The volatile oil was isolated by hydrodistillation at the Jiangxi Hairui Natural Plant Co., Ltd. (Jian, China). The leaves of Artemisiae Argyi were air-dried before steam distillation. The AAEO after extraction was stored in sealed amber ampules at 4 °C until being tested or analyzed.

The volatile oil was analyzed on a GCMS-QP2010 Plus Mass Spectrometer (Shimadzu) equipped with a DB-5 ms capillary column (30.0 m × 0.25 mm; film thickness, 0.25 μm) and mass spectrometry MS detector (MS). GC conditions were the same as described above), and data were recorded in the full-scan acquisition mode with the range of m/z 45–500. The compounds were identified by comparing retention indices (RI) and mass spectra with data from the NIST 11.0 MS library and the literature.

Animal experiments were conducted in accordance with the Hunan Agricultural University Institutional Animal Care and Use Committee (202105). Six-week-old male ICR mice were purchased from SLAC Laboratory Animal Central (Changsha, China). A total of 50 male mice (29.40 ± 1.23 g) were divided into 5 groups at random and housed in polypropylene cages and maintained in standard laboratory conditions of temperature (25 ± 2 °C) and light-dark cycle (12 h/12 h). The diets of mice were fed the control diet (Con), the high-fat diet (HFD), and the HFD-fed mice with the low (LEO, 0.20 ml/kg), medium (MEO, 0.40 ml/kg), and high (HEO, 0.80 ml/kg) doses of AAEO as previous study (33). AAEO dissolved in 4% Tween 80 and administered by gastric administration. Before any treatment, mice were housed for 7 days to allow adaptation. The experiment lasted for 8 weeks. At the end of the experiment the animals were sacrificed under anesthesia.

Prepare 1.5 ml sterile centrifuge tubes (marked animal number in advance), sterile forceps, and transfer the mice to a clean, sterile individual feeding cage after 8 hours on an empty stomach. Pick up fresh stool directly with sterile forceps, avoiding hand contact with the sample. Dispense into marked sterile centrifuge tubes immediately after collection and place kept at −80 °C for examination. The fecal DNA extraction and 16S ribosomal RNA amplification were conducted as previously reported (34, 35). The microbial function prediction analysis process refers to previous studies (36, 37).

100 μl serum sample was added to a 1.5 ml centrifuge tube with 400 μl solution [acetonitrile: methanol = 1:1(v:v)] containing 0.02 mg/ml internal standard (L-2-chlorophenylalanine) to extract metabolites. The samples were mixed by vortex for 30 s and low-temperature sonicated for 30 min (5 °C, 40 KHz). The samples were placed at −20 °C for 30 min to precipitate the proteins. Then the samples were centrifuged for 15 min (4 °C, 13,000 g). The supernatant was removed and blown dry under nitrogen. The sample was then re-solubilized with 100 μl solution (acetonitrile: water = 1:1) and extracted by low-temperature ultrasonication for 5 min (5 °C, 40 KHz), followed by centrifugation at 13,000 g and 4 °C for 10 min. The supernatant was transferred to sample vials for LC-MS/MS analysis. As a part of the system conditioning and quality control process, a pooled quality control sample (QC) was prepared by mixing equal volumes of all samples. The QC samples were disposed and tested in the same manner as the analytic samples. It helped to represent the whole sample set, which would be injected at regular intervals (every 5 samples) in order to monitor the stability of the analysis.

The LC-MS/MS analysis of sample was conducted on a Thermo UHPLC-Q Exactive HF-X system equipped with an ACQUITY HSS T3 column (100 mm × 2.1 mm i.d., 1.8 μm; Waters, USA) at Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The mobile phases consisted of 0.1% formic acid in water: acetonitrile (95:5, v/v) (solvent A) and 0.1% formic acid in acetonitrile: isopropanol:water (47.5:47.5, v/v) (solvent B). The flow rate was 0.40 ml/min and the column temperature were 40 °C. The injection volume was 3 μl. The mass spectrometric data were collected using a Thermo UHPLC-Q Exactive HF-X Mass Spectrometer equipped with an electrospray ionization (ESI) source operating in positive mode and negative mode. The optimal conditions were set as followed: Aux gas heating temperature at 425 °C; Capillary temp at 325 °C; sheath gas flow rate at 50 psi; Aux gas flow rate at 13 psi; ion-spray voltage floating (ISVF) at −3,500 V in negative mode and 3,500 V in positive mode, respectively; Normalized collision energy, 20–40–60 eV rolling for MS/MS. Full MS resolution was 60,000, and MS/MS resolution was 7,500. Data acquisition was performed with the Data Dependent Acquisition (DDA) mode. The detection was carried out over a mass range of 70–1,050 m/z.

Statistical analyses between the means of each group were analyzed using one-way ANOVA (one-way analysis of variance) followed by Duncan comparison range tests through SPSS 22.0. The statistical significance level was set at p < 0.05. The significance between groups represented by ^* was p < 0.05, ^** represented p < 0.01, ^*** represented p < 0.001. For the metabolomics data, the raw data were imported into the Progenesis QI software (Waters Corporation, Milford, USA) for data preprocessing after UHPLC-QE-HFX/MS analyses. Through baseline filtering, peak identification, peak integral, retention time correction, and peak alignment. Then, the data matrix containing sample names, m/z, retention time and peak intensities was exported for further analyses. At the same time, the metabolites were identified by searching database, and the main databases were the HMDB (https://www.hmdb.ca), Metlin (https://metlin.scripps.edu/) and the self-compiled Majorbio Database (MJDB) of Majorbio Biotechnology Co., Ltd. (Shanghai, China). The R package “ropls” (Version 1.6.2) was used to perform principal component analysis (PCA) and orthogonal least partial squares discriminant analysis (OPLS-DA), and 7-cycle interactive validation evaluating the stability of the model. The metabolites with VIP > 1, p < 0.05 were determined as significantly different metabolites based on the Variable importance in the projection (VIP) obtained by the OPLS-DA model and the p-value generated by student's t test.

The chemical composition of AAEO was analyzed by GC/MS as shown in Table 1. Only components with concentrations >0.3% are reported. The obtained results indicated the presence of 25 chemical compounds in the chromatogram, constituting 90.79% of the total components detected. The highest compositions of the compounds were 18.43% of Linalool, followed by Bicyclo[2.2.1] heptan-2-one, 1,7,7-trimethyl-, (1S)- (13.28%), Safroleb (11.96%), D-Limonene (10.86%), Tricyclo[2.2.1.0(2, 6)]heptane, 1,7-dimethyl-7-(4-methyl-3-pentenyl)-, (-)- (6.78%), 3-Cyclohexene-1-methanol, alpha.,4-trimethyl-, (R)- (5.53%), and alpha-Terpineol (5.38%).

As demonstrated in Figure 1, based on PICRUSt prediction, the microbiota function was predicted; the KEGG pathway in top 10 at level 2 about metabolism included global and overview maps, carbohydrate metabolism, amino acid metabolism, energy metabolism, metabolism of cofactors and vitamins, and nucleotide metabolism et al. in three dietary treatments (Figure 1A). Totally, the PICRUSt predicted 295 pathways in level 3 of fecal samples, among them, only 80 pathways were related to metabolism (level 2). The central pathway in top 10 of level 3 contained five pathways about global and overview maps (as shown in Figure 1B), which were metabolic pathways, biosynthesis of secondary metabolites, microbial metabolism in diverse environments, biosynthesis of amino acids, and carbon metabolism respectively. In addition, the pathways of amino sugar and nucleotide sugar metabolism, glycolysis/gluconeogenesis, starch and sucrose metabolism were associated with carbohydrate metabolism. The purine metabolism and pyrimidine metabolism were associated with nucleotide metabolism at level 3. Finally, the top 10 pathways in KEGG levels 2 and 3 were not notably different (p > 0.05). Analysis of the AAEO's effects on the mouse fecal microbiota demonstrated that it influenced the functional potentials of these microbes (Table 2). Compared to the Con group, the HFD group with AAEO significantly raised (p < 0.05) abundance of lysine biosynthesis, tyrosine metabolism, pyruvate metabolism, and glycerolipid metabolism related with amino acid, carbohydrate, and lipid metabolism. On the contrary, the HFD group with AAEO significantly decreased (p < 0.05) abundance of phenylpropanoid biosynthesis, pentose and glucuronate interconversions, oxidative phosphorylation, carbon fixation pathways in prokaryotes, and cyanoamino acid metabolism than Con group. Although HFD significantly increased the abundance of the other glycan degradation pathway associated with glycan biosynthesis and metabolism (p < 0.05), the addition of AAEO to HFD had no effect on it (p > 0.05).

The QC sample positive (Figure 2A) and negative (Figure 2B) ion assessment plots were shown in Figure 2. A QC sample was injected every eight samples to verify the system stability during the whole sample batch. For the overall data, the RSD < 0.3 and the percentage of peak accumulation >70% indicated that the proposed method was stable, and the overall data measured were qualified (dashed line indicates before pretreatment, solid line indicates after pretreatment). It can be seen that when the RSD was 30%, the peak accumulation percentages of positive ions and negative ions of QC samples are 95.8% and 95.2% respectively. It indicated that the data is stable and reliable and can be used for subsequent analysis. The metabolite classification information obtained based on the identification of HMDB database is shown in Figure 2C. It can be proved that the metabolite species of mice blood were mainly concentrated in lipids and lipid-like molecules, organic acids and derivatives, organoheterocyclic compounds, benzenoids, and phenylpropanoids and polyketides. Lipids and lipid-like molecules were the most abundant metabolites with 239, followed by organic acids and their derivatives with 124.

As shown in Table 3, a total of 7,107 positive ion mass spectrometry peaks and 5,106 negative ion mass spectrometry peaks were identified by LC-MS/MS non-targeted metabolomics; 357 positive ion metabolites and 243 negative ion metabolites compared to public databases; and 173 positive ion metabolites and 102 negative ion metabolites compared to KEGG database.

In order to better understand and analyze the intervention effect of AAEO on blood metabolites in mice on HFD, the blood metabolite data from each group of mice were analyzed by PCA in this study. From Figure 3, it can be seen that there was a trend of separation between the samples of mice in the HFD group and those of mice in the AAEO group regardless of the positive (Figure 3A) or negative (Figure 3B) ion mode, indicating that AAEO affected the changes of metabolites in the blood of mice. Between the mice in the Con group and the samples from the HFD group, there was a greater degree of separation within the groups, and there was a crossover between the groups, and the differences were not significant. This may be due to the fact that there were more factors that have an impact on the data, i.e., not only the dietary factors set up in this experiment, but also individual differences in mice, the stability of the instrumentation, and other external factors, resulting in the phenomenon of underfitting the data, and therefore the PLS-DA statistical method is needed to continue the analysis, and thus to ignore the random differences within the groups and to highlight the systematic differences between the groups.

As can be shown from Figures 4A, C, there was no crossover between the samples of the Con group and HFD in either the positive or negative ion mode, and the degree of separation was good. There was no crossover or overlap between the samples of the HFD group and AAEO group. It indicated that the AAEO has a significant effect on the metabolites in the blood of mice on HFD. The replacement test graph was used to speculate whether the model was reliable or not, and the criterion for the evaluation of the replacement test was mainly the intercept of the Q² regression line with the Y-axis, when the intercept was < 0.05, it indicated that the model was robust and reliable. As shown in Figures 4B, D, Q² = −0.558 < 0.05 in the positive ion model and Q² = −0.7065 < 0.05 in the negative ion model, indicating that the model was robust and reliable without overfitting in either ion model.

Orthogonal Partial Least Squares Discriminant Analysis, OPLS-DA was a derivative of PLS-DA. OPLS-DA can better distinguish between subgroups and improve the validity and parsing ability of the model. The OPLS-DA model score plot and permutation test plot between Con vs. HFD and HFD vs. AAEO were shown in Figures 5, 6A, C, respectively, and there was a clear trend of separation between Con vs. HFD and between HFD vs. AAEO in both positive and negative ion modes. And from Figures 5B, D, Q² was < 0.05 in the positive ion mode, and Q² was >0.05 in the negative ion mode, and with the decrease of replacement retention in the negative ion mode, R2 and Q2 decreased, and the regression line showed an upward trend, which could indicate that the replacement test was passed, and the model did not have the overfitting phenomenon. From Figures 6B, D, it could be proved that Q² is < 0.05 in both the positive ion mode and the negative ion mode, indicating that the model was robust and reliable without overfitting in either ion mode.

The parameters of the OPLS-DA models for positive and negative ionization were listed in Table 4. The closer the R²Y and Q² were to 1, the more stable and reliable the model was, Q² > 0.5 indicated a better predictive ability of the model, and Q² < 0.5 indicated a poorer predictive ability of the model. As can be seen from the table, the cumulative explanatory rate of R²Y for the comparison between the two groups in both the positive and negative ion modes exceed 0.9, indicating that at least 90% of the variance in the Y-measure can be explained by these OPLS-DA models. The Q² calculated by cross-validation showed that the predictions of the OPLS-DA models were good for all groups. In particular, the R²Y and Q² between the Con vs. HFD groups were close to 1, indicating that the HFD-induced obesity model in this study, modeling was successful. In addition, Q² > 0.5 between the HFD vs. AAEO groups indicated that ingestion of AAEO could successfully modulate the metabolite composition caused by HFD.

Screening was performed based on the variable importance of projection (VIP) analysis obtained from the OPLS-DA model. Metabolites with VIP value >1 and p-value < 0.05 were considered as significantly different metabolites between the two groups. Table 5 showed the number of significantly different metabolites from the comparative analysis between the two groups. A total of 170 significantly different metabolites were identified in the positive ion mode and 156 significantly different metabolites were identified in the negative ion mode. Compared with the Con group, there were 79 significantly different metabolites in the positive ion mode and 68 in the negative ion mode in the HFD group, and 123 significantly different metabolites in the positive ion mode and 127 in the negative ion mode in the AAEO group. Compared with HFD, there were 91 significantly different metabolites in positive ion mode and 78 significantly different metabolites in negative ion mode in the AAEO group.

Supplementary Table S1 showed the significantly different metabolites in the Con group compared with the HFD group screened according to VIP > 2 and p-value < 0.05. A total of 60 significantly different metabolites were identified, including six amino acids, peptides, and analogs, four fatty acids and conjugates, two glycerophosphocholines, two bile acids, alcohols and derivatives, one glycerophosphocholines, five carbohydrates and carbohydrate conjugates, and so on. Based on the abundance of metabolites after normalized treatment, 10 metabolites were found to be significantly up-regulated (P < 0.05) and 50 metabolites were found to be significantly down-regulated (P < 0.05) in the HFD group compared with Con group.

Supplementary Table S2 showed the significantly different metabolites in the AAEO group compared with the HFD group according to the VIP > 2 and p-value < 0.05 screening. A total of 79 significantly different metabolites were identified, including six Amino acids, peptides, and analogs, four fatty acids and conjugates, three fatty acyl glycosides, three carbohydrates and carbohydrate conjugates, two triterpenoids, and three terpene glycosides, and others. Based on the abundance of metabolites after normalized treatments, 22 metabolites were found to be significantly up-regulated (P < 0.05) and 57 metabolites were found to be significantly down-regulated (P < 0.05) in the AAEO group as compared to HFD group.

The major differential metabolites of AAEO associated with obesity screened according to the Kruskal–Wallis rank sum test are shown in Figure 7. The most abundant differential metabolite was PC (16:0/0:0) [U] (Figure 7A), which was significantly lower in the HFD group than in the Con group (P < 0.05), and it was significantly elevated by AAEO intake (P < 0.05). However, the content of PC (18:0/0:0) in the HFD group was significantly higher than that in the Con group (P < 0.05), and the intake of AAEO significantly decreased its content (Figure 7B) (P < 0.05). In addition, PC [16:0/22:6(4Z,7Z,10Z,13Z,16Z,19Z)], LysoPC [20:4(8Z,11Z,14Z,17Z)], PC [18:2(9Z,12Z)/22:6(4Z,7Z,10Z,13Z,16Z,19Z)], and LysoPC (16:0) were significantly lower than those in the Con group and were restored after AAEO ingestion (P < 0.05) (Figures 7C–F).

Based on the difference metabolites between the two groups screened earlier, the KEGG pathway enrichment in each group was further screened by using the calculation method of num in study/num in pop. The KEGG pathway enrichment in the Con vs. HFD group and the AAEO vs HFD group was shown in Figure 8. It was shown in Figure 8A by analyzing the KEGG pathways in the Con vs. HFD group, we can obtain that there were 20 major pathways enriched. There were 20 major enriched pathways. Among them, linoleic acid metabolism, pyrimidine metabolism, glycerophospholipid metabolism and phenylalanine metabolism were significantly higher than other metabolic pathways (P < 0.05). In addition, some of them focused on the metabolic pathways of fat metabolism regulation, such as cholesterol metabolism, and fatty acid metabolism pathways, such as arachidonic acid metabolism. As shown in Figure 8B, there were 20 major enriched pathways by analyzing the KEGG pathways in the AAEO vs. HFD group. Glycerophospholipid metabolism had the highest enrichment rate and was significantly higher than other pathways (P < 0.05). This was followed by retrograde endocannabinoid signaling, bile secretion, linoleic acid metabolism, Fc gamma R-mediated phagocytosis, pyrimidine metabolism and apelin signaling pathway. Two of them were the main pathways involved in lipid metabolism, including glycerophospholipid metabolism and linoleic acid metabolism.