The expression pattern of miR-29b in the placental tissues from patients with GDM and normal controls was determined by quantitative reverse transcription polymerase chain reaction (qRT-PCR). The overall expression trend of miR-29b in all placental tissues of GDM group was lower than that in control group (P < 0.05; Figure 1A). The GDM patients were grouped into six groups according to the abnormal oral glucose tolerance test (OGTT) values in blood, including 1st hour (h), 2nd h, 0 and 1st h, 0 and 2nd h, 1st and 2nd h, 0 and 1st and 2nd h. The expression of miR-29b was significantly reduced in the groups of 2nd h (P < 0.05), 0 and 2nd h (P < 0.05), 0 and 1st and 2nd h (P < 0.05; Figure 1B). Meanwhile, the GDM placental tissues were grouped into four groups according to age, including ≤ 25 years (≤ 25 Y), 26–30 years (26–30 Y), 31–35 years (31–35 Y), and equal to or over 36 years (≥ 36 Y). There was a trend toward a decrease in the level of expression miR-29b in normal placenta with age (Figure 1C). MiR-29b expression level was significantly down-regulated in the groups of ≤25 Y (P < 0.05), 26–30 Y (P < 0.05), and 31–35 Y (P < 0.05; Figure 1D) as compared with matched controls. The distribution of miR-29b in placenta was detected by in situ hybridization (Figure 1E). The positive staining of miR-29b was found in the surface of the syncytiotrophoblast membrane and villous stroma in both GDM patients and matched controls.

In order to detect the transfection efficiency of miR-29b mimic and inhibitor, miR-29b expression level in HTR8-/SVneo cells transfected with miR-29b mimic, miRNA mimic negative control, miR-29b inhibitor, or miRNA inhibitor negative control was detected by qRT-PCR. MiR-29b mimic significantly enhanced the expression level of mature miR-29b (P < 0.01; Figure S1A). Mature miR-29b expression was evidently reduced by miR-29b inhibitor (P < 0.01; Figure S1B).

The effects of miR-29b on cell growth were analyzed by MTT assay and EdU assay (Figure 2). The results showed that miR-29b over-expression inhibited cell viability and proliferation as compared with control group (P < 0.05; Figures 1B, 2A). There was no significant difference in cell viability and proliferation between miR-29b inhibitor and miRNA inhibitor negative control (Figures 2A,B).

The effects of miR-29b on cell apoptosis were detected by flow cytometry analysis (Figure 3). When trophoblast cells were transfected with miR-29b mimic, the percentage of apoptotic cells was significantly increased at late stage (P < 0.05; Figure 3B) and no statistical difference at early stage as compared with miRNA mimic negative control. No significant difference was observed in cell apoptosis level between miR-29b inhibitor and miRNA inhibitor negative control at both early and late stages (Figure 3C).

In order to further analyze the role of miR-29b in the occurrence of GDM, the effects of miR-29b on trophoblast cell migration and invasion were evaluated by a Transwell chamber. The number of migrating cells was significantly decreased in miR-29b mimic group compared with mimic control (P < 0.05) and markedly increased in miR-29b inhibitor group compared with inhibitor control (P < 0.01; Figure 4A). The number of cell invasion was less in miR-29b mimic group than that in mimic control group (P < 0.01) and more in miR-29b inhibitor group than that in inhibitor control group (P < 0.01; Figure 4B).

To discover the possible molecular mechanisms through which miR-29b may play roles in cell growth and migration, the target genes of miR-29b were predicted online by Targetscan, PicTar and miRanda and a large number of putative miRNA targets were found. We focused on genes associated with cell proliferation and apoptosis. Inhibitor of growth protein-2 (ING2), inhibitor of growth protein-3 (ING3) and hypoxia-inducible factors 3A (HIF3A) were predicted to be the target genes of miR-29b. The 3′-UTR segment of human ING2, ING3, or HIF3A among different species was conservative (Figures S2A, S3A and Figure 5A).

In order to confirm whether these genes were the target genes of miR-29b, the 3′-UTR segment of human ING2, ING3, or HIF3A were cloned into pmirGLO vector, respectively (called ING2-pmirGLO, ING3-pmirGLO, HIF3A-pmirGLO; Figures S2B, S3B, and Figure 5B). ING2-pmirGLO, ING3-pmirGLO or HIF3A-pmirGLO was, respectively, co-transfected with miR-29b mimic or inhibitor to detect the luciferase activity in HeLa cells and HEK-293T cells (Figures S2C, S3C,D and Figures 5C,D). The stronger the luciferase activity, the stronger the expression of the target gene, that is, the weaker the binding of the miR-29b to 3′-UTR of the target gene. When cells were co-transfected with miR-29b mimic or inhibitor and ING2-pmirGLO or ING3-pmirGLO, the luciferase activity was no significant difference compared with control group. When cells were co-transfected with HIF3A-pmirGLO and miR-29b mimic or mimic control, the luciferase activity was weaker in miR-29b mimic group than that in mimic control group in HeLa cells (P < 0.05), but no significant difference in HEK-293T cells. When cells were co-transfected with HIF3A-pmirGLO and miR-29b inhibitor or inhibitor control, the luciferase activity was stronger in miR-29b inhibitor group than that in inhibitor control group in HEK-293T cells (P < 0.05), but no significant difference in HeLa cells. These results suggest that miR-29b may bind to 3′-UTR of HIF3A.

To further confirm whether miR-29b may bind to 3′-UTR of HIF3A, the potential recognition sequences of miR-29b in 3′-UTR of HIF3α mRNA was mutated (named as HIF3A-pmirGLO-Mut; Figure 5E). There were two potential miR-29b recognition sites within 3′-UTR of HIF3A mRNA. Compared with HIF3α-pmirGLO-Mut1 or/and HIF3A-pmirGLO-Mut2, the enzyme activity was significantly reduced in HeLa cells co-transfected with miR-29b mimic and HIF3A-pmirGLO (P < 0.01). Moreover, the luciferase activity was higher in both HIF3A-pmirGLO-Mut1 and HIF3A-pmirGLO-Mut2 group than that in HIF3A-pmirGLO-Mut1 or HIF3A-pmirGLO-Mut2 alone (P < 0.05). These data indicate that miR-29b can bind to the recognition element in 3′-UTR of HIF3A, and that HIF3A is a target gene of miR-29b.

Though miR-29b could bind to the recognition sequences in 3′-UTR of HIF3α, it was unclear whether miR-29b could regulate HIF3α expression in trophoblast cells. Here, we found that the protein level of HIF3α was significantly reduced in miR-29b mimic group as compared with mimic control group (P < 0.05). MiR-29b inhibitor significantly enhanced the protein level of HIF3α as compared with inhibitor control (P < 0.05; Figure 6A). These results show that miR-29b can inversely regulate HIF3α protein level.

The HIF3α expression in the placental tissues from patients with GDM was detected by immunohistochemistry (Figure 6B). The positive signals of miR-29b were located in the apical membrane of the syncytiotrophoblast and villous stroma in different age groups in both the GDM patients and corresponding controls. The mean optical densities (MOD) of positive signals was determined through Image J software with H-score grading system in three different optical fields selected in a random manner for each sample. The MOD of positive signals was stronger in GDM group than that in matched controls in the groups of ≤25 Y and 26–30 Y (P < 0.05; Figure 6B1).

To analyze whether HIF3A was the functional target of miR-29b, we performed the target gene compensation experiment (Figures 7, 8). MiR-29b mimic could inhibit cell proliferation (P < 0.05), migration and invasion. When HIF3A was over-expressed in cells transfected with miR-29b mimic, cell proliferation, late apoptosis, migration and invasion ability were stronger than that in cell transfected with miR-29b mimic alone (P < 0.05), close to control (Figure 7). MiR-29b inhibitor could promote migration and invasion. When HIF3A was knocked down in cells transfected with miR-29b inhibitor, migration and invasion ability were weaker than that in cell transfected with miR-29b inhibitor alone (P < 0.05), close to normal control (Figure 8).

GDM was diagnosed by OGTT. All subjects were orally administrated with 75 g glucose regardless of body weight. Blood samples were collected at times 0, 1, and 2 hour (h) and the blood glucose were measured. If fasting blood glucose >5.1 mmol/L, 1 h postprandial blood glucose >10.0 mmol/L, or 2 h postprandial blood glucose >8.5 mmol/L, it can be diagnosed as GDM. Placentas from 204 GDM patients and 202 normal pregnant women, beyond 37 weeks of gestation, were collected from Maternal and Child Health Hospital of Haidian District in Beijing, from March 2012 to May 2013. Basic clinical data of GDM and control population were showed in Table 1. The study was approved by Ethics Committee of Research Institute for Family Planning (2011-08). The informed consent was got from all participants. Placentas were divided into two parts: one part was fixed in 4% paraformaldehyde (PFA) diluted in 0.1 M phosphate-buffered saline (PBS) for immunohistochemical and in situ hybridization studies; one part was frozen in the −80°C refrigerator for extracting total RNA and protein.

The ING2, ING3, HIF3A 3′-UTR, and HIF3A 3′-UTR-mutant sequences were amplified by PCR from human genomic DNA using the primers in Table 2. After being double digested with XhoI and XbaI, the PCR products were cloned into pGL3 control vector (Invitrogen, Carlsbad, CA, USA). The coding region of HIF3A sequence was amplified and cloned into pcDNA3.1 vector, designated as pcDNA3.1-HIF3A. All the constructs were verified by DNA sequencing. Specific siRNAs for scramble and HIF3A were purchased from RiboBio Co., Ltd. (Guangzhou, Guangdong, China).

The miR-29b mimic, mimic control, miR-29b inhibitor, inhibitor control, scramble siRNA control and HIF3A siRNA were synthesized by GenePharma (GenePharma Co., Ltd., Shanghai, China), and transfected into cells by the lipofectamine 2,000 (Invitrogen, Carlsbad, CA, USA) according to the manufacture's instruction.

Placentas from 204 GDM patients and 202 normal pregnant women were separately arranged into two tissue microarrays according to the match between the case group and the control group. Tissue microarrays were cut to 6 μm thickness. The sections were hybridized with digoxigenin (DIG)-labeled LNA-MiR-29b probe (LNA-miR-29b sequence: 5′ –DIG-aAcaCtgAtttCaaatGgtgCta-3′) overnight at 42°C. The probe was replaced by DIG-labeled LNA-scrambled probe (LNA-scrambled sequences: 5′-DIG-caTtaAtgTcGgaCaaCtcAat-3′) was served as negative control. After blocked by blocking buffer containing 5% bovine serum albumin (BSA), the sections were incubated with alkaline phosphatase labeled anti-DIG-antibody (Roche, Mannheim, Germany, 1: 200) overnight at 4°C, and colorated with bromochloroindolyl phos-phate/nitro blue tetrazolium (BCIP/NBT; Promega, Madison, WI, USA). Samples were photoed by Pannoramic scan digital slice scanner (3DHISTECH Ltd., Budapest, Hungary).

Tissue microarrays including placentas from 204 GDM patients and 202 normal pregnant women were incubated with rabbit anti-HIF3α polyclonal antibody (GeneTex, USA, 1:500), then incubated with HRP-conjugated goat anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA, USA). The sections were developed by diaminobenzidine (DAB; Sigma-Aldrich, St. Louis, MO, USA), and then stained with haematoxylin (Sigma-Aldrich, St. Louis, MO, USA). Normal goat serum instead of primary antibody was served as negative control. Samples were observed by Pannoramic scan digital slice scanner (3DHISTECH Ltd., Budapest, Hungary). The MOD of positive signals was determined through Image J software with H-score grading system in three different optical fields selected in a random manner for each sample. Random 15 pair samples in each age group (≤ 25 Y, 26–30 Y, 31–35 Y, ≥36 Y) were selected for optical density analysis.

Total RNAs from human placentas were isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA). The expression of miR-29b was detected by qRT-PCR using TaqMan MicroRNA Reverse Transcription Kit and TaqMan Univel PCR Master Mix (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. U6 snoRNA was used for normalization. Random 10 pair samples in each age group (OGTT 1st h, 2nd h, 0 and 1st h, 0 and 2nd h, 1st and 2nd h, 0 and 1st and 2nd h) and random 15 pair samples in each age group (≤ 25 Y, 26–30 Y, 31–35 Y, ≥36 Y) were selected for qRT-PCR. Each sample in each experiment was detected in triplicate.

The effects of miR-29b on human trophoblasts cells viability were estimated with MTT assay. HTR8/SVneo cells were seeded in 96-well plate (5,000 cells per well) and allowed to attach overnight. Cells were transfected with 50 nM miR-29b mimic, miRNA mimic negative control, miR-29b inhibitor or miRNA inhibitor negative control, respectively. After 48 h, 20 μL MTT (5 mg/mL; Promega, Madison, WI, USA) was added to each well and incubated for 4 h at 37°C. The supernatant was then removed and the 150 μL dimethyl sulfoxide (Sigma-Aldrich, St. Louis, MO, USA) was added into each well. The absorbance value was recorded at A570 nm with a 96-well plate reader (Bio-Rad 3550; Bio-Rad, Hercules, CA, USA). Each treatment group had three replicates and the experiment was repeated for three times (n = 9).

EdU is a thymidine analog which can infiltrate into the replicated DNA molecule instead of thymidine during cell proliferation to label proliferative cells. The effects of miR-29b on human trophoblasts cells proliferation were analyzed with Cell-Light^TM EdU Apollo®567 in vitro Imaging Kit (RiboBio, Guangzhou, Guangdong, China) according to the manufacturer's instructions. Briefly, HTR8/SVneo cells were exposed to 50 μM EdU for 2 h at 37°C after 48 h transfection of 50 nM miR-29b mimic, miRNA mimic negative control, miR-29b inhibitor or miRNA inhibitor negative control, respectively. Then cell nuclei were stained with Hoechst 33342. The samples were viewed and imaged under a laser-scanning confocal microscope (CarlZeiss LSM 710 META, Germany). Proliferative cells and total cell number were separately computed in three fields (magnification × 400) selected in a random manner. Each treatment group had three replicates and the experiment was repeated for three times (n = 9). The results were represented as a ratio of the number of proliferative cells vs. total cells.

The effects of miR-29b on human trophoblasts cells apoptosis were analyzed with by annexin V-FITC staining kit (Invitrogen, Carlsbad, CA USA) according to the manufacturer's instructions. Briefly, HTR8/SVneo cells were harvested after 48 h transfection of 50 nM miR-29b mimic, miRNA mimic negative control, miR-29b inhibitor or miRNA inhibitor negative control, respectively. Each sample added 5 μL Alexa Fluor® 488 annexin V and 1 μL PI. After 15 min of incubation, each sample added 400 μL 1 × annexin V-binding buffer. The samples were detected at a rate of 8,000 events per second by flow cytometry (BD Biosciences, San Jose CA, USA). The experiment was repeated for three times (n = 3).

Cell migration and invasion were assayed using a Transwell chamber (Corning, NY, USA). HTR8/SVneo cells transfected with 50 nM miR-29b mimic, miRNA mimic negative control, miR-29b inhibitor or miRNA inhibitor negative control were seeded on the top chamber of 24-well plate with or without 40 μL of 1 mg/ml Matrigel (BD, Franklin Lakes, NJ, USA). After 18 h of incubation for migration assay or 24 h for invasion assay, the non-migrated or infiltrated cells on the top of chamber were carefully removed with a cotton swab, and the migrant or invasive cells on bottom of chamber were fixed with 4% formaldehyde. The cells were then stained by hematoxylin and eosin (Sigma-Aldrich, St. Louis, MO, USA). The cells were viewed and imaged under a microscope (BX61; Olympus, Tokyo, Japan). Three different fields (magnification × 400) were selected in a random manner to calculate the number of migration or invasion cells. The results were represented as the average number of cells per field. The experiment was repeated for three times (n = 3).

Extracted protein was boiled in SDS/β-mercaptoethanol sample buffer. Sixty microgram protein was run on 6–10% gradient polyacrylamide gel and transferred to PVDF membranes (Amersham, St Albans, Herts, UK). The membranes were incubated with rabbit anti-HIF3α polyclonal antibody (GeneTex, USA, 1:1000) or mouse anti-β-ACTIN monoclonal antibody (Abcam, Cambridge, MA USA, 1:1000) 2 h at 27°C. Then the membranes were incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG or goat anti-mouse IgG (Jackson Immunoresearch Laboratories, Inc., West Grove, PA, USA, 1:10,000) 1 h at 37°C. ECL The membranes were incubated with ECL reagents 1 min (Millipore, Billerica, MA, USA) and exposed to X-ray film (Kodak, USA). The β-ACTIN was used as a loading control. The experiment was repeated at least three times. The bands were analyzed using Quantity One analyzing system (Bio-Rad, Hercules, CA, USA). The protein level was represented as the relative ratio of the HIF3α signals vs. β-ACTIN signals. The detection of HIF3α protein level in HTR8/SVneo cell transfected with miR-29b mimic and inhibitor was repeated for three times (n = 3).

Partial sequence of the 3′-UTR from HIF3A and mutating miR-29b target site in the 3′-UTR of HIF3A were cloned into the downstream of the firefly luciferase gene in PmirGLO Vector, respectively (Promega, Madison, WI, USA) using primers in Table 2. HEK-293T cells or HeLa cells were seeded in 48-well plates and attached overnight, then transfected using lipofectamine 2,000 (Invitrogen, Carlsbad, CA, USA). 48 h later, cells were detected by the dual-luciferase assay (Promega, Madison, WI, USA). The detection of binding ability of miR-29b and HIF3A 3′-UTR was performed three separate experiments. Each treatment group was repeated for three times for independent triplicate experiments (n = 9). The results were expressed as relative luciferase activity (Firefly LUC/Renilla LUC).

Data are presented as means ± standard deviations (SD) and the error bars represent SD. Comparisons between two groups (paired and unpaired) were performed using a two-tailed Student's t-test. One-way analysis of variance (ANOVA) and Turkey's post-hoc test was used to compare data sets with more than two groups. P < 0.05 was considered statistically significant. All of the statistical analyses were performed with SPSS 16.0 software.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.