All subjects understood and signed the informed consent form prior to participation. Experimental protocols were approved by the Ethics Committee of the Dongfang Hospital Human Ethics Committee, Beijing, China (no. 2011090201).

Decidual tissues were obtained from ten healthy females undergoing an elective termination of a normal pregnancy at between seven and eight weeks of gestation, as determined by the last menstrual period.

Endometrial tissues were collected from biopsies taken during the proliferative phase of the menstrual cycle of females undergoing laparoscopy for benign disease (Dongfang Hospital of Beijing University of Chinese Medicine, Beijing, China). The exclusion criteria were hormonal stimulation, cancerous lesions and irregular menstrual bleeding. There were six volunteers and two endometrial biopsies per volunteer were obtained. Samples from three of the females (six biopsies) were used for the for microarray experiments and samples from the remaining three females (six biopsies) were used for the reverse transcription-quantitative polymerase chain reaction and enzyme-linked immunosorbent assay experiments. Of the two samples taken from each patient, one sample was used as a control and the other was used in the experimental group.

uNK cells were purified as previously described (2). Briefly, decidual tissues were thoroughly washed with Ca²⁺- and Mg²⁺-free Hank’s balanced salt solution (HBSS) containing 100 U/ml penicillin and 100 g/ml streptomycin (Sigma-Aldrich, St. Louis, MO, USA), cut into fragments of 1–2 mm³ using two scalpels and digested for 1 h at 37°C with gentle agitation in HBSS with 0.1% (w/v) collagenase I (Gibco-BRL, Carlsbad, CA, USA). Cell suspensions were layered over Ficoll-Hypaque medium (General Electric, Fairfield, CT, USA) and centrifuged at 800 × g for 25 min. Cells at the interface were washed twice in RPMI-1640 media with 10% fetal calf serum (FCS) and antibiotics. Following incubation for 20 min at 4°C with anti-CD56 micro beads (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), cells were washed in washing buffer [phosphate-buffered saline (PBS), EDTA 2 mM and 0.5% bovine serum albumin(w/v)] and loaded onto a manual cell separation (MS) column in a MiniMACS magnet (MiniMACS^TM Separator System; Miltenyi Biotec GmbH). The MS column was flushed three times and CD56⁺ cells were flushed according to the manufacturer’s instructions. The purity of the uNK cells was >90% CD56⁺CD3⁻ according to flow cytometric analysis. The uNK cells were cultured in RPMI-1640 media with 1% FCS and 10 ng/ml IL-15 (R&D Systems Inc., Minneapolis, MN, USA).

uNK cell-secretion medium was prepared using 200 μl RPMI-1640 media with 1% FCS and IL-15 (10 ng/ml) containing 5×10⁵ of the purified uNK cells and placed into the upper chamber of a 0.4-μm pore hanging cell culture insert (EMD Millipore, Billerica, MA, USA) in a 24-well tissue plate, with 1,300 μl of the same media excluding cells in the lower chamber. Germeyer et al (9) showed that soluble factors from uterine leucocytes had significant effects on endometrial cell gene expression. Thus, a hanging cell culture insert was used so that soluble molecules from the uNK cells were able pass through the filter into the lower chamber, without cells being in direct contact. The control medium comprised 1,500 μl RPMI-1640 media with 1% FCS and 10 ng/ml IL-15. This was used for subsequent experiments. Following incubation for 24 h at 37°C, the uNK cell-secretion medium from the lower chamber and the control media were collected. To reduce interassay variability, the media from several batches was pooled for subsequent experiments and frozen at −80°C. Cells in the upper chamber were collected and the cell viability was measured using a live/dead viability kit (Invitrogen Life Technologies, Carlsbad, CA, USA). Only uNK cell samples containing <35% of dead cells following overnight incubation were used for subsequent experiments.

Human endometrial tissue was dissociated into single cells using 0.1% (w/v) collagenase I (Life Technologies, Carlsbad, CA, USA) for 50–60 min at 37°C. Cell suspensions were filtered using a 40-μm sieve to separate undigested myometrial tissue and debris. Further dissociation of the filtrate was prevented by Dulbecco’s modified Eagle’s medium (DMEM)/F-12 (no Phenol Red; Gibco-BRL) with 10% FBS (Gibco-BRL). To remove erythrocytes, the cells were resuspended in 4 ml DMEM/F12 with 1% FCS, layered over Ficoll-Paque PLUS (General Electric) and centrifuged for 25 min at 800 × g. Endometrial cells were removed from the Ficoll-Paque PLUS medium interface, washed three times and resuspended in 1 ml DMEM/F12 with 1% FCS. Leukocytes were removed with CD45-coated Dynabeads (Invitrogen Life Technologies). Purified stromal and epithelial cell suspensions were then obtained by a further round of magnetic bead sorting using Collection Epithelial Enrich Dynabeads (Invitrogen Life Technologies). Epithelial and stromal cell preparations were >95% pure.

Stromal cells were cultured in DMEM/F12 with 10% FBS and an antibiotic-antimycotic agent (100 U/ml penicillin, 100 g/ml streptomycin, 10 μg/ml gentamicin 0.25 μg/ml amphotericin B; Life Technologies). Epithelial cells were cultured in serum-free bronchial epithelial cell growth medium (final volumes: 2 ml bovine pituitary extract, 0.5 ml insulin, 0.5 ml HC, 0.5 ml GA-1000, 0.5 ml retinoic acid, 0.5 ml transferrin, 0.5 ml triiodothyronine, 0.5 ml epinephrine and 0.5 ml hEGF; Lonza, Walkersville, MD, USA) and an antibiotic-antimycotic agent. The isolated stromal and epithelial cells were separately seeded into six-well plates with 3 ml culture medium per well. Each well contained stromal or epithelial cells from a single patient. After two weeks, cells were passaged into 25 cm² cell culture flasks. Following this, stromal cells were passaged every 4–5 days and epithelial cells were passaged every 9–10 days.

On day 20 following endometrial stromal and epithelial cell generation, cells of each type were seeded onto a Nunc UpCell Surface membrane (Thermo Labsystems, Santa Rosa, CA, USA). The co-culture system was built on these temperature-responsive cell culture surfaces, according to the manufacturer’s instructions. In brief, when cells reached 80% confluence, all medium was aspirated and 500 μl fresh medium was added. The membrane was then placed on top of the stromal cell layer. The Nunc UpCell Surface was maintained at 20°C for 13 min. The membrane and cell layer were then carefully removed from the Nunc UpCell Surface using forceps. The membrane with the attached cell layer was transferred facing downwards onto the epithelial cell surface. Fresh medium was added and samples were incubated at 37°C for 40 min. A further 1 ml of medium was added to the top of the membrane and the membrane was withdrawn from the cell layer. The ratio of stromal to epithelial cells was 1:1, and every co-culture system was built using stromal and epithelial cells from the same participant. Co-cultured cells were maintained in DMEM/F12 with 1% FCS.

Each group, control and uNK cell, contained six co-culture systems. All the groups were washed twice with PBS and placed in serum-free DMEM for 16 h prior to subsequent experiments. DMEM was replaced by 80% uNK cell-secretion medium and 20% DMEM in the uNK cell group, whilst the control group was treated with 80% control medium and 20% DMEM. Following incubation for 6 h, the cells and media from each group were collected. Three pairs of co-culture systems were used for the microarray studies and three pairs for the RT-qPCR experiments.

Total RNA was extracted from the endometrial cells in each co-culture system. RNA was purified with the RNeasy Mini kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. The microarray analysis was performed using the GeneChip^® 3′ IVT Express kit (Affymetrix Inc., Santa Clara, CA, USA). Briefly, total RNA underwent reverse transcription, first strand cDNA synthesis, double strand DNA, in vitro transcription, cRNA synthesis and fragmentation (12). Samples were hybridized onto GeneChip PrimeView Human Gene Expression Array (Affymetrix Inc.). This array covers >36,000 transcripts and variants. Following 16 h hybridization at 45°C, arrays were washed on Fluidics Station 450 (Affymetrix, Inc.) and were scanned with Scanner 3000 (Affymetrix, Inc.) in order to obtain quantitative gene expression levels. The control and uNK cell groups were processed simultaneously throughout. Three chips were analyzed for each group.

A total of four differentially expressed genes were selected for validation of the results from the microarray experiments using RT-qPCR. Cells from the control and uNK cell groups were washed twice with PBS, and total RNA was extracted using TRIzol (Life Technologies). Reverse transcription was performed with 8 μl of total RNA per 20 μl reaction using a standard cDNA Synthesis kit (Takara Bio, Inc., Otsu, Japan). The RT-qPCR primer sequences for target genes were self-designed by this group and ordered from Invitrogen. Primer sequences for target genes are shown in Table I.

For each RT-qPCR experiment, the typical thermal cycling conditions included an initial activation step at 95°C for 5 min, 40 cycles at 95°C each for 30 sec, 56°C for 20 sec and 72°C for 30 sec. PCR reactions were performed on ABI Prism 7700 Sequence Detection system (Applied Biosystems Life Technologies, Foster City, CA, USA). cDNA concentration was normalized to that of GAPDH. The target mRNA expression was analyzed using the 2^−ΔΔCt algorithm.

IL-15 was analyzed using a commercially available ELISA kit (ELH-IL-15, RayBiotech, China) in the uNK cell-secretion medium prior to its use in the co-culture system experiments and in supernatants of the co-culture systems that had been treated with control media or with uNK cell-secretion medium. The analysis was conducted according to the manufacturer’s instructions. Assays were performed in triplicate and concentrations of IL-15 (pg/ml) were compared with standard curves. To determine the quantities of IL-15 secreted by the co-culture systems, the starting IL-15 content of the uNK cell-secretion medium was subtracted. The sensitivity of the kit was 10 pg/ml.

Data were analyzed using analysis of variance using SPSS 17.0 software (SPSS, Inc., Chicago, IL, USA). In the microarray experiments, median fold change ratios between the control and uNK cell groups were derived for each transcript, and genes that were up- or downregulated with a fold change >1 and P<0.001 were selected. In RT-qPCR and ELISA analysis, data are presented as the median ± standard error of the mean. P<0.01 was considered to indicate a statistically significant difference. Graphs of the data were produced using Microsoft Excel software.

Gene expression profiling using a microarray was used to compare transcript expression in the endometrial co-culture systems treated with either control medium or uNK cell-secretion medium.

This analysis identified 155 upregulated genes that exhibited a change of >2-fold in the median expression level in response to uNK cell-secretion medium. No transcript was downregulated >2-fold (Table II). However, certain genes in uNK cell groups were upregulated with a 1.0–2.0-fold change, compared with the control group. Previous studies have shown that these genes have an important functions in uNK cells. For instance, the co-culture system treated with uNK cell-secretion medium showed increased expression of interleukin (IL)15RA (1.6-fold) (13), vascular endothelial growth factor (VEGF)-C (1.35-fold) (14), intercellular adhesion molucule (ICAM) 1 (1.66-fold) (15), superoxide dismutase (SOD)2 (1.09-fold), caspase (CASP)1 (1.85-fold), nuclear factor erythroid 2-related factor 3 (NFE2L3) (1.37-fold), interferon γ receptor 1 (IFNGR1; 1.10-fold) (16), major histocompatibility complex (MHC) class I polypeptide-related sequence A (1.12-fold) and MHC class I polypeptide-related sequence B (1.34-fold) (3,17) and showed decreased expression of IGFBP3 (1.04-fold) (P<0.001).

In order to verify these changes, transcript levels for certain genes were measured by RT-qPCR, including chemokine (C-X-C) motif ligand (CXCL)10, CXCL11, IL-15 and sterile α motif domain containing 9-like (SAMD9L; Fig. 1). The RT-qPCR results confirmed the significant changes in expression that had been indicated by the microarray analysis. The endometrial co-culture system treated with uNK cell-secretion medium showed increased expression of CXCL10 (16.4-fold), CXCL11 (4.3-fold), IL-15 (3.1-fold) and SAMD9L (5.4-fold; P<0.01).

To confirm the observed changes in the IL-15 protein level, the quantity of IL-15 was analyzed by ELISA in the supernatant of the control and uNK cell-secretion medium-stimulated endometrial co-culture systems. There was a significant increase in IL-15 protein levels in the experimental groups compared with the control group (Fig. 2; P<0.01).