The Journal of Immunology
Functional Plasticity of Antigen-Specific Regulatory T Cells
in Context of Tumor
Caroline Addey,*,1Matthew White,*,1Lang Dou,*,1David Coe,* Julian Dyson,†
and Jian-Guo Chai*
Although polyclonal regulatory T cells (Tregs) that once expressed Foxp3 (ex-Tregs) derived from Foxp3+Tregs have been
described in homeostatic and autoimmune settings, little is known regarding the influence of the tumor environment on ex-
Treg development. After adoptive transfer of HY-specific green Tregs (peripheral or thymic) to Rag22/2B6 female mice bearing
syngeneic HY-expressing MB49 tumors, a significant fraction rapidly lost expression of Foxp3. On the second transfer to a Rag22/2
B6 male environment, these ex-Tregs expanded strongly, whereas Tregs that maintained expression of Foxp3 expression did not.
Both FACS and quantitative real-time-PCR analysis revealed that ex-Tregs upregulated genes characteristic of a Th1 effector-
memory phenotype including IFN-g and downregulated a panel of Treg-specific genes. Peripheral HY-specific green Tregs were
adoptively transferred to Rag22/2B6 male mice, to dissect the factors regulating ex-Treg differentiation. Development of ex-Tregs
was more efficient in the mesenteric lymph node (mLN) than peripheral lymph node environment, correlating with a much greater
level of IL-6 mRNA in mLN. In addition, the preferential development of ex-Tregs in mLN was significantly impaired by
cotransfer of HY-specific naive CD4 T cells. Collectively, our study not only demonstrates the plasticity of Ag-specific Tregs in
the context of the tumor environment, but also defines key molecular and cellular events that modulate ex-Treg differentiation.
The Journal of Immunology, 2011, 186: 4557–4564.
are composed of two distinct subsets called natural (nTregs) and
induced Tregs (iTregs). Thymic Tregs purified from thymus are
nTregs, whereas peripheral Tregs purified from lymph node (LN)
and spleen contain both nTregs and iTregs (4). Foxp3 expression
is regulated by a positive autocrine feedback loop (5, 6) and by
epigenetic modification of the Foxp3 locus (7); thus, Tregs have
many features of a stable CD4 subset. Indeed, most Tregs retain
a high Foxp3 expression after adoptive transfer to lymphoreplete
However, the instability of Foxp3 in peripheral Tregs has been
recognized recently. We have previously reported that polyclonal
green Tregs (CD4+CD25+GFP+cells FACS-sorted from Foxp3/
he Foxp3 transcription factor is crucial in directing the
development and maintenance of the suppressive function
and phenotype of regulatory T cells (Tregs) (1–3), which
GFP reporter mice) showed a partial loss of GFP after in vitro
activation (8). Others also have shown that in vitro TCR stimu-
lation with IL-6 led to downregulation of Foxp3 in polyclonal
peripheral green Tregs (9, 10).
The functional plasticity of peripheral Tregs is also evidenced
by in vivo studies. We have recently demonstrated that a signifi-
cant fraction of tumor Ag-specific green Tregs lost Foxp3 expres-
sion after adoptive transfer to tumor-bearing wild-type (WT) mice
(11). Other groups have also shown that a large proportion of
polyclonal green Tregs became GFP2and differentiated to Th1-
like effector cells or follicular Th cells in Rag2/2(12, 13) and
CD32/2recipients (14), respectively. Furthermore, mice with at-
tenuated expression of endogenous Foxp3 develop lymphoproli-
ferative syndromes that are correlated with impaired suppressive
function of Tregs and their conversion to effector cells (15).
Moreover, using an elegant dual-reporter mouse model to distin-
guish T cells that constitutively express or have ceased to express
Foxp3 in vivo, Bluestone and colleagues (16) found that a sub-
stantial fraction of peripheral Tregs lost Foxp3 and obtained an
activated-memory phenotype even under steady-state conditions
in unmanipulated healthy mice. Importantly, ex-Tregs that are
present at a significantly higher frequency in autoimmune diabetes
are autoaggressive proinflammatory effector cells able to transfer
Although the development of ex-Tregs has been studied in
homeostatic (12–14) and autoimmune settings (15, 16), less in-
formation is available in terms of their differentiation in the con-
text of tumor. In addition, adoptive transfer experiments generally
use polyclonal Tregs, which has limited value in studying the role
of Ag. More importantly, little is known regarding the cellular and
molecular mechanisms of ex-Treg generation and regulation (17).
In addition, whether the development of ex-Tregs is restricted to
the CD252vesubset of Tregs is controversial (13, 16). The plas-
ticity of Foxp3 expression and the generation of ex-Tregs by
thymic nTregs has also not been directly assessed.
*Cancer Immunotherapy Group, Section of Immunobiology, Department of Medi-
cine, Imperial College London, Hammersmith Hospital, London W12 0NN, United
Kingdom; and†T Cell Development Group, Section of Immunobiology, Department
of Medicine, Imperial College London, Hammersmith Hospital, London W12 0NN,
1C.A., M.W., and L.D. contributed equally to this work.
Received for publication November 17, 2010. Accepted for publication January 29,
This work was supported by the Cancer Research United Kingdom Senior Cancer
Research Fellowship (to J.-G.C.).
Address correspondence and reprint requests to Dr. Jian-Guo Chai, Section of Im-
munobiology, Department of Medicine, Imperial College London, Hammersmith
Hospital, Du Cane Road, London W12 0NN, United Kingdom. E-mail address:
The online version of this article contains supplemental material.
Abbreviations used in this article: Ct, cycle threshold; dLN, draining lymph node; ex-
Treg, regulatory T cells that once expressed Foxp3; iNL, inguinal LN; iTreg, induced
regulatory T cell; LN, lymph node; MFI, mean fluorescence intensity; mLN, mes-
enteric lymph node; N, naive; nTreg, natural regulatory T cell; pLN, peripheral
lymph node; qPCR, quantitative PCR; qRT-PCR, quantitative real-time PCR; Tregs,
Foxp3+CD4+regulatory T cells; WT, wild-type.
To generate and characterize Ag-specific ex-Tregs, we modified
a recently described in vivo model (11) by adoptively transferring
HY-specific green Tregs (either thymic or peripheral) to MB49
tumor-bearing Rag2/2B6 female recipients. Because of the pre-
ferential development of ex-Tregs in the lymphopenic environ-
ment (12–14), the use of Rag2/2B6 female mice allows sufficient
ex-Tregs to be isolated and characterized.
For exploring the cellular and molecular events that impact on
ex-Treg development, a second invivo model was applied in which
green HY-specific peripheral Tregs were adoptively transferred to
Rag2/2B6 male mice in which HY is expressed ubiquitously.
Materials and Methods
Rag22/2B6, Rag22/2Thy1.2+or Thy1.1+Marilyn (18), Thy1.2+
Foxp3GFPB6 (19), and (Rag22/2Marilyn 3 Foxp3GFP) F1 mice were
described previously (8, 11, 20). All animal experiments were performed
in accordance with Home Office Animals (Scientific Procedures) Act of
1986. In the absence of HY Ags, the development of Tregs in Rag+/2
Marilyn female mice (i.e., [Rag2/2Marilyn 3 Foxp3-GFP] F1) is likely to
involve endogenous Va TCR rearrangements allowing their positive se-
lection on self MHC–peptide complexes (8, 11). In Rag2/2Marilyn female
mice without endogenous Va TCR rearrangement, the frequency of Tregs
is severely reduced. These results are shown in Supplemental Fig. 4. We
have previously demonstrated that CD4+CD25+cells from Rag+/2Marilyn
(8) or CD4+CD25+GFP+cells from (Marilyn 3 Foxp3/GFP) F1 female
mice [Rag+/2] (11) are functionally HY-specific Tregs both in vitro and
Tumor cell line
In vitro culture of HY+MB49 cells (21) was described elsewhere (11, 20).
Purification of HY-specific naive CD4, green peripheral Treg,
and green thymic nTregs
This purification was conducted as described previously (8, 11, 20). Naive
CD4 T cells (CD4+Vb6+CD62LhighCD44low) were FACS sorted from
spleen and LN cells of Rag22/2Marilyn female mice. Green peripheral
Tregs (GFP+CD4+Vb6+CD25+) were FACS sorted from spleen and pooled
LN cells of (Rag22/2Marilyn 3 Foxp3GFP) F1 female mice. Green nTregs
(GFP+CD4+CD82CD25+) were FACS sorted from CD8-depleted thymo-
cytes of (Rag22/2Marilyn 3 Foxp3GFP) F1 female mice.
Adoptive T cell transfer and tumor inoculation
populations into Rag22/2B6 female mice, which were either inoculated or
not with MB49 cells (1 3 105/mouse) by s.c. injection into the right flank
on the same day. In some experiments, Rag22/2B6 male mice were used
Analysis of representation of donor cell populations
For MB49-bearing mice and their controls, analysis was conducted by
staining of inguinal LN (iLN), MB49-draining LN (dLN), spleen, or tumor
mass (MB49) with anti-Vb6PE(or anti-CD4PE, anti-FR4PE, and anti-
CCR7PE), anti-Thy1.1PerCP(or CD4PercP), and anti-CD4allophycocyanin(or
anti-Thy1.2allophycocyanin). The donor cells were identified before analyzing
GFP, FR4, or CCR7 expression. For Rag22/2B6 male mice, analysis was
performed using peripheral LN (pLN), mesenteric LN (mLN), and spleen.
Intracellular cytokine staining
Intracellular cytokine staining was conducted as described previously (8,
Reisolation of ex-Tregs and Tregs
Ex-Tregs (GFP2CD4+Vb6+) and unconverted Tregs (GFP+CD4+Vb6+)
were FACS sorted from spleen and pooled dLN cells of MB49-bearing
Rag22/2B6 female recipients of HY-specific green Tregs on day 12, and
were subjected to either secondary transfer to Rag22/2B6 male mice or
quantitative real-time PCR (qRT-PCR) analysis. After FACS, a fraction of
cells was subjected to intracellular anti-Foxp3 staining to confirm that ex-
Tregs indeed do lose Foxp3.
RNA and cDNA preparation
RNA was extracted from sorted cells using an RNAqueous-4PCR kit
(Applied Biosystems, Warrington, U.K.), including DNase I digestion step.
cDNAwas made from 300 ng RNA using a High-Capacity cDNA Reverse
specific ex-Tregs in vivo. A, Purity of
HY-specific peripheral green Tregs. B
and C, Analysis of ex-Tregs in vivo.
HY-specific green Tregs (2 3 104/
mouse) were adoptively transferred
to MB49-free (B) or MB49-bearing
Rag22/2B6 female mice (C) (4 mice/
group). FACS analysis was conducted
on day 12. Dot plots of one repre-
sentative mouse of each group are
shown. One representative experiment
of three is shown. D–G, The percent-
age (D, E) and absolute number (F, G)
of donor Tregs (D, F) and donor ex-
Tregs (E, G) in iLN of control (black
bars) and dLN of MB49-bearing mice
(white bars) are shown. H, The purity
of HY-specific thymic green nTregs.
I, Analysis of ex-nTregs in vivo. HY-
specific green nTregs (1 3 104/
mouse) were transferred to MB49-
bearing Rag22/2B6 female mice
(n = 4). Analysis was conducted on
day 12. FACS dot plots of one rep-
resentative mouse of four are shown.
One representative experiment of three
Development of HY-
4558 CHARACTERIZATION OF TUMOR-SPECIFIC EX-Tregs
Transcription kit (Applied Biosystems), which facilitates quantitative
conversion of mRNA to cDNA.
Quantitative real-time PCR and Global Pattern Recognition
Mouse T Regulatory Phenotyping 383 StellARray quantitative PCR (qPCR)
array plates (part #00188458; Lonza Sales AG, Basel, Switzerland) were
used, with each plate containing 4 3 96 gene primer pairs. cDNA, Power
Sybr green master mix (Applied Biosystems), and water were mixed, fol-
lowing the manufacturer’s directions, and 10 ml mix was added per well
(1.5 ng cDNA). Plates were run on the 7900HT Real-Time PCR system (Ap-
plied Biosystems), under standard cycling conditions (50˚C for 2 min, 95˚C
for 10 min followed by 40 cycles of 95˚C for 15 sec and 60˚C for 1 min).
The resulting cycle threshold (Ct) data were analyzed using a modifi-
cation of Global Pattern Recognition algorithm, GPR 2.0 (http://array.
lonza.com/gpr/), comparing three replicates of GFP+T Reg (control) with
GFP2T Reg (test) cDNA. When there is no difference between the Ct
values of both groups, the genes are considered as “normalizers.” Each
gene is compared with every normalizer in succession and the DCt is
calculated (DCT Gene= CT Gene2 CT Normalizer). The DCt for each group are
then compared by a two-tailed unpaired Student t test. Results are
expressed as fold change, and those changes with a p value ,0.05 are
Data are presented as mean 6 SEM. The comparison between groups was
performed using a two-tailed, nonparametric Mann–Whitney U test or
Student t test (p , 0.05 was considered as significant).
Generation of HY-specific ex-Tregs in vivo
To develop an in vivo model by which ex-Tregs can be generated,
we transferred highly purified, HY-specific, peripheral green Tregs
(Fig. 1A) into MB49-free (Fig. 1B) or MB49-bearing Rag22/2B6
female mice (Fig. 1C). A significant accumulation of progeny of
donor Tregs was found in tumor dLN, spleen, and tumor tissue on
day 12 (Fig. 1C). However, the representation of donor Tregs in
iLN and spleen in control mice was rather poor because of lack of
Ag (Fig. 1B). Difference in the frequency (Fig. 1D) and absolute
number (Fig. 1F) of donor Tregs between iLN and dLN was
Importantly, a fraction of the donor Tregs in dLN, spleen, and
MB49 of tumor-bearing mice also become GFP2(Fig. 1C), where-
as the donor Tregs in the iLN and spleen of control mice remained
100% GFP+(Fig. 1B). Tregs that have lost Foxp3 are henceforth
referred to as ex-Tregs. Ex-Tregs in dLN have both statistically
significant increased frequency (Fig. 1E) and absolute number
(Fig. 1G), in comparison with those in iLN.
Peripheral Tregs isolated from the LN and spleen are composed
of both nTregs and iTregs, whereas those from thymus are largely,
if not exclusively, nTregs (4). FACS-sorted, HY-specific, thymic
green Tregs (Fig. 1H) were transferred to MB49-bearing Rag22/2
B6 female mice. Ex-Tregs were clearly present in dLN, spleen,
and MB49 on day 12 (Fig. 1I); thus, nTregs can be converted to
ex-Tregs after in vivo activation recapitulating the behavior of
Development of ex-Tregs is not due to overexpansion by GFP2
To rule out the possibility that the generation of ex-Tregs is caused
by contaminating GFP2conventional CD4 T cells, we designed
a “contamination” experiment in which 1000 Thy1.1+HY-specific
naive CD4 T cells without or with 19,000 of Thy1.2+HY-specific
peripheral green Tregs were transferred to two groups of MB49-
bearing Rag22/2B6 female mice.
In the absence of Tregs, a significant accumulation of progeny of
donor naiveCD4 T cells was found in dLN, spleen, and MB49 after
the presence of Tregs, representation of donor naive CD4 T cells
was dramatically impaired with a 93–99% reduction in their ac-
cumulation in dLN, spleen, and MB49 (Fig. 2B). The statistical
analysis of both frequency (Fig. 2C) and absolute number (Fig. 2D)
of donor naive cells in dLN between these two groups also revealed
significant differences. Because the naive CD4 T cells (5%) com-
pletely failed to “grow out” when cotransferred with excess Tregs
(95%), the contribution of Foxp32contaminants must be negligi-
ble, and the emergence of ex-Tregs represents differentiation from
donor Tregs. Using similar approaches, two groups have reached
the same conclusions, although polyclonal T cell populations were
used as donor cells (12, 13).
Ex-Tregs expand more potently than Tregs in Rag22/2B6 male
To compare the function of ex-Tregs and Tregs, we performed
a secondary adoptive transfer. Tregs and ex-Tregs, generated as
described in Fig. 1, were FACS sorted with .99% purity (Fig. 3A,
3B) and separately transferred to two groups of Rag22/2B6 male
mice. Frequency of donor cells in recipients of ex-Tregs (Fig. 3D)
was much greater than that in recipients of Tregs on day 28 (Fig.
3C). These differences were statistically significant (Fig. 3E–G).
In addition, the progeny of donor ex-Tregs showed a higher fre-
quency in mLN but a lower frequency in pLN (Fig. 3D). In-
terestingly, a significant fraction of donor ex-Tregs in spleen but
GFP2contaminants. A and B, Analysis of donor naive CD4 T cells in vivo.
Thy1.1+HY-specific naive CD4 T cells alone (1 3 103/mouse) or mixed
with Thy1.2+HY-specific green Tregs (19 3 103/mouse) were transferred
to two groups of MB49-bearing Rag22/2B6 female mice (4 mice/group).
Analysis was conducted on day 12. FACS dot plots of one representative
mouse of four are shown. One representative experiment of two is shown.
The percentage (C) and absolute number (D) of donor naive CD4 T cells in
dLN of the above two groups were compared. One representative exper-
iment of two is shown. N, naive alone; N+Treg, cotransfer.
Development of ex-Tregs is not due to overexpansion of
The Journal of Immunology4559
not in mLN or pLN re-expressed GFP (Fig. 3H). These differences
were also statistically significant (Fig. 3I).
Ex-Tregs downregulated Treg-specific genes but upregulated
genes characteristic of a Th1 effector-memory phenotype
Tregs and ex-Tregs, generated as described in Fig. 1, were FACS
sorted before comparing gene expression profiles. As expected,
Foxp3 expression was reduced by 10-fold in ex-Tregs, which also
showed a 116-fold decrease of Gpr83 (G protein-coupled receptor
83), a Treg-specific surface marker (22, 23). Other Treg-associated
genes including Ctla4 (CTLA4) (24), Folr4 (FR4) (25), Nt5e
(CD73) (26), and Ccr7 (CCR7) (27) were also downregulated
(Fig. 4A). In contrast, ex-Tregs displayed a significant increase of
Ifng (IFN-g), Cd80 (CD80), Gzmb (granzyme B), and Gzma
(granzyme A) (Fig. 4B). To a lesser extent, Th1 cytokine genes
including Csf2 (GM-CSF), Tnf (TNF-a), and Il2 (IL-2) were also
moderately but significantly upregulated in ex-Tregs. Interestingly,
several Treg-related genes including Runx1 (28, 29) and Gal1
(galectin 1) (30) were maintained in ex-Tregs, indicating that re-
differentiation was incomplete.
Validation of qRT-PCR data by FACS analysis
To confirm that ex-Tregs express more IFN-g than Tregs with
sustained Foxp3 at the protein level, we used intracellular staining.
MB49-bearing Rag22/2B6 female recipients of HY-specific
green Tregs were prepared as described in Fig. 1. On day 12,
dLN cells were subjected to a standard intracellular IFN-g pro-
cedure after a brief in vitro restimulation (Fig. 5A). The frequency
of IFN-g–producing cells in ex-Tregs was significantly higher
than that in Tregs. The mean fluorescence intensity (MFI) of IFN-
g in ex-Tregs was also higher than that in Tregs; thus, overall ex-
Tregs produced more IFN-g than Tregs. Neither IL-4 nor IL-17
was detectable in the same experiment (Supplemental Fig. 1),
indicating that Tregs differentiated into Th1 but not Th2 or Th17
cells. We also confirmed that expression of both FR4 and CCR7
was significantly reduced on ex-Tregs (Fig. 5B).
MB49 tumor mass is an inflamed tissue that favors the
development of ex-Tregs
For the induction of ex-Tregs in vivo, lymphopenic mice are more
efficient than lymphoreplete mice (12–14). Nevertheless, we
spleen (SP) and pooled dLN of MB49-bearing Rag22/2B6 female recipients of HY-specific Tregs at day 12. C and D, Ex-Tregs, but not Tregs, are responsive to
endogenousHYAgs.Tregs and ex-Tregs (23103/mouse) were transferredintotwogroupsof Rag22/2B6male mice(3mice/group).Analysiswasconductedat
shown.Onerepresentative experiment of two isshown.I, Comparisonofthe frequency of donor ex-Tregs,whichregainedGFP expression inpLN, mLN, and SP.
Ex-Tregs are functional effector cells invivo. A and B, Purity of input donor cell populations. Tregs (A) and ex-Tregs (B) were FACS sorted from
4560 CHARACTERIZATION OF TUMOR-SPECIFIC EX-Tregs
previously reported that ex-Tregs were also present in MB49-
bearing WT B6 recipients of HY-specific peripheral green Tregs
(11). Interestingly, the frequency of ex-Tregs in tumor tissues is
higher than that in dLN isolated from the same mice (11), in-
dicating that the tumor provides an environment that favors ex-
Treg development. When gene expression profiles of cytokines/
chemokines/receptors were compared between MB49 tissue and
MB49 cell line, we found that 22 genes were upregulated in tumor
tissue with only one gene being downregulated (IL-13, 17-fold
decrease) (Fig. 6). IL-1b, a well-known proinflammatory cytokine,
showed a 428-fold increase. The highest fold increase in expres-
sion was seen for Cxcl9, which showed a .2600-fold increase,
followed by Ccl3, Ccl4, Ccl12, and Ccl5. Increased expression of
a panel of chemokine receptors including CXCR3, the receptor for
CXCL9, was also observed in MB49 tissue. Taken together, these
qRT-PCR data suggested that MB49 tissue provides a highly in-
HY-specific green Tregs preferentially differentiate into
ex-Tregs in mesenteric but not pLN of Rag22/2B6 male recipients
We also studied the differentiation of ex-Tregs in the second model
in which HY-specific green peripheral Tregs were adoptively
transferred into Rag22/2B6 male mice. Strikingly, the complete
conversion of Tregs into ex-Tregs was seen in mLN but not in
pLN or spleen on day 14 (Fig. 7A, 7B). When mLN cells were
compared with pLN for cytokine/chemokine/receptor gene ex-
pression, seven genes showed significant upregulation in mLN
(Fig. 7C), of which IL-6 mRNA displayed a .300-fold increase.
A moderate increase of IL-1R antagonist, CXCL10, IL-10,
CXCR4, IL-21, and IL-2 was also observed.
Regulation of ex-Treg differentiation by naive CD4 T cells
To explore whether HY-specific naive CD4 T cells could influence
HY-specific ex-Treg development in Rag22/2B6 male recipients,
we performed a cotransfer experiment. As shown in Fig. 7D and
7E, the copresence of naive CD4 T cells (1:1 ratio) led to
a remarked reduction (up to 80%) in the frequency of ex-Tregs in
mLN. A modest but significant decrease (40%) of ex-Treg fre-
quency was also seen in pLN but not in spleen. Thus, feedback
ex-Tregs. A and B, Tregs and ex-Tregs, sorted from pooled dLN and spleen
cells of MB49-bearing Rag22/2B6 female recipients of HY-specific green
Tregs at day 12, were subjected to RNA extraction and cDNA conversion
before running qPCR array assay using mouse Treg Phenotyping 383
StellARray plates. Eight genes were downregulated (A) and 15 genes
upregulated (B) in ex-Tregs compared with Tregs.
Comparison of gene expression profiles between Tregs and
Tregs and ex-Tregs. A, HY-specific green Tregs were transferred to MB49-
bearing Rag22/2B6 female mice (n = 4). Twelve days later, dLN cells
were briefly stimulated with PMA and ionomycin in the presence of bre-
feldin A before surface staining for CD4 and Thy1.2. After fixation and
permeabilization, the cells were stained for Foxp3 and IFN-g. Donor cells
were further divided into Tregs (R4) and ex-Tregs (R3). Both the fre-
quency and MFI of IFN-g+cells in Tregs (black bars) and ex-Tregs (white
bars) were measured. FACS dot plots of one representative mouse of four
are shown. One representative experiment of two is shown. B, HY-specific
green Tregs (Thy1.2) were transferred to MB49-bearing Rag22/2B6 fe-
male mice (n = 3). Twelve days later, dLN cells were stained with anti-
FR4PE(or anti-CCR7PE), anti-CD4PercP, and Thy1.2allophycocyanin. Donor
Tregs identified by CD4+Thy1.2+were distinguished as Treg (R5) or ex-
Tregs (R6) based on GFP expression. Expression of FR4 and CCR7 by
Tregs (open area) and ex-Tregs (closed area) are shown as histograms.
FACS profiles of one representative mouse of three are shown. One rep-
resentative experiment of three is shown. MFI of FR4 and CCR7 by Tregs
(black bars) and ex-Tregs (white bars) were measured.
Comparison of IFN-g, FR4, and CCR7 expression between
The Journal of Immunology4561
from naive CD4 T cells acts to maintain the expression of Foxp3
by Tregs. In the tumor model, the development of ex-Tregs in
MB49 tissues was also significantly impaired when naive CD4
T cells coexisted (Supplemental Fig. 2).
In this study, we have exclusively used Rag2/2B6 mice as re-
cipients of HY-specific green Tregs from B6.Foxp3gfp.Marilyn
TCR-transgenic mice. This is in contrast with the study by Rubtsov
et al. (31), in which lymphoreplete NOD mice were used as re-
cipients of pancreatic-islet-Ag–specific GFP+Tregs from NOD.
Foxp3gfp.BDC2.5 TCR-transgenic mice. The different conclusions
in terms of Foxp3 stability are likely to be due to the use of these
distinct types of recipient mice. In addition, the level of inflam-
mation present in nonmalignant tissue (prediabetic pancreas of 12-
wk-old NOD mice) and malignant tissue might be dramatically
We have examined the expression of intracellular Foxp3 by ex-
Tregs (FACS sorted from MB49-bearing mice based on the loss of
GFP expression), before the secondary transfer (as shown in Fig. 3)
and qRT-PCR analysis (as shown in Fig. 4). These GFP2ex-Tregs
had lost Foxp3 expression since staining with anti-Foxp3 mAb
was also negative, whereas positive Foxp3 staining was seen in the
sorted GFP+population. These results are shown in Supplemental
Fig. 3. Therefore, we can confirm that Tregs that lost GFP are
indeed Foxp32, which appears to be consistent with the RT-PCR
analysis showing that there is a significant downregulation (10-
fold) of Foxp3 mRNA expression in ex-Tregs compared with that
in unconverted green Tregs (Fig. 4A).
In the first part of this study, we characterized HY-specific ex-
Tregs (Figs. 3–5, Supplemental Fig. 1), which have developed
in an in vivo tumor model (Fig. 1), and excluded the possibility
that they represent the outgrowth of contaminating conventional
CD4 cells (Fig. 2). We also identified several unique features of
HY-specific ex-Tregs regarding their origin, differentiation, and
First, ex-Tregs differentiate from CD25+peripheral Tregs (Fig.
1C). This is consistent with one report showing that a fraction of
islet autoantigen-specific CD4+GITRhiGFP+cells (of which most
are CD25+) lose Foxp3 after transfer to TCRa2/2NOD mice (16),
but appears to conflict with another study suggesting that the
plasticity of Tregs is exclusive to the CD252-Treg subset (13).
Failure to detect ex-Tregs in Rag2/2B6 recipients of CD25+
Tregs by Komatsu et al. (13) could be because: 1) the test was
carried out only at an early time point (day 5); and 2) presumably
only pLN (whose environment is less inflammatory than that of
mLN) was analyzed. Indeed, when analysis was conducted at a
later time point (day 28), 90% of donor CD4+Foxp3+Tregs (of
which most are CD25+) became Foxp32recovered from intra-
epithelial lymphocytes (12) and Peyer’s patches (14). It appears
that an inflammatory milieu can promote ex-Treg commitment
(12, 14, 16). Given that IL-2 is critical to maintain stable ex-
pression of Foxp3 in Tregs (32), it has been suggested that the
CD252Treg subset might be more susceptible to Foxp3 loss
(13). But it does not necessarily indicate that the CD252Treg
subset is the sole population from which ex-Tregs are generated.
This is because on homeostatic expansion in lymphopenic mice,
the CD25+Tregs lose CD25 (33), and vice versa, CD252Tregs
also regain CD25 expression (34).
Second, nTregs (i.e., thymic green Tregs) are shown to be able to
differentiate to ex-Tregs (Fig. 1H, 1I). Previous studies exclusively
used peripheral Tregs (10, 12–14, 16), containing both nTregs and
iTregs, which differentially express Helios (4). As a transcription
factor, Helios has a limited value in separating the two subsets
present in peripheral Tregs. The use of thymic green Tregs as an
alternative strategy provides direct evidence that in vivo, nTregs
can be the precursors of ex-Tregs.
Third, ex-Tregs have characteristics of Th1 rather than Th17 or
Th2 cells (Fig. 5A, Supplemental Fig. 1). These data are compatible
with a study by Zhou et al. (16) showing that polyclonal ex-Tregs in
LN produced IFN-g but not IL17, whereas those in Peyer’s patches
made both. Therefore, the cytokine profiles of ex-Tregs also depend
on the location and environment where they are generated.
Finally, ex-Treg population is heterogeneous. qPCR analysis
revealed that expression of some Treg-related genes, including
Runx1 (28, 29) and galectin 1 (30), is maintained in ex-Tregs (Fig.
4). On their secondary transfer to Rag22/2B6 male mice, some
ex-Tregs in spleen but not LN became Foxp3+(Fig. 3H), sug-
gesting that re-expression of Foxp3 also depends on the tissue
environment. Foxp3 re-expression by polyclonal ex-Tregs has
been reported (13), although it is unclear whether ex-Tregs in
spleen are more competent than those in LN at regaining Foxp3.
Treg conversion in lymphocyte-sufficient hosts appears to be
much less efficient than that in lymphocyte-deficient hosts, ac-
cording to our own study (11) and others (12, 13, 16), which raises
concerns over the biological significance of the conversion of
Tregs in vivo. Nevertheless, Tregs can be converted to ex-Treg
under nonlymphopenic conditions. Interestingly, this conversion
takes place in inflamed tissue-expressing Ags, including the MB49
tumor mass (11) and the pancreas of prediabetic NOD mice (16).
From these studies, it has been proposed that the conversion of
Tregs under physiological conditions would contribute to the de-
velopment of autoimmune diseases and cancer surveillance; in
addition, the generation of effector-like ex-Tregs at sites of in-
fected tissue might help initiate clearance viral infection (17).
Exploration of the biological relevance of Treg conversion in WT
mice needs robust model systems and novel approaches. Com-
monly used strategies are to perform the secondary transfer of ex-
Tregs to lymphopenic hosts, as shown by us in this article and by
others (13, 14, 16). To directly examine ex-Treg function in re-
cipient mice, we will be using a Treg depletion system (green Treg
coexpressing Foxp3/GFP/DTR fusion protein).
isolated MB49 tissue and MB49 cell line. MB49 tumor tissue isolated from
WT B6 female mice that were inoculated with MB 49 cells 14 d pre-
viously, as well as in vitro long-term cultured MB49 cells, were subjected
to RNA extraction and cDNA conversion before running qPCR array assay
using mouse cytokine/chemokine/receptor 383 StellARray plates. Twenty-
two genes were upregulated in tumor tissue with only one gene being
downregulated (IL-13, 17-fold decrease). These experiments were repeated
twice and similar results were obtained.
Comparison of gene expression profiles between freshly
4562CHARACTERIZATION OF TUMOR-SPECIFIC EX-Tregs
The second part of this study explored the cellular events con-
trolling the development of ex-Tregs. There are two key obser-
vations. The first is that the microenvironment of mLN but not
pLN favors the generation of ex-Tregs (Fig. 7A, 7B). A subsequent
comparison of gene expression profiles revealed that IL6 mRNA
shows a remarkable increase (.300-fold) in mLN compared with
pLN cells (Fig. 7C). IL-6 alone can downregulate Foxp3 expres-
sion by TCR-stimulated Tregs, which can be further promoted by
the copresence of IL-1 and IL-21 (9, 10, 14). Moreover, IL-6
induces remethylation of the Foxp3 nonintronic upstream CpG-
rich island and closes the chromatin structure of the Foxp3 locus
(35). Although the link between increased IL-6 expression and
efficient ex-Treg differentiation in mLN needs to be confirmed by
in vivo studies, higher IL-6 expression by mLN provides a genetic
basis for greater development of ex-Tregs in this particular tissue,
which significantly extends previous studies showing that in-
flammatory environments such as gut-associated tissues (12, 14)
and islet of NOD mice (16, 17) can facilitate the development of
It is of interest to test whether ex-Tregs that resulted from invivo
conversion would have pathological consequences such as the
development of colitis. However, there is a concern that the ac-
tivities of ex-Tregs would be limited by the coexistence of non-
converted Tregs, especially in pLN where the Treg conversion was
incomplete. Another concern is that the use of HY-specific Tregs
is the main focus in the current investigation, whereas the de-
velopment of colitis in B6 mice appears to require a polyclonal
CD4 T cell response. To explore whether ex-Tregs can induce
a pathological response in vivo, unconverted Tregs would need to
be removed as described earlier. The effector potential of ex-Tregs
of graft-versus-host disease-like disease after transfer to irradiated
male recipients, which we have seen after adoptive transfer of
conventional Marilyn CD4 T cells (J.-G. Chai, unpublished
observations). It is also of interest to use Rag2/2IL6 KO mice to
examine whether these hosts fail to convert the Tregs.
partially prevent ex-Treg development in vivo (Fig. 7D, Supple-
mental Fig. 2). Similar results have been reported by others using
polyclonal Tregs (12, 36). Although the precise molecular mech-
anisms for regulating Foxp3 expression in Tregs by cotransferred
IL-2 and TGF-b have been suggested. Duarte et al. showed that
lymphopenia-induced loss of Foxp3 by polyclonal Tregs was ac-
celerated by anti–IL-2 treatment but prevented by IL-2 adminis-
tration (12). Similarly, IL-2 treatment can restore CD25 expression
and Treg function in inflamed islet of NOD mice (37). TGF-b pro-
duced by Tregs is required for maintaining Foxp3 expression and
protection from apoptosis (38).
What are the implications of this study? First, some features of
Tregs revealed by previous studies might be attributable to alter-
native interpretations because of the instability of Foxp3 expres-
sion. For example, Tregs act as effector cells in lymphopenic hosts
(39); Tregs secrete granzyme B and/or perforin to kill DC, CD8,
favors ex-Treg development. A, HY-specific green Tregs (Thy1.1) (1 3 104/
mouse) were transferred to Rag22/2B6 male mice (n = 4). Analysis
was conducted on day 14 by staining of pLN, mLN, or spleen cells with
anti-Vb6PE, anti-Thy1.1PerCP, and anti-CD4allophycocyanin. Donor Tregs
identified as Vb6+Thy1.1+were analyzed for Foxp3 expression. FACS
profiles of one representative mouse of four are shown. One representative
experiment of three is shown. B, Comparison of the frequency of ex-Tregs
in various tissues. C, Comparison of gene expression profiles between pLN
and mLN. pLN and mLN cells from a Rag22/2B6 male mouse were
subjected to RNA extraction and cDNA conversion before running qPCR
array assay using mouse cytokine/chemokine/receptor 383 StellARray
plates. One representative experiment of two is shown. D, Inhibition of ex-
Microenvironment of mLN in Rag22/2B6 male mice
Treg differentiation by cotransferred naive CD4 cells. HY-specific green
Tregs (Thy1.1) (1 3 104/mouse) mixed with an equal number of HY-spe-
cific naive CD4 T cells (Thy1.2) were transferred to Rag22/2B6 male mice
(n = 4). Analysis was conducted on day 14, as described in Fig. 6A. FACS
profiles of one representative mouse of four are shown. One representative
experiment of three is shown. E, Comparison of the frequency of ex-Tregs
in the absence (black bars) or presence of cotransferred naive CD4 T cells
The Journal of Immunology 4563
and NK cells in tumor-dLN (40) or in tumor tissues (41); and
Tregs preferentially expand in response to cancer vaccines (42).
Second, there is concern over the use of Treg adoptive therapy for
autoimmune diseases (16, 17). Because IL-6 plays a key role in
converting Tregs to ex-Tregs (9, 10), the efficiency of adoptive
Treg therapy may be enhanced by blockade of IL-6 signaling
pathways. Finally, the promotion of ex-Treg development in the
context of tumors can be of therapeutic benefit.
Foxp3/GFP mice, respectively.
The authors have no financial conflicts of interest.
1. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell de-
velopment by the transcription factor Foxp3. Science 299: 1057–1061.
2. 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:
3. 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.
4. Thornton, A. M., P. E. Korty, D. Q. Tran, E. A. Wohlfert, P. E. Murray,
Y. Belkaid, and E. M. Shevach. 2010. Expression of Helios, an Ikaros tran-
scription factor family member, differentiates thymic-derived from peripherally
induced Foxp3+ T regulatory cells. J. Immunol. 184: 3433–3441.
5. Gavin, M. A., J. P. Rasmussen, J. D. Fontenot, V. Vasta, V. C. Manganiello,
J. A. Beavo, and A. Y. Rudensky. 2007. Foxp3-dependent programme of regu-
latory T-cell differentiation. Nature 445: 771–775.
6. Williams, L. M., and A. Y. Rudensky. 2007. Maintenance of the Foxp3-
dependent developmental program in mature regulatory T cells requires con-
tinued expression of Foxp3. Nat. Immunol. 8: 277–284.
7. Floess, S., J. Freyer, C. Siewert, U. Baron, S. Olek, J. Polansky, K. Schlawe,
H. D. Chang, T. Bopp, E. Schmitt, et al. 2007. Epigenetic control of the foxp3
locus in regulatory T cells. PLoS Biol. 5: e38.
8. Chai, J. G., D. Coe, D. Chen, E. Simpson, J. Dyson, and D. Scott. 2008. In vitro
expansion improves in vivo regulation by CD4+CD25+ regulatory T cells. J.
Immunol. 180: 858–869.
9. Xu, L., A. Kitani, I. Fuss, and W. Strober. 2007. Cutting edge: regulatory T cells
induce CD4+CD25-Foxp3- T cells or are self-induced to become Th17 cells in
the absence of exogenous TGF-beta. J. Immunol. 178: 6725–6729.
10. Yang, X. O., R. Nurieva, G. J. Martinez, H. S. Kang, Y. Chung, B. P. Pappu,
B. Shah, S. H. Chang, K. S. Schluns, S. S. Watowich, et al. 2008. Molecular
antagonism and plasticity of regulatory and inflammatory T cell programs. Im-
munity 29: 44–56.
11. Coe, D., C. Addey, M. White, E. Simpson, J. Dyson, and J. G. Chai. 2010. The
roles of antigen-specificity, responsiveness to TGFb and APC subsets in tumour-
induced expansion of regulatory T cells. Immunology 131: 556–569.
12. Duarte, J. H., S. Zelenay, M. L. Bergman, A. C. Martins, and J. Demengeot.
2009. Natural Treg cells spontaneously differentiate into pathogenic helper cells
in lymphopenic conditions. Eur. J. Immunol. 39: 948–955.
13. Komatsu, N., M. E. Mariotti-Ferrandiz, Y. Wang, B. Malissen, H. Waldmann,
and S. Hori. 2009. Heterogeneity of natural Foxp3+ T cells: a committed reg-
ulatory T-cell lineage and an uncommitted minor population retaining plasticity.
Proc. Natl. Acad. Sci. USA 106: 1903–1908.
14. Tsuji, M., N. Komatsu, S. Kawamoto, K. Suzuki, O. Kanagawa, T. Honjo,
S. Hori, and S. Fagarasan. 2009. Preferential generation of follicular B helper
T cells from Foxp3+ T cells in gut Peyer’s patches. Science 323: 1488–1492.
15. Wan, Y. Y., and R. A. Flavell. 2007. Regulatory T-cell functions are subverted
and converted owing to attenuated Foxp3 expression. Nature 445: 766–770.
16. Zhou, X., S. L. Bailey-Bucktrout, L. T. Jeker, C. Penaranda, M. Martı ´nez-
Llordella, M. Ashby, M. Nakayama, W. Rosenthal, and J. A. Bluestone. 2009.
Instability of the transcription factor Foxp3 leads to the generation of pathogenic
memory T cells in vivo. Nat. Immunol. 10: 1000–1007.
17. Zhou, X., S. L. Bailey-Bucktrout, L. T. Jeker, and J. A. Bluestone. 2009. Plas-
ticity of CD4(+) FoxP3(+) T cells. Curr. Opin. Immunol. 21: 281–285.
18. Grandjean, I., L. Duban, E. A. Bonney, E. Corcuff, J. P. Di Santo, P. Matzinger,
and O. Lantz. 2003. Are major histocompatibility complex molecules involved
in the survival of naive CD4+ T cells? J. Exp. Med. 198: 1089–1102.
19. Wang, Y., A. Kissenpfennig, M. Mingueneau, S. Richelme, P. Perrin, S. Chevrier,
C. Genton, B. Lucas, J. P. DiSanto, H. Acha-Orbea, et al. 2008. Th2 lympho-
proliferative disorder of LatY136F mutant mice unfolds independently of TCR-
MHC engagement and is insensitive to the action of Foxp3+ regulatory T cells. J.
Immunol. 180: 1565–1575.
20. Coe, D., S. Begom, C. Addey, M. White, J. Dyson, and J. G. Chai. 2010. De-
pletion of regulatory T cells by anti-GITR mAb as a novel mechanism for cancer
immunotherapy. Cancer Immunol. Immunother. 59: 1367–1377.
21. Summerhayes, I. C., and L. M. Franks. 1979. Effects of donor age on neoplastic
transformation of adult mouse bladder epithelium in vitro. J. Natl. Cancer Inst.
22. Sugimoto, N., T. Oida, K. Hirota, K. Nakamura, T. Nomura, T. Uchiyama, and
S. Sakaguchi. 2006. Foxp3-dependent and -independent molecules specific for
CD25+CD4+ natural regulatory T cells revealed by DNA microarray analysis.
Int. Immunol. 18: 1197–1209.
23. Lu, L. F., M. A. Gavin, J. P. Rasmussen, and A. Y. Rudensky. 2007. G protein-
coupled receptor 83 is dispensable for the development and function of regu-
latory T cells. Mol. Cell. Biol. 27: 8065–8072.
24. Wing, K., Y. Onishi, P. Prieto-Martin, T. Yamaguchi, M. Miyara, Z. Fehervari,
T. Nomura, and S. Sakaguchi. 2008. CTLA-4 control over Foxp3+ regulatory
T cell function. Science 322: 271–275.
25. Yamaguchi, T., K. Hirota, K. Nagahama, K. Ohkawa, T. Takahashi, T. Nomura,
and S. Sakaguchi. 2007. Control of immune responses by antigen-specific reg-
ulatory T cells expressing the folate receptor. Immunity 27: 145–159.
26. Deaglio, S., K. M. Dwyer, W. Gao, D. Friedman, A. Usheva, A. Erat, J. F. Chen,
K. Enjyoji, J. Linden, M. Oukka, et al. 2007. Adenosine generation catalyzed by
CD39 and CD73 expressed on regulatory T cells mediates immune suppression.
J. Exp. Med. 204: 1257–1265.
27. Menning, A., U. E. Ho ¨pken, K. Siegmund, M. Lipp, A. Hamann, and J. Huehn.
2007. Distinctive role of CCR7 in migration and functional activity of naive- and
effector/memory-like Treg subsets. Eur. J. Immunol. 37: 1575–1583.
28. Kitoh, A., M. Ono, Y. Naoe, N. Ohkura, T. Yamaguchi, H. Yaguchi,
I. Kitabayashi, T. Tsukada, T. Nomura, Y. Miyachi, et al. 2009. Indispensable
role of the Runx1-Cbfbeta transcription complex for in vivo-suppressive function
of FoxP3+ regulatory T cells. Immunity 31: 609–620.
29. Klunker, S., M. M. Chong, P. Y. Mantel, O. Palomares, C. Bassin, M. Ziegler,
B. Ru ¨ckert, F. Meiler, M. Akdis, D. R. Littman, and C. A. Akdis. 2009. Tran-
scription factors RUNX1 and RUNX3 in the induction and suppressive function
of Foxp3+ inducible regulatory T cells. J. Exp. Med. 206: 2701–2715.
30. Garı ´n, M. I., C. C. Chu, D. Golshayan, E. Cernuda-Morollo ´n, R. Wait, and
R. I. Lechler. 2007. Galectin-1: a key effector of regulation mediated by CD4
+CD25+ T cells. Blood 109: 2058–2065.
31. 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.
32. Fontenot, J. D., J. P. Rasmussen, M. A. Gavin, and A. Y. Rudensky. 2005. A
function for interleukin 2 in Foxp3-expressing regulatory T cells. Nat. Immunol.
33. Gavin, M. A., S. R. Clarke, E. Negrou, A. Gallegos, and A. Y. Rudensky. 2002.
Homeostasis and anergy of CD4(+)CD25(+) suppressor T cells in vivo. Nat.
Immunol. 3: 33–41.
34. Zelenay, S., T. Lopes-Carvalho, I. Caramalho, M. F. Moraes-Fontes, M. Rebelo,
and J. Demengeot. 2005. Foxp3+ CD252 CD4 T cells constitute a reservoir of
committed regulatory cells that regain CD25 expression upon homeostatic ex-
pansion. Proc. Natl. Acad. Sci. USA 102: 4091–4096.
35. Lal, G., N. Zhang, W. van der Touw, Y. Ding, W. Ju, E. P. Bottinger, S. P. Reid,
D. E. Levy, and J. S. Bromberg. 2009. Epigenetic regulation of Foxp3 expression
in regulatory T cells by DNA methylation. J. Immunol. 182: 259–273.
36. Almeida, A. R., B. Zaragoza, and A. A. Freitas. 2006. Competition controls the rate
of transition between the peripheral pools of CD4+CD252 and CD4+CD25+
T cells. Int. Immunol. 18: 1607–1613.
37. Tang, Q., J. Y. Adams, C. Penaranda, K. Melli, E. Piaggio, E. Sgouroudis,
C. A. Piccirillo, B. L. Salomon, and J. A. Bluestone. 2008. Central role of de-
fective interleukin-2 production in the triggering of islet autoimmune de-
struction. Immunity 28: 687–697.
38. Li, M. O., S. Sanjabi, and R. A. Flavell. 2006. Transforming growth factor-beta
controls development, homeostasis, and tolerance of T cells by regulatory T cell-
dependent and -independent mechanisms. Immunity 25: 455–471.
39. Vendetti, S., T. S. Davidson, F. Veglia, A. Riccomi, D. R. Negri, R. Lindstedt,
P. Pasquali, E. M. Shevach, and M. T. De Magistris. 2010. Polyclonal Treg cells
enhance the activity of a mucosal adjuvant. Immunol. Cell Biol. 88: 698–706.
40. Boissonnas, A., A. Scholer-Dahirel, V. Simon-Blancal, L. Pace, F. Valet,
A. Kissenpfennig, T. Sparwasser, B. Malissen, L. Fetler, and S. Amigorena.
2010. Foxp3+ T cells induce perforin-dependent dendritic cell death in tumor-
draining lymph nodes. Immunity 32: 266–278.
41. Cao, X., S. F. Cai, T. A. Fehniger, J. Song, L. I. Collins, D. R. Piwnica-Worms,
and T. J. Ley. 2007. Granzyme B and perforin are important for regulatory
T cell-mediated suppression of tumor clearance. Immunity 27: 635–646.
42. Zhou, G., C. G. Drake, and H. I. Levitsky. 2006. Amplification of tumor-specific
regulatory T cells following therapeutic cancer vaccines. Blood 107: 628–636.
4564 CHARACTERIZATION OF TUMOR-SPECIFIC EX-Tregs