Hyperglycemia results in significant pathophysiological changes of placental spiral artery remodeling and angiogenesis, further contributing to congenital defects

1. Abstract 2. Introduction 3. Methods 3.1 Chemicals, antibodies, and reagents 3.2 Mouse model of gestational diabetes 3.3 Histology of mouse placenta 3.4 Spiral artery remodeling analysis 3.5 Fetal capillary area (FCA), maternal lacune area (MLA) and IHM thickness 3.6 Western blot 3.7 Realtime-qPCR 3.8 Statistical analysis 4. Results 4.1 STZ injection induced hyperglycemia at E3.5 4.2 Hyperglycemia increased placental and embryonic weights 4.3 Hyperglycemia increased the percentage of glycogen trophoblast (GlyT) cells in the JZ 4.4 SpA remodeling is compromised under hyperglycemia 4.5 Hyperglycemia increases the IHM thickness 4.6 GDM causes congenital malformations and hyperplastic ventricular disease 4.7 The IHM thickness was marginally increased in the placentas from embryos with congenital defects 5. Discussion 6. Conclusions 7. Author contributions 8. Ethics approval and consent to participate 9. Acknowledgment 10. Funding 11. Conflict of interest 12. In-house submissions 13. References


Introduction:
Hyperglycemic conditions achieved during pregnancy have been shown to have detrimental effects to fetal development and increase the prevalence of childhood comorbidities. However, the mechanisms in which diabetic pregnancies affect placental development and subsequently contribute to adverse health effects on the mother and offspring remain unclear. Research design and methods: Streptozotocin was used to induce gestational diabetes in mice. In this model, hyperglycemia was established at embryonic day 3.5 (E3.5). Pregnancy mass was collected at E10.5, E12.5, E14.5, and E16.5 for different assessments. Results: Both placental and embryonic weights were found to be significantly elevated at E16.5. At E14.5, a significantly larger junctional zone with increased number of glycogen trophoblasts was found in the placentas from hyperglycemic pregnancies (HG group) compared to the placentas from normoglycemic pregnancies (NG group). Importantly, the HG placenta exhibited decreased trophoblast giant cell (TGC) association and TUNEL+ cells, and increased expression of α-SMA on the spiral artery, suggesting arterial remodeling was impacted. Moreover, the interhemal membrane of the labyrinth layer, was found to be thicker in the HG placentas. Furthermore, hyperglycemia resulted in more offspring congenital defects, which were associated with a thicker interhemal membrane. Conclusions: Together, these results suggest that gestational diabetes perturbs proper placental development and function, specifically spiral artery remodeling and angiogenesis, thereby negatively impacting embryonic development.

Introduction
Diabetes during pregnancy is known to affect the health of both mothers and their infants. Among the two types of diabetes during pregnancy, gestational diabetes (GDM) accounts for 90% and pre-gestational diabetes (PGDM) comprises the remaining 10% [1,2]. In the United States, the GDM prevalence among pregnant women is as high as 9.2% [3]. With the introduction of insulin, diabeticassociated fetal mortality rates are reduced from 70% to nearly 12%. Unfortunately, the present birth defect rate in diabetic pregnancies (~10%) is still higher than that of the general population (3%), and appears to be ever increasing [2][3][4][5][6]. It has been suggested that placental damages from diabetic pregnancies could be the cause of fetal complications observed in the human population [7].
For a successful and healthy pregnancy, proper spiral artery (SpA) remodeling and labyrinth (LZ) angiogenesis are crucial for the establishment of adequate blood supply. Maternal blood is delivered to the placenta through the maternal SpAs, which reside in the endometrium of the uterus. During early stages of pregnancy, SpAs are classified as high resistance vessels and are surrounded by vascular smooth muscle cells (VSMCs). In order to provide adequate blood to the developing fetus, the maternal SpAs will undergo an invasive remodeling process, in which the VSMCs, possessing contractile properties, are replaced with invasive trophoblast cells [8], resulting in dilated and low resistance arteries [9]. As the intermediate barrier between fetal and maternal circulations, the placenta contains a highly developed blood network in the LZ layer. Different transporters are localized on the interhemal membrane (IHM), which separates the maternal and fetal blood. These establish a vascular network that will allow for the efficient exchange of nutrients and waste [8,10,11].
There is emerging evidence showing that placental malformation is a complication of diabetic pregnancies [12]. Hyperglycemic placentas are characterized by vil-lous immaturity and edema, syncytial nodes, fibrin thrombus, and fibrinoid necrosis [7]. In addition, placentas from diabetic pregnancies have been reported to undergo "premature aging", which is detected by early emergence of placental lobe calcifications [7]. All these problems disrupt placental vasculature and result in inefficient exchange of nutrients and waste [13]. Therefore, it has been speculated that transport efficiency can be impacted by a diabetic pregnancy. Additionally, shallow remodeling of spiral arteries have been observed under gestational diabetic and preeclamptic conditions [14,15]. Despite the known pathological changes in hyperglycemic placentas, details of how this condition affects the process of SpA remodeling and angiogenesis remain unclear.
To gain a detailed insight into the pathology and underlying mechanisms of the placenta under hyperglycemic conditions, a streptozotocin (STZ)-induced hyperglycemic mouse model of pregnancy was employed. The pathological changes of the placenta due to hyperglycemia were evaluated, followed by experiments to investigate the underlying cellular and molecular mechanisms. Additionally, the incidences of several congenital birth defects, caused by the diabetic pregnancy, were also reported.

Mouse model of gestational diabetes
Wildtype (WT), C57BL/6J female mice, between the ages of 8-11 weeks, were maintained with a chow diet (9% fat) and water ad libitum under a 12 h light/dark cycle throughout the experiment. Prior to treatment, mice body weights, basal blood glucose levels and age were recorded. The mice were then randomly selected and administered STZ, prepared in citrate buffer (pH 7.4), at a dose of 100 mg/kg body weight (HG), or citrate buffer only (NG) on day 1 and day 4, intraperitoneally ( Fig. 1A). After the second injection, the female mice were mated with male mice (C57BL/6J) overnight, and the following day at noon was considered as embryonic day 0.5 (E0.5). Random blood glucoses were measured daily after the first injection to identify the initial onset of diabetic conditions. Maternal blood glucose levels higher than 200 mg/dL were defined as a hyperglycemic pregnancy. Embryos and placentas were collected at E10.5, E12.5, E14.5 and E16.5.
All mouse experiments were completed according to a protocol reviewed and approved by the Institutional Animal Care and Use Committee of Texas A&M University, in compliance with the USA Public Health Service Policy on Humane Care and Use of Laboratory Animals.

Histology of mouse placenta
Placentas were sagittally bisected. Half of the placenta was stored at -80 • C, while the embryos and the remaining halves were fixed in 10% formalin at room temper-ature. The formalin fixed embryos and placentas were processed through an ethanol-xylene series and embedded in paraffin. 5 µm thick sections were collected. Routine procedures were used for H&E staining, periodic acid-Schiff (PAS) staining, immunohistochemical (IHC) staining and immunofluorescent (IF) staining. Isotype IgG was used for immunostaining negative control (Supplementary Fig. 1).

Spiral artery remodeling analysis
SpA remodeling during murine pregnancy involves the replacement of VSMCs by SpA-TGCs to form high circulation, low resistance blood vessels. A well-

Acta2
GGCATCCACGAAACCACCTA TTGCGTTCTGGAGGGGCAA remodeled SpA is surrounded by SpA-TGCs, with a dilated lumen to support an increased blood volume during pregnancy. The remodeling process was assessed by the level of TGCs associated with the SpAs at E14.5. Based on the percentage of the arterial wall surrounded by TGCs (Cytokeratin 18 (Cytok) positive), all observed SpAs were categorized as either unremodeled (0-25%), partially remodeled (25-75%), or remodeled (75-100%). Based on the ratio of the remodeled SpAs of each placenta, placentas were further classified as "TGC-associated" if more than 50% of the spiral arteries were remodeled or "TGC-unassociated" if less than 50% of the spiral arteries were remodeled. In addition, the lumen diameter (LD) to outer diameter (OD) ratio of the SpA was measured to determine the lumen size of the SpA. RT-PCR was performed to measure the mRNA expression of TGC migration markers.

Fetal capillary area (FCA), maternal lacune area (MLA) and IHM thickness
The LZ angiogenesis was assessed by measuring the area of the maternal lacunae space, traced by Cytok expression, and the fetal blood space, labeled by CD31 positive staining, at both E14.5 and E16.5. Images of the LZ from CD31 and Cytok co-IF staining were captured with a Leica confocal microscope. In each sample, outlines of fetal capillary and maternal lacune areas in 10,000 µm 2 of the LZ layer were drawn based on CD31 staining (green) and Cytok staining (red), respectively, and the total areas were measured. The barrier between the fetal and maternal circulations, composed of endothelial cells, syncytiotrophoblasts and sinusoidal trophoblast giant cells, is defined as the IHM, and is an important structure for efficient nutrient and gas exchange [16]. To investigate placental nutrient transportation capacity, the IHM thickness was measured. Ten IHMs were randomly selected in each picture and the thinnest portion of the membrane between the fetal capillary and maternal lacunae was measured. All analyses were performed with ImageJ.

Western blot
The total protein from the placentas was extracted using Cell Lysis buffer (Cell Signaling Technology, Cat#: 9803). The concentrations were measured using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, Cat#: 23225). 15-30 µg proteins were loaded on 7-12% SDS-PAGE electrophoresis gels and transferred onto PVDF membranes (Fisher Scientific, Hampton, NH, Cat#: 45-004-110). The membranes were then blocked with 5% fat-free milk, and incubated in the primary antibodies overnight at 4 • C. After incubation with the anti-rabbit secondary antibody (Cell Signaling Technology, Cat#: 7040) for 1 h, the signal was detected with the HRP substrate, and analyzed with the ImageJ.

Realtime-qPCR
The total RNA of mouse placentas was extracted using Trizol and RNA spin columns. cDNA was then synthesized using the SuperScript III Reverse Transcriptase kit (Thermo Fisher, Waltham, MA, Cat#: 18080051). RT-PCR was performed using a SYBR Green PCR master mix (Bio-Rad). Results were analyzed using the comparative C (T) method. The primers are listed in Table 1.

Statistical analysis
Embryos and placentas from at least 5 female mice per group were analyzed in this study. Data was analyzed by the student's t-test, Chi-square test, or one-way ANOVA. Results were considered significant when the p values were less than 0.05. All statistical analyses were carried out using GraphPad Prism.

STZ injection induced hyperglycemia at E3.5
Before pregnancy, female mice demonstrated synonymous age and body weight (Fig. 1B,C). The blood glucose of the HG and NG groups were normal (Fig. 1D). Daily random blood glucose checks identified a significant increase of blood glucose levels in the HG group beginning at E2.5. At E3.5, 80% (8 out of 10) of the HG mice reached a hyperglycemic condition (>200 mg/dL), displaying blood glucose levels of 289 ± 123.9 mg/dL. By E4.5, all of the HG mice had obtained this hyperglycemic condition, which persisted throughout the pregnancy. At E14.5, the blood glucose levels of the HG group were as high as 555.7 ± 32.2 mg/dL, while the NG group remained at 160.3 ± 15.6 mg/dL (Fig. 1E). To be noted, the hyperglycemic status did not affect the litter size in this study (Fig. 1F).

Hyperglycemia increased placental and embryonic weights
Increased placental weights were observed in the HG group beginning at E12.5 and reached significance at E16.5 (Fig. 1G). The HG group also showed a significantly larger embryonic weight at E14.5 and E16.5 (Fig. 1H). The placental efficiency of the HG placentas was lower at E12.5, but this observation was reversed at E16.5 (Supplementary Fig. 1).

Hyperglycemia increased the percentage of glycogen trophoblast (GlyT) cells in the JZ
The effect of GDM on placenta morphology was assessed at E14.5. An increased thickness in the junctional zone (JZ) was observed in the HG placentas ( Fig. 2A, white  arrow). Additionally, the HG placenta displayed a larger portion of the whole placenta to be occupied by the JZ and the LZ to take up a smaller portion, when compared to the NG placenta (Fig. 2B). This resulted in a significantly smaller LZ to JZ area ratio in HG placentas (Fig. 2C). PAS staining was performed to specifically label the glycogen trophoblast cells (GlyTs), which are enriched with glycogen (Fig. 2D). Thus, the PAS negative cell population was majorly composed of spongiotrophoblast cells (SpTs). The PAS staining showed no difference in the SpT area ratios in the JZ area of the HG and NG groups. However, a larger ratio of GlyT area was noticed in the HG group (Fig. 2F). In addition, the total area of the venous sinus was significantly decreased in the JZ of the HG group (Fig. 2G). Tpbpɑ staining (Fig. 2H) showed that the number of SpTs is not changed, while the GlyT cell count increased significantly (Fig. 2J,K). Western blots performed on the whole placenta showed that hyperglycemia enhanced the level of Tpbpɑ (Fig. 2I), a protein highly expressed in GlyT, which is consistent with the increased total area of GlyTs.

SpA remodeling is compromised under hyperglycemia
The decidua (DZ) layer was found to have fewer TGCs, marked by Cytok expression (Cytok+), surrounding the SpAs (CD31+) in the HG placentas (Fig. 3A). When all TGC associations to SpAs were evaluated together, the HG placenta presented more unremodeled SpAs, and less partially and fully remodeled SpAs (Fig. 3B). As shown in Table 2, 9 out of the 10 NG placentas were graded as "TGCassociated", while the HG group only had 3 out of 9 TGCassociated placentas. The placentas from the HG group also had a significantly smaller LD/OD ratio (Fig. 3C,D). The SpA remodeling level was further evaluated by the expression of ɑ-SMA, a marker of VSMCs that surround the SpA. The HG placenta showed a higher ɑ-SMA expression on the SpA (Fig. 3E), and less apoptotic cells surrounding the SpAs (Fig. 3F), suggesting hindered remodeling. Consis- tently, the mRNA expression level of Acta2, which is an alias for ɑ-SMA [17], was found to be significantly higher in the HG placentas at E12.5 (Fig. 3G), suggesting an increased presence of VSMCs, hence decreased remodeling by TGCs.
The reduced number of SpA-TGCs and discrepancies observed in the SpA wall suggested a migration problem of the TGCs. Thus, the influence of GDM on the expression of genes involved in TGC migration was then examined. Mmp2, Mmp9 and Mmp14 are pro-migratory proteins, while Tgfβ is anti-migratory [10,18]. Wnt5a is known to induce cell migration [19,20]. A significant decrease in Mmp2, Mmp9, Mmp14, and Wnt5a mRNA transcripts, and an increase in the Tgfβ levels were observed in the HG placentas (Fig. 3G).

Hyperglycemia increases the IHM thickness
The HG group had a decreased trend of fetal blood space and the maternal lacunae area at both E14.5 and E16.5, however neither showed statistical significance (Fig. 4A-C). The mean of the maternal lacunae and fetal capillary space also did not show any differences between the HG and NG groups (Fig. 4D). The HG placentas were found to display a thicker IHM compared to the NG placentas at E16.5 (Fig. 4E).

GDM causes congenital malformations and hyperplastic ventricular disease
To gain an insight of how hyperglycemia impacts embryonic development, embryos were harvested at E14.5.  The number of congenital defects, including the neural tube defects (NTD), ocular defects, craniofacial defects, and heart defects, were counted (Fig. 5A,B). 11 out of 73 of the HG embryos displayed at least one type of the evaluated birth defects, while no birth defects were observed in the NG group (Table 3, p < 0.001 by Chi-square test). These congenital defects included 6 cases of NTDs, 3 cases of ocular defects and 1 case of a craniofacial defect (Fig. 5A,B). Statistically, the HG group had significantly more incidences of NTDs than the NG group (  (Table 3). Embryonic hearts were further examined for the interventricular septum (IVS) thickness and the presence of cardiac hypertrophy or hyperplasia. Overall, dense ventricles and increased ventricular weight of the embryonic heart were observed in the HG group compared to the NG group (Fig. 5C,D). Specifically, a thicker IVS was observed in the HG embryo (Fig. 5C,E). The HG embryonic heart also had an increased myocardial area of both the left and right ventricle (Fig. 5C,F). IF staining of p-H3S10 was performed on heart sections at E12.5 to determine if proliferation of ventricular cardiomyocytes was affected by hyperglycemia. A significant increase in the percentage of proliferating cardiomyocytes, labeled by p-H3S10 expression, was found in the HG group, when compared to the NG group (Fig. 5G,H). A significant increase in Cdk6, Cyclin A, Cyclin B2 and Cyclin D2 mRNA transcripts was observed in the HG heart ventricle (Fig. 5I). These results suggested cardiac hyperplasia in the HG embryos.

The IHM thickness was marginally increased in the placentas from embryos with congenital defects
To understand if the congenital birth defects were associated with any abnormal feature of the placenta, placental samples with congenital defects, regardless of them being in the NG or HG group, were pooled in a "Congenital Defect" group, while the rest were grouped as "Normal". One-way ANOVA was performed on the data of GlyT/SpT ratio, the LD/OD ratio and the IHM thickness. A Chi-square test was performed on the TGC-SpA association data in terms of congenital defect group versus the normal group. The results showed a marginal significance of IHM thickness and the congenital defects (p = 0.085), but not for the GlyT/SpT ratio (p = 0.200), the LD/OD ratio (p = 0.207) and the TGC-SpA association (p = 0.222) (Fig. 5J-L).

Discussion
In this study, the pathological changes of the placenta and its association with birth defects caused by hyperglycemia was investigated using a diabetic pregnant mouse model. Compared to other mouse studies that generated hyperglycemia before mating [21][22][23], this study utilized a mouse model with hyperglycemia established at E3.5. Prepregnancy diabetes has been known to affect ovulation and implantation [5]. Therefore, this model avoids placental abnormalities associated with pregnancy establishment. Embryos and placentas from hyperglycemic pregnancies were reported to have altered growth metrics. Additionally, placentas under hyperglycemic conditions were found to have a proportionately larger JZ, mainly due to an increased number of GlyTs. Most important, the hyperglycemic placentas displayed defects in SpA remodeling and IHM thickness. Interestingly, IHM thickness appeared to be associated with a higher incidence of birth defects in the embryos.
This study provides in vivo evidence of hyperglycemia disrupting SpA remodeling. This is indicated by decreased TGC association and higher remnants of αSMA on the vascular wall. These results suggest two potential mechanisms leading to hindered SpA remodeling. Firstly, migration alterations of the TGCs delays the process of TGCs replacing the VSMCs. Nteeba et al. [24] reported that STZ-induced hyperglycemic rat placentas had decreased interstitial trophoblast migration into the metrial gland. Consistent with this report, this study showed that hyperglycemic placentas had reduced Mmp2, Mmp9 and Mmp14 mRNA levels and increased Tgfβ mRNA levels, suggesting an inhibition of migratory signaling. Secondly, inhibition of VSMC apoptosis complicates the ability of the TGCs to replace the VSMCs. Through TUNEL staining, a decrease in VSMC apoptosis in HG placentas was observed. Indeed, decreased placental growth factor (PIGF) and uPA activation were observed previously in diabetic placentas, suggesting a potential decrease in apoptosis during SpA remodeling [25][26][27]. Furthermore, defective SpA remodeling has been suggested to result in blood pressure changes [28][29][30][31]. In humans, up to 50% of diabetic pregnancies were reported to develop pregnancy-induced hypertension [29][30][31]. Thus, GDM is considered a major risk factor for preeclampsia. Not only did our mouse GDM model reveal that SpA remodeling was hindered, but also showed a decrease in JZ venous sinus area. There are currently no studies to our knowledge that have examined the physiological consequences of decreased venous sinus in the placenta. However, in the brain, decreased venous sinus causes reduced venous flow, resulting in pressure buildup [32]. Taken together, this data provides a potential pathophysiological link of GDM and the etiology of preeclampsia.
LZ layer angiogenesis is critical for establishing proper blood supply to support a healthy pregnancy. Phenotypic analyses revealed that the IHM was thicker in the hyperglycemic group, consistent with observed thickening of the basement membrane in human GDM placentas [7,33]. The cause of the thickened IHM is still unclear. One potential mechanism of this phenomenon is an extracellular matrix buildup due to an altered microenvironment accompanying inflammation [34]. It is also possible that premature trophoblasts observed in GDM placentas may alter the thickness of their cell membranes, which further affects the IHM [35,36]. Additionally, human GDM placentas often show hypervascularization and villous immaturity [37]. However, these phenotypes were not observed in this murine study. A previous review suggested that GDM affects placental angiogenesis in later stages of pregnancy [38]. Therefore, future studies using later stages of placen-tal development, e.g., E16.5 or E18.5, may elucidate the mechanisms behind hyperglycemic induced temporal dysregulation of placental angiogenesis.
It is suggested that the JZ is the most sensitive placental compartment to diabetic conditions [39][40][41][42]. Consistent with previous reports [40], hyperglycemic conditions caused the relative size of the JZ to increase in this study. The Tpbpɑ IF staining showed that the larger JZ area was mainly due to increased GlyTs and not SpTs. It has previously been reported that hyperglycemia enhances expression of genes involved in JZ development, such as Taf7I, Pappa2, and Plac1 in the rat placenta [24]. Together, this data, along with other reports, suggest that hyperglycemia impacts the JZ development, particularly the cell population of this layer.
The mouse hyperglycemic pregnancies displayed an increased embryonic weight over time. Human GDM pregnancies are often complicated by macrosomia due to accelerated fetal growth in the later pregnancy [43]. In this GDM mouse model, a similar phenotype was observed, in which embryonic weight began to increase at E14.5 and was significantly higher at E16.5. Among the embryos that were collected at E14.5, the rate of congenital malformations, especially the NTDs, was increased in hyperglycemic pregnancies, which is a consistent finding in fetuses of diabetic pregnancies [44,45]. In this study, congenital heart defects were not associated with diabetic pregnancies, which seems to be inconsistent with previous reports in humans [46]. This could be due to a small sample size not producing enough power for this association study, since the incidence rate of congenital heart defects in diabetic pregnancy was around 3% [47,48]. However, hyperplastic ventricles due to increased cell proliferation, with a thickened IVS were observed in our hyperglycemic pregnant mouse model, which was consistent with other reports [45,49]. It has been suggested that placental damages from GDM could be the cause of fetal complications observed in the human population [7]. Moreover, for the first time, this study demonstrated an association between IHM thickness and birth defects, although the same was not observed with GlyT/JZ nor LD/OD. The IHM is where exchange of gas and nutrients occur via diffusion, transporters, and endocytosis/exocytosis [50]. Theoretically, the transport of nutrients, such as sugars, amino acids, and fatty acids, will be affected when its travel distance across the maternal and fetal interface is relatively longer, as the IHM remains thicker under hyperglycemia. This altered availability, either in rate or amount, of these macro-nutrients during development could have an impact on crucial developmental milestones. Further research is needed to confirm this hypothesis.

Conclusions
In summary, we demonstrated that hyperglycemia disrupted SpA remodeling possibly via interference of SpA-TGCs migration. Placentas from hyperglycemic pregnancies had thickened IHM, which was associated with increased birth defects. This study provides insights into the placental changes that occur under hyperglycemic pregnancies in vivo. These observed morphological and molecular changes should be further explored to identify more focused targets affected by hyperglycemia to establish strategies to mitigate congenital malformations and pregnancy complications associated with GDM.

Author contributions
LX conceived the idea and designed the study; YQ, NM, ZD and ZL performed the experiments; KZ, YQ, NM, ZD and LL analyzed the data; LX, YQ, NM, ZD and LL wrote the manuscript.

Ethics approval and consent to participate
All mouse experiments were completed according to a protocol (AUP#2018-0309) reviewed and approved by the Institutional Animal Care and Use Committee of Texas A&M University, in compliance with the USA Public Health Service Policy on Humane Care and Use of Laboratory Animals.

Acknowledgment
Thanks to all the peer reviewers for their opinions and suggestions.

Funding
This project was supported by grants from the National Institutes of Health (NIDDK 1R01DK112368-01 to Drs. Xie and Zhang). This work was funded, in part, by the grant from the National Institute of Environmental Health Sciences (P30 ES029067).

Conflict of interest
The authors declare no conflict of interest.

In-house submissions
Given her role as Guest Editor in special issue: The Effects of Maternal Nutrition on Metabolism of Infants and Children, LX had no involvement in the peer-review of this article and has no access to information regarding its peer-review. Full responsibility for the editorial process for this article was delegated to GP.