Extrathymic Generation of Regulatory
T Cells in Placental Mammals
Mitigates Maternal-Fetal Conflict
Robert M. Samstein,1Steven Z. Josefowicz,1Aaron Arvey,1Piper M. Treuting,2and Alexander Y. Rudensky1,*
1Howard Hughes Medical Institute and Immunology Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA
2Department of Comparative Medicine, University of Washington, Seattle, WA 98195, USA
Regulatory T (Treg) cells, whose differentiation and
function are controlled by X chromosome-encoded
transcription factor Foxp3, are generated in the
thymus (tTreg)and extrathymically
nity. Here, we demonstrate that a Foxp3 enhancer,
conserved noncoding sequence 1 (CNS1), essential
for pTreg but dispensable for tTreg cell generation,
composed of mammalian-wide interspersed repeats
(MIR) that have undergone retrotransposition during
early mammalian radiation. During pregnancy, pTreg
cells specific to a model paternal alloantigen were
generated in a CNS1-dependent manner and accu-
mulated in the placenta. Furthermore, when mated
with allogeneic, but not syngeneic, males, CNS1-
deficient females showed increased fetal resorp-
tion accompanied by increased immune cell infiltra-
tion and defective remodeling of spiral arteries.
Our results suggest that, during evolution, a CNS1-
dependent mechanism of extrathymic differentiation
of Treg cells emerged in placental animals to enforce
The benefits of the adaptive immune system of vertebrates,
which allows for highly efficient protection against invading
pathogens, have come with a substantial trade-off due to
overzealous, or ‘‘unwanted,’’ immune responses and associated
inflammation caused by infectious agents, commensal micro-
biota, autoantigens, and fetal alloantigens during pregnancy
in placental animals. Numerous mechanisms operating within
the mammalian immune system cooperatively limit deleterious
A subset of CD4+T cells known as regulatory T cells express
the X chromosome-encoded transcription factor Foxp3 and
suppress inflammatory immune responses against ‘‘self’’ and
foreign antigens in a variety of physiological and pathological
settings (Littman and Rudensky, 2010). Loss-of-function muta-
systemic immunopathology in both mice and humans, which
reveal the vital role that these cells play in immune homeostasis
(Chatila et al., 2000; Brunkow et al., 2001; Wildin et al., 2001;
Fontenot et al., 2003). Depletion of Treg cells in normal mice
also results in a fatal lympho- and myeloproliferative disorder
with widespread inflammatory lesions (Kim et al., 2007). Analysis
of CD4+T cells expressing a functional Foxp3 reporter allele and
a Foxp3 reporter null allele showed that Foxp3 is essential for
suppressor function of Treg cells (Gavin et al., 2007; Lin et al.,
Recent studies implicated Treg cells in suppression of
different types of inflammatory responses during infection,
autoimmunity, metabolic inflammation, tissue injury, autoin-
flammatory responses at barrier sites, and tumor immunity
(reviewed in Josefowicz et al., 2012). In the thymus, some
thymocytes expressing TCR with a heightened reactivity for
‘‘self’’ antigens upregulate Foxp3 and differentiate into tTreg
cells, whereas pTreg cell generation in the periphery occurs
upon stimulation of naive CD4+T cells with high-affinity cognate
TCR ligands in the presence of TGFb and retinoic acid (Chen
et al., 2003; Zheng et al., 2004; Kretschmer et al., 2005; Hall
et al., 2011). A recent observation that an intronic Foxp3
enhancer CNS1, which contains Smad3- and retinoic acid re-
ceptor (RAR)-binding sites, facilitates TGF-b-dependent Foxp3
induction and pTreg cell differentiation but is dispensable for
tTreg generation suggests that biological functions of these
two Treg cell subsets are distinct (Zheng et al., 2010). Indeed,
in contrast to fatal early-onset inflammatory lesions resulting
from congenital Treg cell deficiency, selective pTreg cell paucity
leads to a rather late onset allergic and asthma-like inflamma-
tion in the gut and lung (Josefowicz et al., 2012). Because the
principal differences between these two Treg cell subsets are
the location and type of antigen that facilitates their differentia-
tion, tTreg cells are likely responsible for tolerance to self-
antigens, whereas pTreg cells restrain immune responses to
nonself antigens such as allergens, commensal microbiota,
Pregnancy represents a physiological situation in which
tolerance to paternal alloantigens is critical for successful re-
production of placental mammals. Treg cells have been sug-
gested to play a role in pregnancy based on their increased
Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc. 29
numbers in pregnant mice and humans (Somerset et al., 2004).
Antibody-mediated depletion of CD25+Treg cells results in
increased resorption of the embryos in allogeneic matings in
mice (Aluvihare et al., 2004; Shima et al., 2010), and women
decreased numbers of CD25+CD4+Treg cells (Munoz-Suano
et al., 2011; Winger and Reed, 2011). These studies left open
a question as to whether a role for Treg cells during pregnancy
is largely due to their general ‘‘housekeeping’’ role in immune
homeostasis and the observed modulation in their numbers
is secondary to altered immune balance or whether there is
an evolutionary selected mechanism of Treg-cell-mediated
maternal-fetal tolerance. We hypothesized that, given the
capacity of pTreg cells to mediate tolerance against nonself anti-
gens, mechanisms supporting their generation arose to mitigate
maternal-fetal conflict caused by the immune response to
paternal alloantigens in placental mammals. We reasoned that
Mouse - CNS1
Zebrafish + CNS1
Oppossum + CNS1
Fold increase ofnormalized
4kb intronic region
Figure 1. Foxp3 CNS1 Element Essential for
Extrathymic Induction of Treg Cells Is
Present Only in Placental Mammals
(A) Schematic of the Foxp3 CNS1 element and
binding sites of transcription factors implicated in
pTreg cell differentiation (not to scale). Overlap
with annotated MIR retrotransposon is indicated.
Phylogenetic tree of vertebrates with spatial
conservation of CNS1 is shown as percent identity
smoothed across a 15 bp window. Scale bars are
noted, and branches with presumed MIR activity
are denoted in green.
(B) Luciferase assay of enhancer activity per-
formed in EL-4 cells using Foxp3 4 kB fragments,
including the first intron of the Foxp3 locus from
indicated species inserted downstream of the
mouse Foxp3 promoter and luciferase (Luc) gene
(schematic above). Error bars represent standard
error; n = 3.
pTreg-cell-mediated suppression might
represent such a mechanism.
In support of this hypothesis, here
we show that CNS1 enhancer is present
only in eutherian mammals and that
CNS1-dependent generation of pTreg
cells during allogeneic pregnancy in mice
plays an important role by preventing
embryo resorption and associated defec-
lation of activated T cells in the placenta.
Our results suggest that extrathymic
generation of regulatory T cells emerged
during evolution as a means of mitigation
of maternal-fetal allogeneic conflict.
CNS1 Emerges in Placental
Extrathymic induction of Foxp3 and
pTreg differentiation is facilitated by
Foxp3 CNS1 enhancer, which contains binding sites for
transcription factors activated downstream of three major
signaling pathways that have been implicated in this process
(Tone et al., 2008; Xu et al., 2010) (Figure 1A). Thus, to test
the hypothesis that pTreg differentiation may have been
gained during evolution to assist tolerance to the fetus, we
examined the conservation of the CNS1 element in a variety
of vertebrates for which annotated genome sequences are
available. Though we observed that CNS1 is highly conserved
throughout placental mammals, no evidence of a CNS1
sequence homolog was found within 100 kb of the transcrip-
tion start site of any forkhead family member in the monotreme
platypus, in marsupials wallaby and opossum, or in nonmam-
mals such as zebrafish (Figure 1A) (Margulies et al., 2007).
Importantly, a Foxp3 homolog was previously identified in
zebrafish, and its forced expression in mouse Foxp3?CD4+
T cells conferred suppressive capacity (Quintana et al.,
30 Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc.
2010). These results led us to consider that species without
identifiable Foxp3 CNS1 sequence homologs lack a proximal
regulatory element conferring efficient TGF-b-mediated Foxp3
induction, a key factor in pTreg differentiation. To test this
notion we cloned 4 kb regions downstream of the promoter
of the zebrafish, opossum, and mouse Foxp3 genes (Foxp3-
4kb) and evaluated their enhancer activities upon TCR-
(PMA/ionomycin) and TGF-b induced activation of the Foxp3
promoter in EL-4 cells in a luciferase assay previously used
for the assessment of CNS1 enhancer activity (Figure 1B)
(Tone et al., 2008). Indeed, only mouse, but not zebrafish or
opossum, Foxp3-4kb sequence markedly augmented the
Foxp3 promoter activity. Furthermore, mouse Foxp3-4kb
sequence devoid of CNS1 was lacking enhancer activity,
whereas incorporation of CNS1 into zebrafish or opossum
Foxp3-4kb sequence reconstituted enhancer activity. Though
the remote possibility remains that, in noneutherians, TGF-b
can facilitate Foxp3 induction in a species-specific manner,
these results support the idea that CNS1-like enhancer activity,
a prerequisite for pTreg differentiation, arose in placental
Interestingly, the majority of the CNS1 element sequence was
contained within an annotated SINE retrotransposon of the
mammalian-wide interspersed repeat (MIR) family (Figure 1A),
suggesting a mechanism for the abrupt emergence of CNS1 in
the Foxp3 locus, because the MIR family is thought to have
been amplified during the Mesozoic era in the course of the radi-
et al., 1995; Smit and Riggs, 1995).
Extrathymic Generation of Fetal Alloantigen-Specific
Treg Cells in Pregnancy
To test whether pTreg cells that are specific for paternal alloan-
tigen are generated during pregnancy in a CNS1-dependent
manner, we used CNS1-sufficient and -deficient Foxp3GFPmice
expressing transgenic (tg) TEa TCR that recognizes Ea52–68
peptide derived from I-Edmolecule bound to MHC class II mole-
H-2bxd(B6 3 BALB/c) F1 mice yet is absent in either parental
strains because B6 do not express antigenic peptide donor,
whereas BALB/c mice lack the appropriate presenting molecule
I-Ab. FACS-purified Foxp3?CD4+T cells from CNS1-deficient
and -sufficient TEa tg Foxp3GFPB6 mice were transferred into
T-cell-deficient B6 recipients, which were subsequently mated
with BALB/c (H-2d) or B6 (H-2b) male. In this experimental setup,
TEa cells were able to recognize endogenous Ea52-68:I-Ab
complex presented by antigen-presenting cells (APC) of the
embryo (H-2dxb) or processed by maternal APC only when recip-
ient B6 females were mated with a BALB/c male (Figure 2A).
Induction of Foxp3 in transferred CNS1-sufficient (WT) TEa
T cells was observed primarily in the draining lymph node (DLN)
and in decidua of females mated with BALB/c, but not B6 males.
In contrast, no significant induction of Foxp3 was observedupon
transfer of CNS1-deficient (KO) TEa T cells (Figure 2b). Although
condition, extrathymic Treg cells can be generated in the
absence of CNS1, these results demonstrate that pTreg cells
that are specific for fetal alloantigens are generated during preg-
nancy in a CNS1-dependent manner.
CNS1+ or CNS1-
Timed mating with
BALB/c or B6 male
B6 mated- WT
BALB/c mated- WT
Figure 2. CNS1-Dependent Generation of pTreg Cells Specific for Fetal Alloantigen during Pregnancy
(B) Percent of CD4+cells that express Foxp3 in indicated tissues on days E13.5–E14.5 of pregnancy in TCRbd?/?mice transferred with CNS1-sufficient (WT)
or -deficient (KO) TEa Foxp3-negative CD4+T cells and mated with B6 or BALB/c males. Error bars indicate SE.
Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc. 31
pTreg Cell Paucity Results in Increased Resorption
A role for CNS1 and extrathymic generation of Treg cells in
maternal-fetal tolerance and fertility was then directly assessed.
CNS1-deficient (KO)and -sufficient (WT)B6females weremated
with allogeneic BALB/c males, and the embryo resorption was
Increased resorption was observed in CNS1-deficient females
compared with the wild-type littermate controls (Figure 3B). Inci-
dence of resorption and its rate per pregnant female were also
increased (Figures 3C and 3D). Consistent with evolutionary
pressure for the emergence of CNS1-dependent pTreg cell
differentiation, the number of nonresorbed fetuses was also
significantly decreased in CNS1-deficient females (Figure S1
available online). Importantly, females impaired in pTreg cell
generation did not show increased resorption of embryos
sired by B6 males expressing syngeneic MHC allele (Figures
3E–3G). The latter observation indicates that CNS1-deficient
%Resorption per mother
% offemales with resorption
%Resorption ofall embryos
%Resorption ofall embryos
% offemales with resorption
%Resorption per mother
Figure 3. CNS1 Deficiency Results in Increased
Resorption of Embryos
(A) Macroscopic evaluation of resorption of allogeneic
embryos in uteri of CNS1-sufficient (WT) and -deficient
(KO) mice on day E14.5. Arrows indicate resorptions.
Representative of 20–30 females is shown.
(B) Percent of resorbed embryos in all CNS1-sufficient
and -deficient pregnancies withBALB/c males. Two-sided
Fisher’s exact test was used to assess the significance.
(C) Incidence of pregnancies with at least one resorption.
Two-sided Fisher’s exact test was used to assess the
(D) Percent resorption observed in individual mothers with
indicated genotype. Error bars indicate SE.
(E) Percent of embryos resorbed in pregnant CNS1-
sufficient and -deficient B6 mice mated with B6 males.
(F) Incidence of pregnancies with at least one resorption.
(G) Percent resorption observed in individual mothers.
Error bars indicate SE.
See also Figure S1.
females are not generally predisposed to
spontaneous abortion but reject embryos ex-
pressing mismatched MHC alleles. CNS1-defi-
cient females were always compared to their
WT counterparts in identical breedings with
syngeneic or allogeneic males to ensure that
the genetic make-up of the embryos is the
same in the two groups compared, as different
strains of mice can vary in rates of embryo loss
due to early fetal death unrelated to immuno-
logic conflicts and because F1 embryos are
frequently more robust and survive better.
To exclude potential effects of congenital
pTreg cell deficiency, we generated heterozy-
gous Foxp3CNS1KO/DTRfemales containing one
CNS1 KO allele and one Foxp3DTRallele. Due
to random X chromosome inactivation, half of
the cells in a Foxp3CNS1KO/DTRmouse express
each allele and are either susceptible to
DT-mediated ablation (Foxp3DTR) or unable to
induce Foxp3 in the periphery (Foxp3CNS1KO). Administration of
DT allows for depletion of all CNS1-dependent pTreg cells
while sparing a pool of CNS1-deficient tTreg cells (Figure S2).
Treatment with DT before and during pregnancy in these mice
resulted in increased resorption compared to control heterozy-
gous Foxp3GFP/DTRfemales in which DT ablation leaves CNS1-
sufficient Foxp3+Treg cells (Figure 4A). The observed increases
were similar to those observed in intact CNS1-deficient females.
Thus, acute pTreg ablation during allogeneic pregnancy results
in increased embryo resorption comparable to that observed in
Extrathymically Generated Treg Cells Play
a Predominant Role in Maternal-Fetal Tolerance
Foxp3CNS1KO/DTRfemales suggested that pTreg cells generated
in a CNS1-dependent manner prevent ‘‘rejection’’ of MHC-
mismatched fetuses, it was possible that their contribution to
32 Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc.
overall Treg cell-mediated suppression of maternal-fetal alloge-
neic conflict was relatively minor. To address this question, we
assessed the effect of essentially complete ablation of Treg cells
in pregnant Foxp3DTRB6 females expressing diphtheria toxin
receptor under control of the endogenous Foxp3 locus (Kim
etal., 2007). Increased resorption of MHC-mismatched embryos
was observed in Foxp3DTRB6 females, compared with control
B6 mice, when mated with BALB/c males and treated with diph-
theria toxin (DT) at days E5.5 and E7.5 (Figure 4B). In contrast to
selective pTreg deficiency, DT-mediated combined ablation of
pTreg and tTreg cells also increased resorption of syngeneic
embryos, most likely due to a loss of restraint of T cell reactivity
against self-antigens (data not shown). Despite widespread
immune mediated inflammation and lympho- and myeloprolifer-
ative syndrome in pregnant Foxp3DTRfemales subjected to
‘‘wholesale’’ Treg ablation, the rates of resorption observed in
these mice were similar to those in CNS1-deficient females and
upon acute ablation of pTreg cells. These findings indicate that
pTreg cells play a predominant role in maternal-fetal tolerance.
CNS1-Deficient Females Exhibit Signs of Inflammation
and Abnormal Spiral Artery Remodeling
Consistent with the impaired pTreg induction during allogeneic
pregnancy in CNS1-deficient females, we observed markedly
reduced Treg cell numbers in the decidua in these mice mated
with BALB/c males (Figure 5A). Proliferative activity assessed
by Ki67 expression was not increased in Treg cell subsets in
the decidua and DLN in CNS1-deficient mice in comparison to
CNS1-sufficient controls (Figure S3). These results suggested
that paucity of pTreg cells is not associated with compensatory
expansion of tTreg cells, in agreement with our recent observa-
tion (Josefowicz et al., 2012). The observed decrease in the Treg
cell population inversely correlated with the increased presence
of activated effector CD62LloCD4+T cells (Figure 5B); however,
no significant changes in effector cytokine production were
detected in the DLN or decidua (Figure S4). It must be noted
that fluid composition and activation state of immune cells in
a rapidly changing placental environment could mask potential
differences. Histologic examination of the placentas of CNS1-
deficient and -sufficient females performed in a blinded fashion
showed that the genotype of the dams can be accurately
predicted by the morphological status of the decidual spiral
arteries; although there was some variability between individual
placentas in any one uterus, CNS1-deficient placentas exhibited
more prominent clusters of thickened blood vessels as com-
pared to CNS1-sufficient littermate controls (Figure 5C). At day
E13.5, placentas of surviving embryos of CNS1-deficient
females exhibited early necrosis of spiral arteries and edema,
whereas resorptions were characterized by embryo loss with
necrotic labyrinths (Figure 5D). Presumptive early resorption
sites exhibited necrosis or thrombosis of decidual vessels and
edema in the placentas and embryos (Figure S5). Consistent
with the immune-mediated pathology, more prominent T cell
presence was noted in CNS1-deficient placentas, where single
T cells were scattered within all layers of the placenta with
clusters prominent in the decidua near spiral arteries (Figure 5E).
In contrast, only rare single T cells were observed in CNS1-
sufficient placentas, and no major changes in the numbers of B
cells and macrophages were detected (Figure S5). These obser-
vations are indicative of an inflammatory pathology associated
that of human pregnancy-associated disorders such as pre-
eclampsia (Redman and Sargent, 2005; Renaud et al., 2011).
Over the last decade, numerous studies have led to the realiza-
tion that suppressive function of Treg cells extends far beyond
%Resorption ofall embryos
%Resorption per mother
% offemales with resorption
Figure 4. Acute Depletion of pTreg and All Treg Cells Results in
Comparable Increase in Embryo Resorption
females treated with DT continuously starting 3 days prior to mating with
BALB/c males; incidence of pregnancies with at least one resorption; percent
resorption observed in individual mothers. Error bars indicate SE.
(B) Percent of resorbed embryos, incidence of resorption, and percent
resorption per mother for wild-type or Foxp3DTRB6 mice mated with BALB/c
males and treated with DT on days E5.5 and E7.5.
See also Figure S2.
Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc. 33
autoimmunity, the originally suggested sphere of their activity.
Treg cells have been implicated in control of acute and chronic
infections, tissue homeostasis at barrier sites populated by
commensal microbiota, allergy, injury response and tissue
repair, metabolic syndrome, and cancer (reviewed in Josefowicz
et al., 2012). In this study, we find that the Foxp3 intronic
%Foxp3+ of CD4+
%Foxp3+ of CD4+
%CD62Llo of CD4+Foxp3-
%CD44hiCD62Llo of CD4+Foxp3-
Figure 5. Decreased Treg Cell Numbers and Histological Features of Immune-Mediated Resorption in Deciduas of CNS1-Deficient
(A) Representative flow cytometric analysis of Foxp3+Treg cells in decidua and analysis of decidua and lymph nodes (LN) of CNS1-sufficient (WT) and -deficient
(KO) mice mated with BALB/c males and analyzed on days E13.5–E14.5. Error bars indicate SE; n = 8–12.
(B) Representative flow cytometric analysis of activated CD62LloFoxp3-negative CD4+T cells within the decidua and analysis of decidua and LN.
(C) Histopathological evaluation of placentas from WT (left) and CNS1 KO (right) females mated with BALB/c males; low-power magnification survey of repre-
sentative sections of H&E stained placenta. The maternal spiral arteries (SA) are more frequently clustered and prominent in KO placenta at day E12.5 (arrows,
upper-right) (representative of 6–8 mice analyzed per group with 4–10 placental sites each).
(D) Analysis of H&E stained sections of the KO placentas at day E13.5; early necrosis of SA (arrow, left) in the decidua (DB) and edema (arrowhead, left) at the
chorionic plate. Resorption sites (lower right) shown in the same animal were characterized by loss of embryo and necrotic labyrinths (L) with variable necrosis in
the trophoblast (T) layer. E, embryo; YS, yolk sac. Scale bars, 500 mm.
(E) Immunohistochemical staining for CD3 in day E12.5 placentas from CNS1-sufficient (WT) and -deficient (KO) females. CD3+T cells (brown staining) are more
numerous in the KO placentas in proximity to maternal spiral arteries (SA). Scale bars, 100 mm.
See also Figures S3, S4, and S5.
34 Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc.
enhancer CNS1, essential for extrathymic differentiation of Treg
or monotremes, and that pTreg cell paucity in CNS1-deficient
females mated to MHC-mismatched males results in increased
spontaneous abortion of embryos. An implication of these
observations is that generation of Treg cells in the thymus
does not afford adequate protection of the fetus expressing allo-
geneic MHC alleles from immune-mediated attack by maternal
T cells. This latter notion is also supported by the comparable
extent of embryo resorption associated with acute depletion of
pTreg and pan-Treg ablation, i.e., elimination of both pTreg
and tTreg cells. This finding implies that extrathymically gener-
ated Treg cells serve as the predominant subset mitigating
maternal-fetal allogeneic conflict. We would propose that,
once in place, extrathymic generation of Treg cells, primarily
driven by the pressure to enforce maternal fetal tolerance, likely
assumed additional functions, including control of responses to
non-self antigens leading to allergy and asthma and to
commensal organisms in the gut (Lathrop et al., 2011; Josefo-
wicz et al., 2012).
Although pTreg cell deficiency in CNS1-deficient mice
resulted in significantly increased resorption of fetuses during
allogeneic pregnancy, its penetrance was incomplete. This
was a rather expected result likely due to multiple mechanisms
that operate during pregnancy to limit encounter of maternal
alloreactive T cells with, or their response to, fetal alloantigens.
These mechanisms include but are not limited to inactivation of
immune cells by tryptophan deprivation by indoleamine 2,3-
dioxygenases (Munn et al., 1998), Fas-Fas ligand-mediated
apoptosis of activated alloreactive T cells (Hunt et al., 1997),
expression of immunosuppressive mediators such as TGF-b
and galectin-1 (Simpson et al., 2002; Blois et al., 2007), entrap-
ment of endometrial dendritic cells (Collins et al., 2009), limited
expression of MHC molecules on trophoblasts (Erlebacher
et al., 2007), and increased expression of inhibitory B7 family
members(PD-L1,B7H3,B7H4) (reviewedinPetroff andPerchel-
let, 2010). It seems reasonable to suggest that pTreg-cell-
mediated suppression enforces maternal-fetal tolerance not
single handedly, but jointly with other numerous immunomodu-
Additional factors, which could influence the degree of
immune-mediated resorption associated with pTreg cell or
pan-Treg cell deficiency include the genetic background, micro-
bial status, and stress exposure. It is likely that, in the absence of
pTreg cells, infection may result in a more severe pregnancy
disruption. It must be also noted that the 3-week-long gestation
period in mice is relatively short; extrathymic generation of Treg
cells may play a more pronounced role in maternal-fetal toler-
ance in mammals with longer gestation times, for which there
would be higher probability of the encounter of alloreactive
T cells of the mother with paternally encoded alloantigens and
for the immune response to develop.
The aforementioned possible influences affecting severity of
pregnancy disruption and differences in experimental design
might account for a varying degree of embryo resorption ob-
served in our experiments and in previous reports employing
adoptive T cell transfers and CD25-antibody-mediated Treg
cell depletion in lymphopenic or lymphoreplete mice (Aluvihare
et al., 2004; Shima et al., 2010). However, we have encountered
pervasive fetal death observed upon continuous DT-mediated
ablation of Treg cells starting at midgestation daily, which was
likely due to secondary effects of the poor health condition of
the mother (Rowe et al., 2011).
Our demonstration of a key role of extrathymic generation of
Treg cells in maternal-fetal tolerance substantially adds to
previous studies demonstrating the general importance of Treg
cells in control of maternal immune responses to the allogeneic
fetus in mice. In humans, Treg cells are present in increased
kinen et al., 2004; Somerset et al., 2004). Decreases in Treg cells
have been associated with frequent human pregnancy disor-
ders, including pre-eclampsia and repeated spontaneous abor-
tions (Arruvito et al., 2009; Darmochwal-Kolarz et al., 2012).
The histological features of allogeneic pregnancy in pTreg-cell-
deficient females were redolent of abnormal spiral artery remod-
eling associated with pre-eclampsia and other complications of
pregnancy inhumans andaccompanying increasedlocalinflam-
mation (Redman and Sargent, 2005; Avagliano et al., 2011;
Renaud et al., 2011). These observations raise an intriguing
accumulation and resulting pathology may be partially due to
defective peripheral induction of Treg cells to paternal antigens,
and potential therapies could be developed to address this
The analysis of CNS1 sequence conservation suggests that
this enhancer was gained during evolution of eutherian
mammals. CNS1 contains binding sites for transcription factors
downstream of three major signaling pathways required for
pTreg generation, and its deletion results in a selective impair-
ment of this differentiation process (Zheng et al., 2010). Thus,
the introduction of this several hundred base-pair-long DNA
sequence into the Foxp3 locus could have been sufficient to
enable the differentiation of pTreg cells. This line of reasoning
implies that the increased interaction between the mother and
fetus during gestation necessitated a mechanism of acquired
active tolerance afforded by peripheral generation of regulatory
T cells. Consistent with this idea, diminished litter size observed
in allogeneic pregnancy in CNS1-deficient versus -sufficient
females suggested that pTreg generation afforded a reproduc-
The process of insertion of CNS1 sequence into the first
intron of the Foxp3 locus appears to have occurred via
a MIR family retrotransposon activity during the Mesozoic
era at a time overlapping with the evolution of placental
mammals. We found that, in the mouse genome, MIR elements
resembling CNS1 are enriched for SMAD- and RXR-binding
sites (data not shown), suggesting that these elements may
have endowed TGF-b and retinoic acid response capacity to
Foxp3 and other genes, in agreement with the idea of exapta-
tion, in which portions of transposable elements acquire a func-
tion that serves their host (Brosius and Gould, 1992). These
elements can then confer novel signaling pathway responsive-
ness to existing genes, thereby augmenting their function.
A mechanism of retrotransposon-mediated exaptation affect-
ing the structure or regulation of pre-existing genes upon
Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc. 35
introduction of novel exons or enhancers has been previously
reported (Bejerano et al., 2006; Mikkelsen et al., 2007). We
suggest that acquisition of CNS1 supporting extrathymic
Treg cell generation in eutherian mammals represents an
example of retrotransposition-mediated innovation in regula-
tion of gene expression, whereby a distinct biological purpose
and novel functionality associated with the CNS1 enhancer are
implicit of the potential evolutionary pressure underlying its
It is noteworthy that the emergence of chorioallantoic
placenta, which allowed for viviparity in therian mammals, was
assisted by appropriation of several retroviral genes, including
retrotransposon-derived Peg10 and Peg11/Rtl1 and syncitin-A
and -B originating from the envelope protein of a defective retro-
virus. These genes are essential for normal function of placenta
and trophoblast fusion, respectively (Mi et al., 2000; Ono et al.,
2006; Sekita et al., 2008; Dupressoir et al., 2011). Taken
together, these results and our findings suggest that, in addition
to facilitating placentation during mammalian evolution, retro-
transposon-mediated innovation helped to alleviate immune
conflict associated with this acquisition.
In conclusion, we suggest that the mechanism of extrathymic
differentiation of Treg cells may have been gained during evolu-
tion to reinforce tolerance to paternal alloantigens presented by
the fetus during the increasingly long gestation period in
placental mammals. This adaptation was realized with the aid
of the Foxp3 CNS1 enhancer responsible for induction of
Foxp3 in peripheral CD4+Foxp3?T cells, which likely emerged
upon capture of a MIR retrotransposon containing TGF-b and
retinoic acid receptor response elements. A role of extrathymi-
cally generated Treg cells in maternal-fetal tolerance may
provide an important insight into potential clinical complications
of human pregnancies.
Mouse Strains and Timed Matings
Foxp3GFP, Foxp3DTR, TEa, and CNS1-deficient mice on a B6 background were
previously described (Grubin et al., 1997; Fontenot et al., 2005; Kim et al.,
2007; Zheng et al., 2010). TCRbd-deficient B6 mice were purchased from
Jackson Laboratory and maintained as a homozygous colony. All of the
mice were bred and housed in the specific pathogen-free animal facility at
the Memorial Sloan-Kettering Cancer Center and were used in accordance
with institutional guidelines. Diphtheria toxin (DT) (Sigma) was administered
twice i.p., as indicated.
One or two female mice were set up in the afternoon with individual males.
Femaleswerechecked daily forthepresenceofavaginal pluginthemornings,
and plugged females were separated from males; the day of plug detection
was considered day E0.5. Plugged females were analyzed for resorbed
fetuses at E14.5, and the resorption was always analyzed in three different
ways to confirm significance (see Statistical Analysis).
Adoptive Cell Transfers
TEa CD4+Foxp3-negative (GFP?) cells were purified using an Aria2 cell sorter
(BD Biosciences) after enrichment of CD4 cells using Dynal CD4 magnetic
beads according to manufacturer instructions (Invitrogen). 4 3 106cells
were injected i.v. into TCRbd-deficient B6 females, and the recipient mice
were time mated with B6 or BALB/c males. Pregnant mice were analyzed on
CNS1 Sequence Analysis
TheFoxp3locus,including the100kbflanking sequence 50and 30of theFoxp3
gene, was found in each species by ENSEMBL annotations. The most CNS1-
like sequence in all species was determined by scanning for mouse CNS1
across the Foxp3 locus using global-local alignment. For scoring alignments,
there was no penalty for opening gaps, ?1 for extending gaps or mismatched
nucleotides, and +1 for matched nucleotides. All of CNS1 was aligned to a
moving window twice the size of CNS1, and the window was moved to 50%
of the size of CNS1. Genome-wide phylogenies were downloaded from
mod, and branch lengths were scaled to the number of substitutions per site
(Nikolaev et al., 2007; Pollard et al., 2010). Phylogenetic tree was unrooted,
and the root was arbitrarily selected. The tree had two scales, as the distance
between noneutherians is much larger than eutherians.
Foxp3 luciferase expression constructs were generated using Infusion cloning
system(Clontech)and wereverifiedby restriction digestsand sequencing.53
106EL4-LAF cells were mixed with 5 mg of indicated vector and 0.8 mg of
pRL-TK control vector in complete RPMI with 20% fetal bovine serum (FBS),
and electroporation was performed using a Biorad electroporator (300V,
1000 mF). Cells were rested for 15 min and then incubated in complete RPMI
supplemented with 10% FBS for 1 hr before addition of PMA, ionomycin,
and TGFb (250 ng/ml, 25 ng/ml, and 4 ng/ml, respectively). After 18–24 hr cells
were lysed, and dual luciferase activity was measured according to manufac-
turer instructions (Promega). Firefly luciferase activity levels were normalized
to Renilla luciferase, and the resulting normalized values with stimulation
were divided by those without to determine the stimulation dependent
enhancement of promoter activity.
Single-cell suspensions from lymph nodes and spleen were prepared by
mechanical disruption after dissection. Deciduas were isolated by careful
dissection from the uterus and separation of the yolk sac containing the fetus
followed by mechanical disruption. Fluorophore-conjugated antibodies were
purchased from BD Biosciences and eBioscience. Intracellular Foxp3 staining
cells were analyzed using an LSRII flow cytometer (BD Biosciences), and data
were analyzed using FlowJo software (Treestar).
Hematoxylin-, eosin-, and periodic acid Schiff-stained placental-fetal units
from days E12.5 to E13.5 were examined histologically for qualitative changes
to morphologically characterize early resorptions. Changes evaluated in-
cluded clustered and thickened decidual spiral arteries, necrosis, thrombosis,
edema, and inflammation in any placental layer and necrotic placenta sites
that lacked embryos (resorptions). For each uterus, all intact placental sites
were evaluated individually and scored for viability (resorptions) and promi-
nence of clustered spiral arteries, necrosis in the metrial gland, and necrosis
and inflammation in the trophoblast layer. A total of 6 WT uteri with 60 gesta-
tional sites and 7 KO uteri with 51 sites were examined.
Placental-embryo units were stained with antibodies against CD3 (T cells;
Clone CD3-12, AbD Serotec), F4/80 (macrophages; CALTAG), and CD31
(blood vessels; Abcam). Appropriate isotype controls and Bond Polymer
Refine (DAB) detection kit including peroxide block, polymer, and hematoxylin
(Leica) were used. Slides were qualitatively examined for signal detected by
DAB staining (brown). Images were captured with Nikon Eclipse 80i with CFI
plan apo-objectives and Nikon Digital Sight DS-Fi1 12 mega pixel camera
and Nikon Basic Elements Software (Nikon). Raw images were edited for
brightness in Adobe Photoshop Elements.
Unless otherwise noted, statistical analysis was performed using the unpaired
two-tailed Student’s t test with Welch’s correction for unequal variances for
individual biological replicates in Prism (GraphPad). The Fisher’s exact test
36 Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc.
was used to assess significance of the ratios of healthy and resorbed fetuses.
This method assumes that each individual fetus represents an independent
event due tophysical separation ofindividualembryosfrom mother’s alloreac-
tive T cells and likely probabilistic factors affecting alloreactive T cell activation
in the placenta. In support of the choice of the Fisher’s exact test as the
adequate way to analyze the data in question, we very rarely observed large
numbers of resorbed embryos in a given female or strings of resorbing
embryos, i.e., frequent resorptions of neighboring embryos. These observa-
tions suggested that resorption of each fetus seems to be an independent
event with a considerable stochastic component. Nevertheless, to exclude
the potential concern of nonindependence, we also performed the Fisher’s
exact test on the incidence of mothers having any resorption event. Lastly,
the Student’s t test was also used to compare the percent resorption of the
fetuses in individual females. Nonparametric tests were also significant, but
p values are not shown.
Supplemental Information includes five figures and can be found with this
article online at http://dx.doi.org/10.1016/j.cell.2012.05.031.
We would like to thank J. Gerard, Y. Liang, D. Canner, and M. Samstein;
Z. Williams for helpful discussion; S. Gelber for review of histology; and
P. Samollow and J. Vandeberg for providing opossum genomic DNA.
R.M.S. was supported by NIH DK091968 and MSTP grant GM07739. A.Y.R.
was supported by NIH grant R37 AI21609. A.Y.R. is an investigator with the
Howard Hughes Medical Institute.
Received: April 18, 2012
Revised: April 26, 2012
Accepted: May 10, 2012
Published: July 5, 2012
Aluvihare, V.R., Kallikourdis, M., and Betz, A.G. (2004). Regulatory T cells
mediate maternal tolerance to the fetus. Nat. Immunol. 5, 266–271.
Arruvito, L., Billordo, A., Capucchio, M., Prada, M.E., and Fainboim, L. (2009).
IL-6 trans-signaling and the frequency of CD4+FOXP3+ cells in women with
reproductive failure. J. Reprod. Immunol. 82, 158–165.
Avagliano, L., Bulfamante, G.P., Morabito, A., and Marconi, A.M. (2011).
Abnormal spiral artery remodelling in the decidual segment during pregnancy:
from histology to clinical correlation. J. Clin. Pathol. 64, 1064–1068.
Bejerano, G., Lowe, C.B., Ahituv, N., King, B., Siepel, A., Salama, S.R., Rubin,
E.M., Kent, W.J., and Haussler, D. (2006). A distal enhancer and an ultracon-
served exon are derived from a novel retroposon. Nature 441, 87–90.
Blois, S.M., Ilarregui, J.M., Tometten, M., Garcia, M., Orsal, A.S., Cordo-
Russo, R., Toscano, M.A., Bianco, G.A., Kobelt, P., Handjiski, B., et al.
(2007). A pivotal role for galectin-1 in fetomaternal tolerance. Nat. Med. 13,
Brosius, J., and Gould, S.J. (1992). On ‘‘genomenclature’’: a comprehensive
(and respectful) taxonomy for pseudogenes and other ‘‘junk DNA’’. Proc.
Natl. Acad. Sci. USA 89, 10706–10710.
Brunkow, M.E., Jeffery, E.W., Hjerrild, K.A., Paeper, B., Clark, L.B., Yasayko,
S.A., Wilkinson, J.E., Galas, D., Ziegler, S.F., and Ramsdell, F. (2001). Dis-
ruption of anew forkhead/winged-helix protein,scurfin, resultsin the fatal lym-
phoproliferative disorder of the scurfy mouse. Nat. Genet. 27, 68–73.
Chatila, T.A., Blaeser, F., Ho, N., Lederman, H.M., Voulgaropoulos, C., Helms,
C., and Bowcock, A.M. (2000). JM2, encoding a fork head-related protein, is
mutated in X-linked autoimmunity-allergic disregulation syndrome. J. Clin.
Invest. 106, R75–R81.
Chen, W., Jin, W., Hardegen, N., Lei, K.-J., Li, L., Marinos, N., McGrady, G.,
and Wahl, S.M. (2003). Conversion of peripheral CD4+CD25- naive T cells to
CD4+CD25+ regulatory T cells by TGF-beta induction of transcription factor
Foxp3. J. Exp. Med. 198, 1875–1886.
Collins, M.K., Tay, C.-S., and Erlebacher, A. (2009). Dendritic cell entrapment
within the pregnant uterus inhibits immune surveillance of the maternal/fetal
interface in mice. J. Clin. Invest. 119, 2062–2073.
Darmochwal-Kolarz, D., Kludka-Sternik, M., Tabarkiewicz, J., Kolarz, B.,
Rolinski, J., Leszczynska-Gorzelak, B., and Oleszczuk, J. (2012). The predom-
inance of Th17 lymphocytes and decreased number and function of Treg cells
in preeclampsia. J. Reprod. Immunol. 93, 75–81.
Dupressoir, A., Vernochet, C., Harper, F., Gue ´gan, J., Dessen, P., Pierron, G.,
and Heidmann,T. (2011). A pair ofco-opted retroviral envelope syncytin genes
is required for formation of the two-layered murine placental syncytiotropho-
blast. Proc. Natl. Acad. Sci. USA 108, E1164–E1173.
Erlebacher, A., Vencato, D., Price, K.A., Zhang, D., and Glimcher, L.H. (2007).
Constraints in antigen presentation severely restrict T cell recognition of the
allogeneic fetus. J. Clin. Invest. 117, 1399–1411.
Fontenot, J.D., Gavin, M.A., and Rudensky, A.Y. (2003). Foxp3 programs the
development and function of CD4+CD25+ regulatory T cells. Nat. Immunol.
Fontenot, J.D., Rasmussen, J.P., Williams, L.M., Dooley, J.L., Farr, A.G., and
Rudensky, A.Y. (2005). Regulatory T cell lineage specification by the forkhead
transcription factor foxp3. Immunity 22, 329–341.
Gavin, M.A., Rasmussen, J.P., Fontenot, J.D., Vasta, V., Manganiello, V.C.,
Beavo, J.A., and Rudensky, A.Y. (2007). Foxp3-dependent programme of
regulatory T-cell differentiation. Nature 445, 771–775.
Grubin, C.E., Kovats, S., deRoos, P., and Rudensky, A.Y. (1997). Deficient
positive selection of CD4 T cells in mice displaying altered repertoires of
MHC class II-bound self-peptides. Immunity 7, 197–208.
Hall, J.A., Grainger, J.R., Spencer, S.P., and Belkaid, Y. (2011). The role of ret-
inoic acid in tolerance and immunity. Immunity 35, 13–22.
Heikkinen, J., Mo ¨tto ¨nen, M., Alanen, A., and Lassila, O. (2004). Phenotypic
characterization of regulatory T cells in the human decidua. Clin. Exp.
Immunol. 136, 373–378.
Hunt, J.S., Vassmer, D., Ferguson, T.A., and Miller, L. (1997). Fas ligand is
positioned in mouse uterus and placenta to prevent trafficking of activated
leukocytes between the mother and the conceptus. J. Immunol. 158, 4122–
Josefowicz, S.Z., Niec, R.E., Kim, H.Y., Treuting, P., Chinen, T., Zheng, Y.,
Umetsu, D.T., and Rudensky, A.Y. (2012). Extrathymically generated regula-
tory T cells control mucosal TH2 inflammation. Nature 482, 395–399.
Jurka, J., Zietkiewicz, E., and Labuda, D. (1995). Ubiquitous mammalian-wide
interspersed repeats (MIRs) are molecular fossils from the mesozoic era.
Nucleic Acids Res. 23, 170–175.
Kim, J.M., Rasmussen, J.P., and Rudensky, A.Y. (2007). Regulatory T cells
prevent catastrophic autoimmunity throughout the lifespan of mice. Nat.
Immunol. 8, 191–197.
Kretschmer, K., Apostolou, I., Hawiger, D., Khazaie, K., Nussenzweig, M.C.,
and von Boehmer, H. (2005). Inducing and expanding regulatory T cell popu-
lations by foreign antigen. Nat. Immunol. 6, 1219–1227.
Lathrop, S.K., Bloom, S.M., Rao, S.M., Nutsch, K., Lio, C.-W., Santacruz, N.,
Peterson, D.A., Stappenbeck, T.S., and Hsieh, C.-S. (2011). Peripheral educa-
tion of the immune system by colonic commensal microbiota. Nature 478,
Lin, W., Haribhai, D., Relland, L.M., Truong, N., Carlson, M.R., Williams, C.B.,
and Chatila, T.A. (2007). Regulatory T cell development in the absence of func-
tional Foxp3. Nat. Immunol. 8, 359–368.
Littman, D.R., and Rudensky, A.Y. (2010). Th17 and regulatory T cells in medi-
ating and restraining inflammation. Cell 140, 845–858.
Margulies, E.H.,Cooper, G.M., Asimenos, G., Thomas, D.J.,Dewey,C.N., Sie-
pel, A., Birney, E., Keefe, D., Schwartz, A.S., Hou, M., et al. (2007). Analyses of
deep mammalian sequence alignments and constraint predictions for 1% of
the human genome. Genome Res. 17, 760–774.
Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc. 37
Mi, S., Lee, X., Li, X., Veldman, G.M., Finnerty, H., Racie, L., LaVallie, E., Tang, Download full-text
X.Y., Edouard,P.,Howes, S., et al.(2000). Syncytin is a captive retroviral enve-
lope protein involved in human placental morphogenesis. Nature 403,
Mikkelsen, T.S., Wakefield, M.J., Aken, B., Amemiya, C.T., Chang, J.L., Duke,
S., Garber, M., Gentles, A.J., Goodstadt, L., Heger, A., et al; Broad Institute
Genome Sequencing Platform; Broad Institute Whole Genome Assembly
Team. (2007). Genome of the marsupial Monodelphis domestica reveals inno-
vation in non-coding sequences. Nature 447, 167–177.
Munn, D.H., Zhou, M., Attwood, J.T., Bondarev, I., Conway, S.J., Marshall, B.,
Brown, C., and Mellor, A.L. (1998). Prevention of allogeneic fetal rejection by
tryptophan catabolism. Science 281, 1191–1193.
Munoz-Suano, A., Hamilton, A.B., and Betz, A.G. (2011). Gimme shelter: the
immune system during pregnancy. Immunol. Rev. 241, 20–38.
Nikolaev, S., Montoya-Burgos, J.I., Margulies, E.H., Rougemont, J., Nyffeler,
B., and Antonarakis, S.E.; NISC Comparative Sequencing Program. (2007).
Early history of mammals is elucidated with the ENCODE multiple species
sequencing data. PLoS Genet. 3, e2.
Ono, R., Nakamura, K., Inoue, K., Naruse, M., Usami, T., Wakisaka-Saito, N.,
Hino, T., Suzuki-Migishima, R., Ogonuki, N., Miki, H., et al. (2006). Deletion of
Peg10, an imprinted gene acquired from a retrotransposon, causes early
embryonic lethality. Nat. Genet. 38, 101–106.
Petroff, M.G., and Perchellet, A. (2010). B7 family molecules as regulators of
the maternal immune system in pregnancy. Am. J. Reprod. Immunol. 63,
Pollard,K.S., Hubisz,M.J., Rosenbloom,K.R., and Siepel, A. (2010). Detection
of nonneutral substitution rates on mammalian phylogenies. Genome Res. 20,
Quintana, F.J., Iglesias, A.H., Farez, M.F., Caccamo, M., Burns, E.J., Kassam,
N., Oukka, M., and Weiner, H.L. (2010). Adaptive autoimmunity and Foxp3-
based immunoregulation in zebrafish. PLoS ONE 5, e9478.
Redman, C.W., and Sargent, I.L. (2005). Latest advances in understanding
preeclampsia. Science 308, 1592–1594.
Renaud,S.J., Cotechini, T.,Quirt,J.S.,Macdonald-Goodfellow, S.K.,Othman,
M., and Graham, C.H. (2011). Spontaneous pregnancy loss mediated by
abnormal maternal inflammation in rats is linked to deficient uteroplacental
perfusion. J. Immunol. 186, 1799–1808.
Rowe, J.H., Ertelt, J.M., Aguilera, M.N., Farrar, M.A., and Way, S.S. (2011).
Foxp3(+) regulatory T cell expansion required for sustaining pregnancy
compromises host defense against prenatal bacterial pathogens. Cell Host
Microbe 10, 54–64.
Sekita, Y., Wagatsuma, H., Nakamura, K., Ono, R., Kagami, M., Wakisaka, N.,
Hino, T., Suzuki-Migishima, R., Kohda, T., Ogura, A., et al. (2008). Role of
retrotransposon-derived imprinted gene, Rtl1, in the feto-maternal interface
of mouse placenta. Nat. Genet. 40, 243–248.
Shima, T., Sasaki, Y., Itoh, M., Nakashima, A., Ishii, N., Sugamura, K., and
Saito, S. (2010). Regulatory T cells are necessary for implantation and mainte-
nance of early pregnancy but not late pregnancy inallogeneic mice. J.Reprod.
Immunol. 85, 121–129.
Simpson, H., Robson, S.C., Bulmer, J.N., Barber, A., and Lyall, F. (2002).
Transforming growth factor beta expression in human placenta and placental
bed during early pregnancy. Placenta 23, 44–58.
Smit, A.F., and Riggs, A.D. (1995). MIRs are classic, tRNA-derived SINEs that
amplified before the mammalian radiation. Nucleic Acids Res. 23, 98–102.
Somerset, D.A., Zheng, Y., Kilby, M.D., Sansom, D.M., and Drayson, M.T.
(2004). Normal human pregnancy is associated with an elevation in the
immune suppressive CD25+ CD4+ regulatory T-cell subset. Immunology
Tone, Y.,Furuuchi, K., Kojima, Y.,Tykocinski, M.L., Greene, M.I., and Tone, M.
(2008). Smad3 and NFAT cooperate to induce Foxp3 expression through its
enhancer. Nat. Immunol. 9, 194–202.
Wildin, R.S., Ramsdell, F., Peake, J., Faravelli, F., Casanova, J.L., Buist, N.,
Levy-Lahad, E., Mazzella, M., Goulet, O., Perroni, L., et al. (2001). X-linked
neonatal diabetes mellitus, enteropathy and endocrinopathy syndrome is the
human equivalent of mouse scurfy. Nat. Genet. 27, 18–20.
regulatory cell levels predict miscarriage risk in newly pregnant women with
a history of failure. Am. J. Reprod. Immunol. 66, 320–328.
Xu, L., Kitani, A., Stuelten, C., McGrady, G., Fuss, I., and Strober, W. (2010).
Positive and negative transcriptional regulation of the Foxp3 gene is mediated
by access and binding of the Smad3 protein to enhancer I. Immunity 33,
Zheng, S.G., Wang, J.H., Gray, J.D., Soucier, H., and Horwitz, D.A. (2004).
Natural and induced CD4+CD25+ cells educate CD4+CD25- cells to develop
suppressive activity: the role of IL-2, TGF-beta, and IL-10. J. Immunol. 172,
Zheng, Y., Josefowicz, S., Chaudhry, A., Peng, X.P., Forbush, K., and
Rudensky, A.Y. (2010). Role of conserved non-coding DNA elements in the
Foxp3 gene in regulatory T-cell fate. Nature 463, 808–812.
38 Cell 150, 29–38, July 6, 2012 ª2012 Elsevier Inc.