The Role of Osteocalcin in Placenta Function in Gestational Diabetes Mellitus

Background: Evidence for osteocalcin role in glucose and energy metabolism is increasing. However, little is known about osteocalcin function in gestational diabetes mellitus. The aim of this study is to examine the associations between osteocalcin and gestational diabetes mellitus. Method: 36 patients with gestational diabetes mellitus and 40 normal glucose tolerance controls were recruited in the Maternal and Child Health Hospital Guangxi Zhuang Autonomous Region from May to August 2018. Total osteocalcin and biochemical indexes of maternal serum and umbilical vein serum were analyzed. Transcriptome of placenta were sequenced. Human trophoblast JAR cells were used for evaluated the affection of osteocalcin on trophoblast In vitro. There were no signicant differences with maternal serum total osteocalcin levels between gestational diabetes mellitus and normal glucose tolerance groups. The gestational diabetes mellitus group has lower umbilical vein serum total osteocalcin (51.46 ng/mL ± 24.29 Vs 67.00 ng/mL ± 25.33, P = 0.008 ), lower adiponectin (1099.72 μg/L ± 102.65 Vs 1235.85 μg/L ± 94.63, P < 0.001), higher leptin (7.41 μg/L ± 0.28 Vs 6.02 μg/L ± 0.31, P (cid:0) 0.001). A signicant relationship existed between umbilical vein serum total osteocalcin levels and leptin (r = -0.456, P = 0.007). Osteocalcin promote JAR trophoblast cells proliferation and HCG synthesis. 36 correlated gene modules of placental transcriptome were identied through weighted gene co-expression network analysis, 2 of them were associated with osteocalcin. Conclusion:


Background
Gestational diabetes mellitus (GDM) is de ned as glucose intolerance that is rst recognized during pregnancy. With the development of living level and the alteration of life style, the morbidity of gestational diabetes is about 30% worldwide [1]. Hyperglycemia leads to complications, such as macrosomia, gestational hypertension, and preterm delivery. Moreover, postpartum patients and GDM offspring often carry a lifelong increased risk of glucose intolerance and obesity [2,3].
Osteocalcin (OCN) is an osteoblast-speci c protein, Osteocalcin is the the most abundant noncollagenous proteins which constitute bone, un carboxylation OCN (ucOCN) released into the circulation, evident shows that ucOCN involved in the regulation of glucose and fat metabolism [4,5]. OCN could decrease serum glucose levels either directly by increases pancreas beta cells insulin secretion or indirectly by promoting glucose consume by liver, adipocyte, muscle and so on [6,7]. Both of these functions are mediated by the G-protein-coupled receptor GPR C6A [8][9][10], inducing the synthesis of adiponectin (ADP) in adipocytes, muscle and other tissues, then facilitating insulin sensitivity [4,5].
It has been con rmed that OCN serum levels are lower in individuals with type 2 diabetes than in normal controls [9,11,12]. Although GDM resembles type 2 diabetes in some ways, there is no consensus regarding the OCN levels in GDM. The results of some epidemiological investigations reveals that maternal OCN,especially the biological activity un carboxylation OCN (ucOCN) levels are signi cantly higher in GDM patients [13][14][15], others reveals that the OCN levels has no signi cant relation with GDM morbidity [11]. It is interesting that some research suggest that ADP levels are signi cantly lower in women with GDM and may help predict the risk of diabetes after GDM [16,17]. Since ADP is the down effect protein of OCN, and during pregnancy, the placenta is one of the major source organs of ADP, so we hypothesized that it couldn't exclude that OCN is involved in the occurrence of insulin resistance via the placenta. Therefore, the aim of this research is to examine the associations between OCN and GDM in GDM both in Maternal peripheral blood and umbilical cord blood. Pre-pregnancy body mass index (BMI) was calculated according to weight before pregnancy as weight (in kilograms)/height (in square meters). Birthweight, neonatal gender, delivery mode, placental size and weight, neonatal peripheral blood glucose levels were obtained from patient medical records.

Biochemical analysis
Maternal blood samples were taken from a cannulated vein before delivery, gestational diabetes mellitus umbilical vein (Uv) blood and umbilical artery (Ua) blood were taken from the umbilical cord venous and arterial storing in the vacuum blood collection tube during the third stage of labor. All blood samples were centrifuged (3000 rpm at room temperature for 5 minutes), and serum were collected in the 1.5 mL Eppendorf tubes. Placental tissue from the central cotyledon about 1 cm 3 was obtained immediately after delivery. Serum aliquots and placental tissues were immediately stored at -80℃ until analysis.
RNA extraction and qRT-PCR RNA extraction and puri cation were conducted using the E.Z.N.A.™ Total RNA Kit I (Omega). cDNA synthesis was performed using the PrimeScript™ RT Reagent Kit (Takara). SYBR Premix Ex Taq (Tli RNaseH Plus) was used with the following PCR parameters, 1 cycle of 30 s at 95 ℃ and then 40 cycles of 95 ℃ for 5 s and 1 cycle of 30s at 60 ℃. qRT-PCR was conducted using a LightCycler 96 (Roche). The primer sequences are presented in Tab. S2. The housekeeping gene β-actin was used as control. The relative levels of the mRNA of the genes of interest were normalized to the β-actin mRNA.

Immunohistochemistry
Placenta tissues were washed 3 times in PBS for 30 seconds and xed in 4% formaldehyde for 24 hours. The placental tissues were then embedded in para n and sectioned at a thickness of 5 μm. The tissue sections were depara nized, subjected to high-temperature antigen exposure, rehydrated in 3% H 2 O 2 , and blocked with 10% normal goat serum for 30 minutes. The sections were then incubated with antibodies to OCN (1:1000) (Santa cruz) or GPRC6A (1:1000) (Invitrogen). The secondary antibodies are provided with the DAB Substrate Kit for Peroxidase (Vector Laboratories). Olympus DP11 camera and Olympus Camedia software were used to produce the images.
Alizarin red stain: Alizarin red stain positive indicate calcium. JAR cells were maintained in 6 well plate until 90% con uency. Osteogenic induction medium was added con uency to 6 well plate to maintain 7 days, and then JAR cells were xed by 4% paraforma ldehyde for 10 minutes. Adding alizarin red for 5 minutes. At each of steps, JAR cells were washed for 3 times by PBS.
Scanning electron microscope: JAR cells were maintained in 24 well plate until 80%. Osteogenic induction medium was added to 24 well plate to maintain 7 days. 3% glutaraldehyde was used to xed JAR cells. The calcium nodules were observed by scanning electron microscope (Czech, TESCAN, VEGA 3 LMU).
Cell proliferation assay: Using the Cell Counting Kit-8 (CCK-8, Solarbio ® ) according to the manufacturer's instructions. This experiment was performed in triplicate.
RNA sequencing and bioinformatics analysis RNA sequencing: Total RNA was extracted from placental tissues using the E.Z.N.A. TM Total RNA Kit I (Omega). mRNA sequencing was performed on an Illumina HiSeq 4000 RNA-sequencing platform, and the results were uploaded to BaseSpace of the Illumina cloud server. The RNA-Seq Alignment and Cu inks Assembly procedures were used in the sequencing analysis.
WGCNA: Co-expression networks were constructed using the WGCNA (v1.47) package in R. Without ltering the genes, gene expression values were imported into WGCNA and used to construct coexpression modules using the automatic network construction function block wise modules with default settings except that the power was set to 2. TOM Type was unsigned, and the minimum module size was 250. The genes were clustered into 36 correlated modules. Module eigengenes were used to calculate the correlation coe cients for samples or sample traits.
Gene ontology (GO) analysis: signi cantly enriched GO terms pathways were de ned by the hypergeometric test using a threshold false discovery rate (FDR) ≤ 0.05.

Statistical analyses
Statistical analyses were undertaken using SPSS 25(IBM, Chicago, IL, USA). The results are reported as the mean ± S.D. unless otherwise noted. Clinical characteristics that followed a normal distribution were compared between the two groups using Student's t-test. Categorical variables were analyzed using the χ 2 test. Spearman's correlation was used to examine associations between serum OCN and metabolic indices. Statistical analysis and graph plotting were performed using Prism 7 software and Adobe Illustrator. P < 0.05 was considered signi cant.

Results
No signi cant differences with maternal serum tOCN levels between GDM and NGT groups in third trimester 36 GMD patients and 40 NGT control participants were recruited. Compared with the NGT controls, GDM patients had higher maternal age, pre-pregnancy BMI, systolic blood pressure, diastolic blood pressure, Spearman's correlation analysis is shown that Uv serum tOCN levels were negatively correlated with LEP and neonatal peripheral blood glucose (R = -0.456, P = 0.007 and R = -0.278, P = 0.026, respectively) and positively correlated with TNF-α (R = 0.395, P = 0.016). It is not signi cantly correlated with birth weight, birth length and chest circumference, or bone metabolism indexes 25(OH)D 3 and TRACP-5b (Tab. 2).

OCN exerts autocrine loop in trophoblast
Since there is no obvious correlation between Uv serum tOCN levels and fetal bone mass, we suspected that some of the OCN present in Uv blood may be produced by trophoblast cells. To test this prediction, rstly, the concentrations of total OCN in maternal peripheral blood and in Ua and Uv were measured in 13 healthy pregnant women, the serum sample was collected during pre-delivery or third stage of labor. The tOCN concentration in Uv serum was signi cantly higher than that in maternal serum and Ua serum (Fig. 1 a). Secondly, OCN and its receptor GPRC6A expression were veri ed by placenta immunohistochemistry staining, which were positive expression in syncytiotrophoblasts (N= Fig. 1b). Thirdly, the ability of trophoblasts to synthesis OCN was further investigated by JAR trophoblast cells human . 7 days after osteoblast induction, qRT-PCR analysis revealed that the bone formationrelated genes OCN, Dmp1, CollA1 were highly expressed in osteogenesis induces differentiated JAR cells (Fig. 1c). JAR cells showed the presence of mineralized nodules within the cells, indicated by Alizarin Red stained or transmission electron microscopic images staining (Fig. 1d).

OCN affected JAR trophoblast cells function
To determine the effects of OCN on JAR trophoblasts cells, 100 ng/ml OCN was added to the culture medium. OCN markedly promoted JAR cells proliferation and HCG synthesis. qRT-PCR analysis is shown that ADP expression is increased while LEP expression is decreased, meanwhile enzymes associated with steroid hormone metabolism such as CYP-11A1, CYP-19A1, CYP-27B1, HSD-17B1, and in ammatory factors IL-6, TNF-α were signi cantly increased than without AND (Fig. 1e-g).
GPRC6A is the receptor of OCN, and GPRC6A was knockdown in JAR trophoblast cells which was con rmed by qRT-PCR (Fig. S1). As expected, shRNAs targeting GPRC6A e ciently attenuated JAR trophoblast cells proliferation and HCG synthesis. LEP, ADP, enzymes associated with steroid hormone metabolism and in ammatory factors were also affected by GPRC6A shRNA (Fig. 1h-j). P<0.001 Placenta mRNA sequence analysis found 6 modules correlated with ADP and LEP mRNA sequence analysis was performed on 39 (GDM=18, NGT=21) placental samples. A total of 18329 genes were detected and selected for subsequent analysis. Co-expression modules were constructed using the WGCNA (v1.47) package in R. Scale-free networks were constructed in which the power value was equal to 2 (Fig. S2a). The grey module, which consisted of genes not belonging to any other module, was excluded, then 36 distinct gene co-expression modules were separated (Fig. S2b).
The enriched GO biological processes molecular function and cellular component were used to show the gene function of two correlations darkmagenta and darkturquoise module. The genes in the darkturquoise module appear to be enriched in glucagon receptor activity, brown fat cell differentiation. darkmagenta module corresponded to muscle system process, cAMP-dependent protein kinase regulator activity (Fig. S3).

Discussion
We found no differences in maternal serum OCN in GDM and NGT groups, in fact, the levels of OCN Uv serum were lower in the GDM group than those in NGT group, these result was consisted with previous research, which demonstrated that in the third trimester or predelivery OCN has no signi cant difference between GDM and control group [11,13]. What we go further than these reports are that we analyze the relationship between OCN and biochemical indexes in Uv serum. Our result suggest that the OCN levels in Uv serum were lower in GDM patient. Uv OCN was negatively correlated with Uv LEP and fetal blood glucose. Based on these founding we guess that maybe the placenta could synthesis OCN by itself, further research reveal that Uv serum concentrations of OCN are higher than that in Ua serum and maternal peripheral blood plasma concentrations, suggesting that the steroidal environment of the placenta differs signi cantly from that in maternal peripheral blood.
By placenta immunohistochemical staining and in vitro JAR trophoblast cells experiment, we identi ed a potential OCN autocrine feed-forward loop in the placenta. OCN is the most abundant non-collagenous protein found in the skeleton and was previously thought to be expressed only in mature osteoblasts. However, there is increasing evidence regarding the ectopic production OCN by other organs. This is not the rst report about that OCN expression was found in placenta tissue. Telejko had reported that OCN mRNA were detected in placenta tissue [18]. Moreover, it has been reported that OCN is expressed in adipose tissue [18][19][20], tendon sheath tissues [21], endothelial progenitor cells [22], and prostate cancer [23] and so on. The function of ectopic production OCN in placenta is worth further investigation.
It is believed that OCN facilitates insulin sensitivity by inducing the expression of ADP through the GPRC6A receptor [4,24]. ADP is a well-known gene involved in glucose metabolism regulation, which promotes sugar uptake and reduces gluconeogenesis in liver and muscle, resulting in reduced insulin resistance and promoting insulin sensitivity [25,26]. The level of insulin resistance during pregnancy is closely associated with proteins secreted by the placenta, such as ADP, in ammatory cytokines and so on [27]. A consensus has been reached that lower ADP levels are signi cantly correlated with insulin secretion and IR indices in GDM [28,29]. However, it is unclear that OCN plays a role in glucose homeostasis through placenta or mediated through ADP during pregnancy.
In this research we reported that OCN synthesis by trophoblast, in vitro JAR trophoblast cells experiment proved that OCN favorable ADP, sex hormone, adipocyte cytokine production. mRNA sequencing of placenta and subsequent WGCNA analysis was performed for further investigate the OCN function in the placenta. It is interesting that 2 of 36 module show correlations with osteocalcin, including darkmagenta and darkturquoise. Hub genes were identi ed in these two modules. Some of these genes have been reported to be associated with diabetes or lipid metabolism, such as MAGI1 [30], ACTC1 [31], ACTA1 [32]. Overall, the result suggested that repress OCN in placenta regulated trophoblast function through ADP. Although some of these genes have been reported to be associated with glucolipid metabolism, their importance in placenta metabolic programming and the relationship with OCN and ADP require further investigation.
The relationship between Uv OCN and neonatal growth index, 25(OH)D 3 were not found out. Uv OCN is negative associated with neonatal peripheral blood glucose. Importantly, LEP is negative correlated with Uv OCN. LEP is mainly secreted by placenta regulating the insulin sensitivity, insulin secretion as well as glucose and lipid metabolism [33]. The level of LEP has an important role in bone metabolism. Previous study report that LEP de cient ob / ob and LEP resistant db / db can increase osteocalcin levels [34]. However, other research shows that LEP increase osteoblast-speci c osteocalcin [35]. One explanation may be the different sources of LEP. Uv LEP may mainly secreted by the placenta to satisfy materal and fetal growth during pregnancy.
In this research, the Uv OCN is positively correlated with TNF-α, and GPRC6A knockdown JAR cells model was stimulated by OCN increased secreting IL-6 and TNF-α. The role of OCN in in ammation in human beings is still unconclusive. The previous study suggest that OCN improves whole -body insulin resistance by decreasing in ammation, and increasing insulin signaling and the expression of Slc2a4/GLUT4 [36]. Moreover, OCN could be a therapeutic target for protecting against chronic in ammation in T2D [37].

Limitations Of The Study
The limitations of this study are as follows. 1) This is a study of a small sample of women in southern China. Although is neonate birth weight and Maternal peripheral blood glucose higher in GDM group, but there was no statistic difference, which may be the sample size is too small, still need a large number of high quality. 2) OCN has two molecular forms, carboxylated or ucOCN, it is ucOCN appears linked endocrine function. Although tOCN may re ect indirectly ucOCN levels, it is still very regret that only tOCN, not speci cally ucOCN was measured in this research. 3) This is a cross-sectional study, the ndings reported here do not necessarily re ect causal associations. 4) OCN were strongly expressed in syncytiotrophoblasts by immunohistochemistry, however, for veri cation the trophoblast OCN synthesis ability, we use Jar cells, which are cytotrophoblasts.

Conclusion
In the third trimester, lower osteocalcin concentration of umbilical venous serum is associated with gestational diabetes mellitus. Osteocalcin exerts autocrine loop in trophoblast may regulate placenta function via adiponectin.   Adjusted for the BMI and age. Figure 1 Osteocalcin exerts an autocrine loop in trophoblast. a: OCN umbilical vein serum levels were signi cantly higher than those in umbilical arterial serum and maternal serum. b: Immunohistochemical staining of placental tissue for OCN and GPRC6A. Seven days after JAR trophoblast cells osteogenic differentiation. gene knockdown by shRNA in JAR trophoblast cells, decrease (h) proliferation, (i) HCG secretion, (j)aromatase-related genes and genes encoding in ammatory factors synthesis. The data are presented as the mean ± SE of the values obtained in 3 independent experiments. * P < 0.05, ** P < 0.01, ***P < 0.001). OCN, osteocalcin; HCG: Human Chorionic Gonadotropin; ADP, adiponectin.  Schematic diagram of OCN autocrine of trophoblast in GDM. Lower trophoblasts synthesis OCN contributes to placenta dysfunction in GDM, Lower OCN resulted in a decrease in HCG, In the same way, IL-6, TNF-α, and ADP levels were reduced, but LEP increased.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.