Gestation stage-dependent mechanisms of invariant natural killer T cell-mediated pregnancy loss

  1. Jonathan E. Boyson*,†,
  2. Nisha Nagarkatti†,
  3. Leena Nizam*,
  4. Mark A. Exley‡, and
  5. Jack L. Strominger†,§
  1. *Division of Transplantation Surgery and Immunology, Department of Surgery, University of Vermont College of Medicine, Burlington, VT 05405;
  2. †Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138; and
  3. ‡Division of Hematology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 02215

  1. Contributed by Jack L. Strominger, December 21, 2005

 
Next Section

Abstract

Stimulation of CD1d-restricted semiinvariant natural killer T cells by using the CD1d ligand α-galactosylceramide (αGalCer) induces pregnancy loss in mice through an ill-defined mechanism involving TNF, IFN-γ, and perforin. In this article, we demonstrate that during early gestation, αGalCer efficiently induced pregnancy loss in C57BL/6J and BALB/cJ mice in a perforin-dependent manner. In contrast, during midgestation perforin was no longer required for pregnancy loss. Concomitant with the loss of a perforin requirement at midgestation was the emergence of strain-dependent variations in susceptibility to αGalCer-induced pregnancy loss. Whereas pregnant C57BL/6J mice remained susceptible to αGalCer at midgestation, pregnant BALB/cJ mice were resistant to its effects. Pregnancy loss during midgestation was correlated with dramatically higher serum cytokine levels, including TNF and IL-2, in the susceptible C57BL/6J strain compared with the resistant BALB/cJ strain. Thus, the stage of gestation defined two distinct mechanisms of pregnancy loss: a perforin-dependent mechanism operating at early gestation and a perforin-independent, cytokine-dominated mechanism operating after midgestation.

Increasing evidence suggests that unexplained cases of recurrent pregnancy loss and idiopathic preterm birth may have an underlying immune etiology (1, 2). Innate immune mechanisms appear to play an especially critical role. Numerous studies suggest that aberrant natural killer (NK) cell activation is associated with recurrent pregnancy loss (3–7), and there is a well-documented association of intrauterine infection with preterm birth (8, 9). In addition, complement activation (10) or improper regulation of the complement system (11) has also been demonstrated to lead to pregnancy loss. A common thread throughout all of these mechanisms is the production of proinflammatory cytokines such as TNF, IL-1β, and IFN-γ, which appear to be critical mediators in the induction of pregnancy loss at all stages of gestation (8, 12–17).

Semiinvariant NKT (iNKT) cells comprise a T cell subset that is present in both the human and mouse decidua (18–21). Unlike conventional T cells that recognize peptides in the context of polymorphic MHC molecules, iNKT cells recognize glycosphingolipids presented by the monomorphic class I MHC-like glycoprotein CD1d (22, 23). Upon recognition of their glycosphingolipid ligands, iNKT cells rapidly produce a wide variety of cytokines including IFN-γ, IL-4, and TNF (22, 24), and they possess the capacity for perforin-dependent cytotoxicity (25, 26). Recently, iNKT cells were demonstrated to recognize cell wall glycosphingolipids from certain strains of bacteria (27–29), adding to the growing body of evidence suggesting that iNKT cells are important in the clearance of pathogenic organisms (30).

Stimulation of iNKT cells in pregnant mice by using the CD1d ligand α-galactosylceramide (αGalCer) rapidly induces pregnancy loss in mice and has been proposed as a model of infection-associated pregnancy pathology (19). In this article, we demonstrate that the mechanism through which iNKT cells mediate pregnancy loss is highly dependent on the stage of gestation, with early gestation pregnancy loss associated with a perforin-dependent lytic mechanism and midgestation pregnancy loss associated with a cytokine-dominated mechanism. We also found that midgestation, but not early gestation, pregnancy loss exhibited strain-dependent variations in pro-inflammatory cytokine production which correlated with susceptibility to iNKT cell-mediated pregnancy loss.

Previous SectionNext Section

Results

iNKT Cell-Mediated Pregnancy Loss.

iNKT cell activation in C57BL/6J mice during early gestation pregnancy has been demonstrated to induce early gestation pregnancy loss (19). We investigated the effect of αGalCer administration on pregnancy during later stages of gestation. Administration of αGalCer to pregnant C57BL/6J mice at midgestation [day 9.5 postcoitus (p.c.)] induced rapid pregnancy loss with complete evacuation of the uterus by 72 h after injection, although remnants of fetoplacental units were occasionally seen (Fig. 1 A). Examination of uteri at earlier time points after αGalCer administration revealed individual resorbing embryos as early as 48 h after αGalCer administration (Fig. 1 B). Administration of αGalCer during late gestation (day 16.5 p.c.) resulted in the rapid onset of preterm birth. Within 18–24 h after receiving αGalCer, all mice delivered preterm fetuses complete with attached amnion and placenta (Fig. 1 C). No effect was observed in vehicle-injected mice (data not shown). Thus, activation of iNKT cells during pregnancy efficiently induced pregnancy loss at all stages of gestation in C57BL/6J mice. Importantly, no effect was observed in mice deficient in iNKT cells (Jα18−/−) or in CD1d-deficient mice, indicating that the effect of αGalCer was solely due to its ability to stimulate iNKT cells in a CD1d-dependent manner (Table 1).

Fig. 1.
View larger version:
Fig. 1.

αGalCer-induces midgestation pregnancy loss and preterm birth. (A) Uteri from C57BL/6J mice that received vehicle (Left) or αGalCer (Right). αGalCer was administered on day 9.5 p.c., and uteri were examined 72 h later. The remnants of two conceptuses can still be seen (arrows) in the αGalCer-treated uterus. (B) Kinetics of αGalCer-induced pregnancy loss. Pregnant C57BL/6J mice were injected with vehicle control or αGalCer on day 9.5 p.c. Mice were euthanized at varying time points after injection, and their uteri were examined for fetal resorption. Data are presented as the percent resorption [(number of resorbing embryos/total number of embryos)·100]. Signs of fetal resorption are evident as early as 48 h after injection of αGalCer. ○, vehicle-injected mice; •, αGalCer-injected mice. (C) Induction of preterm birth at day 16.5 p.c. by αGalCer. Photograph of a single fetus (one of six) delivered prematurely ≈18 h after administration of αGalCer, still encased in the amnion with attached placenta and decidua. All fetuses were delivered preterm (n = 5 pregnant females). (D) Cross-section of a normal mouse implantation site at day 10.5 p.c. stained with hematoxylin/eosin. The sample is oriented with the decidua at the top and the amniotic cavity at the bottom (×40). (E) Implantation site at day 10.5 p.c. of an αGalCer-treated mouse (24 h after administration). Placental development is markedly impaired, and deterioration in the membrane organization is evident (×100). d, decidua; l, labyrinth, m, metrial gland. (F and G) Serial sections of an implantation site 48 h after αGalCer administration stained with normal rabbit antisera (F) or rabbit anti-fibrinogen/fibrin (G) (×40). (H) Uterine cross-section 60 h after αGalCer administration. Fetal and decidual tissue has been completely sloughed from the uterine wall (×40). (I) High-power (×100) view of the uterine wall 60 h after αGalCer administration. The red mass on the lower right is the decidual/fetal tissue in the uterine lumen. Results are representative of histology performed on three mice.


View this table:
Table 1.

Strict dependence on CD1d and iNKT cells in αGalCer-induced pregnancy loss


Examination of the implantation sites after αGalCer administration revealed rapid histological changes. As early as 24 h after administration of αGalCer on day 9.5 p.c., the architecture of the labyrinthine zone of the placenta was noticeably disrupted compared with gestation-age matched controls (Fig. 1 D and E), and there was evidence of increased hemorrhage at the antimesometrial pole (data not shown). By 48 h, signs of deterioration of implantation sites were evident histologically, in part characterized by extensive fibrin deposition throughout the decidua (Fig. 1 F and G). Gross dissection of these fetoplacental units revealed a mass of tissue without any discernible embryo. By 60 h, the presence of necrotic tissue in the lumen was observed (Fig. 1 H), as was the regeneration of the endometrial lining (Fig. 1 I).

Influence of Genetic Background During Early and Midgestation.

Due to the conserved nature of the iNKT T cell receptor/CD1d interactions, αGalCer efficiently stimulates iNKT cells from essentially all inbred mouse strains. Therefore, the ability of αGalCer to induce pregnancy loss was compared between C57BL/6J and BALB/cJ mice during early and midgestation. Although both strains experienced similar rates of pregnancy loss when αGalCer was administered during early gestation (day 6.5 p.c.), only C57BL/6J was susceptible to iNKT cell-mediated pregnancy loss during midgestation (Fig. 2 A). BALB/cJ mice, in contrast, exhibited no overt signs of pregnancy loss at this stage of gestation. To explore this phenomenon further, we examined BALB/cJ uteri at various times after administration of αGalCer on day 9.5 p.c. No increase in the rate of pregnancy loss was seen 3 or 6 days (days 12.5 and 15.5 p.c., respectively) after αGalCer administration (Fig. 2 B). Furthermore, when BALB/cJ females were injected with αGalCer on day 9.5 p.c. and left to proceed to term, there was no significant decrease in litter size (Fig. 2 C). Thus, in contrast to C57BL/6J mice, which were susceptible at all stages of gestation, BALB/cJ mice were susceptible to iNKT cell-mediated pregnancy loss during early gestation but were resistant during midgestation.

Fig. 2.
View larger version:
Fig. 2.

Gestation stage dictates susceptibility to iNKT cell-mediated pregnancy loss. (A) Strain-dependent variations in midgestation, but not early gestation, pregnancy loss. Timed pregnant C57BL/6J and BALB/cJ mice were administered αGalCer or vehicle control during early gestation (day 6.5 p.c) or midgestation (day 9.5 p.c.), followed by examination of the uteri on day 12.5 p.c. Data are presented as follows: (number of females pregnant at the time of euthanasia/number of females with a vaginal plug)·100 (n = 7–10 mice for each condition). ∗, P < 0.05. (B) αGalCer or vehicle control was administered to timed pregnant BALB/cJ mice on day 9.5 p.c. Resorption rate was then calculated at various times after injection. No increase in resorption was observed out to day 15. ○, vehicle-injected mice; •, αGalCer-injected mice. (C) Normal litter size in BALB/cJ mice injected with αGalCer during midgestation. BALB/cJ mice were injected on day 9.5 p.c. with vehicle or αGalCer and allowed to proceed to term. Data are presented as the number of live pups ± SD (n = 5 mice per condition).


To investigate these differences during midgestation further, we examined the effect of αGalCer in other strains of mice. None of the strains we examined exhibited any overt signs of pregnancy loss and, thus, were similar to the BALB/cJ strain (Fig. 3 A). CBA/J mice, however, did exhibit a marked increase in the frequency of resorbing embryos (Fig. 3 B), which corresponded to a significant decrease in the number of viable embryos (Table 2). An increase in resorption may also have occurred in B10.PL/J, although, because of variability, a level of significance was not reached. C3H/HeJ mice, in contrast, were completely resistant to the effect of αGalCer (Fig. 3 B and Table 2). Thus, iNKT cell-mediated pregnancy loss did not exhibit itself as an all-or-none phenomenon. Rather, a range of phenotypes was observed depending on the genetic background.

Fig. 3.
View larger version:
Fig. 3.

Genetic control of susceptibility to iNKT cell-mediated pregnancy loss during midgestation. (A) Pregnancy rate 72 h after administration of αGalCer or vehicle on day 9.5 p.c. (B) Variation in resorption rate among three different inbred strains of mice 72 h after administration of αGalCer or vehicle on day 9.5 p.c. Data are represented as mean resorption rate ± SD. (C) Susceptibility to iNKT cell-mediated pregnancy loss exhibited an incomplete dominant phenotype. Comparison of midgestation resorption frequency among C57BL/6J, BALB/cJ, and (BALB/cJ × B6)F1 hybrid (CB6F1/J). αGalCer or vehicle control was administered to timed pregnant mice at day 9.5 p.c., and the resorption rate was calculated on day 12.5 p.c. ○, vehicle-injected mice; •, αGalCer-injected mice.


View this table:
Table 2.

Strain-dependent variation in iNKT cell-mediated pregnancy loss


To examine more closely the influence of genetic background on iNKT cell-mediated pregnancy loss, we examined pregnant (BALB/cJ × C57BL/6J)F1 mice (CB6F1/J). Although αGalCer administration at day 9.5 p.c. did not result in overt pregnancy loss in these mice (Fig. 3 A), it did result in a significant increase in the frequency of resorbing embryos compared with vehicle-injected controls (Fig. 3 C), which corresponded to a significant decrease in the number of viable embryos (Table 2). Therefore, susceptibility to αGalCer-induced pregnancy loss was inherited as an incompletely dominant trait, suggesting complex genetic control. Collectively, these data suggested that, during midgestation, the genetic background was critical in dictating phenotypes of iNKT cell-mediated pregnancy loss that ranged from complete susceptibility to complete resistance.

The Requirement for Perforin Depends on the Stage of Gestation.

The fundamental differences between early and midgestation iNKT cell-mediated pregnancy loss suggested the possibility that there may be differences in the mechanisms of early and midgestation pregnancy loss induced by αGalCer. Because perforin had been demonstrated to be necessary for iNKT cell-mediated pregnancy loss (19), we investigated its requirement at different stages of gestation. αGalCer was administered to perforin-deficient mice (C57BL/6 background) mice during early (day 6.5 p.c.) and midgestation (day 9.5 p.c.). Examination on day 12.5 p.c. revealed striking differences in their susceptibility to αGalCer. During early gestation, iNKT cell-mediated pregnancy loss was found to be completely dependent on perforin (Fig. 4), confirming the findings of Ito et al. (19). In contrast, administration of αGalCer during midgestation resulted in complete resorption of embryos, suggesting that it was not required for iNKT cell-mediated pregnancy loss during midgestation. No difference in serum cytokine levels between αGalCer-treated pregnant C57BL/6J and perforin-deficient mice was observed, ruling out any indirect effects of perforin on cytokine production (data not shown). Therefore, these data suggested that iNKT cell-mediated pregnancy loss occurs by two distinct mechanisms, a perforin-dependent mechanism operating early in gestation, and a perforin-independent mechanism operating during mid/late gestation pregnancy loss. Taken together, these data indicated that the emergence of strain-dependent effects was inversely correlated with the requirement for perforin.

Fig. 4.
View larger version:
Fig. 4.

A requirement for perforin in early gestation, but not midgestation, pregnancy loss. Perforin−/− mice on a C57BL/6J background were injected with αGalCer or vehicle on either day 6.5 or 9.5 p.c. Implantation sites were examined on day 12.5 p.c. Although perforin was required for the induction of early gestation pregnancy loss, it was completely dispensable during midgestation. Data are represented as mean resorption frequency ± SD (n = 5–7 mice for each condition). ∗∗∗, P < 0.001.


A Correlation Between Pregnancy Loss and Serum Cytokine Levels During Midgestation.

Next, we investigated the underlying differences between the C57BL/6J and BALB/cJ strains that accounted for their differing susceptibility to iNKT cell-mediated pregnancy loss at midgestation. Because IFN-γ and TNF, like perforin, had been implicated in αGalCer-induced pregnancy loss during early gestation (19), we postulated that strain-dependent variation in the elaboration of these and other cytokines between C57BL/6J and BALB/cJ mice may have been responsible for the differential effect of αGalCer in these two strains. Pregnant C57BL/6J and BALB/cJ mice were injected with αGalCer on day 9.5 p.c., and blood was obtained at various times after injection for the measurement of serum cytokine levels. Levels of serum TNF, IL-2, and IL-4 increased rapidly and were highest 2 h after αGalCer administration (Fig. 5). In contrast, serum IFN-γ levels rose more slowly, reaching their highest levels 12 to 24 h after αGalCer administration (Fig. 5), as has been observed in ref. 31. Interestingly, levels of serum TNF, IL-2, and IL-4 were dramatically higher in resorbing C57BL/6J mice compared with resistant BALB/cJ mice. Levels of serum IFN-γ, in contrast, were similar between the two strains. Strain-dependent variation in serum cytokine production was not due to pregnancy because it was also observed in age-matched nonpregnant females (data not shown). Thus, during midgestation when perforin is not required, susceptibility to iNKT cell-mediated pregnancy loss is correlated with increased cytokine production resulting from iNKT cell stimulation.

Fig. 5.
View larger version:
Fig. 5.

Higher levels of serum cytokine production in mice destined for pregnancy loss. αGalCer was administered to pregnant C57BL/6J and BALB/c mice at day 9.5 p.c. Serum samples obtained at various time points after injection were assessed for cytokine levels. Levels of TNF, IL-2, and IL-4 reached their peak 2 h after administration of αGalCer and were present at significantly higher levels in serum from C57BL/6J mice than from BALB/cJ mice (∗∗∗, P < 0.001; ∗∗, P < 0.01). In contrast, the level of serum IFN-γ reached a peak 12–24 h after administration of αGalCer, and there was little difference between the two strains. Mice injected with vehicle control did not exhibit increases in serum cytokines (data not shown). Each time point represents measurements from two to three pregnant mice.


Previous SectionNext Section

Discussion

CD1d-restricted iNKT cells are a recently described unique T cell population known to accumulate in the decidua, but their physiological function there is unknown (18, 19). In this article, we demonstrate that activation of iNKT cells results in both early gestation pregnancy loss and preterm birth and that there are distinct differences between the mechanisms operating at different stages of gestation. The data presented here support a model in which perforin-mediated and cytokine-mediated events are required in early gestation pregnancy loss and in which proinflammatory cytokines alone are sufficient to induce pregnancy loss at midgestation. Our data suggest that this switch from a perforin-dependent mechanism to a cytokine-dominated mechanism renders the pregnant mother susceptible to strain-dependent variations in proinflammatory cytokine production.

The mechanism through which perforin contributes to iNKT cell-mediated pregnancy loss remains unclear. As has been suggested, αGalCer may lead directly to perforin-dependent killing of fetal trophoblasts by iNKT cells (19). Alternatively, NK cells, which have been demonstrated to be indirectly activated by iNKT cells (32), may be the effectors ultimately responsible for mediating pregnancy loss. Perforin-containing NK cells are abundant at the maternal–fetal interface (33–35), as are macrophages (36) and dendritic cells (37), which have been implicated as intermediates in iNKT–NK cell signaling (38). An iNKT-directed NK-dependent mechanism of immune-mediated pregnancy loss is also consistent with a recent article implicating NK cells in the disruption of the reproductive endocrine system through a TNF-dependent mechanism (39). Unlike iNKT cell activation, however, which induced pregnancy loss at all stages of gestation, NK cell activation was only efficient during early gestation, suggesting the possibility that these two models may operate through common pathways in early gestation, but that iNKT cells must use independent mechanisms that allow it to mediate pregnancy loss after midgestation.

Concomitant with the switch from a perforin-dependent mechanism to a perforin-independent mechanism was the emergence of strain-dependent variation in the susceptibility to iNKT cell-mediated pregnancy loss. Susceptibility was correlated with increased production of serum cytokines, including TNF and IL-2, which have been associated with pregnancy loss (12, 40). The observation that IL-4 was also elevated in susceptible C57BL/6 mice argued against a bias in cytokine production. Rather, it suggested either increased activation or elevated numbers of iNKT cells in these mice. Although strain differences in iNKT cell number have been described in certain mouse strains (41, 42), a previous study found no difference in the frequency of iNKT cells between C57BL/6J and BALB/cJ mice (43). Therefore, genetic control of the αGalCer-elicited serum cytokine response may occur downstream of αGalCer-activated iNKT cells. Possible candidates are dendritic cells (44), macrophages (38), and NK cells (31, 45), all of which have been demonstrated to function downstream of activated iNKT cells. Further analysis of these subsets will be necessary to determine their contribution to iNKT cell-mediated pregnancy loss.

The correlation between strain-dependent susceptibility to iNKT cell-mediated pregnancy and elevated serum cytokine levels was apparent at day 9.5 p.c., but not before. In mice, this stage of gestation coincides with the establishment of uteroplacental blood circulation (equivalent to ≈11–12 weeks in humans) (46). It is possible, therefore, that newly developed vascularization of the placenta engenders new avenues of susceptibility to damage initiated by proinflammatory mediators. Possible mechanisms may be thrombotic events initiated by TNF and IFN-γ (17) or direct cytotoxicity of trophoblasts by TNF (47). It is also possible that TNF may induce production of proinflammatory lipid mediators that directly exert their effects on uterine smooth muscle (48, 49). Thus, we speculate that gestation-dependent physiological changes in placental development render the embryo highly susceptible to variations in proinflammatory cytokine production.

The influence of genetic background on iNKT cell-mediated pregnancy loss suggests that it may be a useful model to investigate the links between proinflammatory cytokine production and preterm birth. Preterm birth in humans is linked to intrauterine infection (8, 9). Human epidemiological studies have demonstrated an association between preterm birth and certain TNF gene polymorphisms (14, 50, 51). Taken together, the ability of iNKT cells to rapidly secrete large amounts of cytokines, the recent demonstration that they are capable of recognizing certain bacterial glycosphingolipids (27–29), and our data demonstrating that susceptibility to iNKT cell-mediated pregnancy correlates with strain-dependent variations in TNF production suggests a potential role for these cells in infection-associated preterm birth.

Previous SectionNext Section

Materials and Methods

Mice.

C57BL/6J, BALB/cJ, CBA/J, C3H/HeJ, B10.PL, CB6F1/J, and perforin-deficient (C57BL/6-Prf1tm1sdz/J) mice were purchased from The Jackson Laboratory. Perforin−/− mice were generated on a C57BL/6J background. CD1d-deficient mice were bred onto the C57BL/6 background for six generations (52). Jα18−/− mice were kindly provided by M. Taniguchi (RIKEN Institute, Yokohama City, Japan) and were generated as described in ref. 53. All mice used in these experiments were between 6 and 8 weeks of age and were maintained in ventilated racks in sterile cages in a specific pathogen-free facility. All experiments were performed with the approval of the Institutional Animal Care and Use Committee.

Induction of Abortion by αGalCer.

Timed pregnant syngeneic matings were conducted by using a single pair of mice per cage. The morning of detection of the vaginal plug was considered day 0.5 p.c. Pregnant females were separated in individual cages and given food and water ad libitum. αGalCer (a gift from Kirin Pharmaceuticals, Tokyo) resuspended in vehicle (PBS, pH 7.2/0.5% polysorbate-20) or vehicle alone was delivered i.p. in a 100-μl volume. Except where indicated, αGalCer was administered at a dose of 100 μg/kg. Mice were killed by CO2 asphyxiation, followed by removal of the uterus to assess the status of pregnancy and for macroscopic examination and subsequent histological analysis of the fetoplacental units. Unless otherwise indicated, mice were euthanized on day 12.5 p.c. At this stage of gestation, fetoplacental units undergoing resorption can be clearly distinguished from their viable counterparts on the basis of size and the presence of extensive tissue wasting and hemorrhage. The resorption rate was calculated as follows: (number of resorbing embryos/total number of embryos)·100. The pregnancy rate was calculated as follows: (number of females pregnant at the time of euthanasia/number of females with a vaginal plug)·100.

Histology and Immunohistochemistry.

For hematoxylin/eosin staining, individual fetoplacental units were fixed overnight at 4°C in Bouin’s fixative (Sigma–Aldrich) after which they were washed in an ethanol/xylene series before paraffin embedding. Sections (7 μm) were rehydrated and stained with Gill’s hematoxylin/eosin according to standard procedures.

For immunohistochemical analysis, individual fetoplacental units were fixed overnight at 4°C in freshly prepared 4% paraformaldehyde in PBS, pH 7.2, after which they were washed in an ethanol/xylene series before paraffin embedding. Serial sections (7 μm) were rehydrated in PBS, pH 7.2, after which they were incubated in 3% hydrogen peroxide in methanol for 5 min. Sections were washed in PBS, then stained with rabbit anti-fibrinogen/fibrin (DAKO) or with normal rabbit antisera (Jackson ImmunoResearch) for 2 h at room temperature. Staining was visualized by using a streptavidin ABC system with diaminobenzidine as a substrate (Vectastain; Vector Laboratories).

Serum Cytokine Analysis.

Timed pregnant mice were injected with either αGalCer (100 μg/kg, i.p.) or vehicle control on day 9.5 p.c. Blood was obtained from the tail vein at various times after injection. Separate mice were used for each time point. All serum samples were frozen at −20°C until they were used for cytokine measurements. Serum cytokine levels were measured by using cytometric bead arrays according to the manufacturer’s instructions (BD Biosciences Pharmingen). Diluted serum (50 μl) was incubated with cytometric beads and a mixture of phycoerythrin-labeled anti-cytokine mAbs for 2 h at room temperature. Beads were washed once and then analyzed immediately by FACS (Becton Dickinson). A standard curve of each cytokine was generated in parallel. Collected FACS data were analyzed by using cba software (BD Biosciences Pharmingen).

Statistics.

Pregnancy rates were evaluated in 2 × 2 contingency tables by using Fisher’s exact test. Tests for differences in serum cytokine responses were performed by using two-way ANOVA with Bonferroni post hoc analysis. All other tests were evaluated by using the Mann–Whitney rank-sum test. P values < 0.05 were considered significant.

Previous SectionNext Section

Acknowledgments

We thank Kirin Pharmaceuticals for providing αGalCer; Dr. Masaru Taniguchi (RIKEN Institute, Yokohama City, Japan) for providing Jα18-deficient mice; and S. Brian Wilson, Adrian Erlebacher, and members of the Strominger Laboratory for helpful discussions. This work was supported in part by National Institutes of Health Research Grant AI053330 (to J.L.S.).

Previous SectionNext Section

Footnotes

  • §To whom correspondence should be addressed. E-mail: jlstrom{at}fas.harvard.edu

  • Author contributions: J.E.B. and J.L.S. designed research; J.E.B., N.N., and L.N. performed research; M.A.E. contributed new reagents/analytic tools; J.E.B. analyzed data; and J.E.B. wrote the paper.

  • Conflict of interest statement: No conflicts declared.

  • Abbreviations:
    NK,
    natural killer;
    iNKT,
    invariant natural killer T cell;
    αGalCer,
    α-galactosylceramide;
    p.c.,
    postcoitus.
Previous Section
 

References

  1. ↵
    1. Laird S. M.,
    2. Tuckerman E. M.,
    3. Cork B. A.,
    4. Linjawi S.,
    5. Blakemore A. I.,
    6. Li T. C.
    (2003) Hum. Reprod. Update 9:163–174, pmid:12751778.
    Abstract/FREE Full Text
  2. ↵
    1. Lockwood C. J.,
    2. Kuczynski E.
    (1999) J. Perinat. Med 27:5–20, pmid:10343929.
    CrossRefMedlineISI
  3. ↵
    1. Gendron R. L.,
    2. Farookhi R.,
    3. Baines M. G.
    (1990) J. Reprod. Fertil 89:277–284, pmid:2374121.
    Abstract
  4. ↵
    1. Emmer P. M.,
    2. Nelen W. L.,
    3. Steegers E. A.,
    4. Hendriks J. C.,
    5. Veerhoek M.,
    6. Joosten I.
    (2000) Hum. Reprod 15:1163–1169, pmid:10783371.
    Abstract/FREE Full Text
  5. ↵
    1. Aoki K.,
    2. Kajiura S.,
    3. Matsumoto Y.
    (1995) Lancet 345:1340–1342, pmid:7752757.
    CrossRefMedlineISI
  6. ↵
    1. Emmer P. M.,
    2. Steegers E. A.,
    3. Kerstens H. M.,
    4. Bulten J.,
    5. Nelen W. L.,
    6. Boer K.,
    7. Joosten I.
    (2002) Hum. Reprod 17:1072–1080, pmid:11925408.
    Abstract/FREE Full Text
  7. ↵
    1. de Fougerolles A. R.,
    2. Baines M. G.
    (1987) J. Reprod. Immunol 11:147–153, pmid:3625610.
    CrossRefMedlineISI
  8. ↵
    1. Romero R.,
    2. Espinoza J.,
    3. Mazor M.
    (2004) Fertil. Steril 82:799–804, pmid:15482749.
    CrossRefMedlineISI
  9. ↵
    1. Hillier S. L.,
    2. Martius J.,
    3. Krohn M.,
    4. Kiviat N.,
    5. Holmes K. K.,
    6. Eschenbach D. A.
    (1988) N. Engl. J. Med 319:972–978, pmid:3262199.
    Abstract
  10. ↵
    1. Holers V. M.,
    2. Girardi G.,
    3. Mo L.,
    4. Guthridge J. M.,
    5. Molina H.,
    6. Pierangeli S. S.,
    7. Espinola R.,
    8. Xiaowei L. E.,
    9. Mao D.,
    10. Vialpando C. G.,
    11. Salmon J. E.
    (2002) J. Exp. Med 195:211–220, pmid:11805148.
    Abstract/FREE Full Text
  11. ↵
    1. Xu C.,
    2. Mao D.,
    3. Holers V. M.,
    4. Palanca B.,
    5. Cheng A. M.,
    6. Molina H.
    (2000) Science 287:498–501, pmid:10642554.
    Abstract/FREE Full Text
  12. ↵
    1. Hill J. A.,
    2. Polgar K.,
    3. Anderson D. J.
    (1995) J. Am. Med. Assoc 273:1933–1936.
    Abstract
  13. ↵
    1. Chaouat G.,
    2. Menu E.,
    3. Clark D. A.,
    4. Dy M.,
    5. Minkowski M.,
    6. Wegmann T. G.
    (1990) J. Reprod. Fertil 89:447–458, pmid:2119428.
    Abstract
  14. ↵
    1. Goepfert A. R.,
    2. Goldenberg R. L.,
    3. Andrews W. W.,
    4. Hauth J. C.,
    5. Mercer B.,
    6. Iams J.,
    7. Meis P.,
    8. Moawad A.,
    9. Thom E.,
    10. VanDorsten J. P.,
    11. et al.
    (2001) Am. J. Obstet. Gynecol 184:483–488, pmid:11228507.
    CrossRefMedlineISI
  15. ↵
    1. Hirsch E.,
    2. Saotome I.,
    3. Hirsh D.
    (1995) Am. J. Obstet. Gynecol 172:1598–1603, pmid:7538729.
    CrossRefMedlineISI
  16. ↵
    1. Berman J.,
    2. Girardi G.,
    3. Salmon J. E.
    (2005) J. Immunol 174:485–490, pmid:15611274.
    Abstract/FREE Full Text
  17. ↵
    1. Clark D. A.,
    2. Chaouat G.,
    3. Arck P. C.,
    4. Mittruecker H. W.,
    5. Levy G. A.
    (1998) J. Immunol 160:545–549, pmid:9551885.
    Abstract/FREE Full Text
  18. ↵
    1. Boyson J. E.,
    2. Rybalov B.,
    3. Koopman L. A.,
    4. Exley M.,
    5. Balk S. P.,
    6. Racke F. K.,
    7. Schatz F.,
    8. Masch R.,
    9. Wilson S. B.,
    10. Strominger J. L.
    (2002) Proc. Natl. Acad. Sci. USA 99:13741–13746, pmid:12368486.
    Abstract/FREE Full Text
  19. ↵
    1. Ito K.,
    2. Karasawa M.,
    3. Kawano T.,
    4. Akasaka T.,
    5. Koseki H.,
    6. Akutsu Y.,
    7. Kondo E.,
    8. Sekiya S.,
    9. Sekikawa K.,
    10. Harada M.,
    11. et al.
    (2000) Proc. Natl. Acad. Sci. USA 97:740–744, pmid:10639149.
    Abstract/FREE Full Text
  20. ↵
    1. Tsuda H.,
    2. Sakai M.,
    3. Michimata T.,
    4. Tanebe K.,
    5. Hayakawa S.,
    6. Saito S.
    (2001) Am. J. Reprod. Immunol 45:295–302, pmid:11432404.
    CrossRefMedlineISI
  21. ↵
    1. Wang S.,
    2. Li C.,
    3. Kawamura H.,
    4. Watanabe H.,
    5. Abo T.
    (2002) Cell. Immunol 215:98–105, pmid:12142041.
    CrossRefMedlineISI
  22. ↵
    1. Kawano T.,
    2. Cui J.,
    3. Koezuka Y.,
    4. Toura I.,
    5. Kaneko Y.,
    6. Motoki K.,
    7. Ueno H.,
    8. Nakagawa R.,
    9. Sato H.,
    10. Kondo E.,
    11. et al.
    (1997) Science 278:1626–1629, pmid:9374463.
    Abstract/FREE Full Text
  23. ↵
    1. Zhou D.,
    2. Mattner J.,
    3. Cantu C., 3rd,
    4. Schrantz N.,
    5. Yin N.,
    6. Gao Y.,
    7. Sagiv Y.,
    8. Hudspeth K.,
    9. Wu Y. P.,
    10. Yamashita T.,
    11. et al.
    (2004) Science 306:1786–1789, pmid:15539565.
    Abstract/FREE Full Text
  24. ↵
    1. Spada F. M.,
    2. Koezuka Y.,
    3. Porcelli S. A.
    (1998) J. Exp. Med 188:1529–1534, pmid:9782130.
    Abstract/FREE Full Text
  25. ↵
    1. Kawano T.,
    2. Nakayama T.,
    3. Kamada N.,
    4. Kaneko Y.,
    5. Harada M.,
    6. Ogura N.,
    7. Akutsu Y.,
    8. Motohashi S.,
    9. Iizasa T.,
    10. Endo H.,
    11. et al.
    (1999) Cancer Res 59:5102–5105, pmid:10537282.
    Abstract/FREE Full Text
  26. ↵
    1. Exley M.,
    2. Porcelli S.,
    3. Furman M.,
    4. Garcia J.,
    5. Balk S.
    (1998) J. Exp. Med 188:867–876, pmid:9730888.
    Abstract/FREE Full Text
  27. ↵
    1. Mattner J.,
    2. Debord K. L.,
    3. Ismail N.,
    4. Goff R. D.,
    5. Cantu C., 3rd,
    6. Zhou D.,
    7. Saint-Mezard P.,
    8. Wang V.,
    9. Gao Y.,
    10. Yin N.,
    11. et al.
    (2005) Nature 434:525–529, pmid:15791258.
    CrossRefMedline
  28. ↵
    1. Kinjo Y.,
    2. Wu D.,
    3. Kim G.,
    4. Xing G. W.,
    5. Poles M. A.,
    6. Ho D. D.,
    7. Tsuji M.,
    8. Kawahara K.,
    9. Wong C. H.,
    10. Kronenberg M.
    (2005) Nature 434:520–525, pmid:15791257.
    CrossRefMedline
  29. ↵
    1. Sriram V.,
    2. Du W.,
    3. Gervay-Hague J.,