As a leading cause of maternal and neonatal mortality, preeclampsia (PE) is a complex pregnancy disorder, that affects 3%-8% of pregnancies globally (Abalos et al., 2013; American College of Obstetricians and Gynecologists, 2020). PE not only elevates the risk of adverse pregnancy outcomes (e.g. preterm birth and low birth weight) (Rana et al., 2019) but is also linked to severe maternal and child health issues, such as end-stage kidney disease for mothers (Phipps et al., 2019; Turbeville and Sasser, 2020) and bronchopulmonary dysplasia for offspring (Ratsep et al., 2016; Phipps et al., 2019). Despite numerous studies on PE in recent years, there remain various etiological theories, including placental dysfunction, immune system maladaptation, oxidative stress, angiogenic imbalance, genetic predisposition, and nutritional deficiency (Udenze, 2016; Jabalie et al., 2019). The exact pathogenesis of PE is still not completely understood, and termination of the pregnancy remains the only definitive cure. Therefore, a multi-dimensional investigation into the mechanisms of PE could reveal new potential targets for its early diagnosis and management, which might be crucial in preventing its onset and improving its prognosis.

The gut microbiota, a complex and vast community of microorganism species residing in the digestive tract, plays a crucial role in host metabolism, immunity, and nutrient absorption (Viennois and Chassaing, 2018). Emerging research suggests that the dysbiosis of gut microbiota may be involved in the development of PE. The altered composition of gut bacteria could promote inflammation, metabolic changes, and immune dysregulation, all of which are linked to PE. Compared to healthy controls, significant alterations of gut microbiota and their metabolites were observed in patients with PE (Gomez-Arango et al., 2016; Liu et al., 2017; Lv et al., 2019; Wang et al., 2019). Recent studies have indicated that short-chain fatty acids (SCFAs) are involved in regulating the levels of blood pressure in both patients with PE and PE pregnant rats (Chang et al., 2020; Yong et al., 2022). Moreover, the dysbiosis of gut microbiome disrupted the gut barrier, leading to the colonization of intestinal bacteria in the uterine cavity, thus causing PE (Chen et al., 2020). However, the findings from published studies are inconsistent. Although multiple studies have found that Bifidobacterium has a protective effect against PE (Ahmadian et al., 2020; Chen et al., 2020; Altemani et al., 2021), Altemani et al. observed an increased abundance of Bifidobacterium in patients with PE (Altemani et al., 2021). Miao and Lv et al. identified Blautia as a risk factor for PE (Lv et al., 2019; Miao et al., 2021), whereas Chang and Yu reported the opposite (Chang et al., 2020; Yu et al., 2022). Therefore, it is essential to continue exploring the characteristics of gut microbiome in patients with PE by expanding the sample size and geographical scope.

Except for the dysbiosis of gut microbiota, alterations in the expression of the endocannabinoid system (ECS) are also linked to the development of PE. Compared with healthy pregnant women, women with PE exhibited reduced levels of anandamide (AEA) in both plasma and placental tissues (Molvarec et al., 2015; Maia et al., 2023). Meanwhile, decreased levels of other ECS components, such as N-oleoylethanolamine (OEA) and N-docosahexaenoylethanolamine (DHEA), were also observed in placental tissues (Maia et al., 2023). Interestingly, both ESC-G-protein coupled receptors 1 and 2 (CB1 and CB2) were expressed on enteric nerves, enterocytes, and immune cells in the gut, and the balance of ECS was affected by the gut microbiota. Moreover, the unbalance of ECS could in turn impact the integrity of the intestinal barrier (Moludi et al., 2018). For instance, when CB1 receptor antagonist SR141716A was administered orally to mice with diet-induced obesity (DIO), it led to attenuation in inflammatory cytokines of adipose tissue, changes in gut microbiome composition, and enhancement of mucus layer thickness, compared to DIO mice receiving a vehicle (Mehrpouya-Bahrami et al., 2017). In addition, the administration of Lactobacillus casei in antibiotic-treated mice not only alleviated depressive-like behaviors but also restored the concentrations of N-acyl-serotonin, which in turn normalized the concentrations of AEA in the gastrointestinal tract (Guida et al., 2018). Inspired by the interactions between gut microbiota and ECS components in animal models of human disease, we aimed to explore the associations between gut microbiota and AEA in patients with PE.

Nutritional deficiencies, such as low levels of vitamin D (VitD), also participated in the occurrence and progression of PE. VitD deficiency could impair placental development, angiogenesis, and immune system. Except for the closed linkage between VitD deficiency and cardiovascular diseases or arterial hypertension in observational studies (Pilz et al., 2009; Wimalawansa, 2018), both observational studies (Achkar et al., 2015; Baca et al., 2016; Serrano et al., 2018) and meta-analyses (Wei et al., 2013; Akbari et al., 2018; Serrano-Diaz et al., 2018; Aguilar-Cordero et al., 2020; Hu et al., 2022) have reported a significant association between VitD deficiency and an increased risk of PE. According to the current evidence, the associations between VitD levels and gut microbial structure and function are sufficient (Ooi et al., 2013; Jin et al., 2015; Waterhouse et al., 2019). The maternal gut microbiome is also associated with the intake of dietary VitD by mothers (Mandal et al., 2016). However, it is still unknown whether there are interactions among gut microbial compositions, AEA levels, and VitD levels in late-pregnant women with or without PE. On these bases, we aimed to provide new insights into the interactions of these risk factors by examining the structure and composition of gut microbiota, the plasma levels of AEA, and the serum levels of VitD, which may systematically provide evidence of multi-dimensional pathogenesis and intervention strategies of PE.

Fecal specimens (approximately the size of two soybean grains) were acquired from each subject either by themselves or their family members at home or in the hospital within 3 minutes of defecation. After collection, the fecal samplers (Biotecan, Shanghai, China) were sealed, labeled, and transferred to Biotecan Laboratories at temperatures below 18°C within 2 days. Upon arrival, they were stored at -80°C.

Maternal blood samples were obtained from an antecubital vein into plain tubes with anticoagulant for AEA detection and plain tubes without additives for VitD detection. These blood samples were centrifuged at room temperature with a relative centrifugal force of 4000 rpm for 10 min. The centrifuged plasma and serum were stored at -80°C until further analysis.

Plasma concentrations of AEA were quantified using high-performance liquid chromatography-mass spectrometry (HPLC-MS) methodology. Firstly, 50 μl plasma samples were mixed with 200 μl internal standard solution (100 ng/ml Arachidonoyl Ethanolamide-d8, Cayman Chemical, Art. No. 390050-500). The samples were vortexed for 5 minutes at 1400 rpm and centrifuged at 4000 rpm for 10 minutes. Following centrifugation, 120 μl of the supernatant was extracted for analysis. Quantification was performed using a calibration curve established from blank plasma samples spiked with varying concentrations of AEA, ranging from 0.125 to 4.000 ng/ml. The HPLC system was AB SCIEX Triple Quad™ 4500MD (SCIEX, Boston, USA) with an electrospray ionization source operated in positive ion mode. Separation of analytes was achieved through reversed-phase liquid chromatography using a Waters BEH-C18 column (Waters Corporation, Delaware, United States) with gradient elution. Mobile phase A consisted of 0.1% formic acid solution and mobile phase B was composed of 0.1% formic acid in methanol. The flow rate of the mobile phases was maintained at 450 μl/min during analysis.

Serum levels of 25(OH)D2 and 25(OH)D3 were quantified using the liquid chromatography tandem mass spectrometry (LC-MS/MS) method. Firstly, 50 μl of serum samples were combined with 200 μl of isotope-labeled internal standard solution (20 ng/ml 25(OH)D3-d6, Sigma, Art. No. H-074). Secondly, the samples were vortexed at 1400 rpm for 5 minutes and centrifuged at 4000 rpm for 10 minutes. Thirdly, 100 μl supernatant was transferred into a 96-deep well plate and dried under nitrogen at 45°C. Fourthly, derivatization was performed using a 60 μl acetonitrile solution containing 1 μg/ml PTAD for 5 minutes, and subsequently added 60 μl water for reconstitution. Fifthly, the prepared samples were injected into the UPLC-MS/MS system (Waters Corporation, Delaware, United States). Chromatographic separation was achieved on the Waters ACQUITY UPLC I-class using a binary gradient mobile phase consisting of water with 5 mM methylamine (mobile phase A) and methanol with 5 mM methylamine (mobile phase B). The flow rate of the mobile phases was set at 400 μl/min, and the column temperature was maintained at 40°C. MS/MS detection was conducted using the Xevo TQD in positive electrospray ionization mode with multiple reaction monitoring (MRM) mode.

PE is a complex pregnancy disorder involving hypertension and dysfunction across multiple systems including placental, vascular, renal, and immune systems. It can be classified with or without severe features, and delivery remains the definitive treatment (Robillard et al., 2017). Despite numerous studies highlighting the potential roles of intestinal bacteria, AEA, and VitD, none have established them as reliable biomarkers for predicting the onset of PE. Moreover, their interactions in late-stage pregnancy women remain poorly understood. In our study, α diversity did not significantly differ between the LP and PE groups, whereas β diversity showed notable distinctions. Subdoligranulum, Parabacteroides, Bacteroides had suggestive protective effects against PE, while Prevotella, Erysipelotrichaceae_UCG-003, and Dorea might induce the disease occurrence. Furthermore, women with PE had a significantly lower plasma level in AEA and a marked decrease in serum VitD compared to normal late-pregnant women. Lastly, although the plasma level of AEA was not significantly correlated with VitD or any of the top 6 marker genera, VitD was significantly negatively correlated with the relative abundance of Dorea, a novel finding in this context.

The human gut microbiome exhibited significant variability among individuals, and microbiomes of both healthy and diseased individuals often show overlapping patterns with some degree of differentiation when analyzed using unsupervised learning methods (He et al., 2018). However, our study revealed significant differences in β diversity between the LP group and the PE group, indicating significant differences in the composition of gut microbiota in patients with PE. Of particular interest were the dysbiosis patterns showing significant depletion of Subdoligranulum, Parabacteroides, and Bacteroides in the PE group. Consistent with previous findings, patients with PE exhibited lower abundances of Subdoligranulum compared to healthy pregnant controls (Chang et al., 2020). Interestingly, the decreased level of Subdoligranulum observed in patients with PE was butyric acid-producing gut genus (Louis et al., 2010; Riviere et al., 2016; Qin et al., 2019), suggesting that the reduction in butyric acid in these individuals might be attributed to the deficiency of this gut bacterium. Several studies have reported that the levels of butyrate are lower in PE cases than in controls (Chang et al., 2020; Altemani et al., 2021; Jin et al., 2022). The protective role of butyrate against PE was supported by several biological mechanisms. For instance, animal studies indicated that elevating butyrate levels in the colon significantly reduced blood pressure, potentially by acting on the vagus nerve and GPR41/43 receptors (Onyszkiewicz et al., 2019; Yang et al., 2019), which might help counteract hypertension, a key risk factor for PE. In addition, butyrate was known to modulate the immune response, enhance gut barrier integrity, and support healthy placental development (Hu et al., 2019; Zietek et al., 2021). Furthermore, higher butyrate levels were linked to decreased circulating endotoxins and inflammatory cytokines, both of which were implicated in PE (Zietek et al., 2021). Given the influence of dietary interventions on butyrate, it presented a promising target for preventing PE (Frederick et al., 2005; Qiu et al., 2008).

As anaerobic Gram-negative bacilli, both Parabacteroides and Bacteroides are Bacteroidaceae members, and constitute a substantial proportion of gut microbiota. Consistent with previous findings, patients with PE exhibited lower abundances of Parabacteroides and Bacteroides compared to healthy pregnant controls (Lv et al., 2022; Zhao et al., 2022; Lv et al., 2024), suggesting their protective roles in the occurrence of PE.

Cekanaviciute et al. reported that Parabacteroides distasonis stimulated the expression of anti-inflammatory IL-10 in human CD4⁺CD25⁺T cells and IL-10⁺FoxP3⁺Tregs mouse models (Cekanaviciute et al., 2017). Although limited directed associations have been reported between PE and Parabacteroides, patients with PE exhibited an enhanced inflammatory response accompanied by decreased blood levels of IL-10 in the third trimester of pregnancy in comparison to controls (Sahin et al., 2015; Aggarwal et al., 2019), indicating that Parabacteroides might contribute to the progression of PE by regulating immunity. Meanwhile, some Parabacteroides species could also secret SCFAs, including acetate, propionate, and butyrate (Ahmed et al., 2019; Fu et al., 2019). Except for the pervasive protective effects of butyrate in PE cases, the concentrations of acetate and propionate in PE cases were also significantly lower than in controls (Bock, 1994; Bahado-Singh et al., 2012, 2015; Jin et al., 2022). In addition, many studies have reported the crucial roles of Bacteroides species in immune and metabolic processes (Wexler, 2007; Lv et al., 2016; Zafar and Saier, 2021). Bacteroides are significant contributors to the biosynthesis of lipopolysaccharides (LPS), which can induce inflammation during pregnancy. However, recent studies and our findings indicated significantly lower relative abundances of Bacteroides in the PE group compared to the LP group (Lv et al., 2022; Zhao et al., 2022). Evidence of different beneficial functions from various Bacteroides strains may partly explain this phenomenon. Bacteroides vulgatus and Bacteroides dorei are the predominant species within the Bacteroides genus in the human gut microbiome, and their LPS compounds penta- and tetra-acylated lipid A differ structurally from the hexa-acylated LPS found in Escherichia coli (Vatanen et al., 2016). what’s more, the LPS compounds penta- and tetra-acylated lipid A of these two Bacteroides species could elicit reduced Toll-like receptor 4 (TLR4) responses (Vatanen et al., 2016). The activation and maturation of dendritic cells via TLR signaling promote the upregulation of major histocompatibility complex (MHC) molecules, costimulatory factors, and cytokine production, ultimately leading to T-cell activation (Ardavin et al., 2004; Mills, 2011). In addition, Bacteroides fragilis is also a well-studied representative known for its immune-regulatory abilities. Bacteroides fragilis ATCC25285 has been shown to protect against intestinal inflammatory diseases caused by pathogenic Helicobacter hepaticus (Mazmanian et al., 2008). Another strain, Bacteroides fragilis ZY-312, facilitates the polarization of bone marrow-derived macrophages towards the M1 phenotype, promoting the phagocytosis of pathogens (Deng et al., 2016). Interestingly, Wang et al. also observed differences in the relative abundances of Bacteroides species between patients with PE and healthy controls, with lower levels of Bacteroides stercoris and higher levels of Bacteroides coprocola and Bacteroides fragilis in patients with PE compared to healthy controls (Wang et al., 2019). Taken together, given the diverse roles of Bacteroides species in immunomodulation, transitioning from 16S rRNA sequencing to metagenomics sequencing would be essential for comprehensive understanding in future studies.

Meanwhile, the relative abundances of Prevotella and Erysipelotrichaceae_UCG-003 in the PE group were significantly higher than in the LP group, indicating these bacteria had abilities to induce the occurrence of PE in this study. Inflammation of the chorionic plate is linked to the presence of several bacterial species in the preeclamptic placenta, including Prevotella, Bacillus cereus, Listeria, Salmonella, Escherichia, Klebsiella pneumonia, Anoxybacillus, Variovorax, Porphyromonas, and Dialister (Amarasekara et al., 2015). Moreover, women with severe PE exhibited a higher relative abundance of Prevotella bivia in their vaginal microbiota, along with elevated plasma levels of the pro-inflammatory cytokine TNF-α, compared to healthy controls (Hung et al., 2004; Lin et al., 2020). Although women with gestational diabetes mellitus exhibited a higher abundance of genus Erysipelotrichaceae UCG-003 in comparison with normoglycemic women (Ferrocino et al., 2018; Vavreckova et al., 2022), research about the effects of Erysipelotrichaceae_UCG-003 on PE is still limited. Enhanced relative abundances of Erysipelotrichaceae in patients with colorectal cancer, and animal models of 1, 2-dimethylhydrazine-induced colon cancer or inflammatory bowel diseases, have been reported (Chen et al., 2012; Zhu et al., 2014; Schaubeck et al., 2016), indicating its importance in inflammation-related disorders.

The composition of gut microbiota can be influenced by lifestyle factors, such as VitD intake. VitD deficiency has been associated with dysbiosis of gut microbiota and gastrointestinal inflammation, with studies showing that VitD supplementation can significantly enhance gut microbial diversity and increase the ratio of Bacteroidetes to Firmicutes (Singh et al., 2020). Additionally, VitD intake has also been linked to higher abundances of Akkermansia, Bifidobacterium, Coprococcus, and Lactococcus, while reducing the abundances of Porphyromonas, Ruminococcus, Veillonella, and Erysipelotrichaceae (Kanhere et al., 2018; Naderpoor et al., 2019; Charoenngam et al., 2020; Singh et al., 2020). In this study, although there was no significant reduction in the serum levels of VitD among women with PE compared to normal late-pregnant women, a significant negative correlation was identified between the VitD levels and the relative abundance of Dorea. Consistent with findings from a randomized controlled trial in men with pre-diabetes and VitD deficiency, VitD supplementation could decrease the relative abundance of several genera within the Lachnospiraceae family, including Dorea, Blautia, Roseburia, and Ruminococcus (Ciubotaru et al., 2015). Meanwhile, in an observational cohort study, alcohol-dependent subjects with increased intestinal permeability exhibited higher relative abundances of Dorea, indicating the important role of Dorea in the maintenance of gut barrier integrity (Leclercq et al., 2014). In addition, we also inferred the mechanism underlying how VitD might influence the gut microbiota in patients with PE. VitD contributed to the maintenance of mucosal barrier integrity by enhancing the expression of tight junction and adherent junction proteins and inhibiting epithelial cell apoptosis, thereby preserving gut barrier integrity and function (Wu et al., 2010; Ooi et al., 2012; Clark and Mach, 2016). The disruption of gut barrier function could lead to increased susceptibility to pathogenic infections and heightened inflammation, which, in turn, adversely affects the gut microbiota (Jin et al., 2015; Waterhouse et al., 2019; Li et al., 2023). Moreover, the disruption of gut barrier integrity could also result in the colonization of gut bacteria in the uterine cavity, thus causing PE (Chen et al., 2020). Meanwhile, Vitamin D receptors (VDRs), which are abundantly expressed in intestinal enterocytes, especially in the proximal colon, facilitate the production of antimicrobial peptides like cathelicidins, defensins, claudins, and zonulin occludens (Wu et al., 2010; Ooi et al., 2012; Clark and Mach, 2016). This selective elimination of pathogenic bacteria provides a greater chance for beneficial bacteria to colonize. As a whole, VitD status can influence gut microbiota composition by promoting anti-inflammatory responses and reducing infection risk, making it a potentially modifiable factor in preventing PE by maintaining the balance of gut microbiota (Talsness et al., 2017).

There were several limitations in this study. Firstly, the sample size was relatively small and collected from a single center. Future research with larger sample sizes and data collected from multiple centers will be necessary to validate our findings. Secondly, dietary and nutrient intake information was gathered solely through questionnaires, which may not provide precise assessments. Thirdly, our study primarily observed dysbiosis in gut microbiota among late-pregnant women with PE without delving into the underlying mechanisms. Lastly, we lacked blood and fecal samples from patients with severe PE, as it was challenging to recruit individuals matching the age, gestational age, and BMI criteria of normal late-pregnant women or those with PE. Currently, we are collecting samples from normal late-pregnant women, patients with PE, and severe PE to further substantiate our preliminary findings.

In summary, the profile of gut microbiota differed notably between the PE and LP groups. Gut bacteria involved in regulating immune response and gut barrier integrity (e.g. Subdoligranulum, Bacteroides, and Dorea), showed significant alterations in PE patients, suggesting their potential roles in the onset of PE. Moreover, while the plasma levels of AEA were not significantly correlated with the serum levels of VitD or any of the 6 identified marker genera, there was a significant negative correlation between VitD and Dorea, indicating VitD and gut microbiota have the potential to be combined targets for early diagnosis and management of PE.