Strangers no more: Uterine NK cell recognition of the placenta in mice

Successful pregnancy requires a series of coordinated interactions between the developing placenta, the uterus, and the maternal immune system (1). At the maternal/fetal interface, where the placenta contacts the uterus, a specialized type of lymphocyte known as a uterine natural killer (uNK) cell is thought to play a critical role in mediating the uterine vascular adaptations to pregnancy. Epidemiological data have suggested that this function is modified by the response of uNK cells to the paternally inherited classical class I major histocompatibility complex (MHC I) molecules (i.e., the HLA-A, -B, and -C antigens in humans) expressed by placental trophoblasts invading into the uterus, as inopportune genetic combinations of the mother and father expected to alter uNK cell activity increase the risk of certain obstetrical complications (i.e., preeclampsia, fetal growth restriction, and recurrent miscarriage). In mice, however, direct evidence for such uNK cell-mediated “allorecognition” of the placenta has so far been lacking, and as a result the biomedical researcher's favorite model organism has yet to be used to study how levels of uNK cell activity influence reproductive success. In PNAS, Hemberger and colleagues now present strong evidence that murine uNK cells do indeed recognize paternal MHC I molecules expressed by invasive trophoblasts and that this recognition has significant implications for uterine vascular remodeling and ultimate pregnancy outcome (2). This work thus opens up use of the mouse as a means to study the key interactions that underlie some of the more serious complications of pregnancy.

Uterine NK Cells: Recent Advances

The work of Madeja et al. follows on a series of studies implicating uNK cells as important mediators of the developmental interactions between the placenta and the uterus that establish the placenta as a high-capacity nutrient and gas exchange organ. In both mice and humans, uNK cells have long been known to constitute the vast majority of leukocytes within the decidua, the specialized endometrial stromal tissue that directly contacts the placenta (3). Because of their expression of cytotoxic effector molecules, early studies focused on the potential immunological threat they posed to fetal survival. However, interest in an alternative possibility, namely that uNK cells have a trophic function, was sparked by the work of Croy and coworkers showing that uNK cells are required for decidual vascular remodeling in mice. In a series of seminal findings, this group found that genetically induced uNK cell deficiency leads to a failure of decidual arterioles to remodel into high-conductance vessels, the same vessels that are invaded by the endovascular trophoblasts that co-opt blood flow to the placenta (4).

Work in humans buttressed the idea that uNK cells were key to developmental events in early pregnancy and furthermore

Madeja et al. add the trophic consequences of uNK cell allorecognition of invasive trophoblasts to the list of pathways that can now be studied.

suggested that the activity of these cells was at some level regulated by trophoblasts themselves. The cells were shown to be quite distinct from peripheral NK cells and to express a variety of angiogenic factors and chemoattractants that could potentially influence decidual vascular remodeling and trophoblast invasion (5, 6). They were also shown to express certain nonpolymorphic receptors that would potentially allow them to respond to non-MHC ligands expressed by trophoblasts. Perhaps most importantly, however, in vivo evidence of a role for trophoblast–uNK cell interactions in regulating developmental events at the maternal/fetal interface came from the molecular epidemiological work by Moffett and coworkers. These researchers were studying the pathogenesis of preeclampsia, a mysterious disease seen during human pregnancy that is associated with hypertension, proteinuria, and systemic endothelial cell dysfunction. Although these complications are seen in the second and the third trimester, they are thought to be caused by events in the first trimester that lead to shallow trophoblast endovascular invasion and inadequate decidual vascular remodeling. Moffett and colleagues showed that there was an increased risk of preeclampsia when certain maternal haplotypes of killer Ig-like receptors (KIRs, which are polymorphic class I HLA-interacting receptors expressed by human NK cells) are juxtaposed during pregnancy with certain haplotypes of paternal HLA-C antigens (7). In a more recent study, Moffett and coworkers showed that this same KIR/HLA-C combination also increases the risk of fetal growth restriction and recurrent miscarriage (8). Besides their clinical importance and implication that different complications of pregnancy might share a common pathogenic mechanism, these findings make a strong case that uNK cells are capable of distinguishing polymorphic HLA molecules expressed by trophoblasts.

However, the above evidence that decidual vascular remodeling is regulated by uNK cell allorecognition of trophoblasts has presented a quandary for using mice to dissect underlying cellular and molecular pathways, as a number of studies have indicated that paternal MHC I expression by the developing mouse placenta in early to mid-gestation might be quite minimal (refs. 2 and 9 and references therein). Thus, Madeja et al. start their work reassessing this critical question. After first detecting expression of intact MHC I molecules on the surface of in vitro-differentiated trophoblast giant cells, a trophoblast subtype that contributes to the outer surface of the murine placenta, Madeja et al. next perform immunostaining of tissue sections to show that paternal MHC I is indeed detectable on the surface of these cells in vivo at embryonic day (E) 8.5, when uNK cells are starting to remodel the decidual vasculature. Surprisingly, of the two (H-2K and H-2D) polymorphic MHC I genes present in C57BL/6 (B6) mice, only H-2K is expressed by trophoblast giant cells. Thus, the mouse draws an interesting, and unexplained, parallel with the human, as HLA-C is the only detectable MHC I molecule expressed by invasive trophoblasts.

Next, Madeja et al. determine the spatial relationship between trophoblasts and uNK cells at the maternal/fetal interface. In contrast to later stages of gestation, when uNK cells are somewhat removed from the placenta, Madeja et al. find that uNK cells are in close contact with trophoblasts at the placenta's invasive front on E8.5. Moreover, they measure how many uNK cells on E8.5 expressed Ly49C, an NK cell receptor that binds H-2K^b (i.e., the b haplotype of the H-2K molecule, such as the one found in B6 mice), when BALB/c females are mated to either BALB/c males or B6 males. Because BALB/c mice bear the d haplotype at their H2 locus (which contains the MHC genes), BALB/c uNK cells would have exposure to H-2K^b molecules if they are mated to B6 but not BALB/c males. Strikingly, this former mating combination increases the proportion of uNK cells expressing Ly49C, thus implying a functional interaction between uNK cells and paternal MHC I molecules that could skew the uNK receptor repertoire.

Functional Implications

What are the consequences of these interactions in terms of decidual vascular remodeling and ultimate pregnancy success? To address this question, Madeja et al. first analyze decidual vessel size on E8.5 in different mating combinations of BALB/c and B6 mice. Strikingly, decidual vessel diameters in crosses of B6 males to BALB/c females are about twice as large as those seen in the syngeneic inbred crosses (i.e., B6 × B6 and BALB/c × BALB/c), as well as those seen in the reciprocal intercrosses, i.e., when BALB/c males are mated to B6 females. To determine whether this phenotype was MHC-linked, and thus potentially due to the presence of H-2K^b molecules in BALB/c females, Madeja et al. next use BALB.B mice, a congenic strain of mice that is identical to BALB/c mice except for an H2 locus derived from B6 mice. Remarkably, decidual vessel diameters in BALB.B male × BALB/c female crosses are still 1.5-fold increased compared with those in straight BALB/c × BALB/c crosses. Whereas these results suggest that non-MHC loci derived from the male strain could also influence decidual vessel diameter, they also imply that ~50% of the effect is attributable to the H2 locus itself.

Madeja et al. then ask how the impaired decidual vessel remodeling seen in the BALB.B male × BALB/c female mating combination affects pregnancy outcome at the end of gestation. Strikingly, the fetuses produced by this cross are larger than those produced by the BALB/c male × BALB/c female cross, and the effect is independent of litter size (a key potential confounder because large litters tend to have smaller-sized pups). Interestingly, placentas from allogeneic crosses are also larger, but this effect is much more apparent with small litters. Indeed, fetal/placental weight ratios are significantly increased in large litters. Together, these data suggest that allorecognition plays an important role in regulating fetal growth, most likely through effects on placenta size in small litters, as well as through effects on placental efficiency as litter size increases. In an elegant variation of the experiment, Madeja et al. also analyze fetal and placental weights in crosses between (BALB.B × BALB/c)F₁ males and BALB/c females. In these crosses, all concepti would be genetically identical within a litter except that some would have inherited an H2^b locus plus an H2^d locus and some only an H2^d locus. Remarkably, they observe the same effects on fetal and placental weights, now within single litters. These results thus rule out systemic factors as the cause of altered fetal and placental weights and instead strongly imply that these alterations are due to processes localized to each individual implantation site.

Together, the data of Madeja et al. provide direct evidence that uNK cells in mice detect paternal MHC I molecules expressed by invading trophoblasts and that this detection in turn has functional consequences for their ability to regulate decidual vessel remodeling and influence pregnancy outcome. This is a satisfying result that brings together decade-old work in mice and more recent developments in humans. Perhaps more importantly, however, the results of Madeja raise the possibility that mice might be useful to study certain events at the maternal/fetal interface that have been strongly implicated as causing some of the serious complications of pregnancy. One of the most pressing complications is preeclampsia, the disease discussed above with unknown etiology that affects up to 7% of all pregnancies. Whereas recent work has revealed some of the mechanisms whereby defective placentation in the first trimester leads to the manifestations of the disease when they emerge in the second and third trimester, the pathogenic events of the first trimester themselves have remained very obscure (10). Unfortunately, however, mouse trophoblasts do not invade nearly as deeply into the uterus as they do in humans, even though mice and humans show the same mode of hemochorial placentation. Thus, impairments of trophoblast invasion in mice would not be expected to lead to major changes in the absolute level of invasion and so may not affect final placental function to the extent they do in humans. Indeed, it has been very difficult to induce preeclampsia-like symptoms in mice. Nonetheless, the mouse provides a genetically tractable model organism, and it is likely that they can be used to learn a great deal about the basic biology of the maternal/fetal interface in respects that are highly relevant to human disease. The results of Madeja et al. add the trophic consequences of uNK cell allorecognition of invasive trophoblasts to the list of pathways that can now be studied.