Up-regulation of miR-98 and unraveling regulatory mechanisms in gestational diabetes mellitus

MiR-98 expression was up-regulated in kidney in response to early diabetic nephropathy in mouse and down-regulated in muscle in type 2 diabetes in human. However, the expression prolife and functional role of miR-98 in human gestational diabetes mellitus (GDM) remained unclear. Here, we investigated its expression and function in placental tissues from GDM patients and the possible molecular mechanisms. The results showed that miR-98 was up-regulated in placentas from GDM patients compared with normal placentas. MiR-98 over-expression increased global DNA methylational level and miR-98 knockdown reduced global DNA methylational level. Further investigation revealed that miR-98 could inhibit Mecp2 expression by binding the 3′-untranslated region (UTR) of methyl CpG binding protein 2 (Mecp2), and then led to the expression dysregulation of canonical transient receptor potential 3 (Trpc3), a glucose uptake related gene. More importantly, in vivo analysis found that the expression level of Mecp2 and Trpc3 in placental tissues from GDM patients, relative to the increase of miR-98, was diminished, especially for GDM patients over the age of 35 years. Collectively, up-regulation of miR-98 in the placental tissues of human GDM is linked to the global DNA methylation via targeting Mecp2, which may imply a novel regulatory mechanism in GDM.

Gestational diabetes mellitus (GDM) is a condition in which women without previously diagnosed diabetes exhibit varying degrees of glucose intolerance during pregnancy 1 . In different studies, gestational diabetes mellitus affects approximately 1.1-14.3% of pregnant women [2][3][4] , and has 35.6% to 69% of recurrence risk 5,6 . GDM has adverse effects on the pregnant women and fetus including pre-eclampsia, caesarean section rates, perinatal mortality, birth defects, macrosomia, etc. In longitudinal studies with a duration of at least 5 years, 20% to 65% of women with GDM go on to develop type 2 diabetes (T2DM) 7 . The pathogenesis of GDM is not fully understood, the syndrome has many similarities to T2D that becomes manifest during the course of pregnancy. T2D, also called non-insulin-dependent diabetes (NIDDM), is characterized by hyperglycemia resulting from impairment of insulin secretion and/or defects in insulin action in peripheral tissues. GDM represents a combination of acquired and intrinsic abnormalities of insulin action. The precise mechanisms underlying gestational diabetes are still largely unknown.
MiRNAs are small 19-23 nucleotide RNA molecules that act as negative regulators of gene expression by mediating messenger RNAs degradation or translational arrest. They are potent drivers of differentiation and development in many biological processes. The evidence increasingly shows that miRNA dysregulation has been linked to diabetes in recent years. As is known, miRNAs play roles in type 1 and type 2 diabetes (T1D and T2D), focusing on β -cell biology, insulin resistance and diabetes complications 8 . MiR-98 expression is up-regulated in kidney in response to diabetes complications in mouse in and down-regulated in muscle in type 2 diabetes with insulin resistance in human 9,10 . Also, miR-98 is found to participate in embryo implantation during early pregnancy 11 . Although miR-98 is involved in T2DM and early pregnancy in the available literature [9][10][11] , the relationship between miR-98 and GDM remains unknown, and roles of miR-98 in GDM are still unclear.
In this study, we report the relationship between miR-98 and GDM and investigate the functional roles of miR-98 in GDM. Additionally, we also test the possible molecular mechanisms in which miR-98 is implicated. Demethylation reduces the expression of miR-98. In order to analyze the effect of demethylation on the expression of miR-98, DNA methylation inhibitor 5-aza was used to treat JEG-3 cells, and then the expression level of miR-98 was detected by qRT-PCR ( Fig. 2A). The results indicated that 0.1, 0.5, 1 μ M 5-aza inhibited the expression of miR-98 in a dose-dependent manner. However, only 1 μ M 5-aza significantly inhibited miR-98 expression (P < 0.01). These results suggest that miR-98 may be related with DNA methylation.

MiR-98 enhances the DNA methylation level in vitro.
To verify the relationship between miR-98 and DNA methylation, the global DNA methylation level in JEG-3 cells transfected by miR-98 mimic or inhibitor was estimated by the content of 5-methylcytosine (5-MeC) detected by cells immunohistochemistry (Fig. 2B). The ratio of 5-MeC positive cells was significantly increased in cells transfected by miR-98 mimic compared with mimic control (P < 0.05). MiR-98 inhibitor markedly reduced the ratio of 5-MeC positive cells (P < 0.05). DNA methylation inhibitor 5-aza can reduce the ratio of 5-MeC positive cells. However, transfection of miR-98 mimic Mecp2 is a direct target of miR-98. To figure out the possible molecular mechanisms by which miR-98 may perform in DNA methylation, its target genes were researched. An online search of miR-98 targets by Targetscan, PicTar and miRanda provided a large number of putative miRNA targets. Among them, we focused on Mecp2 for the following reasons: (1) Targetscan, PicTar and miRanda prediction showed that there was a miR-98 responsive element in 3′ -UTR of Mecp2, which is a highly conserved domain among different species (Fig. 3A).
(2) It was reported that Mecp2 was associated with methylation 12 . (3) In this study, we found that the mRNA and protein level of Mecp2 was significantly up-regulated when miR-98 was down-regulated in 5-aza-treated cells (Fig. 3B,C). Thus, an evident inverse relationship is showed between miR-98 and Mecp2 expression levels.
To validate whether Mecp2 was the indeed target gene of miR-98 or not, a human Mecp2 3′ -UTR fragment containing wild-type was cloned into the downstream of the firefly luciferase reporter gene in the pGL3 control vector (designated as Mecp2-pGL3) for the dual-luciferase assay (Fig. 4A). HEK-293T cells were co-transfected with Mecp2-pGL3 and miR-98 mimic or inhibitor. Compared with the mimic control, the luciferase activity was significantly suppressed by the miR-98 mimic, (P < 0.01). Furthermore, the luciferase activity was significantly . Gapdh and β -ACTIN serve as an internal reference for qRT-PCR and western blot, respectively. For western blot, the gels had been run under the same experimental conditions. The bands were analyzed using Quantity One analyzing system (Bio-Rad, Hercules, CA, USA). The histogram represents the optical densities of the signals quantified by densitometric analysis and expressed as MECP2 intensity/β -ACTIN intensity to normalize for gel loading and transfer. *P < 0.05; **P < 0.01.
To further confirm the binding site, base mutation of miR-98 targeting site in 3′ -UTR of Mecp2 (designated as MeCP2-pGL3-Mut) was also conducted. The histogram in Fig. 4B showed that the enzyme activity was significantly reduced in cells co-transfected with miR-98 mimic and Mecp2-pGL3 compared with Mecp2-pGL3-Mut (P < 0.05). These data indicates that miR-98 may suppress Mecp2 expression through binding to miR-98 responsive element in the 3′ -UTR of Mecp2, and Mecp2 may be a direct target of miR-98.

MiR-98 regulates endogenous Mecp2 expression in vitro.
Although Mecp2 was identified as a target gene for miR-98, it was unknown whether miR-98 could regulate endogenous Mecp2 expression. To verify the endogenous effects of miR-98 expression dysregulation on Mecp2, JEG-3 cells were transfected with miR-98 mimic or inhibitor. Compared with corresponding control, the level of MECP2 protein was significantly down-regulated by miR-98 mimic and up-regulated by miR-98 inhibitor (Fig. 5A). Additionally, the mRNA level of Mecp2 detected by qRT-PCR was significantly decreased by miR-98 mimic (P < 0.01) and increased by miR-98 inhibitor (P < 0.01). Compared with miR-98 mimic, miR-98 inhibitor significantly enhanced Mecp2 mRNA level (P < 0.001; Fig. 5B). These results show that the mRNA and protein levels of endogenous Mecp2 are regulated by miR-98.
Mecp2 was down-regulated in GDM placental tissues. Then it was unclear whether miR-98 executed its effects by targeting Mecp2 in GDM in vivo? To verify the phenomenon, we tested the MECP2 expression in the placental tissues from patients with GDM by immunohistochemistry using rabbit anti-MECP2 antibody (Fig. 6A). The MODs of MECP2-positive signals were significantly decreased in the placental tissues from GDM patients in Y30-35 and Y > 35 age group compared with the corresponding controls (P < 0.05), which was further confirmed by qRT-PCR (Fig. 6B). All these facts show that Mecp2 expression level is down-regulated in GDM tissues, while miR-98 expression level is visibly increased in GDM placental tissues, suggesting that miR-98 may execute its effects by targeting Mecp2 in GDM in vivo.

MiR-98 and Mecp2 regulate the protein expression of DNA methyltransferase. Because miR-98
was involved in DNA methylation, we wondered whether miR-98 and its target gene would affect the protein expression of DNA methyltransferase or not (Fig. 7). MiR-98 mimic increased DNMT1 protein level (P < 0.05) and Mecp2 expression vector decreased DNMT1 protein level (P < 0.05) compared with corresponding control. DNMT1 protein level had a downward tendency in cells treated by miR-98 inhibitor and a obvious increase by Mecp2 siRNA (P < 0.05). However, miR-98 and Mecp2 had no significant effects on the protein levels of DNMT3a and DNMT3B.

MiR-98 indirectly regulates the expression of Trpc3 by targeting Mecp2. Previous studies have
identified that canonical transient receptor potential 3 (Trpc3) and secreted frizzled-related protein 4 (Sfrp4), which play roles in T2DM, are the target genes of Mecp2 13,14 . Then it was unknown whether miR-98 affected the expression of Mecp2 target genes? Therefore, the expression of Trpc3 and Sfrp4 was detected by qRT-PCR (Fig. 8A,B). MiR-98 mimic significantly reduced the mRNA level of Trpc3 (P < 0.01). Mecp2 expression vector pMSCV-Mecp2 significantly increased Trpc3 expression (P < 0.01). When cells were co-transfected with miR-98 mimic and pMSCV-Mecp2, the mRNA level of Trpc3 were higher than transfection of miR-98 mimic (P < 0.01) and lower than transfection of pMSCV-Mecp2 (P < 0.01), implying that Trpc3 expression suppressed by miR-98 over-expression was partially rehabilitated by Mecp2 up-regulation. Additionally, miR-98 inhibitor significantly promoted the expression of Trpc3 (P < 0.01). Mecp2 siRNA significantly inhibited the expression of Trpc3 (P < 0.01). However, the mRNA level of Trpc3 in cells co-transfected with miR-98 inhibitor and Mecp2 siRNA was significantly weakened compared with miR-98 inhibitor alone (P < 0.01) and strengthened compared with Mecp2 siRNA alone (P < 0.01), displaying that miR-98 low expression-mediated the up-regulation of Trpc3 was partially attenuated by Mecp2 knockdown (Fig. 8A). MiR-98 mimic down-regulated Sfrp4 mRNA level, while miR-98 inhibitor, Mecp2 expression vector, Mecp2 siRNA also inhibited Sfrp4 expression. So, there was no distinct trend exhibiting the effects of miR-98 and Mecp2 on the expression of Sfrp4 (Fig. 8B).
In order to further confirm that miR-98 affected the expression of Trpc3, the protein level of TRPC3 was detected by western blot (Fig. 8C). Comparable results were obtained by western blot and qRT-PCR. MiR-98 mimic significantly reduced the protein level of TRPC3 (P < 0.05), and pMSCV-Mecp2 significantly increased TRPC3 protein level (P < 0.05). When cells were co-transfected with miR-98 mimic and pMSCV-Mecp2, the protein level of TRPC3 was higher than transfection of miR-98 mimic lone (P < 0.05) and close to the control. MiR-98 inhibitor significantly enhanced TRPC3 protein level (P < 0.05). TRPC3 protein level had a downward trend in cells treated by Mecp2 siRNA. However, the protein level of TRPC3 in cells co-transfected with miR-98 inhibitor and Mecp2 siRNA was significantly weaker than transfected with miR-98 inhibitor alone (P < 0.05) and stronger than transfected with Mecp2 siRNA alone (P < 0.05). Taken together, these results indicate that miR-98 executes functions in GDM partially by targeting Mecp2-Trpc3 pathway. Trpc3 expression in GDM placental tissues. In order to analyze Trpc3 expression in vivo, we tested the Trpc3 expression in the placental tissues from patients with GDM by qRT-PCR and western blot (Fig. 9). The Trpc3 mRNA level was significantly decreased in the placental tissues from GDM patients in Y > 35 age group compared with the normal control (P < 0.05; Fig. 9A). The TRPC3 protein level was significantly decreased in the placental tissues from GDM patients in Y30-35 and Y > 35 age group compared with the corresponding controls (P < 0.05; Fig. 9B).

Discussion
In this study, we found that miR-98 was significantly up-regulated in placental tissues from GDM patients compared with that in normal controls. The trend of miR-98 in four groups, including Y < 25, Y25~30, Y30~35 and Y > 35, is similar, suggesting that up-regulation of miR-98 may be associated with the occurrence of GDM.
Interestingly, DNA methylation inhibitor 5-azacytidine could decrease the miR-98 level in human choriocarcinoma cell line JEG-3. Emerging evidence indicates that GDM has epigenetic effects on genes through DNA methylation, with consequences on fetal growth and development 15 . So we speculate that miR-98 may be able to regulate DNA mathylation in GDM. In vitro cell experiment was used to confirm the relationship between miR-98 and DNA methylation, and found that over-expression of miR-98 increased global DNA methylational level, and It is generally accepted viewpoint that miRNAs function via regulating the expression of their downstream target genes. An online search of miR-98 targets by Targetscan, PicTar and miRanda found that there was a miR-98 responsive element in 3′ -UTR of Mecp2, which was a highly conserved domain among different species. When miR-98 mimic or inhibitor was co-transfected with the recombinant vector Mecp2-pGL3, luciferase activity was reduced by the miR-98 mimic and enhanced by the miR-98 inhibitor, suggesting that Mecp2 may be the target gene of miR-98. Mutation experiment further confirmed that the binding site in the 3′ -UTR of Mecp2 was specific for miR-98. Additionally, over-expression of miR-98 reduced the protein and mRNA level of Mecp2 and knockdown of miR-98 enhanced the protein and mRNA level of Mecp2. These data further confirm that miR-98 not only directly targets Mecp2, but also regulates the endogenous MECP2 expression.
Previous studies have identified that Trpc3 and Sfrp4 are Mecp2 target gene respectively 13,14 . Trpc3 is involved in vasoconstriction and regulation of blood pressure in metabolic syndrome 16 . Sfrp4 reduces insulin secretion and may be a potential biomarker for islet dysfunction in T2D 17 . We found that miR-98 over-expression suppressed Trpc3 expression, which was partially rehabilitated by up-regulation of Mecp2. MiR-98 knockdown promoted Trpc3 expression, which was partially attenuated by depleting Mecp2 expression. There was no significant effect of miR-98 on Sfrp4 expression. A study analyzing insulin-mediated glucose uptake shows that Trpc3 interacts functionally and physically with GLUT4, and Ca(2+ ) influx and modulates insulin-mediated glucose uptake 18 . Therefore, miR-98 may indirectly regulate glucose up-take through targeting Mecp2-Trpc3 pathway.
To our knowledge, this is the first study to examine the relationship between the presence of maternal GDM and miR-98. Our findings demonstrate that miR-98 and its pathways in placental tissues are associated with GDM. MiR-98 not only directly targets Mecp2, but also indirectly regulates the target gene of Mecp2. These results imply that enhanced miR-98 expression may take part in the occurrence of GDM by Mecp2-Trpc3 pathway. Plasmid construction and transfection. The Mecp2 3′ -UTR and Mecp2 3′ -UTR-mutant sequences were amplified by PCR from human genomic DNA using the primers in Table 2. After being double digested with SpeI and XbaI, the PCR products were cloned into pGL3 control vector (Invitrogen, Carlsbad, CA, USA). The coding region of Mecp2 sequence was amplified by RT-PCR from total mRNA of human JEG-3 cells using the primers in Table 2. After being double digested with BamHI and EcoRI, the PCR product was cloned into PMSCV-puro vector, designated as PMSCV-Mecp2. All the constructs were verified by DNA sequencing. Specific siRNAs for scramble and Mecp2 were synthesized as a duplex with the following sequence: scramble siRNA, 5′ -CUUCUUAGGUGGUUUCUGC-dTdT-3′ , Mecp2 siRNA, 5′ -GCAGAAACCACCUAAGAAG-dTdT-3′ .
In situ hybridization. An in situ hybridization procedure was performed using the protocol developed in our laboratory. Placentas were arranged into tissue microarray and cut to a thickness of 6 μ m. After deparaffinized in xylene, rehydrated in descending ethanol series, refixed in 4% PFA, deproteinized with 0.2 M HCl and digested with 20 μ g/ml proteinase K (Tiangen, China), the sections were prehybridized with hybridization buffer (Roche, Mannheim, Germany) at 40 °C for 1 h and then hybridized with digoxigenin (DIG)-labeled LNA-MiR-98 probe (LNA-MiR-98 sequence: 5′ -DIG-aAcaaTaCAaCttaCtAcCtCa-3′ ) overnight at 40 °C. After washed by descending saline-sodium citrate (SSC) and blocked by blocking buffer containing 5% bovine serum albumin (BSA), the sections were incubated with alkaline phosphatase (AP) labelled anti-DIG-antibody (Roche, Mannheim, Germany, 1:250) overnight at 4 °C, and developed with bromochloroindolyl phos-phate/nitro blue tetrazolium (BCIP/NBT; Promega, Madison, WI, USA). BCIP and NBT are the common substrate of alkaline phosphatase. Catalyzed by alkaline phosphatase, BCIP product will be hydrolyzed to produce a strong reactivity, and then the product will react with NBT and form an insoluble blue NBT-formazan. The probe was replaced by DIG-labeled LNA-scrambled probe (LNA-scrambled sequences: 5′ -DIG-caTtaAtgTcGgaCaaCtcAat-3′ ) as negative control. Samples were viewed by Nikon TE 2000-U microscope (NIKON, Tokyo, Japan).
Immunohistochemistry. Sections of the tissues microarray were deparaffinized in xylene and rehydrated in descending ethanol series. Antigen retrieval was accomplished with 1 N HCl. Slides were incubated with rabbit anti-MECP2 polyclonal antibody (GeneTex, USA, 1:500) and mouse anti-5 methylcytosine monoclonal

Quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR). Total RNAs from
tissues and cells were extracted using Trizol (Invitrogen, Carlsbad, CA, USA) accordint to the manufacturer's protocols. 1 μ g of total RNA was subjected to reverse transcription of mRNAs using dT18 as primer and reverse transcription kit (TakaRa Biotechnology (Dalian) Co., Ltd. Dalian, Liaoning, China) to generate total cDNA. Then the mRNA quantitative PCR was carried using primers in Table 2