A total of 78 specific pathogen free (SPF) grade female Institute of Cancer Research (ICR) mice (6–8 weeks old, body mass 25–35 g) and 30 SPF grade male ICR mice (8–10 weeks old, body mass 30–35 g) were purchased from Shanghai Sippe-Bk Lab Animal Co., Ltd. [Laboratory Animal License No. SCXK (Shanghai) 2018-0006]. The mice were housed in Zhejiang Chinese Medical University Laboratory Animal Research Center, SPF grade animal laboratory: 22–24°C, relative humidity 55–65%, 12-h light/dark cycle and food and drink supplied ad libitum.

The present study was approved by the Animal Ethical and Welfare Committee of Zhejiang Chinese Medical University (ethics approval no. IACUC-20220919-25).

Lipopolysaccharide (LPS; MilliporeSigma; cat. no. L2630) was diluted using pH 7.2–7.4 phosphate-buffered saline (PBS), which had been filtered through a membrane to remove bacteria. The LPS solution was prepared at a concentration of 50 µg/ml. Quercetin (MilliporeSigma; cat. no. Q4951; purity ≥95%) was initially dissolved in a small quantity of anhydrous ethanol. It was subsequently diluted with PBS to create concentrations of 0.25, 1.25 and 2.5 mg/ml (25,26).

Mice were acclimated for 1 week and were included in the experiment after observing a normal estrous cycle. Females and males were mated 2:1 and cages were combined at 17:00 pm. Females were checked for vaginal plugs at 8:00 am the following morning. Those without vaginal plugs were recorded and continued to mate with males, while those with vaginal plugs were designated as the 1st day of gestation (D1). Pregnant mice confirmed by vaginal plugs were randomly divided into 6 groups, including control group, model group, quercetin + PBS group, low-dose quercetin group, medium-dose quercetin group and high-dose quercetin group, with 13 mice in each group.

On days 3–8 of gestation, mice in the control and model groups were gavaged with PBS; mice in the quercetin + PBS group and the high-dose quercetin group were gavaged with high-dose quercetin (2.5 mg/ml); mice in the low-dose and medium-dose quercetin group were gavaged with low-dose (0.25 mg/ml) and medium-dose (1.25 mg/ml) quercetin, respectively. Gavage was performed twice at 09:00 and 15:00, with each time 0.2 ml and a total of 0.4 ml each day. On day 7 of gestation, mice in both the control group and quercetin + PBS group received with a 0.2 ml injection of PBS into the tail vein and mice in the model group, low-dose quercetin group, medium-dose quercetin group and high-dose quercetin group were injected with 0.2 ml of LPS (50 µg/ml). The specific procedure was shown in Fig. 1.

Humane endpoints were as follows: Inability to eat or drink without anesthesia or sedation or stand for up to 24 h; mice in the absence of anesthesia or sedation exhibits poor condition including hypothermia (body temperature <37°C); mice show the appearance of central nervous system depression, tremor, paralysis, pain that does not respond to analgesics. 5 female mice were sacrificed prematurely according to the humane endpoints.

The body mass of each group of pregnant mice was measured in the morning at gestation day 1, 3, 7 and 9. After tail vein administration, the general state [including appearance, activity, autonomic activities, response to stimuli, degree of eye closure (27), and eye secretions; Table SI], diarrhea, vaginal bleeding and pregnancy discharge of pregnant mice in each group were observed and recorded in time. The body weight loss in D9 (compared with that in D7) was calculated as: Degree of weight loss (%)=[body weight (D9)-body weight (D7)]/body weight (D7) ×100%.

Mice were sacrificed by cervical dislocation under anesthesia on the morning of day 9 of gestation and complete uterine tissues were removed by dissecting the uterus. The death was confirmed by cessation of heartbeat, respiration, congestion and temperature of the skin. Embryos were observed and the gross weight and net weight of the uterus, the number of normal embryos and the number of dead embryos were recorded (or the number of implantation sites if the embryos were resorbed or expelled prior to the execution).

Criteria for determining dead embryos were uterus with ‘bamboo-like’ changes and embryos of different sizes with black or purplish-brown color. Following the intrauterine death, the size of the embryo does not continue to grow but shrinks, resulting in a bamboo-like appearance of the uterus, section by section without excessive expansion and with a knot between each section.

Embryo loss rate (%)=total number of dead embryos/(total number of dead embryos + total number of normal embryos) ×100%.

Mean embryo weight=(gross uterine weight-net uterine weight)/number of surviving embryos.

Implantation site tissue including embryo specimens from the control and model group were fixed with 4% paraformaldehyde at room temperature for 48 h. Specimens were embedded in paraffin, sectioned at 3 mm, and routinely stained at room temperature with hematoxylin (Solarbio, cat. no. G1150) for 3–8 min and eosin (Solarbio, cat. no. G1100) for 1–3 min. 100× and 200× light microscopes (Olympus, cat. no. BX53) were used to observe the pathological changes of the implantation sites.

The 2D structure diagram of quercetin (SDF format) was achieved from PubChem (28) and then was uploaded to PharmMapper platform (29) to predict potential targets within Homo sapiens. Uniprot database (30) was applied to correct and normalize all the retrieved target names.

SA-related genes were obtained from the databases of OMIM (31), GeneCards (32) and DisGeNet (33) using the keywords ‘spontaneous abortion’ and ‘pregnancy loss’ within ‘Homo sapiens’. The SA-related genes from the searches were combined and duplicates were removed. Gene names were standardized by the Uniprot database (30).

Venn online tool (34) was utilized to generate a Venn diagram and obtain mutual targets by intersecting the targets of quercetin with the targets of SA.

The overlapping targets were submitted to the STRING database (35) to retrieve the PPI network within the context of ‘Homo sapiens’ and at a medium confidence level (0.04). The nodes that were discrete from the main network were hided. Next, the PPI network in TSV file format was imported into Cytoscape (v3.7.1) (36) for visualization. The cytoNCA plugin was employed to calculate the degree for each target and the core targets were identified by applying a degree value exceeding the average degree (37).

The Metascape database (38) was used to analyze the enriched GO and KEGG pathways with the intersecting targets. P<0.01 was defined as the criteria for statistical significance. The top 20 KEGG pathways and the top 20 terms within the categories of molecular function (MF), biological process (BP) and cellular component (CC) were collected based on the smallest to largest P-value. The above results were then visualized using the Microbiome platform (http://www.bioinformatics.com.cn/).

The gene expression levels were quantified using RT-qPCR. RNA extraction, cDNA synthesis, and qPCR were performed according to the manufacturer's protocols. Total RNA was isolated from uterus tissues with TRIzol Plus RNA Purification Kit (Thermo Fisher Scientific, cat. no. 12183-555). Ultraviolet spectrophotometer and electrophoresis were used to test RNA purity and quantification. The RNA was reverse transcribed into cDNA using the SuperScript III First-Strand Synthesis SuperMix (Thermo Fisher Scientific, cat. no. 11752-050). RT-qPCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems, cat. no. 4367659) with the CFX384 Touch Real-Time PCR Detection System (Bio-Rad), the reaction volume of which was 20 µl. PCR cycling conditions (denaturation, annealing and extension, times and temperatures) are as follows:

For initial denaturation: 95°C for 60 sec, followed by 40 cycles: 95°C for 15 sec for denaturation; 63°C for 25 sec for annealing and extension. GAPDH) was served as the internal reference, and the 2^−ΔΔCq (39) method was applied to calculate the relative expression. The experiments were performed in biological triplicate for each group.

All primers were purchased from Sangon Biotech Co., Ltd. (Shanghai, China). Primer sequences were as follows: GAPDH were F: 5′-GAAGGTCGGTGTGAACGGATTTG-3′; R: 5′-CATGTAGACCATGTAGTTGAGGTCA-3′. Primer sequences for Akt1 were F: 5′-GAAGGTCGGTGTGAACGGATTTG-3′; R: 5′-GCATGAGGTTCTCCAGCTTCA-3′. Primer sequences for ESR1 were F: 5′-CAGGCTTTGGGGACTTGAATC-3′; R: 5′-CCAGACGAGACCAATCATCAGA-3′. Primer sequences for JAK2 were F: 5′-GCTACCAGATGGAAACTGTGCG-3′; R: 5′-GCCTCTGTAATGTTGGTGAGATC-3′. Primer sequences for MAPK1 were F: 5′-TCAAGCCTTCCAACCTCCTGCT-3′; R: 5′-AGCTCTGTACCAACGTGTGGCT-3′. Primer sequences for MAPK3 were F: 5′-CAACACCACCTGCGACCTT-3′; R: 5′-CCACATACTCCGTCAGAAAGC-3′. Primer sequences for PGR were F: 5′-GTCCGAGTTATGAGAACCCTTGA-3′; R: 5′-GATTTGGTGAAAAGGTGATTCTCTGG-3′. Primer sequences for PI3K were F: 5′-GATGTGGCTGACGCAGAAAG-3′; R: 5′-GGTTGCTGCTCCCGACATT-3′. Primers for SGK1 were F: 5′-GCTCGATTCTACGCAGCTGAA-3′; R: 5′-CCCTGGGAGTCTAGGAGAA-3′.

Microsoft Office Excel 2021 (Microsoft Corporation) was used to create database and SPSS 26.0 statistical analysis software (IBM Corp.) was used for data processing and analysis. Quantitative data that conformed to normal or near-normal distribution were described by mean ± standard deviation. Then t-test or one-way ANOVA was conducted and Tukey's or Scheffe's post hoc comparison was used to evaluate differences between groups. In case the data deviated from a normal distribution, they were represented by the median (upper quartile, lower quartile), namely M (P25, P75), analyzed by Kruskal-Wallis H test and multiple comparisons were performed by Kruskal-Wallis one-way ANOVA. Count data were described as percentages, compared using the χ² test or Fisher's exact probability method. Multiple comparisons were conducted using the Bonferroni method with a corrected test level of α'=0.008. P<0.05 was considered to indicate a statistically significant difference.

The pregnant mice in the control group and the quercetin + PBS group were in a good health, had shiny fur and were responsive, whereas the pregnant mice in the model group and medium-dose quercetin group had obviously rough and dull fur and were unresponsive, accompanied by diarrhea. About half of them had blood stains and pregnancy residues on their vaginas. In the high- and low-dose quercetin group, the pregnant mice showed an improved state than those in the model group and medium-dose quercetin group, but some of them had rough fur, diarrhea and were less responsive, accompanied by blood stains and pregnancy residues on their vaginas. After autopsy, it was found that 45.92% of the embryos from abortion mice in the model group had been absorbed or discharged. 53.57% of aborted embryos in the low-dose quercetin group, 84.78% in the medium-dose quercetin group and 55.88% in the high-dose quercetin group had been absorbed or discharged.

As shown in the Fig. 2B, the body weights of pregnant mice in the control group and the model group exhibited a steady increase from day 3 to day 5 of gestation. Conversely, pregnant mice in the medium- and high-dose quercetin group displayed a decline pattern in their body weights during this period. Subsequent to the tail vein injection, the body weights of mice in the control group and quercetin + PBS group continued to show consistent growth. In contrast, the other four groups exhibited varying degrees of weight decline after the injection.

Following the tail vein injection on gestation day 7, the body weights of pregnant mice in low-, medium-, high-dose quercetin and the model groups descended markedly (P<0.0001). Notably, the model group exhibited the most substantial decrease in body weight with the rate of 10.94±3.13%. Among three quercetin groups, the medium-dose quercetin group experienced the greatest reduction in body weight while the low-dose one showed the lowest weight loss rate of 5.92±2.47%. However, the body weight growth rate of the control group and the quercetin + PBS group remained similar, with slight growth rates of 0.50±4.14% and 1.03±1.90%, respectively (P=0.998; Table I).

In the control group and the quercetin + PBS group, the implantation sites were evenly distributed, with normal embryonic development. The uterus was in red and shaped like a bead, with a small amount of bruising in the uterine cavity. In the model group, the development of embryos was not synchronized with each other and the size of the implantation sites was reduced significantly. Some disappeared or were necrotic, leaving dark brown or dark red bamboo-like appearances of the uteri, section by section with a knot between each section, and obvious bruising was seen in the uterine cavity. The low- and medium-dose groups showed shrinkage in size and a whitening color of the implantation sites, but no bleeding spot or bruising was seen in the uterine cavity. While the high-dose quercetin group displayed a superior embryo status, there was a small amount of bleeding spot or bruising in the uterine cavity (Fig. 2A).

Compared with the model group, the number of embryo deaths decreased in both the control group and quercetin + PBS group, while the number of surviving embryos increased in the quercetin + PBS group (P=0.028, P=0.028 and P=0.034, respectively). Meanwhile, the medium-dose quercetin group exhibited a markedly higher number of embryo deaths in comparison to the control group (P=0.035). No significant difference was observed in the numbers of surviving and dead embryos between the control group and quercetin + PBS group (P=1.000 and P=1.000, respectively; Table II).

Regarding the mean weight of surviving embryos, both the control group and quercetin + PBS group exhibited significantly higher values than the model group (P<0.0001), which were 0.0267±0.0019, 0.0258±0.0025 g and 0.0174±0.0008 g, respectively. The high-, medium- and low-dose quercetin groups all demonstrated higher mean weight than the model group, with the high-dose quercetin group displaying the highest mean weight of 0.0218±0.0014 g among three quercetin groups (P=0.049, P=0.376 and P=0.457, respectively). Moreover, the mean weight in the high-, medium- and low-dose quercetin groups were all lower than that in the quercetin + PBS group (P=0.0096, P=0.0011 and P<0.0001, respectively). The results are shown in Fig. 2C and Table SII.

The embryo loss rate of the model group was 59.39%, while there was no embryo loss in the control group and the quercetin + PBS group, showing a statistically significant difference (P<0.0001). As for the three quercetin groups, the medium-dose quercetin group shared the highest rate of embryo loss (51.69%), accompanied by the rates of 15.73 and 38.64% in the low- and high-dose quercetin groups. Statistically speaking, the high- and medium-dose quercetin groups displayed an increased number of dead embryos and a higher rate of embryo loss in comparison to the low-dose quercetin group (P<0.0001 and P<0.0001, respectively). The low-dose quercetin group exhibited a reduction in both the quantity of dead embryos and the rate of embryo loss. Moreover, the high-dose quercetin group showed a significantly elevated number of dead embryos and a higher rate of embryo loss when compared with the high-dose quercetin + PBS group (P<0.0001). Interestingly, no significant difference was found in the rate of embryo loss between the medium-dose quercetin group and the model group (P=0.159; Table III).

To confirm that quercetin does prevent SA, the results of another set of experiments showed that all pregnant mice in each group could produce offspring, and there was no significant difference in litter size and birthweight of the pups (Table SIII and Fig. S1).

Fig. S2 shows the representative hematoxylin and eosin staining for the implantation site tissue sections of the control and model group, including the embryo. The findings showed that the embryos of normal mice were structurally intact with regularly and tightly arranged cells, while those of aborted mice were characterized by irregularly and loosely arranged cells with the significant infiltration of inflammatory cells and red blood cells, the structures of which were not as intact as the former.

A total of 291 candidate quercetin targets were obtained using PharmMapper platform and 53, 1,498 and 197 SA-related targets were estimated in OMIM, GeneCards and DisGeNet databases respectively by using the keywords ‘spontaneous abortion’ and ‘pregnancy loss’. Finally, a total of 1,622 disease-associated targets were determined after removing 126 duplicates.

The Venn online tool was applied to visualize the Venn diagram. A total of 80 targets of quercetin intersection with SA were obtained.

The PPI network of the 80 intersecting targets was generated by the STRING database (Fig. 3A). After removing the nodes that were discrete from the main network, 79 targets and 579 edges were finally presented. The TSV format result was downloaded and input into Cytoscape (3.7.1) for visualization. The degree value of each target was obtained using the cytoNCA plug-in and the average degree value of 14.658 was calculated using Excel software. Consequently, 34 core targets that had higher degree value than the average value were screened out (Fig. 3B and Table SIV). Among them, the top three targets in terms of degree value were AKT1, albumin (ALB) and caspase-3 (CASP3).

GO and KEGG enrichment analysis were performed by importing the 80 intersecting targets into the Metascape database.

The GO enrichment analysis of quercetin treating SA yielded a total of 1,062 terms (P<0.01), including 923 entries for BP, 44 entries for CC and 95 entries for MF. BP was mainly involved in ‘hormone responses’, ‘regulation of defense responses’, ‘inflammatory responses’, ‘cell migration’ and ‘phosphatidylinositol 3-kinase signaling’. CC was mainly involved in ‘outer plasma membrane’, ‘extracellular matrix’, ‘serine peptidase complex’ and ‘serine endopeptidase complex’. MF was mainly comprised of ‘endopeptidase activity’, ‘serine hydrolase activity’, ‘serine peptidase activity’, ‘serine endopeptidase activity’, ‘peptidase activity’, ‘organic acid binding’, ‘nuclear receptor activity’ and ‘carboxylic acid binding’. The Fig. 3C shows the top six entries in BP, CC and MF sorted from the smallest to largest P-value.

A total of 130 KEGG pathways were responsible for in the treatment of SA (P<0.01). To unravel the basic molecular mechanisms leading to quercetin in the treatment of SA, the top 20 pathways were selected in order of P-value from smallest to largest. As shown in Fig. 3D, quercetin was not only related to cancer and diabetes, but also associated with signaling pathways of PI3K-Akt, Ras, Rap1, MAPK, TNF, VEGF and C-type lectin receptor.

The relative mRNA expressions of AKT1, MAPK1, PGR and SGK1 in the model group were lower than those in the control group (P<0.0001, P=0.012, P=0.0002 and P=0.001, respectively), and the expressions of the same targets in the quercetin groups were higher than those in the model group. But the expression changes of ESR1 and MAPK3 showed a reverse trend: the expressions in the model group were higher than those in the control group (P=0.013 and P=0.0004, respectively), while the ones of quercetin groups were lower than that in the model group. There was no statistical difference between the expressions of JAK2 in the model and the control group, as well as PI3K. LD-QUE group shared the highest expression of JAK2 among the six groups. Data and pictures are shown in Table SV and Fig. S3.