T cell apoptosis at the maternal–fetal interface
in early human pregnancy, involvement
Hernan D. Kopcowa, Florencia Rosettia, Yiuka Leunga, David S. J. Allana, Jeffrey L. Kutokb, and Jack L. Stromingera,1
aDepartment of Molecular and Cellular Biology, Harvard University, 7 Divinity Avenue, Cambridge, MA 02138; andbPathology Department, Brigham and
Women’s Hospital, Harvard Medical School, Boston, MA 02115
Contributed by Jack L. Strominger, September 24, 2008 (sent for review August 20, 2008)
The human fetus is not rejected by the maternal immune system
despite expressing paternal antigens. Natural killer cells, the major
lymphocyte population of the human decidua (dNKs), express
genes with immunomodulatory potential. These include galectin-1
and Th17 CD4?cells. Although many cell types at the maternal–
fetal interface also produce gal1, its production by dNKs has been
used here to study its function in pregnancy. Media conditioned by
dNKs containing gal1 induced apoptosis of activated T cells. This
effect was blocked by anti-gal1 antibodies. Decidual T (dT) cells but
not peripheral T (pT) cells bound gal1 and presented a distinct
glycophenotype compatible with sensitivity to gal1. Annexin V
staining, TUNEL, and hypodiploidy showed a substantial propor-
tion of apoptotic dT cells. Immunohistochemistry revealed wide-
that colocalized with dNKs. Thus, secretion of gal1 by dNKs and
other decidual cells contributes to the generation of an immune-
privileged environment at the maternal–fetal interface.
maternal–fetal tolerance ? NK cells ? lectin ? decidua
maternal immune system during successful pregnancies (1).
During the first trimester of gestation, maternal immune cells
account for 30% of human decidual cells (2). This infiltrate is
abundant in natural killer cells (?80%) and relatively scarce in
CD3?T cells (?10%). In contrast, NK cells and CD3?T cells
represent about 10% and 80% of lymphocytes, respectively, in
peripheral blood (2–5). Various mechanisms to suppress poten-
tially alloreactive placental T cells and tolerize the maternal
immune system toward the fetus have been proposed in mice and
humans. These include placental expression of Fas and Fas
ligand (6, 7), T cell starvation through tryptophan depletion by
indolamine 2,3-dioxygenase (8), ligation of the inhibitory ligand
polarization (10), suppression by CD4?CD25?regulatory T cells
(11, 12), and progesterone-induced Th2-type responses (13). A
number of redundant immunosuppressive systems exist at the
maternal–fetal interface, not very surprisingly in a system so
essential for species survival.
Human decidual NK cells (dNKs) are phenotypically distinct
from peripheral blood NK cells (pNKs) (4). They play a role in
placental vasculature remodeling (14, 15). These CD56bright
dNKs overexpress many genes compared with CD56brightand
CD56dimpNK subsets (4), and display reduced cytotoxic activity
(16). One of the genes most up-regulated in dNKs encodes
galectin-1 (gal1) (4). Gal1 is a 14-kDa secreted protein in the
galectin family, whose members have 1 or 2 carbohydrate
recognition domains with affinity for poly-N-acetyllactosamine
(?-1,4-galactosyl-N-acetylglucosamine) moieties in cell surface
glycoproteins (17). The immunosuppressive properties of gal1
have been studied extensively, with particular attention to its
effects on T cells. Gal1 induces apoptosis of activated CD8 T
uman pregnancy is an immunological paradox. Despite
expressing paternal antigens, the fetus is not rejected by the
cells (18) and induces Th2 cytokine shifts. Expression of C2GnT
(core 2 ?-1,6-N-acetylglucosaminyltransferase) by T cells, which
initiates a second branch on O-glycans, later elongated with
poly-N-acetyllactosamine moieties, renders activated T cells
susceptible to gal1-induced apoptosis (19, 20). Differential gly-
cosylation of Th1, Th2, and Th17 CD4 T cells results in differ-
ential sensitivity to gal1-induced apoptosis. Th1 and Th17 cells
present complex N-glycans with poly-N-acetyllactosamine moi-
eties that make them susceptible to gal1 (21). The ?-2,6-
sialylation of poly-N-acetyllactosamine moieties by ST6Gal1
interferes with gal1 binding (22), protecting Th2 cells from
gal1-induced cell death (21). Many other effects of gal1 on cell
adhesion, T cell proliferation, monocytes, and neutrophils have
been reported (17, 23–26).
Fifteen galectins have been described in mammals, of which
gal1, gal3, gal9, and gal13 occur in the human placenta (4,
27–31). Cycling endometrium produces gal1 during the secretory
phase, and its expression is augmented during early pregnancy.
It is expressed by various cells in the human placenta, including
placental stromal cells, CD45?cells (29), cytotrophoblasts in cell
columns, and syncytiotrophoblasts (28). Expression levels are
reduced in patients with early pregnancy loss (27), suggesting
that gal1 may play a role in maintaining pregnancy.
Given the abundance of NK cells compared with T cells in the
human decidua (2–5), the overexpression of gal1 by dNKs at the
transcriptional level (4), and the abundance of gal1 produced by
several cell types in the human placenta (28, 29), we explored the
immunosuppressive role of gal1 in protecting the fetus from po-
tentially alloreactive T cells in the human decidua. The data
presented here suggest a critical role for gal1 in human maternal–
Human dNKs and Decidual Macrophages Express gal1. Gene expres-
sion profiles showed that gal1 mRNA is overexpressed by dNKs
compared with CD56bright(20 times) and CD56dim(5–10 times)
pNKs (4). Differences in gal1 expression also were found at the
protein level by Western blot analysis (Fig. 1A). Western blot
analysis and RT-PCR also showed gal1 expression by decidual
macrophages, the second most populous leukocyte in the de-
cidua (data not shown). Although many cells in addition to dNKs
in the human decidua produce gal1, its secretion by dNKs is used
here to study further its effects in human pregnancy.
Conditioned media (CM) from dNKs or CD3?CD56?pNKs
performed research; H.D.K., F.R., Y.L., D.S.J.A., J.L.K., and J.L.S. analyzed data; and H.D.K.,
F.R., and J.L.S. wrote the paper.
The authors declare no conflict of interest.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
November 25, 2008 ?
vol. 105 ?
no. 47 www.pnas.org?cgi?doi?10.1073?pnas.0809233105
were analyzed by sandwich ELISA assays. After 60 h in culture
(with IL-15), 4 ?g/mL gal1 was detected in CM from dNKs but
not from pNKs (Fig. 1B). After 90 h, gal1 also was detected in
not ex vivo pNKs also secreted gal1 in the absence of IL-15 in
18-h cultures (data not shown). The cross-reactivity of the
antibody used in the ELISA assays was tested against 2 other
galectins present in the human placenta that were available to us
in recombinant form. The antibody cross-reacted slightly with
recombinant human galectin-3 (gal3) at high protein concentra-
tions, but not with recombinant human galectin-9 (gal9) by
ELISA (Fig. 1C). However, only gal1 was immunoprecipitated
from CM from dNKs. Anti-gal1 mAb but not anti-gal3 mAb
produced bands in Western blots of the immunoprecipitate (Fig.
1D), confirming that dNKs secrete gal1.
Media Conditioned by Human dNKs Have Apoptotic Activity on Acti-
vated T Cells.Gal1hasapoptoticactivityonactivatedTcells.Both
gal1 and CM from dNKs induced apoptosis of MOLT4 T cells,
a gal1-susceptible T cell tumor line. Apoptosis induction was
blocked in the presence of 50 mM lactose, which competes with
gal1 for binding to poly-N- acetyllactosamines (18) (Fig. 2A), and
anti-gal1 antibodies (Fig. 2B), indicating that gal1 is the major
contributor to the apoptotic activity of CM from dNKs. Similar
results were obtained using CD3?T cells activated for 5 days
with phytohemagglutinin (PHA) and IL-2.
Decidual and Peripheral Human T Cells Have Different Glycopheno-
types and gal1-Binding Capacity. Secretion of gal1 in the decidua
by dNKs and other cells (28, 29) might contribute to an immu-
nosuppressive microenvironment at the maternal–fetal inter-
face, thereby protecting fetal cells from activated alloreactive
maternal T cells. The gal1-binding capacity of decidual T cells
(dTs) and peripheral blood T cells (pTs) was therefore explored
using biotinylated gal1. CD3?dTs, but not pTs from pregnant
and nonpregnant female donors, bound biotinylated gal1, and
binding was glycan-dependent because it could be competed
with 50 mM lactose (Fig. 3B and statistics in Fig. 3D).
Gal1 binds poly-N-acetyllactosamine moieties on N-glycans or
core 2 O-glycans. On N-glycans, it binds to unsialylated or
?-2,3-sialylated poly-N-acetyllactosamine moieties. The latter
can be detected with the lectin Maackia amurensis agglutinin
(MALII) (21). The ?-2,6-sialylation of N-acetyllactosamines on
N-glycans reduces gal1 binding and can be identified with the
lectin Sambucus nigra agglutinin (SNA) (Fig. 3A). Both dTs and
pTs bound biotinylated SNA equally and did not bind MALII
(Fig. 3C), suggesting that these cells do not bind gal1 through
On O-glycans, the lectin peanut agglutinin (PNA) recognizes
asialo core 1 O-glycans, the substrate on which the enzyme
blot of fresh CD56dimpNK, CD56brightpNK, and dNK lysates developed with
anti-gal1 mAb (Lower) and anti-?-actin mAb (Upper). Gal1 was visualized as a
analyzed. Similar results were obtained with decidual macrophages (data not
dNKs and CD56?CD3?pNKs for 60 h (filled bars) and 90 h (open bars).
Incubations were done in the presence of 12 ng/mL IL-15. The average of 3
B with recombinant human gal1 (diamonds), gal3 (squares), and gal9 (trian-
by dNKs (CM), control media not exposed to dNKs (M), recombinant human
gal1 (gal1), and recombinant human gal3 (gal3) with the same antibody as
used in B (anti-gal1) or with an isotype control (Ig). Immunoprecipitates were
resolved by Western blotting with anti-gal1 (Left) or anti-gal3 (Right) mAbs.
The additional bands are derived from immunoglobulins and protein G. One
experiment of 3 is shown.
Gal1 is expressed and secreted by dNKs but not by pNKs. (A) Western
v i t a l e
l l e
medium + Lactose
gal-1 + lactosa
medium + rabbit serum
medium + anti-gal-1
v i t a l e
l l e
v i t a l e
l l e
PHA T cell blasts
activity on T cells and MOLT4 cells. (A) Annexin V staining of MOLT4 cells
treated (open histogram) with gal1 (Left) or medium conditioned by dNKs
(Right) in the presence (dotted line histogram) or absence (solid line histo-
gram) of 50 mM lactose. Shaded histograms correspond to cells incubated
with control medium. Incubations (30 min at 37°C) were done in the presence
induction of MOLT4 cells (Left) and peripheral T cell blasts (Right) by dNK
conditioned medium with anti-gal1 polyclonal antibodies. Control medium
(shaded histogram), dNK conditioned medium (solid line histogram), condi-
tioned medium plus nonimmune rabbit serum (dotted line histogram), and
conditioned medium plus anti-gal1 rabbit serum (dashed line histogram) are
Medium conditioned by dNKs and containing gal1 has apoptotic
Kopcow et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
elongated with poly-N-acetyllactosamine moieties, to which gal1
binds. The dTs bound significantly more PNA than the pTs (Fig.
3C and statistics in Fig. 3E). Furthermore, only dTs expressed
C2GnT mRNA (Fig. 3F) and had core 2 O-glycosylated CD43,
stained with antibody 1D4 (Fig. 3C). These differences in
binding capacity of the 2 cell types.
A Population of Apoptotic T Cells Is Present in the Human Decidua.
studied. Staining freshly isolated decidual lymphocytes with
anti-CD56 or anti-CD3 mAbs and Annexin V revealed that
CD3?dTs (Fig. 4A Center) but not the majority of CD56?dNKs
was detected on CD3?T cells from peripheral blood lymphocyte
preparations exposed to an isolation procedure similar to that
used to obtain decidual lymphocytes (Fig. 4A Right). Statistical
comparison showed that differences in the percentage of An-
nexin V-positive CD3?T cells were highly significant (P ? 6.7 ?
10?6) (Fig. 4A?). DNA fragmentation and hypodiploidy, 2 late
apoptosis events, were evaluated by TUNEL and propidium
iodide (PI) staining, respectively. TUNEL revealed that up to
60% of CD3?dTs were apoptotic (Fig. 4 B and B?). However,
total decidual lymphocytes, most of which are NK cells, were
negative for TUNEL labeling (Fig. 4B Lower Right). Analysis of
CD3?pTs, isolated in the same fashion as dTs by anti-CD3
FACS sorting, showed that on average, only 10% of CD3?pTs
were positive for TUNEL staining, in contrast to 45% of CD3?
dTs (P ? 0.0016) (Fig. 4B?). Statistically significant differences
also were observed in TUNEL analyses of dTs and pTs isolated
as CD14?, CD56?, and CD4?/CD8?lymphocytes to avoid
B?). PI staining showed similar results (Fig. 4 C and C?).
Human dTs Form Periglandular Apoptotic Foci. Anti-CD3 and
TUNEL staining of serial sections of decidual tissue revealed
that CD3?dTs formed periglandular foci (Fig. 5 A and B) that
colocalized with foci of apoptotic lymphocytes identified by
TUNEL staining (Fig. 5 A? and B?). Away from these dense
periglandular foci, interstitial CD3?cells were identified, and
apoptosis identified by TUNEL was seen only in scattered single
cells. Tissue samples analyzed were fixed immediately after
sample collection, ruling out the possibility of apoptosis induc-
tion due to tissue manipulations. Periglandular apoptotic foci
were also infiltrated by CD56?NK cells (Fig. 5B??). Efforts to
costain single tissue sections for TUNEL and CD3 were unsuc-
cessful because of high nonspecific background generated by the
combination of the staining techniques. Nevertheless, the pat-
tern of TUNEL staining correlates with CD3 staining in serial
sections, and the histochemical data combined with flow cyto-
metric analysis of Annexin V staining, TUNEL, and hypodip-
loidy confirm the apoptotic nature of a major population of dTs
but not of dNKs.
In some histological sections, T cell aggregates with relatively
low levels of apoptosis were noted. This may reflect that not all
dTs are apoptotic, in agreement with the finding that a major
proportion of but not all dTs were apoptotic in Annexin V
stainings, TUNEL, and hypodiploidy analyses (Fig. 4).
core 2 O-glycans indicating the structures recognized by lectins SNA, PNA, MALII, and gal1. Crossed structures are absent from dTs and pTs based on data shown
2 O-glycans are recognized by gal1 on dT cells (in bold). (B) Binding of biotinylated gal1 to dTs (Right) and pTs from pregnant (Center) and nonpregnant (Left)
female donors in the presence (dashed line histograms) or absence (solid line histograms) of 50 mM lactose. (C) Binding of biotinylated SNA, MALII, and PNA
cells. Bound biotinylated lectins were detected by staining with fluorescently labeled streptavidin. Gray shaded histograms correspond to CD3?gated cell
preparations incubated only with fluorescently labeled streptavidin. Statistics comparing the staining with biotinylated gal1 (D) and biotinylated lectins SNA,
PNA, and MALII (E) in dTs and pTs as described in B and C are shown. Asterisks indicate statistical significance in t test comparisons involving the groups denoted
by the overlying lines (*, P ? 0.05). Results correspond to 8 decidual and 4 peripheral blood samples. Error bars represent standard error. (F) Expression of C2GnT
by dTs and pTs evaluated by RT-PCR.
Differential gal1 binding and glycophenotype of decidual and peripheral blood CD3?T cells. (A) Biosynthetic pathways of tetraantenary N-glycans and
www.pnas.org?cgi?doi?10.1073?pnas.0809233105 Kopcow et al.
Staining of decidual sections revealed widespread expression
of gal1 in cells with different morphology (Fig. 5 B???), indicating
that many cell types in addition to dNKs may contribute to the
generation of an immunosuppressive environment through gal1
Various immunosuppressive mechanisms have been proposed to
protect the fetus from potentially alloreactive T cells (6–13).
Here, we described a novel mechanism likely involved in the
induction of apoptosis of dTs in the human placenta mediated by
gal1. The dTs express CD69, and ?50% are HLA-DR?, indi-
cating an activated phenotype (2). Gal1 has the capacity to
induce apoptosis of activated T cells (18), and dTs have the
capacity to bind gal1 (Fig. 3B). Furthermore the glycophenotype
of dTs, distinct from that of pTs, is compatible with their
activated profile, differentially binding PNA, expressing C2GnT,
and presenting core 2 O-glycans, and suggests that gal1 binds
these cells through O-glycans (Fig. 3).
In serial sections from early human placentas, CD3?T cells
formed periglandular foci (Fig. 5) that colocalized with the foci
of TUNEL-positive apoptotic lymphocytes. The combined anal-
ysis of immunohistochemical sections and flow cytometric anal-
yses of Annexin V, PI, and TUNEL staining support the
presence of apoptotic dTs and nonapoptotic dNKs at this site.
Scattered interstitial T cells also were present, some of which
were nonapoptotic, and a few of which were apoptotic. Many
other decidual cells also expressed gal1, contributing to the
generation of a local immunosuppressive environment.
Media conditioned by dNKs contained gal1 at a lower con-
centration (1–4 ?g/mL; Fig. 1B) than that of recombinant gal1
used to induce T cell apoptosis. It is possible that other proteins
secreted by dNKs could synergize with the apoptotic effect of
gal1 secreted by dNKs. PP14, a glycoprotein overexpressed by
dNKs (4) that shares immunosuppressive properties with gal1, is
n i x
s l l e
n l l e
v i t a l e
s l l e
s l l e
c d i o l p i d
n l l e
v i t a l e
Center) staining of freshly isolated decidual lymphocytes, and peripheral blood
cytes positive for Annexin V staining. (B and C) BrdU staining for TUNEL analysis
decidual lymphocytes. Numbers indicate the percentage of cells with DNA frag-
cell preparations. Peripheral T cells shown were obtained from a single prepara-
tion. (A?, B?, and C?) Statistics comparing the abundance of Annexin V?(A?),
TUNEL?(B?), and hypodiploid cells (C?) in CD3?or CD4?/8?T cells preparations
from decidual and peripheral blood lymphocytes analyzed as described in A, B,
groups denoted by the overlying lines (**, P ? 0.01;*, P ? 0.05). Numbers in
parentheses indicate the number of samples analyzed. Error bars represent
Freshly isolated decidual lymphocytes contain a population of Annexin
staining (A? and B?) of 4-?m serial histological sections (A with A?; and B with
B?, B??, and B???) of first-trimester human decidua of 6 weeks’ gestational age.
Images are representative of 2 samples from 2 different donors. EG, endome-
trial gland; D, decidua.
Infiltrating T cells form periglandular apoptotic foci in the human
Kopcow et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
a candidate for such an interaction. Like gal1, PP14 also induces
T cell apoptosis (32), colocalizes with CD45 on the cells to which
it binds (33), and reacts with N-acetyllactosamines (34). Inter-
estingly, PP14 is also expressed by glandular epithelium around
which apoptotic T cells are seen, and has poly-N-acetyllac-
tosamine moieties (35) to which gal1 could potentially bind. An
interesting difference between the 2 proteins is that whereas
?-2,6-sialylation of CD45 glycans on T cells negatively regulates
gal1-induced apoptosis, it favors the activity of PP14 (22, 34).
The presence of both proteins at the maternal–fetal interface
may modulate a broad range of T cells.
Annexin V staining revealed the externalization of phospha-
tidylserine (PS) due to loss of membrane asymmetry during early
apoptosis. DNA fragmentation, as evidenced by TUNEL, and
hypodiploidy are late apoptosis events. It has been proposed that
gal1-induced PS exposure can contribute to leukocyte ho-
meostasis by phagocytic recognition (36). The lower percentage
of TUNEL-positive or hypodiploid decidual T cells compared
with Annexin V-positive decidual T cells (Fig. 4) could result
DNA fragmentation and hypodiploidy can be detected.
To establish whether gal1 is necessary for maintaining a
successful pregnancy, animal models are required. Mice that are
gal1?/?breed normally (37), but a syngeneic mating places little
immunological stress on pregnancy. Experiments comparing
syngeneic and allogeneic matings [CBA/Caj (H2k), C3H/j (H2k),
and Balb/cJ (H2d) males] failed to show differences in the litter
size or embryo resorption rates of pregnant C57BL/6 (H2b)
gal1?/?females (data not shown). However, it was reported
recently that gal1?/?129P3/J females (H2b) mated with alloge-
neic DBA/J2 males (H2d) present greater fetal resorption rates
than wild-type 129P3/J females, whereas no differences were
observed in syngeneic matings (38), thus supporting an impor-
tant role for gal1 in maternal–fetal immune tolerance. The fact
that augmented embryo resorption rates in gal1?/?pregnant
females are evident only in matings with certain strain combi-
nations is indicative of the presence of redundant systems to
prevent fetal rejection. This redundancy may be important only
in genetic backgrounds that pose an increased immunological
stress. As multiple galectins are present in the human decidua
(28–31), multiple galectins present in the murine placenta (39)
could, conceivably, compensate for each other.
Particular attention has been paid recently to the role of
regulatory T cells in maternal–fetal tolerance (11, 12). Interest-
ingly, CD4?CD25?regulatory T cells have been shown to
mediate their suppressive activity through gal1 (40).
The periglandular localization of apoptotic dTs is intriguing,
given that glandular epithelial cells in the decidua basalis are the
frontier to the intervillous space. Furthermore, it has been
reported that some uterine glandular epithelial cells have MHC
class II expression (41) and antigen-presenting capacity (42).
Speculatively, presentation of fetal antigens by these cells in a
galectin-rich environment could promote the survival of fetal
antigen-specific tolerogenic T cells and the death of activated
antigen-specific Th1, Th17, and reactive CD8 T cells.
In conclusion, human dTs are shown here to have a particular
glycophenotype that gives them the capacity to bind gal1. A
significant proportion of those cells are apoptotic and colocalize
with NK cells that, among other decidual cells, secrete apopto-
tically active gal1. These data, together with recent findings on
the role of gal1 in murine pregnancies, point to an important role
for gal1 in human maternal–fetal tolerance.
obtained from discarded material of elective first-trimester terminations and
processed as described previously (4). The dNKs were FACS sorted as
CD56brightCD16?CD3?cells, and dTs as CD3?or as CD4?/CD8?CD14?CD56?
cells using a mixture of anti-CD4 and anti-CD8 mAbs conjugated to the same
Peripheral lymphocytes were isolated from leukopacks obtained from
anonymous donors (Massachusetts General Hospital, Boston, MA) using den-
sity gradients (Ficoll-Hypaque; Amersham Biosciences) directly or, for NK cell
preparations, after NK cell enrichment using Rosettesep (StemCell Technolo-
gies Inc.) following the manufacturer’s instructions. For generation of pNK
CM, cells were further isolated by FACS sorting as CD56?CD3?lymphocytes.
For Western blotting, CD56brightpNKs were isolated as CD56brightCD16?CD3?
cells, and CD56dimpNKs as CD56dimCD16?CD3?cells.
Mock isolation of peripheral lymphocytes as if they were decidual lympho-
cytes was done by exposing peripheral lymphocytes, after density gradient
separation, to the same procedure used to obtain decidual cells (4), including
DNA content analyses, pTs were further isolated by FACS sorting by using the
same staining scheme as described above for dTs.
Media Conditioned by NK Cells. FACS-sorted dNKs or CD56?pNKs were
FCS for 60 or 90 h in the presence of 12 ng/mL IL-15, or for 18 h without IL-15.
presence of IL-15 were used for induction of T cell apoptosis.
ELISA Assays. Inmunolon 2 ELISA plates (DynatechLaboratories Inc.) were
covered with polyclonal anti-gal1 antibodies, 10 ?g/mL (Peprotech Inc.), and
washed with washing buffer (0.05% Tween-20 in PBS). Plates were blocked
with 1% BSA in PBS. Individual wells were incubated with serial dilutions of
(0.3 ?g/mL in PBS 0.1% BSA; Peprotech Inc.). Plates were washed 6 times and
incubated with HRP-conjugated streptavidin (Zymed). After 6 additional
washes, a colorimetric reaction was developed with TMB substrate (Pierce).
The reaction was stopped by adding one volume of 2 N H2SO4. Absorbance at
450 nm was recorded.
with polyclonal anti-human gal1 antibody (10 ?g/mL; Peprotech) or control
IgG. A total of 20 ?L coated beads was added to 100 ?L dNK CM. Beads were
collected 12 h later by centrifugation. After 4 washes with PBS, SDS loading
buffer was added, and samples were boiled. Beads were precipitated by
centrifugation, and the bead-free supernatant was analyzed by Western
blotting with anti-gal1 mAb (NCL Gal1; Novocastra) or anti-gal3 mAb (clone
9C4; Lab Vision).
Apoptosis Induction by Recombinant gal1 and Media Conditioned by dNKs.
MOLT4 cells or peripheral blood lymphocytes activated for 5 days in RPMI
media containing 10% FCS plus 2 ?g/mL PHA and 200 U/mL IL-2 were incu-
bated at 37°C with fresh dNK CM or recombinant gal1 (13 ?M). Incubations
were done in the presence of 0.5 mM DTT with or without the addition of 50
mM lactose, anti-gal1 rabbit serum, or nonimmune rabbit serum. Apoptosis
induction was interrupted at 45 min of incubation by the addition of lactose
over untreated CD3?control cells was calculated by Overton histogram accu-
mulative substraction (43) with the software Flow Jo (Treestar).
TUNEL and Hypodiploidy Assays. Total decidual lymphocytes, peripheral blood
lymphocytes, or FACS-sorted dTs were processed with the APO-BRDU apopto-
samples were also stained with propidium iodide (PI) to quantitate DNA
ploidy. Fluorescein-Br-dUTP incorporation and PI staining were measured in a
FACScalibur flow cytometer (Becton–Dickinson).
Biotinylation of gal1 and Lectin Staining. Recombinant gal1 was biotinylated
by incubation with Sulfo-NHS-Biotin (Pierce) according to the manufacturer’s
instructions. Unconjugated biotin was removed by dialysis. Cells were incu-
bated with biotinylated gal1 (15 ?M), biotinylated SNA (20 ?g/mL; E-Y Lab-
(20 ?g/mL; Vector Laboratories) (21). Bound biotinylated lectins were de-
tected with phycoerithrin-conjugated streptavidin (Sigma–Aldrich). Nonspe-
cific binding was determined with phycoerithrin-conjugated streptavidin
www.pnas.org?cgi?doi?10.1073?pnas.0809233105 Kopcow et al.
Antibodies, Lectins, and Recombinant Proteins. Recombinant human gal1 was Download full-text
kindly provided by Linda Baum (University of California, Los Angeles, CA) and
Gabriel Rabinovich (Instituto de Biología y Medicina Experimental, Consejo
Nacional de Investigaciones Científicas y Técnicas, Buenos Aires, Argentina);
was from Peprotech Inc.
The following murine mAb and the proteins conjugated with FITC, PE,
CyChrome, were used in FACS analysis or sorting: anti-CD3, anti-CD56, anti-
CD16, anti-CD4, anti-CD8, anti-CD14, IgG isotype controls, and Annexin V
were all from BD PharMingen. Anti-core 2 O-glycosylated CD43 antibody 1D4
was from MBL International. Rabbit anti-gal1 antibody used in immuohisto-
chemistry was provided by Gabriel Rabinovich (44). Blocking anti-gal1 rabbit
serum was kindly provided by Linda Baum (45). This serum has no cross-
with gal9 in Western blot analysis (data not shown).
Immunohistochemistry, Western blotting, and RT-PCR are described in
detail in supporting information (SI) Methods.
ACKNOWLEDGMENTS. We thank Linda Baum and Gabriel Rabinovich, who
kindly provided recombinant gal1 and anti-gal1 immune serum. This research
was funded by National Institutes of Health Grant AI053330 to J.L.S.
1. Medawar PB (1953) Some immunological and endocrine problems raised by evolution
of viviparity in vertebrates. Symp Soc Exp Biol 7:320–338.
2. Loke YW, King A (1955) Human Implantation: Cell Biology and Immunology (Cam-
bridge Univ Press, Cambridge, UK).
3. Cooper MA, Fehniger TA, Caligiuri MA (2001) The biology of human natural killer-cell
subsets. Trends Immunol 22:633–640.
4. Koopman LA, et al. (2003) Human decidual natural killer cells are a unique NK cell
subset with immunomodulatory potential. J Exp Med 198:1201–1212.
5. Erlebacher A, Vencato D, Price KA, Zhang D, Glimcher LH (2007) Constraints in antigen
presentation severely restrict T cell recognition of the allogeneic fetus. J Clin Invest
6. Pearson H (2002) Reproductive immunology: Immunity’s pregnant pause. Nature
7. Vacchio MS, Hodes RJ (2005) Fetal expression of Fas ligand is necessary and sufficient
for induction of CD8 T cell tolerance to the fetal antigen H-Y during pregnancy.
J Immunol 174:4657–4661.
8. Munn DH, et al. (1998) Prevention of allogeneic fetal rejection by tryptophan catab-
olism. Science 281:1191–1193.
tolerance. J Exp Med 202:231–237.
10. Piccinni MP, et al. (1998) Defective production of both leukemia inhibitory factor and
type 2 T-helper cytokines by decidual T cells in unexplained recurrent abortions. Nat
11. Aluvihare VR, Kallikourdis M, Betz AG (2004) Regulatory T cells mediate maternal
tolerance to the fetus. Nat Immunol 5:266–271.
12. Tilburgs T, et al. (2008) Evidence for a selective migration of fetus-specific
CD4?CD25brightregulatory T cells from the peripheral blood to the decidua in human
pregnancy. J Immunol 180:5737–5745.
during pregnancy: Endocrine-immune cross talk in mammalian species and the role of
stress. Am J Reprod Immunol 58:268–279.
14. Guimond MJ, et al. (1997) Absence of natural killer cells during murine pregnancy is
associated with reproductive compromise in TgE26 mice. Biol Reprod 56:169–179.
15. Hanna J, et al. (2006) Decidual NK cells regulate key developmental processes at the
human fetal-maternal interface. Nat Med 12:1065–1074.
16. Kopcow HD, et al. (2005) Human decidual NK cells form immature activating synapses
and are not cytotoxic. Proc Natl Acad Sci USA 102:15563–15568.
in tuning the immune response: Lessons from experimental models of inflammatory
disease, autoimmunity and cancer. Scand J Immunol 66:143–158.
18. Perillo NL, Pace KE, Seilhamer JJ, Baum LG (1995) Apoptosis of T cells mediated by
galectin-1. Nature 378:736–739.
expression of core 2 O-glycans. J Immunol 167:5697–5707.
20. Cabrera PV, et al. (2006) Haploinsufficiency of C2GnT-I glycosyltransferase renders T
lymphoma cells resistant to cell death. Blood 103:2399–2406.
selectively regulates susceptibility to cell death. Nat Immunol 8:825–834.
22. Amano M, Galvan M, He J, Baum LG (2003) The ST6Gal I sialyltransferase selectively
phosphatase modulation, and T cell death. J Biol Chem 278:7469–7475.
23. Correa SG, Sotomayor CE, Aoki MP, Maldonado CA, Rabinovich GA (2003) Opposite
effects of galectin-1 on alternative metabolic pathways of L-arginine in resident,
inflammatory, and activated macrophages. Glycobiology 13:119–128.
24. Karmakar S, Cummings RD, McEver RP (2005) Contributions of Ca2?to galectin-1-
induced exposure of phosphatidylserine on activated neutrophils. J Biol Chem
25. Lowe JB (2001) Glycosylation, immunity, and autoimmunity. Cell 104:809–812.
26. Rubinstein N, et al. (2004) Targeted inhibition of galectin-1 gene expression in tumor
cells results in heightened T cell-mediated rejection: A potential mechanism of tumor-
immune privilege. Cancer Cell 5:241–251.
27. Liu AX, et al. (2006) Proteomic analysis on the alteration of protein expression in the
placental villous tissue of early pregnancy loss. Biol Reprod 75:414–420.
28. Maquoi E, van den Bru ˆle FA, Castronovo V, Foidart JM (1997) Changes in the distribu-
tion pattern of galectin-1 and galectin-3 in human placenta correlates with the
differentiation pathways of trophoblasts. Placenta 18:433–439.
30. Popovici RM, Krause MS, Germeyer A, Strowitzki T, von Wolff M (2005) Galectin-9: A
in humans. J Clin Endocrinol Metab 90:6170–6176.
31. Than NG, et al. (2004) Functional analyses of placental protein 13/galectin-13. Eur
J Biochem 271:1065–1078.
32. Mukhopadhyay D, Sundereshan S, Rao C, Karande AA (2001) Placental protein 14
induces apoptosis in T cells but not in monocytes. J Biol Chem 276:28268–28273.
of T cell activation by placental protein 14 is mediated by the tyrosine phosphatase
receptor CD45. J Biol Chem 278:14059–14065.
34. Ish-Shalom E, et al. (2006) alpha 2,6-Sialylation promotes binding of placental protein
14 via its Ca2?-dependent lectin activity: Insights into differential effects on CD45RO
and CD45RA T cells. Glycobiology 16:173–183.
a human glycoprotein with potent immunosuppressive and contraceptive activities.
J Biol Chem 270:24116–24126.
36. Dias-Baruffi, et al. (2003) Dimeric galectin-1 induces surface exposure of phosphati-
dylserine and phagocytic recognition of leukocytes without inducing apoptosis. J Biol
37. Poirier F, Robertson EJ (1993) Normal development of mice carrying a null mutation in
the gene encoding the L14 S-type lectin. Development 119:1229–1236.
38. Blois SM, et al. (2007) A pivotal role for galectin-1 in fetomaternal tolerance. Nat Med
39. Phillips B, et al. (1996) Differential expression of two beta-galactoside-binding lectins
in the reproductive tracts of pregnant mice. Biol Reprod 55:548–558.
T cells. Blood 109:2058–2065.
41. Bulmer JN, Johnson PM (1985) Immunohistological characterization of the decidual
leucocyticinfiltrate related to endometrial gland epithelium in early human preg-
nancy. Immunology 55:35–44.
42. Wallace P, et al. (2001) MHC class II expression and antigen presentation by human
endometrial cells. J Steroid Biochem Mol Biol 76:203–211.
43. Overton WR, (1988) Modified histogram subtraction technique for analysis of flow
cytometry data. Cytometry 9:7.
44. Rodig SJ, et al. (2008) AP1-dependent galectin-1 expression delineates classic hodgkin
and anaplastic large cell lymphomas from other lymphoid malignancies with shared
molecular features. Clin Cancer Res 14:3338–3344.
45. He J, Baum LG (2004) Presentation of galectin-1 by extracellular matrix triggers T cell
death. J Biol Chem 279:4705–4712.
Kopcow et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?