Ethics statement
This study was approved by the Research Committee for Human Subjects, Shanghai Tongji University School of Medicine. All participants provided their written informed consent to participate in this study.
Patients and tissues
Thirty pregnant women recruited in this study were categorized into three groups (ten pregnancies in each group) according to the serum bile acid level at gestational age of 32 weeks: healthy group with serum bile acid concentration < 10 μM; mild ICP, with serum bile acid concentration ranging from 10–40 μM; and severe ICP, with bile acid concentration >40 μM (Figure 1A). ICP was diagnosed with the following symptoms: 1) presence of classical pruritus with no rash and quick disappearance after delivery; 2) raised maternal serum bile-acid concentration (>10 μM); 3) absence of itching skin disease; and 4) absence of any other causes of liver dysfunction, including preeclampsia; hemolysis, elevated liver enzymes and low platelets (HELLP) syndrome; acute fatty liver of pregnancy; primary biliary cirrhosis; viral hepatitis; and any ultrasound abnormality. All of the ICP patients were not treated with UDCA under their consents. Exclusion criteria for recruitment of healthy pregnant women were similar to that for ICP cases.
To avoid artefact from effects of labor, only placental samples obtained from women who had not undergone labor were selected. All samples were collected immediately after Caesarean sections, cut near the center zone of maternal surface villous lobule after deciduas and amnionic membranes were removed, and then rinsed with saline to remove maternal blood.
Total serum bile acids analysis
Fasting Serum samples were taken from women and measured twice every week since the gestational age of 32 weeks until caesarean section. Total bile acids were assessed by using an enzymatic method (Kit No Bl 3863, Randox Laboratoires Ltd, United Kingdom) on a Hitachi 7180 biochemistry analyzer (Hitachi, Tokyo, Japan).
RNA preparation
Freshly isolated placental tissues were divided and dissected separately on ice-cooled RNase-free surfaces. Each sample was placed into an individual tube before addition of 1 mL of TRIzol reagent (Invitrogen), followed by homogenization with a Polytron mixer (Kinematica). Total RNA was extracted according to manufacturer’s instructions (Invitrogen). For DNA microarray, ten pregnant women in each group were randomly divided into two sub-groups and equal amounts of RNA from five individuals in each sub-group were pooled for one array to reduce the variations among different individuals [14, 15].
DNA microarray and hybridization
Agilent Human 4 × 44 K Gene Expression Arrays, which target 27 958 Entrez Gene RNAs, were employed in this study. Sequences were compiled from a broad survey of sources and were subsequently verified and optimized by alignment to the assembled human genome. Coupled with Agilent’s probe selection and robust validation processes, this design delivers increased data quality and less redundancy in gene coverage.
Sample labeling and array hybridization were performed according to the Agilent One-Color Microarray-Based Gene Expression Analysis protocol (Agilent Technology). Briefly, total RNA from each sample was linearly amplified and labeled with Cy3-UTP. The labeled cRNAs were purified using the RNeasy Mini Kit (QIAGEN). The concentration and specific activity of the labeled cRNAs (pmol Cy3/μg cRNA) were measured using a NanoDrop ND-1000. One microgram of each labeled cRNA was fragmented by addition of 11 μl 10× Blocking Agent and 2.2 μl 25× Fragmentation Buffer followed by heating at 60°C for 30 min. Finally, 55 μl 2× GE Hybridization buffer was added to dilute the labeled cRNA. One hundred microliters of hybridization solution was dispensed into the gasket slide, which was assembled with the gene expression microarray slide, followed by incubation for 17 hours at 65°C in an Agilent Hybridization Oven. The hybridized arrays were washed, fixed, and scanned with the Agilent DNA Microarray Scanner (part number G2505B).
Data analysis and differentially expressed genes sorting
Agilent Feature Extraction software (version 11.0.1.1) was used to analyze acquired array images. Quantile normalization and subsequent data processing were performed using the GeneSpring GX v12.0 software package. Genes that had flags in at least two out of six arrays were chosen for further analysis. Detailed gene expression data are available at the GEO website under accession number GSE46157.
To sort the differentially expressed genes in mild ICP and severe ICP compared with healthy pregnancy, microarray data were analyzed by using linear models and empirical Bayes method, which is considered to be able to generate a reliable statistical inference when the number of arrays is small [16, 17]. This method is similar to a standard t test for each probe except that the SES are moderated across genes to get more stable results, which prevents a gene with a very small fold change from being judged as differentially expressed due to an accidentally small residual SD. The resulting p values were further adjusted using the BH FDR algorithm [18]. Subsequently, differentially expressed genes were sorted according to fold change (±1.5-fold or greater), followed by secondary selection based on p-value and FDR threshold (p < 0.05, FDR < 0.05, controlling the expected FDR to no more than 5%).
Hierarchical clustering of differentially expressed genes
Hierarchical clustering was performed using the Agilent GeneSpring GX software (version 12.0), and the results were expressed as a dendrogram in which genes and samples with a similar expression pattern formed clusters.
Gene Ontology (GO) analysis
Gene Ontology (GO) analysis was applied to characterize the primary functions of the differentially expressed genes based on the key functional classification of NCBI [19]. Fisher’s exact test was used to classify the GO category. A p-value was calculated for each GO term for all differentially expressed genes and subsequently corrected by false-discovery rate (FDR) [20]. A smaller FDR is associated with a smaller error in judging the p-value. Next, enrichment analysis was performed to find GO terms with more specific functions. The enrichment Re was given by Re = (n
f
/n)/(N
f
/N), where n
f
is the number of differential genes within the particular category, n is the total number of genes within the same category, N
f
is the number of differentially expressed genes in the entire microarray, and N is the total number of genes on the microarray.
Gene co-expression network
To identify modules of co-expressed genes [21], we built the gene co-expression network according to the normalized signal intensities of specific genes. We calculated the Pearson’s correlation for each pair of genes and chose those whose values fit the selection criterion to construct the network [22]. Degree of centrality, defined as the number of links between one node and the others, was used as the simplest and most important measure of the centrality of a gene within a network, which in turn determines that gene’s relative importance. In addition, the k-core (from graph theory) was introduced to simplify the graph-topology analysis for the purposes of locating the core regulatory factors (genes) in the gene co-expression network. The k-core of a gene co-expression network usually contains a cohesive group of genes [23–25]. A k-core sub-network with a higher k-core level is considered to have core status within a large-scale gene network.
Quantitative real-time PCR (qPCR)
Total RNA of each placental sample was prepared as described in RNA preparation. Synthesis of cDNA was carried out using Reverse Transcriptase M-MLV (RNase H-; TaKaRa) and oligo-dT as the primer. QPCR was performed on an Applied Biosystems 7500 Fast Real-Time PCR System using Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Primer pairs used were listed in Additional file 1: Table S1. Reactions were performed in triplicate under standard thermocycling conditions (one cycle of 94°C for 4 min, followed by 45 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 40 s), and the mean threshold cycle number was used.
Histology and immunohistochemistry
Freshly isolated placental tissues were fixed with 4% PFA overnight at 4°C, embedded in paraffin, sectioned at 4-μm thickness, and followed by hematoxylin and eosin staining.
Immunohistochemical analyses were performed using the LSAB kit (ZSGB-Bio, Beijing, China). Briefly, paraffin sections were deparaffinized in xylene and rehydrated sequentially in ethanol. The sections were microwaved in EDTA buffer (pH 8.0) to retrieve antigens, incubated in 3% H2O2 (10 min, room temperature) to inactivate endogenous peroxidase. Sections were incubated with blocking solution (15 min, 37°C), followed by incubation in mouse anti-human CD45 antibody (1:200, 304002, Biolegend) diluted in blocking buffer (overnight, 4°C). The samples were then incubated with biotinylated secondary antibody (15 min, 37°C), followed by incubation with streptavidin conjugated with HRP (15 min, 37°C) and staining with DAB substrate. Nuclei were lightly counterstained with hematoxylin. Finally, slides were dehydrated sequentially in ethanol, cleared with xylenes, and mounted with neutral resin.
For immunofluorescence staining, tissue was fixed in 4% PFA overnight at 4°C, immersed in sucrose solutions, embedded in OCT (Tissue Tek) and sectioned. Frozen sections were incubated with 10% FBS (PBS) to block nonspecific binding sites (1 h, RT) and further incubated with mouse anti-humam CD19 antibody (Alexa Fluor® 700 conjugated, HIB19, 1:200, BD Pharmingen) or mouse anti-human CD3 antibody (FITC conjugated, APA1/1, 1:200, BD Pharmingen) at 4°C overnight. Nuclei were stained with DAPI.
The stained sections were imaged using an Olympus BX51 microscope coupled with an Orca CCD camera (Q-IMAGING).
Quantifications of blood vessels and immune cells
The HE staining image was used for blood vessel quantification. Ten sections of the terminal villi at the same position in each placenta were examined and 5 random fields in each slide were selected (50 fields total for each placenta) to avoid areas of placental infarction and intervillous fibrin deposition, arterial vessels forming the primary stem and anchoring villi; and histological artifact. The selected fields were then analyzed using ImageJ to count the number of capillaries in each villus. Only terminal villi showing outlines completely within the microscopic field were analyzed. The number of capillaries per villus was calculated by the total number of capillaries divided by the number of terminal villi in 50 fields for each placental sample.
For immune cell quantification, ten sections of the terminal villi at the same position in each placenta were examined and the number of different immune cells in 5 random fields in each section were counted (50 fields total for each placenta). The average of cell number per field was calculated for each placenta.
Statistical analysis
P values of qPCR results, immune cell quantification and blood vessel quantification were calculated by analysis of variance (ANOVA) with Dunnett post-tests. Significant differences are indicated as: *, p < 0.05; **, p < 0.01. Each experiment was repeated independently for three times.