making this a
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–78 77
suitable procedure for sorting of live cells. Previously,
sorting of antigen-specific CD8+ T cells ex vivo has
been limited to methods utilizing either tetramer stain-
ing or IFNg capture systems. Tetramer-based sorts,
while very rapid and specific, are limited by the need
for very specific reagents and restricted peptide-MHC
combinations. While IFNg capture is not limited by
reagent availability, it is inherently more difficult to
perform, and is limited to IFNg producing cells. Unlike
tetramer-based methods, sorting based on CD107a and
b expression is not limited by reagent availability, or to
prior knowledge of HLA type or peptide recognition,
since mixtures of overlapping peptides can be used.
Additionally, sorting based on CD107a and b is not
limited only to those cells capable of producing a
certain cytokine, as is the IFNg capture assay.
In conclusion, we have described a novel method to
assess CD8+ T cell effector function based on the
ability of these cells to degranulate. This assay is sim-
ple, rapid, sensitive, and can be adapted for use in
combination with both tetramer and intracellular cyto-
kine assays. Assessment of degranulation alongside
cytokine production and phenotypic characterization
will greatly enhance our knowledge of the functionality
of antigen-specific CD8+ Tcells both in disease as well
as in vaccine models.
References
Altman, J.D., Moss, P.A.H., Goulder, P.J.R., Barouch, D.H.,
McHeyzer-Williams, M.G., Bell, J.I., McMichael, A.J., Davis,
M.M., 1996. Phenotypic analysis of antigen-specific T lympho-
cytes. Science 274, 94–96.
Appay, V., Nixon, D.F., Donahoe, S.M., Gillespie, G.M., Dong, T.,
King, A., Ogg, G.S., Spiegel, H.M., Conlon, C., Spina, C.A.,
Havlir, D.V., Richman, D.D., Waters, A., Easterbrook, P.,
McMichael, A.J., Rowland-Jones, S.L., 2000. HIV-specific
CD8(+) T cells produce antiviral cytokines but are impaired in
cytolytic function. J. Exp. Med. 192, 63–75.
Appay, V., Papagno, L., Spina, C.A., Hansasuta, P., King, A., Jones,
L., Ogg, G.S., Little, S., McMichael, A.J., Richman, D.D., Row-
land-Jones, S.L., 2002. Dynamics of T cell responses in HIV
infection. J. Immunol. 168, 3660–3666.
Barry, M., Bleackley, R.C., 2002. Cytotoxic T lymphocytes: all
roads lead to death. Nat. Rev. Immunol. 2, 401–409.
Betts, M.R., Casazza, J.P., Patterson, B.A., Waldrop, S., Trigona, W.,
Fu, T.M., Kern, F., Picker, L.J., Koup, R.A., 2000. Putative
immunodominant human immunodeficiency virus-specific
CD8(+) T-cell responses cannot be predicted by major histocom-
patibility complex class I haplotype. J. Virol. 74, 9144–9151.
Bossi, G., Griffiths, G.M., 1999. Degranulation plays an essential
part in regulating cell surface expression of Fas ligand in T cells
and natural killer cells. Nat. Med. 5, 90–96.
Brenchley, J.M., Douek, D.C., Ambrozak, D.R., Chatterji, M.,
Betts, M.R., Davis, L.S., Koup, R.A., 2002. Expansion of acti-
vated human naive T-cells precedes effector function. Clin. Exp.
Immunol. 130, 432–440.
Brunner, K.T., Mauel, J., Cerottini, J.C., Chapuis, B., 1968. Quan-
titative assay of the lytic action of immune lymphoid cells on
51-Cr-labelled allogeneic target cells in vitro; inhibition by iso-
antibody and by drugs. Immunology 14, 181–196.
Douek, D.C., Betts, M.R., Brenchley, J.M., Hill, B.J., Ambrozak,
D.R., Ngai, K.L., Karandikar, N.J., Casazza, J.P., Koup, R.A.,
2002. A novel approach to the analysis of specificity, clonality,
and frequency of HIV-specific T cell responses reveals a poten-
tial mechanism for control of viral escape. J. Immunol. 168,
3099–3104.
Fukuda, M., 1991. Lysosomal membrane glycoproteins. Structure,
biosynthesis, and intracellular trafficking. J. Biol. Chem. 266,
21327–21330.
Goepfert, P.A., Bansal, A., Edwards, B.H., Ritter Jr., G.D., Tellez,
I., McPherson, S.A., Sabbaj, S., Mulligan, M.J., 2000. A sig-
nificant number of human immunodeficiency virus epitope-spe-
cific cytotoxic T lymphocytes detected by tetramer binding do
not produce gamma interferon. J. Virol. 74, 10249–10255.
Kannan, K., Stewart, R.M., Bounds, W., Carlsson, S.R., Fukuda, M.,
Betzing, K.W., Holcombe, R.F., 1996. Lysosome-associated
membrane proteins h-LAMP1(CD107a) and h-LAMP2
(CD107b) are activation-dependent cell surface glycoproteins
in human peripheral blood mononuclear cells which mediate cell
adhesion to vascular endothelium. Cell. Immunol. 171, 10–19.
Kern, F., Surel, I.P., Brock, C., Freistedt, B., Radtke, H., Scheffold,
A., Blasczyk, R., Reinke, P., Schneider-Mergener, J., Radbruch,
A., Walden, P., Volk, H.D., 1998. T-cell epitope mapping by
flow cytometry. Nat. Med. 4, 975–978.
Liu, L., Chahroudi, A., Silvestri, G., Wernett, M.E., Kaiser, W.J.,
Safrit, J.T., Komoriya, A., Altman, J.D., Packard, B.Z., Fein-
berg, M.B., 2002. Visualization and quantification of T cell-
mediated cytotoxicity using cell-permeable fluorogenic caspase
substrates. Nat. Med. 8, 185–189.
Mollenhauer, H.H., Morre, D.J., Rowe, L.D., 1990. Alteration of
intracellular traffic by monensin; mechanism, specificity and re-
lationship to toxicity. Biochim. Biophys. Acta 1031, 225–246.
Peters, P.J., Borst, J., Oorschot, V., Fukuda, M., Krahenbuhl, O.,
Tschopp, J., Slot, J.W., Geuze, H.J., 1991. Cytotoxic T lympho-
cyte granules are secretory lysosomes, containing both perforin
and granzymes. J. Exp. Med. 173, 1099–1109.
Roederer, M., Methods for fluorescent conjugation of monoclonal
antibodies (http://www.drmr.com/abcon).
Roederer, M., Bowser, R., Murphy, R.F., 1987. Kinetics and tem-
perature dependence of exposure of endocytosed material to
proteolytic enzymes and low pH: evidence for a maturation
model for the formation of lysosomes. J. Cell. Physiol. 131,
200–209.
Sandberg, J.K., Fast, N.M., Nixon, D.F., 2001. Functional hetero-
geneity of cytokines and cytolytic effector molecules in human
CD8+ T lymphocytes. J. Immunol. 167, 181–187.
M.R. Betts et al. / Journal of Immunological Methods 281 (2003) 65–7878
Shankar, P., Russo, M., Harnisch, B., Patterson, M., Skolnik, P.,
Lieberman, J., 2000. Impaired function of circulating HIV-spe-
cific CD8(+) T cells in chronic human immunodeficiency virus
infection. Blood 96, 3094–3101.
Sheehy,M.E.,McDermott, A.B., Furlan, S.N., Klenerman, P., Nixon,
D.F., 2001. A novel technique for the fluorometric assessment of
T lymphocyte antigen specific lysis. J. Immunol. Methods 249,
99–110.
Trapani, J.A., Smyth, M.J., 2002. Functional significance of the
perforin/granzyme cell death pathway. Nat. Rev. Immunol. 2,
735–747.