The Journal of Immunology
T Cell-Signaling Network Analysis Reveals Distinct
Differences between CD28 and CD2 Costimulation Responses
in Various Subsets and in the MAPK Pathway between
Resting and Activated Regulatory T Cells
Maria Elisabeth Kalland, Nikolaus Gu ¨nter Oberprieler,1Torkel Vang, Kjetil Taske ´n, and
Knut Martin Torgersen2
To uncover signaling system differences between T cell stimuli and T cell subsets, phosphorylation status of 18 signaling proteins
at six different time points following TCR triggering and CD28/CD2 costimulation was examined in human T cell subsets by
phospho-epitope–specific flow cytometry of fluorescent cell barcoded samples, thereby providing a high-resolution signaling map.
Compared with effector/memory T cells, naive T cells displayed stronger activation of proximal signaling molecules after TCR
triggering alone. Conversely, distal phosphorylation events, like pErk and pS6-ribosomal protein, were stronger in effector/
memory subsets. CD28 costimulation specifically induced signaling necessary for proper NF-kB activation, whereas CD2 signaled
more strongly to S6-ribosomal protein. Analysis of resting regulatory T cells (rTregs; CD4+CD45RA+FOXP3+) and activated
regulatory T cells (actTregs; CD4+CD45RA2FOXP3++) revealed that, although rTregs had low basal, but inducible, Erk activity,
actTregs displayed high basal Erk phosphorylation and little or no Akt activation. Interestingly, the use of Mek inhibitors to block
Erk activation inhibited activation-dependent FOXP3 upregulation in rTregs, their transition to actTregs, and the resulting
increase in suppressive capacity. In summary, our systems approach unraveled distinct differences in signaling elicited by
CD28 and CD2 costimulation and between rTregs and actTregs. Blocking rTreg transition to highly suppressive actTregs by
Mek inhibitors might have future therapeutic applications. The Journal of Immunology, 2011, 187: 5233–5245.
receptors, such as cytokine receptors (1–3). Triggering of the TCR
complex leads to activation of the lymphocyte-specific protein
tyrosine kinase (Lck) (4, 5), which phosphorylates ITAMs (6) in
the cytoplasmic portions of the TCR-associated CD3 and z-chains
(7). Phosphorylated ITAMs serve as docking sites for the tandem
evelopment and differentiation of CD4+T cells are
controlled by signals provided by the clonotypic TCR,
costimulatory molecules (e.g., CD28 and CD2), and other
Src homology 2 domains of the protein tyrosine kinase ZAP70 (8).
Importantly, this positions ZAP70 in the vicinity of Lck, which
activates the former through phosphorylation of key tyrosine res-
idues (9). Activated ZAP70 phosphorylates the transmembrane
adaptor molecule linker for activation of T cells (LAT) (10) and,
subsequently, the cytosolic adaptor molecule Src homology 2
domain-containing leukocyte-specific phosphoprotein of 76 kDa
(Slp-76), leading to formation of a LAT/Slp-76 nucleated signal-
ing complex (3). This complex orchestrates the activation of sig-
naling proteins involved in a number of processes important for
T cell activation, including reorganization of the cytoskeleton
through activation of phospholipase Cg1(PLCg1), which controls
the calcium- and diacylglycerol-mediated activation of calcineurin,
NFAT, protein kinase C u (PKCu), and Ras guanyl nucleotide-
releasing protein (RasGRP). RasGRP is a positive regulator of
Ras and, hence, the Ras–Raf–Mek–Erk-signaling pathway, leading
to activation of the transcription factor AP-1 (3). In contrast, PKCu
is implicated in signaling events, eventually activating NF-kB.
However, in contrast to NFAT and AP-1, NF-kB activation is
very weak when only the TCR is triggered (11).
The presence of costimulation lowers the threshold for TCR
signaling but also increases the probability that activated cells will
commit to proliferation and subsequent differentiation (12). The
underlying molecular mechanisms for the effects of costimula-
tion have been elusive, but most probably involve augmentation of
signaling pathways that are also activated by TCR stimulation
alone. Importantly, NF-kB seems to be a key player downstream of
costimulatory molecules. For instance, costimulation through CD28
induces potent activation of NF-kB (11, 13). Two signaling path-
ways have been implicated to explain the link between CD28 and
NF-kB, and there may even be cross-talk between these. The first
pathway involves Grb2/Vav-mediated activation of PLCg1/PKCu
Center for Molecular Medicine, Nordic European Molecular Biology Laboratory
Partnership, University of Oslo, N-0318 Oslo, Norway; and Biotechnology Center
of Oslo, University of Oslo, N-0318 Oslo, Norway
1Current address: AstraZeneca Norway AS, Oslo, Norway.
2Current address: Pfizer AS, Oslo, Norway.
Received for publication June 20, 2011. Accepted for publication September 15,
This work was supported by grants from the Norwegian Functional Genomics Pro-
gram, the Research Council of Norway, and the Norwegian Cancer Society. M.E.K. is
a Ph.D. fellow of the Norwegian Cancer Society.
Address correspondence and reprint requests to Prof. Kjetil Taske ´n, Centre for Mo-
lecular Medicine Norway, Nordic European Molecular Biology Laboratory Partner-
ship, University of Oslo, P.O. Box 1137 Blindern, N-0318 Oslo, Norway. E-mail
The online version of this article contains supplemental material.
Abbreviations used in this article: 7-AAD, 7-amino-actinomycin D; actTreg, acti-
vated (CD4+CD45RA2FOXP3++) regulatory T cell; arcsinh, inverse hyperbolic sine;
Ax647, Alexa Fluor 647; BD, Becton Dickinson; CNS, conserved noncoding se-
quence; CsA, cyclosporin A; FCB, fluorescent cell barcoding; LAT, linker for acti-
vation of T cells; Lck, lymphocyte-specific protein-tyrosine kinase; MFI, mean
fluorescent intensity; mTOR, mammalian target of rapamycin; phospho-flow, phos-
pho-epitope–specific flow; PKCu, protein kinase C u; PLCg1, phospholipase Cg1;
rTreg, resting (CD4+CD45RA+FOXP3+) regulatory T cell; Slp-76, Src homology
2 domain-containing leukocyte-specific phosphoprotein of 76 kDa; S6-Rp, S6-
ribosomal protein; Tconv, conventional T cell; Treg, regulatory T cell.
(14, 15), whereas the second encompasses activation of PI3K,
phosphatidylinositol-dependent kinase 1, and Akt. The relatively
mild phenotype associated with CD28 deficiency suggests that
other costimulatory molecules can compensate for the loss of
CD28, and CD2 has been proposed to play such a role (16–18).
CD2 can act both as an adhesion molecule and a signaling mol-
ecule. In fact, CD2 cross-linking can induce proliferation of
T cells, as well as cytokine secretion, both in a ZAP70-dependent
manner (19). Independently of TCR stimulation, interactions be-
tween CD2 and its ligand CD58 can also induce signaling through
formation of CD2 clusters in distinct membrane microdomains
(20). Moreover, transgenic expression of CD2 in developing thy-
mocytes results in increased apoptosis at the double-positive stage,
suggesting that CD2 generates signals that resemble the ones de-
rived from the TCR (21). Along these lines, both Lck and Fyn
have been shown to interact with CD2 and to become activated
in response to CD2 cross-linking (22, 23). Similarly to CD28-
deficient T cells, CD2-deficient T cells are only mildly affected,
indicating some redundancy (24). However, the combined lack of
CD28 and CD2 leads to profound defects in activation and pro-
liferation of T cells (16), suggesting the necessity of at least one of
these costimulatory molecules for proper T cell function.
Activation of CD4+T cells is required for a majority of adaptive
immune responses, and a balanced reaction is essential to avoid
excessive tissue damage and autoimmunity. Consequently, con-
ventional CD4+T cells are kept in check by regulatory T cells
(Tregs) (25). Tregs are characterized by the expression of the
transcription factor FOXP3 and their ability to suppress conven-
tional T cells (Tconvs) (26–29). A recent paper suggested that
CD4+T cells in human blood can be divided into five categories
based on the expression of CD45RA, CD25, and FOXP3 (30).
These subsets are conventional naive CD4+T cells (CD45RA+
CD252FOXP32), conventional effector/memory CD4+T cells
(CD45RA2CD252FOXP32), CD4+FOXP3+effector T cells with
cytokine-producing capabilities (CD45RA2CD25+FOXP3+), rest-
ing Tregs (rTregs; CD45RA+CD25+FOXP3+), and activated Tregs
(actTregs; CD45RA2CD25++FOXP3++). Given proper stimulatory
conditions, naive Tconvs can proliferate and differentiate to become
effector/memory Tconvs. In a similar manner, rTregs can be con-
sidered precursors of actTregs. The observation that, in response
to activation, rTregs can proliferate and differentiate into Ki-67+
FOXP3++CD45RO+cells with suppressive abilities comparable to
actTregs supports this notion (30).
Because the abovementioned CD4+FOXP3+T cell subsets
combined constitute ,10% of all CD4+T cells in peripheral blood,
traditional biochemical studies of signal transduction in these cells
have been challenging. However, recent technical developments,
including fluorescent cell barcoding (FCB) (31) and a growing
number of phospho-epitope–specific Abs, have made it possible to
use phospho-epitope–specific flow (phospho-flow) cytometry-based
methods to study signaling processes at single-cell resolution in
several phenotypically defined T cell subsets simultaneously (32–
34). Furthermore, these developments have enabled an increase in
the resolution to a level at which signaling differences can be linked
to functional properties in small subsets of cells. To understand
systems-level signaling differences between different T cell stimuli
and in various T cell subsets, we used this technique to investigate
signaling and the role of CD28 and CD2 costimulation in different
human CD4+T cell subsets, including the CD4+FOXP3+subsets.
We were able to reproducibly identify signaling characteristics and
define distinct differences between CD28 and CD2 in the activation
of NF-kB and the S6-ribosomal protein (S6-Rp) transcriptional
program, respectively. In addition, differences in Akt versus Mek1/
Erk signaling between rTregs and actTregs, combined with func-
tional studies with signaling inhibitors, suggested that the activity
level of Mek1/Erk and Akt pathways is involved in defining a
functional switch between these two subsets.
Materials and Methods
Cyclosporin A (cat. no. 239835), PI-103 (cat. no. 528100), Akt Inhibitor
VIII (cat. no 124017), rapamycin(cat. no. 553211), and SB 203580 (cat. no.
559395) were purchased from Calbiochem; U0126 (cat. no. 9903) and
wortmannin (cat. no. 9951) were purchased from Cell Signaling; and
PD0325901 (cat. no. 1408) was purchased from Axon Medchem. Abs used
for T cell stimulation included anti-CD3 (clone OKT3) custom produced
from the hybridoma by Diatec, anti-CD28 (cat. no. 13-0289; eBioscience),
from Invitrogen, and 7-amino-actinomycin D (7-AAD) was from Becton
Dickinson (cat. no. 559925; BD). Abs used to detect the phosphorylation of
CD3z(Y142) (cat. no. 558489), LAT(Y171) (cat. no. 558518), MEK1
(S298) (cat. no. 560043), NF-kB p65(S529) (cat. no. 558422), Slp-76
(Y128) (cat. no. 558438), STAT3(Y705) (cat. no. 557815), Rb(S807/S811)
(cat. no. 558590), ZAP70/Syk(Y319/Y352) (cat. no. 557817), and the
isotype control IgG1 k (cat. no. 557783) were from BD. Abs used to detect
the phosphorylation of Akt/PKB(S473) (cat. no. 4075), histone H3(S10)
(cat. no. 9716), NF-kB p65(S536) (cat. no. 4887), MAPKAPK-2(T334)
(cat. no. 4320), S6-Rp(S235/236) (cat. no. 4851), tyrosine (Y100) (cat. no.
9415), 44/42 MAPK(T202/Y204) (cat. no. 4375), and p38 MAPK(T180/
Y182) (cat. no. 4552) were from Cell Signaling Technology. Ab used to
detect the phosphorylation of ATF-2(T71) (cat. no. sc-8398) was from
Santa Cruz. Furthermore, Abs used to detect the expression of CD3 (cat.
no. 345766), CD4 (cat. no. 348809 and 557922), CD25 (cat. no. 557741),
CD45RA (cat. no. 555489), CD45RO (cat. no. 555493), and FOXP3 (cat.
no. 560047 and 560045) were from BD, whereas the Ab used to detect the
expression of CD8 (cat. no. 9536-09) was from Southern Biotech.
T cell purification, stimulation, and fixation
Buffy coats were obtained from healthy blood donors (Oslo University
Hospital Blood Centre, Oslo, Norway), and studies were approved by the
Regional Ethics Review Board. T cells, either CD3+or CD4+, were purified
by negative selection from buffy coats using RosetteSep Enrichment kits
(StemCell Technologies), according to the manufacturer’s instructions.
Unless otherwise stated, purified T cells were resuspended in RPMI 1640
GlutaMAX (Life Technologies) with 1% FCS. Prior to stimulation, cells
were pre-equilibrated at 37˚C for 5 min. Thereafter, biotinylated Abs were
added (different concentrations of anti-CD3 alone [range, 1 ng/ml–10 mg/
ml] or a fixed concentration of anti-CD3 [1 mg/ml] alone or together with
anti-CD28 [5 mg/ml], anti-CD2 [5 mg/ml], or anti-CD28 plus anti-CD2
[each, 5 mg/ml]). Two minutes later, avidin (50 mg/ml) was added to allow
cross-linking, and incubations were continued for different time periods. In
some experiments (specifically stated in the figure legends), cells were
stimulated using a different protocol that included incubation on ice with
biotinylated Abs for 30 min, one wash, addition of prewarmed avidin, and
incubation at 37˚C for different time periods. All harvested samples were
fixed immediately using prewarmed BD Phosflow Fix Buffer I (BD Bio-
sciences) for 10 min at 37˚C, followed by centrifugation (830 3 g, 4˚C,
5 min), one wash with flow washing solution (PBS containing 1% FCS and
0.09% sodium azide), and one wash with PBS.
Fluorescent cell barcoding
Three-dimensional FCB was carried out, as previously described (34, 35).
In brief, fixed cells were incubated with varying concentrations of esters
conjugated to Pacific Blue (100, 25, 6.3, 0.7 pg/ml; Molecular Probes, In-
vitrogen), Pacific Orange (1000, 250, 41.7, 4.2 pg/ml; Molecular Probes,
Invitrogen), and Alexa Fluor 488 (50,12.5, 3.1, 0.3 pg/ml;Molecular Probes,
Invitrogen)for 20 min atroomtemperature.Eachsamplewas incubated with
a unique combination of dye concentrations, allowing identification of each
sample based on its color coding combination. Following FCB, all samples
werecentrifuged and washedseparately oncein PBS with 3%FCS.Later, all
samples were combined and washed once with flow-washing solution, per-
stored at 280˚C (henceforth called FCB cell stock).
Staining of samples and analysis by phospho-flow cytometry
FCB cell stocks were rehydrated with PBS and washed once in flow-washing
solution. Aliquots of FCB cells were then incubated with different combi-
nations of Alexa Fluor 647 (Ax647)-conjugated phospho-epitope–specific
5234 T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
Abs and fluorescently labeled cell surface-marker Abs (30 min at room
temperature), washed twice with flow-washing solution, and made ready
for flow cytometric analysis by resuspension in flow-washing solution.
FOXP3 staining, when included, was performed on FCB cell stocks prior
to permeabilization and storage at 280˚C, using a FOXP3 staining kit from
BD Pharmingen (cat. no. 560098). Finally, all samples were analyzed
using a BD FACSCanto II (4-2-2) cytometer equipped with 405-, 488-, and
633-nm lasers. For each sample, $200,000 events were recorded (corre-
sponding to .3000 events per square in the heatmaps for prevalent cell
populations and .300 events for less prevalent populations). For fluoro-
chrome compensation, PerCP-conjugated CD3 Ab, PE-Cy7–conjugated
CD4 Ab, and PE-conjugated CD8, CD45RA, or CD45RO Ab staining was
used on unstimulated non-FCB cells. Ax647-conjugated CD3 Ab was used
to compensate for the phospho-epitope–specific Abs conjugated to Ax647.
The data-analysis program Cytobank (http://cytobank.stanford.edu) and
FlowJo 8.8.2 (TreeStar, Ashland, OR) were used for further analysis and
visualization of data.
Sorting of CD4+T cell subsets
Purified CD4+T cells in PBS with 2% FCS were incubated with fluo-
rescently labeled Abs (anti–CD45RA-PE and anti–CD25–PE-Cy7, anti–
CD4-Alexa Fluor 700 was also used unless subsequent suppression assays
were to be performed) for 30 min on ice. Thereafter, cells were washed
once in PBS with 2% FCS, resuspended in PBS with 2% FCS, and sorted
on a BD FACSAria IIu cytometer (5-2) equipped with 488- and 633-nm
lasers and set up with a sheath pressure of 70 pound-force per square inch
and a 70-mm nozzle.
In vitro stimulation of rTregs
rTregs, defined as CD4+CD45RA+CD25+T cells, were sorted as described.
Sorted cells were resuspended at 1 3 106cells/ml in complete medium
(RPMI 1640 containing 10% FCS, 100 U/ml penicillin, 0.1 mg/ml strep-
tomycin, 1 mM sodium pyruvate, and nonessential amino acids), incubated
with or without specific inhibitors of PI3K (0.5 mM PI-103 in combination
with 0.1 mM wortmannin), Akt (either 1 mM Akt1/2 inhibitor or 10 nM of
the mammalian target of rapamycin (mTOR) inhibitor, rapamycin), Mek
(either 1 mM PD0325901 or 10 mM U0126), p38 (5 mM SB 203580), or
NFAT (0.1 mM of the calcineurin inhibitor, cyclosporin A [CsA]) pathways
for 20 min at 37˚C, followed by stimulation with aCD3/CD28/CD2-coated
MicroBeads (Miltenyi Biotec; bead/cell ratio of 1:5, which was used in all
bead-based experiments presented in Figs. 5 and 6) for different time
periods. The same experimental setup was also used in additional studies,
using purified CD4+T cells to examine all of the inhibitors’ effect on
viability and for titration of optimal concentrations of inhibitors on
stimulation-induced upregulation of the fractions of rTregs and actTregs.
The defined optimal inhibitor concentrations were used in subsequent
experiments. We next tested to what extent inhibition of the Mek–Erk- or
Akt-signaling pathways affected the stimulation-induced upregulation of
FOXP3 in rTregs. One sample was left unstimulated as a reference.
Unstimulated, sorted actTregs (defined as CD4+CD45RA2CD25++T cells)
and naive Tconvs (defined as CD4+CD45RA+CD252T cells) were in-
cluded as controls. After stimulation for the intended time periods, samples
were washed once in flow-washing solution before staining with 7-AAD.
Later, cells were fixed with Buffer A from a FOXP3 staining kit (BD
Pharmingen), according to the manufacturer’s protocol, and stored at 280˚C.
Subsequently, all samples were stained for FOXP3 and surface markers and
analyzed as described above.
Sorted rTregs were either added directly into a suppression assay or pre-
treated (30 min at 37˚C) or not with specific inhibitors of PI3K, Akt, Mek,
p38, and NFAT pathways, followed by 36 h of culture in complete medium
alone or in the presence of aCD3/CD28/CD2-coated MicroBeads. After
two rounds of washing, cells were mixed with CFSE-stained, purified
CD4+T cells (called responder cells) at a 1:1 ratio and stimulated
with aCD3/CD28/CD2-coated MicroBeads for 84 h. Cells were then
stained with 7-AAD and subjected to flow cytometric analysis using a BD
FACSCanto II. The suppressive capacity of different rTreg populations was
determined by the level of CFSE dilution in responder cells using FlowJo
Processing of data and statistical analysis
Changes in phosphorylation of signaling proteins following activation of
T cells were calculated using the inverse hyperbolic sine (arcsinh) of the
median fluorescence intensity (MFI) of stimulated versus unstimulated cell
populations. The reason for choosing arcsinh for calculating changes was
explained by Irish et al. (36). Comparison of the effects of preincubation
of rTregs with different specific inhibitors on FOXP3 expression was
analyzed using one-way ANOVA, specifying unstimulated rTregs as the
control group. Comparisons of the effects of specific inhibitors on the
suppressive capacity of rTregs were also analyzed using one-way ANOVA;
stimulated rTregs preincubated in the absence of an inhibitor was used as
the control group. Differences in mean values were considered statistically
significant when p , 0.05. The statistical analyses were conducted using
SigmaPlot 11.2 (Systat Software).
High-throughput analysis of T cell-signaling profiles
Our experimental setup is outlined in Fig. 1A. After subjecting
cells to different stimulatory conditions and fixation with form-
aldehyde, cells from each stimulatory condition were stained with
a unique combination of FCB reagents and, therefore, could be
tracked from all other sample populations in subsequent assays.
Thus, FCB allows for combining all cell samples prior to staining
with fluorescently labeled Abs against intracellular phospho-
epitopes and cell surface markers, thereby analyzing all sam-
ples with the same baseline, minimizing intra-assay variability
and allowing for high-throughput analysis (31).
To define the sensitivity of the experimental system, initial as-
says involvedincubation of purified CD3+human T cells (95 6 3%
purity, data not shown) with different concentrations of anti-CD3
Ab (1 ng/ml–10 mg/ml) on ice, followed by one round of washing,
and then cross-linking with avidin at 37˚C for up to 60 min. Fig.
1B shows the subsequent analysis of these cells with regard to
the phosphorylation status of five selected signaling intermediates
downstream of the TCR (see Supplemental Fig. 1 for 13 additional
signaling parameters and data from more individuals). TCR-
proximal–signaling molecules (such as z-chain, ZAP70, LAT, and
Slp-76) were considerably activated/phosphorylated only at anti-
CD3 concentrations $1 mg/ml, and they peaked after 1–3 min(Fig.
1B). In accordance with the principle of signal-cascade amplifi-
cation, signaling mediators located more downstream (e.g., Erk,
p38, NF-kB, and S6-Rp) were activated at lower levels of stimu-
lation and with delayed kinetics compared with TCR-proximal–
signaling molecules (Fig. 1B, Supplemental Fig. 1). These obser-
vations were apparent for both CD4+and CD8+T cells. Moreover,
combination of data from three independent donors revealed a
consistent pattern, testifying to the robustness of the method
Alternatively, purified CD3+T cells were pre-equilibrated at
37˚C prior to addition of anti-CD3 Ab, followed 2 min later by
cross-linking withavidinand continuedincubation for upto 60min
(Supplemental Fig. 2). Distinct differences were observed when
analyzing the whole data set with regard to signaling responses and
comparing data obtained from cells incubated on ice with cells
treated at 37˚C (Supplemental Fig. 2). Phosphorylation of TCR-
proximal molecules (such as z-chain and ZAP70) appeared stron-
ger in cells that had been incubated on ice compared with cells only
treated at 37˚C. This could be due to a lower level of CD3 cross-
linking under the latter condition, slower and/or weaker effects of
inhibitory molecules (e.g., protein tyrosine phosphatases) in cells
that had been incubated on ice, or a better synchronized effect of
the activation giving stronger synergy in the costimulation when
allowing Abs to first bind on ice. However, phosphorylation of
TCR-distal molecules, such as Mek1, p38, and NF-kB, were ele-
vated at time 0 following incubation on ice, suggesting a direct
effect of the temperature changes on several signaling processes.
As a result of this, and to optimize signal-to-noise across the full
panel of signal markers, all further experiments were conducted on
cells that had been pre-equilibrated at 37˚C.
The Journal of Immunology5235
Comparison of T cell-signaling profiles for different naive and
effector/memory T cell subsets
for a set of markers across relevant signal pathways in T cell acti-
subsets. Three different stimulatory conditions were used: cross-
linking of anti-CD3 alone, cross-linking of anti-CD3/anti-CD28,
and cross-linking of anti-CD3/anti-CD28/anti-CD2. For all these
experiments,a suboptimalconcentration ofanti-CD3(1mg/ml)was
used to capture effects of the different costimulatory conditions. In
addition to the panel of 18 phospho-specific Abs described above,
staining for CD45RO was included so that naive T cells could be
respectively). Because of the limited number of channels in the
FACS analysis, CD8 staining had to be omitted in this setup. Still,
.90% of CD3+CD42peripheral T cells were CD3+CD8+T cells
(data not shown), indicating that CD4 negativity within the pop-
ulation of peripheral T cells was a good surrogate marker for
CD8 positivity. Therefore, CD3+CD42CD45RO2and CD3+CD42
CD45RO+T cells are referred to as naive and effector/memory
CD8+T cells, respectively, whereas CD3+CD4+CD45RO2and
CD3+CD4+CD45RO+T cells denote naive and effector/memory
CD4+T cells. The signaling in each subset was initially analyzed
relative to the control sample for the same subset (Fig. 2A). Com-
pared with analysis of the entire populations of CD4+and CD8+
peripheral T cells, the addition of CD45RO-based subgating of
CD4+and CD8+T cells revealed differences in signaling responses.
signaling responses seemed to be stronger in naive cells compared
with effector/memory cells. This was particularly evident when
considering Mek1 and Erk, both of which demonstrated a strong
signaling response in naive CD4+T cells but less so in effector/
differently barcoded cell populations. B, Anti-CD3 concentration-dependent increase in T cell signaling in both CD4+and CD8+T cells assessed by
phospho-flow cytometry. Primary human T cells were incubated with different anti-CD3 concentrations on ice, washed once, and then incubated in the
presence of avidin at 37˚C for the indicated time periods. Subsequently, cells were barcoded, stained with fluorescently labeled Abs, and analyzed by FACS.
Data are presented as heatmaps where warmer colors (yellow) indicate an increase in phospho-signals, and colder colors (blue) represent a decrease (for
intensities, see scale bars on the right). All values in a given row are relative to the value in the first column of that row. The data are representative of
experiments for T cells from three individuals (see Supplemental Fig. 1 for all individual data). C, Experiment as in B, but data for T cells from three
individuals are combined. Data are mean 6 SEM (n = 3) of arcsinh median differences of the phospho-epitope–specific fluorescence intensity signals.
High-throughput analysis of T cell-signaling profiles. A, Diagram showing workflow for phospho-flow cytometry. Inset depicts separation of
5236 T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
memory cells. The addition of CD28 costimulation only had a
modest effect on the signaling amplitudes in all four subsets and
then preferentially on downstream mediators, such as S6-Rp. In
contrast, combined costimulation of CD28 and CD2 consistently
gave high signaling amplitudes for all signaling molecules tested
in all of the subsets (Fig. 2A).
cross-linking of the indicated combinations of Abs with avidin and incubated for different time periods. Subsequently, cells were barcoded, stained with
fluorescently labeled Abs, and analyzed by FACS. Phospho-signal profiles for the different T cell subsets are presented as heatmaps, as explained in Fig. 1.
The panel for each signaling molecule consists of two parts: the upper one distinguishes between CD3+CD4+and CD3+CD42T cells, whereas the lower
one also includes naive (CD45RO2) and effector/memory (CD45RO+) subsets within these CD4+and CD42T cell subsets. Data are representative of
experiments for T cells from three individuals (see Supplemental Fig. 2 for all individual data). B, Experiment as in A, but all values are relative to the
phospho-signals obtained for the control sample of the naive CD4+T cell population. Data are representative of experiments with T cells from three
individual blood donors. C, Experiment as in B, but data for T cells from three individuals are combined. Data are mean 6 SEM (n = 3) of arcsinh median
differences of the phospho-epitope–specific fluorescence intensity signals.
Comparison of T cell-signaling profiles for different naive and effector/memory T cell subsets. A, Primary human T cells were stimulated by
The Journal of Immunology 5237
actTregs. A, Gating strategy for distinguishing between CD4+effector and regulatory T cell subsets using CD45RA and FOXP3 Ab staining. B, Primary
human T cells were stimulated by cross-linking of the indicated combinations of Abs with avidin and incubated for different time periods. Cells were then
barcoded, stained with fluorescently labeled Abs, and analyzed by FACS. Upper part of each panel, Phospho-epitope–specific (Figure legend continues)
Analysis of signaling in conventional and regulatory CD4+T cell subsets reveals elevated Erk activation and decreased Akt activation in
5238T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
Next, signaling responses in all subsets where analyzed using
unstimulated naive CD4+T cells as a reference (Fig. 2B, 2C). With
such an approach, several interesting features were observed.
First, compared with CD3 stimulation alone, the addition of CD28
costimulation resulted in higher signaling responses in naive
CD4+T cells, even at the level of TCR-proximal molecules, such
as z-chain and ZAP70. Second, all four subsets displayed higher
signaling responses with two rather than one costimulatory factor
present and with an especially strong effect of CD2 costimulation.
Third, although phosphorylation of TCR-proximal–signaling
molecules (such as z-chain, ZAP70, and Slp-76) generally was
stronger in naive T cells than in effector/memory T cells (both
CD4+and CD8+cells), the opposite was the case for more
downstream mediators (e.g., Erk and S6-Rp). Fourth, in naive
T cells (both CD4+and CD8+), phosphorylation of Mek1 peaked
after 1 min of stimulation and subsequently displayed a second
wave of activation in the presence of CD2 costimulation. This
suggested involvement of a positive-feedback loop as a result of
coreceptor signaling. Somewhat surprisingly, the signaling pat-
terns of Erk did not relate directly to their upstream activator
Mek1. Finally, the relatively high basal phosphorylation of S6-Rp
in effector/memory subsets, and especially for CD8+T cells, in-
dicated an activated transcription and translation program, as ex-
pected for effector cell populations (Fig. 2C). We observed that
addition of CD2 costimulation produced a reproducible increase in
S6-Rp phosphorylation in all of the subsets, irrespective of basal
status, supporting a role for CD2-induced signals in the regulation
of translation at the level of S6-Rp (37). Although the CD2-
regulated S6-Rp phosphorylation was in agreement with the ki-
netics of Akt activation in naive T cells, albeit with a different
amplitude consistent with downstream amplification at the S6-Rp
level, this did not appear to be the case in effector/memory sub-
sets, indicating the influence of other components of a CD2-
regulated signal network in effector/memory T cells (Fig. 2B,
2C, bottom panels), as further addressed in Figs. 3 and 4 below.
Analysis of signaling in conventional and regulatory CD4+
T cell subsets reveals elevated Erk activation and decreased
Akt activation in actTregs
As previously described for humans (30, 38), CD4+Tregs can be
divided into two functionally distinct subsets based on CD45RA
and FOXP3 expression: CD4+CD45RA+FOXP3+rTregs and
CD4+CD45RA2FOXP3++actTregs (Fig. 3A). We hypothesized
that this functional delineation would be reflected in signal-
transduction processes and used the established phospho-flow
cytometry protocol to investigate signaling in these subsets. For
global overview purposes, the analyses also included naive
(CD45RA+FOXP32) and effector/memory (CD45RA2FOXP32)
CD4+Tconvs, as well as the CD4+CD45RA2FOXP3+effector
T cell subset with cytokine-secreting ability (30).
As shown in Fig. 3B and 3D, phosphorylation levels of z-chain
(as well as other TCR-proximal–signaling molecules; Supple-
mental Fig. 3) were comparable among actTregs, rTregs, and naive
CD4+Tconvs, both in response to CD3 stimulation and when
different types of costimulation were added. The same observation
was made for rTregs and naive CD4+Tconvs with regard to Mek1
phosphorylation, whereas the signals for actTregs were weaker and
comparable to the ones seen for effector/memory CD4+Tconvs.
Interestingly, levels of Erk activation in actTregs were very high,
both in unstimulated cells and after CD3 stimulation with or
without costimulation. For all other CD4+T cell subsets tested,
robust Erk activation was observed in response to stimulation, but
the signalswere always significantlyweaker than those inactTregs.
ActTregs also differed from most other subsets with regard to Akt
signaling (Fig. 3C, 3D), which was nearly absent in actTregs, even
in the presence of the strongest stimulus tested (combined cross-
linking of CD3/CD28/CD2). In contrast, CD4+Tconvs, as well as
rTregs, displayed potent Akt activation, especially in the presence
of costimulation. Again, all subsets significantly increased the
levels of S6-Rp phosphorylation when CD2 costimulation was
added, even when the basal activity was elevated, as seen for CD4+
effector/memory T cells, which also differed from the kinetics of
Akt activation. The basal phosphorylation levels and the patterns of
induced responses were reproducible inall threedonors tested (Fig.
3D, 3E, Supplemental Fig. 3).
CD28 and CD2 costimulation trigger overlapping but distinct
To delineate differences between CD28 and CD2 in costimulation,
we examined more carefully the relative contribution of CD2
and CD28 by cross-linking, either alone or combined, with CD3.
Costimulation with CD28 or CD2 separately increased the am-
plitude of proximal TCR-signaling events at the level of z-chain
and Slp-76 phosphorylation compared with CD3 stimulation
alone (Fig. 4). Similar responses were seen for phosphorylation
of ZAP70, Mek1, and histone 3 (Supplemental Fig. 4). Additive
effects of CD28 and CD2 were generally observed (as also seen in
Fig. 3B–D). However, at the level of Erk activation, the addition of
CD28 and/or CD2, compared with CD3 alone, had only a modest
effect on the signaling amplitudes in the different subsets tested,
indicating that Erk phosphorylation depends mainly on the TCR
signal (Fig. 4). Interestingly, the phosphorylation responses of Akt
in naive CD4+T cells and rTregs and in all subsets tested for NF-
kB phosphorylation depended mainly on CD28, indicating that
the CD28 signal may be essential for proper activation of an Akt–
NF-kB pathway in naive CD4+T cells and rTregs. In contrast,
the CD28-dependent activation of NF-kB in other subsets did
not seem to be mediated by Akt (Fig. 4B, Supplemental Fig. 4).
Stimulation through the CD2 receptor appeared to result in
stronger activation of S6-Rp compared with CD28 in the effector
T cell subsets and in actTregs. A clear additive effect of the ac-
tivation of S6-Rp was observed when both costimulators were
Mek-Erk–dependent upregulation of FOXP3 in rTregs
rTregs most likely represent a thymus-derived population that,
become actTregs,which are characterized by enhanced suppressive
capabilities (30). Key events in this maturational process are in-
creased FOXP3 expression and proliferation. Given the significant
differences between rTregs and actTregs with respect to activation
of Akt and Erk (as described in Figs. 3, 4), we aimed to address
signal profiles for the different T cell subsets are presented as heatmaps with the first column of each row set as reference, as explained in Fig. 1. Lower part
of each panel, Heatmaps in which all values are relative to the phospho-signals obtained for the control sample of the total naive CD4+T cell population
allowing for comparison of subsets. Data are representative of experiments conducted on T cells from three individuals (see Supplemental Fig. 3 for all
individual data). C, As in B, but additional phospho-signals were analyzed. D, Experiment as in the lower parts of the panels in B and C, but data from three
individuals are combined. Data are presented as mean 6 SEM (n = 3) of arcsinh median differences of the phospho-epitope–specific fluorescence intensity
signals. E, As in D, but only unstimulated cells are included in the analysis.
The Journal of Immunology5239
cross-linking of the indicated combinations of Abs with avidin and incubated for different time periods. Cells were then barcoded, stained with fluo-
rescently labeled Abs, and analyzed by FACS. The data are presented as in Fig. 3 and are representative of experiments from three separate blood donors
(see Supplemental Fig. 4 for all individual data). B, Amalgamated data with normalization to the CD4+CD45RA+FOXP32subset, as in the lower panels in
A. Data are mean 6 SEM (n = 3) of arcsinh median differences of the phospho-epitope–specific fluorescence intensity signals.
Costimulation with CD28 is essential for proper activation of NF-kB–related signaling. A, Primary human CD4+T cells were stimulated by
5240T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
the importance of different signaling pathways in the transition of
rTregs to actTregs. To first assess the induction of FOXP3 protein
in rTregs upon activation, CD4+CD25+CD45RA+T cells were
sorted and stimulated with aCD3/CD28/CD2-coated Microbeads
in vitro for up to 92 h, followed by flow cytometry analysis of
FOXP3 levels. As seen in Fig. 5A and 5B, FOXP3 expression in
rTregs increased markedly in a time-dependent manner in re-
sponse to stimulation. Peak levels, which were reached after 36–
44 h, even exceeded the FOXP3 levels observed for actTregs
isolated directly from blood (Fig. 5A). At later time points, FOXP3
expression in stimulated rTregs decreased, suggesting that tran-
siently high expression of FOXP3 was necessary to drive the
transcriptional program necessary for the maturation of these
cells. No significant stimulation-induced upregulation of FOXP3
was observed in conventional naive and effector/memory CD4+
T cell subsets under the same conditions (data not shown).
We next tested towhat extent inhibition of the Mek–Erk- or Akt-
signaling pathways affected the stimulation-induced upregulation
of FOXP3 in rTregs. To do so, all inhibitors were first tested for
toxicity, and their effect on upregulation of rTreg and actTreg
subsets was assessed over a range of concentrations (see Fig. 5C
for titration of the Mek inhibitor PD 0325901). Subsequently,
sorted rTregs (purity .98%, data not shown) were incubated with
the panel of inhibitors at defined concentrations prior to stimula-
tion for 36 h, as shown in Fig. 5D. Interestingly, Mek inhibitors
(PD 0325901 and U0126) that would prevent activation of Erk
potently inhibited the stimulation-induced upregulation of FOXP3
(.90%) without any effect on viability (data not shown), whereas
pretreatment with inhibitors against PI3K (PI-103 in combination
with wortmannin), mTOR (rapamycin), Akt (Akt1/2-inhibitor), or
calcineurin (CsA) reduced the FOXP3 induction by ∼50%. These
effects were consistent between several donors (Fig. 5E), indi-
cating a crucial role for Mek-related signaling in activation-
induced upregulation of FOXP3 in rTregs. In comparison, inhi-
bition of p38 (SB 203580) had no significant effect.
Induction of rTreg-suppressive capacity is Mek dependent
We next tested the functional consequence of blocking the
activation-induced upregulation of FOXP3 in rTregs. In these
assays, Treg function was defined as the ability to suppress the
proliferation of CFSE-labeled purified CD4+T cells. Notably,
suppressive function could not be demonstrated for sorted rTregs
indicated time periods, followed by FOXP3 staining and FACS analysis. Sorted actTregs were included as controls. Data are representative of experiments
with T cells from three separate blood donors. B, Experiments as in A were performed with T cells from three individuals. The bars represent relative
increase in FOXP3 expression after stimulation for the indicated time periods. Relative increase in FOXP3 expression was calculated using MFI and the
following formula: ΔMFI = [MFI (stimulated) – MFI (unstimulated)]/MFI (unstimulated). Data are given as mean 6 SEM (n = 3). C, Effects of various
concentrations of Mek inhibitor on stimulation-induced increase in rTreg (left panel) and actTreg (right panel) subsets. Purified CD4+T cells were in-
cubated with indicated concentrations of Mek inhibitor (PD0325901, 20 min) and then stimulated (aCD3/CD28/CD2-coated MicroBeads) for 36 h. One
sample was kept unstimulated and used as a reference. Thereafter, FOXP3 staining and FACS analysis were conducted. The bars (mean 6 SEM; n = 3)
represent increases in rTreg and actTreg subsets after stimulation relative to the unstimulated reference sample. Data are representative of three individual
blood donors analyzed in duplicate. D, Experiment as in A, but sorted rTregs were incubated with the indicated inhibitors of Mek (either PD0325901 or
U0126), p38 (SB 203580), PI3K (PI-103 in combination with wortmannin), Akt (Akt1/2-inhibitor), mTOR (rapamycin), or calcineurin (CsA) for 20 min,
followed by stimulation with aCD3/CD28/CD2-coated MicroBeads for 36 h. E, Experiments as in D were performed with T cells from three individuals
and otherwise analyzed as outlined in B. Data are presented as mean 6 SEM (n = 3). **p , 0.01, ***p , 0.001.
Mek-dependent upregulation of FOXP3 in rTregs. A, Sorted rTregs were stimulated or not with aCD3/CD28/CD2-coated MicroBeads for the
The Journal of Immunology5241
added back on the day of sorting (Fig. 6; purity .98%, data not
shown). Further studies of suppressive capacity of rTregs from six
individual blood donors compared with that of actTregs revealed
little or no suppression by rTregs without prior stimulation (Table
I). However, when rTregs were stimulated for 36 h (to assure
proper FOXP3 upregulation) before initiation of the CFSE assay,
significant suppression was observed. Finally, incubation of rTregs
with a Mek inhibitor prior to stimulation and subsequent initiation
of the CFSE assay completely blocked the ability of these cells to
become suppressive. In comparison, inhibition of Akt resulted
in 50% reduction in suppressive capacity. These results were con-
sistent among all donors tested (Fig. 6B) and suggest that the
degree to which these inhibitors prevent FOXP3 upregulation
defines their potency to restrain induction of suppressive function
T cell signaling has been extensively studied for decades, and the
various signaling pathways involved have been well characterized.
However, there is a need for an overview of how distinct signal
pathways integrate and cross-talk to form signaling networks.
Furthermore, with the increasing number of different T cell subsets
being defined, especially among the group of CD4+T cells, and the
possibility of signaling differences between various subsets, there
is a need for more subset-specific analyses to link signaling ac-
tivity to functional properties. Because several CD4+T cell sub-
sets are relatively low in abundance (e.g., rTregs and actTregs),
such analyses are challenging using traditional biochemical
methods. Recent technical advances have made it possible to
obtain cell-signaling data even from a relatively low number of
cells. One such approach is phospho-flow cytometry in combi-
nation with FCB. In this report, we described how this technique
can be exploited to assess the phosphorylation status of 18 im-
portant signaling intermediates downstream of the TCR and co-
stimulatory molecules in various CD4+and CD8+T cell subsets.
Of note, this assay system captured clear signaling differences
between CD28 and CD2 with respect to activation of NF-kB and
the S6-Rp program, as well as between rTregs and actTregs where
our results indicated that a Mek–Erk-signaling pathway is nec-
essary for the maturation and functional control of human act-
The comprehensive analysis of T cell signaling described in this
report revealed time-dependent activation/inactivation of the sig-
naling intermediates tested. Generally, TCR-proximal signaling
proteins were transiently activated within 1–3 min after TCR cross-
linking; the time kinetics for intermediate signaling molecules
were slower; and, finally, the most TCR-distal molecules analyzed
(e.g., histone 3) displayed the slowest activation kinetics, peaking
after 10–30 min and with signals persisting throughout the dura-
tion of the assay (60 min). These findings are in agreement with
data obtained with conventional methods (e.g., immunoblotting)
in the past, indicating that the phospho-flow approach is suffi-
ciently sensitive to capture relevant and established alterations in
the phosphorylation status of various signaling molecules.
A, Sorted rTregs were either directly added to CFSE-labeled CD4+T cells
or pretreated with indicated inhibitor and then stimulated with aCD3/
CD28/CD2-coated MicroBeads for 36 h before being added to CFSE-la-
beled CD4+T cells. As a control sample, purified CD4+T cells were mixed
with CFSE-labeled CD4+T cells (marked 0:1 in the panels). As soon as
rTregs (or purified CD4+T cells) had been mixed with CFSE-labeled CD4+
T cells (1:1 ratio), aCD3/CD28/CD2-coated MicroBeads were added, and
incubations continued for 84 h, followed by assessment of CFSE dilution
in viable cells. Data are representative of experiments conducted on T cells
from three different donors. B, Data from three experiments as in A were
combined. The suppressive capacity was calculated as Δproliferation =
% proliferation (responder cells alone, 0:1) 2 % proliferation (responder
cells in presence of Tregs, 1:1). Data are presented as mean 6 SEM (n =
3). **p , 0.01, ***p , 0.001. Day 0 rTreg, sorted rTregs directly mixed
with responder cells; 36 h srTreg, sorted rTregs pretreated with inhibitor or
not and then stimulated with aCD3/CD28/CD2-coated MicroBeads for
36 h before being added to responder cells.
Induction of rTreg suppressive capacity is Mek dependent.
not equally well by rTregs
Proliferation of Tconvs is strongly suppressed by actTregs but
Unstained Add Back Cells
Sorted CD4+Tconvs (CD45RA+CD252and CD45RA2CD252) were CFSE
stained and mixed at 1:1 ratio with non-CFSE–stained add back cells (sorted act-
Tregs, sorted rTregs, or sorted CD4+Tconvs). Thereafter, cells were stimulated with
aCD3/CD28-coated T cell expander beads (bead/cell ratio 1:10) for 90 h, followed
by assessment of CFSE dilution in cells. Numbers indicate the percentage of prolif-
erating CFSE-labeled CD4+Tconvs.
p = 0.001, add back of actTregs versus Tconvs; p = 0.099, add back of rTregs
versus Tconvs; p = 0.001, add back of actTregs versus rTregs; paired two-tailed t test.
5242T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
For all T cell subsets tested (both among CD4+and CD8+), our
experiments revealed that signaling was clearly augmented in
the presence of CD28 costimulation, in agreement with previous
observations. As expected, we also found that signaling was even
stronger with the additional presence of CD2 costimulation. Based
on studies of mice deficient in CD28, CD2, or both (16–18), it is
clear that some level of costimulation is necessary for proper
T cell activation and commitment; but, these studies also indicated
that the loss of CD28 can be compensated for by CD2, and vice
versa. However, we also unraveled clear signaling differences in
this regard. Indeed, CD28 was necessary and sufficient to trigger
activation of NF-kB, indicating that although there may be re-
dundancy in activation of PLCg1pathways, activation of NF-kB
may be more exclusively controlled by CD28. Furthermore, it is
interesting to note that augmented phosphorylation of many sig-
nal molecules is observed when both costimulatory pathways are
triggered simultaneously. One possible explanation for this may
be that the combined action of both costimulatory molecules is
necessary for fine-tuning of some T cell responses. The costim-
ulatory effects of either CD28 or CD2 are believed to be mediated
by signaling molecules that are also activated downstream of the
TCR. One such candidate is PLCg1(39, 40), which was previously
suggested to be the most TCR-proximal molecule activated by
CD28. However, another upstream candidate is Lck. It is note-
worthy that the SH3 domain of Lck can bind proline-rich regions
in the cytoplasmic parts of both CD28 and CD2 (41, 42). Because
only ∼50% of Lck molecules in a CD4+T cell are bound to CD4
and, hence, are directly involved in TCR signaling, the remaining
50% of Lck molecules are available for interactions with other
proteins (43). Mobilization and activation of some of these latter
Lck molecules may significantly boost T cell activation. This
notion is supported by our data showing that TCR-proximal sig-
naling events, such as z-chain phosphorylation and ZAP70 acti-
vation, both Lck-dependent processes, are much stronger when
CD28, CD2, or combined CD28/CD2 stimulation is added to TCR
stimulation. In addition to the stronger TCR-proximal signals seen
when CD2 costimulation was added to TCR/CD28 stimulation, we
noticed a robust increase in S6-Rp phosphorylation by CD2 alone,
especially in effector/memory subsets. This observation supports
a specific role for CD2-induced signals in S6-Rp–mediated reg-
ulation of translation, which appears independent of the activation
of Akt, a well-known upstream activator of S6-Rp.
Our analyses indicated clear differences in signaling patterns
between naive and effector/memory cells. This was seen for both
CD4+and CD8+Tconvs. Generally, for a given stimulus, naive cells
displayed stronger TCR-proximal responses than did effector/
memory cells. The opposite was the case when TCR-distal pro-
cesses were analyzed. The most obvious explanation for this ob-
servation is altered expression and/or regulation of certain signaling
molecules in differentiated cells compared with naive cells, thereby
shifting the activation threshold. Such differences may also provide
a molecular explanation for why effector/memory cells, compared
with naive cells, generally respond more rapidly and strongly to
rechallenge with a previously encountered Ag. Because effector/
memory responses are potent and potentially harmful to the host,
the activation of these cells is tightly controlled, both by cell-
extrinsic mechanisms (e.g., controlled by other cells, such as
Tregs) and mechanisms that are intrinsic to the effector/memory
T cells. In this respect, we recently observed that effector/memory
cells have a constitutively active protein kinase A-signaling node
(34), which could be important to prevent aberrant activation of
CD8+effector/memory T cells, possibly through protein kinase A-
mediated phosphorylation and activation of C-terminal Src kinase.
A similar, but weaker, trend was seen for CD4+effector/memory
T cells, indicating that additional mechanisms are also operative.
Because our starting material was buffy coats, the Treg pop-
ulations that we studied were mainly thymus-derived Tregs, also
known as natural Tregs. A recent publication (44) demonstrated that
these cells are remarkably stable and continue to express the Treg-
defining transcription factor FOXP3, despite being subjected to an
inflammatory environment. Still, differences in FOXP3 expression,
as well as differential expression of CD25 and CD45RA, have led
to the division of natural Tregs into two groups, rTregs (FOXP3+)
and suppressive actTregs (FOXP3++); the former is considered
a precursor of the latter (30). Our phospho-flow analyses of rTregs
and actTregs revealed clear signaling differences between these
subsets. The most striking feature was that activation-induced Akt
signaling was present in rTregs and diminished in actTregs. In
contrast, actTregs displayed high basal Erk activation, indicating an
active Mek. Previous reports suggested an inverse relationship be-
tween Akt signaling and the suppressive function of Tregs (45, 46).
Furthermore, it was recently shown that the Foxo proteins Foxo1
and Foxo3A, both of which are inactivated by Akt, drive FOXP3
expression (47–49). Combined with our data, this suggested that
a well-functioning Akt-signaling pathway in rTregs keeps FOXP3
expression in check and, hence, controls the function of rTregs until
proper activation and Mek-Erk–dependent maturation into actTregs.
Transcription of the FOXP3 gene is controlled by a plethora
of transcription factors acting on the promoter region (e.g., AP-1,
SP1, NF-kB, NFAT, Foxo1, Foxo3a, STAT5, and Runx) and con-
served noncoding sequence (CNS)1 (NFAT, Smad 2/3), CNS2
(Foxo1, Foxo3a, NF-kB, CREB, ATF, Runx, STAT5, and FOXP3
itself), and CNS3 (NF-kB) (50). The transition of rTregs into
highly suppressive actTregs involves upregulation of FOXP3, as
well as several other proteins. We observed that TCR/CD28/CD2-
mediated upregulation of FOXP3 in rTregs peaked after 36 h in
a Mek-dependent manner. Importantly, at this time, these “acti-
vated” rTregs had acquired a suppressive capability comparable to
that of actTregs. In contrast, sorted rTregs that were added back
without prior stimulation were only mildly suppressive. Hence,
our data indicated that potent FOXP3 upregulation may be nec-
essary for these cells to acquire strong suppressive capacity. In
a recent publication, Miyara et al. (30) demonstrated that rTregs
also exert suppression prior to their transition to actTregs. This
suppression was substantially weaker than that of actTregs, which
is in line with our observations. Although we used anti-CD3/
CD28/CD2– or anti-CD3/CD28–coated beads for stimulation of
responder cells in our suppression assays, Miyara et al. used plate-
bound anti-CD3 in the presence of irradiated autologous accessory
cells. Because the level of Treg-mediated suppression depends on
the strength and quality of the stimulus provided, these differences
in experimental conditions may explain the observed variation in
the suppressive capacity of rTregs.
Because actTregs in the phospho-flow analysis displayed ele-
vated basal Erk activation, we speculated that the differentiation
of rTregs into actTregs involves a shift in Mek-Erk–mediated
signaling necessary to keep FOXP3 levels high, perhaps through
AP-1. However, with the experimental evidence presented in
this article, it is difficult to define cause and action. Based on
the abovementioned inhibitory effects of Akt signaling on Treg
function, it was somewhat surprising that Akt inhibitors did not
augment TCR/CD28/CD2-mediated upregulation of FOXP3 in
rTregs. In fact, the presence of Akt inhibitors resulted in lower
TCR/CD28/CD2-mediated FOXP3 upregulation and subsequent
lower suppressive capacity. Combined with the data obtained with
Mek inhibitors, this suggested that the differentiation of rTregs
into actTregs is a complex process involving several pathways and
The Journal of Immunology5243
possibly distinct stages where the Akt and Mek/Erk pathways play
The functional interplay between Tconvs and Tregs is important
for balancing immune activation and immune control. Disturbance
of this balance can contribute to autoimmunity, inappropriate in-
flammation, and cancer. Hence, a more detailed view of the sig-
naling processes in the different functional T cell subsets might
increase our understanding of the complex molecular mechanisms
forming the basis for specific subset functions and reveal disease
adaptability and potential escape mechanisms that prevent an ef-
ficient immune response. Overall, the data from this report indi-
cated that inhibitors of the Mek/Erk-signaling pathway might be
used therapeutically to control the function of Tregs and peripheral
tolerance. Further studies are needed to clearly define the precise
roles of Mek and Erk in immune suppression by Tregs.
We thank Drs. Jonathan Irish and Garry Nolan for access to the Cytobank
flow cytometry analysis software at the Stanford server and Dr. Eirik A.
Torheim for technical advice on suppression assays.
The authors have no financial conflicts of interest. The University of Oslo
Technology Transfer Office, Inven2 AS, has filed a pending patent applica-
tion on methods to inhibit rTreg activation.
1. Mustelin, T., and K. Taske ´n. 2003. Positive and negative regulation of T-cell
activation through kinases and phosphatases. Biochem. J. 371: 15–27.
2. Samelson, L. E. 2002. Signal transduction mediated by the T cell antigen re-
ceptor: the role of adapter proteins. Annu. Rev. Immunol. 20: 371–394.
3. Smith-Garvin, J. E., G. A. Koretzky, and M. S. Jordan. 2009. T cell activation.
Annu. Rev. Immunol. 27: 591–619.
4. Straus, D. B., and A. Weiss. 1992. Genetic evidence for the involvement of the
lck tyrosine kinase in signal transduction through the T cell antigen receptor.
Cell 70: 585–593.
5. Veillette, A., J. B. Bolen, and M. A. Bookman. 1989. Alterations in tyrosine
protein phosphorylation induced by antibody-mediated cross-linking of the CD4
receptor of T lymphocytes. Mol. Cell. Biol. 9: 4441–4446.
6. Reth, M. 1989. Antigen receptor tail clue. Nature 338: 383–384.
7. Samelson, L. E., M. D. Patel, A. M. Weissman, J. B. Harford, and
R. D. Klausner. 1986. Antigen activation of murine T cells induces tyrosine
phosphorylation of a polypeptide associated with the T cell antigen receptor. Cell
8. Weiss, A., and D. R. Littman. 1994. Signal transduction by lymphocyte antigen
receptors. Cell 76: 263–274.
9. Chan, A. C., M. Iwashima, C. W. Turck, and A. Weiss. 1992. ZAP-70: a 70 kd
protein-tyrosine kinase that associates with the TCR zeta chain. Cell 71: 649–662.
10. Zhang, W., J. Sloan-Lancaster, J. Kitchen, R. P. Trible, and L. E. Samelson.
1998. LAT: the ZAP-70 tyrosine kinase substrate that links T cell receptor to
cellular activation. Cell 92: 83–92.
11. Kane, L. P., P. G. Andres, K. C. Howland, A. K. Abbas, and A. Weiss. 2001. Akt
provides the CD28 costimulatory signal for up-regulation of IL-2 and IFN-
gamma but not TH2 cytokines. Nat. Immunol. 2: 37–44.
12. Iezzi, G., K. Karjalainen, and A. Lanzavecchia. 1998. The duration of anti-
genic stimulation determines the fate of naive and effector T cells. Immunity 8:
13. Schulze-Luehrmann, J., and S. Ghosh. 2006. Antigen-receptor signaling to nu-
clear factor kappa B. Immunity 25: 701–715.
14. Tarakhovsky, A., M. Turner, S. Schaal, P. J. Mee, L. P. Duddy, K. Rajewsky, and
V. L. Tybulewicz. 1995. Defective antigen receptor-mediated proliferation of B
and T cells in the absence of Vav. Nature 374: 467–470.
15. Zhang, R., F. W. Alt, L. Davidson, S. H. Orkin, and W. Swat. 1995. Defective
signalling through the T- and B-cell antigen receptors in lymphoid cells lacking
the vav proto-oncogene. Nature 374: 470–473.
16. Green, J. M., V. Karpitskiy, S. L. Kimzey, and A. S. Shaw. 2000. Coordinate
regulation of T cell activation by CD2 and CD28. J. Immunol. 164: 3591–3595.
17. Sasada, T., and E. L. Reinherz. 2001. A critical role for CD2 in both thymic
selection events and mature T cell function. J. Immunol. 166: 2394–2403.
18. Shahinian, A., K. Pfeffer, K. P. Lee, T. M. Ku ¨ndig, 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.
19. Meinl, E., D. Lengenfelder, N. Blank, R. Pirzer, L. Barata, and C. Hivroz. 2000.
Differential requirement of ZAP-70 for CD2-mediated activation pathways of
mature human T cells. J. Immunol. 165: 3578–3583.
20. Kaizuka, Y., A. D. Douglass, S. Vardhana, M. L. Dustin, and R. D. Vale. 2009.
The coreceptor CD2 uses plasma membrane microdomains to transduce signals
in T cells. J. Cell Biol. 185: 521–534.
21. Melton, E., N. Sarner, M. Torkar, P. A. van der Merwe, J. Q. Russell, R. C. Budd,
C. Mamalaki, M. Tolaini, D. Kioussis, and R. Zamoyska. 1996. Transgene-
encoded human CD2 acts in a dominant negative fashion to modify thymocyte
selection signals in mice. Eur. J. Immunol. 26: 2952–2963.
22. Carmo, A. M., D. W. Mason, and A. D. Beyers. 1993. Physical association of the
cytoplasmic domain of CD2 with the tyrosine kinases p56lck and p59fyn. Eur. J.
Immunol. 23: 2196–2201.
23. Danielian, S., A. Alcover, L. Polissard, M. Stefanescu, O. Acuto, S. Fischer, and
R. Fagard. 1992. Both T cell receptor (TcR)-CD3 complex and CD2 increase the
tyrosine kinase activity of p56lck. CD2 can mediate TcR-CD3-independent and
CD45-dependent activation of p56lck. Eur. J. Immunol. 22: 2915–2921.
24. Killeen, N., S. G. Stuart, and D. R. Littman. 1992. Development and function of
T cells in mice with a disrupted CD2 gene. EMBO J. 11: 4329–4336.
25. Vignali, D. A., L. W. Collison, and C. J. Workman. 2008. How regulatory T cells
work. Nat. Rev. Immunol. 8: 523–532.
26. Fontenot, J. D., M. A. Gavin, and A. Y. Rudensky. 2003. Foxp3 programs the
development and function of CD4+CD25+ regulatory T cells. Nat. Immunol. 4:
27. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell de-
velopment by the transcription factor Foxp3. Science 299: 1057–1061.
28. Khattri, R., T. Cox, S. A. Yasayko, and F. Ramsdell. 2003. An essential role for
Scurfin in CD4+CD25+ T regulatory cells. Nat. Immunol. 4: 337–342.
29. Sakaguchi, S., M. Miyara, C. M. Costantino, and D. A. Hafler. 2010. FOXP3+
regulatory T cells in the human immune system. Nat. Rev. Immunol. 10:
30. Miyara, M., Y. Yoshioka, A. Kitoh, T. Shima, K. Wing, A. Niwa, C. Parizot, C.
Taflin, T. Heike, D. Valeyre, et al. 2009. Functional delineation and differenti-
ation dynamics of human CD4(+) T cells expressing the FoxP3 transcription
factor. Immunity 30: 899–911.
31. Krutzik, P. O., and G. P. Nolan. 2006. Fluorescent cell barcoding in flow
cytometry allows high-throughput drug screening and signaling profiling. Nat.
Methods 3: 361–368.
32. Krutzik, P. O., and G. P. Nolan. 2003. Intracellular phospho-protein staining
techniques for flow cytometry: monitoring single cell signaling events. Cytom-
etry A 55: 61–70.
33. Krutzik, P. O., J. M. Irish, G. P. Nolan, and O. D. Perez. 2004. Analysis of
protein phosphorylation and cellular signaling events by flow cytometry: tech-
niques and clinical applications. Clin. Immunol. 110: 206–221.
34. Oberprieler, N. G., S. Lemeer, M. E. Kalland, K. M. Torgersen, A. J. R. Heck,
and K. Tasken. 2010. High-resolution mapping of prostaglandin E2-dependent
signaling networks identifies a constitutively active PKA signaling node in CD8
+CD45RO+ T cells. Blood 116: 2253–2265.
35. Krutzik, P. O., J. M. Crane, M. R. Clutter, and G. P. Nolan. 2008. High-content
single-cell drug screening with phosphospecific flow cytometry. Nat. Chem. Biol.
36. Irish, J. M., J. H. Myklebust, A. A. Alizadeh, R. Houot, J. P. Sharman,
D. K. Czerwinski, G. P. Nolan, and R. Levy. 2010. B-cell signaling networks
reveal a negative prognostic human lymphoma cell subset that emerges during
tumor progression. Proc. Natl. Acad. Sci. USA 107: 12747–12754.
37. Hagner, P. R., A. Schneider, and R. B. Gartenhaus. 2010. Targeting the trans-
lational machinery as a novel treatment strategy for hematologic malignancies.
Blood 115: 2127–2135.
38. Valmori, D., A. Merlo, N. E. Souleimanian, C. S. Hesdorffer, and M. Ayyoub.
2005. A peripheral circulating compartment of natural naive CD4 Tregs. J. Clin.
Invest. 115: 1953–1962.
39. Michel, F., G. Attal-Bonnefoy, G. Mangino, S. Mise-Omata, and O. Acuto. 2001.
CD28 as a molecular amplifier extending TCR ligation and signaling capa-
bilities. Immunity 15: 935–945.
40. Kanner, S. B., N. K. Damle, J. Blake, A. Aruffo, and J. A. Ledbetter. 1992. CD2/
LFA-3 ligation induces phospholipase-C gamma 1 tyrosine phosphorylation and
regulates CD3 signaling. J. Immunol. 148: 2023–2029.
41. Holdorf, A. D., J. M. Green, S. D. Levin, M. F. Denny, D. B. Straus, V. Link,
P. S. Changelian, P. M. Allen, and A. S. Shaw. 1999. Proline residues in CD28
and the Src homology (SH)3 domain of Lck are required for T cell costimula-
tion. J. Exp. Med. 190: 375–384.
42. Bell, G. M., J. Fargnoli, J. B. Bolen, L. Kish, and J. B. Imboden. 1996. The SH3
domain of p56lck binds to proline-rich sequences in the cytoplasmic domain of
CD2. J. Exp. Med. 183: 169–178.
43. Veillette, A., M. A. Bookman, E. M. Horak, and J. B. Bolen. 1988. The CD4 and
CD8 T cell surface antigens are associated with the internal membrane tyrosine-
protein kinase p56lck. Cell 55: 301–308.
44. Rubtsov, Y. P., R. E. Niec, S. Josefowicz, L. Li, J. Darce, D. Mathis, C. Benoist,
and A. Y. Rudensky. 2010. Stability of the regulatory T cell lineage in vivo.
Science 329: 1667–1671.
45. Crellin, N. K., R. V. Garcia, and M. K. Levings. 2007. Altered activation of AKT
is required for the suppressive function of human CD4+CD25+ T regulatory
cells. Blood 109: 2014–2022.
46. Liu, G., S. Burns, G. Huang, K. Boyd, R. L. Proia, R. A. Flavell, and H. Chi.
2009. The receptor S1P1 overrides regulatory T cell-mediated immune sup-
pression through Akt-mTOR. Nat. Immunol. 10: 769–777.
47. Harada, Y., Y. Harada, C. Elly, G. Ying, J. H. Paik, R. A. DePinho, and Y. C. Liu.
2010. Transcription factors Foxo3a and Foxo1 couple the E3 ligase Cbl-b to the
induction of Foxp3 expression in induced regulatory T cells. J. Exp. Med. 207:
5244T CELL SIGNAL MAPS DIFFER BY STIMULUS AND SUBSET
48. Ouyang, W., O. Beckett, Q. Ma, J. H. Paik, R. A. DePinho, and M. O. Li. 2010.
Foxo proteins cooperatively control the differentiation of Foxp3+ regulatory
T cells. Nat. Immunol. 11: 618–627.
49. Kerdiles, Y. M., E. L. Stone, D. R. Beisner, M. A. McGargill, I. L. Ch’en,
C. Stockmann, C. D. Katayama, and S. M. Hedrick. 2010. Foxo transcription
factors control regulatory T cell development and function. [Published erratum
appears in 2011 Immunity 34: 135.] Immunity 33: 890–904.
50. Merkenschlager, M., and H. von Boehmer. 2010. PI3 kinase signalling blocks
Foxp3 expression by sequestering Foxo factors. J. Exp. Med. 207: 1347–
The Journal of Immunology5245