B7−CD28 Costimulatory Signals Control the
Survival and Proliferation of Murine and
Human γδ T Cells via IL-2 Production
This information is current as
of June 18, 2017.
Julie C. Ribot, Ana deBarros, Liliana Mancio-Silva, Ana
Pamplona and Bruno Silva-Santos
J Immunol published online 25 June 2012
http://www.jimmunol.org/content/early/2012/06/25/jimmun
ol.1200268
http://www.jimmunol.org/content/suppl/2012/06/25/jimmunol.120026
8.DC1
Subscription
Information about subscribing to The Journal of Immunology is online at:
http://jimmunol.org/subscription
Permissions
Email Alerts
Submit copyright permission requests at:
http://www.aai.org/About/Publications/JI/copyright.html
Receive free email-alerts when new articles cite this article. Sign up at:
http://jimmunol.org/alerts
The Journal of Immunology is published twice each month by
The American Association of Immunologists, Inc.,
1451 Rockville Pike, Suite 650, Rockville, MD 20852
Copyright © 2012 by The American Association of
Immunologists, Inc. All rights reserved.
Print ISSN: 0022-1767 Online ISSN: 1550-6606.
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
Supplementary
Material
Published June 25, 2012, doi:10.4049/jimmunol.1200268
The Journal of Immunology
B7–CD28 Costimulatory Signals Control the Survival and
Proliferation of Murine and Human gd T Cells via IL-2
Production
Julie C. Ribot,* Ana deBarros,* Liliana Mancio-Silva,*,† Ana Pamplona,* and
Bruno Silva-Santos*,‡
g
d T cells have been evolutionarily conserved as a lymphocyte lineage whose signature TCR (TCRgd) does not
obey the paradigm of MHC restriction (1). In fact, we
know very little about Ag specificity and recognition via TCRgd
(2). Although this has not precluded advances in exploiting the
functional properties of these lymphocytes, particularly in cancer
immunotherapy (3), or in dissecting their pathological contribution to autoimmunity (4–6), it remains a priority to understand
how gd cells are activated in vivo. As part of our efforts to elucidate the contribution of costimulatory receptors to this process
(7, 8), we addressed the role of the Ig superfamily protein CD28.
Much of what we know about T cell costimulation has come
from studies on CD28 and its ligands, B7.1 (CD80) and B7.2
(CD86), which are typically found on professional APCs, such as
*Unidade de Imunologia Molecular, Instituto de Medicina Molecular, Faculdade de
Medicina, Universidade de Lisboa, 1649-028 Lisboa, Portugal; †Unidade de Malária,
Instituto de Medicina Molecular, Faculdade de Medicina, Universidade de Lisboa,
1649-028 Lisboa, Portugal; and ‡Instituto Gulbenkian de Ciência, 2781-901 Oeiras,
Portugal
Received for publication January 23, 2012. Accepted for publication May 24, 2012.
This work was supported by Fundação para a Ciência e Tecnologia (PTDC/SAU-MII/
104158/2008) and the European Molecular Biology Organization (Young Investigator Programme). L.M.-S. is supported by a European Molecular Biology Organization fellowship (ALTF960-2009).
Address correspondence and reprint requests to Prof. Bruno Silva-Santos or Dr. Julie
Ribot, Unidade de Imunologia Molecular, Instituto de Medicina Molecular, Avenida
Prof. Egas Moniz, 1649-028 Lisboa, Portugal. E-mail addresses: [email protected]
(B.S.-S.) and [email protected] (J.R.)
The online version of this article contains supplemental material.
Abbreviations used in this article: B6, C57BL/6; DC, dendritic cell; HMB-PP, 4hydroxy-3-methyl-but-2-enyl pyrophosphate; ICOSL, ICOS ligand; LN, lymph node;
OX40L, OX40 ligand; PbA, Plasmodium berghei ANKA; WT, wild-type.
Copyright Ó 2012 by The American Association of Immunologists, Inc. 0022-1767/12/$16.00
www.jimmunol.org/cgi/doi/10.4049/jimmunol.1200268
dendritic cells (DCs) or B cells (9). B7–CD28 signaling in ab
T cells was shown to have both qualitative and quantitative effects
that lower activation thresholds, promote cell proliferation, and
enhance functional activity in vitro and in vivo (reviewed in Refs.
9–11).
In contrast to this well-established function in ab T cells,
the role of CD28 in gd T cell activation has remained controversial because of discrepant results of previous studies (7).
In particular, resting murine gd splenocytes (12) and various
intraepithelial lymphocytes subsets (13–15) were reported to be
devoid of CD28 expression, which was observed, however, upon
cellular activation (12). This pattern contrasted with that of human
Vd2+ PBLs, which significantly downregulated CD28 following
activation (16). Moreover, human Vd1+ cells, unlike their Vd2+
counterparts, failed to express CD28 (17). Furthermore, although
CD28 signals promoted the in vitro proliferation of mouse
splenocytes (12, 15) and human gd lymphocytes (18), the alloreactivity of murine Vg2+ transgenic thymocytes was normal in the
absence of CD28 signaling (19). Critically, none of these studies
examined the role of CD28 costimulation during physiological gd
cell responses to infection.
In this study, we used the last generation of specific mAbs and
genetically manipulated mice to unequivocally assess the impact of
B7–CD28 signals in the context of gd cell responses to Plasmodium parasites, the infectious agents that cause malaria. Infection
by Plasmodium is known to cause striking expansions of gd cells
both in mice (20–22) and in humans (23–26). In fact, in patients
infected with either Plasmodium falciparum (23, 24) or Plasmodium vivax (25), gd cells frequently expand to 30–40% of all
peripheral blood T cells (compared with 1–5% in healthy donors).
Moreover, gd cells were shown to be the major cellular source of
the key proinflammatory cytokine IFN-g in patients in endemic
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
gd T cells play key nonredundant roles in immunity to infections and tumors. Thus, it is critical to understand the molecular
mechanisms responsible for gd T cell activation and expansion in vivo. In striking contrast to their ab counterparts, the
costimulation requirements of gd T cells remain poorly understood. Having previously described a role for the TNFR superfamily
member CD27, we since screened for other nonredundant costimulatory receptors in gd T cell activation. We report in this article
that the Ig superfamily receptor CD28 (but not its related protein ICOS) is expressed on freshly isolated lymphoid gd T cells and
synergizes with the TCR to induce autocrine IL-2 production that promotes gd cell survival and proliferation in both mice and
humans. Specific gain-of-function and loss-of-function experiments demonstrated a nonredundant function for CD28 interactions
with its B7 ligands, B7.1 (CD80) and B7.2 (CD86), both in vitro and in vivo. Thus, gd cell proliferation was significantly enhanced
by CD28 receptor agonists but abrogated by B7 Ab-mediated blockade. Furthermore, gd cell expansion following Plasmodium
infection was severely impaired in mice genetically deficient for CD28. This resulted in the failure to mount both IFN-g–mediated
and IL-17–mediated gd cell responses, which contrasted with the selective effect of CD27 on IFN-g–producing gd cells. Our data
collectively show that CD28 signals are required for IL-2–mediated survival and proliferation of both CD27+ and CD272 gd T cell
subsets, thus providing new mechanistic insight for their modulation in disease models. The Journal of Immunology, 2012, 189:
000–000.
CD28 COSTIMULATION OF gd T CELLS
2
areas (27), and this correlated with reduced incidence of clinical
episodes (28). Therefore, malaria constitutes one of the most
relevant physiological contexts in which to investigate gd cell
activation. We performed a comprehensive series of gain-offunction and loss-of-function studies that demonstrates a critical
role for B7–CD28 interactions in the activation and expansion of
murine and human proinflammatory gd T cell subsets in response
to Plasmodium Ags.
Materials and Methods
Mice
Preparation of supernatants from P. falciparum-infected
erythrocyte cultures
P. falciparum 3D7 blood-stage parasites were cultured using modifications
to the method described by Trager and Jensen (32). Parasites were grown
in human erythrocytes (5% hematocrit, 5% parasitemia) in RPMI 1640
medium containing L-glutamine supplemented with 0.5% (w/v) AlbuMAX
II, 25 mM HEPES, and 0.05 mg/ml gentamicin at 37˚C in a 5% CO2
environment. Supernatant of synchronized cultures in mature late-stage
schizonts (40–50 h after reinvasion) was obtained by centrifugation at
2000 rpm for 5 min. Synchronization of cultures consisted of treatment
with 5% (w/v) D-sorbitol, one or two cycles before supernatant collection.
Culture of human PBLs
Human peripheral blood was collected from anonymous healthy volunteers
in accordance with the guidelines of the Declaration of Helsinki. Total
PBMCs were prepared and cultured as previously described (33, 34). For
purification of gd PBLs, cells were sorted on a MiniMACS separator using
an anti-TCRgd MicroBead Kit (Miltenyi Biotec). Specific Vg9Vd2 T cell
activation was accomplished by continuous stimulation (up to 6 d) of total
PBMCs or sorted gd-PBLs with 10 nM 4-hydroxy-3-methyl-but-2-enyl
pyrophosphate (HMB-PP; Echelon Biosciences) and 25–100 U/ml IL-2
(Roche Applied Science). Alternatively, PBMCs or sorted gd-PBLs were
cultured in the presence of collected supernatant from P. falciparum cultures. When indicated, anti-human blocking mAbs (or isotype controls)
were added at a concentration of 10 mg/ml.
Flow cytometry and cell sorting
Cells were sorted electronically using a FACSAria (BD Biosciences, San
Jose, CA). For cell surface stainings, cells were incubated for 15 min on ice
in 2.4G2 (anti-FcgR mAb) hybridoma supernatant and then incubated for
15 min with saturating concentrations of the indicated mAbs. For intracellular cytokine staining, cells were stimulated with PMA (50 ng/ml) and
ionomycin (1 mg/ml) (both from Sigma) for 4 h at 37˚C; 10 mg/ml brefeldin A (Sigma) was added during the last 2 h. Cells were stained for the
indicated cell surface markers, and intracellular staining was performed
using fixation/permeabilization and permeabilization buffers (both from
eBioscience), following the manufacturer’s instructions. Samples were
analyzed using Fortessa or FACSCalibur (both from BD Biosciences).
In vitro functional assays
Lymphocyte activation and apoptosis were assessed between days 1 and 6 of
culture by staining for CD69 and for annexin V (BD Pharmingen), respectively, according to the manufacturer’s instructions. IL-2 secretion was
quantified using Cytometric Bead Array as described (33). For analysis of
cell proliferation, cells were stained with 5 mM the cytoplasmic dye CFSE
(Molecular Probes) for 5 min at 37˚C, cultured for 3 d, and analyzed by
flow cytometry for CFSE dilution. Alternatively, 1 mCi [3H]thymidine
(Amersham) was added for the last 18 h, and [3H]thymidine incorporation
was measured using the MicroBeta TriLux scintillation counter (PerkinElmer).
Quantitative real-time PCR
RNA was prepared and analyzed by quantitative real-time PCR, as previously
described (21, 22). The primers used were: Il2-Fwd, 59-GCTGTTGATGGACCTACAGGA-39; Il2-Rev, 59-TTCAATTCTGTGGCCTGCTT-39;
Efa1-Fwd, 59-ACACGTAGATTCCGGCAAGT-39; and Efa1-Rev, 59-AGGAGCCCTTTCCCATCTC-39. Transcripts were quantified by the standard
curve method, normalized to Efa1, and expressed in arbitrary units.
Statistical analysis
Statistical significance of differences between populations was assessed
using the Mann–Whitney test.
Culture of murine lymphocytes
Cells from spleen, lymph nodes (LNs), or peritoneal cavity were prepared as
described (21). Cells were cultured in RPMI 1640 medium with 10% FCS,
50 mM 2-ME, L-glutamine, nonessential amino acids, 10 mM HEPES,
penicillin, and streptomycin in 96-well round-bottom plates. When indicated, 30–300 U/ml rIL-2 (Sigma) was added to the cultures. Cells were
stimulated with 0.5 mg/ml soluble anti-CD3ε Ab, in the presence of 5 3
105 APCs (3000 rad-irradiated splenocytes)/5 3 104 responding cells.
Alternatively, cells were cultured on plate-bound anti-CD3ε (0.1–10 mg/
ml) and anti-CD28 (10 mg/ml) or soluble rCD70 (5 mg/ml; kindly provided
by Dr. Jannie Borst). When indicated, anti-mouse agonist or blocking
mAbs were added to the culture medium at a concentration of 1–5 mg/ml.
mAbs
The following anti-mouse mAbs were purchased from eBioscience (San
Diego, CA) and used for flow cytometry: FITC-labeled anti-TCRd
(eBioGL3), anti-CD69 (H1.2F3), and anti–IL-17 (eBio17B7); PE-labeled
anti-TCRd (eBioGL3), anti-CD11c (N418), and anti-IFN-g (XMG1.2);
PerCP-Cy5.5–labeled anti-CD3ε (145.2C11); PE-Cy7–labeled anti-CD19
(eBio1D3); allophycocyanin-labeled anti-TCRd (eBioGL3), anti-CD11b
(M1/70), and anti–IL-17 (eBio17B7); eFluor 450-labeled anti-CD4
(RM4-5); biotin-conjugated anti-CD28 (37.51); and streptavidin-PE.
In murine cell cultures, anti-mouse mAbs (from eBioscience) specific for
IL-2 (JES6-1A12), ICOS (7E.17G9), CD80 (16-10A1), and CD86 (GL1)
Results
B7–CD28 interactions promote the proliferation and survival
of murine gd T cells
This study began with the observation that lymphoid (spleen and
LN) gd cells, in contrast to their intraepithelial counterparts (13–
15), constitutively express CD28 at similar levels to ab cells (Fig.
1A). Moreover, in vitro activation with anti-CD3 (aCD3ε) mAb
significantly upregulated CD28 levels on isolated (FACS-purified)
gd cells (Fig. 1B), whereas the addition of B7-expressing APCs
provoked CD28 downregulation (Supplemental Fig. 1A). In contrast, the CD28-related Ig superfamily member, ICOS, was not
expressed in either resting or activated gd cells (Fig. 1B).
To functionally test the role of B7–CD28 interactions in gd cell
activation, we used CD28 agonists (anti-CD28 mAb) or antagonists (anti-B7.1/CD80 and anti-B7.2/CD86 mAbs) in cocultures of
gd cells (10%) and irradiated splenocytes (90%). After 3 d of
activation (with aCD3ε mAb), the percentage of live gd cells
recovered from these cultures was markedly higher in the presence
of CD28 agonists and lower in the presence of CD28 antagonists,
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
All mice were adults 4–10 wk of age. C57BL/6 (B6), B6.TCRa-deficient
(Tcra2/2), and B6.CD28-deficient (Cd282/2) mice were described previously (29, 30) and were obtained from The Jackson Laboratory. B6.CD27deficient (Cd272/2) mice were a kind gift from Dr. Jannie Borst (Netherlands Cancer Institute, Amsterdam, The Netherlands). Mice were bred
and maintained in the specific pathogen-free animal facilities of Instituto
de Medicina Molecular. When stated, mice were infected i.p. with 106
Plasmodium berghei ANKA (PbA)-infected erythrocytes and monitored as
described (31). All experiments involving animals were performed in
compliance with the relevant laws and institutional guidelines and were
approved by the local ethics committees.
were used as blocking reagents; anti-mouse mAbs for CD3ε (145.2C11) and
CD28 (37.51) were used as agonists.
The following anti-human mAbs were used for flow cytometry: anti–
CD28-FITC (CD28.2; eBioscience); anti–CD69-PE (FN50), anti–CD70PE (ki-24), anti–CD80-PE (L307.4), and anti–CD86-allophycocyanin
(2331) (all from BD Pharmingen, San Diego, CA); and anti–OX40 ligand
(OX40L)-PE (11C3.1) and anti–ICOS ligand (ICOSL)-PE (2D3) (both
from BioLegend, San Diego, CA).
The mAbs used as blocking reagents in human cell cultures were antiCD70 (ki-24) and anti-CD86 (IT2.2) (both from BD Pharmingen); antiICOSL (9F.8A4; BioLegend); and anti-CD80 (37711) and anti-OX40L
(159403) (both from R&D Systems). IgG1 (MOPC-21), IgG2 (MPC-11),
and IgG3 (MG3-35) were used as isotype controls (all purchased from
BioLegend).
The Journal of Immunology
3
whereas blocking ICOS had no effect (Fig. 1C). This correlated
well with the activation phenotype of gd cells, as evaluated by the
expression of the activation marker CD69 (Fig. 1D).
To assess the precise role of CD28 signaling on gd cell proliferation, we performed [3H]thymidine incorporation and CFSE-
dilution assays. CD28 agonists enhanced gd cell proliferation,
whereas CD28 antagonists inhibited it completely (Fig. 1E, 1F).
Consistent with this, CD28-deficient gd cells showed impaired
activation and proliferation in vitro (Supplemental Fig. 1B, 1C).
Furthermore, CD28 signaling on gd cells also affected their sur-
FIGURE 2. CD28 costimulation of gd cells induces secretion of IL-2 required for proliferation. gd cells were sorted using FACS from pooled spleen and
LN of Tcra2/2 mice and activated with the indicated amounts of plate-bound anti-CD3 mAb in the presence of 10 mg/ml plate-bound anti-CD28 mAb
(aCD28; in green) or isotype control (iso ctrl). (A) CFSE dilution profiles after 2 d under the indicated concentrations of anti-CD3 (left panel) and extracted
numbers of cells/division at 0.1 mg/ml anti-CD3 (right panel). (B) IL-2 concentration was measured at the indicated time points in the supernatant of
cultures stimulated with 0.1 mg/ml anti-CD3. (C) Quantitative real-time PCR for Il2 expression in purified gd cells cultured for 6 h with 0.1 mg/ml anti-CD3
mAb in media alone, media supplemented with soluble rCD70 (sCD70) or anti-CD28 mAb (aCD28), or isotype control (iso ctrl). Expression was normalized to the housekeeping gene Efa1 and expressed in arbitrary units (a.u.). CFSE dilution profiles (D) and extracted numbers of cells/division (E) after
3 d of stimulation with 0.1 mg/ml plate-bound anti-CD3 in the presence of 10 mg/ml isotype control (iso ctrl) (gray), 10 mg/ml plate-bound anti-CD28
(green), 5 mg/ml anti–IL-2 (white), or 300 U/ml rIL-2 (black). Error bars represent SD (n = 3). Data are representative of three independent experiments.
*p , 0.05, **p , 0.01, ***p , 0.001.
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
FIGURE 1. CD28 signals promote murine gd T cell activation and proliferation. (A) Cells pooled from spleens and LN of adult WT B6 mice were
stained with mAbs specific for CD3ε, TCRgd, and CD28. The graph shows the expression of CD28 on pregated CD3ε+ TCRgd+ cells, CD3ε+ TCRgd2
cells, and CD3ε2 cells. Peripheral lymphoid cells from Cd282/2 mice (shaded curve) are shown as a negative control. (B) gd cells were sorted from pooled
spleen and LNs of Tcra2/2 mice and cultured with plate-bound anti-CD3ε mAb. Mean fluorescence intensity (MFI) of CD28 or ICOS stainings at indicated
time points are shown. Dashed line represents isotype control staining. (C–G) gd cells were sorted by FACS from pooled spleen and LNs of Tcra2/2 mice,
labeled with CFSE, and cultured for 3 d in the presence of APCs and soluble anti-CD3 mAb, without mAbs (ctrl), or with the addition of the following
mAbs: blocking anti-ICOS, blocking anti-CD80 and anti-CD86 (aB7), or agonist anti-CD28. (C) Representative forward scatter (FSC)/side scatter (SSC)
plots. Numbers refer to percentages of cells within the indicated regions. (D) CD69 expression on gd cells at day 2 (upper panel) or at the indicated time
points (lower panel). Color code is the same as used in (G). (E) [3H]thymidine incorporation for 18 h between days 2 and 3, measured in a scintillation
counter (cpm). (F) CFSE dilution at day 3 (left panel) and extracted numbers of cells per division (right panel). (G) Percentage of annexin V+ cells at day 3.
Error bars represent SD (n = 3). All data are representative of three to five independent experiments.*p , 0.05, **p , 0.01, ***p , 0.001.
4
CD28 COSTIMULATION OF gd T CELLS
vival following activation, as indicated by annexin V staining of
day-3 cultures (Fig. 1G). Collectively, these data demonstrate that
B7–CD28 signals control the activation, survival, and proliferation
of murine gd T cells.
CD28 costimulation is necessary for autocrine IL-2 production
by gd T cells
CD28 signals are required for gd cell responses to
Plasmodium infection in vivo
To establish the importance of CD28 signaling on gd cell activation
in vivo, we resorted to a mouse model of severe malaria, induced
by PbA infection of B6 mice (31). Importantly, we had previously
documented very robust gd cell responses in this animal model (8,
21, 22).
We followed the phenotype of gd cells during the course of
infection and detected a transient upregulation of CD28 at day 3,
whereas ICOS was not induced on gd cells (Fig. 3A). In PbAinfected mice, the modulation of CD28 expression on gd cells was
accompanied by the upregulation of its inducible ligand, CD86, on
professional APCs, such as DCs (Fig. 3B). This led us to investigate the functional consequences of B7–CD28 costimulation
in vivo. For this, we analyzed CD28-deficient mice and observed
their failure to expand gd cells on infection, in contrast to wildtype (WT) controls (Fig. 3C). This correlated with a comparatively small increase in the pool of CD69+ (activated) gd cells in
infected CD28-deficient mice (Fig. 3D). Importantly, the effects of
CD28 deficiency on the numbers of total or activated gd cells were
not phenocopied in CD27-deficient animals (Fig. 3C, 3D), thus
demonstrating that the two costimulatory pathways play distinct
roles in gd T cell expansion in vivo.
FIGURE 3. CD28 signals are required for murine gd cell responses to
Plasmodium infection in vivo. B6 (WT), Cd282/2, or Cd272/2 mice were
infected with PbA, and their spleens were harvested and analyzed by flow
cytometry at the indicated days postinfection. (A) Mean fluorescence intensities (MFI) for CD28 and ICOS expression in gd T cells from WT
mice. Dashed line represents isotype control staining. (B) MFI for CD86
expression in DCs from WT, Cd282/2, or Cd272/2 mice at day 3 postinfection. Dashed line indicates reference MFI in naive WT animals.
Absolute numbers of total gd cells (C) and activated CD69+ gd cells (D) at
day 3 postinfection. (E and F) gd cells were sorted using FACS and stained
intracellularly for IFN-g and IL-17. Absolute numbers of IFN-g+ gd cells
(E) or IL-17+ gd cells (F) are shown. Error bars represent SD. All data are
from three independent experiments involving 6–10 mice. *p , 0.05,
**p , 0.01, ***p , 0.001. ns, Not statistically significant (p $ 0.05).
In PbA-infected CD28-deficient animals, the reduced (compared
with WT) gd T cell compartment also contained smaller pools of
both IFN-g–producing and IL-17–producing subsets (Supplemental Fig. 2). In fact, the CD28 deficiency completely prevented
the accumulation of IFN-g+ (Fig. 3E) and IL-17+ (Fig. 3F) gd
cells upon malaria infection. We observed an additional difference
in comparison with CD27-deficient animals: the expansion of IL17+ gd cells was selectively impaired in CD282/2 mice (Fig. 3F).
In contrast, both mutant models exhibited a defect in the IFN-g
response (Fig. 3E). This led us to assess the combined effect of
CD27/CD28 signals on IFN-g production by gd cells in vivo. For
this, we decided to block B7–CD28 signaling in CD272/2 mice
during the course of infection. The efficient mAb-mediated neutralization of CD80 and CD86 caused a significant decrease in the
numbers of activated IFN-g+ gd cells in the CD27-deficient
background (Supplemental Fig. 3). Thus, CD28 acts nonredundantly and synergistically with CD27 in the activation of IFNg+ gd cells following malaria infection.
These data collectively demonstrate that CD28 signals are required for the expansion of murine proinflammatory gd cell subsets
in vivo.
Dynamic expression of CD28 and its B7 ligands on activated
human gd cells
We next investigated the role of CD28 costimulation on human gd
cells. Freshly isolated Vg9Vd2 PBLs were previously shown to
express CD28 (18, 38). Moreover, Vg9Vd2 PBLs are known to be
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
Because the previous (standard) assays used total irradiated
splenocytes as “feeders,” which typically included ∼30% of ab
T cells that also express CD28, it was critical to assess whether
B7–CD28 signals operated directly on gd cells. For this, we
established cultures of highly (.99%) purified gd cells, activated
by plastic-bound agonist Abs to TCR/CD3 and CD28. We observed that CD28 costimulation provided a marked proliferative
advantage at low amounts of TCR/CD3 triggering (i.e., 0.1 mg/ml
anti-CD3ε mAb) (Fig. 2A). This effect was lost at saturating (10fold higher) concentrations of anti-CD3ε mAb (Fig. 2A).
To gain further mechanistic insight into the direct effects of
CD28 on gd cell activation, we considered that CD28 costimulation of ab cells can enhance their production of IL-2 (35, 36),
which is a key determinant of gd cell expansion (33, 37).
Therefore, we measured IL-2 protein levels in the supernatants of
the gd cell cultures. Strikingly, IL-2 was only detected when these
were supplemented with CD28 agonists (Fig. 2B). The induction
of IL-2 expression in CD28-costimulated gd cells was also observed at the mRNA (transcriptional) level (Fig. 2C). Importantly,
CD27 triggering using soluble recombinant ligand (7, 8) did not
upregulate Il2 expression (Fig. 2C), suggesting that this mechanism is specifically downstream of CD28 signaling in gd cells.
The interesting dynamics of IL-2 production by gd cells and
their proliferation following TCR/CD28 stimulation (Supplemental Fig. 1D) prompted us to test whether autocrine IL-2 production was essential for gd cell proliferation. Upon addition of
neutralizing Abs against murine IL-2, a strong reduction in gd cell
expansion was observed (Fig. 2D, 2E). Conversely, the addition of
saturating doses of IL-2 bypassed the need for CD28 costimulation, both in isolated gd cell cultures (Fig. 2D, 2E) and in
cocultures with splenic APCs (Supplemental Fig. 1E). These
results show that CD28 costimulation directly controls gd cell
expansion through the induction of IL-2 production.
The Journal of Immunology
CD28 costimulation controls survival and proliferation of
activated human gd cells
To directly assess the B7–CD28 costimulation requirements of
HMB-PP–activated Vg9Vd2 PBLs, we used monoclonal blocking
Abs specific to CD80 and CD86 in Vg9Vd2 PBL cultures. We
observed a clear impairment in cell proliferation, as assessed by
CFSE-dilution assays (Fig. 5A), compared with control cultures
incubated with isotype Abs. Moreover, CD80/CD86 blockade also
affected the survival of HMB-PP–activated Vg9Vd2 PBLs, as
indicated by the accumulation of annexin V+ cells (Fig. 5B). This
combined survival/proliferation effect of CD80/CD86 inhibition
resulted in a highly significant reduction in thymidine incorporation in Vg9Vd2 PBL cultures (Fig. 5C). Moreover, and consistent
with the murine data (Fig. 2D, 2E), the need for CD28 costimulation was bypassed by saturating amounts of exogenous IL-2
(Fig. 5C).
Interestingly, the inhibition of ICOSL (CD275; B7-H2) or
OX40L (CD252) had no impact on their proliferation (Fig. 5A,
5C). Conversely, and in agreement with our previous results (40),
the inhibition of CD70–CD27 signaling prevented efficient
Vg9Vd2 PBL expansion (Fig. 5A, 5C). Thus, CD27 and CD28
provide independent and nonredundant costimulatory signals that
determine human Vg9Vd2 T cell survival and proliferation upon
activation.
Discussion
A critical step toward the development of more efficient therapeutic
strategies based on gd T cells is improvement of our understanding
of the molecular mechanisms that control their activation and
expansion. In this study, we showed that gd cell activation critically requires CD28 costimulation in both mice and humans. The
selective effect of CD28, but not of its related Ig superfamily
comember ICOS, emphasizes the biological relevance of B7–
CD28 interactions in gd T cell activation. Importantly, the role of
CD28 signaling was established through in vitro systems, as well
FIGURE 4. Dynamic expression of CD28, CD80, and CD86 in human
Vg9Vd2 T cells. (A) MACS-sorted gd PBLs of two healthy donors were
analyzed by flow cytometry for CD28 expression (gated on Vg9+ cells),
either before (non-stim) or after 2 d of stimulation with supernatants of P.
falciparum-infected erythrocyte cultures. The gates were set based on
isotype control stainings. Numbers refer to percentages of cells within the
indicated quadrants. Vg9+ cells within total PBMC cultures were analyzed
by flow cytometry for CD80 and CD86 expression before (ns) or after 2 d
of stimulation with supernatants of P. falciparum (Pf)-infected erythrocyte
cultures (B) or after the indicated days of stimulation with 10 nM HMB-PP
(C). All data are representative of three independent experiments with
similar results. ns, Nonstimulated.
FIGURE 5. B7–CD28 interactions are necessary for survival and proliferation of activated human Vg9Vd2 T cells. MACS-sorted gd PBLs
were stimulated for 4 d with 10 nM HMB-PP (in the presence of 25 U/ml
recombinant human IL-2 [rhIL-2]). Blocking Abs specific for CD80 and
CD86, ICOSL, OX40L, or CD70 (or isotype controls) were added at the
beginning of the cultures. (A) Cells were labeled with CFSE before being
cultured for 4 d and then analyzed by flow cytometry. Cells from isotype
control cultures are shaded in gray. (B) Annexin V staining was performed
on day 4 and analyzed by flow cytometry (gating on Vg9+ cells). (C) [3H]
thymidine was added during the last 18 h of culture, and its incorporation
was measured in a scintillation counter (cpm). All data are representative
of three independent experiments with similar results. Error bars represent
SD. *p , 0.05, **p , 0.01. IL-2high, Cells cultured in the presence of 100
U/ml rhIL-2 (in addition to CD80/CD86 blockade).
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
highly activated by soluble metabolites produced by apicomplexan protozoa like P. falciparum (39). To study the expression
dynamics of CD28 and its B7 ligands, CD80 and CD86, on
Vg9Vd2 PBLs exposed to microbial Ags, we established an experimental system in which either total PBMCs or purified gd
PBLs were incubated for 2–6 d with supernatants collected from
P. falciparum-infected erythrocyte cultures. We observed a
marked downregulation of CD28 after 2 d of culture with
P. falciparum-derived supernatants (Fig. 4A), concomitantly with
the induction of CD80 and CD86 expression on Vg9Vd2 PBLs
(Fig. 4B). This induction was a direct effect of the Plasmodium
Ags on Vg9Vd2 PBLs, because it was also observed on isolated
Vg9Vd2 PBLs (Supplemental Fig. 4A).
A strong candidate P. falciparum Ag was HMB-PP, the most
potent natural Vg9Vd2 TCR agonist known (33, 39), because it is
the product of HMB-PP synthase (GcpE) present in all Plasmodium spp. and is predominantly expressed on mature blood-stage
parasites (http://plasmodb.org; Gene ID: PF10_0221). Therefore,
we established similar cultures of PBMCs or purified Vg9Vd2
PBLs on medium supplemented with HMB-PP. This resulted in
the marked induction of both CD80 and CD86 on Vg9Vd2 PBLs
from various healthy donors (Fig. 4C, Supplemental Fig. 4B). The
upregulation of B7 ligands and downregulation of CD28 receptor
on P. falciparum-activated or HMB-PP–activated Vg9Vd2 PBLs
prompted us to test the functional consequences of CD28 engagement in human gd cells.
5
CD28 COSTIMULATION OF gd T CELLS
6
future research should examine the potential of manipulating
pathogenic IL-17+ gd cell responses through the inhibition of B7–
CD28 costimulatory signals.
Acknowledgments
We thank M.M. Mota, T. Hänscheid, J.P. Simas, J. Decalf, A.E. Sousa, L.
Graca, and J. Borst for reagents and suggestions and the staffs of the Flow
Cytometry and Animal facilities of Instituto de Medicina Molecular for
technical assistance.
Disclosures
The authors have no financial conflicts of interest.
References
1. Hayday, A. C. 2000. [gamma][delta] cells: a right time and a right place for
a conserved third way of protection. Annu. Rev. Immunol. 18: 975–1026.
2. Hayday, A. C. 2009. Gammadelta T cells and the lymphoid stress-surveillance
response. Immunity 31: 184–196.
3. Gomes, A. Q., D. S. Martins, and B. Silva-Santos. 2010. Targeting gd
T lymphocytes for cancer immunotherapy: from novel mechanistic insight to
clinical application. Cancer Res. 70: 10024–10027.
4. Roark, C. L., J. D. French, M. A. Taylor, A. M. Bendele, W. K. Born, and
R. L. O’Brien. 2007. Exacerbation of collagen-induced arthritis by oligoclonal,
IL-17-producing gamma delta T cells. J. Immunol. 179: 5576–5583.
5. Sutton, C. E., S. J. Lalor, C. M. Sweeney, C. F. Brereton, E. C. Lavelle, and
K. H. Mills. 2009. Interleukin-1 and IL-23 induce innate IL-17 production from
gammadelta T cells, amplifying Th17 responses and autoimmunity. Immunity
31: 331–341.
6. Petermann, F., V. Rothhammer, M. C. Claussen, J. D. Haas, L. R. Blanco,
S. Heink, I. Prinz, B. Hemmer, V. K. Kuchroo, M. Oukka, and T. Korn. 2010. gd
T cells enhance autoimmunity by restraining regulatory T cell responses via an
interleukin-23-dependent mechanism. Immunity 33: 351–363.
7. Ribot, J. C., A. debarros, and B. Silva-Santos. 2011. Searching for “signal 2”:
costimulation requirements of gd T cells. Cell. Mol. Life Sci. 68: 2345–2355.
8. Ribot, J. C., M. Chaves-Ferreira, F. d’Orey, M. Wencker, N. Gonçalves-Sousa,
J. Decalf, J. P. Simas, A. C. Hayday, and B. Silva-Santos. 2010. Cutting edge:
adaptive versus innate receptor signals selectively control the pool sizes of
murine IFN-g- or IL-17-producing gd T cells upon infection. J. Immunol. 185:
6421–6425.
9. Duttagupta, P. A., A. C. Boesteanu, and P. D. Katsikis. 2009. Costimulation
signals for memory CD8+ T cells during viral infections. Crit. Rev. Immunol. 29:
469–486.
10. Smith-Garvin, J. E., G. A. Koretzky, and M. S. Jordan. 2009. T cell activation.
Annu. Rev. Immunol. 27: 591–619.
11. Acuto, O., and F. Michel. 2003. CD28-mediated co-stimulation: a quantitative
support for TCR signalling. Nat. Rev. Immunol. 3: 939–951.
12. Sperling, A. I., P. S. Linsley, T. A. Barrett, and J. A. Bluestone. 1993. CD28mediated costimulation is necessary for the activation of T cell receptor-gamma
delta+ T lymphocytes. J. Immunol. 151: 6043–6050.
13. Ohteki, T., and H. R. MacDonald. 1993. Expression of the CD28 costimulatory
molecule on subsets of murine intestinal intraepithelial lymphocytes correlates
with lineage and responsiveness. Eur. J. Immunol. 23: 1251–1255.
14. Rakasz, E., M. Hagen, M. Sandor, and R. G. Lynch. 1997. Gamma delta T cells
of the murine vagina: T cell response in vivo in the absence of the expression of
CD2 and CD28 molecules. Int. Immunol. 9: 161–167.
15. Witherden, D. A., P. Verdino, S. E. Rieder, O. Garijo, R. E. Mills, L. Teyton,
W. H. Fischer, I. A. Wilson, and W. L. Havran. 2010. The junctional adhesion
molecule JAML is a costimulatory receptor for epithelial gammadelta T cell
activation. Science 329: 1205–1210.
16. Lafont, V., J. Liautard, A. Gross, J. P. Liautard, and J. Favero. 2000. Tumor
necrosis factor-alpha production is differently regulated in gamma delta and
alpha beta human T lymphocytes. J. Biol. Chem. 275: 19282–19287.
17. De Rosa, S. C., D. K. Mitra, N. Watanabe, L. A. Herzenberg, L. A. Herzenberg,
and M. Roederer. 2001. Vdelta1 and Vdelta2 gammadelta T cells express distinct
surface markers and might be developmentally distinct lineages. J. Leukoc. Biol.
70: 518–526.
18. Takamizawa, M., F. Fagnoni, A. Mehta-Damani, A. Rivas, and E. G. Engleman.
1995. Cellular and molecular basis of human gamma delta T cell activation. Role
of accessory molecules in alloactivation. J. Clin. Invest. 95: 296–303.
19. Penninger, J. M., E. Timms, A. Shahinian, A. Jezo-Bremond, H. Nishina,
J. Ionescu, S. M. Hedrick, and T. W. Mak. 1995. Alloreactive gamma delta
thymocytes utilize distinct costimulatory signals from peripheral T cells. J.
Immunol. 155: 3847–3855.
20. Yañez, D. M., J. Batchelder, H. C. van der Heyde, D. D. Manning, and
W. P. Weidanz. 1999. Gamma delta T-cell function in pathogenesis of cerebral
malaria in mice infected with Plasmodium berghei ANKA. Infect. Immun. 67:
446–448.
21. Ribot, J. C., A. deBarros, D. J. Pang, J. F. Neves, V. Peperzak, S. J. Roberts,
M. Girardi, J. Borst, A. C. Hayday, D. J. Pennington, and B. Silva-Santos. 2009.
CD27 is a thymic determinant of the balance between interferon-gamma- and interleukin 17-producing gammadelta T cell subsets. Nat. Immunol. 10: 427–436.
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
as in vivo, in the context of the immune response to Plasmodium
parasites.
Using a model of severe malaria induced by PbA infection of B6
mice (31), we previously showed robust gd cell expansion and
cytokine (IFN-g and IL-17) production (8, 21, 22). In this study,
we demonstrated that these in vivo gd cell activities are strictly
dependent on CD28 signaling, which promotes their survival and
proliferation on activation. Furthermore, we showed that similar
rules apply to human gd cells incubated with soluble Ags derived
from cultures of P. falciparum-infected erythrocytes. We believe
that these results provide important novel knowledge for the
therapeutic manipulation of gd cells in the context of malaria.
In recent years, we have significantly improved our understanding on how gd cells acquire and exert their cytokineproducing capacities (2, 41). In particular, we (21) and other
investigators (42) demonstrated that murine gd cells are programmed in the thymus to become either IFN-g or IL-17 producers. Various molecular mechanisms, including TCR (42–44),
CD27 (21), LTbR (45), and TGFbR (46) signaling, have been
proposed to participate in the “developmental preprogramming”
(2) of gd cells. Once in the periphery, innate TLR-mediated signals are key for the expansion of IL-17+ gd cells. Thus, TLR
stimulation of myeloid cells led to the secretion of IL-1b and IL23, which were the direct inducers of IL-17+ gd cell proliferation
(5, 8). By contrast, IFN-g+ gd cells were refractory to IL-1b/IL-23
stimulation but typically relied on TCR and CD27 signals for their
expansion (8, 40, 47).
These data led us previously to propose a model in which IFN-g+
gd cells followed an “adaptive-like” mode of activation (requiring
TCR plus costimulation), whereas IL-17+ gd cells behaved as
innate-like lymphocytes (8). Importantly, the data presented in this
article call for a significant revision of this model. Thus, we
provide clear evidence that IL-17+ gd cells, like IFN-g+ gd cells,
depend on CD28 costimulatory signals for their expansion in vivo.
This costimulation requirement of IL-17+ gd cells is also likely to
reflect an underappreciated dependence on TCR-mediated activation (5, 48, 49). Thus, although hampered by the lack of known
TCRgd ligands, future research should readdress the role of TCR
stimulation on the development and activation of IL-17+ gd cells.
Of note, the effects of CD28 costimulation can be bypassed
in vitro by saturating TCR (Fig. 2A) signals. Our finding that gd
cell expansion (on infection) in vivo is impaired in CD28-deficient
mice importantly suggests that physiological TCRgd triggering
(by yet mostly unknown ligands) relies on intermediate avidities,
at which the TCR acts synergistically with CD28.
Importantly, our detailed examination of CD28-deficient and
CD27-deficient mice (Fig. 3) established that the two costimulatory pathways play independent (and nonredundant) roles in gd
cell activation. Mechanistically, we showed that the induction
of IL-2 production is a major (Fig. 2B) and specific (Fig. 2C)
function of CD28 (but not CD27) costimulation in gd cells, which
are known to strongly benefit from IL-2 signals for their expansion
(33, 37). The fact that gd cells themselves can produce high levels
of IL-2 strictly upon CD28 costimulation defines important rules
for their expansion in situ. This notwithstanding, B7–CD28 signals can additionally contribute to gd cell expansion through costimulation of ab cells, as we previously showed that coculture
with IL-2–producing CD4+ CD252 cells enhanced gd cell proliferation (22).
Finally, although this study focused on infection, gd cells were
also shown to be key players in autoimmunity, because their secretion of IL-17 promotes the development of experimental autoimmune encephalomyelitis (5, 6), collagen-induced arthritis (4),
and psoriatic plaques (50) in animal disease models. Therefore,
The Journal of Immunology
36. Lindstein, T., C. H. June, J. A. Ledbetter, G. Stella, and C. B. Thompson. 1989.
Regulation of lymphokine messenger RNA stability by a surface-mediated T cell
activation pathway. Science 244: 339–343.
37. Casetti, R., G. Perretta, A. Taglioni, M. Mattei, V. Colizzi, F. Dieli, G. D’Offizi,
M. Malkovsky, and F. Poccia. 2005. Drug-induced expansion and differentiation
of V gamma 9V delta 2 T cells in vivo: the role of exogenous IL-2. J. Immunol.
175: 1593–1598.
38. Testi, R., and L. L. Lanier. 1989. Functional expression of CD28 on T cell antigen
receptor gamma/delta-bearing T lymphocytes. Eur. J. Immunol. 19: 185–188.
39. Morita, C. T., C. Jin, G. Sarikonda, and H. Wang. 2007. Nonpeptide antigens,
presentation mechanisms, and immunological memory of human Vgamma2Vdelta2 T cells: discriminating friend from foe through the recognition of prenyl
pyrophosphate antigens. Immunol. Rev. 215: 59–76.
40. DeBarros, A., M. Chaves-Ferreira, F. d’Orey, J. C. Ribot, and B. Silva-Santos.
2011. CD70-CD27 interactions provide survival and proliferative signals that
regulate T cell receptor-driven activation of human gd peripheral
blood lymphocytes. Eur. J. Immunol. 41: 195–201.
41. Bonneville, M., R. L. O’Brien, and W. K. Born. 2010. Gammadelta T cell effector functions: a blend of innate programming and acquired plasticity. Nat.
Rev. Immunol. 10: 467–478.
42. Jensen, K. D., X. Su, S. Shin, L. Li, S. Youssef, S. Yamasaki, L. Steinman,
T. Saito, R. M. Locksley, M. M. Davis, et al. 2008. Thymic selection determines
gammadelta T cell effector fate: antigen-naive cells make interleukin-17 and
antigen-experienced cells make interferon gamma. Immunity 29: 90–100.
43. Turchinovich, G., and A. C. Hayday. 2011. Skint-1 identifies a common molecular mechanism for the development of interferon-g-secreting versus
interleukin-17-secreting gd T cells. Immunity 35: 59–68.
44. Mahtani-Patching, J., J. F. Neves, D. J. Pang, K. V. Stoenchev, A. M. AguirreBlanco, B. Silva-Santos, and D. J. Pennington. 2011. PreTCR and TCRgd signal
initiation in thymocyte progenitors does not require domains implicated in receptor oligomerization. Sci. Signal. 4: ra47.
45. Powolny-Budnicka, I., M. Riemann, S. Tänzer, R. M. Schmid, T. Hehlgans, and
F. Weih. 2011. RelA and RelB transcription factors in distinct thymocyte populations control lymphotoxin-dependent interleukin-17 production in gd T cells.
Immunity 34: 364–374.
46. Do, J. S., P. J. Fink, L. Li, R. Spolski, J. Robinson, W. J. Leonard, J. J. Letterio,
and B. Min. 2010. Cutting edge: spontaneous development of IL-17-producing
gamma delta T cells in the thymus occurs via a TGF-beta 1-dependent mechanism. J. Immunol. 184: 1675–1679.
47. Born, W. K., and R. L. O’Brien. 2011. gd T cells develop, respond and survive with a little help from CD27. Eur. J. Immunol. 41: 26–28.
48. Martin, B., K. Hirota, D. J. Cua, B. Stockinger, and M. Veldhoen. 2009.
Interleukin-17-producing gammadelta T cells selectively expand in response to
pathogen products and environmental signals. Immunity 31: 321–330.
49. Kapsenberg, M. L. 2009. Gammadelta T cell receptors without a job. Immunity
31: 181–183.
50. Pantelyushin, S., S. Haak, B. Ingold, P. Kulig, F. L. Heppner, A. A. Navarini, and
B. Becher. 2012. Rorgt+ innate lymphocytes and gd T cells initiate psoriasiform
plaque formation in mice. J. Clin. Invest. 10.1172/JC161862.
Downloaded from http://www.jimmunol.org/ by guest on June 18, 2017
22. Gonçalves-Sousa, N., J. C. Ribot, A. deBarros, D. V. Correia, I. Caramalho, and
B. Silva-Santos. 2010. Inhibition of murine gammadelta lymphocyte expansion
and effector function by regulatory alphabeta T cells is cell-contact-dependent
and sensitive to GITR modulation. Eur. J. Immunol. 40: 61–70.
23. Ho, M., H. K. Webster, P. Tongtawe, K. Pattanapanyasat, and W. P. Weidanz.
1990. Increased gamma delta T cells in acute Plasmodium falciparum malaria.
Immunol. Lett. 25: 139–141.
24. Roussilhon, C., M. Agrapart, J. J. Ballet, and A. Bensussan. 1990. T lymphocytes
bearing the gamma delta T cell receptor in patients with acute Plasmodium
falciparum malaria. J. Infect. Dis. 162: 283–285.
25. Perera, M. K., R. Carter, R. Goonewardene, and K. N. Mendis. 1994. Transient
increase in circulating gamma/delta T cells during Plasmodium vivax malarial
paroxysms. J. Exp. Med. 179: 311–315.
26. Dieli, F., M. Troye-Blomberg, S. E. Farouk, G. Sireci, and A. Salerno. 2001.
Biology of gammadelta T cells in tuberculosis and malaria. Curr. Mol. Med. 1:
437–446.
27. D’Ombrain, M. C., D. S. Hansen, K. M. Simpson, and L. Schofield. 2007.
gammadelta-T cells expressing NK receptors predominate over NK cells and
conventional T cells in the innate IFN-gamma response to Plasmodium falciparum malaria. Eur. J. Immunol. 37: 1864–1873.
28. D’Ombrain, M. C., L. J. Robinson, D. I. Stanisic, J. Taraika, N. Bernard,
P. Michon, I. Mueller, and L. Schofield. 2008. Association of early interferongamma production with immunity to clinical malaria: a longitudinal study
among Papua New Guinean children. Clin. Infect. Dis. 47: 1380–1387.
29. Shahinian, A., K. Pfeffer, K. P. Lee, T. M. Kündig, K. Kishihara, A. Wakeham,
K. Kawai, P. S. Ohashi, C. B. Thompson, and T. W. Mak. 1993. Differential
T cell costimulatory requirements in CD28-deficient mice. Science 261: 609–
612.
30. Silva-Santos, B., D. J. Pennington, and A. C. Hayday. 2005. Lymphotoxinmediated regulation of gammadelta cell differentiation by alphabeta T cell
progenitors. Science 307: 925–928.
31. Pamplona, A., A. Ferreira, J. Balla, V. Jeney, G. Balla, S. Epiphanio, A. Chora,
C. D. Rodrigues, I. P. Gregoire, M. Cunha-Rodrigues, et al. 2007. Heme
oxygenase-1 and carbon monoxide suppress the pathogenesis of experimental
cerebral malaria. Nat. Med. 13: 703–710.
32. Trager, W., and J. B. Jensen. 1976. Human malaria parasites in continuous
culture. Science 193: 673–675.
33. Correia, D. V., F. d’Orey, B. A. Cardoso, T. Lança, A. R. Grosso, A. deBarros,
L. R. Martins, J. T. Barata, and B. Silva-Santos. 2009. Highly active microbial
phosphoantigen induces rapid yet sustained MEK/Erk- and PI-3K/Akt-mediated
signal transduction in anti-tumor human gammadelta T-cells. PLoS ONE 4:
e5657.
34. Correia, D. V., M. Fogli, K. Hudspeth, M. G. da Silva, D. Mavilio, and B. SilvaSantos. 2011. Differentiation of human peripheral blood Vd1+ T cells expressing
the natural cytotoxicity receptor NKp30 for recognition of lymphoid leukemia
cells. Blood 118: 992–1001.
35. Fraser, J. D., B. A. Irving, G. R. Crabtree, and A. Weiss. 1991. Regulation of
interleukin-2 gene enhancer activity by the T cell accessory molecule CD28.
Science 251: 313–316.
7