Cutting Edge: Regulation of Uterine NKT Cells by a Fetal Class I Molecule Other Than CD11

The peri-implantation uterus contains an expanded population of NK1.1+ Vα14+ TCRint (NKT) lymphocytes. Although these cells bear the above features in common with other NKT cells populations in thymus, bone marrow, liver, and spleen, they differ from these other populations in terms of an altered Vβ repertoire and absence of a CD4+ component. In this study, we demonstrate that the uterine population also differs from other NKT cell populations because they recognize a class I/class I-like molecule other than CD1, whereas most previously described Vα14+ NKT cells are CD1-restricted. Moreover, the class I/class I-like molecule leading to the uterine NKT cell expansion may be supplied by the fetus. These data demonstrate a novel mechanism whereby the fetus is capable of modulating the maternal immune system.

A s the fetal trophoblasts invade the uterine wall at implantation, maternal immune cells, either circulating or resident within the uterine lining (decidua), may be exposed to fetal Ags. The ensuing maternal immune response and its potential positive or negative effects upon the developing fetus are of great interest in perinatal medicine. It has been hypothesized that the fetus may in fact modulate the nature and constituents of this immune response, and recently examples of this phenomenon have been documented. For example, human trophoblasts express HLA-G, an oligomorphic class Ib molecule that appears to regulate maternal macrophages, NK, and T cells at the maternal-fetal interface. (1)(2)(3) Additionally, in separate transgenic murine models, T cells expressing the transgenic TCR specific for either a paternally derived Ag (H-Y) (4) or a paternally derived class I molecule (K b ) (5) were both deleted and tolerized in an Ag-specific fashion during pregnancy, possibly via placental expression of Fas ligand (5,6).
Recently, we demonstrated a large increase in uterine NK1.1 ϩ V␣14 ϩ CD4 Ϫ /CD8 Ϫ (NKT) cells in the maternal decidua at peri-implantation. (7) Here, we demonstrate that a fetal Ag regulates this increase. This Ag is most likely a class I/Ib molecule since ␤ 2 -microglobulin-deficient (␤ 2 m Ϫ/Ϫ ) 3 mice do not have increased numbers of NKT cells at peri-implantation. However, the Ag does not appear to be CD1, the class I-like molecule recognized by NKT cells in other tissues, as CD1 Ϫ/Ϫ mice have normal numbers of decidual NKT cells.

Mice
Pathogen-free CD1 ϩ/Ϫ C57BL/6 and BALB/c (8) mice were obtained from the laboratory of C.-R. Wang at the University of Chicago (Chicago, IL), bred in the animal care facility at Swedish Medical Center (Denver, CO), and screened, as described, by PCR to obtain CD1 ϩ/ϩ and CD1 Ϫ/Ϫ mice of each strain for use in additional experiments (8). Six-to 8-wk-old ␤ 2 m Ϫ/Ϫ C57BL/6 and BALB/c (9) mice were obtained commercially from The Jackson Laboratory (Bar Harbor, ME). All experiments were approved by the HealthONE Institutional Animal Care and Utilization Committee, which acts for the Swedish Medical Center.
For cytofluorographic analysis, cells (0.1-1 ϫ 10 6 /well in 96-well plates) were preincubated with rat anti-mouse Fc␥R Ab (24G2, 1 g/ml) (14) or undiluted normal mouse serum to block nonspecific binding and one-or two-color staining was performed. Cells were analyzed by flow cytometry using a Becton Dickinson FACSort cytometer (Becton Dickinson, San Jose, CA), and data plots were generated using the CellQuest version 1.2 software package supplied by the manufacturer. In all experiments, set up and calibration were performed with nylon wool prepared splenocytes, and these cells were also used to set gates for thymocytes and hepatic and placental/decidual lymphocytes. In all cases, appropriate negative control experiments were performed to verify staining specificity.

A fetal ␤ 2 m-associated molecule regulates expansion of placental/decidual NKT cells
The experiments performed indicate that the expansion of placental/decidual NKT cells largely relies on the expression of a ␤ 2 massociated molecule other than CD1. The next experiment was designed to assess whether fetal (placental) expression of the ␤ 2 massociated molecule is sufficient for expansion of placental/decidual NKT cells. To address this question, we bred C57BL/6 ␤ 2 m Ϫ/Ϫ females with WT BALB/c males and analyzed NKT cells from the resulting gestations. Although thymic, splenic, and hepatic NKT cell numbers remained at the very low levels seen in (␤ 2 m Ϫ/Ϫ ϫ ␤ 2 m Ϫ/Ϫ ) gestations, the numbers of NKT cells from these (␤ 2 m Ϫ/Ϫ ϫ WT) pregnancies were restored to those of WT matings. Typical staining data are shown in Fig. 3. Thus, even in the absence of maternal ␤ 2 m, fetal ␤ 2 m derived from the paternal genome is sufficient to expand the placenta/decidual NKT cell population. These data are summarized in Fig. 4.
To determine whether paternal ␤ 2 m expression is necessary for the expansion of placenta/decidua NKT cells, (WT ϫ ␤ 2 m Ϫ/Ϫ ) gestations were analyzed and normal numbers of NKT cells were found (data not shown). Thus, paternal ␤ 2 m expression is sufficient, but not necessary, for NKT cells expansion.

Discussion
Multiple studies in recent years have characterized a unique cell population known as NKT cells. (16,20,21). These cells, found primarily in thymus, bone marrow, spleen. and liver, are characterized by several unique features. In addition to coexpression of NK1.1 and ␣␤-TCR, they uniformly express V␣14-J␣281 with a canonical VJ junction, have a limited V␤ repertoire (predominantly V␤8.2 in addition to smaller proportions of V␤2 and V␤7), express TCR at an intermediate level, secrete large amounts of IL-4 upon engagement of the TCR (8,22), and recognize the class I-like CD1 molecule. (15,23). In a recent study, we added the peri-implantation pregnant murine uterus to the list of sites with a large NKT cell population (7). In that study, we speculated that NKT cells might play an important role at peri-implantation by inducing a Th2-type environment via elaboration of IL-4. Such a Th2 environment has been shown to be critical for reproductive success (24 -27).
Ito et al. (28) have also recently confirmed the presence of NKT cells in the pregnant uterus. These authors speculated that NKT cells might mediate spontaneous abortion based upon experiments involving in vivo stimulation of NKT cells with ␣-galactosylceramide, a known ligand of V␣14 ϩ NKT cells. The physiological significance of these findings remains to be demonstrated. In our description of NKT cells in the pregnant uterus and their expansion at peri-implantation, we noted two features that suggested these cells might develop independently from NKT cells in other organs. First, unlike other populations of NKT cells that contain both CD4 ϩ and CD4 Ϫ /CD8 Ϫ (DN) components, uterine NKT cells are entirely CD4 Ϫ . Additionally, although uterine NKT cells are V␣14 ϩ , their V␤ repertoire is different from other organs, with an apparent predominance of V␤3 ϩ cells and few V␤8 ϩ cells. 5 We did note expression of CD1 within the pregnant uterus and speculated that these cells might nonetheless recognize CD1. Although experiments have not yet been done directly assessing cognate recognition of CD1 by uterine NKT cells, the experiments reported herein suggest that development of this uterine population is regulated by a class I/class I-like molecule other than CD1.
Additionally, these findings indicate that the relevant ␤ 2 m-associated molecule can be supplied by the paternal genome, presumably via placental expression. In the HLA-G experiments discussed in the introduction, HLA-G likely mediates its effects via placental expression, as its expression is highest in the trophoblasts (29). It seems likely that the relevant fetal Ags (H-Y, K b ) in the  transgenic models also mediate the observed effects via placental expression, as the placental trophoblast are the cells are most likely to be encountered by maternal immune cells. In this regard, it is interesting to note that the paternal genome is preferentially expressed in the placental tissues, presumably due to imprinting (30). Since we have not specifically identified the ␤ 2 m-associated molecule that regulates the expansion of uterine NKT cells, we cannot yet assess whether it is imprinted or not. We did observe that maternal ␤ 2 m expression is capable of regulating uterine NKT cell expansion. Whether this effect is mediated by placental expression of the relevant ␤ 2 m-associated maternal allele or by expression by adjacent maternal tissues is unknown at present.
In these experiments, as in previous reports, thymus, liver, and spleen of ␤ 2 m Ϫ/Ϫ females contain very few NKT cells (16 -19, 23). Similarly, the uteri of pregnant ␤ 2 m Ϫ/Ϫ females contain very few NKT cells when crossed with ␤ 2 m Ϫ/Ϫ males. In contrast, when crossed with ␤ 2 m ϩ/ϩ males, normal numbers of uterine NKT cells are found, while numbers of NKT cells in the other organs remain very low. Current data support both thymic and extrathymic development of NKT cells (reviewed in Ref. 23). It is unclear from these data whether the local expansion of uterine NKT cells represents de novo development of uterine NKT cells from a resident progenitor or a local expansion of a small resident pool. The data do not support thymic development and migration. This model may provide an interesting system for the study of NKT cell development.
Interactions between the mother and fetus have long been known to occur via a variety of steroid and peptide hormones. These data document another mechanism whereby maternal-fetal interaction occurs via their immune systems. Unlike similar situations that occur in tumor and transplant models, for example, these phenomena take place within the physiologic setting of the hemochorial placenta and emphasize the specialized nature of maternal-fetal immunology.