Professional Documents
Culture Documents
Moffett2004 Immunological Paradox PDF
Moffett2004 Immunological Paradox PDF
doi:10.1016/S0143-4004(03)00167-X
CURRENT TOPIC
The survival of the allogeneic conceptus has long been an immunological paradox. Medawar was the first to propose an evasive
mechanism based on the concept of self/non-self recognition described in classical transplantation immunology. Since then,
several newer models of self/non-self recognition have been proposed, such as the PAMP/PRR system, the Missing Self and the
Danger Hypothesis. The present paper considers the fetal-maternal relationship in the context of all these models. The conclusion
reached is that none of them is really appropriate because the interface between trophoblast cells of the fetal placenta and the
leukocytes of the maternal decidua is unique. Pregnancy is not simply a case of acceptance or rejection like a transplant. The
immunological mechanism must provide a balanced environment whereby the conceptus is nurtured by the mother and yet
prevented from excessive invasion. Future identification of trophoblast ligands and their respective receptors on uterine Natural
Killer cells and other leukocytes is likely to offer the best insight as to how this symbiotic state is achieved.
Placenta (2004), 25, 1–8 2003 Elsevier Ltd. All rights reserved.
INTRODUCTION because there are many features about pregnancy which are
unique and which are not found in any other immunological
In the light of the discovery of the Major Histocompatibility situation.
Complex (MHC) and its role in transplantation, the seminal
essay written by Medawar in 1953 drew a logical comparison
between an allograft and a fetus (Medawar, 1953). Despite
both being non-self, the fetus grows and yet the transplant IMMUNOLOGICAL MODELS
is rejected. Medawar himself performed no experiments to The classical self/non-self model formulated by Burnet and
address this conundrum but suggested that the placenta must Medawar proposed that each lymphocyte expresses a single
play a central role in fetal acceptance as it is the trophoblast receptor for a foreign antigen and signalling through this
cells which interface with the mother. However, his other triggers an immune response (Burnet, 1959) (Figure 1a).
proposals—maternal immunosuppression and immaturity of Lymphocytes with receptors for non-self in the form of
fetal antigens—together with subsequent diverse hypotheses pathogens and allogeneic cells will be present and, if these are
by others have not withstood the test of time. Why is encountered, elimination of the infectious agent or rejection of
the solution to this problem so elusive? We believe that the the allograft results (Billingham, Brent and Medawar, 1953).
analogy using self/non-self discrimination to compare the The random generation of these clonally distributed receptors
graft/host and placental/maternal relationship has become a by gene rearrangement during fetal development means that
misleading framework in which to conduct experiments. the T and B lymphocytes which acquire receptors reactive to
Recently, alternative concepts have been proposed for how antigens present in the fetus need to be deleted. In this way
the immune system operates to maintain the integrity of the tolerance to self and infectious or other stimuli encountered
host (Figure 1). In this essay we consider whether these before birth is acquired.
newer models are more appropriate to explain the success of The revelation that it is not only T and B cells which have
the maternal-fetal interaction. We conclude that although receptors for non-self has now changed our views of self/non-
many of these do offer valuable insights none of them, self recognition. Medzhitov and Janeway have shown that
as they stand, explain Medawar’s immunological paradox microbes can be detected by the conserved molecular patterns
that are essential to their physiology but not a part of their
a
To whom correspondence should be addressed. Tel.: +44-1223- vertebrate hosts (Figure 1b). These are known as pathogen
333727; Fax: +44-1223-765065; E-mail: am485@cam.ac.uk associated molecular patterns (PAMPs) (Janeway, 1992;
0143-4004/04/$–see front matter 2003 Elsevier Ltd. All rights reserved.
2 Placenta (2004), Vol. 25
Medzhitov and Janeway, 2002). This system permits the microbes. Importantly, the receptors for PAMPs, known as
host to distinguish between microbial non-self and normal pattern recognition receptors (PRR), are germ line encoded
self but as all microbes posses PAMPs it will not allow and are mainly found on antigen presenting cells
discrimination between pathogenic and non-pathogenic (APC)—dendritic cells and macrophages. Once a PRR is
Moffett and Loke: Reproductive Immunology Revisited 3
(Georgiades et al., 2001), endogenous retroviral products (Krieg, 2000; Ohashi et al., 2000). Although bacteria lack CpG
(Mi et al., 2000) and oncofetal proteins (Hammarstrom, 1999). methylation enzymes most mammalian DNA is methylated.
The EVT has a genetic contribution from the father and is Perhaps, the unmethylated DNA characteristic of extraembry-
thus an allogeneic normal cell in the mother, but unlike other onic trophoblast could act as a PRR ligand mimicking
situations where allogeneic cells come into contact, such as bacteria.
transplantation, there are important differences. The highly Furthermore, there is the intriguing finding that human
polymorphic HLA class I (HLA-A and HLA-B) and class II endogenous retroviral (HERV) products are expressed by
molecules displayed by somatic cells which initiate recognition trophoblast (Taruscio and Mantovani, 1988; Harris, 1998) so
of the allograft and its subsequent rejection are not displayed in some respects it could be considered an ‘infected’ cell and
by trophoblast cells, although the less polymorphic HLA-C is might be discerned by PRR such as Toll-like receptor (TLR)
present (Moffett-King, 2002). It is therefore, not a transplant 3 which binds double-stranded (ds) RNA (Akira and Hemmi,
in terms of Burnet and Medawar’s original concept of self/ 2003). The HERV could act in other ways to influence the
non-self because MHC non-self is not present. Furthermore, it maternal immune system. For example, they might provide
cannot be considered ‘missing self’ because it does have a trophoblast specific peptides for HLA class I molecules or
genetic contribution from the mother. EVT expresses an encode MHC-like molecules that bind NK receptors. These
unusual combination of three HLA class I molecules two of are speculative ideas but there is support from a protein,
which (HLA-E and HLA-C) are known to be dominant EBI-3, whose expression is induced in EBV-transformed B
ligands for NK cells. Indeed HLA-E is virtually monomorphic cells (Devergne et al., 2001). This protein is normally
and would be perceived as ‘self’ by maternal NK cells expressed at high levels in the placenta and interestingly, the
expressing the inhibitory receptor CD94/NKG2A thus dominant peptide eluted from trophoblast HLA-G is derived
preventing cytotoxicity (King et al., 2000). from this molecule (Lee et al., 1995). Other peptides predomi-
Matzinger has considered the problem of the fetal allograft nate if HLA-G is expressed in cell lines and thus the cell
in terms of the Danger Model and has suggested that the expressing HLA-G may alter binding to its receptor. EBI-3
difference in outcome is because the surgical trauma and can also form heterodimers with members of the IL-12 family
hypoxia of the iatrogenic transplant situation results in tissue (Devergne, Birkenbach and Kieff, 1997; van Seventer, Nagai
damage and necrosis, whereas as pregnancy is a natural and van Seventer, 2002). This trophoblast derived IL-12-like
phenomenon, with intermingling of normal healthy cells, cytokine might direct the nature of the maternal immune
alarm signals are not sent (Matzinger, 2002). However, at a response by acting on uterine DC, T or NK cells.
cellular level the placental bed is not a harmonious site, indeed
it has been described as a ‘battle-ground’ (Robertson, 1987).
There is necrosis at the uterine surface where the anchoring THE MATERNAL UTERINE IMMUNE SYSTEM
villi attach (Nitabuch’s layer) as well as the characteristic
‘fibrinoid’ necrosis of the spiral arterial media (Kaufmann and The maternal uterine immune system is thus meeting an
Burton, 1994). The trophoblast cells around these arteries have apparently ‘normal’ but invading and selectively destructive
a spidery appearance when stained for cytokeratin suggesting extraembryonic cell with unmethylated DNA. It has
cell damage (Loke and King, 1995). Thus, both fetal and distinctive expression of MHC molecules, endogenous retro-
maternal cells are damaged although it is remarkable how the viral products and other placental-specific molecules and
invasion through the decidua basalis between the spiral arteries hormones. It will not readily fit into any category of
is accompanied by relatively little damage compared to that non-self, allogeneic, foreign or infected. However, it may be
seen with tumour invasion. It is only at specific anatomical ‘alarming’ to the mother by the damage wrought in the decidua
sites in the placental bed that necrosis is seen. This battle- and spiral arteries. How does the mother discern the presence
ground cannot be considered ‘danger’ free and from the of these trophoblast cells in the uterus and what is the
viewpoint of the Danger Model leukocytes should be alerted. response?
It would appear that viewing the fetus as an allograft in the The absence of B cells and paucity of T cells in the uterine
terms of the Self/non-self, Danger or Missing Self hypotheses mucosa suggests that the adaptive immune system is not as
is not illuminating and may even be distracting. important as the innate system. Indeed, NK cells and macro-
At first sight it would seem non-self PAMP recognition by phages are the main types of maternal leukocyte present and
APC is also not relevant to pregnancy since this was proposed NKT cells have also been described (King, 2000; Boyson et al.,
as a mechanism to distinguish between non-infected self and 2002). Do any of these leukocytes have receptors which could
microbial non-self. Clearly trophoblast is not microbial. How- perceive trophoblast? Despite the emphasis placed in the
ever, PRR may have evolved initially to bind ligands present literature on the necessity for the avoidance of maternal T cell
in the host and have then been subsequently co-opted by reactivity during pregnancy it has always been hard to define
microbes to gain entry into target cells. Precedents for this are how T cells might directly react to human trophoblast given
well known and include HIV binding to CD4 and EBV to the absence of HLA class II and HLA-A and HLA-B
CR2. Indeed, several PRR have been shown to bind to molecules. Although HLA-C is polymorphic, it has low levels
endogenous ligands such as Hsp and DNA CpG sequences of surface expression and HLA-C-reactive T cells are rare in
Moffett and Loke: Reproductive Immunology Revisited 5
transplant recipients. Indirect recognition by uptake of other by both gene content and polymorphism, means unrelated
trophoblast molecules by decidual APC would appear more individuals rarely have the same KIR genotype (Vilches and
likely and in this case the ability of the decidual DC to tolerize Parham, 2002). The KIR gene family includes inhibitory
or stimulate will be crucial. Characterization of decidual DC receptors (KIR2DL1, KIR2DL2, KIR2DL3) that have
reveals they are an immature myeloid type similar to those specificities for polymorphic determinants of all HLA-C allo-
found at other body surfaces such as the dermis, lung or gut types; the latter are known as the KIR epitopes of HLA-C. All
but their functional capabilities are still unknown (Gardner HLA-C allotypes have either asparagine or lysine at position
and Moffett, 2003). 80 of the 1 domain and this dimorphism distinguishes the two
Decidual NKT cells have also been identified at appreciably KIR epitopes. Potential NK cell alloreactions can occur if a
higher numbers than in the peripheral blood. These cells NK cell from one individual has a receptor for an HLA-C KIR
express an invariant TCR capable of binding to an MHC-like epitope that is lacking on cells from another individual (miss-
molecule, CD1D. Definitive confirmation that CD1D is ing self). This is a quite different form of non-self recognition
expressed by trophoblast is still lacking but other ligands could based on NK cell KIR and not TCR or PRR expressed by T
be present that are capable of binding to these cells (Boyson cell and APC respectively.
et al., 2002). As yet there is virtually no information on the The importance of NK cell allorecognition is seen in bone
possible expression of PRR such as Toll-like receptors (TLR) marrow transplantation where the dominant pattern of NK cell
on DC, macrophages and other leukocytes in the decidua alloreactivity has been shown to be due to the two HLA-C
nor is it known whether stress molecules are induced on epitopes (Parham and McQueen, 2003). Indeed, NK cell
maternal and trophoblast cells by the placental invasion and alloreactions are beneficial in patients with acute myeloid
destruction. leukaemia (AML) when HLA-mismatched haploidentical
NK cells are heterogenous due to differential expression of transplants are performed. The donor is a family member who
various inhibitory and activating receptors. Decidual NK cells is identical for one HLA haplotype and mismatched for the
express a wide array of receptors but the ligands for only a few other, a situation with obvious resonance to pregnancy. Some
of these are known. However, a mechanism by which maternal of the donor NK cells may possess a KIR receptor for an
leukocytes could directly recognize trophoblast has now been HLA-C epitope lacking in the recipient but present in donor.
identified. This has been demonstrated for the receptor for These NK cells are not inhibited by ‘self’ MHC and kill the
HLA-E, CD94/NKG2A, by using HLA-E tetramers which leukaemic cells of the recipient (Ruggeri et al., 2002). In
bind decidual NK cells (King et al., 2000). How HLA-G, the this scenario, therefore, NK cell allorecognition is actually
only trophoblast-specific HLA class I molecule, is discerned beneficial. Could the same be true in pregnancy?
is still controversial. Besides members of the ILT family Not all KIR with specificities for HLA-C KIR epitopes are
expressed by decidual macrophages there are likely to be inhibitory. For example, KIR2DS1 is an activating receptor
receptors for HLA-G on NK cells such as KIR2DL4, a that binds all the HLA-C allotypes which have lysine at
member of the Killer-cell immunoglobulin-like receptor (KIR) position 80 (HLA-Clys80). A possible situation that could arise
family (Allan et al., 1999; Rajagopalan and Long, 1999). is the presence of both KIR2DS1 in the recipient (mother) and
Trophoblast ligands for another KIR, KIR3DL3, the mRNA its ligand, the HLA-Clys80 KIR epitope, in the donor (father).
of which is detectable in uterine but not blood CD56dim NK This KIR epitope would be absent in the mother. Maternal
cells (unpublished), and for activating NK receptors such as NK cell activation could then occur but only in this particular
NKG2D, NKp30 and NKp46 await identification. maternal KIR paternal HLA-C combination.
The potential for NK cell alloreactions to occur during
pregnancy is, therefore, present and this has not been consid-
NK CELL ALLORECOGNITION ered by any of the four models described earlier in this essay.
Maternal NK cell allorecognition of the fetus may not only be
In many respects unravelling the interaction between HLA-C important at the implantation site, but could also influence the
and the KIR2D receptors present on decidual NK cells is engraftment or clearance of fetal leukocytes which may occur
likely to be particularly rewarding because unlike HLA-E when fetal cells are spilt into the maternal circulation at
and HLA-G, HLA-C is polymorphic and the father will delivery. Persistence of fetal cells can be associated with later
therefore assert his genetic difference in the trophoblast of his onset of diseases in the mother such as scleroderma (Nelson,
fetus by donating one of his HLA-C allotypes. In terms of 2002).
allorecognition in the context of reproduction, trophoblast
non-self HLA-C is likely to be of particular importance
(Figure 3). MATERNAL IMMUNE EFFECTOR
The KIR gene family is also polymorphic with variation in MECHANISMS
the number of KIR genes present in different individuals at the
KIR locus on chromosome 19. The combination of genes on A separate consideration from the question of how the
one chromosome is the KIR haplotype. There is also allelic maternal immune system detects the presence of trophoblast is
polymorphism of individual genes and this diversity, created the type of immune response that is generated. The CD56bright
6 Placenta (2004), Vol. 25
Figure 3. NK allorecognition.
NK cells produce an array of cytokines but little is known DC will either stimulate or tolerize T cells after migration
about how the profile alters in the shift from the non-pregnant from the periphery to the lymph nodes depending on the
to pregnant state and after interaction with trophoblast. Atten- signals received at their original surface location. Stimulation of
tion needs to be directed to the distinctive nature of the maternal T cells would not be expected without signals to the
decidual tissue which must guide both innate and adaptive DC inducing costimulatory molecules. Interestingly, many of
effector immune responses. Apart from the unique compos- the factors known to induce tolerogenic DC are also integral to
ition of leukocytes, decidua is characterized by the presence of the decidualizing process and include Vitamin D3, and PGE-2
plump stromal cells secreting copious amounts of proteins (e.g. (Harizi et al., 2002; Griffin and Kumar, 2003). With regard to
renin, IGFBP-1, prolactin) and extracellular matrix proteins NK cells the unique microenvironment will be important in the
(King, 2000). How will this decidual microenvironment drive proliferation and differentiation of these cells in utero. The
the immune responses initiated there? difference in phenotype of decidual NK cells compared with
Moffett and Loke: Reproductive Immunology Revisited 7
their peripheral blood counterparts, particularly with regard to lymphocytes can direct both inflammatory responses and also
the difference in KIR repertoire, is likely to be induced by the promote tissue healing (Born and O’Brien, 2002).
tissue they inhabit (King 2000; King et al., 2000). Our conclusion is that it is unwise to design experiments
exploring the nature of the immunological paradox of
pregnancy based on any of the existing paradigms. The
CONCLUSION pregnant situation is a biological conundrum that should be
studied with an open mind exploring further the unusual
The local decidual immune system must contribute to the
features of both the trophoblast and the maternal uterine
nurturing environment of the uterus which functions to mould
immune system. It is particularly important to move away
the placental cells as they migrate, allowing the optimum
from the view that the trophoblast is like a conventional
delivery of nutrients to the fetus without endangering the
allograft that must resist rejection. Instead, we should consider
mother. The idea that immune responses can be nurturing and
that the maternal immune response may be providing a
beneficial has come from another tissue, the skin, where epi-
nurturing balanced environment which curbs excessive or
dermal T cells with an invariant receptor, bind a self mol-
unsocial behaviour by both placenta and mother leading to
ecule induced by the damage from wounding. The response is
a state of peaceful coexistence between the two allogeneic
to generate cytokines (e.g. KGF) that stimulate epidermal
tissues.
growth and healing. It appears in this situation that these tissue
ACKNOWLEDGEMENTS
This paper was based on a lecture given by Professor Y.W. (Charlie) Loke at the IFPA meeting held in Melbourne, Australia 2002.
Nelson JL (2002) Microchimerism: incidental byproduct of pregnancy or Rossant J, Sanford JP, Chapman VM & Andrews GK (1986) Under-
active participant in human health? Trends Mol Med, 8, 109–113. methylation of structural gene sequences in extraembryonic lineages of the
Ohashi K, Burkart V, Flohe S & Kolb H (2000) Heat shock protein 60 is a mouse. Dev Biol, 117, 567–573.
putative endoegenous ligand of the toll-like receptor-4 complex. J Rossant J & Cross JC (2001) Placental development: lessons from mouse
Immunol, 164, 558–561. mutants. Nat Rev Gen, 2, 538–548.
Ohgane J, Hattori N, Oda M, Tanaka S & Shiota K (2002) Differentiation Ruggeri L, Capanni M, Urbani E, Perruccio K, Shlomchik WD, Tosti
of trophoblast lineage is associated with DNA methylation and A, Posati S, Rogaia D, Frassoni F, Aversa F, Martelli MF & Velardi
demethylation. Biochem Biophys Res Commun, 290, 701–706. A (2002) Effectiveness of donor natural killer cell alloreactivity in mis-
Oldenburg P-A, Zheleznyak A, Fang Y-F, Lagenaur CF, Gresham HD matched hematopoietic transplants. Science, 295, 2097–2100.
& Lindberg FP (2000) Role of CD47 as a marker of self on red blood
Taruscio D & Mantovani A (1988) Human endogenous retroviral se-
cells. Science, 288, 2051–2054.
quences: possible roles in reproductive physiopathology. Biol Reprod, 59,
Parham P & McQueen KL (2003) Alloreactive killer cells: hindrance and
713–724.
help for haematopoietic transplants. Nat Rev Immunol, 3, 108–122.
Potgens AJ, Schmitz U, Bose P, Versmold A, Kaufmann P & Frank HG van Seventer JM, Nagai T & van Seventer GA (2002) Interferon-beta
(2002) Mechanisms of syncytial fusion: a review. Placenta, Suppl A, differentially regulates expression of the IL-12 family members p35, p40,
S107–S113. p19 and EBI3 in activated human dendritic cells. J Neuroimmunol, 133,
Rajagopalan S & Long EO (1999) A human histocompatibility leukocyte 60–71.
antigen (HLA)-G-specific receptor expressed on all natural killer cells. J Vilches C & Parham P (2002) KIR: diverse, rapidly evolving receptors of
Exp Med, 189, 1093–1100. innate and adaptive immunity. Ann Rev Immunol, 20, 217–251.
Robertson WB (1987) Pathology of the pregnant uterus. In Haines and
Taylor, Obstetrical and Gynaecological Pathology (Ed.) Fox H, pp. 1149–
1176. Churchill Livingstone.