Cell experiments
Somatic cell pellets were harvested by scraping. The hiPSCs were incubated at 37°C, in a solution containing 1 mg/ml collagenase IV (Invitrogen, Carlsbad, CA), 1 mM CaCl2, 20% KNOCKOUTTM Serum Replacement (KSR), and 10% ACCUMAX (Innovative Cell Technologies, Inc., San Diego, CA). When the edges of the colonies started to dissociate from the bottom of the dish, the collagenase solution was removed and the cells were washed with medium. Colonies were then picked up and collected.
MRC-5 and amniotic mesodermal (AM) cells were maintained in POWEREDBY 10 medium (MED Shiratori Co., Ltd., Tokyo, Japan). The human placental artery endothelial (PAE) cells were harvested from human placenta. To isolate the arterial endothelium, we used the explant culture method, in which the cells were outgrown from pieces of the placenta’s arterial vessels. Briefly, arterial vessels were separated from arteries in the chorionic plate, and chopped into approximately 5-mm3 pieces. The pieces were washed in endothelial basal medium-2 (EBM-2; Cambrex, Walkersville, MD) and cultured in EGM-2MV medium (Cambrex), which consisted of EBM-2, 5% fetal bovine serum (FBS), and the supplemental growth factors VEGF, bFGF, EGF, and IGF. The arterial vessels attached to the substrata of the culture dishes (BD Falcon; Becton Dickinson, San Jose, CA). Cells migrated out from the surface of the tissues after about 20 days of incubation at 37°C in 5% CO2. The cells were harvested in PBS containing 0.1% trypsin and 0.25 mM EDTA, and were re-seeded at a density of 3 × 105 cells in a 10-cm dish. Confluent monolayers of cells were subcultured. The culture medium was replaced every 3-4 days. Human uterine endometrium (UtE) was harvested from a patient with endometriosis. The endometrium was sterilized in PBS and cut into small pieces with dissection scissors. These pieces were placed in Dulbecco’s Modified Eagle’s Medium (Sigma Chemical Co. St. Louis, MO), supplemented with 10% FBS and an antibiotic-antimycotic (100×) solution (Invitrogen), and incubated for 10-14 days at 37°C in a humidified 5% CO2 atmosphere. Subconfluent adherent cells were harvested in PBS containing 0.06% trypsin and 0.005% EDTA, and were subcultured. The culture medium was replaced every 4 days. This study was approved by the Ethical Committee of the National Institute for Child Health and Development. The purpose of this study was explained thoroughly to the patients, who gave their written informed consent.
hiPSCs were cultivated on irradiated MEFs in iPSellon medium (Cardio, Osaka, Japan), supplemented with 10 ng/ml recombinant human bFGF (Wako Pure Chemicals, Osaka, Japan). hiPSCs were established from MRC-5 and AM cells, as previously described [21, 22]. In addition, hiPSCs were established from PAE and UtE cells in the present study. Briefly, 1 × 105 cells were infected overnight with pooled viral supernatants, obtained by the transfection of HEK293FT cells (TransIT-293 reagent; Mirus, Madison, WI) with the retroviral vector pMXs, which encodes the cDNAs for OCT3/4, SOX2, KLF4, and c-MYC, together with the packaging plasmids pCLGagPol and pHCMV-VEV-G (a gift from T. Kiyono, National Cancer Center Research Institute, Tokyo, Japan). Four days after infection, the cells were split, plated on irradiated MEFs in 100-mm dishes, and maintained in iPSellon medium until colonies formed.
The immunocytochemical analysis was performed as described previously [22, 23]. Human cells were fixed with 4% paraformaldehyde in PBS for 10 min at 4 °C. After washing with 0.1% Triton X-100 in PBS (PBST), the cells were prehybridized in blocking buffer for 1–12 h at 4 °C, and then incubated for 6–12 h at 4°C with the following primary antibodies: anti-SSEA4 (1 : 300 dilution; Chemicon, Temecula, CA), anti-TRA-1–60 (1 : 300; Chemicon), anti-Oct4 (1 : 50; Santa Cruz Biotechnology, Santa Cruz, CA), anti-Nanog (1 : 300; ReproCELL, Tokyo, Japan), and anti-Sox2 (1 : 300; Chemicon). The cells were then incubated with anti-rabbit IgG, anti-mouse IgG or anti-mouse IgM conjugated with Alexa Fluor 488 or Alexa Fluor 546 (1: 500; Molecular Probes, Eugene, OR) in blocking buffer for 1 h at room temperature. The cells were counterstained with DAPI, and then mounted using a SlowFade light antifade kit (Molecular Probes).
Teratoma formation was performed as described previously [22, 23]. The 1:1 mixtures of the AM-hiPSC suspension and Basement Membrane Matrix (BD Biosciences, San Jose, CA) were implanted subcutaneously, at 1.0 × 107 cells / site, into immunodeficient, non-obese diabetic (NOD)/severe combined immunodeficiency (SCID) mice (CREA, Tokyo, Japan). Teratomas were surgically dissected out 6–10 weeks after implantation, and were fixed with 4% paraformaldehyde in PBS and embedded in paraffin. Sections of 10-μm thickness were stained with hematoxylin-eosin.
Gene expression analysis
Total RNA samples were extracted using ISOGEN (NipponGene). The global gene expression patterns and changes in mRNA levels were monitored using Agilent Whole Human Genome Microarray chips (G4112F) with one-color (Cyanine 3) dye. This microarray chip covers 41,000 well-characterized human genes and transcripts. The raw microarray data were submitted to the GEO (Gene Expression Omnibus) microarray data archive (http://www.ncbi.nlm.nih.gov/geo/) at the NCBI (accession number: GSE 20750). After background correction using a Normal plus Exponential convolution model, which adjusts the foreground to the background, we used an offset to dampen the variation of the log-ratios for intensities close to zero.
Among the 41,000 probes, 16,483 representative probes corresponding to MAQC unique genes were used for the following analyses [44]. Global array clustering was performed by the complete linkage method with Euclidean distance, and was visualized using the Java TreeView 1.1.0 software; the gene expression values are displayed as normalized log ratios. Cell line similarities were measured using Pearson correlation coefficients. To further validate whether the global gene expression is different in each origin cell, we evaluated the classification accuracy by leave-one-out cross-validation (LOOCV) on the nearest-neighbor classifier, based on Pearson's correlation distance. To obtain the expression signatures, we performed a differential analysis for each origin cell: differences between the two arbitrary datasets were evaluated by the Student’s t-test for the expression of each gene. Thereafter, the false discovery rate (FDR) was estimated using the Benjamini–Hochberg procedure. Differentially expressed genes were selected if they satisfied both FDR <0.05 and a 2.0-fold change in the average values for the cell lines being compared. The gene ontology analysis was performed using the GO Term Finder Perl script [45] (http://go.princeton.edu/cgi-bin/GOTermFinder), with EBI human GO annotations and generic GO slim annotations (http://www.geneontology.org/).
Network screening
Network screening was performed as described previously [21]. This analysis is based on the procedure for estimating the consistency of a network structure (directed acyclic graph) with the measured data for the constituent variables in the graph. The joint density function for a given network (reference network) was recursively factorized into conditional density functions, according to the parent-child relationship in the graph. The conditional functions were quantified into log-likelihoods, using linear regression for the measured data, with the assumption that the data followed a normal distribution. The probability of the log-likelihood for the network structure (graph consistency probability; GCP) was then estimated from the distribution of log-likelihoods for 2,000 networks, generated under the condition that the networks shared the same numbers of nodes and edges as those of the given network. The significance probability of the given network was set at 0.05 in this analysis.
In the present study, the GCP was estimated for the ensemble of reference networks, to extract the candidate activated networks in the hiPSCs, in a process termed ‘network screening’. The reference networks were constructed using the ChIP-on-Chip data and the classification scheme for gene function. The genes bound by four factors were cited from a previous report [20], and were divided into sub-networks according to the functional gene sets previously defined in the Molecular Signatures Database (MSigDB) [24]. The sub-networks that included at least one gene of the expression signature were then selected. The set of selected sub-networks was used as the reference network for network screening.
Glycan analysis
We analyzed cell surface glycans with a lectin microarray [31]. The 43 lectins were dissolved at a concentration of 0.5 mg/ml in spotting solution (Matsunami Glass, Osaka, Japan), and were spotted onto epoxysilane-coated glass slides (Nexterion Slide E Epoxysilane-coated Substrate 25 × 75.6 × 1 mm; Schott, Mainz, Germany) attached to a silicone rubber sheet, using a non-contact microarray printing robot (MicroSys 4000; Genomic Solutions, Ann Arbor, MI). The lectins were spotted in triplicate, with a spot diameter of 500 μm. The glass slides were incubated at 25°C for 3 h, to allow lectin immobilization. The lectin-immobilized glass slides were then washed with probing buffer (25 mM Tris-HCl [pH 7.5], 140 mM NaCl, 2.7 mM KCl, 1 mM CaCl2, 1 mM MnCl2, 1% [v/v] Triton X-100), and incubated with the blocking reagent N102 (NOF, Tokyo, Japan) at 20°C for 1 h. Finally, the lectin-immobilized glass slides were flooded with TBS containing 0.1% NaN3 and stored at 4°C. The cell membrane faction was prepared using the CelLytic MEM Protein Extraction Kit (Sigma-Aldrich, Tokyo, Japan), and the protein concentration was determined using the MicroBCA Protein Assay Reagent kit (Thermo Fisher Scientific, Waltham, MA). After dilution in PBST (10 mM PBS [pH 7.4], 140 mM NaCl, 2.7 mM KCl, 1% Triton X-100), the cell membrane fraction was labeled with Cy3 NHS ester (GE Healthcare Ltd., Buckinghamshire, England). After dilution in probing buffer to the desired concentration, the Cy3-labeled cell membrane fraction was applied to the lectin microarray and incubated at 20°C overnight. After washing with the probing buffer, fluorescence images were acquired using an evanescent-field activated fluorescence scanner (SC-Profiler; GP BioScience, Kanagawa, Japan). The fluorescence signal of each spot was quantified using the Array Pro Analyzer ver. 4.5 software (Media Cybernetics, Bethesda, MD), and the background value was subtracted. The values shown for the lectin signals represent the average of triplicate spots.