Cd1d and Invariant NKT Cells at the Human Maternal–Fetal Interface

CD1d and invariant NKT cells at the human maternal–fetal interface

Jonathan E. Boyson*, Basya Rybalov*, Louise A. Koopman*, Mark Exley†, Steven P. Balk†, Frederick K. Racke‡, Frederick Schatz§, Rachel Masch§, S. Brian Wilson¶, and Jack L. Strominger*¶ʈ

*Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138; †Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 02115; ‡Department of Pathology, Johns Hopkins University, Baltimore, MD 21287; §Department of Obstetrics and Gynecology, New York University Medical Center, New York, NY 10016; and ¶Cancer Immunology and AIDS, Dana Farber Cancer Institute, Boston, MA 02115

Contributed by Jack L. Strominger, August 15, 2002 Invariant CD1d-restricted natural killer T (iNKT) cells comprise a selves play an active immunological role in regulating the small, but significant, immunoregulatory T cell subset. Here, the immune response to the fetal allograft because they express presence of these cells and their CD1d ligand at the human immunologically relevant molecules such as IL-10 (25, 26) and maternal–fetal interface was investigated. Immunohistochemical granulocyte͞macrophage colony-stimulating factor (GM-CSF) staining of human decidua revealed the expression of CD1d on receptor (27). both villous and extravillous trophoblasts, the fetal cells that Here, we demonstrate in humans the expression of CD1d on invade the maternal decidua. Decidual iNKT cells comprised 0.48% both villous and invasive extravillous fetal trophoblasts. In of the decidual CD3؉ T cell population, a frequency 10 times addition, human decidual iNKT cells, present at a frequency Ϸ10 greater than that seen in peripheral blood. Interestingly, decidual times that seen in peripheral blood, exhibited a marked Th1-like CD4؉ iNKT cells exhibited a striking Th1-like bias (IFN-␥ produc- cytokine bias as well as a striking polarization toward GM-CSF -tion), whereas peripheral blood CD4؉ iNKT clones exhibited a production. These data suggest that decidual iNKT cell inter Th2-like bias (IL-4 production). Moreover, compared to their pe- actions with CD1d-expressing trophoblasts play a specific role at ripheral blood counterparts, decidual iNKT clones were strongly the maternal–fetal interface, possibly in the acceptance of the polarized toward granulocyte͞macrophage colony-stimulating fetal allograft and͞or in placental development. factor production. The demonstration of CD1d expression on fetal trophoblasts together with the differential pattern of cytokine Materials and Methods expression by decidual iNKT cells suggests that maternal iNKT cell Abs and Fluorescence-Activated Cell Sorter (FACS) Analysis. The interactions with CD1d expressed on invading fetal cells may play following Abs were used in FACS analysis and immunohisto- an immunoregulatory role at the maternal–fetal interface. chemistry: 6B11 (an mAb specific for the invariant V␣24J␣Q CDR3 loop (M.E., F.K.R., J.E.B., J.L.S., S.P.B., and S.B.W., ␣ nvariant CD1d-restricted natural killer T cells (hereafter unpublished results); anti-CD1d 42.1 (28); anti-V 24 and anti- ␤ Idesignated as iNKT cells) appear to play an important immu- V 11 (Coulter); anti-CD3 used in proliferation assays, T3D; noregulatory role in the immune system. iNKT cells possess anti-CD4, anti-CD8, anti-CD16, anti-CD56, anti-CD57, anti- ␣ ␣ ␣ CD69, anti-CD161, anti-CD45RO, anti-CD45RA, anti-CD94, invariant TCR chains, in which a V 24J Q junction is formed ␥ without N or P additions, preferentially paired with noninvariant IgG isotype controls, anti-IFN- , anti-IL-4, anti-GM-CSF, anti- V␤11 TCR␤ chains (1). An endogenous natural activating ligand IL-10, and anti-IL-2 were all from BD Biosciences (Los Ange- has yet to be identified, although the marine sponge-derived les); NOR3.2 (BioSource International, Camarillo, CA), and glycolipid ␣-galactosylceramide (␣GalCer), when bound and anti-c-erbB-2 (NCL-CB11; NovoCastra, Newcastle, U.K.). For presented by CD1d, activates iNKT cells (2, 3). Activated iNKT FACS staining, cells were washed in staining buffer [PBS (pH cells are able to secrete a wide variety of cytokines and this may, 7.2) supplemented with 2% FCS and 0.1% sodium azide]. Cells in part, explain their ability to exert their effects over a wide were incubated with mAb on ice for 30 min, then washed twice immunological spectrum, such as in the onset of diabetes in with staining buffer. humans and in NOD mice (4–8), the clearance of certain tumors (9, 10), the immune response to certain viruses (11–13), and as Tissues and Cell Culture. Peripheral blood mononuclear cells were mediators of anterior chamber acquired immune deviation obtained from leukocytes discarded from aphoresis or from (ACAID) in which specific tolerance is generated to antigens peripheral blood obtained by venipuncture. First trimester de- introduced into the anterior chamber of the eye (14, 15). cidua from patients undergoing elective abortion were collected Notably, iNKT cells also have been implicated in allograft at the New York University Medical Center, New York, and the Brigham and Women’s Hospital, Boston. Decidual tissue was survival (16–19). ␮ ͞ During pregnancy, hemi-allogeneic fetal trophoblast cells washed extensively in PBS supplemented with 50 g ml genta- migrate from the chorionic villi and invade the maternal decid- micin before mincing with sterile scissors. Decidual lymphocytes ual tissue, eventually invading the maternal spiral arteries. The were released by digesting the tissue with 0.1% collagenase and mechanisms that regulate trophoblast invasion remain poorly 0.05% DNase I (both from Sigma). Lymphocytes were purified defined. Extravillous trophoblasts do not express HLA-A or -B by density gradient centrifugation (Ficoll-Hypaque; Amersham molecules (20), but they do express HLA-C (21) as well as Pharmacia and Upjohn), and cells were allowed to adhere to HLA-G (22), a nonclassical MHC class I molecule whose tissue culture plates for 2–18 h at 37°C in a humidified 5% CO2, 37°C incubator. iNKT cell clones were derived by FACS sorting function is unknown. In mice, active induction of maternal ␣ ϩ ␤ ϩ ϩ ␣ ϩ tolerance to trophoblast-expressed antigens has been observed either V 24 V 11 or 6B11 V 24 T cells at 1 cell per well in transgenic models (23) and is probably the result of multiple in 96-well round-bottom plates. Peripheral blood mononuclear immunological mechanisms present at the maternal–fetal interface to prevent rejection of the fetus, such as the regulation of Abbreviations: iNKT cell, invariant natural killer T cell; ␣GalCer, ␣-galactosylceramide; tryptophan catabolism by indoleamine 2,3-dioxygenase in de- GM-CSF, granulocyte͞macrophage colony-stimulating factor; FACS, fluorescence- cidual macrophages and the expression by fetal trophoblasts of activated cell sorter. IMMUNOLOGY complement regulatory proteins (24). Trophoblasts may them- ʈTo whom correspondence should be addressed. E-mail: [email protected]. www.pnas.org͞cgi͞doi͞10.1073͞pnas.162491699 PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21 ͉ 13741–13746 Fig. 1. CD1d is expressed on human trophoblasts. Immunohistochemistry on paraformaldehyde-fixed decidual tissue (8 wk; 6-␮m sections). (A) Hematoxylin͞ eosin staining of the maternal–fetal interface where trophoblasts (arrow) have migrated from chorionic villi toward the decidua. The eosinophilic (pink) material is fibrin deposited at the maternal–fetal interface. Staining of serial sections demonstrated that CD1d was expressed on the same cell population which expressed an extravillous trophoblast marker, c-erbB-2 (E and F). CD1d expression also was detected on villous trophoblasts (G) as well as on individual cells that had migrated from the villi (D and H). Isotype-matched control IgG mAb staining (B and C), CD1d staining by using NOR3.2 (D and F–H), and staining with anti-c-erbB-2 (E). DEC, decidua; IVS, intervillous space; VIL, villous. Arrows denote trophoblasts. cells (␥-irradiated; 5,000 rads) were used as feeders, and phy- after the last restimulation. Target cells were labeled with tohemagglutinin (Remel, Lenexa, KS) was added to a final [35S]methionine and culturing iNKT effector cells with targets concentration of 1 ␮g͞ml. Cells were cultured in RPMI medium for5hateffector-to-target cell ratios of 50, 25, and 12.5 to 1. 1640 supplemented with 15% human AB serum (Atlanta Bio- NaOH (1 N) was added to induce maximum release. When logicals, Norcross, GA), penicillin͞streptomycin, 100 mM so- ␣GalCer was used, it was added to the target cells the night dium pyruvate, 1% nonessential amino acids, 2 mM L-glutamine, before the assay at 100 ng͞ml final concentration. Percentage of and 2-mercaptoethanol (all from Life Technologies, Rockville, specific lysis was calculated by the following formula: [(experi- MD). IL-2 (National Cancer Institute–Frederick Cancer Re- mental cpm Ϫ spontaneous release cpm)͞(maximum release search and Development Center, Frederick, MD) was added to cpm Ϫ spontaneous release cpm)] ϫ 100. Percentage of specific a final concentration of 100 units͞ml. In some cases, peripheral lysis was always Ͻ25%. All measurements were performed in iNKT clones were grown in 20 units͞ml IL-2 plus 15 ng͞ml IL-7 triplicate. (R & D Systems). No difference was observed in functional assays between the two growth conditions. Clones were restim- Immunohistochemistry. Decidual tissue was identified macroscop- ulated with 50,000 ␥-irradiated feeders and phytohemagglutinin ically and washed extensively with PBS, pH 7.2. Decidual pieces (1 ␮g͞ml) every 3–4 wk. were fixed overnight at 4°C in 4% paraformaldehyde, after which they were washed in PBS, dehydrated in ethanol followed by Functional Analyses. For iNKT cell activation assays, 5 ϫ 104 iNKT xylene, and embedded in paraffin. For staining, 6-␮m sections cells were plated in a 96-well round-bottom plate with either were rehydrated, and they were either stained with Gill’s hema- medium alone or with 5 ϫ 104 stimulator cells that had been toxyolin and eosin or they underwent antigen retrieval (Vector ͞ ␣ treated with 0.1 mg ml mitomycin C for1hat37°C. GalCer Laboratories). Slides were incubated in 3% H2O2 in methanol (KRN7000, a gift from Kirin Brewery, Tokyo), resuspended in before staining to block endogenous peroxidase activity. DMSO was added at a final concentration of 100 ng͞ml. Either NOR3.2 and anti-c-erbB-2 mAb binding was detected by using phytohemagglutinin (1 ␮g͞ml) or plate-bound anti-CD3 (1 the streptavidin͞peroxidase and visualized by using diamino- ␮g͞ml) was used as a positive control. For stimulation with benzidine substrate (R.T.U. Vectastain kit; Vector Laboratories). CD1d transfectants, IL-2 was included in the culture medium at a final concentration of 20 units͞ml. C1R B-LCLs and HeLa Results cells transfected with CD1d have been described (29). Typically, CD1d Expression on Trophoblasts. To determine whether CD1d was proliferation and cytokine release assays were performed 3–4wk expressed in the human placenta, immunohistochemical staining after the last restimulation. Proliferation was measured in a 96-h on paraffin-embedded sections of 8-, 10-, 38-, and 39-wk-old assay in which 1 ␮Ci [3H]thymidine (1 Ci ϭ 37 GBq) was added human decidua was performed. CD1d expression detected by during the last 18 h of incubation, followed by harvesting and the CD1d-specific NOR3.2 mAb was clearly apparent in both scintillation counting. Cytokine levels were measured by collect- villous and extravillous trophoblasts (Fig. 1). Extravillous tro- ing supernatants after 48 h of incubation and testing by ELISA. phoblast were identified by morphology in serial hematoxylin͞ Abs for GM-CSF detection were fromR&DSystems, and Abs eosin sections (Fig. 1A) and by the expression of the c-erbB-2 for IFN-␥ and IL-4 were from BioSource. Streptavidin-POD protein, which is expressed on extravillous but not villous (Roche Diagnostics) was used for detection with TMB substrate trophoblast (Fig. 1E; ref. 30). Staining of serial sections with (Pierce). Cytotoxicity assays were performed no later than 10 d NOR3.2 demonstrated that CD1d was expressed on the c-

13742 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.162491699 Boyson et al. viduals and stained with a panel of mAbs. Importantly, 6B11ϩ iNKT cells were always V␣24ϩV␤11ϩ. Interestingly, although FACS analysis demonstrated that Ͻ40% of decidual iNKT cells were CD4ϩ (the remainder being CD4Ϫ), only CD4ϩ clones could be expanded. The reason for this is unclear, but it is possible that the decidual CD4Ϫ population may have some specific growth impairment or that additional factors may be required to expand these cells. Therefore, in this manuscript all decidual iNKT clones referenced are CD4ϩ clones. Like peripheral iNKT cells, decidual iNKT clones were all CD45ROϩ and CD45RAϪ. Examination of the clones with a panel of mAb directed to NK markers revealed that the expression of CD161 (the human homolog of murine NK1.1) was quite variable and that although none of the decidual iNKT clones expressed the CD16 or CD94 molecules, many clones expressed low levels of CD57 and significant levels of CD56.

CD1d Reactivity of Decidual iNKT Cells. Lymphocytes that populate the decidua are thought to have altered functional capabilities (31). Because the hallmark of iNKT cells is their ability to recognize CD1d, especially ␣GalCer bound by CD1d, the ability of decidual iNKT clones to recognize CD1dϩ target cells was tested. In a 96-h standard proliferation assay, ␣GalCer, pulsed onto C1R cells transfected with CD1d, induced proliferation of decidual iNKT cells to a level similar to that of peripheral iNKT clones (Fig. 3A) and decidual iNKT clones efficiently killed ␣GalCer-loaded CD1dϩ target cells (Fig. 3B), as previously shown for peripheral iNKT cells (3). In addition, decidual iNKT Fig. 2. A high frequency of invariant NKT cells in the decidua. (A) Repre- clones cocultured with either ␣GalCer-pulsed C1R͞CD1d or ␣ sentative contour plots of CD3-gated lymphocytes stained with anti-V 24 and ␣GalCer-pulsed HeLa͞CD1d transfectants secreted both IFN-␥ anti-V␤11 mAbs. (B) Contour plot of CD3-gated decidual lymphocytes stained with the anti-CDR3␣ mAb 6B11 and anti-Va24. (C) Summary of iNKT cell and IL-4, demonstrating that the decidual iNKT clones were ϩ CD1d-restricted (unpublished data). Interestingly, some clones frequencies as a percentage of CD3 T cells in the decidua and in the periphery. ͞ Mean percentages of iNKT frequencies are indicated by bars (C). exhibited low levels of reactivity to C1R CD1d transfectants in the absence of ␣GalCer (clone 15-3 in Fig. 3A, and clones 21-29 and 21-10 in Fig. 3B), which presumably reflects reactivity erbB-2ϩ population (Fig. 1 E and F). CD1d also was present on toward endogenous ligands bound by CD1d. villous trophoblasts (Fig. 1G) and on trophoblasts, which had The ability of decidual iNKT cells to secrete IL-10 was been trapped in the fibrinoid deposits at the maternal–fetal determined because IL-10 was demonstrated in the mouse to be interface (Fig. 1 D and H). Additional experiments will be an important regulatory cytokine produced by iNKT cells (15). needed to determine whether CD1d is expressed in all villous Intracellular staining of decidual iNKT clones after stimulation trophoblasts or only on those differentiating to the ‘‘intermedi- with phorbol 12-myristate 13-acetate and ionomycin revealed ate’’ phenotype that comprises cell columns. A low level of little or no IL-10 production (unpublished data). syncytiotrophoblast staining with the NOR3.2 mAb was observed but it was judged to be inconclusive because a similar Decidual iNKT Cells Exhibit a Th1-Like Bias and a Polarization Toward amount of syncytiotrophoblast staining was seen by using the GM-CSF Production. Because iNKT cells appear to function, in isotype-matched control. part, by secreting a wide variety of cytokines, we asked whether there was a difference in the cytokine production of resident Invariant NKT Cells in the Decidua. To determine the frequency and decidual and peripheral blood iNKT cells. Peripheral iNKT phenotype of iNKT cells in the human decidua, FACS analysis clones were FACS-sorted from four additional donors. Be- cause all of the decidual clones were CD4ϩ iNKT cells, only of lymphocytes isolated from first trimester (7–10 wk) decidual CD4ϩ peripheral iNKT cells were selected for comparison. tissue was performed. V␣24ϩV␤11ϩ cells comprised 0.48 Ϯ Three weeks after restimulation, quiescent clones were ei- 0.05% (n ϭ 4) of the CD3ϩ decidual lymphocytes, much higher Ϯ ϭ ther incubated with medium alone or stimulated with either than the 0.04 0.02% (n 4) frequency seen in human IL-2, plate-bound anti-CD3, or ␣GalCer-loaded C1R͞CD1d peripheral blood (Fig. 2 A and C). Similar results were obtained ␣ transfectants. Cell culture supernatants were harvested after when anti-V 24 mAb was used in conjunction with 6B11, a mAb 48 h and tested by ELISA for GM-CSF, IFN-␥, and IL-4 raised against a cyclic peptide corresponding to the invariant ␣ production. CDR3 loop of iNKT cells (Fig. 2B). Thus, the decidual T cell Surprisingly, decidual iNKT clones exhibited a striking bias population was greatly enriched in iNKT cells compared to the toward IFN-␥ secretion as opposed to IL-4 (Fig. 4). Peripheral peripheral blood. iNKT clones, in contrast, were biased toward IL-4 production. ϩ ␣ ϩ To verify that the 6B11 V 24 decidual T cells were indeed Although some peripheral clones appeared predisposed to se- iNKT cells, these cells were cloned by single cell FACS sorting crete IFN-␥, and some decidual clones exhibited little if any bias, of 6B11ϩV␣24ϩ decidual T cells. PCR amplification and direct there was a striking difference when the two populations were sequencing across the TCR CDR3␣ region indicated that the compared as a whole. The IFN-␥͞IL-4 difference between the clones possessed characteristic TCR V␣24J␣Q junctions without peripheral and decidual populations was detectable even in N͞P additions. Because it is difficult to obtain T or iNKT cells IL-2-stimulated clones, although the absolute level of cytokine in large numbers from decidual tissue for phenotypic analysis, a secretion from these clones was very low. Similar results were IMMUNOLOGY total of 32 iNKT clones were isolated from six different indi- obtained whether the clones were stimulated with anti-CD3 or

Boyson et al. PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21 ͉ 13743 Fig. 3. Decidual iNKT cells are restricted by CD1d. iNKT clones were assayed for their ability to recognize CD1d-expressing cells. (A) Proliferation of peripheral (B2) and decidual (15-3 and 33-10) iNKT clones cocultured with mock-transfected C1Rneo cells and C1R cells transfected with CD1d, with and without ␣GalCer. Error bars represent the SD from the mean. Results are representative of 25 decidual iNKT clones. (B) Lysis by iNKT clones of C1Rneo and C1R͞CD1d transfectants that had been pulsed with ␣GalCer or vehicle. Six representative decidual iNKT clones from two different patients are shown at an effector-to-target ratio of 50:1. Error bars represent the SD from the mean. Results are representative of 14 decidual iNKT clones. Fig. 4. Stimulation of decidual iNKT clones reveals a Th1-like bias in cytokine secretion. Decidual and peripheral iNKT clones were stimulated with IL-2, with C1R͞CD1d transfectants pulsed with ␣GalCer, or with anti-CD3. Supernatants with ␣-GalCer-loaded C1R͞CD1d transfectants (Fig. 4). Inter- were collected after 48 h and tested for IFN-␥ and IL-4 by ELISA. Data are estingly, in both peripheral and decidual iNKT clones, ␣GalCer expressed as log ratios of the amount of secreted cytokines. Each bar repre- stimulation resulted in a more pronounced Th2-like bias than did sents one individual clone. Peripheral and decidual clones were derived from three and six different individuals, respectively. anti-CD3 stimulation—i.e., the Th2-like bias in peripheral iNKT clones was enhanced by using ␣GalCer, whereas the Th1-like bias in the decidual iNKT clones was diminished (Fig. 4). decidual iNKT clones produced higher relative levels of IFN-␥ Next, GM-CSF levels were assessed because GM-CSF was and GM-CSF transcripts than peripheral iNKT clones, whereas previously demonstrated to be a highly regulated iNKT cell peripheral iNKT clones exhibited higher levels of IL-4 tran- cytokine by microarray gene expression profiling (7). When scripts (unpublished data). iNKT clones were stimulated with either ␣GalCer or with anti-CD3, GM-CSF production by the decidual clones was Discussion significantly greater than that by the peripheral clones (Fig. 5). Numerous reports have demonstrated that iNKT cells are key The difference in mean GM-CSF production was greatest in regulators of autoimmunity and immune tolerance (4–8, 14, anti-CD3-stimulated cells (decidual, 11,700 pg͞ml; peripheral, 15, 32, 33) and may be important in graft survival (16–19). 2,750 pg͞ml, P Ͻ 0.001) and slightly less so in ␣GalCer- Active tolerance induction occurs at the maternal–fetal inter- stimulated cells (decidual, 3,680 pg͞ml; peripheral, 1,640 pg͞ml, face, suggesting it is one site in which iNKT cell–CD1d P Ͻ 0.001). Upon IL-2 stimulation, the peripheral iNKT clones interactions may occur. In pregnant mice, ␣GalCer efficiently produced minimally higher mean levels of GM-CSF than the induces abortion via an iNKT-dependent mechanism (34), decidual iNKT clones, 140 and 55 pg͞ml, respectively (Fig. 5). possibly as the result of overstimulation of a normal physio- The ELISA data were confirmed by comparing peripheral and logical mechanism that may be analogous to the systemic decidual CD4ϩ iNKT clone cytokine transcript levels by using release of cytokines induced by superantigens. Evaluating the ribonuclease protection assays. In accordance with the ELISA, possible functions of iNKT cells at the maternal–fetal interface

13744 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.162491699 Boyson et al. CD4ϩ and CD4Ϫ iNKT cells, yet only CD4ϩ cells were recovered after cloning. Whether this reflects a growth impairment on the part of CD4ϪCD8Ϫ decidual iNKT cells is unknown; although we have observed that decidual NK cells demonstrate diminished proliferative and cytolytic capacities compared to peripheral blood NK cells (J.E.B. and L.A.K., unpublished data). Examination of decidual iNKT cell phenotypic markers revealed few differences from peripheral blood iNKT cells. One significant difference, however, was that all decidual iNKT clones were CD94Ϫ, whereas CD94 is expressed on Ϸ50% of peripheral blood iNKT cells. This difference probably reflects the fact that only CD4ϩ clones were examined here, because a recent report using ␣GalCer-loaded tet- ramers suggests that CD94 is preferentially expressed on CD4Ϫ clones (39). Because previous reports suggest that iNKT cells may function, in part, by differential cytokine expression, the expression profile of decidual CD4ϩ iNKT clones and peripheral CD4ϩ iNKT clones was compared. Intracellular staining for cytokine expression revealed the production of GM-CSF, IFN-␥, and IL-4. When expression of these cytokines was Fig. 5. Decidual iNKT cells produce much higher amounts of GM-CSF than compared between peripheral and decidual iNKT clones, two their peripheral counterparts. GM-CSF was measured by ELISA from decidual striking differences were observed. First, GM-CSF production, and peripheral iNKT clones cultured with IL-2, or stimulated with anti-CD3 or measured by ELISA, was much higher in the decidual iNKT with ␣GalCer-pulsed C1R͞CD1d transfectants. E, Individual clones (decidual, clones than in the peripheral iNKT clones. These data were n ϭ 24; peripheral n ϭ 18), and the geometric mean of the GM-CSF level for corroborated by ribonuclease protection assay analysis, which each condition is denoted by a bar. The geometric means of GM-CSF levels for demonstrated higher GM-CSF transcript levels in decidual decidual clones were: IL-2-stimulated, 55 pg͞ml; ␣GalCer-stimulated, 3,680 iNKT cells than in peripheral iNKT cells. GM-CSF is a ͞ ͞ pg ml; anti-CD3-stimulated 11,700 pg ml. Geometric means of GM-CSF levels pleiotropic cytokine, which has a demonstrated ability to affect for peripheral clones were: IL-2-stimulated, 140 pg͞ml; ␣GalCer-stimulated, Ϫ͞Ϫ 1,640 pg͞ml; anti-CD3-stimulated, 2,750 pg͞ml. pregnancy outcome. GM-CSF mice show selective impairment in their ability to reproduce (40), whereas administration of GM-CSF to mice in the abortion-prone CBA͞J ϫ ͞ is difficult because CD1d expression has not been evaluated. DBA 2 mating combination results in a dramatic decrease in Therefore, to investigate whether maternal iNKT cell–CD1d fetal resorption (41). Interestingly, mice lacking the GM-CSF ␤ interactions may play a role in normal physiological processes receptor common chain possess fewer iNKT cells than at the human maternal–fetal interface, the presence of CD1d control mice as well as a defect in iNKT maturation (42). was examined and decidual iNKT cells were compared to their GM-CSF receptor is expressed on extravillous trophoblast peripheral blood counterparts. cells, the very cells on which CD1d is expressed (27), suggest- Interestingly, CD1d was expressed on trophoblasts, the fetal ing that GM-CSF production in response to iNKT–trophoblast cells that invade the maternal tissue. CD1d expression has interaction may have implications for the function or fate of previously been reported in myeloid lineage cells, B cells, and the trophoblast. Alternatively, GM-CSF production may play a role in the maturation of myeloid lineage cells in the decidua. activated T cells (28) as well as on intestinal and reproductive Decidual macrophages, in turn, have the capability to suppress tract epithelia (35, 36). Immunohistochemical staining dem- allogeneic T cell responses (43, 44). onstrated CD1d expression on both villous and extravillous A dramatic difference between decidual and peripheral trophoblasts [as judged by costaining of serial sections for the iNKT clones in IFN-␥ and IL-4 production also was observed. epidermal growth factor-like c-erbB-2, an extravillous tropho- Decidual iNKT clones exhibited a clear bias toward IFN-␥ blast marker (30)], but not syncytiotrophoblast. These data are secretion vs. IL-4 secretion. In contrast, peripheral blood consistent with previous work demonstrating CD1d transcripts CD4ϩ iNKT cells appeared to be biased toward IL-4 secretion in trophoblasts and choriocarcinoma cell lines, although full- in agreement with a previous report (45). These observations length transcripts were never identified (37). Attempts to are consistent with RT-PCR results from mouse decidual detect CD1d on JEG-3 choriocarcinoma cells by FACS and to ␥ iNKT cells (34), suggesting that a similar skewing toward up-regulate its expression by using IFN- were unsuccessful Th1-like cytokines occurs in both species. Curiously, IFN-␥, (J.E.B., unpublished data). Human trophoblasts do not ex- together with tumor necrosis factor-␣, mediates ␣GalCer- press HLA-A and HLA-B molecules (20), presumably to induced abortion in the mouse (34). Why would decidual iNKT escape allo-recognition by the maternal immune system. Upon cells possess a phenotype (Th1-like) associated with abortion their differentiation to the invasive extravillous phenotype, induction? One possible explanation is that the abortions trophoblasts begin to express the nonclassical molecule observed after ␣GalCer administration results from a systemic HLA-G (22, 38), whose function is unknown. Thus, it is overproduction of these cytokines by iNKT cells in the pe- probable that CD1d and HLA-G are coexpressed on the riphery (as well as an overproduction at the maternal–fetal invasive extravillous trophoblast subset, though this has not interface). Indeed, direct injection of IFN-␥ or tumor necrosis been formally demonstrated. factor-␣ effectively induces abortions in mice (41, 46). Con- The frequency of iNKT cells observed in decidua was con- versely, IFN-␥ appears to be necessary for the proper devel- siderably higher than that found in peripheral blood. A similar opment of the decidua (47). Thus, any severe disruption to the frequency of decidual V␣24ϩCD161ϩ cells, presumably iNKT tightly regulated balance of IFN-␥ and͞or other cytokines may cells, was reported recently (50), but these cells were not further be deleterious to the maintenance of pregnancy. characterized. In the present study, FACS staining of freshly Qualitative and quantitative differences in the cytokines se- IMMUNOLOGY isolated decidual lymphocytes revealed the presence of both creted by iNKT cells have been reported in both prostate cancer

Boyson et al. PNAS ͉ October 15, 2002 ͉ vol. 99 ͉ no. 21 ͉ 13745 and type I diabetes (6, 7, 48). Recently, phenotypic differences subsets may account for the ability of iNKT cells to function in between the CD4ϩ and CD4Ϫ iNKT cell subsets were described a variety of contexts. by two groups (39, 49). Whereas CD4Ϫ iNKT cells produce primarily Th1-type cytokines such as IFN-␥ and tumor necrosis We thank Ilaria Potolicchio, Hiroyuki Nishimura, and members of the factor-␣, CD4ϩ iNKT cells were reported to produce both Strominger laboratory for helpful discussion, Drs. Rapin Osathanondh and Dan Schust for supplying decidual samples, Maris Handley and Herb Th1-type and Th2-type cytokines. The present results suggest ϩ Levine for assistance in flow cytometry, and Dr. Raymond Genest for that within the CD4 iNKT cell subset, there may be differences advice. We also thank Kirin Brewery for providing the KRN7000 in decidual and peripheral blood Th1͞Th2 cytokine phenotypes. (␣-galactosylceramide). This research was supported by National Insti- A shift of this balance in a manner similar to classical T helper tutes of Health Grants CA-47554 and AI-50207 (to J.L.S.).

1. Porcelli, S., Gerdes, D., Fertig, A. M. & Balk, S. P. (1996) Hum. Immunol. 48, 25. Bennett, W. A., Lagoo-Deenadayalan, S., Whitworth, N. S., Brackin, M. N., 63–67. Hale, E. & Cowan, B. D. (1997) Early Pregnancy 3, 190–198. 2. Kawano, T., Cui, J., Koezuka, Y., Toura, I., Kaneko, Y., Motoki, K., Ueno, H., 26. Roth, I., Corry, D. B., Locksley, R. M., Abrams, J. S., Litton, M. J. & Fisher, Nakagawa, R., Sato, H., Kondo, E., et al. (1997) Science 278, 1626–1629. S. J. (1996) J. Exp. Med. 184, 539–548. 3. Spada, F. M., Koezuka, Y. & Porcelli, S. A. (1998) J. Exp. Med. 188, 1529–1534. 27. Loke, Y. W., King, A., Gardner, L. & Carter, N. P. (1992) J. Reprod. Immunol. 4. Sharif, S., Arreaza, G. A., Zucker, P., Mi, Q. S., Sondhi, J., Naidenko, O. V., 22, 33–45. Kronenberg, M., Koezuka, Y., Delovitch, T. L., Gombert, J. M., et al. (2001) 28. Exley, M., Garcia, J., Wilson, S. B., Spada, F., Gerdes, D., Tahir, S. M., Patton, Nat. Med. 7, 1057–1062. K. T., Blumberg, R. S., Porcelli, S., Chott, A. & Balk, S. P. (2000) Immunology 5. Hong, S., Wilson, M. T., Serizawa, I., Wu, L., Singh, N., Naidenko, O. V., Miura, 100, 37–47. T., Haba, T., Scherer, D. C., Wei, J., et al. (2001) Nat. Med. 7, 1052–1056. 29. Exley, M., Garcia, J., Balk, S. P. & Porcelli, S. (1997) J. Exp. Med. 186, 109–120. 6. Wilson, S. B., Kent, S. C., Patton, K. T., Orban, T., Jackson, R. A., Exley, M., 30. Jokhi, P. P., King, A. & Loke, Y. W. (1994) Cytokine 6, 433–442. Porcelli, S., Schatz, D. A., Atkinson, M. A., Balk, S. P., et al. (1998) Nature 391, 31. King, A., Gardner, L. & Loke, Y. W. (1996) J. Reprod. Immunol. 30, 67–74. 177–181. 32. Singh, A. K., Wilson, M. T., Hong, S., Olivares-Villagomez, D., Du, C., Stanic, 7. Wilson, S. B., Kent, S. C., Horton, H. F., Hill, A. A., Bollyky, P. L., Hafler, A. K., Joyce, S., Sriram, S., Koezuka, Y. & Van Kaer, L. (2001) J. Exp. Med. D. A., Strominger, J. L. & Byrne, M. C. (2000) Proc. Natl. Acad. Sci. USA 97, 194, 1801–1811. 7411–7416. 33. Miyamoto, K., Miyake, S. & Yamamura, T. (2001) Nature 413, 531–534. 8. Naumov, Y. N., Bahjat, K. S., Gausling, R., Abraham, R., Exley, M. A., 34. Ito, K., Karasawa, M., Kawano, T., Akasaka, T., Koseki, H., Akutsu, Y., Kondo, Koezuka, Y., Balk, S. B., Strominger, J. L., Clare-Salzer, M. & Wilson, S. B. E., Sekiya, S., Sekikawa, K., Harada, M., et al. (2000) Proc. Natl. Acad. Sci. USA (2001) Proc. Natl. Acad. Sci. USA 98, 13838–13843. 97, 740–744. 9. Kawano, T., Cui, J., Koezuka, Y., Toura, I., Kaneko, Y., Sato, H., Kondo, E., 35. Blumberg, R. S., Terhorst, C., Bleicher, P., McDermott, F. V., Allan, C. H., Harada, M., Koseki, H., Nakayama, T., et al. (1998) Proc. Natl. Acad. Sci. USA Landau, S. B., Trier, J. S. & Balk, S. P. (1991) J. Immunol. 147, 2518–2524. 95, 5690–5693. 36. Sallinen, K., Verajankorva, E. & Pollanen, P. (2000) J. Reprod. Immunol. 46, 10. Kawano, T., Nakayama, T., Kamada, N., Kaneko, Y., Harada, M., Ogura, N., 91–101. Akutsu, Y., Motohashi, S., Iizasa, T., Endo, H., et al. (1999) Cancer Res. 59, 37. Jenkinson, H. J., Wainwright, S. D., Simpson, K. L., Perry, A. C., Fotiadou, P. 5102–5105. & Holmes, C. H. (1999) Immunology 96, 649–655. 11. Kakimi, K., Guidotti, L. G., Koezuka, Y. & Chisari, F. V. (2000) J. Exp. Med. 38. McMaster, M. T., Librach, C. L., Zhou, Y., Lim, K. H., Janatpour, M. J., 192, 921–930. DeMars, R., Kovats, S., Damsky, C. & Fisher, S. J. (1995) J. Immunol. 154, 12. Exley, M. A., Bigley, N. J., Cheng, O., Tahir, S. M., Smiley, S. T., Carter, Q. L., 3771–3778. Stills, H. F., Grusby, M. J., Koezuka, Y., Taniguchi, M., et al. (2001) J. Leukocyte 39. Lee, P. T., Benlagha, K., Teyton, L. & Bendelac, A. (2002) J. Exp. Med. 195, Biol. 69, 713–718. 637–641. 13. Exley, M. A., He, Q., Cheng, O., Wang, R. J., Cheney, C. P., Balk, S. P. & Koziel, M. J. (2002) J. Immunol. 168, 1519–1523. 40. Robertson, S. A., Roberts, C. T., Farr, K. L., Dunn, A. R. & Seamark, R. F. 14. Sonoda, K. H., Exley, M., Snapper, S., Balk, S. P. & Stein-Streilein, J. (1999) (1999) Biol. Reprod. 60, 251–261. J. Exp. Med. 190, 1215–1226. 41. Chaouat, G., Menu, E., Clark, D. A., Dy, M., Minkowski, M. & Wegmann, T. G. 15. Sonoda, K. H., Faunce, D. E., Taniguchi, M., Exley, M., Balk, S. & Stein- (1990) J. Reprod. Fertil. 89, 447–458. Streilein, J. (2001) J. Immunol. 166, 42–50. 42. Sato, H., Nakayama, T., Tanaka, Y., Yamashita, M., Shibata, Y., Kondo, E., 16. Ikehara, Y., Yasunami, Y., Kodama, S., Maki, T., Nakano, M., Nakayama, T., Saito, Y. & Taniguchi, M. (1999) Proc. Natl. Acad. Sci. USA 96, 7439–7444. Taniguchi, M. & Ikeda, S. (2000) J. Clin. Invest. 105, 1761–1767. 43. Munn, D. H., Zhou, M., Attwood, J. T., Bondarev, I., Conway, S. J., Marshall, 17. Seino, K. I., Fukao, K., Muramoto, K., Yanagisawa, K., Takada, Y., Kakuta, S., B., Brown, C. & Mellor, A. L. (1998) Science 281, 1191–1193. Iwakura, Y., Van Kaer, L., Takeda, K., Nakayama, T., et al. (2001) Proc. Natl. 44. Munn, D. H., Shafizadeh, E., Attwood, J. T., Bondarev, I., Pashine, A. & Acad. Sci. USA 98, 2577–2581. Mellor, A. L. (1999) J. Exp. Med. 189, 1363–1372. 18. Sonoda, K. H. & Stein-Streilein, J. (2002) Eur. J. Immunol. 32, 848–857. 45. Takahashi, T., Nieda, M., Koezuka, Y., Nicol, A., Porcelli, S. A., Ishikawa, Y., 19. Sonoda, K. H., Taniguchi, M. & Stein-Streilein, J. (2002) J. Immunol. 168, Tadokoro, K., Hirai, H. & Juji, T. (2000) J. Immunol. 164, 4458–4464. 2028–2034. 46. Clark, D. A., Chaouat, G., Arck, P. C., Mittruecker, H. W. & Levy, G. A. (1998) 20. Faulk, W. P. & Temple, A. (1976) Nature 262, 799–802. J. Immunol. 160, 545–549. 21. King, A., Boocock, C., Sharkey, A. M., Gardner, L., Beretta, A., Siccardi, A. G. 47. Ashkar, A. A. & Croy, B. A. (1999) Biol. Reprod. 61, 493–502. & Loke, Y. W. (1996) J. Immunol. 156, 2068–2076. 48. Tahir, S. M., Cheng, O., Shaulov, A., Koezuka, Y., Bubley, G. J., Wilson, S. B., 22. Kovats, S., Main, E. K., Librach, C., Stubblebine, M., Fisher, S. J. & DeMars, Balk, S. P. & Exley, M. A. (2001) J. Immunol. 167, 4046–4050. R. (1990) Science 248, 220–223. 49. Gumperz, J. E., Miyake, S., Yamamura, T. & Brenner, M. B. (2002) J. Exp. Med. 23. Tafuri, A., Alferink, J., Moller, P., Hammerling, G. J. & Arnold, B. (1995) 195, 625–636. Science 270, 630–633. 50. Tsuda, H., Sakai, M., Michimata, T., Tanebe, K., Hayakawa, S. & Saito, S. 24. Mellor, A. L. & Munn, D. H. (2000) Annu. Rev. Immunol. 18, 367–391. (2001) Am. J. Reprod. Immunol. 45, 295–302.

13746 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.162491699 Boyson et al.