The Choline Metabolite TMAO Inhibits NETosis and Promotes Placental Development in GDM of Humans and Mice

Gestational diabetes mellitus (GDM) is one of the most common complications during pregnancy. The report released by the International Diabetes Federation shows that approximately 20 million live births are affected by hyperglycemia during pregnancy, of which 84% are related to GDM (1). Because the extent of glucose tolerance impairment and extent of overweight/obesity have been significantly increased in the offspring of mothers with GDM (2–5), exploring the mechanism by which GDM affects offspring is necessary.

Trimethylamine N-oxide (TMAO) is a choline metabolite. Most of the trimethylamine produced by choline catalysis in intestinal bacteria is absorbed into the blood and then rapidly oxidized to TMAO by flavin-containing monooxygenase 3 in the liver. Red meat, eggs, dairy products, and saltwater fish are rich in choline, lecithin, and l-carnitine, and all are potential food sources of TMAO (6). TMAO levels are higher in people with diabetes, and increased glucose tolerance impairment, disturbance of liver insulin signaling pathways, and increased adipose tissue inflammation have been observed in mice receiving TMAO supplementation (7–10). More interestingly, recent clinical studies have found that TMAO is associated with GDM risk and that the TMAO concentrations in the peripheral and cord blood of women with GDM are significantly increased (11–13). Changes in the intrauterine environment are closely related to the development of fetuses, and whether TMAO contributes to the development of fetuses is an important question worthy of research.

Neutrophil extracellular traps (NETs) are reticular structures released by neutrophils and can trap and kill invasive pathogens. They are comprised of double-stranded DNA (dsDNA) and antibacterial proteins embedded in the dsDNA backbone (14). However, excessive or dysregulated NETs formation is associated with many inflammatory diseases. Diabetes is associated with chronic low-grade inflammation, mainly caused by activation of the innate immune system, and circulating dsDNA can overactivate the innate immune system, resulting in a vicious cycle of inflammation-induced DNA damage and DNA-induced inflammation (15,16). Existing studies have shown that neutrophils in a high glucose environment are more likely to release NETs spontaneously (17–19). However, the role and significance of NETs in GDM have not been clearly elucidated. The placenta is the most direct organ that connects mothers and fetuses and has become the target of pregnancy pathology. Exploring the effects and mechanisms of NETs on the placenta will help provide new evidence for adverse pregnancy outcomes caused by GDM. In this study, we focused on the effect of TMAO and NETs on the placenta and revealed their roles in GDM.

Relevant studies from PubMed, China National Knowledge Infrastructure (CNKI), VIP Journal Integration Platform (VJIP), and WANFANG MED were searched through December 2018. The inclusion criteria were as follows: 1) case-control studies and language limited to English and Chinese, 2) plasma choline levels as the exposure factors, and 3) diabetes as the outcome of interest. Pooled means and SDs were calculated to evaluate the association between plasma choline levels and diabetes risk. If only the medians and interquartile ranges were provided in the original studies, the method established by Wan et al. (20) was used to convert the values to means and SDs. The Cochran Q and I² statistics were used to evaluate between-study heterogeneity, and random-effects models were used with P < 0.10 or I² statistics ≥50%. All of the above statistical analyses were conducted with Stata, version 12.0 (StataCorp, College Station, TX).

This study was approved by the ethics committee of The First Affiliated Hospital of Chongqing Medical University (2020-565), and informed consent was obtained from all subjects. Two designs were used in the current study. The first design included 16 women with GDM and 17 age-matched control subjects at the time of cesarean section. The cord blood that they provided was used to detect the content of TMAO, dsDNA, and myeloperoxidase (MPO), and the placental tissue was used to detect the expression of MPO. The inclusion criteria of the participants were as follows: age ≥20 years, firstborn single pregnancy, no previous diagnosis of diabetes, no history of medications affecting glucose metabolism, no infectious diseases, and no clinically significant neurological, endocrinological, or other systemic disease. Another design included 20 women with GDM and 20 age-matched control subjects, who provided fasting blood samples to detect the content of TMAO, dsDNA, and MPO during the oral glucose tolerance test at 24–28 weeks and were followed up until delivery. The placental thickness refers to the maximum thickness of the placenta from the maternal side to the fetal side measured vertically by type-B ultrasonic before delivery, and this value can reflect the size of the placenta. The newborn weight refers to the value obtained by weighing of the newborn after delivery. All participants were recruited from The First Affiliated Hospital of Chongqing Medical University.

Peripheral blood samples (at 24th–28th week of gestation) and cord blood samples (at delivery) were collected into EDTA anticoagulation tubes, and the plasma was immediately stored at −80°C until analysis. Placental tissues (at delivery) were collected from both the maternal side and the fetal side, immediately immersed in 4% formaldehyde solution, fixed at 4°C for 24 h, and then embedded in paraffin for preparation of 4-μm-thick sections.

The MPO levels in plasma were detected with ELISA (Cusabio, Wuhan, China). Quant-iT PicoGreen dsDNA Reagent (Thermo Fisher Scientific, Waltham, MA) was used to quantify NETs. Briefly, 100 μL standard or sample and 100 μL PicoGreen dsDNA reagent were added to a black opaque 96-well plate. After incubation at room temperature in the dark for 5 min, the fluorescence was detected with a fluorescence microplate reader at excitation and emission settings wavelengths of 485 and 528 nm, respectively, and dsDNA in the sample was quantified by a standard curve with known dsDNA concentrations.

The placental slices were dewaxed with xylene and dehydrated by a gradient series of ethanol (100%, 95%, 85%, 70%, 50%). After antigen repair with citrate buffer, 3% H₂O₂ was added to eliminate catalase, and the sections were blocked with goat serum. The diluted primary antibody against MPO (1:200, Research Resource Identifier [RRID] AB_307322, cat. no. ab9535; Abcam, Cambridge, U.K.) was added and incubated overnight at 4°C. After incubation with a fluorescent secondary antibody and DAPI in the dark, images were acquired with an inverted microscope.

The neutrophils from healthy donors (n = 6) were separated by Polymorphprep (Axis-Shield, Oslo, Norway) and directly seeded in 48-well plates after counting by a hemocytometer. Neutrophils from each healthy donor were independently tested. Neutrophils were treated with TMAO (300 μmol/L; Sigma-Aldrich, St. Louis, MO), phorbol 12-myristate 13-acetate (PMA) (50 nmol/L; Sigma-Aldrich), and lipopolysaccharide (LPS) (250 ng/mL; Sigma-Aldrich). The samples were stained with DAPI for 5 min, and fluorescence images were taken under an inverted microscope.

Neutrophils were isolated as previously described and then counted and seeded in cell culture dishes. After PMA stimulation for 4 h, the medium was removed. The bottom of the dish was washed repeatedly with cold PBS, and the solution was transferred to a centrifuge tube. Centrifugation was performed at 450g for 10 min at 4°C to obtain a cell-free supernatant rich in NETs, and the NETs content was quantified with Quant-iT PicoGreen dsDNA Reagent.

The human placental extravillous trophoblast cell line HTR-8/Svneo was cultured in RPMI medium supplemented with 10% FBS at 37°C in a 5% CO₂ cell incubator. The experiment was conducted under conditions of serum starvation. HTR-8/Svneo cells were treated with TMAO (300 μmol/L), NETs (200 ng/mL), and NETs in combination with DNase 1 (Sigma-Aldrich) for 24 or 48 h. The concentrations of TMAO and NETs were confirmed with the Cell Counting Kit-8 (CCK-8) assay and previous literature reports.

CCK-8 Cell Viability Assay was used to assess the effects of TMAO and NETs on the viability of HTR-8/Svneo cells. HTR-8/Svneo cells were seeded into 96-well plates, serum starved for 24 h, and then treated with TMAO at concentrations of 0, 100, 200, 300, 400, and 500 μmol/L for 48 h or NETs at concentrations of 0, 100, 200, 300, 400, and 500 ng/mL for 24 h. CCK-8 reagent (10 μL; Dojindo, Kumamoto, Japan) was added to each well and then incubated at 37°C for 1 h in the dark. Finally, the optical density value was read at a wavelength of 450 nm with a microplate reader.

HTR-8/Svneo cells were incubated with the specified concentrations of TMAO, NETs, and NETs in combination with DNase 1 at 37°C for 24 or 48 h. Then, the cells were seeded in sixwell plates and cultured overnight. The wound area was constructed manually by scraping with a 200-μL pipette tip. The degree of wound healing was photographed with an inverted microscope at 0, 24, and 48 h.

The treated cells and 200 μL RPMI medium were transferred to the transwell chamber with or without Matrigel, and 600 μL complete medium was added to the 24-well plate as a container for the transwell chamber. After 24 h, the cells were immersed in 4% paraformaldehyde for 30 min, stained with crystal violet for 20 min, and washed twice with PBS. After drying, an inverted microscope was used to take pictures and directly count the stained cells.

The treated cells and 500 μL RPMI medium were transferred to a 48-well plate with Matrigel. After 4 h, an inverted microscope was used to take pictures.

The plasma TMAO levels were assayed using liquid chromatography–mass spectrometry. Plasma samples were prepared by the addition of the internal standard deuterated TMAO (Cambridge Isotope Laboratories, Andover MA) and ice-cold acetonitrile to 100 μL plasma for extraction of metabolites and precipitate proteins. A standard curve for plasma analytes was prepared with addition of TMAO standards and the same amount of internal standard as used for sample preparation. Both technical support and equipment were provided by The Mass Spectrometry Center of the Laboratory of Maternal and Fetal Medicine of Chongqing Medical University.

Wild-type and PAD4^−/− C57BL/6J female mice (4–5 weeks old; RRID IMSR_JAX:030315) were randomly divided into four groups: control, TMAO, PAD4^−/−, and PAD4^−/−+TMAO; 0.12% TMAO was added to their drinking water. All mice were housed under a 12-h:12-h light/dark cycle at a controlled temperature. After 6 weeks, the female mice were mated with healthy males, and streptozotocin (110 mg/kg i.p.; Sigma-Aldrich) was injected on the second and eighth days after mating. The blood glucose rose to 10–20 mmol/L, which marked successful establishment of GDM mice model, and the vaginal plug was used to indicate gestation day (GD)0.5. The mice were sacrificed at GD14.5 or GD18.5 by researchers who were blinded to the groups of mice. Plasma and placental samples were collected, and the fetal and placental numbers and weight were recorded. The PAD4^−/− female mice were gifted by Professor Li Qifu (Department of Endocrinology, The First Affifiliated Hospital of Chongqing Medical University, Chongqing, China), and all the experimental procedures were approved by the animal ethics committee.

Data were analyzed with SPSS 25.0 (SPSS, Chicago, IL) or GraphPad Prism 7 (GraphPad Software, San Diego, CA). If the data showed a Gaussian distribution, Student t test was used to compare two groups, one-way ANOVA was used to compare multiple groups, and the Pearson correlation test was used to calculate correlations; otherwise, the nonparametric and Spearman rank correlation tests were adopted. The data are expressed as the mean ± SEM without special instructions. P < 0.05 was considered statistically significant.

No data sets were generated or analyzed during the current study. The rabbit polyclonal antibody against MPO (RRID AB_307322) and the PAD4^−/− (RRID IMSR_JAX:030315) mice analyzed during the current study are available in the RRID repository.

A meta-analysis was conducted to assess the changes in plasma choline levels in people with diabetes. The detailed characteristics of the included studies are shown in Supplementary Table 1. Compared with control subjects, people with diabetes showed higher choline levels (weighted mean difference [WMD]   0.95 μmol/L, 95% CI  0.03–1.86), lower betaine levels (WMD  −2.28  μmol/L, 95% CI  −3.47 to −1.10), and higher TMAO levels (WMD 1.45  μmol/L, 95% CI 0.73–2.17) in plasma (Fig. 1A–C). Further analysis revealed that the plasma TMAO levels in women with GDM were significantly higher than those in control pregnant women (WMD   0.14  μmol/L, 95% CI  0.09–0.20) (Fig. 1D). These results indicated that changes in choline metabolism, especially those in TMAO levels, may be involved in the impact of GDM on offspring.

The characteristics of the study participants are summarized in Supplementary Tables 2 and 3. The TMAO levels in the peripheral plasma of women with GDM were increased significantly (Fig. 2A), which was consistent with the results of our meta-analysis. Correlation tests showed that the TMAO levels in the peripheral plasma were positively correlated with the fasting blood glucose (FBG), 1-h postprandial blood glucose (1h-PBG), and 2-h postprandial blood glucose (2h-PBG) in pregnant women (Fig. 2B–D). The cord plasma TMAO levels in women with GDM were also significantly higher than those in the control group (Fig. 2E), indicating that the fetuses affected by GDM had higher circulating TMAO levels. BMI, systolic blood pressure, and BMI before pregnancy of pregnant women were significantly positively correlated with the cord blood TMAO levels (Fig. 2F–H). The above results indicated that the metabolic changes in pregnant women were correlated with the increase of TMAO levels in peripheral and cord blood.

For clarification of the role of TMAO during pregnancy, correlation tests were first performed, revealing that the cord blood TMAO levels in pregnant women were positively correlated with the placental thickness and the newborn birth weight (Fig. 2I and J). The peripheral blood TMAO levels in pregnant women were also significantly positively correlated with the placental thickness and the newborn birth weight (Fig. 2K and L). To further analyze for correlation between TMAO and placental thickness and newborn weight, we used multiple linear regression analysis. After adjustment for BMI, age, FBG, 1h-PBG, and 2h-PBG, the TMAO levels in the peripheral blood of pregnant women were still positively correlated with the placental thickness (β = 0.429, P = 0.026). The TMAO levels in peripheral blood (β = 0.348, P = 0.047) and cord blood (β = 0.626, P = 0.017) of pregnant women were also significantly positively correlated with newborn weight (Supplementary Tables 4 and 5). The placenta is the most direct organ connecting the mothers and fetuses, and the establishment of the maternal-fetal cycle in early pregnancy depends on extravillous trophoblast cells. Therefore, we chose the extravillous trophoblast cell line HTR-8/Svneo to explore the effect of TMAO on the placenta. First, we investigated the effects of different doses of TMAO on cell viability at different time points (24 or 48 h). As shown in Fig. 3A, HTR-8/Svneo cells treated with TMAO showed an increased viability. The proliferation, migration, and invasion abilities of extravillous trophoblast cells are the basic elements necessary for ensuring the successful remodeling of blood vessels and the completion of pregnancy. Therefore, we further explored the effect of TMAO on the biological function of HTR-8/Svneo cells. TMAO treatment significantly enhanced the ability of HTR-8/Svneo cells to migrate, invade, and form blood vessels (Fig. 3B–D). These data indicated that TMAO could enhance the viability and biological function of HTR-8/Svneo cells.

The excessive or abnormal formation of NETs plays a role in the pathologies of many inflammatory diseases, and the neutrophils of women with GDM show obvious activation (21). Therefore, we wanted to know whether TMAO played a role in the formation of NETs. We observed that neutrophils treated with PMA and LPS formed a large number of NETs, while TMAO had no obvious effect on neutrophils. However, after TMAO pretreatment of neutrophils, neutrophils obviously resisted the induction of NETs formation by PMA and LPS (Fig. 4A). The results of the dsDNA content analysis were also consistent (Fig. 4B). These results proved that TMAO could inhibit the release of NETs from neutrophils.

To explore the changes in NETs release in GDM, we measured the levels of dsDNA and MPO in the peripheral blood and observed that their levels in women with GDM were significantly higher than those in the control group (Fig. 5A and B), which suggested that although pregnancy itself promotes NETs formation (22), GDM further increases the generation of NETs. Correlation tests showed that the peripheral plasma dsDNA and MPO content were positively correlated with the FBG, 1h-PBG, and 2h-PBG in pregnant women (Fig. 5C–H). In particular, the content of dsDNA and MPO in plasma were positively correlated with BMI before pregnancy (Fig. 5I and J), indicating that the metabolism of pregnant women was correlated with the release of NETs during pregnancy.

To clarify the role of NETs during pregnancy, we first detected the NETs content in cord blood and found that the dsDNA and MPO levels in women with GDM were significantly higher than those in the control group (Fig. 6A and B), indicating that the fetuses affected by GDM had relatively high circulatory NETs levels. Correlation analysis showed that the dsDNA and MPO content in cord blood were positively correlated with the BMI before pregnancy (Fig. 6C and D), indicating that the metabolism of pregnant women was correlated with the cord blood NETs levels. Therefore, we wanted to know whether NETs formation in the placenta is also increased in women with GDM. Immunofluorescence analysis of placental tissues confirmed this phenomenon, and the MPO expression levels on both the maternal and fetal sides of the placenta were increased significantly in women with GDM (Fig. 6E and F).

To clarify the effect of NETs on the placenta, we observed their effects on the HTR-8/Svneo cells. As shown in Fig. 7A, NETs treatment reduced the cell viability after 24 h in a dose-dependent manner. DNase 1 hydrolyzes NETs and was used to determine the effect of NETs on HTR-8/Svneo cells function. NETs significantly reduced the migration and invasion abilities of HTR-8/Svneo cells and hindered their ability to form blood vessels. However, the addition of DNase 1 restored the biological functions of HTR-8/Svneo cells (Fig. 7B–G). The above results proved that NETs could inhibit the viability and biological functions of HTR-8/Svneo cells.

To further explore the roles of TMAO and NETs in GDM, we established a GDM model in wild-type and PAD4^−/− mice, as shown in Fig. 8A. Peptidyl arginine deiminase 4 (PAD4) is essential for the generation of NETs. The dsDNA content in PAD4^−/− mice plasma was significantly lower than that in the control group, and the plasma dsDNA content in the mice administered TMAO was further reduced (Fig. 8B). This result was consistent with the vitro experiments and proved that TMAO can also inhibit NETs formation in mice with GDM. TMAO did not affect the prepregnancy weight of mice (Supplementary Fig. 1A), but it significantly increased the weight gain of mice with GDM during pregnancy (Fig. 8C), which suggested that TMAO mainly played a role during pregnancy. Therefore, we observed the fetuses and placenta in the second (GD14.5) and third (GD18.5) trimesters and found that TMAO supplementation significantly increased the fetal and placental weight. Knockout of PAD4 did not affect the fetal weight but significantly increased the placental weight (Fig. 8D–I). Changes in placental structure are closely related to its function, and we observed that the promotional effect of TMAO placental development did not affect the placental structure (Fig. 8J–L).

Choline metabolites are necessary to promote fetal brain development and tissue expansion, and choline deficiency during pregnancy increases the risk of adverse pregnancy outcomes, including impaired neurodevelopment and birth defects (23–26). Recent clinical studies have found significantly increased TMAO levels in women with GDM (11,12), but there are no reports on the role of TMAO in GDM. In this study, we found for the first time that TMAO can not only promote placental and fetal development of GDM, but also inhibit the formation of NETs in vivo and in vitro. A recent study found that the neutrophils of women with GDM were significantly activated (21). We found that NETs inhibited the viability and biological functions of extravillous trophoblast cells and that knocking out PAD4 can promote placental development in mice with GDM. Our results provide evidence that TMAO and NETs affect the placental development of GDM.

In our study we observed that the peripheral plasma TMAO levels in women with GDM were increased significantly and positively correlated with their blood glucose levels. TMAO is mainly produced by the metabolism of dietary choline by the gut microbiota, and changes in the gut microbiota are related to a variety of diseases. Numerous studies have shown that the gut microbiota composition in women with GDM is changed (27–29), which may explain their elevated levels of TMAO. We also found the cord plasma TMAO levels in women with GDM to be increased significantly, indicating that the fetuses affected by GDM had relatively high circulatory levels of TMAO. The TMAO levels in cord blood were positively correlated with BMI, systolic blood pressure, and prepregnancy BMI in pregnant women, indicating that metabolic changes in pregnant women promoted the transfer of TMAO to the fetuses. The placenta plays an important role in the growth and development of the fetuses, and the viability and migration and invasion functions of extravillous trophoblast cells are the basic elements necessary for ensuring the smooth completion of vascular remodeling and pregnancy (30). We found that TMAO treatment significantly enhanced the viability, migration, invasion, and angiogenesis of extravillous trophoblast cells. In further analysis of the placental and newborn data we found that TMAO was significantly positively correlated with placental weight and newborn birth weight. The placental and fetal weight were also significantly increased in mice with GDM administered TMAO. These results indicate that TMAO can promote placental and fetal development in GDM.

We also found significantly increased peripheral plasma NETs levels in women with GDM, which were positively correlated with the blood glucose levels in pregnant women. A high-glucose environment can activate neutrophils and promote the release of NETs, which may be induced by the NADPH oxidase–dependent pathway (31,32). The NETs levels in the cord blood of women with GDM were also increased significantly, indicating that their fetuses also had relatively high circulating NETs levels. Therefore, we assessed the effect of NETs on the placenta and found increased NETs formation in the placentas of women with GDM, and NETs can damage the viability and biological functions of extravillous trophoblast cells. In vivo experiments also revealed a significant increase in the placental weight of PAD4^−/− mice. Autophagy is essential for inducing NETosis, and PMA stimulates neutrophils to form NETs and increases autophagosome formation. Inhibition of autophagy prevents the condensation of intracellular chromatin and the formation of NETs (33–36). TMAO can inhibit the expression of the autophagy-related protein ATG16L1 in a dose- and time-dependent manner and subsequently reduce the formation of LC3II and p62 (37). Therefore, we assessed the effect of TMAO on NETs formation and found that it could inhibit the release of NETs by neutrophils both in vivo and in vitro. These findings indicate that in addition to promoting placental development, TMAO may also protect the placenta by inhibiting NETs formation.

In summary, we provide evidence that higher TMAO levels and the formation of abnormal NETs were associated with GDM. TMAO not only could promote the development of the placenta and fetuses but also could inhibit the formation of NETs to protect the placenta.

Acknowledgments. The authors give special thanks to Professor Li Qifu for help in donating PAD4^−/− female mice.

Funding. This work was supported by the National Natural Science Foundation of China (82071734 and 81871222 to X.X.), the Fundamental Science and Advanced Technology Research of Chongqing (CSTC2015jcyjB0146 to X.X.), and Basic Research and Frontier Exploration of Science and Technology Commission by Chongqing Municipality (CSTC2018jcyjAX0788 to Y.C.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. X.L. and Y.Z. designed the experiments. X.L., Y.Z., X.H., Y.C., N.C., J.L., M.W., Y.L., H.Y., L.F., and Y.H. performed the experiments. X.L. and Y.Z. wrote the manuscript. X.L., Y.Z., and J.L. analyzed data. C.W., H.Q., and H.Z. provided the reagents and analyzed data. X.X. designed, supervised, and revised the manuscript. All authors read and approved the final manuscript. X.X. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.