L-Cit, (purity ≥ 98%) was obtained from Hebei Hongtao Biological Engineering Co., Ltd. Cloprostenol Injection (PG, 0.2 mg per vial) was purchased from Ningbo No.2 Hormone Factory. Enzyme-linked immunosorbent assay (ELISA) kits for gonadotropin-releasing hormone (GnRH, catalog number: ml410381), follicle-stimulating hormone (FSH, catalog number: ml497803), luteinizing hormone (LH, catalog number: ml440758), and estradiol (E₂, catalog number: ml450381) were sourced from Shanghai Enzyme-linked Biotechnology Co., Ltd.

The equipment included a veterinary B-mode ultrasound scanner (Kaixin Brand, Model RKU 10, Xuzhou Kaixin Technology Co., Ltd.), a microplate reader (Model iMark, Bio-Rad Laboratories, Inc.), a centrifuge (Model 5,424, Eppendorf AG, Germany), insemination gun, vacuum blood collection tubes, electronic balance, disposable syringes, gloves, and long-arm disposable gloves.

On Days 1 and 11, prior to morning feeding, six cows from each group were randomly selected, and 10 mL of blood was collected from the caudal vein using plain blood collection tubes. To prevent metabolite degradation, the blood samples were centrifuged at 720 × g within 30 min of collection. The supernatant (serum) was then aliquoted into 5 mL cryopreservation tubes within 1 h after centrifugation, flash-frozen in liquid nitrogen, and stored for subsequent analysis.

After Day 10, six cows from each group were randomly selected. At 10 h post-mounting, 10 mL of rumen fluid was collected by repeatedly inserting and repositioning a rumen cannula sampler at multiple sites within the rumen digesta. Concurrently, 10 g of fecal samples were obtained via rectal collection. All samples were immediately aliquoted into multiple 5 mL cryopreservation tubes, flash-frozen in liquid nitrogen, and subsequently transferred to a − 80 °C freezer for long-term storage pending microbiome analysis.

Serum levels of GnRH, FSH, LH, and E₂ were quantified using ELISA kits. Liquid chromatography-mass spectrometry (LC–MS) analysis of serum metabolites was performed by Beijing Huatai Boao Biological Gene Co., Ltd., following the protocol described by Klepacki et al. (2016). Sequencing of 16S rRNA in rumen fluid and fecal microbiota was also conducted by the same company. Polymerase chain reaction (PCR) amplification employed bacterial primers (338F: 5’-ACTCCTACGGGAGGCAGCAG-3′ and 806R: 5’-GGACTACHVGGGTWTCTAAT-3′), and the detection procedure followed the method of Klepacki et al. (2016).

Table 2 shows that no significant variation was detected in any alpha diversity indices between the control and treatment groups, indicating that L-Cit supplementation did not alter ruminal microbial diversity. Sample coverage across all groups exceeded 99.9%, confirming that sequencing depth was adequate to capture the microbial composition of rumen fluid with high reliability. Furthermore, as depicted in Figure 3, the sample points from different groups largely overlapped, and their confidence ellipses intersected. A PERMANOVA (Adonis) test yielded a p-value greater than 0.05, indicating no significant difference in the overall microbial community structure among the three groups.

Figure 3 presents the PLS-DA (Partial Least Squares-Discriminant Analysis) score plots, which visually represent the classification efficiency of the model. In the comparison between the Control Group (CON) and Experimental Group I, the degree of separation between sample clusters reflected clear group discrimination, with the first and second principal components explaining 41.7 and 8.12% of the variance, respectively.

Similarly, in the comparison of CON with Experimental Group II, the sample distribution also demonstrated distinct segregation, with explained variances of 42.67% for the first principal component and 7.04% for the second.

Figure 4A illustrates the phylum-level distribution of ruminal microbiota under L-Cit supplementation. The 10 most abundant phyla were Bacteroidota, Firmicutes, Proteobacteria, Verrucomicrobiota, Patescibacteria, Fibrobacterota, Desulfobacterota, Spirochaetota, Cyanobacteria, and Actinobacteriota. Bacteroidota and Firmicutes represented the predominant phyla across all groups.

Notably, Bacteroidota exhibited the highest relative abundance in Experimental Group II (52.16%), whereas Firmicutes were most abundant in the CON group (39.87%).

As shown in Figure 4B, analysis at the family level revealed that the 10 most abundant ruminal taxa included Prevotellaceae, Rikenellaceae, Lachnospiraceae, F082, Oscillospiraceae, Bacteroidales_RF16_group, Succinivibrionaceae, Acidaminococcaceae, Hungateiclostridiaceae, and Ruminococcaceae. Prevotellaceae, Ruminococcaceae, and Lachnospiraceae represented the dominant families in both the control (CON) and experimental groups, with higher relative abundances observed in Experimental Group I.

Prevotellaceae and Ruminococcaceae reached 20.38 and 12.25%, respectively. In contrast, Firmicutes exhibited a higher relative abundance in the CON, reaching 10.36%.

At the genus level Figure 4C, the top 10 genera were Prevotella, Rikenellaceae_RC9_gut_group, F082_norank, Bacteroidales_RF16_group_norank, Prevotellaceae_UCG-001, Succiniclasticum, Saccharofermentans, UCG-011_norank, Prevotellaceae_UCG-003, and Muribaculaceae_norank. Among them, Prevotella, Rikenellaceae_RC9_gut_group, and F082_norank were particularly enriched in Experimental Group I, indicating their dominance within the ruminal microbiota.

Prevotella reached the highest relative abundance in Experimental Group I at 23.42%. Ruminococcus also displayed higher levels in the experimental groups compared with the CON, with relative abundances of 12.47% in Experimental Group I, 11.74% in Experimental Group II, and 8.37% in the CON.

Figure 4D illustrates that 12 microbial taxa displayed significant differences between the control group and the experimental groups, comprising 1 phylum, 1 class, 2 orders, 2 families, and 6 genera. In the control group, 7 taxa were enriched, including Verrucomicrobiota (phylum), Lentisphaeria (class), Oligosphaeraceae (family), Oligosphaerales (order), g_horsej_a03, Paeniclostridium (genus), and g_Z20. Experimental Group I exhibited 3 taxa, namely Victivallales (order), vadinBE97_g_norank, and vadinBE97, whereas experimental Group II contained 2 taxa, Eubacterium_ruminantium_group and Prevotellaceae_g_norank. Effects of dietary supplementary feeding of L-Citrulline on intestinal microbiota of Simmental cows.

Table 3 shows that no significant differences were observed in any alpha diversity indices of the intestinal microbiota between the control and treatment groups, indicating that L-Cit supplementation did not alter the overall intestinal microbial diversity. The species coverage in all groups exceeded 99.9%, confirming that the sequencing depth was sufficient to adequately capture the microbial composition of the fecal samples. As illustrated in Figure 5, sample points from different groups exhibited extensive overlap, and their confidence ellipses intersected. A PERMANOVA (Adonis) test yielded a p-value greater than 0.05, indicating no significant difference in the microbial community structure among the three groups.

Figure 5 presents the PLS-DA score plot, which visually represents the classification performance of the model. Greater separation among the three groups corresponds to more distinct classification. The first principal component accounted for 33.68% of the variance, while the second explained 5.23%.

As illustrated in Figure 6A, the dietary supplementation of L-Cit influenced the intestinal microbiota composition of Simmental cows at the phylum level. The eight phyla with the highest relative abundance were Firmicutes, Bacteroidota, Spirochaetota, Proteobacteria, Actinobacteriota, Verrucomicrobiota, and Cyanobacteria (note: although described as “top 8,” only seven phyla are listed, consistent with the original text).

Firmicutes and Bacteroidota exhibited consistently high proportions across the Control Group, Experimental Group I, and Experimental Group II, thereby representing the dominant phyla. In Experimental Group II, the relative abundance of Firmicutes reached 66.24%, exceeding the levels observed in the other groups. Conversely, Bacteroidota demonstrated the greatest relative abundance in the Control Group, reaching 30.77%, which was higher than that observed in the experimental groups.

As shown in Figure 6B, the analysis of dietary L-Cit supplementation on the intestinal microbiota of Simmental Cows at the family level revealed that the top 10 families in relative abundance were Oscillospiraceae, Lachnospiraceae, Rikenellaceae, UCG-010, Prevotellaceae, Peptostreptococcaceae, coprostanoligenes_group, Bacteroidaceae, Christensenellaceae, and Ruminococcaceae.

Oscillospiraceae, Lachnospiraceae, and Rikenellaceae represented the major families across the Control Group, Experimental Group I, and Experimental Group II. Oscillospiraceae and Lachnospiraceae showed the highest proportions in Experimental Group II, with relative abundances of 15.92 and 12.21%, respectively. In contrast, Rikenellaceae displayed its highest relative abundance in Experimental Group I, accounting for 8.98%.

As illustrated in Figure 6C, analysis of the intestinal microbiota at the genus level following dietary supplementation with L-Cit in Simmental cows revealed that the 10 genera with the highest relative abundance were UCG-005, UCG-010_norank, Rikenellaceae_RC9_gut_group, coprostanoligenes_group_norank, Bacteroides, Prevotellaceae_UCG-003, Alistipes, Christensenellaceae_R-7_group, Romboutsia, and Lachnospiraceae_unclassified.

Among them, UCG-005, UCG-010_norank, and Rikenellaceae_RC9_gut_group displayed greater proportions across the Control Group, Experimental Group I, and Experimental Group II, thereby representing the dominant genera. In Experimental Group II, UCG-005 showed the highest relative abundance (10.74%), while UCG-010_norank (7.86%) and Rikenellaceae_RC9_gut_group (4.75%) were more prevalent in Experimental Group I compared with the other groups.

As shown in Figure 6D, 20 microbial taxa with significant differences were detected between the Control Group and the experimental groups, including 1 phylum, 2 orders, 6 families, and 11 genera. Within the Control Group, 13 taxa were enriched, including Marinifilaceae, an unclassified genus of Erysipelotrichales,an unclassified family of Erysipelotrichales; Victivallaceae, Sanguibacteroides, Lachnospiraceae_ND3007_group, an unclassified genus of Victivallaceae, UCG_012, an unclassified genus of Alcaligenaceae, Alcaligenaceae, Mailhella, Desulfovibrionales, Desulfovibrionia, Desulfovibrionaceae, and Desulfobacterota.

In Experimental Group I, 6 taxa were identified: Lachnospiraceae, Lachnospirales, Sediminispirocheata, Marvinbryantia, an uncultured genus of Rickettsiales, an uncultured family of Rickettsiales, and an uncultured genus of Ruminococcaceaea. Effects of dietary supplementary feeding of L-Citrulline on serum metabolites of Simmental cows

Figure 7 illustrates the PLS-DA (Partial Least Squares-Discriminant Analysis) score plots used to visualize model-based group discrimination. In the comparison between CON and Experimental Group I, clear separation was observed, indicating effective group differentiation; the first and second principal components accounted for 11.1 and 16.8% of the variance, respectively.

In the comparison between CON and Experimental Group II, the separation was similarly distinct, with the first and second principal components explaining 14.8 and 15.8% of the variance, respectively.

Figure 8 presents the screening strategy for differential serum metabolites, which applied the thresholds of variable importance in projection (VIP) > 1 and p < 0.05 within the orthogonal partial least squares-discriminant analysis (OPLS-DA) framework. Metabolites satisfying both criteria were classified as differential trend metabolites. Following identification, 345 metabolites were characterized as differential, including 186 up-regulated and 159 down-regulated species.

Figure 9A illustrates that, among the top 20 enriched pathways differentiating the Control Group from Experimental Group I, nine metabolic pathways exhibited highly significant differences (p < 0.01). These included the Two-component system, Inflammatory mediator regulation of TRP channels, Efferocytosis, Central carbon metabolism in cancer, Biosynthesis of various other secondary metabolites, Arginine biosynthesis, Insect hormone biosynthesis, Staurosporine biosynthesis, and Alanine, aspartate and glutamate metabolism.

Figure 9B shows that, within the top 20 enriched pathways between the Control Group and Experimental Group II, nine metabolic pathways also displayed highly significant differences (p < 0.01). These comprised Central carbon metabolism in cancer, Two-component system, Arginine biosynthesis, Biosynthesis of various antibiotics, Biosynthesis of 12-, 14-, and 16-membered macrolides, Alanine, aspartate and glutamate metabolism, ABC transporters, Biosynthesis of various other secondary metabolites, and Efferocytosis.

The analysis suggests that the majority of differential metabolites were functionally associated with the biological processes governed by these nine pathways.

In Figure 9C, comparison between the Control Group and Experimental Group I revealed 13 metabolic pathways exhibiting an upward trend, including Rheumatoid arthritis, Renal cell carcinoma, Human Papillomavirus infection, Meiosis-yeast, Arginine biosynthesis, Non-alcoholic fatty liver disease, Efferocytosis, Staurosporine biosynthesis, Central carbon metabolism in cancer, Glucagon signaling pathway, Phenylalanine metabolism, Carbon fixation pathways in prokaryotes, and the Citrate cycle. Among them, Rheumatoid arthritis, Renal cell carcinoma, Human Papillomavirus infection, Meiosis-yeast, Arginine biosynthesis, Non-alcoholic fatty liver disease, and Efferocytosis displayed stronger up-regulation trends relative to the remaining pathways. In contrast, three pathways demonstrated a downward trend, namely Linoleic acid metabolism, Basal cell carcinoma, and TGF-beta signaling pathway.

In Figure 9D, comparison between the Control Group and Experimental Group II indicated three pathways with an up-regulated tendency, specifically Inflammatory mediator regulation of TRP channels, Biosynthesis of 12-, 14-, and 16-membered macrolides, and Retrograde endocannabinoid signaling. Seventeen pathways exhibited a down-regulated tendency, including Linoleic acid metabolism, Efferocytosis, Alanine, aspartate and glutamate metabolism, Biosynthesis of various antibiotics, Central carbon metabolism in cancer, Two-component system (commonly referring to bacterial signal transduction), ABC transporters, Synaptic vesicle cycle, Basal cell carcinoma, TGF-beta signaling pathway, Staurosporine biosynthesis, Histidine metabolism, Protein digestion and absorption, D-Amino acid metabolism, Biosynthesis of various other secondary metabolites, and Arginine biosynthesis.

The correlation heatmap depicts associations between species and environmental variables, reflecting both the magnitude and significance of relationships across multiple environmental factors and microbial taxa. In Figure 10A, analysis of rumen and intestinal microbiota at the phylum level in Simmental cows revealed the following: Bdellovibrionota abundance displayed a highly significant correlation with WPS-2 abundance (p < 0.001); Verrucomicrobiota abundance was negatively correlated with Firmicutes (0.01 < p ≤ 0.01); Firmicutes abundance exhibited a strong positive correlation with Bacteroidota (p < 0.001); and reciprocally, Bacteroidota abundance was also positively correlated with Firmicutes (p < 0.001).

As illustrated in Figure 10B, family-level analysis indicated that Eggerthellaceae abundance correlated positively with Synergistaceae (0.01 < p ≤ 0.01). Unclassified Bacteria abundance showed a negative correlation with WPS-2_norank (0.01 < p ≤ 0.01) while being positively correlated with Synergistaceae. Selenomonadaceae abundance was positively correlated with Synergistaceae (0.01 < p ≤ 0.01) but negatively correlated with Clostridia_vadinBB60_group_norank (0.01 < p ≤ 0.01). In addition, Erysipelatoclostridiaceae abundance demonstrated a significant positive correlation with unclassified Bacteroidales (0.01 < p ≤ 0.01).

As illustrated in Figure 10C, correlation analysis of ruminal and intestinal microbiota at the genus level in Simmental cows revealed several significant associations. The abundance of uncultured Ruminococcaceae exhibited a highly significant positive correlation with Lachnospiraceae_ND3007_group (p < 0.001) and a significant negative correlation with unclassified Bacteroidales (0.01 < p ≤ 0.05). Lachnospiraceae_NK3A20_group abundance was strongly and positively correlated with Desulfovibrio (p < 0.001), while showing a negative association with Treponema (0.001 < p ≤ 0.01). Gastranaerophilales_norank displayed a highly significant positive correlation with Veillonellaceae_UCG-001 (p < 0.001). Furthermore, uncultured Oscillospiraceae abundance correlated positively with probable_genus_10 (p < 0.001) and uncultured Prevotellaceae (0.001 < p ≤ 0.01), but negatively with unclassified Bacteroidales (0.01 < p ≤ 0.05).

As shown in Figure 11A, several significant associations were identified. The abundance of 4-Methylcatechol exhibited a strongly positive correlation with Synergistota (p < 0.001). Taurohyocholate abundance displayed a strongly positive correlation with unclassified Bacteria (p < 0.001), a highly positive correlation with Elusimicrobiota (0.001 < p ≤ 0.01), and a positive correlation with Campylobacterota (0.01 < p ≤ 0.05). PURPURIN abundance was strongly correlated with Cyanobacteria (p < 0.001). Avermectin B2b aglycone abundance showed a highly negative correlation with Desulfobacterota (0.001 < p ≤ 0.01);

Behenic acid abundance was negatively correlated with Patescibacteria (0.001 < p ≤ 0.01) and also showed a weaker negative correlation with Firmicutes (0.01 < p ≤ 0.05). Cis,cis-3-Carboxymuconic acid abundance exhibited a highly positive correlation with Bacteroidota (0.001 < p ≤ 0.01);

Decanoylcarnitine (C10) abundance demonstrated strong positive correlations with unclassified Bacteria (p < 0.001), Elusimicrobiota (0.001 < p ≤ 0.01), and Campylobacterota (0.001 < p ≤ 0.01).

Figure 11B Demonstrates that specific metabolites exhibited significant correlations with microbial taxa. The abundance of 16-methoxy-2,3-dihydro-3-hydroxytabersonine was positively associated with Peptococcaceae (0.01 < p ≤ 0.01), whereas 4-methylaminobutanal abundance showed a negative association with Barnesiellaceae (0.01 < p ≤ 0.01). A positive association was also identified between 2-phenylethanol abundance and Peptococcaceae (0.01 < p ≤ 0.01). In contrast, 2-iminobutanoic acid abundance correlated negatively with Izimaserumtales_norank (0.01 < p ≤ 0.01). Additionally, behenic acid abundance displayed negative associations with both Defluviitaleaceae and Selenomonadaceae (0.01 < p ≤ 0.01).

As presented in Figure 11C, several associations were identified. Veillonellaceae_UCG-001 abundance exhibited significant positive correlations with both 4-Methylaminobutanal (0.01 < p ≤ 0.01) and PUROMYCIN (0.01 < p ≤ 0.01). UCG-005 abundance displayed a significant negative correlation with Behenic acid (0.01 < p ≤ 0.01), while Anaerovorax abundance also showed a significant negative correlation with Behenic acid (0.01 < p ≤ 0.01); Treponema abundance demonstrated significant positive correlations with Taurocholate (0.01 < p ≤ 0.01) and FA 18_1,0 (0.01 < p ≤ 0.01). Ruminococcaceae_NK3A20_group abundance correlated negatively with both cis,cis-3-Carboxymuconic acid (0.01 < p ≤ 0.01) and FA 18_1,0 (0.01 < p ≤ 0.01).