REVIEWS

Immunology of placentation
in eutherian mammals
Ashley Moffett and Charlie Loke
Abstract | The traditional way to study the immunology of pregnancy follows the
classical transplantation model, which views the fetus as an allograft. A more recent
approach, which is the subject of this Review, focuses on the unique, local uterine
immune response to the implanting placenta. This approach requires knowledge of
placental structure and its variations in different species, as this greatly affects the type
of immune response that is generated by the mother. At the implantation site, cells from
the mother and the fetus intermingle during pregnancy. Unravelling what happens here
is crucial to our understanding of why some human pregnancies are successful whereas
others are not.

Trophoblast cells
Trophoblast cells are the
earliest extra-embryonic cells
to differentiate from the cells of
the mammalian embryo. They
surround the conceptus
throughout gestation and are
in direct contact with maternal
tissues.
Blastocyst
After fertilization, the potential
embryo undergoes mitotic
division and, at the 128-cell
stage in humans, two distinct
cell lineages are present.
Trophoblast cells are derived
from the trophectoderm that
surrounds the blastocyst and
the inner cell mass gives rise
to the embryo.
King’s College, Cambridge
CB2 1ST, UK.
Correspondence to A.M.
e-mail: am485@cam.ac.uk
doi:10.1038/nri1897
For immunologists, ruminations about the immune sys-
tem during pregnancy are mostly centred on the acqui-
sition of maternal tolerance to the allogeneic fetus1,2.
This view is probably too simplistic because it does not
take into consideration the anatomical fact that it is the
maternal relationship with the placenta rather than with
the fetus that holds the key to our understanding of the
‘immunological paradox’ of pregnancy. In particular,
the focus should be on the intermingling of placental and
maternal cells in the uterine wall, as this is where direct
tissue contact occurs during placentation. Failure to dis-
tinguish between the local uterine immune response to
the placenta and the systemic immune response to fetal
cells (which usually cross to the mother at delivery) has
led to a great deal of confusion.
To understand maternal uterine immune responses
to the placenta requires knowledge of the sequen-
tial anatomical and physiological events that occur
during placentation. Herein lies a difficulty, in that
each species has developed its own strategy and this
results in a great divergence of types of placentation
in mammals3–5. One of the most obvious differences
is the extent of invasion into the uterus by placental
trophoblast cells. This can range from no invasion at all
(as in epitheliochorial placentation) to very extensive
invasion (as in haemochorial placentation), whereby
trophoblast cells penetrate through uterine blood ves-
sels to come into direct contact with maternal blood.
Humans have haemochorial placentae, as do many
laboratory animals, such as mice, rats, guinea pigs and
rabbits, but even among this group, the human placenta
is particularly invasive.
Comparison of divergent placental strategies must
also encompass the maternal reaction that each placental
type evokes. Here there is also much diversity. In haemo-
chorial placentation, the uterine mucosa is transformed
into a highly specialized tissue known as the decidua
(a process referred to as decidualization). This does
not occur in species with non-invasive epitheliochorial
placentae. In primates, decidualization correlates closely
with the degree of invasion, so the most marked decidual
change is seen in those species with the most invasive
placentae. A conspicuous feature of the decidua is the
influx of a distinctive lymphocyte population of maternal
uterine natural killer (NK) cells6. NK cells are emerging
as important players in the uterine immune response to
invasive forms of placentation, although the precise role
they have is still unclear.
The cells that define the boundary between the
mother and fetus are trophoblast cells7. These cells are
derived from the outer layer of the blastocyst and have
many unusual characteristics that tend to be ignored
by immunologists8 (BOX 1). Because trophoblast cells are
freed from the developmental constraints that affect the
rest of the embryo, they have a unique pattern of pater-
nal and maternal gene expression. Of most relevance to
immunologists is the expression of MHC and MHC-like
genes by trophoblast cells, which would be the poten-
tial ligands for immune receptors on uterine NK cells,
lymphocytes and myelomonocytic cells. Human tropho-
blast cells have been studied extensively and express a
unique and intriguing array of HLA class I molecules, the
functions of which might hold the key to the successful
temporary coexistence of two individuals.

584 | AUGUST 2006 | VOLUME 6 www.nature.com/reviews/immunol

© 2006 Nature Publishing Group
 REVIEWS

Eutherian placenta It is therefore clear that a detailed knowledge of the crucial modification from the anamniote to amniote egg
Eutherian mammals include all anatomical and molecular interactions between the pla- is the development of four extra-embryonic membranes,
mammalian species except centa and the uterus at the implantation site is necessary consisting of the yolk sac, amnion, chorion and allantois. The
marsupials and egg-laying if we are to understand nature’s allograft. The starting egg shell was subsequently lost during the evolution of
monotremes. The eutherian
placenta is well developed
point of this Review will be the eutherian placenta. The viviparous animals, but all amniote embryos retain these
compared with the marsupial evolution from the extra-embryonic membranes of extra-embryonic membranes. Only minor modifications
placenta and has a great amniote eggs to the formation of the definitive placenta were then required for the evolution of these into the
diversity of forms. is traced and the diverse characteristics of placentae that definitive placenta.
are seen in extant mammalian species is emphasized. Mammals can be divided into three subclasses
Amniote egg
The unique characteristics of trophoblast cells and the that became separated from the reptile-like mammals
Eggs of amniote vertebrates
provide an interface between
adaptation of the uterine mucosa by transformation into 120 million years ago, and these are known as the
the embryo and its immediate decidua are described. Also, the immunological implica- monotremes (for example, duckbill platypus), marsupi-
environment, therefore tions of these divergent placental forms are considered. als (for example, kangaroos) and eutherians (for exam-
allowing increased respiratory It is hoped that this approach will provide a more solid ple, humans). Monotremes are oviparous and the egg is
and excretory capacity as well
as nutrient provision.
framework on which to discuss the immunology of mam- retained in the oviduct until shortly before the young
malian reproduction, especially from the standpoint of hatches. The eggs of marsupials hatch in the oviduct at
Yolk sac the success or failure of human pregnancy. the 10-somite stage of development, when the embryo
The first of the four extra- implants briefly and superficially with a simple placenta.
embryonic membranes of Evolution of viviparity The eutherians have the most complex placental develop-
amniote eggs to form during
Viviparity (the bearing of live young) has evolved inde- ment. The main evolutionary change in mammalian
embryogenesis. It surrounds
the mass of yolk in reptile and
pendently many times in many groups of vertebrates, placental development was the emergence of trophoblast
bird eggs and is connected to including fish, reptiles and mammals9–11. The selective cells as a distinctive cell type from the outer epithelium
the midgut by the yolk stalk. pressures for viviparity include protection of offspring (chorion) of the amniote egg.
The yolk sac is also formed in from cold, from inhospitable environments and from
mammals, despite the absence
of yolk.
predators. The spectrum of viviparity seen today ranges Anatomy of placentation in eutherian mammals
from a mother simply holding yolky eggs in her body until Bringing order to the seeming chaos of placental
Amnion they hatch (ovoviviparity) to the development of a complex diversity is difficult, but for immunologists the most
The innermost membranous placenta that extracts nutrients from the mother. important consideration is the invasive potential of
sac of amniote eggs. It is filled The placenta is formed when fetal membranes become trophoblast cells in each species and how this is regu-
with a serous fluid and
closely attached to the uterine wall to facilitate physiologi- lated. Traditionally, the various complex types of placen-
encloses the embryo of an
amniote (reptile, bird or
cal exchange of gases, nutrients and waste products. The tation seen in eutherian mammals have been viewed as
mammal). first step in the emergence of placentation was the evolu- three simplified groups, based on the number of inter-
tion of the amniote egg, which was an important verte- vening cellular layers between the maternal and fetal
brate innovation10,11. This paved the way for the transition circulations13 (FIG. 1). Trophoblast cells are always the
from oviparity to viviparity and a shift from yolk-sac outermost layer of fetal cells that overlie an inner core
nutrition to nourishment delivered by the mother 4,12. The of mesenchyme and fetal capillaries. In epitheliochorial
placentation, the trophoblast cells can attach (and even
fuse with) the surface epithelium of the uterus but there
Box 1 | Characteristics and functions of trophoblast cells is no invasion by the trophoblast cells. Trophoblast-cell
Characteristics infiltration through the surface epithelium of the uterus
• Paternal X chromosome inactivation is characteristic of other placental forms. For example,
• Unmethylated DNA trophoblast cells can migrate to abut maternal blood
• Expression of endogenous retroviral products (such as syncytin) vessels (in endotheliochorial placentation). The most
• Expression of oncofetal proteins (such as carcinoembryonic antigen, α-fetoprotein invasive form is seen when trophoblast cells infiltrate
and human placental lactogen) through the maternal vessels to come into direct contact
• Formation of multinucleated cells by fusion or endoreduplication with maternal blood (in haemochorial placentation). In
• Lack of expression of MHC class II antigens and variable expression of MHC class I
this haemochorial form, trophoblast cells disrupt the
antigens endothelial cells and, in some cases, the muscle coat
(media) of the uterine arteries as well.
Functions Molecular phylogenetics has allowed placental struc-
• Anchoring the placenta to the uterine wall ture to be viewed in a new context, although some ques-
• Transport of nutrients and oxygen to the fetus tions still remain14. All extant eutherian mammals can
• Removal of waste products be grouped into four superorders (or clades): Afrotheria,
• Secretion of hormones and other placental proteins Xenarthra, Laurasiatheria and Euarchontoglires (or
• Physical barrier between maternal and fetal circulations Supraprimates)15. Although the relationship between the
• Site of contact between the maternal immune system and the conceptus
clades is still disputed, by studying retroposed elements,
the basic eutherian divergence was found to be between
• Transfer of maternal immunoglobulins to the fetus*
Xenarthra and the other clades16. All placentae examined
• Phagocytosis of red blood cells for acquisition of iron* so far from eutherian mammals of the Xenarthra clade
*Species-specific functions. For more details, see REFS 8,85. (such as armadillos and sloths) are either haemochorial or
endotheliochorial, whereas non-invasive epitheliochorial

NATURE REVIEWS | IMMUNOLOGY VOLUME 6 | AUGUST 2006 | 585

© 2006 Nature Publishing Group
REVIEWS
Chorion
In birds and reptiles, the
chorion adheres to the shell
and is highly vascularized to
function in gas exchange. In
mammals, it forms the fetal
contribution to the placenta,
made by an outer layer of
trophoblast cells and inner
layer of extra-embryonic
mesoderm, which contains
blood vessels that allow
exchange of materials with the
maternal circulation.
Allantois
The extra-embryonic
membrane that emerges as a
sac from the posterior part of
hindgut of the embryo. It fuses
with the chorion to form the
chorio-allantoic placenta. The
connection it makes between
the embryo and the placenta
becomes the umbilical cord.
Retroposed elements
Retroposons randomly insert
into the genomes with little
likelihood of the same element
integrating into the
orthologous position in
different species. Analysis of
the patterns of presence or
absence of retroposons is a
reliable method for studying
the evolutionary history of
organisms.
Convergent evolution
The process whereby
organisms that are not closely
related independently acquire
similar characteristics while
evolving in separate and
sometimes varying
ecosystems.
Haemolytic disease of
the newborn
If there is rhesus-blood-group
incompatibility between the
mother and her fetus,
the mother makes an antibody
response against fetal red
blood cells that access the
mother’s circulation at delivery.
These IgG antibodies cross the
placenta during a subsequent
pregnancy, which results in the
destruction of fetal red blood
cells, leading to haemolytic
disease of the new born.
Maternal and fetal
microchimerism
The presence of fetal cells in
the mother or maternal cells
in the fetus. Fetal or maternal
cells generally cross the
placenta at delivery and might
persist for many years.
586 | AUGUST 2006 | VOLUME 6
a Epitheliochorial b Endotheliochorial c Haemochorial
Fetal vessel Fetal vessel Fetal vessel
Fetal
mesenchyme Trophoblast cell Trophoblast cell Cytotrophoblast cell
Uterine Maternal Maternal blood
epithelium uterine vessel in intervillous Syncytiotrophoblast
space layer
Maternal
Maternal uterine vessel
uterine vessel Endometrium
Figure 1 | Types of placentation. Schematic representation of the three main types of placentation, showing the
relationship between the fetal trophoblast cells and maternal blood. a | Epitheliochorial. Trophoblast cells of the placenta
are in direct apposition with the surface epithelial cells of the uterus but there is no trophoblast-cell invasion beyond this
layer. b | Endotheliochorial. The uterine epithelium is breached and trophoblast cells are in direct contact with endothelial
cells of maternal uterine blood vessels. c | Haemochorial. Maternal uterine blood vessels are infiltrated by trophoblast cells
causing rupture and release of blood into the intervillous space. The outer layer of the chorionic villi (syncytiotrophoblast)
is now bathed in blood ‘like a mop in a bucket of blood’.
placentation (as is found in marsupials) is not seen17,18. It Maternal uterine response to placentation
is still uncertain what the primordial form of eutherian Concomitant with the marked degree of placental
placentation is and there are compelling arguments that diversity in different species, there is also variation
this was the endotheliochorial form14,17. However, a recent in the uterine response to placentation that correlates
phylogenetic analysis combined with morphological and closely with the extent of trophoblast-cell invasion.
molecular data indicates that the ancestral placenta was In epitheliochorial placentation, there are minimal
haemochorial and invasive19. changes in the stroma of the uterine mucosa during
The observation that haemochorial placentae are pregnancy, apart from local angiogenesis, which is
found in diverse species belonging to all four eutherian needed to increase the blood flow and deliver nutri-
superorders is consistent with convergent evolution and ents to the uterine surface. By contrast, haemochorial
the presence of strong selective pressures that favour placentation is characterized by two changes in the
this condition, presumably to provide the fetus with easy uterus, the differentiation of the endometrium into
access to nutrients directly from the maternal blood. decidua and the transformation of the uterine spiral
However, the disadvantage of this form of placentation arteries. In the two extremes of placental types, the
is that the mother and fetus are no longer separated by non-invasive epitheliochorial form and the invasive
an intact layer of epithelial cells and this allows exposure haemochorial form, it is obvious that the mechanisms
of the trophoblast cells to potential allogeneic immune for increasing the blood flow to the feto–placental
responses by the mother. Uterine immune responses unit are completely different. In epitheliochorial
must therefore allow the placenta access to maternal placentation, this is achieved by expansion of the size
supplies but at the same time prevent excessive invasion. of the vascular bed in the uterus by angiogenesis. By
In addition, the transfer of cells between the mother and contrast, in human haemochorial placentation, there
the fetus becomes more likely in haemochorial placen- is lowering of resistance in the vessels of the placental
tation. In humans, fetal cells invariably cross into the bed caused by modification of the walls of pre-existing
maternal circulation at birth and a maternal antibody arteries, resulting in increased low-pressure blood
response to incompatible red blood cell antigens such as flow22,23 (FIG. 2).
Rhesus can result in haemolytic disease of the newborn in The changes of decidualization involve all the cel-
subsequent pregnancies20. A long-term consequence of lular elements of the uterine mucosa and are most
this cellular deportation is microchimerism, in which fetal pronounced in humans. The most obvious features
cells persist in the mother for several decades. The pres- are enlargement of the uterine stromal cells and the
ence of increased numbers of fetal cells is associated with presence of a distinct lymphocyte population of uter-
diseases such as systemic sclerosis, giving rise to the idea ine NK cells6 (BOX 2). In all species, the hallmark of
that such diseases might have an alloimmune rather than the decidua might indeed be the presence of uterine
autoimmune pathogenesis21. Haemochorial placentation NK cells. Similar cells are not found in other tissues,
can be viewed as a trade-off between the risk of these and whenever decidual tissue is formed uterine NK
adverse immunological reactions and the need for an cells are present (even in ectopic locations such as in
efficient way of obtaining nutrients from the mother. endometriotic foci)24.
www.nature.com/reviews/immunol
© 2006 Nature Publishing Group
 REVIEWS

Reproductive failure in humans
In humans, it is clear that disruption of the normal balance
between the itinerant trophoblast cells and the uterine tis-
sues they colonize during placentation can result in vari-
ous clinical problems. These conditions give insight into
how the delineation of the territorial boundary between
two individuals is achieved. Because of the close corre-
lation between the invasion of trophoblast cells and the
extent of decidualization, it was argued that the decidual
tissue has a permissive influence that favours trophoblast-
cell invasion into the uterus25. The alternative view was
that the decidua provides a defensive riposte to the highly

a
Myometrium Chorion Amnion Amniotic cavity
Decidua parietalis

Cervical
canal

Allantoic vessels
in umbilical cord
Remnants
Placenta vascularized of yolk sac
by allantoic vessels
Decidua basalis
Radial artery Arcuate artery

Uterine artery

b Normal pregnancy Fetus c Pre-eclampsia and fetal growth restriction

Placenta Placenta Placental villous

tree has fewer
branches
Villous because
trophoblast of altered
cell blood flow
characteristics
Maternal
blood in
intervillous
space
Decidua
Spiral arterial Extravillous basalis
wall replaced trophoblast
by trophoblast cells
cells (interstitial)
(endovascular) Placental Spiral artery
bed giant remains
Decidua narrowed
basalis cells
in this segment
Basal
artery
Systemic sclerosis
Media Media
A chronic autoimmune disease Myometrium
that causes a hardening of the
skin. The skin thickens because Endothelium Endothelium
of increased deposits of
collagen. Compared with the
localized form of the disease
Radial artery Radial artery
(scleroderma), systemic
sclerosis causes more
widespread skin changes and Arcuate artery Arcuate artery
can be associated with damage Figure 2 | Disorders of human pregnancy resulting from abnormal placentation. a | The blood supply to the human
to the lungs, heart and kidneys.
pregnant uterus is shown. b | Normal pregnancy. The spiral arteries of the placental bed are converted to uteroplacental
arteries by the action of migratory extravillous trophoblast cells. Both the arterial media and the endothelium are
Endometriotic foci
disrupted by trophoblast cells, converting the artery into a wide calibre vessel that can deliver blood to the intervillous
Foci of endometrial tissue
outside the endometrium or
space at low pressure. The small basal arteries are not involved and remain as nutritive vessels to the inner myometrium
myometrium (muscle wall) of and decidua basalis. c | Pre-eclampsia and fetal growth restriction. When trophoblast-cell invasion is inadequate, there is
the uterus. They are usually deficient transformation of the spiral arteries. The disturbed pattern of blood flow leads to reduced growth of the
found in the peritoneum. branches of the placental villous tree, which results in poor fetal growth.

NATURE REVIEWS | IMMUNOLOGY VOLUME 6 | AUGUST 2006 | 587

© 2006 Nature Publishing Group
REVIEWS
Box 2 | Characteristics and functions of human uterine NK cells
• CD56hiCD16– uterine natural killer (NK) cells are similar to the minor CD56hiCD16–
NK-cell population in the blood but with phenotypic differences from both CD56low
and CD56hi blood NK-cell subsets.
• They represent 70% of leukocytes at the implantation site.
• They might arise ab initio from a separate lymphoid lineage or differentiate in the
endometrial microenvironment from blood CD56hi NK cells.
• They produce a range of soluble products, including angiogenic cytokines (such as
angiopoietin-2 and vascular endothelial growth factor C88) and lytic enzymes (such as
granzymes and perforin).
• The diagnostic tests used to evaluate NK-cell phenotype and activity in the peripheral
blood of women with reproductive failure give no information regarding uterine
NK-cell function89.
• Therapeutic regimes to downregulate NK-cell ‘activities’ or numbers (including
steroids or intravenous immunoglobulins) to treat pregnancy failure have little
scientific basis89.
• Similar cells are found in species with haemochorial placentation. These have been
called granulated metrial gland cells in rodents or Kurloff cells in guinea-pigs90–92.
In all species, they are always associated with the spiral arteries that supply the
placenta but their spatial association with trophoblast cells is more variable.
• Their functions are unknown, but possible roles include: first, to maintain the
mucosa and stability of blood vessels (in the non-pregnant endometrium). NK cells
are only found in the non-pregnant endometrium in menstruating primates when
there is an associated pronounced decidual reaction93. Second, to modify the walls
of spiral arteries (in the decidua). Third, to control trophoblast-cell invasion of the
decidua, myometrium and arteries (at the implantation site).
For more details, see REFS 6,86,87.
invasive trophoblast cells26. Mothers compromise, and it
Tubal pregnancy now seems probable that both ideas are correct and that
An ectopic pregnancy occurs the decidua allows orderly access of trophoblast cells
when the blastocyst implants to the maternal nutrient supply by achieving the right
at a site outside the uterus.
balance between under- and over-invasion.
Most ectopic pregnancies
occur in the fallopian tube so Trophoblast-cell penetration of the uterine epithe-
the terms ectopic pregnancy lium and invasion into the uterine wall and arteries is
and tubal pregnancy are nearly potentially highly dangerous, particularly in humans.
synonymous. Uncontrolled trophoblast-cell invasion is seen when
the decidua is deficient or absent, as in tubal pregnancy
Placenta creta
A condition when placental
or when the placenta implants on scar tissue from
trophoblast cells invade deeply a previous Caesarian section, a condition known as
into the muscle coat placenta creta27. Without medical intervention, these
(myometrium) of the uterus, conditions result in maternal death from haemorrhage.
usually because of the absence
Early studies in which trophoblast cells were trans-
of decidua. This can lead to
uterine rupture, torrential planted to ectopic sites in mice and pigs showed the
haemorrhage and failure of the inherent invasive proclivities of trophoblast cells26. The
placenta to separate after decidua can be considered to behave as a Procrustean
delivery. bed, violently forcing conformity on its guests what-
Procrustean bed
ever their shape or size — a harsher view of maternal
In Greek mythology, Procrustes compromise28.
(whose name means he At the opposite extreme, excessive restraint of tro-
who stretches) was a host who phoblast cells by the decidua can result in pregnancies
adjusted his guests to their
in which trophoblast-cell invasion into the arteries
bed. If they were longer than
the bed, he cut off the and uterine wall is inadequate. In this case, the territo-
redundant part; if shorter, he rial boundary has moved in favour of the mother and
stretched them till they fitted the blood supply to the fetus becomes poor. The main
it. Any attempt to reduce men problems that result from such reduced blood supply
to one standard, one way of
thinking or one way of acting,
are fetal prematurity, fetal growth restriction, still-birth
is called placing them on a and pre-eclampsia, and in many of these pregnancies
Procrustean bed. the main defect is reduced trophoblast-cell infiltration
588 | AUGUST 2006 | VOLUME 6
© 2006 Nature Publishing Group
into the uterus22 (FIG. 2c). The extent to which all of these
conditions occur in apes and monkeys is difficult to
ascertain, but pre-eclampsia seems to be restricted to
humans. Pre-eclampsia has a high maternal and fetal
mortality rate and mainly affects first-time mothers.
Why should such a devastating disease be maintained
despite the strong selective pressures for reproductive
success? The answer probably lies in the delicate nego-
tiation between trophoblast-cell invasion and decidua
that is required during every human pregnancy. The few
mothers dying from pre-eclampsia can be viewed as an
evolutionary consequence, or indeed sacrifice, because
of the need to control the aggressive behaviour of human
trophoblast cells.
Placentation in primates
Interestingly, not all primates have haemochorial pla-
centae but they are seen in all higher primates such as
monkeys, apes and humans, although the pattern of
trophoblast-cell invasion differs29 (FIG. 3a). In particular,
interstitial invasion into the decidual stroma and myo-
metrium is a prominent feature in humans, whereas only
vascular migration has been seen in Rhesus monkeys30.
The pattern of trophoblast-cell infiltration in the great
apes is not known. Humans clearly have evolved a
uniquely invasive form of placentation that is potentially
dangerous to the mother. So, what selective pressures
might be driving these changes? The most obvious dif-
ferences between humans and apes are bipedalism and
enlarged brain size. Both these characteristics could
influence reproductive strategies.
The physiological responses that are necessary
to redistribute blood flow to the uterus are likely to
alter with bipedalism31. The cardiac output is affected
because of compression of the inferior vena cava by the
larger uterus and also by the increased sympathetic
tone (that is, increased peripheral vascular resistance
and heart rate) required to ensure perfusion of blood
to the brain against the pull of gravity. These conse-
quences of bipedalism might place selective pressures to
increase structural transformation of the uterine arter-
ies by trophoblast cells to ensure that the uteroplacental
blood flow that is required for fetal development can be
achieved throughout pregnancy.
In the third trimester of human pregnancies, to sup-
port development of a large brain, 60% of total nutri-
tional needs are directed to the fetal brain compared
with only ∼20% in other mammalian pregnancies32.
However, large brains are not limited to species with
haemochorial placentation, as dolphins (which have
epitheliochorial placentae) have the second largest brain
after humans. Evolution of the large human brain, at a
time when the form of haemochorial placenta in higher
primates was already in place, required modifications of
the primate haemochorial placenta to allow increased
delivery of oxygen and nutrients through the uterine
arteries. To achieve this, trophoblast cells might have to
invade deeper into the uterine wall to modify the struc-
ture of the uterine arteries so that they are converted to
the high-conductance vessels that are required for the
development of our large brains.
www.nature.com/reviews/immunol
 REVIEWS

a Implantation site of rhesus monkey b Implantation site of mouse at day 13 of pregnancy

Cord and fetus Cord and fetus

Fetal vessel

Placenta
Central arterial
canal lined by
Maternal trophoblast cells
Intervillous space
blood
Maternal blood Fetal vessel
Placental villous
space
tree covered
Placenta
by villous
syncytiotrophoblast

Trophoblast-cell Labyrinth
shell area
Decidua

Trophoblast cells
infiltrate the
endothelium
and media Myometrium
of spiral arteries Junctional zone
to replace the wall with glycogen cells
Spongiotrophoblast
cell
Decidua basalis Mesometrium
Myometrium
Mesometrial triangle
Media containing NK cells

Endothelium
Uterine artery
Radial artery Radial artery Radial artery

Figure 3 | Placentation in rhesus monkeys and mice. a | In contrast to humans, the structural and destructive changes
to the uterine vessels of Rhesus monkeys are not so marked (FIG. 2). The arteries are only invaded in an endovascular
manner, with trophoblast cells migrating down the lumen of the blood vessels and replacing the endothelium and
eventually the media94. Interstitial trophoblast-cell invasion and the decidual reaction is very limited and no invasion of the
myometrium occurs. Instead, there is a well-developed trophoblast-cell shell that forms a clear demarcation line between
the placenta and uterine tissues. A similar pattern is found in baboons95. b | In mice, the area of placental exchange of
nutrients is the labyrinth, which is provided by extensive branching of the chorionic villi and is analogous to the villous
placenta in humans. Natural killer (NK) cells are abundant in the decidua basalis on days 8–10 of pregnancy. By day 13 of
pregnancy, there are very few NK cells remaining in the decidua basalis and they are found in the mesometrial triangle, an
area formed by the two layers of the myometrium.

Pre-eclampsia
Eclampsia (in Greek meaning
bolt from the blue) describes
grand mal seizures (epileptic
fits) occurring towards the end
of pregnancy. Pre-eclampsia
describes the symptoms that
precede eclampsia, which
include oedema, proteinuria
and hypertension.
Inferior vena cava
The large vein that carries
de-oxygenated blood from
the lower half of the body
to the heart.
Placenta–uterine immune interaction
So, what does placental development have to do with the
immunology of reproduction? The standard approach
of immunology text books is to view the fetus as an
allograft, a situation that can be associated with strong
antibody and T-cell-mediated responses to the allogeneic
MHC molecules expressed by the vascularized graft. This
approach does not distinguish between fetal cells in the
circulation or trophoblast cells in the uterus nor does it
encompass the different placental forms. For example, in
haemochorial placentation, although the decidua avoids
classic allogeneic rejection of trophoblast cells, the depth
of trophoblast-cell invasion is regulated. To understand
the mechanisms involved, both the anatomy of placenta-
tion and the maternal leukocytes present in the lining of
the uterus are clearly important considerations. In addi-
tion, it is crucial that the MHC status of trophoblast cells
is considered, as these molecules can function as ligands
for uterine immune cells, including T cells, NK cells and
myelomonocytic cells. Trophoblast-cell populations are
always negative for MHC class II expression, indicating
that they cannot present antigen directly to maternal
CD4+ T cells. The most definitive data on MHC expres-
sion are for the human placenta, whereas the picture in
other species is limited and confusing. Obviously, these
species differences in placentation and MHC expression
mean that humans and other primates need to be consid-
ered separately from species using other placental strat-
egies (particularly those without decidua), as the local
uterine immune responses are likely to be dissimilar.
Immune responses in epitheliochorial placentation. In
most cases of epitheliochorial placentation, the allan-
tochorion trophoblast cells that contact the uterine
epithelium lack expression of both MHC class I and II
molecules. There are however species-specific excep-
tions, with MHC-class-I-expressing trophoblast cells
described at certain sites and stages of gestation in
several species. For example, in horses, MHC-class-I-
positive trophoblast cells do invade into the uterus to form

NATURE REVIEWS | IMMUNOLOGY VOLUME 6 | AUGUST 2006 | 589

© 2006 Nature Publishing Group
REVIEWS
Endometrial cups
A focal collection of
trophoblast cells that
penetrate the uterus of horses.
These cells are responsible for
secretion of equine chorionic
gonadotrophin.
Ectoplacental cone
A core of rapidly dividing
trophoblast cells with an outer
layer of giant cells that is
present in the developing
mouse conceptus at 7.5 days
post-coitum.
590 | AUGUST 2006 | VOLUME 6
Table 1 | Differences between mouse and human placentation
Characteristics Mouse
Intravascular trophoblast Minimal*
-cell invasion
Arterial transformation Largely independent of trophoblast
cells
Decidua Formed after implantation only at site of
placentation
NK cells Infiltrate the media of arteries and
disrupt vascular architecture
NK-cell receptors KIR genes not functional. The Ly49
family fulfils the same function as
human KIRs, but Ly49 expression by
uterine NK cells and MHC expression by
trophoblast cells are unknown
*In guinea-pigs91, trophoblast cells migrate through the media out of the uterus into the mesometrial artery. This species has a long
gestation period. Hamsters90 have prominent granulated lymphocytes that form a sheath around the arteries. In rats, in late gestation39,
trophoblast cells extend as far as the mesometrial triangle. KIR, killer-cell immunoglobulin-like receptor; NK, natural killer.
transient endometrial cups that are surrounded and even-
tually destroyed by maternal lymphocytes. In addition,
maternal antibody responses to paternal MHC class I
antigens are often generated in horses, but cytotoxic
T-cell responses against paternal alloantigens are reduced
compared with such responses before pregnancy, indi-
cating an asymmetric immune response to the fetus33.
Subpopulations of bovine trophoblast cells seem to
express mRNAs that are encoded by both classical and
non-classical MHC class I genes34. In sheep, binucleate
trophoblast cells fuse with the epithelial cells that line the
uterus, creating a condition known as synepitheliochorial
placentation. Although the binucleate cells express MHC
class I molecules35, the potential immunological effects of
this fusion between two allogeneic cells are not known.
In species that use epitheliochorial placentae, the
simple apposition between the placenta and uterine
epithelium might not provoke any damaging immune
responses by the mother. The conceptus in these species
could be regarded as similar to commensal bacteria in
the gut, generating minimal immune recognition by the
host unless they breach the epithelial-cell barrier. In other
words, the conceptus is non-self, settled innocuously in
an epithelial-cell-lined lumen. Although intraepithelial
granulated lymphocytes, which are characteristic of
mucosal surfaces, have been described, the endometrial
stroma in epitheliochorial placentation lacks NK cells36.
This indicates that a different immune response to the
placenta occurs in species that use epitheliochorial pla-
centae than in those species that have trophoblast-cell
invasion and decidualization.
Immunology of rodent placentation. Haemochorial pla-
centation is a feature of most rodents (FIG. 3b) but it differs
from that in humans with regard to the depth of inva-
sion by trophoblast cells and the pattern of distribution
of uterine NK cells around the spiral arteries that supply
the placenta37–40 (TABLE 1). Even among rodents, there are
significant variations. In mice, the presence of uterine NK
© 2006 Nature Publishing Group
Human
Extensive
Caused by perivascular and endovascular
trophoblast cells
Formed in late-secretory non-pregnant
endometrium, involving entire uterine mucosa
Encircle adventitia of arteries. Probably have
some direct effect on arterial function but mainly
act through indirect effects on trophoblast-cell
invasion
Highly diverse KIR genes. The KIR ligands HLA-G
and HLA-C are present on migratory trophoblast
cells
cells in the media of the arteries indicates that they might
have a direct physiological role in regulating the blood
pressure and flow to the placenta. In support of this, preg-
nant mice with no uterine NK cells retain the vascular
architecture that is typical of the non-pregnant state41.
Mouse uterine NK cells might also indirectly modify the
blood flow through an effect on trophoblast-cell behav-
iour (as seems to be the case in humans), because direct
contact occurs between uterine NK cells and trophoblast
cells when the ectoplacental cone moves into the decidual
tissue on day 8 of gestation. The receptors expressed by
uterine NK cells might give insights into how these cells
function, but information on the expression by uterine
NK cells of members of the Ly49 family (which carry out
the same function as killer-cell immunoglobulin-like
receptors (KIRs) in primates) and about their cognate
ligands on trophoblast cells is sparse. It is therefore not
clear whether uterine NK cells have the same role or use
the same molecular mechanisms in mice and humans.
T-cell recognition of paternal alloantigens expressed by
the fetus was shown to occur in mice in which all T cells
expressed a transgenic T-cell receptor specific for pater-
nal alloantigens, and this resulted in transient tolerance
of the transgenic T cells42. But where in the feto-placental
unit are these alloantigens expressed? The most glaring
omission that has so far prevented a clear understanding
of mouse reproductive immunology is the lack of defini-
tive information on the MHC expression status of mouse
trophoblast cells. It seems that the labyrinthine trophoblast
cells are MHC class I and class II negative. By contrast, the
spongiotrophoblast cells that separate the labyrinth layer
from the decidua have been shown to express polymor-
phic paternally derived MHC class I molecules43,44. It is not
known which MHC class I loci encode these products nor
whether any non-classical MHC molecules are present45.
Disruption of many immunological pathways can
lead to reproductive failure in mouse models, but cau-
tion is needed in interpreting the results because it is
often unclear whether the failure is caused by a classical
www.nature.com/reviews/immunol

Syncytiotrophoblast
The outermost trophoblast-cell
layer covering the chorionic villi
that is formed by fusion of the
underlying layer of
mononuclear trophoblast cells
to become a multinucleated
syncytium, which forms a
barrier between the fetus and
the mother.
NATURE REVIEWS | IMMUNOLOGY
allogeneic response or due to some other immunological
mechanism such as inflammation. Paradoxically, the ‘nor-
mal’ controls used in many of these models are syngeneic
pregnancies, clearly something of an oxymoron when
thinking of normal human pregnancy 46–48. Furthermore,
in another classic mouse model of abortion — CBA/J
mice (H2k) crossed with DBA/2 mice (H2d) — the nor-
mal control mating is with a BALB/c male (which is also
H2d), so it is not certain whether the fetal losses in CBA/J
× DBA/2 matings have an alloimmune basis owing to
MHC differences49–51. In another model, T-cell reactivity
to the conceptus could be implicated because ligation
of the T-cell-expressed co-stimulatory molecule CD40
ligand led to pregnancy loss47. However, the mechanism
of pregnancy failure proved to be caused by dysfunc-
tional ovaries (ovarian insufficiency) resulting from
excessive inflammation in the ovary. Despite these
caveats, lessons can be learnt.
When syngeneic matings are used, analysis of gene-
knockout animals has indicated that pregnancy failure
(resorption) results from the lack of genes that seem to
have functions that prevent excessive inflammation at
the implantation site52,53. These genes encode proteins
such as the complement regulator Crry (complement-
receptor-related protein) and CD95 ligand (CD95L, also
known as FASL)52,53. Other models using allogeneic mat-
ings have helped to explain how adverse T-cell responses
might be avoided; for example, by mechanisms involving
indoleamine 2,3-dioxygenase (IDO), T-cell co-stimulatory
molecules (such as CD80, CD86 and programmed death
ligand 1 (PDL1)) and immune deviation to T helper 2
(TH2)-type responses46,50,51,54,55. There is presumably
redundancy in the system, because mice deficient in IDO
or TH2-type cytokines reproduce normally56,57. Notably,
many of these pathways might affect the generation of
regulatory T cells, which are known to be increased in
mouse as well as human pregnancies both systemically
and in the uterus48,58,59. However, regulatory T cells seem
to be driven by hormonal rather than antigen-dependent
mechanisms, as they are also increased in number in
syngeneic pregnancies. Depletion of regulatory T cells
leads to failure of pregnancies following allogeneic but
not syngeneic matings, indicating that regulatory T cells
might regulate damaging allospecific effector T-cell
responses48. Crucially, it is still not established whether
the failed pregnancies occur because of T-cell reactivity
to either trophoblast-cell antigens or to fetal antigens.
Furthermore, the effector mechanisms are unclear, as it
has not been shown that uterine T cells can kill murine
trophoblast cells.
Adaptive immune responses in humans and other pri-
mates. In primates, placental trophoblast cells encounter
the maternal immune system in two main areas — the
villous trophoblast cells interact with the maternal blood
and the extravillous trophoblast cells interact with the
uterine tissue. The first area of interaction is between
the layer of syncytiotrophoblast that overlies the chorionic
villi and is bathed by maternal blood that is delivered by
the spiral arteries into the intervillous space (FIG. 3a). In
humans, the syncytiotrophoblast is therefore in contact
© 2006 Nature Publishing Group
REVIEWS
with the systemic but not the uterine immune com-
ponents of the mother. The syncytiotrophoblast expresses
no MHC antigens on its surface, which is consistent
with the concept that the placenta is immunologically
neutral6. Indeed, it has been difficult to demonstrate any
systemic maternal T- or B-cell responses to trophoblast
cells (as opposed to fetal cells that cross into the maternal
circulation) during human pregnancy 60. Hints that there
are qualitative differences in all systemic T- and B-cell
responses in pregnancy come from the altered clinical
course of autoimmune diseases and viral infections dur-
ing pregnancy. For example, the symptoms of rheumatoid
arthritis (which is TH1-cell mediated) improve during
pregnancy, whereas those of systemic lupus erythema-
tosus (which is TH2-cell mediated) worsen and this is
presumably caused by the bias away from TH1- towards
TH2-cell responses61,62. Notably, these responses are to all
antigens, not just to those expressed by components of
the feto-placental unit. This shift to TH2-cell responses in
pregnancy might be an epiphenomenon that is secondary
to the flux of hormones and cytokines that are secreted
into the circulation, because there is no evidence that it
is essential for pregnancy success in humans. Overall,
it is improbable that classical allogeneic rejection of the
villous placenta is responsible for reproductive failure.
The second area of contact is between invasive extra-
villous trophoblast cells and immune cells in the decidua.
In contrast to the syncytiotrophoblast, extravillous tropho-
blast cells express an unusual combination of HLA-C,
HLA-G and HLA-E molecules6. High level expression
of HLA-G is restricted to the trophoblast cells that
infiltrate the uterus. The polymorphic HLA-A and
HLA-B molecules, which initiate allograft rejection, are
not expressed, and HLA-C is the only HLA molecule
expressed by trophoblast cells that shows any appreciable
polymorphism. In those species that have been studied
in detail, such as humans and mice, there is no large
influx of T or B cells to the implantation site in normal
pregnancies. Any T cells present in failed pregnancies
might be recruited following the demise of the fetus
and the resulting inflammatory changes. As in mice,
an important role for T-cell damage to trophoblast cells
infiltrating the decidua that results in pregnancy loss in
humans has not been established.
So, how are adverse maternal T-cell responses to
paternally expressed HLA-C molecules or other unidenti-
fied trophoblast-cell antigens avoided? MHC-class-II-
expressing macrophages and dendritic cells (DCs) are
present in the placental bed and could present trophoblast-
cell-derived antigens indirectly to the maternal immune
system63. These decidual antigen-presenting cells might
be pivotal in the expansion of both CD4+CD25+ and CD8+
regulatory T-cell populations that are present in utero
during human pregnancy 59,64. Interestingly, the CD8+
regulatory T cells in the uterus are not MHC restricted
but are specific for a member of the carcinoembryonic
antigen family, an oncofetal trophoblast molecule, and
selectively use the T-cell receptor Vβ964. Another possible
mechanism to explain the lack of uterine T-cell activa-
tion in normal pregnancies depends on the high-avidity
binding of HLA-G to leukocyte immunoglobulin-like
VOLUME 6 | AUGUST 2006 | 591
REVIEWS
receptors (LILRs) expressed by myelomonocytic cells65.
Increased expression of LILRB1 is associated with the
induction of a ‘tolerogenic’ population of DCs, which, in
a transplantation setting, results in tolerance66,67. Recent
data have indicated that this HLA-G-induced tolerance
was due to decreased MHC class II peptide presentation
by the tolerogenic DCs68. The idea that the placenta itself
is modifying the maternal immune reactivity, locally in
the uterus, through a trophoblast-cell-specific monomor-
phic HLA molecule or an oncofetal protein to downregu-
late T-cell responses during pregnancy is attractive. In the
non-pregnant endometrium, T-cell responses are normal
as evidenced by rapid production of granulomas follow-
ing infection of the endometrium with Mycobacterium
tuberculosis69.
Of the non-human primates studied, Rhesus monkeys
(Macaca mulatta) express a MHC molecule (Mamu-AG)
that has many of the characteristics of HLA-G, includ-
ing that of having a soluble variant70. A similar MHC
molecule is also present in baboons71. However, the pat-
tern of expression is different, as the baboon MHC-like
molecules are expressed by the syncytiotrophoblast.
This might reflect the limited interstitial invasion by
extravillous trophoblast cells in these species. The role
of these HLA-G-like molecules in immunomodulation
is unexplored.
Uterine NK-cell recognition of trophoblast cells.
Predecidual changes in the endometrium and the influx
of uterine NK cells, which occur before implantation, are
unique to primates. Given the lack of evidence for T-cell
responses to trophoblast cells, it is compelling to think
that uterine NK cells provide the main mechanism by
which the maternal immune system recognizes tropho-
blast cells. In humans, uterine NK cells express an array of
receptors, some of which are known to bind to the HLA
class I molecules expressed by extravillous trophoblast
cells6. Unlike blood NK cells, all uterine NK cells express
high levels of the C-type lectin family member CD94–
NKG2A, which binds to HLA-E resulting in inhibition
of NK-cell cytotoxicity72. All NK cells also express the
KIR-family member KIR2DL4, which can bind HLA-G.
HLA-G is endocytosed into KIR2DL4-containing endo-
somal compartments. The subsequent interaction results
in upregulation of expression of pro-inflammatory and
pro-angiogenic cytokines, indicating a mechanism by
which the placenta can increase its own blood supply73. In
addition, any soluble HLA-G molecules in the maternal
circulation could bind KIR2DL4 expressed by blood NK
cells and as a result could contribute to the inflammatory
and vascular changes that are characteristic of all preg-
nancies74. Therefore, a trophoblast-cell MHC molecule
can signal to the decidual innate immune system through
both KIR2DL4 on NK cells and LILRB1 (or LILRB2) on
myelomonocytic cells. By alerting two different cell types,
HLA-G might be acting as a ‘placental’ signal that induces
pregnancy-specific functions in the uterus.
HLA-C is the only known polymorphic MHC or
MHC-like molecule that is expressed by trophoblast cells
and is the dominant ligand for the members of the KIR
family of receptors that have two immunoglobulin-like
592 | AUGUST 2006 | VOLUME 6
© 2006 Nature Publishing Group
domains (KIR2D). These might be activating (KIR2DS)
or inhibitory (KIR2DL) receptors. KIR haplotypes com-
prise two groups, A and B, the main difference being
that there are more activating receptors in the B hap-
lotype75. In any pregnancy, the maternal KIR genotype
could be AA (no activating KIR) or AB/BB (presence
of between one and five activating KIRs). The HLA-C
ligands for KIRs on trophoblast cells can belong to two
groups, HLA-C1 and HLA-C2, which are defined by
a dimorphism at position 80 of the α1 domain. This
maternal–fetal immunological interaction that occurs
at the site of placentation, therefore involves two poly-
morphic gene systems, maternal KIRs and fetal HLA-C
molecules. NK-cell function is therefore likely to vary in
each pregnancy. Some KIR/HLA-C combinations might
be more favourable to trophoblast-cell invasion, result-
ing in a greater increase in in utero placental blood flow
than other combinations as a result of the overall signals
that the NK cell receives.
This hypothesis is supported by a recent study show-
ing that the occurrence of pre-eclampsia is associated
with an increased frequency of maternal KIRs of the
AA genotype but only when this is combined with
the presence of an HLA-C2 allotype in the fetus76. How
do these genetic results translate to functional events
at the implantation site? The KIR phenotype of uterine
NK cells is skewed towards increased expression of the
KIR2D receptors that bind to the two HLA-C groups
compared with blood NK cells77. Stronger inhibitory sig-
nals are delivered to NK cells by the HLA-C2–KIR2DL1
interaction compared with the HLA-C1–KIR2DL2 or
HLA-C1–KIR2DL3 interactions75. We propose that in
pregnancies with a fetus that expresses HLA-C2, the
strong inhibitory signal needs to be overcome for suf-
ficient trophoblast-cell invasion to occur and this will
happen if the mother has activating KIRs, otherwise the
feto-placental blood supply will be inadequate. When
trophoblast cells are homozygous for HLA-C1, there is
only weak inhibition that does not require the presence
of compensatory activating KIRs. To summarize, uterine
NK cells do express KIRs that are specific for HLA-C
ligands expressed by trophoblast cells, and genetic
polymorphisms of this system can affect reproduc-
tive outcome. This predicts that there is strong selec-
tion against those HLA-C–KIR combinations that are
detrimental to reproduction. Population analysis has
demonstrated reciprocal frequencies of HLA-C2 and
KIR AA genotypes in different human populations76.
Because they segregate independently, a situation might
have evolved so that pregnancies with HLA-C2–KIR AA
combinations (which are associated with pre-eclampsia)
do not occur too frequently in any population. However,
individuals must not only reproduce but their offspring
need to survive, and balancing selection (which would
maintain the gene frequencies) for KIRs and HLA-C
might come from immune recognition of pathogens78.
Comparison of the human KIR-gene family with that
of chimpanzees, gorillas, bonobos, orangutans and rhesus
macaques indicates that this system is rapidly evolving79–83.
With regard to placentation, the lineage of KIR genes that
recognizes MHC-C molecules is only present in apes.
www.nature.com/reviews/immunol
 REVIEWS

In rhesus macaques, in which there is a well-defined tro-
phoblast-cell shell, minimal infiltration of the decidual
stroma and modification of the arteries only by endovas-
cular trophoblast cells, MHC-C–KIR interactions do not
occur 80. MHC-C is only present in half of orangutans and
all the alleles belong to the C1 group. In this species, the
KIR genes that are predicted to bind to MHC-C would
all bind the C1 epitope and there are none specific for
MHC-C2 (REF. 81). Chimpanzees, gorillas and bonobos
have KIRs that can bind MHC-C of both C1 and C2
groups. This shows species-specific co-evolution of both
KIR and MHC-C genes. The MHC-C C1 group, when
in combination with the KIR AA genotype, seems to be
neutral as far as the risk of pre-eclampsia is concerned.
This is the only combination that occurs in orangutans
and so the strong KIR inhibition mediated by MHC-C
C2 is a later addition in the great apes. This has possibly
arisen as a result of selective pressures imposed by the
increasingly dangerous placental invasion.
Concluding remarks
Although structural variations in eutherian placentae
provide endless fascination for comparative anatomists,
they can present difficulties when extrapolating results
from animal studies to human pregnancy. Structural
characteristics are important in the study of pregnancy
immunology because the more invasive the placenta,
the greater the interaction it is likely to have with the
maternal immune system. Placental anatomical varia-
tion is reflected in the considerable difference in the gene
repertoire for both immunity and reproduction in the
mouse and human genome84. The two gene systems that
have diverged most are the MHC genes and the NK-cell-
receptor genes and these now seem to have important
roles in both reproduction and immunity.
Although the immunological characteristics of human
placentation are fairly well documented, the situation in
other species, including mice, is still sparse and often con-
flicting. It is clear that the placenta is not immunologically
neutral because MHC antigens are expressed by tropho-
blast-cell subpopulations in most of the species studied. In
humans, these are ligands for receptors on innate immune
cells, and whether MHC-restricted T-cell recognition of
trophoblast cells occurs in normal or abnormal pregnan-
cies is unclear. It will be a challenge to determine how
regulatory T cells, HLA-G, oncofetal antigens and other
potential mechanisms to avoid adverse T-cell responses
to trophoblast cells are generated and whether failure of
T-cell control ever does have a role in pregnancy failure.
The role of NK cells in pregnancy is also uncertain,
although in humans, there is an indication that HLA-C–
KIR interactions between trophoblast cells and NK cells
do regulate the depth of trophoblast-cell invasion. It is also
probable that there is a direct effect of uterine NK cells
on spiral artery structure and function (possibly modified
by soluble trophoblast-cell-derived factors). The relative
importance of interactions between the three components
— uterine NK cells, trophoblast cells and arteries — prob-
ably vary in different species. Whatever mechanisms are
involved, the maternal immune system must provide
a balance between the need for fetal intrusion into the
mother’s resources and the need to protect the mother
from excessive fetal greed. In studying this, the view of
the uterus as a ‘privileged site’ is no longer valid, as all
anatomical sites have unique immune features and this
applies particularly to mucosal surfaces. The comparison
of the uterine mucosa to the gut or the nose (in which
CD56hi NK-like cells are also frequently found) would
seem far more informative than to the traditional sites of
immune privilege, such as the eye, brain or testis.

1. Medawar, P. B. Some immunological and
endocrinological problems raised by the evolution
of viviparity in vertebrates. Symp. Soc. Exp. Biol.
7, 320–338 (1953).
The seminal paper that introduced the concept
of the fetus as an allograft.
2. Trowsdale, J. & Betz, A. G. Mother’s little helpers:
mechanisms of maternal-fetal tolerance.
Nature Immunol. 7, 241–246 (2006).
3. Steven, D. H. (ed) Comparative Placentation
(Academic Press, New York, 1975).
4. Mossman, H. W. Vertebrate fetal membranes.
(Rutgers University Press, New Brunswick, 1987).
5. Wooding, F. B. P. & Flint, A. P. F. in Marshall’s
Physiology of Reproduction Vol. 3 Ch. 4 (ed. Lamming,
G. E.) 230–466 (Chapman and Hall, London, 1994).
6. Moffett-King, A. Natural killer cells and pregnancy.
Nature Rev. Immunol. 2, 656–663 (2002).
7. Johnson, M. Origins of pluriblast and trophoblast
in the eutherian conceptus. Reprod. Fertil. Dev.
8, 699–709 (1996).
8. Moffett, A., Loke, Y. W. & McLaren, A. The Biology
and Pathology of Trophoblast (Cambridge University
Press, Cambridge, 2006).
9. Blackburn, D. G. Reconstructing the evolution
of viviparity and placentation. J. Theor. Biol.
192, 183–190 (1998).
10. Romer, A. S. Major steps in vertebrate evolution.
Science 158, 1629–1637 (1967).
11. Luckett, W. P. in Major Patterns of Vertebrate Evolution
(eds Hecht, M. K., Goody, P. C. & Hecht, B. M.)
439–516 (Plenum Press, New York 1977).
12. Hubrecht, A. A. W. Studies in mammalian
embryology. 1. The placentation of Erinaceus
europaeus, with remarks on the phylogeny of the
placenta. Q. J. Microsc. Sci. 119, 283–404 (1889).
13. Amoroso, E. C. in Marshall’s Physiology of
Reproduction Vol. 2 Ch. 15 (ed.Parkes, A. S.)
127–311 (Longman, Green and Co., London,
1952).
14. Vogel, P. The current molecular phylogeny of
Eutherian mammals challenges previous
interpretations of placental evolution. Placenta 26,
591–596 (2005).
15. Murphy, W. J. et al. Resolution of the early placental
mammal radiation using Bayesian phylogenetics.
Science 294, 2348–2351 (2001).
16. Kriegs, J. O. et al. Retroposed elements as archives
for the evolutionary history of placental mammals.
PLoS Biol. 4, e91 (2006).
17. Mess, A. & Carter, A. M. Evolutionary transformations
of fetal membrane characters in Eutheria with special
reference to Afrotheria. J. Exp. Zool. B. 306, 140–
163 (2006).
18. Enders, A. C. & Carter, A. M. What can comparative
studies of placental structure tell us? — A review.
Placenta 25 (Suppl. A), S3–S9 (2004).
19. Wildman, D. E. et al. Evolution of the mammalian
placenta revealed by phylogenetic analysis.
Proc. Natl Acad. Sci. USA 103, 3203–3208 (2006).
20. Clarke, C. A. Prevention of rhesus iso-immunisation.
Lancet 2, 1–7 (1968).
21. Nelson, J. L. Maternal-fetal immunology and
autoimmune disease. Is some autoimmune disease
auto-allo or allo-auto immune? Arthritis Rheum.
39, 191–194 (1996).
22. Pijnenborg, R., Vercruysse, L. & Hanssens, M.
The uterine spiral arteries in human pregnancy:
facts and controversies. Placenta 17 February 2006
[epub ahead of print].
The definitive paper of the placental bed based
on a life-time’s study.
23. Jauniaux, E., Poston, L. & Burton, G. J. Placental-
related diseases of pregnancy: involvement of
oxidative stress and implications in human evolution.
Hum. Reprod. Update 8 May 2006 [epub ahead of
print].
24. King, A. et al. Uterine leukocytes and decidualization.
Hum. Reprod. Update 6, 28–36 (2000).
25. Robertson, W. B. in Obstetrical and Gynaecological
Pathology (ed. Fox, H.) 1149–1176 (Churchill
Livingstone, 1987).
26. Kirby, D. R. S. in The Early Conceptus, Normal and
Abnormal (ed. Park, W. W.) 68–73 (Proceedings of
Symposium, Queen’s College, Dundee, 1965).
27. Khong, T. Y. & Robertson, W. B. Placenta creta and
placenta praevia creta. Placenta 8, 399–409 (1987).
28. McLaren, A. in The Early Conceptus, Normal and
Abnormal (ed. Park, W. W.) 27–33 (Proceedings
of Symposium, Queen’s College, Dundee, 1965).
29. Martin, R. D. Human reproduction: a comparative
background for medical hypotheses. J. Reprod.
Immunol. 59, 111–135 (2003).
30. Ramsey, E. M., Houston, M. L. & Harris, J. W. S.
Interactions of the trophoblast and maternal tissues
in three closely related primate species. Am. J. Obstet.
Gynecol. 124, 647–652 (1976).
This paper emphasizes that, even among closely
related species of primates, the ways trophoblast
cells interact with maternal tissues show many
differences.
31. Rockwell, L. C., Vargas, E. & Moore, L. G. Human
physiological adaptation to pregnancy: inter- and
intraspecific perspectives. Am. J. Hum. Biol.
15, 330–341 (2003).
32. Chaline, J. Increased cranial capacity in hominoid
evolution and preeclampsia. J. Reprod. Immunol.
59, 137–152 (2003).

NATURE REVIEWS | IMMUNOLOGY VOLUME 6 | AUGUST 2006 | 593

© 2006 Nature Publishing Group
REVIEWS
33. Baker, J. M., Bamford, A. I. & Antczak, D. F.
Modulation of allospecific CTL responses during
pregnancy in equids: an immunological barrier to
interspecies matings? J. Immunol. 162, 4496–4501
(1999).
34. Davies, C. J., Eldridge, J. A., Fisher, P. J. &
Schlafer, D. H. Evidence for expression of both
classical and non-classical major histocompatibility
complex class I genes in bovine trophoblast cells.
Am. J. Reprod. Immunol. 55, 188–200 (2006).
35. Bainbridge, D. R., Sargent, I. L. & Ellis, S. A. Increased
expression of major histocompatibility complex (MHC)
class I transplantation antigens in bovine trophoblast
cells before fusion with maternal cells. Reproduction
122, 907–913 (2001).
36. Stewart, I. J. Granulated metrial gland cells in ‘minor’
species. J. Reprod. Immunol. 40, 129–146 (1998).
37. Georgiades, P., Ferguson-Smith, A. C. & Burton, G. J.
Comparative developmental anatomy of the murine
and human definitive placentae. Placenta 23, 3–19
(2002).
38. Ain, R., Canham, L. N. & Soares, M. J. Gestation
stage-dependent intrauterine trophoblast cell invasion
in the rat and mouse: novel endocrine phenotype and
regulation. Dev. Biol. 260, 176–190 (2003).
39. Vercruysse, L., Caluwaerts, S., Luyten, C. &
Pijnenborg, R. Interstitial trophoblast invasion in the
decidua and mesometrial triangle during the last
third of pregnancy in the rat. Placenta 27, 22–33
(2005).
40. Hemberger, M., Nozaki, T., Masutani, M. & Cross, J. C.
Differential expression of angiogenic and vasodilatory
factors by invasive trophoblast giant cells depending
on depth of invasion. Dev. Dyn. 227, 185–191
(2003).
41. Croy, B. A., Chantakru, S., Esadeg, S., Ashkar, A. A. &
Wei, Q. Decidual natural killer cells: key regulators of
placental development (a review). J. Reprod. Immunol.
57, 151–168 (2002).
Comprehensive review discussing uterine NK cells
in mice.
42. Tafuri, A., Alferink, J., Moller, P., Hammerling, G. J.
& Arnold, B. T cell awareness of paternal
alloantigens during pregnancy. Science 270,
630–633 (1995).
43. Redline, R. W. & Lu, C. Y. Localization of fetal major
histocompatibility complex antigens and maternal
leukocytes in murine placenta. Implications for
maternal-fetal immunological relationship. Lab. Invest.
61, 27–36 (1989).
44. Zuckermann, F. A. & Head, J. R. Expression of MHC
antigens on murine trophoblast and their
modulation by interferon. J. Immunol. 137,
846–853 (1986).
45. Erlebacher, A., Lukens, A. K. & Glimcher, L. H. Intrinsic
susceptibility of mouse trophoblasts to natural killer
cell-mediated attack in vivo. Proc. Natl Acad. Sci. USA
99, 16940–16945 (2002).
46. Munn, D. H. et al. Prevention of allogeneic fetal
rejection by tryptophan catabolism. Science
281, 1191–1193 (1998).
47. Erlebacher, A., Zhang, D., Parlow, A. F. &
Glimcher, L. H. Ovarian insufficiency and early
pregnancy loss induced by activation of the innate
immune system. J. Clin. Invest. 114, 39–48 (2004).
48. Aluvihare, V. R. et al. Regulatory T cells mediate
maternal tolerance to the fetus. Nature Immunol.
5, 266–271 (2004).
Paper describing a potential role for regulatory
T cells in allogeneic murine pregnancies.
49. Sollwedel, A. et al. Protection from abortion by
heme oxygenase-1 up-regulation is associated with
increased levels of Bag-1 and neuropilin-1 at the fetal-
maternal interface. J. Immunol. 175, 4875–4885
(2005).
50. Zhu, X. Y. et al. Blockade of CD86 signaling facilitates
a Th2 bias at the maternal-fetal interface and expands
peripheral CD4+CD25+ regulatory T cells to rescue
abortion-prone fetuses. Biol. Reprod. 72, 338–345
(2005).
51. Zenclussen, A. C. et al. Regulatory T cells induce a
privileged tolerant microenvironment at the fetal-
maternal interface. Eur. J. Immunol. 36, 82–94 (2006).
52. Xu, C. et al. A critical role for murine complement
regulator crry in fetomaternal tolerance. Science
287, 498–501 (2000).
53. Hunt, J. S., Vassmer, D., Ferguson, T. A. & Miller, L.
Fas ligand is positioned in mouse uterus and placenta
to prevent trafficking of activated leukocytes between
the mother and the conceptus. J. Immunol.
158, 4122–4128 (1997).
594 | AUGUST 2006 | VOLUME 6
54. Guleria, I. et al. A critical role for the programmed
death ligand 1 in fetomaternal tolerance. J. Exp. Med.
202, 231–237 (2005).
55. Krishnan, L., Guilbert, L. J., Wegmann, T. G.,
Belosevic, M. & Mosmann, T. R. T helper 1 response
against Leishmania major in pregnant C57BL/6 mice
increases implantation failure and fetal resorptions.
Correlation with increased IFN-γ and TNF and reduced
IL-10 production by placental cells. J. Immunol.
156, 653–662 (2006).
56. Fallon, P. G. et al. IL-4 induces characteristic Th2
responses even in the combined absence of IL-5,
IL-9, and IL-13. Immunity 17, 7–17 (2002).
57. Baban, B. et al. Indoleamine 2,3-dioxygenase
expression is restricted to fetal trophoblast giant
cells during murine gestation and is maternal
genome specific. J. Reprod. Immunol. 61, 67–77
(2004).
58. Somerset, D. A., Zheng, Y., Kilby, M. D., Sansom, D. M.
& Drayson, M. T. Normal human pregnancy is
associated with an elevation in the immune
suppressive CD25+CD4+ regulatory T-cell subset.
Immunology 112, 38–43 (2004).
59. Sasaki, Y. et al. Decidual and peripheral blood
CD4+CD25+ regulatory T cells in early pregnancy
subjects and spontaneous abortion cases. Mol. Hum.
Reprod. 10, 347–353 (2004).
60. Grabowska, A. Placental antigenicity and maternal
immunoregulation in human and murine pregnancy.
Ph.D. Thesis, Univ. Cambridge (1989).
61. Ostensen, M. Sex hormones and pregnancy in
rheumatoid arthritis and systemic lupus erythematosus.
Ann. NY Acad. Sci. 876, 131–144 (1999).
62. Buyon, J. P. The effects of pregnancy on autoimmune
diseases. J. Leukoc. Biol. 63, 281–287 (1998).
63. Gardner, L. & Moffett, A. Dendritic cells in the human
decidua. Biol. Reprod. 69, 1438–1446 (2003).
64. Shao, L., Jacobs, A. R., Johnson, V. V. & Mayer, L.
Activation of CD8+ regulatory T cells by human
placental trophoblasts. J. Immunol. 174, 7539–7547
(2005).
65. Shiroishi, M. et al. Efficient leukocyte Ig-like receptor
signaling and crystal structure of disulfide-linked
HLA-G dimer. J. Biol. Chem. 281, 10439–10447
(2006).
66. Chang, C. C. et al. Tolerization of dendritic cells by
TS cells: the crucial role of inhibitory receptors ILT3
and ILT4. Nature Immunol. 3, 237–243 (2002).
67. Velten, F. W., Duperrier, K., Bohlender, J.,
Metharom, P. & Goerdt, S. A gene signature of
inhibitory MHC receptors identifies a BDCA3+
subset of IL-10-induced dendritic cells with reduced
allostimulatory capacity in vitro. Eur. J. Immunol.
34, 2800–2811 (2004).
68. Ristich, V., Liang, S., Zhang, W., Wu, J. & Horuzsko, A.
Tolerization of dendritic cells by HLA-G. Eur. J. Immunol.
35, 1133–1142 (2005).
69. Sherman, M. E., Mazur, M. T. & Kurman, R. J. in
Blaustain’s Pathology of the Female Genital Tract
(ed. Kurman, R. J.) 428 (Springer, New York, 2002).
70. Ryan, A. F., Grendell, R. L., Geraghty, D. E. &
Golos, T. G. A soluble isoform of the rhesus monkey
nonclassical MHC class I molecule Mamu-AG is
expressed in the placenta and the testis. J. Immunol.
169, 673–683 (2002).
71. Stern, P. L. et al. Class I-like MHC molecules
expressed by baboon placental syncytiotrophoblast.
J. Immunol. 138, 1088–1091 (1987).
72. King, A., et al. HLA-E is expressed on trophoblast
and interacts with CD94/NKG2 receptors on
decidual NK cells. Eur. J. Immunol. 30, 1623–1631
(2000).
73. Rajagopalan, S. et al. Activation of NK cells by an
endocytosed receptor for soluble HLA-G. PLoS. Biol.
4, e9 (2006).
A careful study describing stimulation of NK cells
by soluble HLA-G that is endocytosed by KIR2DL4.
74. Redman, C. W. & Sargent, I. L. Preeclampsia and the
systemic inflammatory response. Semin. Nephrol.
24, 565–570 (2004).
75. Parham, P. MHC class I molecules and KIRs in human
history, health and survival. Nature Rev. Immunol.
5, 201–214 (2005).
A comprehensive reference covering all aspects
of KIRs.
76. Hiby, S. E. et al. Combinations of maternal KIR and
fetal HLA-C genes influence the risk of preeclampsia
and reproductive success. J. Exp. Med. 200, 957–965
(2004).
This paper provides genetic evidence for a role for
NK cells in regulating trophoblast-cell invasion.
© 2006 Nature Publishing Group
77. Verma, S., King, A. & Loke, Y. W. Expression of
killer-cell inhibitory receptors (KIR) on human uterine
NK cells. Eur. J. Immunol. 27, 979–983 (1997).
78. Martin, M. P. & Carrington, M. Immunogenetics of
viral infections. Curr. Opin. Immunol. 17, 510–516
(2005).
79. Abi-Rached, L. & Parham, P. Natural selection
drives recurrent formation of activating killer cell
immunoglobulin-like receptor and Ly49 from
inhibitory homologues. J. Exp. Med. 201, 1319–1332
(2005).
80. LaBonte, M. L., Hershberger, K. L., Korber, B. &
Letvin, N. L. The KIR and CD94/NKG2 families of
molecules in the rhesus monkey. Immunol. Rev.
183, 25–40 (2001).
81. Guethlein, L. A., Flodin, L. R., Adams, E. J. &
Parham, P. NK cell receptors of the orangutan
(Pongo pygmaeus): a pivotal species for tracking the
coevolution of killer cell Ig-like receptors with MHC-C.
J. Immunol. 169, 220–229 (2002).
82. Rajalingam, R., Parham, P. & Abi-Rached, L. Domain
shuffling has been the main mechanism forming new
hominoid killer cell Ig-like receptors. J. Immunol.
172, 356–369 (2004).
83. Khakoo, S. I. et al. Rapid evolution of NK cell receptor
systems demonstrated by comparison of chimpanzees
and humans. Immunity 12, 687–698 (2000).
84. Emes, R. D., Goodstadt, L., Winter, E. E. &
Ponting, C. P. Comparison of the genomes of human
and mouse lays the foundation of genome zoology.
Hum. Mol. Genet. 12, 701–709 (2003).
85. Selwood, L. & Johnson, M. H. Trophoblast and
hypoblast in the monotreme, marsupial and
eutherian mammal: evolution and origins. BioEssays
28, 128–145 (2006).
86. Koopman, L. A. et al. Human decidual natural
killer cells are a unique NK cell subset with
immunomodulatory potential. J. Exp. Med.
198, 1201–1212 (2003).
87. Bulmer, J. N. & Lash, G. E. Human uterine natural
killer cells: a reappraisal. Mol. Immunol. 42, 511–521
(2005).
88. Li, X. F. et al. Angiogenic growth factor messenger
ribonucleic acids in uterine natural killer cells.
J. Clin. Endocrinol. Metab. 86, 1823–1834 (2001).
89. Moffett, A., Regan, L. & Braude, P. Natural killer cells,
miscarriage, and infertility. BMJ 329, 1283–1285
(2004).
90. Pijnenborg, R., Robertson, W. B. & Brosens, I.
The arterial migration of trophoblast in the uterus of
the golden hamster, Mesocricetus auratus. J. Reprod.
Fertil. 40, 269–280 (1974).
91. Nanaev, A. et al. Physiological dilation of uteroplacental
arteries in the guinea pig depends on nitric oxide
synthase activity of extravillous trophoblast.
Cell Tissue Res. 282, 407–421 (1995).
92. Pijnenborg, R., Robertson, W. B., Brosens, I. &
Dixon, G. Trophoblast invasion and the establishment
of haemochorial placentation in man and laboratory
animals. Placenta 2, 71–91 (1981).
93. Finn, C. Menstruation: a nonadaptive consequence
of uterine evolution. Q. Rev. Biol. 73, 163–173
(1998).
94. Blankenship, T. N. & Enders, A. C. Modification of
uterine vasculature during pregnancy in macaques.
Microsc. Res. Technol. 60, 390–401 (2003).
95. Pijnenborg, R., D’Hooghe, T., Vercruysse, L. &
Bambra, C. Evaluation of trophoblast invasion
in placental bed biopsies of the baboon, with
immunohistochemical localisation of cytokeratin,
fibronectin and laminin. J. Med. Primatol.
25, 272–281 (1996).
Acknowledgements
The authors thank D. Antczak, G. Burton, S. Ellis, S. Murphy,
P. Parham, R. Pijneneborg, A. Sharkey and P. Wooding for
helpful comments.
Competing interests statement
The authors declare no competing financial interests.
DATABASES
The following terms in this article are linked online to:
Entrez Gene: http://www.ncbi.nlm.nih.gov/entrez/query.
fcgi?db=gene
CD40 ligand | CD80 | CD86 | CD95L | HLA-A | HLA-B | HLA-C |
HLA-E | HLA-G | KIR2DL4 | LILRB1 | NKG2A | PDL1
Access to this links box is available online.
www.nature.com/reviews/immunol