IL-10-Null Mice Inflammation-Induced Fetal Demise in Uterine NK Cells Mediate

Uterine NK Cells Mediate Inflammation-Induced Fetal Demise in IL-10-Null Mice

This information is current as Shaun P. Murphy, Loren D. Fast, Nazeeh N. Hanna and of September 27, 2021. Surendra Sharma J Immunol 2005; 175:4084-4090; ; doi: 10.4049/jimmunol.175.6.4084 http://www.jimmunol.org/content/175/6/4084 Downloaded from

References This article cites 52 articles, 13 of which you can access for free at: http://www.jimmunol.org/content/175/6/4084.full#ref-list-1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

by guest on September 27, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2005 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Uterine NK Cells Mediate Inﬂammation-Induced Fetal Demise in IL-10-Null Mice

Shaun P. Murphy,* Loren D. Fast,† Nazeeh N. Hanna,‡ and Surendra Sharma2*

Specialized NK cells are recruited in high numbers to the mammalian embryo implantation sites, yet remain pregnancy compatible. It is not well understood whether uterine NK (uNK) cells become adversely activated and mediate fetal demise, a common complication of early pregnancy. In this study we show that mating of IL-10؊/؊ mice resulted in fetal resorption or intrauterine growth restriction in response to very low doses of LPS. Pregnancy in congenic wild-type mice was normal even at 10-fold higher LPS doses. Fetal resorption in IL-10؊/؊ mice was associated with a signiﬁcant increase in uNK cell cytotoxic activation and invasion into the placenta. Depletion of uNK cells, TNF-␣ neutralization, or IL-10 administration rescued pregnancy in LPS- treated IL-10؊/؊ animals. Our results identify an immune mechanism of fetal demise involving IL-10 deﬁciency, NK cells, and

inﬂammation. These results may provide insight into adverse pregnancy outcomes in humans. The Journal of Immunology, 2005, Downloaded from 175: 4084–4090.

pproximately half of all human blastocyst implantations populations of macrophages, T cells, and other immune cells in the result in failed pregnancy. Multiple factors may contrib- uterus during early stages of pregnancy (11). These cells populate A ute to this failure, including genetic and developmental the uterus after implantation in mice and during the proliferative anomalies of the embryo. However, many cases of pregnancy fail- stage of the menstrual cycle in humans, where they increase in http://www.jimmunol.org/ ure are thought to be associated with maternal immune-mediated number until midgestation. Thereafter, their numbers decline rap- mechanisms. A successful pregnancy is marked by an intricate idly (11, 12). Most studies to date have shown a pregnancy-com- regulation of the immune system at the maternal-fetal interface, patible role for uNK cells in reproduction, mainly through their resulting in tolerance of the semiallogeneic fetus (1–3). It is be- regulation of decidualization, production of pregnancy-compatible lieved that disruptions in this regulation may result in pregnancy cytokines, and cross-talk with the trophoblast. During pregnancy, failures. In this context, an array of factors may impel the maternal NK cell-deficient mice display abnormalities in decidual artery immune system toward antifetal responses. Indeed, infection and remodeling and trophoblast invasion, possibly due to the lack of other inflammatory insults are associated with a host of pregnancy uNK cell-derived IFN-␥ (13–16). In humans, it has been suggested complications in humans (4, 5). that defective trophoblast invasion and placental development are by guest on September 27, 2021 The maternal-fetal interface constitutes a unique environment associated with altered uNK cell function and preeclampsia (17, for innate and adaptive immune responses (6, 7). However, the 18). Curiously, although uNK cells display an activated phenotype implanted embryo and developing fetus during normal pregnancy (19, 20), to date, no in vivo role for uNK cell cytotoxicity has been are capable of suppressing these immune responses (8). There also identified. It is tempting to propose that although uNK cells nor- appear to be pregnancy-compatible alterations in the maternal im- mally contribute to the success of pregnancy, they may exert a mune system that protect against a graft-vs-host reaction from the negative role given aberrant intrauterine conditions. fetal immune system (8). However, aberrant immune regulation For the most part, tolerogenic processes that control the maternal may result in adverse pregnancy outcomes. In mice, disruptions in immune system and protect the fetus are probably local and temporal T cell regulation have been shown to result in immune-mediated at the maternal-fetal interface. The search for components of the in- loss of the fetal allograft (9, 10). In normal pregnancy, no adverse trauterine milieu that contribute to successful pregnancy outcome and immune responses are mounted despite the presence of high num- control detrimental innate and adaptive inflammatory immune re- 3 bers of uterine NK (uNK) cells (up to 65–70%) as well as smaller sponses has implicated cytokines, neuroendocrine immunomodula- tors, complement regulators, and nutrition-based factors (9, 21, 22). Among cytokines, the anti-inflammatory cytokine IL-10 is especially † *Departments of Pediatrics and Pathology, Women and Infants’ Hospital, and De- attractive for a critical role in pregnancy because of its regulatory partment of Medicine, Rhode Island Hospital-Brown University, Providence, RI 02905; and ‡Division of Neonatology, University of Medicine and Dentistry of New relationship with other intrauterine modulators and its wide range of Jersey-Robert Wood Johnson Medical School, New Brunswick, NJ 08903 immunosuppressive activities (23). Significantly, its local production Received for publication May 27, 2005. Accepted for publication July 11, 2005. by gestational tissues is well documented (24–26). We observed that The costs of publication of this article were defrayed in part by the payment of page IL-10 expression by the human placenta was gestational age depen- charges. This article must therefore be hereby marked advertisement in accordance dent, with significant expression through the second trimester coupled with 18 U.S.C. Section 1734 solely to indicate this fact. with attenuation at term (26). IL-10 expression was also found to be 1 This work was supported by National Institutes of Health Grants HD41701 and P20RR018728 (to S.S.). poor in decidual and placental tissues from unexplained spontaneous 2 Address correspondence and reprint requests to Dr. Surendra Sharma, Department abortion cases (27) and from deliveries associated with preterm labor of Pediatrics, 101 Dudley Street, Women and Infants’ Hospital of Rhode Island, and pre-eclampsia (our unpublished observations). However, the Providence, RI 02905. E-mail address: [email protected] mechanism(s) by which IL-10 protects the fetus remains poorly un- 3 Abbreviations used in this paper: uNK, uterine NK; DB, deciduas basalis; gd, ges- derstood. Although IL-10Ϫ/Ϫ mice suffer no pregnancy defects when tational day; IUGR, intrauterine growth restriction; MMP, matrix metalloproteinase; NRS, nonimmune rabbit serum; P, placental labyrinth; PAS, periodic acid-Schiff; mated under pathogen-free conditions (28), these mice eventually de- UMC, uterine mononuclear cell. velop colitis and fail to control intrinsic inflammatory responses (29,

30). It is then plausible that in addition to IL-10 deficiency, an un- the supernatant from target cells lysed with 1 N HCl. The flow cytometry- regulated inflammatory insult resulting from genital tract infections, based cytotoxicity assay was performed according to the manufacturer’s Ј environmental factors, and/or hormonal anomalies during gestation protocol (Molecular Probes). Briefly, target cells were treated with 3,3 - dioctadecyloxacarbocyanine for 20 min at 37°C. Effector cells were mixed may lead to adverse pregnancy outcomes. with target cells as described above, and propidium iodide was added to A likely potential mechanism for IL-10-mediated protection of effector/target cell cultures, which were incubated for2hat37°C in 10% pregnancy is through direct action on decidual immune cells. In CO2. Cellular events were then immediately acquired on a flow cytometer. this study we demonstrate a novel regulatory relationship among Histochemistry IL-10 deficiency, inflammation, enhanced uNK cell activity, and pregnancy loss. Exposure of IL-10Ϫ/Ϫ, but not wild-type, mice to Individual, intact utero-placental units were isolated and fixed with 10% low doses of LPS provokes vigorous uNK cell cytotoxicity and buffered formalin for 24 h. Tissue was processed for histological staining with hematoxylin and periodic acid-Schiff reagent (PAS) as previously invasiveness into the placental zone, leading to dose-dependant described (32) fetal demise or intrauterine growth restriction (IUGR). Immu- nodepletion of uNK cells, IL-10 administration, or treatment with Results anti-TNF-␣ Ab, but not anti-IFN-␥ Ab, reverses these LPS-in- IL-10Ϫ/Ϫ mice experience pregnancy loss in response to very duced pregnancy defects. Taken together, these results imply a low doses of LPS crucial regulatory cross-talk between IL-10 and uNK cells in the We studied the effect of IL-10 on syngeneic and allogeneic preg- prevention of inflammation-induced adverse pregnancy outcomes. nancies in a mouse model of IL-10 deficiency coupled with the inflammatory signal provided by LPS administration. Timed syn- Materials and Methods Ϫ Ϫ Downloaded from geneic matings between C57BL/6 IL-10 / mice or allogeneic Mice Ϫ Ϫ matings between female C57BL/6 IL-10 / mice (H-2b) and male Ϫ Ϫ The mice used in this study, C57BL/6, C57BL/6 IL-10 / , NOD, and NOD IL-10Ϫ/Ϫ mice (H-2k) were conducted. Matings using con- Ϫ/Ϫ NOD IL-10 , were obtained from The Jackson Laboratory. All mice genic wild-type (IL-10ϩ/ϩ) mice were set up in parallel as a con- were housed in a specific pathogen-free facility supervised by the Central Research Department of Rhode Island Hospital. All protocols were ap- trol for normal pregnancy outcome (Table I). In the preliminary proved by the Lifespan Animal Welfare Committee and conducted accord- experiments, LPS doses Ͼ1 ␮g/mouse were found to result in http://www.jimmunol.org/ ing to its guidelines. severe placental pathology in IL-10Ϫ/Ϫ mice. Thus, LPS doses of 0.5 or 0.2 ␮g/mouse have been used in all experiments. Pregnant In vivo treatments mice were injected i.p. with 0.5 or 0.2 ␮g of LPS or saline (100 ␮l) Mice received i.p. injections of serotype 026:B6 Escherichia coli LPS on gd 6.5 as described in Materials and Methods. The mice were ␮ (Sigma-Aldrich) at doses of 0.2, 0.5, or 1 g/mouse on gestational day (gd) either killed on gd 10–13 and their uterine horns examined or were 6.5 or an equivalent volume (100 ␮l) of saline. For NK cell depletion, mice received i.p. LPS injections of 0.5 ␮g as described above and i.p. injection allowed to deliver. As shown in Table I, pregnancy progressed Ϫ/Ϫ of rabbit anti-asialo-GM1 (100 ␮l; Wako USA) or anti-NK1.1 (PK-136; normally in matings of IL-10 mice in the absence of LPS treat- 250 ␮g; BD Biosciences) on gd 4, 6.5, and 9. Nonimmune rabbit serum ment. At 0.5 ␮g of LPS/mouse, both allogeneically and syngenei- (NRS; Antibodies, 100 ␮l) or irrelevant isotype Ab was injected in parallel Ϫ/Ϫ cally mated IL-10 mice experienced total fetal loss. However, by guest on September 27, 2021 as a control. IL-10 was administered i.p. in a single dose of 500 ng/mouse on gd 6.5 immediately before LPS administration. Anti-IFN-␥ (XMG1.2) at the same dose, pregnant wild-type mice had successful preg- and anti-TNF-␣ (G281-2626) mAbs (BD Biosciences) were administered nancy with no resorptions. In addition, pregnancy in wild-type i.p. at 750 and 300 ␮g, respectively, on gd 5 and 7 with LPS injection on mice appeared unaffected even at a 10-fold higher dose (data not gd 6.5. shown). The condition of the uterine horns of both wild-type and Ϫ/Ϫ Cell preparation IL-10 mice receiving saline injections was consistent with normal pregnancy (Fig. 1). In contrast, although pregnancy was un- Uterine mononuclear cells (UMC) were obtained by mincing and mechan- affected by administration of 0.5 ␮g of LPS/mouse in wild-type ical dispersion of whole gd 10–13 uterus and placenta (containing entire uterus and placenta with fetus removed) in RPMI 1640 supplemented with mice (Fig. 1), this dose resulted in fetal resorption with an obvious Ϫ/Ϫ 10% FBS, penicillin/streptomycin, and L-glutamine. Single-cell suspen- necrotic appearance in IL-10 mice. By comparison, injection of Ϫ Ϫ sions from uterine horns from three mice were pooled and subsequently 0.2 ␮g of LPS/mouse in syngeneically mated IL-10 / mice (n ϭ subjected to density gradient separation using Fico-Lite LM (Atlanta 4) resulted in IUGR with no obvious resorption or necrosis (Fig. Biologicals). 1). Similar results were observed in allogeneic matings (data not Flow cytometry Isolated UMC were washed in PBS and resuspended in PBS containing Table I. Pregnancy outcome in IL-10Ϫ/Ϫ mice 0.1% sodium azide and 1% BSA. FITC-conjugated anti-CD45 (30-511), PE-conjugated anti-NK1.1 (PK136), PerCP-conjugated anti-CD3 (145- 2C11; BD Biosciences), and PerCP-conjugated anti-F4/80 (Serotec) were Fetal Mating Combination Treatment Resorptiona then added simultaneously and allowed to incubate at 4°C for 30 min. Fluorochome-conjugated isotype Abs of irrelevant specificity were used as C57BL/6 IL-10Ϫ/Ϫ ϫ C57BL/6 IL-10Ϫ/Ϫ LPS 34/34b,c controls. Saline 0/11

Cytotoxicity assays C57BL/6 IL-10Ϫ/Ϫ ϫ NOD IL-10Ϫ/Ϫ LPS 9/9 The uNK cell activity was measured using a standard chromium release Saline 0/4

assay as previously described (31) or using a ﬂow cytometry-based system. ϩ ϩ ϩ ϩ ϫ 3 51 C57BL/6 IL-10 / ϫ C57BL/6 IL-10 / LPS 0/24 Target YAC-1 cells (5 10 ) were labeled with 0.15 mCi of Na2( CrO4) (PerkinElmer) for1hat37°C. Effector UMC were added to target cells at Saline 0/6 descending half-fold E:T cell ratios of 50:1 to 1.56:1 in RPMI 1640 plus ϩ/ϩ ϩ/ϩ 10% FBS, and supernatants were harvested after an incubation of5hat C57BL/6 IL-10 ϫ NOD IL-10 LPS 0/15 Saline 0/5 37°C in 10% CO2 and read on a gamma radiation counter. The percent lysis was calculated as: [(sample count Ϫ spontaneous release)/(maximal re- a Ϫ ϫ Fetal resorption refers to 100% fetal demise, as assessed by visual observation lease spontaneous release)] 100. Spontaneous release was assessed by of uterine horns or failure to deliver. the radioactivity detected in the supernatant from target cells incubated b Four animals that received 0.2 ␮g/mouse LPS exhibited IUGR. alone, and maximal release was assessed by the radioactivity detected in c Numbers refer to the number of vaginal plug-positive females. 4086 uNK CELL-MEDIATED FETAL LOSS IN IL-10Ϫ/Ϫ MICE

FIGURE 2. Increase in proportion of uNK cells in response to LPS treatment in IL-10Ϫ/Ϫ mice. A, UMC from saline- or LPS-treated wild-type FIGURE 1. Effect of LPS administration on pregnancy outcome in or IL-10Ϫ/Ϫ mice were isolated on gd 10–13 and subjected to FACS anal- Ϫ Ϫ Ϫ Ϫ wild-type and IL-10 / mice. Pregnant wild-type or IL-10 / C57BL/6 ysis as described in Materials and Methods. The proportion of uterine (Ut) mice were injected with LPS or saline on gd 6.5. Mice were killed on gd NK cells (NK1.1ϩCD3Ϫ), but not that of splenic (Sp) NK cells, increased 12. Upper left, Representative healthy uterine horns from a wild-type in IL-10Ϫ/Ϫ mice. The results are representative of ﬁve experiments, each mouse injected with saline. Upper right, Uterine horns from a wild-type performed with cells pooled from three mice. Downloaded from mouse injected with 0.5 ␮g of LPS. Bottom left, Uterine horns from a saline-injected IL-10Ϫ/Ϫ mouse. Bottom right, Representative pathology Ϫ/Ϫ observed in IL-10 mice injected with either 0.2 ␮g (IUGR) or 0.5 ␮g Uterine NK cells from IL-10Ϫ/Ϫ mice become invasive upon (fetal resorption) of LPS. The arrow indicates a site of fetal resorption. LPS treatment Uterine NK cells are primarily localized to the mesometrial de-

cidua in pregnant wild-type mice through gd 14 (13). At this time, http://www.jimmunol.org/ shown). Our findings demonstrate that pregnancy in IL-10Ϫ/Ϫ trophoblast cells invade the mesometrial zone and replace uNK mice is exceptionally sensitive to the detrimental effects of very cells, which are normally absent from this region during later low doses of LPS. Furthermore, the severity of the effect of LPS on stages of pregnancy. To assess the role of uNK cells in the ob- pregnancy in these animals is dose dependent. served fetal demise, we examined the localization of these cells in response to LPS administration. The high carbohydrate content of Ϫ/Ϫ LPS-induced abortion in IL-10 mice is associated with an the perforin-containing granules of uNK cells can be easily visu- increase in uNK cell number and cytotoxicity alized by staining with PAS (32). Fixed uterine tissue from the Ϫ Ϫ Although uNK cells appear to be primed for cytotoxic activity due syngeneically or allogeneically mated C57BL/6 IL-10 / or wild- to their abundant perforin-containing granules, they remain preg- type congenic females (gd 10 or 12) was examined by PAS/he- by guest on September 27, 2021 nancy compatible during normal gestation. Recent in vitro studies matoxylin staining. In wild-type mice, uNK cells were restricted to have suggested that human peripheral blood NK cells become cy- the mesometrial decidua in response to saline treatment, with no totoxically activated and proliferate in response to LPS, and these effect on their localization or migration after LPS treatment (Fig. Ϫ Ϫ responses are enhanced in the presence of anti-IL-10R-blocking 4). Similarly, uNK cells from IL-10 / mice receiving saline in- Abs (33). Thus, uNK cells from LPS-treated mice were assessed jections also remained localized to the mesometrial decidua (see Ϫ Ϫ for their natural killing activity. UMC were obtained on gd 10–13 Fig. 6, A–D). In contrast, uNK cells from IL-10 / mice treated from syngeneically mated C57BL/6 IL-10Ϫ/Ϫ or congenic wild- with 0.5 ␮g of LPS/mouse began to accumulate at the decidua type mice that had received an i.p. dose of LPS at 0.5 ␮g/mouse or basalis (DB) by gd 10 (Fig. 5, A–D) and penetrated the labyrinth saline on gd 6.5 as described above. In response to LPS adminis- zone by gd 12 (Fig. 5, E–H). In gd 12 tissue, PAS-stained cells tration, the proportion of uNK (NK1.1ϩCD3Ϫ) cells as a percent- were quantitatively fewer. This is probably due to advanced pla- age of the total immune cell (CD45ϩ) population significantly in- cental pathology on gd 12 in response to LPS treatment (Fig. 5, creased in IL-10Ϫ/Ϫ mice (24.9–51.8%) in contrast to congenic E–H). Interestingly, upon administration of 0.2 ␮g of LPS/mouse, wild-type mice (27.8–36.1%) (Fig. 2). Interestingly, at this dose of uNK cells invaded the DB region and the placental zone only by LPS, neither wild-type nor IL-10Ϫ/Ϫ splenic NK cells showed any gd 13 (Fig. 6, E–H), with no apparent fetal pathology. This is increase (Fig. 2). LPS had little effect on the NK cytotoxicity of consistent with the observation of a more moderate effect at this wild-type UMC, as measured by 51Cr release assay (Fig. 3A)ora dose with resultant fetal growth restriction. It is thus possible that flow cytometry-based method (data not shown) on NK cell-spe- delayed penetration of uNK cells into the placenta may lead to cific target YAC-1 cells, whereas cells from LPS-treated IL-10Ϫ/Ϫ IUGR or preterm delivery (our unpublished observations). mice acquired significant cytotoxic activity. Results from two al- Depletion of uNK cells rescues pregnancy in LPS-treated IL- logeneic mating experiments were similar to those described above Ϫ/Ϫ (data not shown). Splenic mononuclear cells from neither IL- 10 mice 10Ϫ/Ϫ nor wild-type mice displayed increased cytotoxicity in re- The increase in uNK cell cytotoxicity and invasiveness observed in sponse to LPS, suggesting a selective effect on uNK cells of LPS LPS-treated IL-10Ϫ/Ϫ mice suggest a role for uNK cells in fetal administration (data not shown). Notably, UMC from animals re- demise. To examine this possibility, we conducted syngeneic mat- ceiving 0.2 ␮g of LPS/mouse displayed an intermediate NK cy- ings as described above. In addition to injection of 0.5 ␮g of LPS/ totoxicity index between animals receiving 0.5 ␮g of LPS/mouse mouse on gd 6.5, we conducted NK cell depletion. Anti-asialo- and animals receiving saline alone (Fig. 3B). Thus, the dose-de- GM1 Ab treatment has been widely used for in vivo NK cell pendant gross pathology observed in LPS-treated IL-10Ϫ/Ϫ mice is depletion in mice. However, the asialo-GM1 Ag can be expressed associated with a graded increase in uNK cell cytotoxic activity as on minor subpopulations of other immune cells. Thus, to delineate well as an increase in the proportion of uNK cells. the specific role of uNK cells in pregnancy pathology, NK cell The Journal of Immunology 4087

FIGURE 3. Increase in cytotoxicity of uNK cells in response to LPS treatment in IL-10Ϫ/Ϫ mice. A, LPS (0.5 ␮g) or saline was injected on gd 6.5, and UMC were isolated from pregnant mice. The uNK cell cytotox- FIGURE 5. Localization of PAS-stained uNK cells in uteri from LPS- icity was tested against YAC-1 target cells as described in Materials and injected IL-10Ϫ/Ϫ mice. Photomicrographs of gd 10 (A–D)orgd12(E–H) Methods. B, Pregnant IL-10Ϫ/Ϫ mice were injected with saline, 0.2 ␮gof uteroplacental tissue from LPS-injected IL-10Ϫ/Ϫ mice are shown. A–D, LPS, or 0.5 ␮g of LPS, and uNK cell cytotoxicity was examined. In all Lower (A) and higher (B–D) power images of gd 10 utero-placental tissue experiments, UMC were isolated from gd 10–13 tissue as described in from LPS-injected IL-10Ϫ/Ϫ mice showing two mesometrial regions (M1

Materials and Methods. Data are the mean Ϯ SD of five (A) and three (B) and M2), DB, and placental labyrinth (P). B and C, uNK cells (arrowheads) Downloaded from experiments. are found in the mesometrium and DB, but not the placenta. E–G, Lower (E) and higher (F–G) power images of gd 12 utero-placental tissue from LPS-injected IL-10Ϫ/Ϫ mouse M1 and M2 and placental labyrinth. F, uNK cells are no longer found in the placental-distal mesometrium. The placen- depletion was performed using either anti-asialo-GM1 polyclonal tal labyrinth and placental-proximal mesometrium by this time contain or anti-NK1.1 mAbs i.p. on gd 4, 6.5, and 9. NRS or irrelevant uNK cells (arrowheads). Bars in A and E, 150 ␮m; bars in B–D and F–H, isotype control Ab was used in parallel as a control. Pregnant mice 15 ␮m. http://www.jimmunol.org/ were then killed on gd 11–13, UMC were isolated, and some of the feto-placental units were fixed for histology. Both anti-asialo-GM1 and anti-NK1.1 treatments successfully depleted uNK cells, as demonstrated by the complete reduction in PAS-positive cells activation and placental migration of uNK cells contribute to fetal (data shown only for anti-asialo-GM1 treatment) in the uterus of demise in the context of IL-10 deficiency. treated animals (Fig. 7C). The successful depletion of uNK cells IL-10 administration or TNF-␣ neutralization abrogates LPS- was also confirmed by flow cytometric demonstration of the re- Ϫ Ϫ induced fetal resorption in IL-10 / mice duction in the NK1.1ϩ population (Fig. 7A, top). Our data also demonstrate that collateral F4/80ϩ macrophage depletion did not NK cell functions are mediated by cytokines such as IFN-␥ and by guest on September 27, 2021 occur as a result of NK cell depletion (Fig. 7A, bottom). Consistent TNF-␣ as well as by the perforin/granzyme cytotoxic pathway (14, with successful NK cell depletion, UMC NK cytotoxicity was 19, 34). The fact that NK cell depletion rescued pregnancy in LPS- Ϫ/Ϫ shown to be abrogated after anti-asialo-GM1 or anti-NK1.1 deple- treated IL-10 mice (Fig. 8 and Table II) prompted us to inves- tion (Fig. 7B). These observations prompted us to assess the effect tigate the roles of IFN-␥ and TNF-␣ in LPS-induced fetal resorp- of NK cell depletion by both anti-asialo-GM1 and anti-NK1.1 tion by in vivo Ab-mediated neutralization of these cytokines as treatment on pregnancy outcome in IL-10Ϫ/Ϫ mice. As shown in Fig. 8 and Table II, NK cell depletion by both Abs successfully rescued pregnancy in LPS-treated animals, implying that cytotoxic

FIGURE 6. Analysis of uNK cell migration from IUGR utero-placental FIGURE 4. Localization of PAS-stained uNK cells in uteri from LPS- tissue. Photomicrographs of gd 13 uteroplacental tissue from saline-in- injected wild-type mice. Photomicrographs of gd 12 uteroplacental tissue jected (A–D) or LPS-injected (E–H) IL-10Ϫ/Ϫ mice are shown. A–D, from saline-injected (B and C)or0.5␮g of LPS-injected (A, D, and E) Lower (A) and higher (B–D) power images of utero-placental tissue from wild-type mice are shown. A–C, Lower (A) and higher (B and C) power saline-injected IL-10Ϫ/Ϫ mice showing three mesometrial regions (M1– images of utero-placental tissue from wild-type mice, showing two me- M3) and DB. As in saline-treated wild-type mice, M1-M3 show numerous sometrial regions (M1 and M2) and DB. M1 and M2 show numerous uNK uNK cells (arrowheads), which are absent from the DB. E–H, Lower (E) cells (arrowheads), which are absent from the DB. D and E, Higher power and higher (F–H) power images of utero-placental tissue from 0.2 ␮gof images of utero-placental tissue from LPS-injected IL-10Ϫ/Ϫ mice. As in LPS-injected IL-10Ϫ/Ϫ mice showing the mesometrium (M) and two pla- wild-type mice, uNK cells in IL-10Ϫ/Ϫ mice remain in the mesometrium cental labyrinthine regions (P1 and P2). The uNK cells (arrowheads) are and do not invade the DB. P, placental labyrinth. Bars in A and E, 150 ␮m; found in the mesometrium, but also have invaded the placental labyrinth. bars in B and C and D and E,15␮m. Bars in A and E, 150 ␮m; bars in B–D and F–H,15␮m. 4088 uNK CELL-MEDIATED FETAL LOSS IN IL-10Ϫ/Ϫ MICE

FIGURE 7. Depletion of uNK cells in LPS-treated IL-10Ϫ/Ϫ mice. A, Expression of NK1.1 and CD3 on UMC preparations from mice treated with LPS or with LPS and anti- asialo-GM1 and expression of NK1.1 and F4/80 on UMC prepared from mice treated with LPS or LPS and anti- NK1.1. Data are representative of five experiments (A, top) and three experiments (A, bottom), each performed with pooled cells from three mice. B, Cytotoxicity of UMC from LPS- and NRS-treated, LPS- and anti- asialo-GM1-treated, or LPS- and anti- NK1.1-treated IL-10Ϫ/Ϫ mice. A and B, UMC were pooled from gd 11–13. B, Data are representative of at least five experiments. C, Representative PAS-hematoxylin-stained histologi- Downloaded from cal sections of gd 12 utero-placental tissue from animals treated with LPS or both LPS and anti-asialo-GM1. Scale bar, 150 ␮m; insets, scale bar, 15 ␮m. http://www.jimmunol.org/ described in Materials and Methods. In parallel, mouse recombi- major accessory LPS receptor. Likewise, NK cell expression of the nant IL-10 was administered i.p. immediately before LPS admin- LPS-signaling TLR4 has not been reported. As such, LPS may istration. Examination of excised uterine horns on gd 11–13 re- exert its effect on NK cells indirectly, perhaps through uterine vealed that IL-10 administration restored normal utero-placental macrophages, dendritic cells, or placental trophoblast cells. In- unit morphology in LPS-treated mice, consistent with normal preg- deed, the LPS-induced NK cell proliferation by human PBMCs nancy (Fig. 8 and Table II). Interestingly, treatment with anti- appears to require dendritic cells (33), possibly via the induction of IFN-␥ mAb failed to restore pregnancy, whereas treatment with NK cell stimulatory cytokines. LPS has been shown to up-regulate by guest on September 27, 2021 anti-TNF-␣ mAb significantly reduced the number of fetal resorp- the expression of IL-12 in placental trophoblast (37) as well as tion sites (Fig. 8). This was confirmed by term delivery outcome both IL-12 and IL-18 in macrophages (38, 39). by IL-10Ϫ/Ϫ mice in response to these treatments. These data sug- Our findings uniquely suggest that IL-10 may be a regulatory gest that TNF-␣, not IFN-␥, is likely to play a critical role in factor for uNK cells in the context of inflammation, as shown by uNK-cell mediated LPS-induced fetal demise in IL-10Ϫ/Ϫ mice. rescue of pregnancy by this cytokine in LPS-treated IL-10Ϫ/Ϫ mice (Fig. 8 and Table II). The precise function of IL-10 in lim- Discussion iting uNK cell-induced pathology is unclear. IL-10 has been im- Our data implicate uNK cells as critical mediators in inflammation- plicated in the control of LPS-induced NK cell proliferation in induced fetal demise. Moreover, IL-10 appears to play a crucial role vitro (33). Moreover, IL-10 can inhibit the production of cytokines in the protection of the fetus from inflammation-associated uNK cell produced in response to LPS that enhance NK cell activity, such as aggression in this model. These findings suggest at least a two-hit mechanism for adverse pregnancy outcomes. Most likely, both inflammation and IL-10 deficiency are required for initiation of antifetal immune responses. During normal pregnancy, mild inflammation either systemically or locally at the maternal-fetal interface, may not lead to fetal demise due to the protective anti-inflammatory effects of the physiological presence of IL-10 or other modulators. However, with insufficient IL-10 production, inflammatory processes may pro- ceed unchecked, leading to increased uNK cell cytotoxic activation, invasiveness, and eventual fetal demise. The restoration of pregnancy in LPS-treated IL-10Ϫ/Ϫ mice upon NK cell depletion suggests that uNK cells are targets for LPS activity leading to fetal demise (Fig. 5). Several studies have shown LPS-mediated effects on NK cells. Treatment of in vitro cultures of isolated human PBMC with LPS induces the proliferation and cytotoxic activation of NK cells (33, 35). Additionally, ␥ Ϫ Ϫ IFN- responses to LPS administration are markedly reduced in FIGURE 8. Restoration of pregnancy in LPS-treated IL-10 / mice. NK cell-deficient mice (36). It is not entirely clear whether the NK Shown are representative gd 12 uterine horns from mice receiving the cell functional effects of LPS treatment are direct or require ac- indicated treatments. Arrows indicate the location of residual fetal-resorb- cessory cells. Neither peripheral nor uNK cells express CD14, a ing sites in anti-TNF-␣ Ab-treated animals. The Journal of Immunology 4089

Table II. NK cell depletion, anti-TNF-␣ treatment, or IL-10 administration rescues pregnancy

Mating Combination Treatment Fetal Resorptiona

C57BL/6 IL-10Ϫ/Ϫ ϫ C57BL/6 IL-10Ϫ/Ϫ LPS ϩ anti-asialo-GM1 0/17b LPS ϩ anti-NK1.1 0/6 LPS ϩ NRS 5/5 LPS ϩ mrIL-10c 0/4 LPS ϩ saline 3/3 LPS ϩ anti-IFN-␥ 4/4 LPS ϩ anti-TNF-␣ 0/6d LPS ϩ isotype Ab 3/3

C57BL/6 IL-10Ϫ/Ϫ ϫ NOD IL-10Ϫ/Ϫ LPS ϩ anti-asialo-GM1 0/8 LPS ϩ NRS 4/4 LPS ϩ mrIL-10 0/3 LPS ϩ saline 2/2

a Fetal resorption refers to 100% fetal demise, as assessed by visual observation of uterine horns. b Numbers refer to the number of vaginal plug-positive females. c mrIL-10, mouse rIL-10. d Three of six uterine horns had a majority of viable fetuses with a small number of resorptions of between one and four fetuses each. Downloaded from

IL-12 and TNF-␣ (23). A possible explanation for the increased rophage-derived TNF-␣ is involved in the observed fetal demise. This invasiveness of uNK cells in response to LPS in the absence of does not, however, rule out a cross-talk between uterine macrophages IL-10 may be through altered chemokine expression. LPS induces and uNK cells, with the latter cells playing a dominant role. In addi-

the expression of a variety of chemokines in macrophages (40), tion, IL-10 administration can rescue pregnancy in LPS-treated IL- http://www.jimmunol.org/ and this effect is antagonized by IL-10 (41, 42). LPS and IL-10 10Ϫ/Ϫ mice, suggesting that this potent anti-inflammatory cytokine also exert antagonistic effects on the expression of matrix metal- can counteract pregnancy-incompatible activities of TNF-␣ and cy- loproteinases (MMPs) in macrophages and trophoblasts. LPS in- totoxic uNK cells. We speculate that uNK cells, in addition to their duces the expression of MMPs (43, 44), whereas IL-10 inhibits well-studied role in decidualization, may serve as a fail-safe mecha- expression of MMPs and induces the expression of tissue inhibitor nism to terminate pregnancy when excessive inflammatory or other of metalloproteinase (45, 46). Therefore, reduced IL-10 production insults are experienced, sparing the mother from the energy waste of could potentially lead to increased MMP expression in uNK cells, carrying a damaged or abnormal fetus. thus enhancing their invasiveness. The increase in cytotoxicity associated with fetal demise described The influx of a large number of potentially cytotoxic NK cells into in this study indicates a novel function for uNK cells in the inflam- by guest on September 27, 2021 the uterus during pregnancy seems counterintuitive. In this regard, mation-induced termination of pregnancy. During normal pregnancy, two major functions of uNK cells have been suggested in the context these cells may contribute to gestational success by contributing to the of pregnancy. Mice deficient in NK cells display defects in decidual IL-10 pool and aiding in the remodeling of spiral arteries, whereas in spiral artery remodeling (32, 47), which appears to be due to the lack response to inflammatory insults, they may contribute to the fetal de- of uNK cell-derived IFN-␥ production (48). However, NK cell-defi- mise or premature termination of pregnancy (our manuscript in prep- cient mice show a normal pregnancy outcome. Another proposed aration). Furthermore, we demonstrate that IL-10 plays a positive role uNK cell function is regulation of trophoblast invasion into the de- in the protection of pregnancy from these effects. IL-10 may play a cidua. Recent evidence supporting this hypothesis demonstrates a re- dual role of protecting trophoblast cells against apoptosis (our unpub- ciprocal relationship between location of uNK cells and trophoblasts lished observations) as well as controlling the effects of TNF-␣ and during gestation (13). IFN-␥ is thought to play a major role in this activated uNK cells. In humans, NK cell activity is increased in some process, because IFN-␥ treatment inhibits trophoblast migration in women experiencing recurrent spontaneous abortion compared with placental explants. Although these studies suggest a temporal role for that during normal pregnancy (52). Our findings suggest that an as- IFN-␥ in reproduction, exogenous IFN-␥ administration at high doses sociation among inflammation, poor IL-10 production, and cytotoxic is known to result in fetal resorption in mice (49). However, IFN-␥ uNK cell activation may explain the etiology of unexplained adverse does not appear to play a significant role in the fetal resorption ob- pregnancy outcomes. served in the IL-10Ϫ/Ϫ mouse model of LPS-induced pregnancy failure. Interestingly, NK cell-deficient mice are resistant to fetal loss due Acknowledgments ␣ to systemic up-regulation of TNF- and reproductive endocrine dys- We thank Drs. James Padbury and Sunil Shaw for critical reading of this function mediated by anti-CD40 Ab administration during early manuscript. We also thank the Lifespan Animal Care Facility for assistance stages of pregnancy (50). In this regard, our data (Fig. 8 and Table II) with husbandry of our mouse colonies. clearly suggest that TNF-␣ is a key cytokine leading to uNK cell- mediated fetal resorption. Consistent with this are our observations of Disclosures ␣ ␥ ϩ an increase in TNF- -producing, but not IFN- -producing, CD45 The authors have no financial conflict of interest. UMC from LPS-treated pregnant IL-10Ϫ/Ϫ mice (our unpublished observations). Macrophages are recruited at the maternal-fetal inter- ␣ References face and are major producers of TNF- (51). As such, they would 1. Medawar, P. B. 1953. Some immunological and endocrinological problems appear to be prime candidate cells responsible for the deleterious pro- raised by evolution of viviparity in vertebrates. Symp. Soc. Exp. Biol. 7: 320–338. duction of TNF-␣ in response to LPS. However, because NK cell- 2. Mellor, A. L., and D. H. Munn. 2000. Immunology at the maternal-fetal interface: lessons for T cell tolerance and suppression. Annu. Rev. Immunol. 18: 367–391. depleted mice retain uterine macrophages (Fig. 7A, bottom) yet do not 3. Norwitz, E. R., D. J. Schust, and S. J. Fisher. 2001. Implantation and the survival experience fetal resorption in response to LPS, it is unlikely that mac- of early pregnancy. N. Engl. J. Med. 345: 1400–1408. 4090 uNK CELL-MEDIATED FETAL LOSS IN IL-10Ϫ/Ϫ MICE

4. Matovina, M., K. Husnjak, S. Milutin, S. Ciglar and M. Grce. 2004. Possible role 30. Berg, D. J., R. Kuhn, K. Rajewsky, W., Muller, S. Menon, N. Davidson, of bacterial and viral infections in miscarriages. Fertil. Steril. 81: 662–669. G. Grunig, and D. Rennick. 1995. Interleukin-10 is a central regulator of the 5. Romero, R., T. Chaiworapongsa, H. Kuivaniemi, and G. Tromp. 2004. Bacterial response to LPS in murine models of endotoxic shock and the Shwartzman re- vaginosis, the inflammatory response and the risk of preterm birth: a role for action but not endotoxin tolerance. J. Clin. Invest. 96: 2339–2347. genetic epidemiology in the prevention of preterm birth. Am. J. Obstet. Gynecol. 31. Fast, L. D. 1990. Identification of a single host non-H-2 gene regulating graft- 190: 1509–1519. versus-host disease response. J. Immunol. 144: 4177–4182. 6. Raghupathy, R. 1997. Th1-type immunity is incompatible with successful preg- 32. Croy, B. A., A. A. Ashkar, R. A. Foster, J. P. DiSanto, J. Magram, D. Carson, nancy. Immunol. Today 18: 478–482. S. J. Gendler, M. J. Grusby, N. Wagner, W. Muller, et al. 1997. Histological 7. Sacks, G., I. Sargent, and C. Redman. 1999. An innate view of human pregnancy. studies of gene-ablated mice support important functional roles for natural killer Immunol. Today 20: 114–118. cells in the uterus during pregnancy. J. Reprod. Immunol. 35: 111–113. 8. Gill, T. J., III. 1986. Immunological and genetic factors influencing pregnancy 33. Goodier, M. R., and M. Londei. 2000. Lipopolysaccharide stimulates the prolif- and development. Am. J. Reprod. Immunol. Microbiol. 10: 116–120. eration of human CD56ϩCD3Ϫ NK cells: a regulatory role of monocytes and 9. Mellor, A. L., J. Sivakumar, P. Chandler, K. Smith, H. Molina, D. Mao, and IL-10. J. Immunol. 165: 139–147. D. H. Munn. 2001. Prevention of T cell-driven complement activation and in- 34. Vujanovic, N. L. 2001. Role of TNF family ligands in antitumor activity of flammation by tryptophan catabolism during pregnancy. Nat. Immunol. 2: 64–68. natural killer cells. Int. Rev. Immunol. 20: 415–437. 10. Aluvihare, V. R., M. Kallikourdis, and A. G. Betz. 2004. Regulatory T cells 35. Miranda, D., J. Puente, L. Blanco, M. E. Wolf, and A. D. Mosnaim. 1998. In vitro mediate maternal tolerance to the fetus. Nat. Immunol. 5: 266–271. effect of bacterial lipopolysaccharide on the cytotoxicity of human natural killer 11. Moffett-King, A. 2002. Natural killer cells and pregnancy. Nat. Rev. Immunol. 2: cells. Res. Commun. Mol. Pathol. Pharmacol. 100: 3–14. 656–663. 36. Kim, S., K. Iizuka, H. L. Aguila, I. L. Weissman, and W. M. Yokoyama. 2000. 12. Bulmer, J. N., and G. E. Lash. 2005. Human uterine natural killer cells: a reap- In vivo natural killer cell activities revealed by natural killer cell-deficient mice. praisal. Mol. Immunol. 42: 511–521. Proc. Natl. Acad. Sci. USA 97: 2731–2736. 13. Ain, R., L. N. Canham, and M. J. Soares. 2003. Gestation stage-dependent intrauterine trophoblast cell invasion in the rat and mouse: novel endocrine phe- 37. Abrahams, V. M., P. Bole-Aldo, Y. M. Kim, S. L. Straszewski-Chavez, notype and regulation. Dev. Biol. 260: 176–190. T. Chaiworapongsa, R. Romero, and G. Mor. 2004. Divergent trophoblast re- 14. Ashkar, A. A., and B. A. Croy. 2001. Functions of uterine natural killer cells are sponses to bacterial products mediated by TLRs. J. Immunol. 173: 4286–4296. mediated by interferon ␥ production during murine pregnancy. Semin. Immunol. 38. D’Andrea, A., M. Rengaraju, N. M. Valiante, J. Chehimi, M. Kubin, M. Aste, Downloaded from 13: 235–241. S. H. Chan, M. Kobayashi, D. Young, E. Nickbarg, et al. 1992. Production of 15. Barber, E. M., and J. W. Pollard. 2003. The uterine NK cell population requires natural killer cell stimulatory factor (interleukin 12) by peripheral blood mono- IL-15 but these cells are not required for pregnancy nor the resolution of a Lis- nuclear cells. J. Exp. Med. 176: 1387–1398. teria monocytogenes infection. J. Immunol. 171: 37–46. 39. Okamura, H., H. Tsutsui, T. Komatsu, M. Yutsudo, A. Hakura, T. Tanimoto, 16. Ashkar, A. A., and B. A. Croy. 2001. Functions of uterine natural killer cells are K. Torigoe, T. Okura, Y. Nukada, K. Hattori, et al. 1995. Cloning of a new mediated by interferon ␥ production during murine pregnancy. Semin. Immunol. cytokine that induces IFN-␥ production by T cells. Nature 378: 88–91. 3: 235–241. 40. Kopydlowski, K. M., C. A. Salkowski, M. J. Cody, N. van Rooijen, J. Major,

17. Hiby, S. E., J. J. Walker, K. M. O’shaughnessy, C. W. Redman, M. Carrington, T. A. Hamilton, and S. N. Vogel,. 1999. Regulation of macrophage chemokine ex- http://www.jimmunol.org/ J. Trowsdale, and A. Moffett. 2004. Combinations of maternal KIR and fetal pression by lipopolysaccharide in vitro and in vivo. J. Immunol. 163: 1537–1544. HLA-C genes influence the risk of preeclampsia and reproductive success. J. Exp. 41. Sherry, B., M. Espinoza, K. R. Manogue, and A. Cerami. 1998. Induction of the Med. 200: 957–965. chemokine ␤ peptides, MIP-1␣ and MIP-1␤, by lipopolysaccharide is differen- 18. Parham, P. 2004. NK cells and trophoblasts: partners in pregnancy. J. Exp. Med. tially regulated by immunomodulatory cytokines ␥-IFN, IL-10, IL-4, and TGF-␤. 200: 951–955. Mol. Med. 4: 648–657. 19. Parr, E. L., M. B. Parr, and J. D Young. 1987. Localization of a pore-forming 42. Biswas, R., S. Datta, J. D. Gupta, M. Novotny, J. Tebo, and T. A. Hamilton. 2003. protein (perforin) in granulated metrial gland cells. Biol. Reprod. 37: 1327–1335. Regulation of chemokine mRNA stability by lipopolysaccharide and IL-10. J. Im- 20. Gentron, R. L., and M. G. Baines. 1988. Infiltrating decidual natural killer cells munol. 170: 6202–6208. are associated with spontaneous abortion in mice. Cell. Immunol. 113: 261–267. 43. Tanaka, A., Y. Yamane, and H. Matsuda. 2001. Mast cell MMP-9 production 21. Clark, D. A., P. C. Arck, R. Jalali, F. S. Merali, J. Manuel, G. Chaouat, enhanced by bacterial lipopolysaccharide. J. Vet. Med. Sci. 63: 811–813. J. L. Underwood, and J. F. Mowbray. 1996. Psycho-neuro-cytokine/endocrine path- 44. Lai, W. C., M. Zhou, U. Shankavaram, G. Peng, and L. M. Wahl. 2003. Differ- ways in immunoregulation during pregnancy. Am. J. Reprod. Immunol. 35: 330–337. ential regulation of lipopolysaccharide-induced monocyte matrix metalloprotein- by guest on September 27, 2021 22. Makrigiannakis. A., E. Zoumakis, S. Kalantaridou, C. Coutifaris A. N. Margioris, ase (MMP)-1 and MMP-9 by p38 and extracellular signal-regulated kinase 1/2 G. Coukos, K. C. Rice, A. Gravanis, G. P. Chrousos. A. 2001. Corticotropin- mitogen-activated protein kinases. J. Immunol. 170: 6244–6249. releasing hormone promotes blastocyst implantation and early maternal toler- 45. Lacraz, S., L. P. Nicod, R. Chicheportiche, H. G. Welgus, and J. M. Dayer. 1995. ance. Nat. Immunol. 2: 1018–1024. IL-10 inhibits metalloproteinase and stimulates TIMP-1 production in human 23. Moore, K. W., R. de Waal Malefyt, R. L. Coffman, and A. O’Garra. 2001. In- mononuclear phagocytes. J. Clin. Invest. 96: 2304–2310. terleukin-10 and the interleukin-10 receptor. Annu. Rev. Immunol. 19: 683–765. 46. Roth, I., and S. J. Fisher. 1999. IL-10 is an autocrine inhibitor of human placental 24. Trautman, M. S., D. Collmer, S. S. Edwin, W. White, M. D. Mitchell, and cytotrophoblast MMP-9 production and invasion. Dev. Biol. 205: 194–204. D. J. Dudley. 1997. Expression of interleukin-10 in human gestational tissues. 47. Ashkar, A. A., and B. A. Croy. 1999. Interferon-␥ contributes to the normalcy of J. Soc. Gynecol. Invest. 4: 247–253. murine pregnancy. Biol. Reprod. 61: 493–502. 25. Bennett, W. A., S. Lagoo-Deenadayalan, N. S. Whitworth, M. N. Brackin, E. Hale, and B. D. Cowan. 1997. Expression and production of interleukin-10 by 48. Guimond, M. J., B. Wang, and B. A. Croy,. 1998. Engraftment of bone marrow from human trophoblast: relationship to pregnancy immunotolerance. Early Pregnancy severe combined immunodeficient (SCID) mice reverses the reproductive deficits in 3: 190–198. natural killer cell-deficient tg epsilon 26 mice. J. Exp. Med. 187: 217–223. 26. Hanna, N., I. Hanna, M. Hleb, E. Wagner, J. Dougherty, D. Balkundi, J. Padbury, 49. Vassiliadis, S., D. Tsoukatos, and I. Athanassakis. 1994. Interferon-induced class II and S. Sharma. 2000. Gestational age-dependent expression of IL-10 and its expression at the spongiotrophoblastic zone of the murine placenta is linked to fetal receptor in human placental tissues and isolated cytotrophoblasts. J. Immunol. rejection and developmental abnormalities. Acta Physiol. Scand. 151: 485–495. 164: 5721–5728. 50. Erlebacher, A., D. Zhang, A. F. Parlow, and L. H. Glimcher. 2004. Ovarian 27. Plevyak, M., N. Hanna, S. Mayer, S. Murphy, H. Pinar, L. Fast, C. Ekerfelt, insufficiency and early pregnancy loss induced by activation of the innate im- J. Ernerudh, G. Berg, L. Matthiesen, et al. 2002. Deficiency of decidual IL-10 in mune system. J. Clin. Invest. 114: 39–48. first trimester missed abortion: a lack of correlation with the decidual immune cell 51. Vince, G., S. Shorter, P. Starkey, J. Humphreys, L. Clover, T. Wilkins, I. Sargent, and profile. Am. J. Reprod. Immunol. 47: 242–250. C. Redman. 1992. Localization of tumour necrosis factor production in cells at the 28. White, C. A., M. Johansson, C. T. Roberts, A. J. Ramsay, and S. A. Robertson. materno/fetal interface in human pregnancy. Clin. Exp. Immunol. 88: 174–180. 2004. Effect of interleukin-10 null mutation on maternal immune response and 52. Yamada, H., E. H. Kato, G. Kobashi, Y. Ebina, S. Shimada, M. Morikawa, reproductive outcome in mice. Biol. Reprod. 70: 123–131. N. Sakuragi, and S. Fujimoto. 2001. High NK cell activity in early pregnancy 29. Kuhn, R., J. Lohler, D. Rennick, K. Rajewsky and W. Muller. 1993. Interleukin- correlates with subsequent abortion with normal chromosomes in women with 10-deficient mice develop chronic enterocolitis. Cell 75: 263–274. recurrent abortion. Am. J. Reprod. Immunol. 46: 132–136.