Effects of L-selenomethionine supplementation on nutrient digestibility and metabolism, antioxidant capacity, hormone levels, and fecal microbiota diversity in pregnant Yili mares during mid- to late gestation

Abstract

Background

L-selenomethionine (L-SeMet), a highly bioavailable organic form of selenium, plays a critical role in maintaining antioxidant homeostasis, regulating reproductive hormone secretion, and improving intestinal microbial ecology. Previous studies have demonstrated that appropriate supplementation with L-SeMet can significantly enhance the production performance and health status of ruminants. However, the nutritional regulatory mechanisms and physiological effects of L-SeMet in monogastric herbivores, particularly horses during mid- to late gestation, remain inadequately understood. Therefore, this study investigated the effects of different levels of L-SeMet supplementation on nutrient digestibility and metabolism, antioxidant capacity, reproductive hormone profiles, and fecal microbiota diversity in pregnant Yili mares.

Results

The results showed that selenium (L-SeMet) supplementation at 0.4, 0.6, or 0.8 mg Se mare⁻¹ day⁻¹ significantly increased apparent crude protein digestibility and serum glutathione peroxidase (GSH-Px) activity in pregnant mares compared with controls. Compared with the control group, the 0.6 and 0.8 mg Se mare⁻¹ day⁻¹ groups exhibited significantly higher neutral detergent fiber (NDF) digestibility, nitrogen metabolism rate, total antioxidant capacity (T-AOC), catalase (CAT) activity, progesterone, and estradiol levels, while malondialdehyde (MDA) and urinary estrone levels were reduced. Fecal microbiota analysis further revealed an increased relative abundance of methanogens and Actinobacteriota, particularly in the 0.6 mg Se mare⁻¹ day⁻¹ group. Functional predictions indicated enrichment of microbial metabolic pathways related to carbohydrates and energy metabolism.

Conclusions

Collectively, these findings indicate that selenium supplementation (provided as L-SeMet) enhances nutrient utilization, antioxidant defenses, and the endocrine milieu during pregnancy, with 0.6–0.8 mg Se mare⁻¹ day⁻¹ appearing to confer the broadest benefits; dose optimization and long term outcomes warrant further investigation.

Introduction

Pregnancy is a dynamic physiological process involving changes in maternal reproductive and physiological functions, as well as fetal development. In female livestock, pregnancy is accompanied by alterations in nutritional requirements, digestive physiology, blood biochemical parameters, and gut microbiota composition [1, 2]. Maintaining these dynamic factors within normal ranges is crucial for the smooth progression of pregnancy. Increased pregnancy-related stress may adversely affect maternal physiology, thereby endangering both maternal and fetal health [3]. The biochemical indices and physiological status of pregnant mares show adaptive changes. According to Vincze et al., reproductive hormones undergo specific changes during mid-to-late pregnancy: progesterone levels are relatively low during mid-gestation but gradually increase 30–60 days before parturition, while estrogen peaks between the 7th and 8th months of pregnancy and then gradually declines until foaling [4–6]. In addition, the gut microbiota of pregnant livestock changes as pregnancy progresses, showing a decrease in microbial diversity [7]. The physiological adaptations of pregnancy may also induce oxidative stress, with lipid peroxidation product levels increasing up to twofold before parturition. In mid-to-late pregnancy, rapid fetal growth and increased metabolic demands lead to the generation of excessive reactive oxygen species (ROS), which may overwhelm the antioxidant defense system, resulting in oxidative stress [8] and compromising maternal health.

Selenium (Se), an essential trace mineral for mammals, plays an irreplaceable role in various physiological functions. It is indispensable for maintaining normal physiological activities, growth, reproduction, immunity, antioxidant defense, nutrient metabolism, and hormonal regulation [9, 10]. Studies have shown that selenium supplementation improves rumen fermentation efficiency and nutrient metabolism in ruminants [11–13]. In equines, different selenium sources have distinct effects on immune regulation, such as modulating IL-5 expression in lymphocytes and IL-1/IL-8 expression in neutrophils [14]. In Garza ewes, pre-breeding supplementation with organic or inorganic selenium and vitamin E enhanced metabolite and reproductive hormone levels, likely due to optimized cholesterol, glucose, and hormone metabolism [15]. Notably, supplementing pregnant sows with DL-2-hydroxy-4-methylselenobutyric acid (HMSeBA) not only improves redox balance by increasing glutathione peroxidase (GSH-Px) activity but also regulates placental inflammation and autophagy, thereby protecting fetal development [16]. Compared to inorganic selenium, organic selenium (such as selenomethionine and selenium-enriched yeast) has higher bioavailability due to its methionine-like structure [17], and also offers advantages in tissue deposition and environmental sustainability [18, 19]. Its efficacy stems from maintaining epithelial cell activity [20], which can translate into measurable improvements in livestock performance, such as enhanced growth rate, milk yield, and egg production [21–26].

Therefore, pregnant Yili mares in mid-to-late gestation were used as the experimental model. L-selenomethionine (L-SeMet) was added to the total mixed ration (TMR) to support maternal health and mitigate abortion risk during this critical period. The objective was to evaluate effects on maternal nutrient digestibility and whole-body nitrogen metabolism, antioxidant status, hormone profile, and fecal microbiota diversity, thereby providing a theoretical basis for the management of pregnant mares.

Materials and methods

Ethical considerations

This study was conducted in strict accordance with all applicable institutional, national, and international guidelines and regulations for animal research.The protocols were approved by the Laboratory Animal Welfare Ethics Committee of Xinjiang Agricultural University (Approval No. 2022020). The study was conducted at Zhaosu Horse Farm, Yili Kazakh Autonomous Prefecture, Xinjiang, China.

Animals, management, and treatments

In the experiment, 32 Yili mares in mid- to late-pregnancy with similar expected due dates (aged 9–10 years, weighing 431.06 ± 9.39 kg, with a parity of 6–7) were selected from Zhaosu Horse Farm in Yili, Xinjiang, China. The mares underwent a two-week adaptation period before entering a 60-day experimental period. Based on the mares’ weights and parities, they were randomly divided into four groups: the control group, Treatment group I, Treatment group II, and Treatment group III (n = 8). Under identical feeding and management, the Control group received no supplemental selenium (Se), whereas Treatment groupI, II, and III provided 0.4, 0.6, and 0.8 mg Se mare⁻¹ day⁻¹, respectively (source: L-selenomethionine; corresponding doses: 0.2, 0.3, and 0.4 g mare⁻¹ day⁻¹). The supplementation doses were set based on the selenium requirement range for pregnant mares with a mature body weight of 400 kg, as outlined in the NRC (2007). The basic feed formula and nutrient composition of this experiment are shown in Table 1. During the entire experimental period, 11.20 kg of total mixed ration (TMR) was fed daily, and drinking water was always freely available. L-selenomethionine was supplemented at 10:00 every day to ensure the accurate supplementation amount for each horse and to ensure that they consumed it all. Pregnant Yili mares in mid to late gestation were kept in the same barn and handled by the same staff on a fixed schedule. A basal TMR was offered once daily with ad libitum water; the only difference between group was the amount of supplemental selenium.

Table 1.

The composition and nutritional level of the diet

Items	Content (%)	Nutritional levelb	Content (%)
Whole-plant corn silage	30.00	CP (%)	8.37
Alfalfa	18.00	EE (%)	1.73
Oat straw	30.00	NDF (%)	34.69
Wheat straw	9.00	ADF (%)	24.41
Corn	5.00	Ca (%)	0.56
Oats	4.50	P (%)	0.27
Wheat	2.50	DE (MJ/kg)	16.04
Salt	0.10
Premixa	0.90
Total	100

^a The premix provided the following for per kg concentrate: VA 6 500 IU, VD3 1 400 IU, VB1 21.26 mg, VB2 333.20 mg, VB6 1.20 mg, VE 850 mg, biotin 5 mg, pantothenic acid 20.46 mg, nicotinamide 84.85 mg, Cu (as copper sulfate) 35.00 mg, Fe (as ferrous sulfate) 127.62 mg, Mn(as manganese sulfate) 133.54 mg, Zn༈(as zinc sulfate) 150.00 mg, I (as potassium iodide) 21.46 mg, Co (as cobalt chloride) 4.11 mg

^b Nutrient levels were determined values

ATTD and nitrogen, energy, and calcium/phosphorus metabolism rates

During the formal experimental period, two digestion and metabolism trials were conducted on days 26–30 and 56–60, with a 5-day collection of feces and urine for each trial. During the digestion and metabolism trials, all experimental mares were individually tethered in stalls within the barn. Each day, before the morning feeding (10:00), the feed (TMR) for each mare was weighed, and a 500 g sample of forage was collected using the five point sampling method. The forage samples were airdried, reweighed, then ground and sealed in plastic bags for storage until analysis.

During the digestion and metabolism trials, fecal samples from the experimental mares were collected using the total fecal collection method. The feces collected from each mare throughout the day were thoroughly mixed. A 10% portion of the total fecal volume was then taken using the quartering method. This portion was split into two parts: one part (10% of the total fecal sample after sampling) was mixed with 10% sulfuric acid solution (10 mL of 10% sulfuric acid per 100 g of fecal sample) for nitrogen fixation. After thorough mixing, the sample was frozen at -20 °C for crude protein determination. The other part (90% of the total fecal sample after sampling) was dried in an oven at 65 °C. After drying, the sample was ground through a 40-mesh sieve, thoroughly mixed, and stored at -20 °C for the determination of other nutritional components.

Urine samples were collected using a urine collection device from the mare’s own urine collection system. The urine from each mare was thoroughly mixed in a collection barrel, and then 10% sulfuric acid solution was added to prevent nitrogen loss. The urine was filtered through four layers of gauze, and a 10% portion of the total urine volume was transferred into a 50 mL sampling tube, which was then frozen at -20 °C for storage.

Components of feed, fecal and urine samples in each group, including DM (method 930.15), crude protein (CP; method 984.13), ash (method 924.05), calcium (method 927.02), and phosphorus (method 984.27) were analyzed according to the method of the AOAC (2007) [27] Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were measured using filter bags and fiber analysis equipment (Ankom A200, Ankom Technology Corp., Fairport, NY, USA) following a modification of the procedure of Van Soest et al. [28]. The NDF was assayed without a heat-stable α-amylase but with sodium sulphite. Both NDF and ADF were expressed inclusive of residual ash. Acid-insoluble ash was analyzed based on the method of Van Keulen and Young [29]. Energy was determined using an automatic adiabatic oxygen bomb calorimeter (Parr 1281 Automatic Energy Analyzer). The sample was completely combusted in an oxygen filled oxygen bomb. The heat released during the combustion process raised the temperature of the surrounding medium (usually water). By measuring the change in water temperature and combining with the relevant parameters of the instrument, the gross energy of the sample was calculated.

On a dry matter basis (DM), the apparent total tract digestibility (ATTD) for given nutrient X was calculated as:

where X denotes a specific nutrient (e.g., DM, CP, NDF, ADF).

On a dry matter basis (DM), the apparent metabolism rate for a given nutrient/element X was calculated as:

Where Intake_x =DMI×Diet_x, Fecal_x=Fecal DM×Feacl_x, and Urinary_x=Urine volume×Urine_x.

Serum sample collection

Serum samples were collected in the morning on days 0, 30, and 60 of the formal experimental period. The specific procedure was as follows: 10 mL of blood was collected into a vacuum serum collection tube without anticoagulant through jugular venipuncture using an 18 gauge needle. The blood samples, without the addition of heparin, were drawn with a syringe and placed into sterile vials. Subsequently, the samples were centrifuged at 3500 rpm for 10 min to separate the serum. The separated serum was then transferred into 2 mL Eppendorf tubes and stored at -20 °C.

Serum total antioxidant capacity (T-AOC; ABTS method), superoxide dismutase (SOD; xanthine oxidase method), glutathione peroxidase (GSH-Px; colorimetric), catalase (CAT; colorimetric), and malondialdehyde (MDA; TBA colorimetric method) were measured with commercial kits (Beijing Huaying, Beijing, China) according to the manufacturer’s instructions: T-AOC Cat. HY60021; SOD Cat. HY60001; GSH-Px Cat. HY60005; CAT Cat. HYM0018; MDA Cat. HY60003. Assay endpoints were analyzed on an automated biochemical analyzer (Mindray BS-420; Shenzhen Mindray Bio-Medical Electronics Co., Ltd., Shenzhen, China).

The concentrations of testosterone, follicle-stimulating hormone (FSH), luteinizing hormone (LH), progesterone (P), estradiol (E₂), and gonadotropin-releasing hormone (GnRH) in serum are measured using enzyme-linked immunosorbent assay (ELISA) [30]. All hormone assays are conducted using the ELISA method, following the instructions provided in the respective assay kits.

Hormone and conjugated estrogen levels in urine

Urine samples from the mares were collected on the 0th, 30th, and 60th days of the formal experimental period. The urine from each mare throughout the day was collected in a bucket, thoroughly mixed, and then 10 mL of the mixture was transferred into a sampling tube. After labeling, the samples were stored frozen at -20 °C for subsequent measurement of hormones and conjugated estrogen markers.

The levels of E₂ (Cat. HY10029, Beijing Huaying, Beijing, China) and E₁ (Cat. HY10038, Beijing Huaying, Beijing, China) in urine were detected using ELISA kits. Both detections were carried out by the enzyme linked immunosorbent competitive assay and analyzed according to the manufacturer’s instructions.

Conjugated estrogens in the urine of pregnant mares, including ESS, EQS, and 17α-DHEQS, were determined using ultra high performance liquid chromatography (HPLC).

HPLC conditions and sample preparation. Analyses were performed on a C18 column (Ultimate XB-C18, 4.6 × 100 mm, 5 μm; Welch Materials). The UV detection wavelength was 215 nm; column temperature 20 °C; flow rate 1.0 mL/min; injection volume 10 µL. The mobile phase consisted of phosphate buffer/acetonitrile/methanol (680:240:80, v/v/v). The phosphate buffer (pH 7.00 ± 0.01) was prepared by dissolving K₂HPO₄·3 H₂O (2.00 g) and KH₂PO₄ (1.00 g) in water and making up to 1.000 L, followed by pH adjustment with HCl or KOH as needed. The mobile phase was mixed, filtered (0.45 μm), and degassed by ultrasonication (60 min). Urine samples were thawed, vortex-mixed, and subjected to stepwise dilution: 1:5 with water, centrifugation (12,000 rpm, 10 min, 4 °C), 1:2 dilution of the supernatant with water, and then 1:1 (v/v) with HPLC-grade methanol. The resulting injection solution was 50% methanol with an overall 1:20 dilution relative to the original sample; 10 µL were injected for HPLC-UV analysis. Unless otherwise stated, reagents were chromatographic grade and water was double distilled.

Microbial diversity and composition

On the last day of the experiment, fecal samples were collected by rectal sampling from all 32 pregnant mares (n = 8 per treatment), immediately snap-frozen in liquid nitrogen, transported on dry ice to Beijing Nuohe (Beijing, China), and subjected to 16 S rRNA gene sequencing (V3–V4 region) to assess microbial diversity and composition. Total genomic DNA was extracted using the QIAamp Fast DNA Stool Mini Kit (Tianjing, China). The V3-V4 regions of the bacterial 16 S rRNA gene were amplified using specifically primers 338 F(5’-ACTCCTACGGGAGGCAGCA G-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’). The PCR commercial were sequenced at a com-mercial lab (Shanghai Majorbio Biomedical Technology Co, China) using the Illumina Miseq platform.The optimized sequences were clustered into operational taxonomic units (OTUs) at 97% similarity. A random rarefaction of sample reads according to the smallest value was performed. The OTU sequences were taxonomically classified in phylum, class, order, family, genus, and species using the Silva database. The number of reads in the OTU table was standardized using the total sum approach to generate relative abundance (proportion). Sobs, Shannon, Simpson, Ace, Chao, and coverage indexes were calculated for alpha diversity evaluation for cecal microbiota. The principal coordinate analysis (PCoA) based on Bray Curtis dissimilarity metrics was conducted for beta diversity. The linear discriminant analysis (LDA) was performed to show the magnitude of the effect of the signature species identified between groups. The LDA threshold was set to 3.5 and subsequently applied for the Linear discriminant analysis Effect Size (LEfSe) analysis.

Statistical analyses

Graphs were prepared using GraphPad Prism 9.5.1 (GraphPad Software, San Diego, CA, USA) and Origin 2024 (OriginLab, Northampton, MA, USA). Data (apparent total tract digestibility, metabolism rates, serum biochemistry, antioxidant indices, hormones, pH, volatile fatty acids [VFA], and microbiota diversity metrics) were first preprocessed in Microsoft Excel (Microsoft, Redmond, WA, USA) and then analyzed in IBM SPSS Statistics v26.0 (IBM, Armonk, NY, USA). For serum biochemistry, antioxidant indices, hormones, pH, and VFA, one-way ANOVA was performed with treatment (Se level) as the fixed effect. ATTD and metabolism rates (nitrogen, energy, calcium, phosphorus) were analyzed using a two-way ANOVA with fixed effects of treatment and collection period (and their interaction). When a main effect or interaction was significant, means were separated using Duncan’s multiple range test. Results are reported as mean ± SD; statistical significance was declared at P < 0.05 and considered highly significant at P < 0.01 (two-tailed). Pearson rank correlation was used for correlation analyses, and correlation heatmaps were generated in Origin 2024.

Results

Apparent digestibility and metabolic rate of nutrients

The apparent digestibility and metabolic rate of nutrients for pregnant mares in each group are presented in Table 2. There were no significant changes in the digestibility of DE and P, ME, and the metabolic rates of Ca and P (P > 0.05). All experimental groups significantly increased the crude protein (CP) digestibility of mares (P < 0.05). Treatment groups II and III significantly increased the neutral detergent fiber (NDF) digestibility and nitrogen metabolic rate of mares (P < 0.05). Treatment group II significantly increased the dry matter (DM), acid detergent fiber (ADF), and calcium (Ca) digestibility of mares (P < 0.05).

Table 2.

Apparent digestibility and apparent metabolic rate of nutrients in pregnant mares of different treatment groups

Items	C	T1	T2	T3	SEM	P-value
						TRT	Date	TRT*Date
Apparent Digestibility of Nutrients
DM %	59.62b	60.03ab	61.63a	61.07ab	0.61	0.09	0.56	0.31
OM %	62.56	63.18	64.26	63.96	0.65	0.25	0.01	0.47
DE MJ/kg	10.49	10.41	10.61	10.54	1.54	0.89	0.13	0.89
CP %	64.54b	68.21a	69.48a	71.10a	0.99	0.00	0.00	0.72
NDF %	39.28b	42.21ab	45.05a	43.58a	1.05	0.00	< 0.01	0.17
ADF %	36.53b	38.21ab	41.36a	39.76ab	1.75	0.05	< 0.01	0.98
Ca %	47.54b	48.11ab	50.14a	49.45ab	0.79	0.09	0.65	0.31
P %	26.37	27.25	28.20	27.58	1.54	0.69	0.34	0.94
Apparent Metabolic Rate of Nutrients
ME MJ/kg	9.47	9.42	9.59	9.44	0.27	0.91	0.22	0.93
N %	34.31b	35.39b	42.00a	41.81a	1.41	< 0.01	0.03	0.51
Ca %	33.95	34.51	35.61	35.41	1.32	0.79	0.00	0.57
P %	22.36	23.24	25.44	24.59	1.01	0.15	0.15	0.29

Different lowercase superscripts within a row indicate significant differences (P < 0.05), and different uppercase superscripts indicate highly significant differences (P < 0.01); identical letters denote no significant difference; the same notation applies to the subsequent tables

Serum antioxidant indicators

The antioxidant indices in the serum of pregnant mares in each group are shown in Fig. 1. There were no significant changes in the SOD concentration in the serum of pregnant mares in each group (P > 0.05). Compared with the control group, all Treatment groups significantly increased the GSH-Px concentration in the serum of mares (P < 0.05). The T-AOC and CAT concentrations in the serum of mares in Treatment groups II and III were significantly higher than those in the control group, while the MDA concentration was significantly lower than that in the control group (P < 0.05).

Fig. 1.

Effects of different levels of L-SeMet on serum antioxidant indices in pregnant mares. (A) Total antioxidant capacity (T-AOC, U/mL). (B) Superoxide dismutase activity (SOD, U/mL). (C) Glutathione peroxidase activity (GSH-Px, U/mL). (D) Catalase activity (CAT, U/mL). (E) Malondialdehyde concentration (MDA, mmol/mL). Note: The symbol “*” indicates a significant difference between groups (LDS test, P < 0.05); same below

Serum estrogen levels

The estrogen levels in the serum of pregnant mares in each group are shown in Fig. 2. There were no significant changes in the concentrations of T, LH, and GnRH among the groups (P > 0.05). The concentrations of P and E₂ in the serum of mares in Treatment groups II and III were significantly higher than those in the control group (P < 0.05). The concentration of FSH in the serum of mares in Treatment group II was significantly higher than that in the control group (P < 0.05).

Fig. 2.

Effects of different levels of L-SeMet on serum estrogen in pregnant mares.(A) Testosterone (T, ng/mL). (B) Progesterone (P, ng/mL). (C) Estradiol (E₂, pg/mL). (D) Follicle-stimulating hormone (FSH, mIU/mL). (E) Luteinizing hormone (LH, mIU/mL). (F) Gonadotropin-releasing hormone (GnRH, pg/mL)

Levels of estrogen and conjugated estrogen in urine

The levels of estrogen and conjugated estrogen in the urine of pregnant mares in each group are shown in Fig. 3. There were no significant changes in the concentration of EQS in the urine of mares in any group (P > 0.05). All experimental groups significantly increased the concentration of ESS in the mares’ urine (P < 0.05). The concentration of 17α-DHEQS in the urine of mares in Treatment group II was significantly higher than that in the control group (P < 0.05). Treatment groups II and III significantly decreased the concentration of E₁ in the urine of mares compared to the control group (P < 0.05), with Treatment group II also significantly decreasing the concentration of E₂ in the mares’ urine (P < 0.05).

Fig. 3.

Effects of different levels of L-SeMet on estrogen and conjugated estrogen in the urine of pregnant mares. (A) E₁ (pg/mL). (B) E₂ (pg/mL). (C) ESS (µg/mL). (D) EQS (µg/mL). (E) 17α-DHEQS (µg/mL)

Fecal microbiota diversity

The volatile fatty acids (VFA) and pH in the feces of pregnant mares in each group are shown in Table 3. There were no significant differences in Acetate, Propionate, Butyrate, Isobutyrate, Valerate, Isovalerate, Total VFA, and pH (P > 0.05).

Table 3.

Effects of different levels of L-SeMet on fecal VFA and pH of pregnant mares (ug/ml)

Items	C	T1	T2	T3
Acetate	630.8 ± 163.37	686.76 ± 138.69	644.9 ± 179.55	635.05 ± 114.84
Propionate	189.62 ± 49.29	193.8 ± 34.84	194.59 ± 60.18	192.11 ± 47.41
Butyrate	54.31 ± 5.51	56.09 ± 5.31	57.45 ± 5.46	57.19 ± 5.42
Isobutyrate	74.68 ± 12.56	75.34 ± 15.72	82.46 ± 17.86	81.24 ± 22.33
Valerate	57.46 ± 4.43	55.57 ± 4.33	56.6 ± 7.35	55.24 ± 6.26
Isovalerate	39.7 ± 2.73	38.86 ± 2.39	39.65 ± 3.28	38.42 ± 1.90
Total VFA	1046.56 ± 221.62	1106.42 ± 189.86	1075.64 ± 261.93	1059.26 ± 178.71
pH	6.58 ± 0.24	6.48 ± 0.32	6.51 ± 0.31	6.54 ± 0.35

The Alpha-diversity of fecal microbiota in pregnant mares of each group is shown in Table 4. There were no significant differences in the chao1, dominance, goods_coverage, observed_otus, pielou_e, shannon, and simpson indices (P > 0.05).

Table 4.

Effects of different levels of L-SeMet on Alpha - diversity of pregnant mares

Items	C	T1	T2	T3
chao1	1879 ± 248.13	1748.76 ± 191.9	1909.99 ± 271.14	1958.58 ± 171.95
dominance	0.01 ± 0.01	0.01 ± 0.01	0.01 ± 0.00	0.01 ± 0.00
goods_coverage	1.00 ± 0.00	1.00 ± 0.00	1.00 ± 0.00	1.00 ± 0.00
observed_otus	1842.13 ± 249.7	1716.88 ± 188.73	1877.88 ± 268.92	1925.63 ± 168.84
pielou_e	0.84 ± 0.05	0.81 ± 0.04	0.85 ± 0.03	0.84 ± 0.03
shannon	9.07 ± 0.69	8.73 ± 0.52	9.22 ± 0.52	9.13 ± 0.39
simpson	0.99 ± 0.01	0.99 ± 0.01	0.99 ± 0.00	0.99 ± 0.00

The effects of different level of L-SeMet on the β-diversity of fecal microbiota in pregnant mares are shown in Fig. 4. As can be seen from Fig. 4, the addition of different levels of L-SeMet to the diet can affect the β-diversity of fecal microbiota in pregnant mares to varying degrees. In order to compare the similarities and differences of fecal microbiota in pregnant mares with different level of L-SeMet, Venn diagram analysis and UPGMA cluster tree based on OUT were carried out. As can be seen from Fig.4 (A), on the left is the structure of the UPGMA cluster tree, and on the right is the distribution map of the relative abundance of species at the phylum level in each sample. The branch lengths of the Treatment Group II and Treatment Group III were shorter than those of the control group, indicating that the bacterial communities of the Treatment Group II and Treatment Group III were similar. Therefore, the Treatment Group II and Treatment Group III cluster together first, then cluster with the control group, and finally cluster with the Treatment Group I.

Fig. 4.

Effects of different levels of L-SeMet on the beta-diversity of fecal microbiota in pregnant mares. (A) UPGMA Cluster Tree Based on unweighted_unifrac Distance,(B) Venn Diagram Analysis Based on OUT,(C) and (D) Principal Component Analysis

As can be seen from the Venn diagram in Fig. 4 (B), the number of common OTUs in pregnant mares of each group was 2439. The number of unique OTUs in the control group was 27,420, in the Treatment Group I was 2629, in the Treatment Group II was 3357, and in the Treatment Group III was 2458. The number of unique OTUs in the Treatment Group II was higher than in the other groups, indicating that more bacterial species were detected in the feces of the II group.

The principal component analysis is shown in Fig. 4C and D. As can be seen from the figure, according to the results of PCoA (Principal Coordinate Analysis), the horizontal axis represents one principal component, contributing 7.24% to the sample variation, and the vertical axis represents other principal components, contributing 5.76% to the sample variation. The similarity of bacteria in each group was relatively high. The results of PCA (Principal Component Analysis) showed that the horizontal axis was the principal component, contributing 5.1% to the sample variance. The vertical axis represents the second principal component, contributing 4.76% to the sample variation. The bacteria in each group also had a relatively high similarity.

The results of the effects of different dietary L-SeMet levels on the relative abundance of fecal microbiota in pregnant mares are shown in Fig. 5. As shown in Fig. 5-A, the top 10 species in terms of relative abundance at the phylum level in pregnant mares were Synergistota, Halobacterota, Patescibacteria, Spirochaetota, Actinobacteriota, Verrucomicrobiota, Euryarchaeota, Bacteroidota, Proteobacteria, and Firmicutes. Among them, Firmicutes is the main dominant microbiota in each group, with abundances of 57.75%, 62.15%, 61.72%, and 60.75%, respectively. As shown in Fig. 6-A and 6-B, the relative abundance of Euryarchaeota in pregnant mares in Treatment Group I was significantly higher than that of the control group (P < 0.05), and the relative abundance of Actinobacteriota in Treatment Group II was significantly higher than that of Treatment Group III (P < 0.05), and extremely significantly higher than that of the control group and Treatment Group I (P < 0.01).

Fig. 5.

Effects of different levels of L-SeMet on the relative abundance of fecal microbiota in pregnant mares. A: Histogram of community structure at the phylum level; (B): Histogram of community structure at the genus level; (C): Circos graph of community structure at the phylum level; (D): Circos graph of community structure at the genus level

Fig. 6.

Differential bacteria in the feces of pregnant mares with different levels of L-SeMet. Relative abundances of (A) Euryarchaeota, (B) Acidobacteriota, (C) Methanobacteria, (D) Kiritimatiellae, (E) Methanobacteriales, (F) Methanobacteriaceae, (G) Methanobrevibacter, (H) Clostridium_butyricum, and (I) Lactobacillus_hayakitensis

As shown in Fig. 5-B, the top 10 species in terms of relative abundance at the class level in pregnant mares were Kiritimatiellae, Spirochaetia, Negativicutes, Coriobacteriia, Verrucomicrobiae, Methanobacteria, Bacteroidia, Bacilli, Gammaproteobacteria, and Clostridia. Among them, Clostridia is the main dominant microbiota in each group, with abundances of 47.82%, 49.28%, 52.31%, and 48.65%, respectively. As can be seen from Fig. 6-CA and Fig. 6-D, the relative abundance of Methanobacteria in pregnant mares in Treatment Group I was significantly higher than that of the control group (P < 0.05), and the relative abundance of Kiritimatiellae in Treatment Group II was significantly higher than that of Treatment Group I (P < 0.05).

As shown in Fig. 5C, the top 10 species in terms of relative abundance at the order level in pregnant mares were Coriobacteriales, Christensenellales, Verrucomicro-biales, Lactobacillales, Methanobacteriales, Bacillales, Bacteroidales, Lachnospirales, Pseudomonadales, and Oscillospirales. As can be seen from Fig. E, the relative abundance of Methanobacteriales in pregnant mares in the Treatment Group I was significantly higher than that of the control group (P < 0.05).

As shown in Fig. 5D, the top 10 species in terms of relative abundance at the family level in pregnant mares were Akkermansiaceae, p-251-o5, Streptococcaceae, Rikenellaceae, F082, Methanobacteriaceae, Oscillospiraceae, Planococcaceae, Lachnospiraceae, and Moraxellaceae. As can be seen from Fig. 6F, the relative abundance of Methanobacteriaceae in pregnant mares in the Treatment Group I was significantly higher than that of the control group (P < 0.05).

As shown in Fig. 5E, the top 10 species in terms of relative abundance at the genus level in pregnant mares were UCG-002, Psychrobacter, Christensenellaceae R-7_group, Lachnospiraceae UCG-009, Akkermansia, Streptococcus, Rikenellaceae RC9_gut_group, Methanobrevibacter, and Acinetobacter. As can be seen from Fig.6G, the relative abundance of Methanobrevibacter in pregnant mares in the Treatment Group I was significantly higher than that of the control group (P < 0.05).

As shown in Fig. 5F, the top 10 species in terms of relative abundance at the genus level in pregnant mares were Ruminococcus_sp_HUN007, Lactobacillus_hayakitensis, Clostridiales_bacterium_Firm_14, Ruminococcaceae_bacterium_AB4001, Ruminococca-ceae_bacterium_AE2021, rumen_bacterium_NK4A65, Clostridium_butyricum, Bacteroi-dales_bacterium_Bact_22, Lactobacillus_equi, and Acineto66bacter_movanagherensis. As can be seen from Fig. 6H and I, the relative abundance of Clostridium_butyricum in pregnant mares of the Treatment Group I was significantly higher than that of the Treatment Group III (P < 0.05), and extremely significantly higher than that of the control group (P < 0.01). The relative abundance of Lactobacillus_hayakitensis in the Treatment Group III was significantly higher than that of the Treatment Group I, and extremely significantly higher than that of the control group and Treatment Group I.

The T-test for intergroup differences and LEfSe analysis of fecal microbiota are shown in Fig.6 and Fig. 7. In order to specifically identify the species with significant differences in abundance changes between groups, a T-test was first carried out. The results showed that there was one microorganism that was significantly upregulated at the Phylum, Class, Order, and Family levels. To further search for biomarkers with statistical differences between groups, an LDA Effect Size analysis was conducted. There were 7 species with significant differences among the groups, and 6 species with differences in the Treatment Group II, namely Actinobacteriota, Actinobacteria, Actinomyces, Rhodococcus, Nocardiaceae, and Actinomycetales. The species with differences in the Treatment Group I was s_bacterium_XPD3003.

Fig. 7.

Analysis of T-test intergroup differences and LDA analysis for pregnant mares in each group. (A) Cladogram showing the differentially abundant taxa between groups; (B) LDA score (log10) histogram of discriminative taxa identified by LEfSe. o_: order; c_: class; g_: genus; f_: family; p_: phylum

The results of the effects of different L-SeMet levels on the functional prediction of fecal microbiota in pregnant mares are shown in Fig.8 . As can be seen from Fig.8, different L-SeMet levels have a significant impact on the functions of fecal microbiota in pregnant mares.

Fig. 8.

Tax4Fun Functional Prediction of Bacteria in the Feces of Pregnant Mares in Different L-SeMet Level Treatment Groups. (A): β-diversity analysis, (B): Venn diagram,(C): Level 2 functional prediction,(D): Level 2 functional clustering heatmap, (E): KEGG pathway annotation

From the principal component analysis diagram (Fig. 8A), the contribution value of the first principal component to the sample differences was 32.17%, and the contribution value of the second principal component to the sample differences was 20.45%, indicating a large inter group difference.

As shown in Fig. 8B, the number of unique OTUs in the fecal microbiota of pregnant mares in the control group is 5, in the Treatment Group I was 10, in the Treatment Group II was 1, and in the Treatment Group III was 3.

As can be seen from Fig. 8C and D, at the Level 2 functional level, the fecal microbiota of pregnant mares under different L-SeMet supplementation levels mainly exhibited functions related to energy metabolism, signal transduction, membrane transport, nucleotide metabolism, replication and repair, translation, glycan biosynthesis and metabolism, metabolism of cofactors and vitamins, amino acid metabolism, and carbohydrate metabolism. Specifically, in Treatment Group I, the predominant functions of the fecal microbiota were focused on signal transduction, membrane transport, replication and repair, and translation. In Treatment Group II, the primary microbial functions were centered on energy metabolism. In Treatment Group III, the fecal microbiota functions were mainly concentrated on membrane transport, nucleotide metabolism, and replication and repair.

As can be seen from Fig. 8E, the functional information obtained from the KEGG comparison based on the feces of pregnant mares contains a total of 43 metabolic functions. Among them, the KEGG metabolic pathway analysis showed that the top 5 secondary functions are Carbohydrate metabolism, Membrane transport, Translation, Replication and repair, and Amino acid metabolism.

Correlation analysis

The correlation analysis is shown in Fig. 9. To further explore the relationships among the fecal microbiota of pregnant mares, phenotypic data, and the hormonal levels in serum and urine, a correlation analysis was conducted using the Pearson correlation coefficient as a measure.

Fig. 9.

Correlation analysis. (A) Correlation analysis between differential bacteria and serum biochemical indicators,(B) Correlation analysis between differential bacteria and serum antioxidant indicators,(C) Correlation analysis between differential bacteria, volatile fatty acids, and pH,(D) Correlation analysis between hormonal levels in the serum and urine of pregnant mares in the group,(E) Correlation analysis between functional prediction and serum antioxidant indicators,(F) Correlation analysis between functional prediction and serum antioxidant indicators,(G) Correlation analysis between functional prediction, volatile fatty acids, and pH. (H) Correlation Analysis of Serum Antioxidant Indicators, Volatile Fatty Acids, and pH. “*” indicates a significant correlation (P≤0.05), “**” indicates an extremely significant correlation (P≤0.01). Blue and red colour gradients indicate a positive or negative in correlation coefficient, respectively

As can be seen from Fig. 9A, ALB, AST, and c_Actinobacteria showed a significant positive correlation (P < 0.05); ALT was significantly positively correlated with o_Actinomycetales, g_Actinomyces, and s_bacterium_XPD3003 (P < 0.05); Mg²⁺ was significantly positively correlated with s_bacterium_XPD3003 (P < 0.05); Ca²⁺ was significantly positively correlated with f_Nocardiaceae (P < 0.05); CREA was significantly negatively correlated with c_Actinobacteria (P < 0.05).

As shown in Fig. 9B, SOD was significantly positively correlated with s_Clostridium_butyricum and was significantly negatively correlated with c_Actinobacteria (P < 0.05); MDA was significantly negatively correlated with c_Kiritimatiellae (P < 0.05).

From Fig. 9C, Butyrate was significantly positively correlated with o_Actinomycetales and g_Actinomyces (P < 0.05); Valerate was significantly positively correlated with c_Kiritimatiellae, while being significantly negatively correlated with s_bacterium_XPD3003 (P < 0.05); c_Actinobacteria was extremely significantly positively correlated with Isobutyrate (P < 0.01) and significantly negatively correlated with pH (P < 0.05).

As indicated in Fig. 9D, ESS and GnRH were significantly positively correlated (P < 0.05); E₂ in the urine of pregnant mares was significantly negatively correlated with E₂ in the serum (P < 0.05), and was extremely significantly negatively correlated with LH (P < 0.01).

As shown in Fig. 9E, TP and GLB were extremely significantly positively correlated with Translation and Nucleotide_metabolism (P < 0.01), and were significantly negatively correlated with Signal_transduction (P < 0.05); Replication_and_repair was extremely significantly positively correlated with TP (P < 0.01) and significantly positively correlated with GLB (P < 0.05); CREA was significantly positively correlated with Signal_transduction (P < 0.05); UREA was significantly positively correlated with Ami-no_acid_metabolism (P < 0.05); T-Bil and Ca²⁺ were significantly positively correlated with Glycan_biosynthesis_and_metabolism (P < 0.05); ALT was significantly positively correlated with Membrane_transport (P < 0.05), and was significantly negatively correlated with Glycan_biosynthesis_and_metabolism (P < 0.05); ALP was significantly positively correlated with Translation and Nucleotide_metabolism (P < 0.05), CK was extremely significantly positively correlated with Translation (P < 0.01), and was significantly positively correlated with Nucleotide_metabolism (P < 0.05), while being significantly negatively correlated with Signal_transduction (P < 0.05); AST/ALT was significantly positively correlated with Glycan_biosynthesis_and_metabolism (P < 0.05).

From Fig. 9F, there was no correlation among the indicators of each group (P > 0.05).

As shown in Fig. 9G, Butanoic acid was significantly negatively correlated with Glycan_biosynthesis_and_metabolism (P < 0.05).

As can be seen from Fig. 9H, CAT was significantly positively correlated with Butyrate (P < 0.05).

Discussion

Gestation is the process from conception to parturition in the female body. The gestation period of horses is generally about 11 months and is divided into two stages: the early gestation stage and the late gestation stage. The first eight months constitute the early gestation stage, mainly involving the development of the mare herself, as well as the uterus, placenta, and amniotic membranes. The last three months are the late gestation stage [31]. The late gestation stage is the primary period for fetal weight gain. During this stage, due to the rapid development of the fetus, the gastrointestinal tract is compressed, resulting in a decrease in the dry matter intake of the dam [32]. Meanwhile, the dam’s demand for nutrients increases significantly. The high metabolic state of gestational tissues (fetus and mammary glands) can generate large amounts of reactive oxygen species (ROS), causing oxidative stress [33], which damages the structure and function of cells and membranes, thereby adversely affecting fetal growth and development [34]. Therefore, improving the oxidative stress status of mares during this stage and increasing nutrient intake are crucial for maintaining normal pregnancy and meeting fetal demands. NRC (2007) [35] pointed out that for a 400 kg pregnant mare, nutrient requirements increase progressively during gestation. In the first five months, the DE was 13.3 Mcal, CP was 504 g, Lys was 21.7 g, Ca was 16.0 g, and P was 11.2 g. By the 11th month, DE rose to 17.1 Mcal, CP to 714 g, Lys to 30.7 g, Ca to 28.8 g, and P to 21.0 g. Under the same dietary composition and nutritional levels, we supplemented L-SeMet to 400 kg pregnant mares. The results showed that L-SeMet supplementation improved the apparent nutrient digestibility and metabolic rate of the mares. Dietary additions of 0.4 mg Se, 0.6 mg Se, and 0.8 mg Se as L-SeMet significantly increased crude protein (CP) digestibility and calcium metabolic rate. Additions of 0.6 mg Se and 0.8 mg Se significantly improved NDF and ADF digestibility and nitrogen metabolism rate. Dry matter (DM) digestibility in the 0.6 mg Se·mare L-SeMet group was significantly higher than in the control group. Shi [36] and Wang [37] studied the effects of selenium on feed conversion rate and rumen fermentation and found that selenium can stimulate rumen microbial activity and enzyme activity, increase dry matter intake, and improve rumen fermentation and feed utilization. When animals are exposed to adverse environmental factors, their feed intake may decrease or stop. The L-SeMet stored in the body releases selenium during protein catabolism, providing a selenium source for synthesizing glutathione peroxidase and other selenoproteins, thus preventing selenium deficiency. In this study, dry matter intake in the experimental groups increased to varying degrees, and the intake in experimental group II was significantly higher than that in the control group. This may be related to selenium’s ability to enhance certain intestinal digestive enzyme activities, improve the digestive and absorptive capacity of intestinal epithelial cells, and thus increase the mare’s appetite. Li et al. [38] found that adding selenium to sow diets increased the activities of GSH-Px, T-SOD, and CAT in the intestine, elevated serum immunoglobulin levels and expression of immune-related selenoprotein genes, reduced MDA and pro-inflammatory factors, and promoted the growth of beneficial intestinal bacteria. In herbivorous monogastric animals, the cecum is the main site of fiber digestion. Selenium may improve the digestibility of fibrous substances in pregnant mares by promoting the proliferation of beneficial bacteria, such as Kiritimatiellaeota, in the cecum, thereby altering the cecal microbiota and enhancing ADF and NDF digestibility [39]. During late gestation, protein is the most important nutrient for maternal health and fetal development. Nitrogen metabolism reflects the efficiency of dietary protein utilization by animals. In this study, L-selenomethionine supplementation increased CP digestibility and nitrogen metabolism rate in mares. This may be because selenomethionine enhanced the activity of protein-digesting enzymes or improved intestinal health, thus boosting crude protein digestion and absorption efficiency, reducing protein breakdown and nitrogen excretion, and improving nitrogen utilization efficiency [40, 41].

Oxidative stress refers to the host’s endogenous antioxidant protection mechanisms, which are activated upon exposure to stressors or xenobiotics to help scavenge reactive oxygen species (ROS) [42, 43]. Selenium is an essential trace element in humans and animals, and it is a key component of glutathione peroxidase, thyroid hormone deiodinase, selenoprotein K, and others. Selenium functions by catalyzing the reaction of H₂O₂ and fatty acid alcohols, thus helping to protect cell membranes from oxidative damage. Glutathione peroxidase (GSH-Px) is the first selenium-containing enzyme proven to prevent oxidative damage to cell membranes, and its activity serves as a marker of serum selenium utilization. Shi et al. [36] pointed out in their study that adding organic selenium to the diet of late-pregnant Taihang black goats not only increased the activity of GSH-Px, superoxide dismutase (SOD), and total antioxidant capacity (T-AOC) in the serum of ewes, but also stimulated the production of estradiol, progesterone, and T4, while promoting the metabolism of major nutrients. The results of this study show that, compared to the control group, adding selenium at three different levels significantly increased the activity of GSH-Px. However, it is worth noting that the activity of GSH-Px in the Experimental Group II was higher than that in Experimental Group I and Experimental Group III, though the difference was not statistically significant. Therefore, for pregnant mares in the Zhaosu area of Xinjiang, the optimal selenium supplementation amount is 0.3 g/d·mare of L-SeMet. Supplementing with 0.2 g/d·mare may not meet the selenium needs of pregnant mares, while 0.4 g/d·mare may lead to selenium overdose and potential inhibitory effects. SOD is not only a superoxide anion-scavenging enzyme but also primarily generates H₂O₂, playing a key role in the biological antioxidant system. Various antioxidants and antioxidant enzymes in the serum constitute the total antioxidant capacity. The level of lipid peroxidation can be reflected by measuring MDA levels. Therefore, these indicators indirectly reflect the impact of selenium on the antioxidant capacity of pregnant mares.The results of this study show that adding different levels of L-SeMet to the diet increased the activity of SOD in the serum of mares, significantly enhanced the activities of T-AOC, GSH-Px, and catalase (CAT), and significantly reduced the MDA activity in the serum. The enhancement of these enzyme activities further strengthens the antioxidant defense system of mares, helping to resist the potential harm caused by oxidative stress. Additionally, the study observed that supplementing medium- and high-doses of L-SeMet significantly reduced MDA activity in the serum of mares. MDA is a product of lipid peroxidation, and the decrease in its content indicated that the degree of lipid peroxidation in mares has been effectively controlled, thus further protecting the integrity and stability of cell membranes [44].

Reproductive hormones such as progesterone and estrogen play a crucial and continuous role throughout the pregnancy of mares. Selenium, a trace element, significantly impacts animal reproductive performance. During pregnancy, it can increase the levels of reproductive hormones, reduce the risk of miscarriage, improve fetal survival rates, and positively influence fetal development and growth. Between 140 and 170 days of pregnancy, due to the disappearance of the endometrial cups and the eventual reduction of eCG in maternal serum, the plasma LH concentration begins to decline. It starts to rise steadily again around 210 days of pregnancy, peaks around 280 days, and then decreases after 280 days, remaining at a relatively low level until just before parturition. The change in FSH concentration follows a similar pattern, reaching its maximum around 280 days of pregnancy before gradually decreasing [45, 46]. GnRH is secreted at low levels during the middle and late stages of pregnancy. At this time, the mare’s reproductive system is primarily regulated to maintain pregnancy. The stable, low-level secretion of GnRH helps to keep the ovaries in a relatively quiescent state, ensuring that the pregnancy process is not disrupted. Ren Youshe [47] demonstrated in his research that the addition of selenium significantly affects the concentration of FSH during late pregnancy but does not significantly impact LH concentration. In this study, the addition of L-SeMet significantly influenced the FSH level in the serum but had no significant effect on the LH level, which aligns with previous findings. This may be because selenium promotes the secretory function of adenohypophysis cells. Selenium participates in intracellular redox reactions, maintains redox balance within pituitary cells, affects their functional state, and enhances their sensitivity to GnRH, thereby promoting FSH synthesis and secretion [48]. Lekatz et al. [49] showed that increasing dietary selenium in ewes during mating and pregnancy led to higher serum progesterone levels on the 60th, 90th, and 110th days of pregnancy compared to the control group, thus protecting the fetus. In this study, the serum P content in mares in Trial II group and Trial III group increased significantly. ROS can damage the luteinizing hormone (LH) receptor and inhibit the transfer of cholesterol to the mitochondria for progesterone synthesis [50]. Selenium plays a vital role in the luteal function of mares. The corpus luteum is the main site of progesterone secretion. Selenium promotes progesterone secretion by influencing luteal cell function. Adequate selenium increases GSH-Px activity in luteal cells, reduces oxidative stress-induced luteal cell damage, maintains normal luteal cell metabolism and function, and ensures the continuous secretion of progesterone and other progestogens, helping to maintain pregnancy in mares [51].

Estrogen levels continue to rise in the middle of pregnancy. Around 100–150 days of pregnancy, estrogen levels accelerate. Estrogen promotes angiogenesis in the uterus and placenta, increases blood flow in the uteroplacenta, and ensures that the fetus receives adequate oxygen and nutrients. As the luteal-placental transition progresses, the placenta takes over estrogen secretion. Estrogen levels begin to rise again around the 100th day of pregnancy, reach their peak around the 210th day, and then slowly decrease [52, 53]. Makkawi et al. [54]. showed that adding selenium to the diet of Awassi sheep increases white blood cell count, protects gonadotropin receptors from oxidative damage, and significantly improves estrogen secretion. In this study, E2 levels in the serum of mares in Trial II and Trial III groups increased significantly. This may be because selenium regulates estrogen metabolism. Research suggests that an appropriate amount of selenium affects enzymes related to estrogen metabolism in the liver, such as the cytochrome P450 enzyme system. Among these, P450scc (cholesterol side-chain cleavage enzyme, CYP11A1) is a key enzyme in initiating estrogen synthesis [55]. The estrogens in the urine of pregnant mares mainly exist in the form of conjugated estrogens, including ESS, EQS, and 17α-DHEQS. Studies have shown that the content of conjugated estrogens is at least 100 times that of their unconjugated forms [56], with ESS accounting for the highest proportion in the urine of pregnant mares, reaching over 50% [57]. The sulfation metabolic pathway of estrogens primarily depends on estrogen sulfotransferase (EST), which converts free estrogens into conjugated estrogens, enhancing their water solubility and promoting metabolic excretion [58]. The key donor in the sulfation process is 3′-phosphoadenosine-5′-phosphosulfate (PAPS), which is generated from ATP and sulfate ions under the catalysis of an enzyme to form adenosine-5′-phosphosulfate (APS) and PAPS [59]. Trace elements, especially copper, zinc, iron, manganese, and selenium, play important roles in estrogen metabolism. Li Xiaobin et al. [60]. showed that adding these trace elements to the diet of pregnant mares significantly increased the levels of ESS, EQS, and 17α-DHEQS in their urine. This may be because trace elements promote the generation of free sulfate ions in body tissues, which in turn increases the synthesis of PAPS, thereby enhancing the sulfation efficiency of estrogens. This study further confirms the role of L-SeMet in the metabolism of conjugated estrogens in pregnant mares, with results showing that L-SeMet supplementation significantly increased the content of conjugated estrogens in the urine. Selenium is closely related to the metabolism of sulfur-containing amino acids, which produce sulfate ions during the metabolic process [61]. These sulfate ions can further convert into PAPS during the ATP-mediated biosynthesis process and serve as a sulfur donor to promote the estrogen sulfation reaction. Moreover, selenium may also enhance the rate of estrogen conjugated metabolism by regulating the gene expression of estrogen sulfotransferase (EST). Studies have shown that selenium regulates the expression of EST mRNA in the liver, with the EST mRNA levels in selenium-deficient rats being only 7.3% of the normal levels [62], indicating that selenium may indirectly enhance the sulfation efficiency of estrogens by promoting EST transcription and translation. On the other hand, the antioxidant effects of selenium may also play an indirect regulatory role in the metabolism of conjugated estrogens. During pregnancy, the enhanced metabolic activity, rapid growth of the placenta, uterus, and fetus lead to elevated oxidative stress levels, which may inhibit the activity of enzymes related to estrogen metabolism, thereby affecting hormonal balance [63]. As a key component of GSH-Px and Epp1, selenium can effectively scavenge excess reactive oxygen species (ROS), reduce lipid peroxidation damage, and maintain the stability of estrogen metabolic pathways. Therefore, supplementation with L-SeMet may indirectly increase the efficiency of PAPS production and the metabolic rate of estrogen sulfation by reducing oxidative stress, thus raising the levels of conjugated estrogens and promoting the clearance of free estrogens. Additionally, selenium’s effect on placental function and maternal hormonal regulation may be one of the potential mechanisms for promoting the metabolism of conjugated estrogens. The accumulation of selenium in placental tissues and its regulatory effect on placental endocrine function have been confirmed in several studies [64]. Selenium may enhance the activity of sulfation enzymes in placental tissues, improve the metabolic capacity for estrogen sulfation, and accelerate the degradation and excretion of estrogens by the placenta, thereby lowering the levels of free estrogens in maternal blood.

Changes in the composition and abundance of the microbial community can have either beneficial or harmful effects on the body [65]. In the intestine, many bacteria are beneficial to the host and play important roles in multiple physiological processes such as nutrient absorption, metabolism, immune homeostasis, and maintenance of overall health [66]. The microbial community regulates these processes by producing different metabolites, such as short-chain fatty acids (acetic acid, propionic acid, and butyric acid), which are mainly produced by the fermentation of anaerobic bacteria in the large intestine [67]. In this study, both short chain fatty acids and total volatile fatty acids increased at different levels, indicating that the addition of selenium to the diet may promote the proliferation of microorganisms in the large intestine and improve intestinal fermentation. Research has pointed out that organic selenium can affect the α-diversity of rectal microbiota at the class, order, family, genus, and species levels [68], and increase the relative abundances of Lachnospiraceae, Carnobacterium, Dysgonomonas, Lachnospiraceae XPB1014, and Rikenellaceae RC9 gut group in the rumen microbiota. Moreover, Tax4Fun metagenomic analysis showed that the activities of genes related to carbohydrates and other amino acids and the metabolic pathways in the rumen of sheep in the Se-supplemented group were improved [69]. The results of this experiment showed that 0.4 mg Se of selenium significantly increased the relative abundances of methanogens in the intestine [70, 71], including Euryarchaeota, Methanobacteriales, Methanobacteria, Methanobacteriaceae, and Methanobrevibacter, as well as Clostridium butyricum. Most selenoproteins in archaea are involved in methanogenesis [72]. Hendawy et al. [73] believed that selenium can change the activity and composition of digestive tract microorganisms, thereby increasing the bio-energy supply of the animal body. Meanwhile, intestinal microorganisms may compete with methanogens for hydrogen utilization of selenium as an energy source, thus affecting the production of short chain fatty acids, mainly the production of butyrate [74, 75]. In this study, the functions of intestinal microorganisms in pregnant mares in the 0.4 mg/Se group were mainly concentrated in carbohydrate metabolism. This may be related to the fact that methanogens in the intestine can produce methane using hydrogen and carbon dioxide, and the addition of selenium may enhance its carbohydrate metabolism process [76, 77]. The relative abundance of Actinobacteriota in the intestine of pregnant mares in the 0.6 mg·Se·mare of L-SeMet group was significantly higher than that in the control group. Actinobacteriota can participate in the host’s metabolic processes by producing bioactive substances [78]. The mechanism is that after acetic acid produced by Actinobacteriota is absorbed by the colon, it can cross the blood-brain barrier, activate hypothalamic neurons, activate the parasympathetic nerve, and finally stimulate the release of neurohormones, thus increasing the animal’s feed intake [79]. This is consistent with the upward trend of dry matter intake and acetic acid content in pregnant mares in the 0.6 mg·Se·mare of L-SeMet group. At the same time, combined with functional prediction, it was found that the fecal microbial functions of pregnant mares in the 0.6 mg·Se·mare of L-SeMet group were mainly concentrated in aspects such as energy metabolism. Therefore, these results suggest that adding selenium may enhance the activities of beneficial microorganisms such as Actinobacteriota and their ability to produce bioactive substances. These bioactive substances may have various functions such as improving the animal’s digestive tract environment, increasing feed utilization rate, and promoting animal health, thus indirectly or directly increasing the animal’s feed intake. This positive effect of selenium may help optimize the nutrient intake of pregnant mares, supporting their physiological needs and the normal development of the fetus. Research has pointed out that Kiritimatiellaeota can participate in carbohydrate metabolism [80] and utilize formate, glucose, and hydrogen to reduce methane production [81]. The reason why the relative abundance of Kiritimatiellaeota in the fecal microbiota of the 0.6 mg·Se·mare of L-SeMet group was higher than that in the control group and other experimental groups may be that the addition of selenium not only affected the activities of microbiota such as Actinobacteriota but also influenced microbiota such as Kiritimatiellaeota that are involved in fiber digestion. At the same time, this is consistent with the results of the apparent digestibility of ADF and NDF in nutrient digestion and metabolism [82]. In conclusion, adding selenium to the diet can regulate the intestinal microbial community of pregnant mares in the middle and late stages of pregnancy, optimize nutrient absorption and metabolism, and contribute to the health of pregnant mares and the development of the fetuses.

Conclusions

Compared with the control group, adding 0.6 mg/d·mare of Se to the diet of pregnant mares in the midlate pregnancy stage can increase the dry matter intake of pregnant mares, as well as the apparent digestibility of DM, CP, NDF, and ADF, and improve the nitrogen metabolism rate. It can improve the oxidative status of mares by increasing the activities of T-AOC, GSH-Px, and CAT in the blood, and by reducing the activity of MDA. It can also maintain the hormone levels of pregnant mares in the mid-late pregnancy stage by reducing the levels of E₁ and E₂ in the blood and increasing the contents of ESS and 17α-DHEQS. In addition, selenium supplementation can optimize the intestinal microbial community by improving the hindgut fermentation status of mares.

Authors’ contributions

ML and JL: Contributed significantly to article conception and design, data acquisition, data analysis, and interpretation.CM , GW and QH : Performed animal feeding and sample collection. All authors read and approved the final manuscript.XL: Participated in critical revisions of important knowledge and content in the manuscript.

Funding

This research was funded by the Major Science and Technology Special Project of Xinjiang Uygur Autonomous Region (2022A02013-2-2), the Undergraduate Innovation Project of Xinjiang Agricultural University (2024), and the Postgraduate School-level Scientific Research Innovation Program of Xinjiang Agricultural University (XJAUGRI2024027).

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

All animal procedures conducted in this study strictly followed the guidelines for the ethical treatment of animals in scientific research. The entire experimental protocol was reviewed and approved by the Animal Experiment Ethics Committee of Xinjiang Agricultural University (Permit Number: 2020024). Before the commencement of the study, all pregnant mare owners were thoroughly informed of the purpose, procedures, potential risks, and benefits of the study. They voluntarily signed informed consent forms, agreeing to the inclusion of their animals in the research. All efforts were made to ensure the welfare and humane treatment of the animals throughout the experimental period.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.