Trimester Extravillous Cytotrophoblast Decidual NK Cells Alter in Vitro First

Decidual NK Cells Alter In Vitro First Trimester Extravillous Cytotrophoblast Migration: A Role for IFN-γ

This information is current as Yuxiang Hu, Jan P. Dutz, Colin D. MacCalman, Paul Yong, of September 25, 2021. Rusung Tan and Peter von Dadelszen J Immunol 2006; 177:8522-8530; ; doi: 10.4049/jimmunol.177.12.8522 http://www.jimmunol.org/content/177/12/8522 Downloaded from

References This article cites 46 articles, 9 of which you can access for free at: http://www.jimmunol.org/content/177/12/8522.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 25, 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 © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Decidual NK Cells Alter In Vitro First Trimester Extravillous Cytotrophoblast Migration: A Role for IFN-␥1

Yuxiang Hu,*¶ Jan P. Dutz,†¶ Colin D. MacCalman,*¶ Paul Yong,†§¶ Rusung Tan,‡¶ and Peter von Dadelszen2*¶

Abnormal placentation results in either inadequate (consequences: recurrent miscarriage, intrauterine growth restriction, and preeclampsia) or overzealous (consequences: placenta accreta, increta, and percreta) placentation. NK cells dominate in ﬁrst trimester decidua and probably control extravillous cytotrophoblast (EVT) invasion. We examined this interaction in a novel way, using NK cells and villous explants from concordant ﬁrst trimester pregnancies cocultured using a new collagen (two-dimensional) model of placentation. Decidual NK (dNK) cells exerted contact-independent inhibition of normal cytotrophoblast migration, associated with changes in the cytotrophoblast expression of metalloproteases-2 and -9, and plasminogen activator inhibitor-1. dNK cells did not affect EVT proliferation and apoptosis, and cell column formation. dNK cell effects were partially reversed by Downloaded from neutralizing Abs against IFN-␥. We provide ex vivo human evidence of a direct role for dNK in modulating EVT differentiation as they form columns and then migrate from anchoring villi. The Journal of Immunology, 2006, 177: 8522–8530.

he successful and tightly regulated interaction between vasion (1, 6). Severe reduction of trophoblast cells, in combination the maternal immune system and fetal tissue is a prereq- with a strong increase of maternal lymphocytes and dNK cells, is T uisite for normal pregnancy outcomes. Insufficient acqui- found in intrauterine growth restriction (7). http://www.jimmunol.org/ escence by the maternal immune system may lead to the inade- Extravillous cytotrophoblast (EVT) express novel HLA-Ib pro- quate placentation of recurrent pregnancy loss, intrauterine growth teins, HLA-G, and HLA-E (8, 9). These molecules have special restriction, and preeclampsia (1). Overzealous fetal invasion may roles in protecting trophoblast cells from maternal immune cells, lead to morbid adherence of the placenta, placenta accreta, increta, especially dNK cells. In vitro studies have shown that the cytotoxic and percreta (2). reactivity of dNK cells against HLA-G expression target cells of 3 It has been postulated that maternal decidual NK (dNK) cells various origins was inhibited (10). HLA-E is a major ligand for the play an important role in the immune regulation of trophoblast NK inhibitory receptor CD94/NKG2A (11), which mediates inhi- invasion. There is considerable evidence to support that view (3). bition of killing of dNK cells (9). In the mouse, IL-15 was essential by guest on September 25, 2021 NK cells dominate during the early phase of gestation when pla- for the support of NK cell differentiation in the decidualizing cental trophoblast cells invade into decidua (4), and are found par- uterus (12). dNK maintained with IL-15 in vitro have no cytotox- ticularly around infiltrating trophoblast. This activity suggests a icity against trophoblast (13). This result suggests that, in normal role in regulating the degree of invasion and differentiation of the pregnancy, dNK are not cytotoxic against cytotrophoblast. trophoblast (5). Preeclampsia, and the related fetal growth restric- During implantation, cytotrophoblast leave the tips of anchoring tion, are interpreted as consequences of impaired trophoblast in- villi and become EVT. Initially, the EVT form cell columns, then invade and migrate from the columns into the maternal tissue, the ␣ ␤ † ‡ decidua. The EVT at the villus tip are 6 4 integrin positive, the *Department of Obstetrics and Gynaecology, Department of Medicine, Department ␣ ␤ of Pathology and Laboratory Medicine, and §Department of Medical Genetics, Uni- EVT in the columns are 5 1 integrin positive, and the invading versity of British Columbia, Vancouver, British Columbia, Canada; and ¶Child and ␣ ␤ EVT are 1 1 integrin positive (14, 15). Family Research Institute, Centre for Healthcare Innovation and Improvement, and Programs in Infection and Immunity and Reproductive Health, Vancouver, British A well-established experimental model of placentation involves Columbia, Canada the use of chorionic villi from first trimester terminations of preg- Received for publication March 15, 2006. Accepted for publication September nancy cultured as tissue explants on either collagen (two-dimen- 19. 2006. sional model) (16, 17) or Matrigel (three-dimensional model) (18). The costs of publication of this article were defrayed in part by the payment of page ␣ ␣ In the two-dimensional model, the integrin 1 and 5 changes of charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. invading EVT are not noted (16), whereas they are with the Ma- trigel model (18). 1 This study was funded by a Pilot Grant from the Infection and Immunity Program, Child and Family Research Institute. Y.H. is funded by grants from the Canadian IFN-␥, a cytokine secreted by both activated T lymphocytes and Institutes for Health Research, from whom P.v.D. receives salary support. C.D.M. and NK, is best known for its immunological functions. It has been P.v.D. receive salary support from the Child and Family Research Institute. P.v.D. and ␥ R.T. receive salary support from the Michael Smith Foundation for Health Research. used to treat viral infections (19, 20). However, the use of IFN- 2 Address correspondence and reprint requests to Dr. Peter von Dadelszen, Depart- in pregnancy might exert a deleterious effect, such as increased ment of Obstetrics and Gynaecology, University of British Columbia, 2H30-4500 miscarriage (21). IFN-␥ halts trophoblast migration in explant cul- Oak Street, Vancouver, British Columbia V6H 3N1, Canada. E-mail address: ture (22, 23), as does TNF-␣. Others have reported that the fol- [email protected] 3 lowing factors also contribute trophoblast migration inhibition: Abbreviations used in this paper: dNK, decidual NK; EVT, extravillous cytotro- ␤ ␤ phoblast; TIMP, tissue inhibitor of metalloprotease; MTS, 3-(4,5-dimethylthiazol-2- TGF- 1, TGF- 2, IL-10, and tissue inhibitor of metalloproteases yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium; PAI, plasmino- (TIMP)-1 (17, 22, 24–29). The role of dNK in controlling EVT gen activator inhibitor; CM, dNK-derived conditioned medium; HF, hollow fiber; IQR, interquartile range; MMP, matrix metalloprotease; uPA, urokinase-type protease differentiation and invasion has not been fully determined in a activator. human model of placentation.

In this study, we have tested the hypothesis that coculturing first PAI-1 (clone MA-56A7C10; University of Leuven, Leuven, Belgium) was trimester villus explants with dNK will alter in vitro EVT migra- also tested at 10 ␮g/ml, both individually and in combination. tion. Using a new model in which dNK and villus explants were cocultured, we show that dNK cells limit trophoblast migration but Assessment of cell columns, migration, and trophoblast not proliferation, apoptosis, and differentiation. This activity is differentiation mainly through the action of soluble mediators. Using Abs to neu- EVT outgrowth and migration occurs largely across the surface of the gel tralize IFN-␥ partially reversed this effect. as described previously (32, 33). Outgrowth was examined qualitatively taking into account the size, distance, morphology of the constituent cells, and their interrelationships, including cell-cell adhesion and sheet integrity, Materials and Methods although size varied greatly among tips (Nikon Eclipse TS100-F micro- Preparation and culture of dNK cells scope). Semiquantitative evaluation of cytotrophoblast outgrowth from explants was conducted by analyzing individual sites at villous tips. Images We obtained decidual tissue from healthy women undergoing elective were taken at different time points with a digital Nikon Coolpix5400 Cam- pregnancy termination of a normal pregnancy at 6- to 12-wk gestation, era. The distance of the cell outgrowth was measured using the arbitrary after informed consent. dNK cells were isolated as described previously scale provided by the aperture grid installed in microscope, and measured (26), with minor modifications. Specimens were extensively washed with from villous tips to the edge of the sheets. The distances were then cor- PBS, minced thoroughly into fragments of ϳ1mm3, and digested for 1 h rected to the mean distance in control villous tips for each placental sample. at 37°C in RPMI 1640 medium with 1.5 mg/ml collagenase and 2 mg/ml This process was done to reduce the effect of variation in growth potential hyaluronidase (Sigma-Aldrich). Cell suspensions were washed and plated between individual samples, providing observed/expected data for in dishes with 2 ng/ml IL-15 (R&D Systems), and, after an overnight analysis. incubation at 37°C, nonadherent cells were harvested from the culture For example, in sample 3, the migration distances for the four control dishes, leaving behind adherent stromal cells and macrophages. The cell tips were 35, 30, 30, and 35 U, the mean was 32.5 U. For the five hollow Downloaded from suspension obtained was then separated from dead cells and RBC by Fi- fiber (HF) contained dNK-exposed tips, the migration distance were 12, 20, coll-Hypaque gradient 1077 (Amersham Biosciences). NK cells were then 20, 13, and 20 U. Therefore, for those five tips, the distance was corrected enriched with human NK cell enrichment mixture following the manufac- and calculated to be 0.37, 0.62, 0.62, 0.4, and 0.62, respectively, using the turer’s instructions (StemCell Technologies). On average, 6 ϫ 105 en- control tip mean as unity. For sample 5, the control tip distances were 70, riched dNK per gram of decidual tissue were isolated. 50, 50, and 60 (mean 57.5) U, and the four HF dNK-exposed tip distances We performed flow cytometer analysis (FACScan flow cytometer; BD were 35 (0.61), 60 (1.04), 20 (0.35), and 45 (0.78) U. The magnitude

Biosciences) with enriched dNK cells prepared in this way. Aliquots of and direction of effect was similar between samples; using the observed/ http://www.jimmunol.org/ ϳ105 cells were incubated with fluorescence-labeled Abs (BD Bio- expected approach facilitated interpretation and analysis of the data. sciences) for 30 min on ice. In each sample 10,000 cells were evaluated. To Cell outgrowth patterns were examined by two individuals blinded to increase the purity of dNK cells, and to confirm highly purified dNK cells the experimental group, and described on an arbitrary scale of 1 (tightly have the same effect on trophoblast, allophycocyanin-conjugated anti- packed cells through the outgrowth especially at the edge of the column, no CD56 Ab-labeled NK cells were sorted by FACS. The purity of CD56ϩ gaps visible between EVT) to 3 (majority of cells radially oriented and cells reached 96.3% after sorting. migrated beyond the column, large gaps visible between EVT). The score Enriched dNK were cultured in 200 ␮l of DMEM/F12 supplemented 2 was assigned when cell outgrowth patterns were intermediate between 1 with 10% FBS (Invitrogen Life Technologies) and incubated for 3 days and 3. The data are presented as median (interquartile range; IQR). at 37°C with 5% CO2. IL-15 was added at 0–50 ng/ml concentrations. The purity of EVT was assessed by fluorescence microscopy to deter- dNK proliferation was measured by 3-(4,5-dimethylthiazol-2-yl)-5-(3- mine the level of fibroblast contamination of EVT columns and migrating carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) (Promega) cells. Explant cultures fixed by 70% methanol were double-stained with by guest on September 25, 2021 staining. rabbit anti-human cytokeratin (1/500 dilution; DakoCytomation) and mouse anti-human vimentin (1/250 dilution; DakoCytomation) for 1 h, Explant culture washed, and then incubated with Alexa Fluor 594 goat anti-rabbit IgG (HϩL) or Alexa Fluor 488 goat anti-mouse IgG (HϩL) (Invitrogen Life We prepared 30 first trimester placental explants as described previously Technologies). Stromal cells harvested from decidua were used as a pos- (30). All dNK cells and villi were from concordant pregnancies; dNK from itive control for vimentin Ab. different individuals were not pooled. Small pieces of tissue (2–3 mm) from the periphery of the placental debris were dissected (usually 4–8 Immunohistochemistry staining for integrins villus tips per sample). For preparation of rat-tail collagen I-coated plates (BD Biosciences), 0.8 ml of collagen I was mixed with 0.1 ml of 10ϫ Immunocytochemistry staining was used to assess whether NK cells af- ␮ DMEM (1:10) and 0.1 ml of 7.5% NaHCO4 (1:10). For each explant, 80 l fected integrin expression in EVT. Briefly, 70% methanol-fixed cultured ␣ of collagen was placed in the center of a 12-well culture dish. After for- explants were incubated with mouse anti-human integrin 1 or anti-integrin ␣ mation of gels (30 min at 37°C), the dissected tissue pieces were carefully 5 (Chemicon International), overnight at 4°C, and washed extensively put on the top of each gel drop, covered with 20 ␮l of medium, and in- with PBS. HRP-labeled secondary Abs (Bio-Rad) were added, and the ␣ cubated for 2–4 h to allow anchorage. Subsequently, explants were flooded plates were incubated for1hatroom temperature for integrin 5 stain- with 1 ml of DMEM/F12 supplemented with 10% FBS (DMEM/F12/FBS) ing; or biotinylated link universal secondary Ab and streptavidin-HRP ␣ and incubated overnight at 37°C with 5% CO2. After an initial period of (DakoCytomation) was used for integrin 1. After washing with PBS, 24 h, during which cytotrophoblast proliferation occurs at the villous tips 3-amino-9-ethylcarbazole substrate was applied. Images were taken at to produce anchoring cell columns, the explant cultures were exposed to a ϫ4 magnification. variety of experimental conditions, including Jurkat cells (a human T cell ␣ ␥ leukemia line; American Type Culture Collection), TNF- , and IFN- Total RNA preparation and first-strand cDNA synthesis (R&D Systems) (described in Results). The villi that did not show good attachment were discarded. The day when cultures were exposed to various Villi were carefully dissected from EVT. Total RNA was prepared from conditions was time 0. EVT after EVT were lysed and pooled from same assay groups according In preliminary experiments, we investigated the attachment of villus to the manufacturer’s protocol (RNeasy mini kit; Qiagen). For this type of explants to collagen-noncoated surfaces, and found that the explants did assay, EVT were always from the HF contained NK-treated group when not attach efficiently; EVT outgrowth rate was much lower (6 vs 50% with coculturing system was involved. cDNA synthesis was followed as de- collagen-coated surface). Also, EVT differentiation (including column for- scribed by the manufacturer (ThermoScript RT-PCR System; Invitrogen mation, then migration) was poor. This result is consistent with previous Life Technologies). findings (31). ␣ ␤ ␤ To assess potential involvement of TNF- , TGF- 1, TGF- 2, IL-10, Real-time quantitative PCR TIMP-1, plasminogen activator inhibitor (PAI)-1, and IFN-␥ secreted by dNK, explants were incubated with dNK in the presence of neutralizing The SYBR Green real-time quantitative PCR was performed to quantify Abs (R&D Systems) at 10 ␮g/ml (TIMP-1 was used at 2 ␮g/ml) individ- the mRNA expression using the ABI prism 7300 Sequence Detection Sys- ually, or in combination, added daily. dNK-derived conditioned medium tem (Applied Biosystems). After 2 min at 50°C and 10 min at 95°C, the (CM) were preincubated with 10 ␮g/ml (TIMP-1, 2 ␮g/ml) Abs for1hon samples were cycled 40 times at 95°C for 15 s and 60°C for 60 s. The ice, then exposed to explant cultures for a total of 96 h. Neutralizing Ab to relative quantification of gene expression was calculated using 18S gene 8524 dNK CELLS ALTER IN VITRO FIRST TRIMESTER EVT MIGRATION Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 1. Isolated dNK were Ͼ80% pure and IL-15-dependent in culture; EVT were Ͼ99% pure. OD (490 nm): OD for MTS staining at 490 nm. a, Flow cytometry: Ͼ94% decidual lymphocytes were CD45ϩ, of which 87.3% were dNK cells (CD56ϩ/CD16Ϫ). CD3ϩ and CD14ϩ cells were all Ͻ0.3%, and CD16ϩ cells were Ͻ3%. b, Enriched dNK cells cultured in 200 ␮l of DMEM/F12 supplemented with 10% FBS and incubated for 3 days at 37°C with Ն 5% CO2. IL-15 was added at 0–50 ng/ml. dNK proliferation was measured by MTS staining (OD at 490 nm). dNK viability maintained 2 ng/ml IL-15. c, More than 99% cell column (top panel) and migrating (bottom panel) EVT were cytokeratin positive (red) and vimentin negative (green). d, Isolated decidual stromal cells were cytokeratin negative and vimentin positive. Scale bars, 70 ␮m.

expression as endogenous control, as described previously (34), and pre- detected with IMUBIND Tissue PAI-1 ELISA Kit (American Diagnos- sented as fold changes in EVT in treated groups compared with the me- tica). The expression of MMP-2 and -9, uPA, and PAI-1 are presented as dium control group. The following proteins were evaluated: matrix metal- fold changes in the treated groups compared with the medium control loprotease (MMP)-2 and -9, urokinase-type protease activator (uPA), group. PAI-1, caspase-3, and E-cadherin. (Applied Biosystems). The primer se- quences were listed as described below. MMP-2 and -9, uPA, and PAI-1 The determination of EVT apoptosis by caspase-3 assay and were a gift from Dr. P. Leung’s (Vancouver, British Columbia, Canada) annexin-V staining, E-cadherin expression by flow cytometry, laboratory. and proliferation assay by MTS staining The primers used were as follows: MMP-2, (sense) CCGCAGTGACGG AAAGATGT and (antisense) CACTTGCGGTCGTCATCG-TA; MMP-9, Caspase-3 and annexin-V were used to determine whether EVT went ap- (sense) GGACGATGCCTGCAACGT and (antisense) CAAATACAG optosis after exposure to dNK or CM. Caspase-3 activity was tested by CTGG-TTCCCAATCT; uPA, (sense) CGCTTTCTTGCTGGTTGT and Caspase 3/CPP32 Colorimetric assay kit (BioVision). Fold increase in CPP (antisense) CCCAGTCTCTTCTTACAGCTGAT; PAI-1, (sense) CCG activity was determined by comparing EVT activity in HF dNK or CM- CCGCC-TCTTCCA and (antisense) GCCATCATGGGCACAGAGA; treated groups with the level in medium-treated control. Annexin V ex- caspase-3, (sense) GCCTACAGCCCATTTCTCCAT and (antisense) pression was measured by flow cytometry following the manufacturer’s GCGCCCTGGCAGCAT; and E-cadherin, (sense) AGGTGACAGAGCC suggestions (BD Pharmingen). In these two assays, EVT from the same TCTGGATAGA and (antisense) GGATGACACAGCGTGAGAGAAG. group were pooled after villi were dissected and then assayed. Staurospor- ine was added to EVT in medium control group for an additional4hatthe The measurement of protein activity and level in explant culture end of explant culture and then used as positive EVT apoptosis inducer. supernatants E-cadherin expression was used to determine whether dNK and CM changed E-cadherin expression in EVT. After removing villi, EVT from The supernatants of explant cultures were pooled from the same group and the same assay group were pooled and labeled with rabbit Ab against used to measure the protein activity or level. The total and active MMP-2 E-cadherin (Santa Cruz Biotechnology) followed by FITC-conjugated goat and -9 activity were measured using MMP-2 and -9 Biotrak Activity Assay anti-rabbit Ig. The E-cadherin expression was measured by flow cytometry. System (Amersham Biosciences); TIMP-1 was detected by TIMP-1, hu- MTS staining was applied to measure whether HF dNK or CM affected man, Biotrak ELISA System (Amersham Biosciences); uPA activity was the EVT proliferation. After removing the villi, collagen gels attached with tested by uPA Activity Assay Kit (Chemicon International); and PAI-1 was EVT were carefully transferred to 96-well plates individually. Medium The Journal of Immunology 8525 Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 2. dNK inhibit EVT migration through a contact-independent manner at 96 h of coculture. C, Cell column; ctrl, control medium; M, migrating EVT; T, villous explant tip. a, Control villus explant culture over 96 h revealed normal cell column formation over the ﬁrst 48 h, followed by migration of EVT at 72 and 96 h. b, EVT cell column formation and migration at 96 h varied between groups. For ctrl, IL-15, and Jurkat cell-exposed samples, cell column formation was followed by migration (as in a); however, for free dNK and HF dNK, and CM, TNF-␣, and IFN-␥ groups, the cell columns became packed with EVT, and migration was limited. For the TNF-␣ group, anti-TNF-␣ reversed the effect. For HF dNK and CM, anti-IFN-␥ permitted expansion of the cell column but did not reverse migration inhibition (see Table I for statistical analyses). For free dNK, the arrows are at the limit of EVT migration, the other cells being dNK ﬂoating in the medium. Scale bars, 250 ␮m.

(100 ␮l) was added back to each well containing the gel. Then, EVT were positive and vimentin negative (Fig. 1c); cytokeratin stains for exposed to 25 ␮l of MTS and incubated for 5 h. OD at 490 nm was trophoblast, an epithelial cell line, whereas vimentin stains for measured. cells of fibroblast origin. Isolated stromal cells from decidua were vimentin positive (Fig. 1d). Statistics Groups were compared using either parametric one-way ANOVA, with dNK inhibit in vitro first trimester EVT migration Student’s t tests as posttests, or nonparametric Kruskal-Wallis ANOVA, We examined the effect of dNK on EVT column formation and with Mann-Whitney U posttest, as appropriate, using GraphPad Prism 4.0 migration under the following experimental conditions: 1) control software. For posttests, significance was set at 0.01, to adjust for multiple comparisons, because Bonferroni corrections were deemed too strict. medium; 2) 10 ng/ml IL-15 in medium; and 3) Free dNK (enriched dNK with 10 ng/ml IL-15 suspended at 5 ϫ 104 Ϫ 6 ϫ 105 cells/ explant in culture medium). Results Upon seeding on collagen I (i.e., two-dimensional culture), we Purity and maintenance of dNK cell preparations, and purity of observed attachment and flattening at the distal parts of the tips and cell column and migrating villous outgrowth cells outgrowth of EVT in explant cultures, as described previously (32, We observed that Ͼ94% of decidual lymphocytes were CD45ϩ. 33). For explant cultures in medium or IL-15 control groups, EVT Of the CD45ϩ cells, 87.3% were dNK (CD56ϩ/CD16Ϫ), CD3ϩ formed columns and multilayered sheets of cells in 24–48 h, and and CD14ϩ cells were all Ͻ0.4%, and CD16ϩ cells were Ͻ2.1% migrated out in a radially oriented direction especially from the (BD Biosciences) (Fig. 1a). At IL-15 concentrations, Ͼ2 ng/ml column edge, usually in 48–96 h (Fig. 2 and Table I). In the pres- dNK viability in culture was well maintained for 72 h (Fig. 1b). ence of free dNK, proliferation and outgrowth occurred and in- For all experiments, 10 ng/ml IL-15 was used. We found that cells creased in extent over several days, but migration of EVT away forming cell columns and migrating on collagen were cytokeratin from column was not observed. For one experiment in which flow 8526 dNK CELLS ALTER IN VITRO FIRST TRIMESTER EVT MIGRATION

Table I. Decidual NK cells inhibit extravillous trophoblast migration, but not proliferation, in a contact-independent manner

Corrected Migration Migration/Cell Culture Conditions Distance Packing Scorea (p ,ء) p) (median; IQR) Bonferroni ,ء) number of tips examined) (median; IQR) MWu)

Control (229) 1.01 (0.8, 1.2) 3.0 (3, 3) Control ϩ IL-15 (51) 0.92 (0.66, 1.1) NS 3.0 (3, 3) NS (vs control) (vs control) Free dNK (59) 0.66 (0.42, 0.83) Ͻ0.0001 1.0 (1, 1) Ͻ0.0001 (vs control) (vs control) Free dNK ϩ anti-IFN-␥ (19) 0.85 (0.82, 0.98) 0.0047 1.0 (1, 1) NS (vs free dNK) (vs free dNK) Free dNK ϩ anti-TNF-␣ (7) 0.70 (0.37, 0.76) NS 1.0 (1, 1) NS (vs free dNK) (vs free dNK) Empty hollow ﬁbers (7) 0.98 (0.93, 1.1) NS 3.0 (3, 3) NS (vs control) (vs control) HF dNK (106) 0.58 (0.47, 0.70) Ͻ0.0001 1.0 (1, 1) Ͻ0.0001 (vs control) (vs control) HF dNK ϩ anti-IFN-␥ (16) 0.84 (0.61, 0.93) 0.0015 1.0 (1, 1) NS (vs HF dNK) (vs HF dNK) HF dNK ϩ anti-TNF-␣ (12) 0.58 (0.41, 0.63) NS 1.0 (1, 1) NS (vs HF dNK) (vs HF dNK) Downloaded from HF dNK ϩ anti-PAI-1 (22) 0.66 (0.48, 0.76) NS 1.0 (1, 1) NS (vs HF dNK) (vs HF dNK) Conditioned medium (137) 0.63 (0.46, 0.85) Ͻ0.0001 1.0 (1, 1) Ͻ0.0001 (vs control) (vs control) CM ϩ anti-IFN-␥ (49) 0.79 (0.52, 1.05) 0.0089 1.0 (1, 1) NS (vs CM) (vs CM) CM ϩ anti-TNF-␣ (53) 0.74 (0.62, 0.92) NS 1.0 (1, 1) NS http://www.jimmunol.org/ (vs CM) (vs CM) CM ϩ anti-PAI-1 (13) 0.73 (0.59, 0.82) NS 1.0 (1, 1) NS (vs CM) (vs CM) ϩ ␤ CM anti-TGF- 2 (11) 0.68 (0.56, 0.74) NS 1.0 (1, 1) NS (vs CM) (vs CM) CM ϩ anti-TIMP-1 (7) 0.51 (0.47, 0.57) NS 1.3 (0.5) NS (vs CM) (vs CM) IFN-␥ (81) 0.48 (0.36, 0.65) Ͻ0.0001 1.0 (1, 1) Ͻ0.0001 (vs control) (vs control) IFN-␥ ϩ anti-IFN-␥ (46) 0.86 (0.66, 1.08) Ͻ0.0001 3.0 (2.5, 3) Ͻ0.0001

(vs IFN-␥) (vs IFN-␥) by guest on September 25, 2021 TNF-␣ (24) 0.68 (0.40, 0.79) Ͻ0.0001 1.0 (1, 1) Ͻ0.0001 (vs control) (vs control) TNF-␣ ϩ anti-TNF-␣ (15) 1.03 (0.81, 1.24) 0.0033 2.5 (2, 3) Ͻ0.0001 (vs TNF-␣) (vs TNF-␣) Jurkat cells (6) 0.95 (0.63, 1.04) NS 3.0 (3, 3) NS (vs control) (vs control)

a Semiquantitative evaluation of cytotrophoblast outgrowth from explants was carried out by analyzing individual sites at villous tips. Corrected migration distance: farthest cell outgrowth from villous tips using aperture grid; corrected against mean migration distance for all control tips for each sample. Migration/cell packing score: arbitrary scale of 1 (tightly packed cells through the outgrowth especially at the edge of the column) to 3 (majority of cells radially oriented and migrated beyond the column). anti-, Neutralizing Ab; MWu, Mann-Whitney U test. .((p Ͻ 0.0001, Kruskal-Wallis; MWu tests performed (significance set at 0.01 (multiple comparisons ,ء cytometry-sorted dNK were used (purity 96.3%), the similar in- clinical drug efficacy tests in oncology (35); 2) HF dNK were hibitory effect on trophoblast migration was observed (data not cocultured with villous explants (enriched dNK cells (3 ϫ 105 shown). The outgrowths in these sheets were densely packed, es- cells/explant) with 10 ng/ml IL-15 suspended in culture medium); pecially at the anticipated limits of normal cell columns from con- and 3) The HFs were placed at the periphery of the culture plates, trol cultures, and outgrowth distance was limited (Fig. 2 and to avoid close proximity between dNK and EVT. However, empty Table I). HFs were also tested and did not influence EVT migration. The dNK-dependent migration inhibition was cell number de- The results showed that cell outgrowth and column formation pendent. When the concentration of dNK was Ն100,000 cells/ was not interrupted, but migration distance and pattern were explant, EVT migration was invariably limited; when the dNK affected similar to free dNK cells (Fig. 2 and Table I). Enriched concentration was Յ50,000 cells/explant, then EVT migration was dNK conditioned medium (enriched dNK cells were cultured not affected (data not shown). for 48 h at 0.5 ϫ 106 cells/ml with 10 ng/ml IL-15; conditioned Therefore, we focused on whether the observed effect was con- medium was added to explant culture at 1:1 ratio with medium) tact-dependent or contact-independent. caused similar migration inhibition as free or HF dNK (Fig. 2 and Table I). Instead of dNK, Jurkat cells, at 3 ϫ 105 cells/ dNK inhibit EVT migration in a contact-independent manner explant culture, were used to assess cell contact-dependent To address the question of whether dNK-dependent migration in- changes in EVT differentiation. Normal EVT migration was hibition was due to the contact inhibition, we undertook the fol- observed (Fig. 2 and Table I). The assay indicated that inter- lowing studies: 1) dNK contained within sealed protein permeable, action between EVT and Jurkat cells did not reveal contact but cell impermeable, HFs (MWCO 500,000, used mainly in pre- inhibition. The Journal of Immunology 8527

FIGURE 3. dNK affect EVT protease gene expression and protein levels. a,HF NK and CM reduced mRNA for total and active forms of MMP-2. b, Although HF dNK and CM increased mRNA for and total MMP-9, the active form was reduced. c and Downloaded from d, HF dNK and CM increased mRNA for and protein concentrations of uPA and PAI-1, respectively. Data are presented as fold increase or decrease compared with mean control tip mRNA or protein expression, as relevant, per pregnancy sample. For all graphs, ANOVA p Ͻ 0.01; Student’s t http://www.jimmunol.org/ -p Ͻ 0.001, com ,ءءء p Ͻ 0.01 and ,ءء test pared with control (ctrl). by guest on September 25, 2021

␤ Therefore, we investigated whether cytokines could be respon- TGF- 2, IL-10, PAI-1, and TIMP-1 did not reverse migration in- sible for the observed effects of dNK on EVT. hibition (Table I). Because the PAI-1 level was increased after exposure to dNK (Fig. 3), neutralizing Ab to PAI-1 was also Assessment of the potential involvement of the soluble mediators tested, and no reverse effect was observed (Table I). secreted by dNK Anti-IFN-␥ was associated with reversal of the farthest distance dNK appeared to inhibit trophoblast migration through soluble me- of EVT spread from villous tips (Table I). There was no discern- diators. TNF-␣ has been reported to restrict EVT migration (17). ible effect of anti-IFN-␥ on migration inhibition/cell packing As a positive control, we determined whether or not TNF-␣ influ- scores (Table I). Therefore, we tested the influence of IFN-␥ on enced the alterations in EVT migration noted above. Briefly, 4 EVT. Briefly, 5 ng/ml recombinant human IFN-␥ was added into ng/ml TNF-␣ was added into explant cultures causing the migra- explant cultures causing migration inhibition; neutralizing Ab to tion inhibition; neutralizing Ab to TNF-␣ reversed the effect (Fig. IFN-␥ reversed the effect (Fig. 2 and Table I). Recombinant human 2 and Table I). dNK produced 21.5 (Ϯ13.5 SD) pg/ml TNF-␣ in IFN-␥ showed this effect when the dose was as low as 200 pg/ml. coculture medium after 96 h, as measured by ELISA. However, dNK produced 207 (Ϯ122 SD) pg/ml IFN-␥ in coculture medium dNK-dependent migration inhibition could not be reversed by a after 96 h, as measured by ELISA. neutralizing anti-TNF-␣ Ab (Table I). By MTS assay, HF dNK and CM did not alter EVT cell pro- ␥ ␤ ␤ dNK also produce IFN- , TGF- 1, TGF- 2, IL-10, and TIMP-1 liferation compared with controls (median (IQR): 0.39 (0.29, (23–26, 36), and these cytokines may either reduce MMP expres- 0.358; n ϭ 10), 0.44 (0.25, 0.78; n ϭ 10), and 0.49 (0.35, 0.59), sion (27), MMP autoconversion into active forms (37), or inhibit respectively), whereas IFN-␥ (5 ng/ml) did (0.22 (0.15, 0.35; n ϭ EVT migration and invasion (28, 29). We found that IL-15-main- 13), Mann-Whitney U test, p ϭ 0.0115 (vs control)). The effect of tained dNK released TIMP-1 into conditioned medium (n ϭ 10; IFN-␥ on proliferation was reversed by neutralizing anti-IFN-␥ Ab Ϯ ␤ ϭ 17.9 11.6 (SD) ng/ml). However, neutralizing Abs to TGF- 1, (0.40 (0.29, 0.51; n 6)). 8528 dNK CELLS ALTER IN VITRO FIRST TRIMESTER EVT MIGRATION

FIGURE 4. E-cadherin expression by EVT was up-regulated following coculture with dNK, but caspase-3 expression was down-regulated. Integrin expression was unaltered. a, EVT E-cadherin gene and protein expression were increased following exposure to HF dNK and ␣ ␣ CM. b, 1(top row) and 5 (bottom row) integrins were expressed by both cell column and migrating EVT, and were not affected by exposure to either free dNK or CM. c, EVT caspase-3 gene expression was de-

creased by exposure to CM, whereas Downloaded from EVT caspase 3 protein concentration and annexin V surface expression were unaltered by either HF dNK or CM, but was increased by staurosporine (positive control). ANOVA p Ͻ 0.001 for all compar-

/p Ͻ 0.01 http://www.jimmunol.org ,ءء ,isons; Student’s t test p Ͻ 0.001, compared with ,ءءء and control. Scale bars, 250 ␮m. by guest on September 25, 2021 ␣ dNK and CM influenced MMP-2 and -9, uPA, and PAI-1 and 1-integrin expression was weakly positive in both column expression EVT and migrating EVT, in the medium and IL-15 control groups. ␣ We examined the effect of dNK exposure on EVT gene expression But 1 integrin was also strongly positive following the introduc- ␣ ␣ and protein levels of the following proteases: MMP-2 (total and tion of the biotin-streptavidin system. 1 and 5 integrin expres- active), MMP-9 (total and active), uPA, and PAI-1 (Fig. 3). These sion was not affected by exposure to either dNK or CM (Fig. 4b). proteases have been shown by us and others to be involved in EVT Because the trophoblast from dNK or CM-cocultured explants migration and differentiation. For MMP-2, mRNA was reduced still formed columns and grew bigger in extent over several days following exposure to HF dNK (reversed by anti-IFN-␥; data not (Fig. 2), and EVT proliferation from dNK or CM did not show shown), CM, and IFN-␥ (reversed by anti-IFN-␥); this was mir- inhibition compared with medium control by MTS staining, this rored by both total and active MMP-2 protein, for free dNK and indicated that dNK mainly inhibited trophoblast migration but not CM, but not HF dNK. Although mRNA message and total MMP-9 proliferation. Caspase-3 gene expression varied between groups, were increased following HF dNK and CM exposure, however, the being significantly reduced in EVT exposed to CM (Fig. 4c). active form was not increased or even decreased with exposure to Caspase-3 activity in EVT was unchanged after exposure to either free and HF dNK and CM. Both PAI-1 and uPA message RNA HF dNK or CM, but was significantly increased after staurosporine were increased following exposure to HF dNK and CM; however, treatment, a positive control for apoptosis induction. Annexin V HF dNK did not cause elevated levels of PAI-1, whereas free dNK expression was unaltered in EVT exposed to either HF dNK or and CM did. HF dNK caused elevated uPA expression. CM, but was increased by staurosporine (Fig. 4c). Because the cell columns in the three dNK groups (free, contained, and conditioned medium) appeared tightly packed with Discussion EVT, we investigated whether or not either EVT adhesion mole- We have introduced a new model of the critical interaction be- cule expression or apoptosis was influenced. tween dNK and first trimester trophoblast that closely simulates the in vivo state. We found that dNK limit migration of EVT, dNK and CM increased EVT E-cadherin expression, did not through changes in protease activity and E-cadherin expression, ␣ ␣ alter 1 and 5 integrin expression, and had no cytotoxicity and that the inhibition of EVT expansion was reversed by anti- against cytotrophoblast IFN-␥. Also, we have introduced the use of HFs into the villus E-cadherin gene and protein expression were increased in EVT explant model, showing that dNK-dependent migration inhibition exposed to both HF dNK and CM (Fig. 4a). Cytotrophoblast in is contact independent. ␣ ␣ explant cultures expressed 1 and 5 integrins in the collagen dNK limit trophoblast migration in a dose-dependent manner. model as reported previously (16). In the absence of the biotin- When dNK cells were Ն100,000 cells per explant culture, tropho- ␣ streptavidin system, 5 integrin expression was strongly positive blast migration was restricted, but no inhibition was observed The Journal of Immunology 8529 when dNK cells were Յ50,000 cells per explant culture. HF dNK This implies that part of the dNK effect on MMP-2 activation is and CM, but not Jurkat cells, had the same inhibitory effects on due to large solutes (molecular mass Ͼ500 kDa). Such solutes EVT migration as free dNK. This indicated that dNK-dependent might be complexed proteoglycans (e.g., serglycin or decorin). migration inhibition was contact independent and is mediated by The observed increase in uPA mRNA and protein probably rep- soluble factors produced by dNK. resents a compensatory response by EVT, because increased uPA We have determined that IL-15 is required to maintain dNK expression would be expected to increase EVT migration (44). cells in vitro, but does not affect EVT proliferation and migration. We have also confirmed the finding that EVT cultured on col- dNK activated by IL-15 do not have cytotoxicity against cytotro- lagen do not undergo the integrin expression changes noted both in phoblast in vitro (13). We have found that trophoblast cocultured vivo and during culture on Matrigel, which permits true invasion, with dNK or CM are able to proliferate. The cell column expanded rather than migration (16, 18). dNK had no discernible effects on ␣ ␣ over several days of culture, with increased cellular packing within 1 and 5 integrin expression in EVT in vitro. the columns, related to the observed up-regulation of E-cadherin In vivo, the interaction between dNK and EVT may be modu- gene expression. There was no evidence of apoptosis induction in lated by Ag-presenting dendritic cells, which are present in prox- EVT by dNK measured by caspase-3 and annexin V; indeed, CM imity to dNK in human first trimester decidua (45, 46), as well as seemed to reduce caspase-3 gene expression in EVT. However, other factors such as other cytokines, growth factors, and preg- because migration was limited, dNK alter EVT transition to a mi- nancy hormones. grating phenotype, but not proliferation, and without cytotoxicity. There are some limitations to this study. Primarily, as noted, the We have confirmed the findings of Bauer et al. (17) that TNF- model that we chose for these experiments was two-dimensional,

␣-treated EVT remain as highly packed cell columns. This impor- with explants being cultured on collagen, rather than three-dimen- Downloaded from tant cytokine, produced by dNK, may regulate at least some of the sional, using Matrigel; we anticipate being able to replicate the dNK effects on EVT differentiation. However, neutralizing Ab to ﬁndings with Matrigel. Second, the scale used to measure EVT TNF-␣ added to dNK coculture did not reverse the effect, indicat- migration was intrinsically arbitrary, and, therefore, prone to bias. ing that other soluble mediators must be involved. However, the assessment of EVT migration was blinded in terms Neutralizing the effect of IFN-␥ reversed the EVT expansion of experimental conditions and was averaged between two observ-

associated with dNK exposure. IFN-␥ is a major dNK cytokine ers; the findings presented reflect this approach. As we refine the http://www.jimmunol.org/ (23), being secreted spontaneously by dNK isolated from human model to the use of Matrigel, we will apply objective invasion decidua and at higher concentrations than peripheral blood NK assays. cells (38, 39). We found that IFN-␥-treated EVT remained as tight Despite the limitations, these data show ex vivo human evidence columns. However, unlike exposure to either HF dNK or CM, of a direct role for dNK in modulating EVT differentiation as they IFN-␥ reduced EVT proliferation, as measured by MTS assay. We form columns and then migrate from anchoring villi. The control believe that this due to the effect induced by dNK or CM results of this interaction, partially through IFN-␥, is important in the from multiple modulators, rather than a single inhibitory factor. In processes that lead to either normal placentation, or deficient pla- the mouse, dNK are recruited to the decidua, and dNK-derived centation and its sequelae (e.g., preeclampsia and intrauterine IFN-␥ plays a key role in the control of trophoblast migration and growth restriction), and morbid adherence of the placenta. by guest on September 25, 2021 spiral artery modification (40). However, the subjective assessment of migration and cell packing remained significantly different Acknowledgments between the anti-IFN-␥-treated group and controls, implying that We thank Drs. Bruce Verchere, Poul Sorenson, Nelly Auersperg, and Peter the change in measured distance may have modulated through ei- Leung for access to laboratory equipment and supplies. We also thank the ther a reduction in tight inter-EVT adhesion (imperceptible to the CARE Program, BC Women’s Hospital and Health Centre, for their as- sistance in gaining access to reproductive tissue, and Dr Isabella Caniggia light microscope-aided eye) or an increase in the dimensions of the for her helpful advice as we established the model. cell column due to EVT proliferation, without permitting change to the migrating phenotype. Disclosures ␤ dNK can produce other soluble molecules like TGF- 1, TGF- The authors have no financial conflict of interest. ␤ 2, and IL-10, all of which have been reported to limit trophoblast migration and invasion (24, 26–29, 41). However, in these studies, References neutralizing Abs to these molecules did not reverse the effect. dNK 1. Khong, T. Y., F. De Wolf, W. B. Robertson, and I. Brosens. 1986. Inadequate maternal vascular response to placentation in pregnancies complicated by pre- produce a variety of cytokines and other immunomodulatory pro- eclampsia and by small-for-gestational age infants. Br. J. Obstet. Gynaecol. 93: teins (23, 24, 42). The observed effect of dNK on EVT almost 1049–1059. certainly does not act through a single molecule, such as IFN-␥.A 2. Bulmer, J. N. 1992. Immune aspects of pathology of the placental bed contrib- uting to pregnancy pathology. Baillieres Clin. Obstet. Gynaecol. 6: 461–488. protein array investigation of the CM would facilitate the identi- 3. Parham, P. 2004. NK cells and trophoblasts: partners in pregnancy. J. Exp. Med. fication of other cytokines or immunomodulatory proteins to be 200: 951–955. investigated. 4. Starkey, P. M., L. L. Sargent, and C. W. G. Redman. 1988. Cell populations in human early pregnancy decidua: characterization and isolation of large granular Among other MMPs, MMP-9 is known to be self-activating, lymphocytes by flow cytometry. Immunology 65: 129–134. with the cleavage of the propeptide leading to the formation of an 5. King, A., and Y. W. Loke. 1991. On the nature and function of human uterine granular lymphocytes. Immunol. Today 12: 432–435. active enzyme (43). TIMP-1, the natural inhibitor of MMP-9 and 6. Meekins, J. W., R. Pijnenborg, M. Hanssens, I. R. McFadyen, and A. van Assche. MMP-2 (43), is secreted by dNK (25, 36), a finding that we have 1994. A study of placental bed spiral arteries and trophoblast invasion in normal confirmed. TIMP-1 can block the activation of pro-MMP-9 in skin and severe pre-eclamptic pregnancies. Br. J. Obstet. Gynaecol. 101: 669–674. 7. Stallmach, T., G. Hebisch, P. Orban, and X. Lu. 1999. Aberrant positioning of (37). Given our findings of decreased active MMP-9, despite in- trophoblast and lymphocytes in the feto-maternal interface with pre-eclampsia. creased MMP-9 mRNA and total MMP-9 protein, in association Virchows Arch. 434: 207–211. with dNK exposure, we support the proposed down-regulation of 8. King, A., S. E. Hiby, L. Gardner, S. Joseph, J. M. Bowen, S. Verma, T. D. Burrows, and Y. W. Loke. 2000. Recognition of trophoblast HLA class I mole- the proteolytic activation of MMP-9 in human tissue by TIMP-1 cules by decidual NK cell receptors: a review. Placenta 21(Suppl. A): S81–S85. (37). However, we could not reverse the phenotypic effect on EVT 9. King, A., D. S. Allan, M. Bowen, S. J. Powis, S. Joseph, S. Verma, S. E. Hiby, A. J. McMichael, Y. W. Loke, and V. M. Braud. 2000. HLA-E is expressed on migration with a blocking mAb to TIMP-1. Also, whereas free trophoblast and interacts with CD94/NKG2 receptors on decidual NK cells. Eur. dNK and CM reduced EVT-free MMP-2 protein, HF dNK did not. J. Immunol. 30: 1623–1631. 8530 dNK CELLS ALTER IN VITRO FIRST TRIMESTER EVT MIGRATION

10. Le Gal, F. A., B. Riteau, C. Sedlik, I. Khalil-Daher, C. Menier, J. Dausset, J. G. 29. Karmakar, S., and C. Das. 2002. Regulation of trophoblast invasion by IL-1␤ and Guillet, E. D. Carosella, and N. Rouas-Freiss. 1999. HLA-G-mediated inhibition of TGF-␤1. Am. J. Reprod. Immunol. 48: 210–219. antigen-specific cytotoxic T lymphocytes. Int. Immunol. 11: 1351–1356. 30. Vivovac, L., C. J. Jones, and J. D. Aplin. 1995. Trophoblast differentiation during 11. Lee, N., M. Llano, M. Carretero, A. Ishitani, F. Navarro, M. Lopez-Botet, and formation of anchoring villi in a model of the early human placenta in vitro. D. E. Geraghty. 1998. HLA-E is a major ligand for the natural killer inhibitory Placenta 16: 41–56. receptor CD94/NKG2A. Proc. Natl. Acad. Sci. USA 95: 5199–5204. 31. Aplin, J. D., T. Haigh, C. J. Jones, H. J. Church, and L. Vicovac. 1999. Devel- 12. Ashkar, A. A., G. P. Black, Q. Wei, H. He, L. Liang, J. R. Head, and B. A. Croy. opment of cytotrophoblast columns from explanted first-trimester human placen- ␤ 2003. Assessment of requirements for IL-15 and IFN regulatory factors in uterine tal villi: role of fibronectin and integrin alpha5 1. Biol. Reprod. 60: 828–838. NK cell differentiation and function during pregnancy. J. Immunol. 171: 32. Lacey, H., T. Haigh, M. Westwood, and J. D. Aplin. 2002. Mesenchymally- 2937–2944. derived insulin-like growth factor 1 provides a paracrine stimulus for trophoblast 13. Verma, S., S. E. Hiby, Y. W. Loke, and A. King. 2000. Human decidual natural migration. BMC Dev. Biol. 2: 5–12. killer cells express the receptor for and respond to the cytokine interleukin 15. 33. Leach, R. E., B. Kilburn, J. Wang, Z. Liu, R. Romero, and D. R. Armant. 2004. Biol. Reprod. 62: 959–968. Heparin-binding EGF-like growth factor regulates human extravillous cytotro- ␣ ␤ phoblast development during conversion to the invasive phenotype. Dev. Biol. 14. Aplin, J. D. 1993. Expression of integrin 6 4 in human trophoblast and its loss from extravillous cells. Placenta 14: 203–215. 266: 223–237. 15. Damsky, C. H., M. L. Fitzgerald, and S. J. Fisher. 1992. Distribution patterns of 34. Fink, L., W. Seeger, L. Ermert, J. Hanze, U. Stahl, F. Grimminger, W. Kummer, extracellular matrix components and adhesion receptors are intricately modulated and R. M. Bohle. 1998. Real-time quantitative RT-PCR after laser-assisted cell during first trimester cytotrophoblast differentiation along the invasive pathway, picking. Nat. Med. 4: 1329–1333. in vivo. J. Clin. Invest. 89: 210–222. 35. Hall, L. A., C. M. Krauthauser, R. S. Wexler, M. G. Hollingshead, A. M. Slee, 16. Aplin, J. D., T. Haigh, C. J. Jones, H. J. Church, and L. Vicovac. 1999. Devel- and J. S. Kerr. 2000. The hollow fiber assay: continued characterization with opment of cytotrophoblast columns from explanted first-trimester human placen- novel approaches. Anticancer Res. 20: 903–911. ␣ ␤ 36. Kim, M. H., R. P. Kitson, P. Albertsson, U. Nannmark, P. H. Basse, P. J. Kuppen, tal villi: role of fibronectin and integrin 5 1. Biol. Reprod. 60: 828–838. 17. Bauer, S., J. Pollheimer, J. Hartmann, P. Husslein, J. D. Aplin, and M. Knofler. M. E. Hokland, and R. H. Goldfarb. 2000. Secreted and membrane-associated 2004. Tumor necrosis factor-␣ inhibits trophoblast migration through elevation of matrix metalloproteinases of IL-2-activated NK cells and their inhibitors. J. Im- munol. 164: 5883–5889. plasminogen activator inhibitor-1 in first-trimester villous explant cultures. ␣ Downloaded from J. Clin. Endocrinol. Metab. 89: 812–822. 37. Han, Y. P., Y. D. Nien, and W. L. Garner. 2002. Tumor necrosis factor- -induced proteolytic activation of pro-matrix metalloproteinase-9 by human skin is controlled 18. Caniggia, I., C. V. Taylor, J. W. K. Ritchie, S. J. Lye, and M. Letarte. 1997. by down-regulating tissue inhibitor of metalloproteinase-1 and mediated by tissue- Endoglin regulates trophoblast differentiation along the invasive pathway in hu- associated chymotrypsin-like proteinase. J. Biol. Chem. 277: 27319–27327. man placental villous explants. Endocrinology 138: 4977–4988. 38. Tsuda, H., M. Sakai, T. Michimata, K. Tanebe, S. Hayakawa, and S. Saito. 2001. 19. Bocher, W. O., S. Herzog-Hauff, J. Schlaak, K. H. Meyer zum Buschenfeld, and Characterization of NKT cells in human peripheral blood and decidual lympho- H. F. Lohr. 1999. Kinetics of hepatitis B surface antigen-specific immune re- cytes. Am. J. Reprod. Immunol. 45: 295–302. sponses in acute and chronic hepatitis B or after HBs vaccination: stimulation of 39. Lidstrom, C., L. Matthiesen, G. Berg, S. Sharma, J. Ernerudh, and C. Ekerfelt. the in vitro antibody response by interferon ␥. Hepatology 29: 238–244.

2003. Cytokine secretion patterns of NK cells and macrophages in early human http://www.jimmunol.org/ 20. Dallalio, G., M. North, and R. T. Means, Jr. 1996. Inhibition of marrow CFU-E pregnancy decidua and blood: implications for suppressor macrophages in de- colony formation from human immunodeﬁciency virus-infected patients by ␤- cidua. Am. J. Reprod. Immunol. 40: 444–452. and ␥-interferon. Am. J. Hematol. 53: 118–120. 40. Croy, B. A., H. He, S. Esadeg, Q. Wei, D. McCartney, J. Zhang, A. Borzychowski, 21. Haddad, E. K., A. J. Duclos, E. Antecka, W. S. Lapp, and M. G. Baines. 1997. ␥ A. A. Ashkar, G. P. Black, S. S. Evans, et al. 2003. Uterine natural killer cells: Role of interferon- in the priming of decidual macrophages for nitric oxide insights into their cellular and molecular biology from mouse modelling. Reproduc- production and early pregnancy loss. Cell. Immunol. 181: 68–75. tion 126: 149–160. 22. Hunt, J. S., R. Atherton, and J. L. Pace. 1990. Differential responses of rat tro- 41. Lash, G. E., H. A. Otun, B. A. Innes, J. N. Bulmer, R. F. Searle, and S. C. Robson. phoblast cells and embryonic ﬁbroblasts to cytokines that regulate proliferation 2005. Inhibition of trophoblast cell invasion by TGFB1, 2, and 3 is associated and class I MHC antigen expression. J. Immunol. 145: 184–189. with a decrease in active proteases. Biol. Reprod. 74: 403–409. ␥ 23. Redline, R. W. 2000. Role of uterine natural killer cells and interferon in 42. Koopman, L. A., H. D. Kopcow, B. Rybalov, J. E. Boyson, J. S. Orange, F. Schatz, placental development. J. Exp. Med. 192: F1–F4. ϩ R. Masch, C. J. Lockwood, A. D. Schachter, P. J. Park, and J. L. Strominger. 2003.

24. Clark, D. A., G. Vince, K. C. Flanders, H. Hirte, and P. Starkey. 1994. CD56 Human decidual natural killer cells are a unique NK cell subset with immunomodu- by guest on September 25, 2021 lymphoid cells in human ﬁrst trimester pregnancy decidua as a source of novel latory potential. J. Exp. Med. 198: 1201–1212. transforming growth factor-␤2-related immunosuppressive factors. Hum. Reprod. 43. Rouy, D., I. Ernens, C. Jeanty, and D. R. Wagner. 2005. Plasma storage at Ϫ80 9: 2270–2277. degrees C does not protect matrix metalloproteinase-9 from degradation. Anal. 25. Albertsson, P., K. M. Kim, L. E. Jonges, R. P. Kitson, P. J. Kuppen, Biochem. 338: 294–298. B. R. Johansson, U. Nannmark, and R. H. Goldfarb. 2000. Matrix metallopro- 44. Liu, J., C. Chakraborty, C. H. Graham, Y. P. Barbin, S. J. Dixon, and P. K. Lala. teinases of human NK cells. In Vivo 14: 269–276. 2003. Noncatalytic domain of uPA stimulates human extravillous trophoblast 26. Vigano, P., B. Gaffuri, E. Somigliana, M. Infantino, M. Vignali, and A. M. migration by using phospholipase C, phosphatidylinositol 3-kinase and mitogen- Di Blasio. 2001. Interleukin-10 is produced by human uterine natural killer cells activated protein kinase. Exp. Cell Res. 286: 138–151. but does not affect their production of interferon-␥. Mol. Hum. Reprod. 7: 45. Kammerer, U., M. Schoppet, A. D. McLellan, M. Kapp, H. I. Huppertz, 971–977. E. Kampgen, and J. Dietl. 2000. Human decidua contains potent immunostimu- 27. Roth, I., and S. J. Fisher. 1999. IL-10 is an autocrine inhibitor of human placental latory CD83ϩ dendritic cells. Am. J. Pathol. 157: 159–169. cytotrophoblast MMP-9 production and invasion. Dev. Biol. 205: 194–204. 46. Kammerer, U., A. O. Eggert, M. Kapp, A. D. McLellan, T. B. Geijtenbeek, 28. Irving, J. A., and P. K. Lala. 1995. Functional role of cell surface integrins on J. Dietl, Y. van Kooyk, and E. Kampgen. 2003. Unique appearance of prolifer- human trophoblast cell migration: regulation by TGF-␤, IGF-II, and IGFBP-1. ating antigen-presenting cells expressing DC-SIGN (CD209) in the decidua of Exp. Cell Res. 217: 419–427. early human pregnancy. Am. J. Pathol. 162: 887–896.