2.1 Materials

CT was provided by Xinjiang Life Nuclear Power High-Tech Co., Ltd., and was identified as the dried flower head of Coreopsis tinctoria Nutt. by Xiaopeng Chen associate professor of Tianjin University of Traditional Chinese Medicine. The samples are kept in the sample room of the Institute of Traditional Chinese Medicine, Tianjin University of Traditional Chinese Medicine. Chlorogenic acid, protocatechuic acid, flavanomarein, isookanin, marein, okanin, and quercetin-7-O-β-D-glucoside standards (batch numbers DST190906-021, DST180428-081, DST191102-240, DST200509-453, DST200311-132, DST191210-105, DST200315-029) were purchased from Chengdu Desite Bio-Technology Co., Ltd. (Chengdu, China). The standards of neochlorogenic acid (wkq19021413) and coreopsin (wkq20061703) were obtained from Sichuan Victory Biological Technology Co., Ltd. (Sichuan, China), and caffeic acid standard (MUST-18032003) was purchased from Chengdu Must Bio-Technology Co., Ltd.(Chengdu, China). The mass fractions of all standard reagents were ≥ 98%.

2.2 CT water extract preparation and analysis

Following the pre-treatment method of teabag test in the National Food Safety Standard (GB15193.21–2014). Weigh 2 g CT samples into a 100 mL beaker, add 50 mL boiling water, and cover with tin foil, brew for 30 min, filter out the tea soup. Add 50 mL of boiling water to brew for 30 min once more, filter out the tea soup, mix the tea soup, shake well, and concentrate. When the temperature of the tea soup drops, fix the volume to 40 mL and filter. The sample is stored in the refrigerator below 4 ℃ for later use.

The qualitative and quantitative analysis of the components of the CT water extract was carried out on UPLC-IMS-QTOF-MS (Waters, USA) and HPLC-PDA (Waters, USA), respectively. The instrument characteristics and chromatographic conditions are provided in the supplementary material.

2.3 Network pharmacology analysis

Based on the ingredients from section 2.2, the active ingredient targets and predicted targets of CT were obtained from TCMSP, Swiss Target Prediction, and PharmMapper Server databases. The related targets of T2DM were searched with “type 2 diabetes mellitus” in the Gene Cards database, and combined with the literature reports, the duplicated data were deleted and the diabetes target genes were screened. The intersection of the above two-part targets is the key target genes of CT in the treatment of T2DM. Input genes into the STRING database, and get the PPI network of CT acting on T2DM. Import the network into Cytoscape software, and screen the potential genes of the compound by node value, intermediate center degree, and near center degree. Input the data into DAVID 6.8 database to carry out GO biological process enrichment analysis and KEGG pathway enrichment analysis.

2.4 Animal study

SD rats, 180–200 g, male, purchased from Beijing Huafukang Biotechnology Co., Ltd., license number: SCXK (Beijing) 2019–0008. Basic feed and high-fat feed were purchased from Beijing Huafukang Biotechnology Co., Ltd. The rat model of T2DM was established by high-fat diet (HFD) combined with a low-dose injection of streptozotocin (STZ). After 3 days of adaptive feeding, SD male rats were fed with HFD for 8 weeks, resulting in insulin resistance. After passing the glucose tolerance test, STZ 35 mg/kg (prepared by the citric acid buffer of pH 4.5) was injected into the tail vein and fed for 2 weeks to establish the T2DM model. The blood glucose higher than 16.7 mmol/L is taken as the success model. A normal control group using conventional feed and the same feeding conditions was set at the same time. The model rats were randomly divided into four groups: model group (n = 8), positive control group (250 mg/kg of metformin, n = 8), low dose group (1000 mg/kg of CT, n = 8), high dose group (2000 mg/kg of CT, n = 8), and normal control group (n = 8). The rats were given intragastric administration and the normal control and model group were given the same volume of normal saline for 8 weeks. During the dosing, the rats were fed with conventional feed in the controlled environmental conditions of 12 h day and night replacement, ambient temperature 22 ± 2 ℃, and ambient humidity 35 ± 5%. All experimental protocols strictly followed the U.K. Animals (Scientific Procedures) Act, 1986 and associated guidelines, EU Directive 2010/63/EU for animal experiments, the Guidance Suggestions for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China (2006), and the Animal Use and Care Committee of Tianjin University of Traditional Chinese Medicine (No. IRMDWLL-201903).

2.5 Sample collection and biochemical index determination

After dosing for 8 weeks, the rats fasted in metabolic cages, collected 24-hour urine and packed in centrifuge tubes for metabonomic analysis. Then rubbed the abdomen of rats, collected the feces of rats in each group, packed them in aseptic frozen tubes, and stored them in the refrigerator at −80 ℃ after quick freezing with liquid nitrogen. The body weight of rats was measured and recorded regularly every week before and during dosing, and the changes were calculated. Before and during dosing, the rats fasted for 12 h and the blood of the orbital venous plexus was taken every week. Blood glucose, the levels of serum total cholesterol (TC) and triglyceride (TG) were detected by Mindray BS-240VET animal automatic biochemical instrument.

2.6 Intestinal microflora

The genomic DNA of feces was extracted by the CTAB method, and then the purity and concentration of DNA were detected by agarose gel electrophoresis. An appropriate amount of sample DNA was taken out to the centrifuge tube, and the sample was diluted to 1 ng/μL with sterile water. Then the diluted genomic DNA was used as the template, and the 16SrDNA variable region primers were applied to PCR using the specific primers 515F (5′-ACTCTACGGAGCAGCAGCAGCA-3′) with Barcode and 806R (5′-GGACTACGGGTWTCTAAT-3′). The amplification procedure is as follows: initial denaturation of 1 min at 98 ℃, then denaturation at 98 ℃, annealing at 50 ℃, annealing at 50 ℃, extension of 30 s at 72 ℃ for 30 cycles, and final extension of 5 min at 72 ℃. The qualified PCR products were purified by magnetic beads, quantified by enzyme labeling, and the same amount of samples was mixed according to the concentration of PCR products. After full mixing, 2% agarose gel electrophoresis was used to detect PCR products, and the gel recovery kit provided by the Qiangen company was used to recover the products. The library was constructed with TruSeq ®DNAPCR-Free Sample Preparation Kit library construction kit, and the library was quantified by Qubit and QPCR. After the library was qualified, NovaSeq6000 was used for computer sequencing. Construction of the 16S rDNA gene amplicon sequence library, statistical analysis, and data manipulation are described in Supporting Information.

2.7 Urine metabolomics

The urine samples stored in the −80 °C refrigerator were thawed on the ice. 100 μL of urine was taken into the 1.5 mL EP tube, 20 μL of the internal standard solution was added, and the vortex oscillated for 1 min. Then add 400 μL acetonitrile solution to precipitate the protein, continue to scroll for 1 min, then centrifuge for 15 min at 4 °C, 19,721 g. Suck the supernatant nitrogen to dry, add 100 μL acetonitrile solution to re-dissolve, and centrifuge for 15 min at 4 °C, 19,721 g. Take the supernatant in the automatic injection bottle, to be analyzed. The QC samples were mixed with 10 μL of all the samples. At the beginning of the sequence, the QC samples were run 5–10 times to balance the whole detection system and analyzed periodically in the whole analysis process. The QC samples were used to monitor and examine the stability of the instrument and the repeatability of the method. See Supplementary Materials for metabolomics data acquisition and analysis conditions.

2.8 Data analysis and statistics

SPSS 24.0 software and Origin 2021 were used for data analysis. The data were expressed by mean ± standard deviation, and the significant levels were *p < 0.05 and ***p < 0.01, respectively. Using the file format converter of Labsolutions software (Shimadzu, Japan), the original offline data of MS was converted into mzML format data file. The obtained data file was imported into MsDial software (RIKEN, Japan), which is used to set noise, perform baseline correction, peak extraction, and peak alignment to obtain a preliminary metabonomic data matrix, and use MetaboAnalyst database (https://www.metaboanalyst.ca/) to process the data. The not available value is filled with 1/2 of the minimum value, and the peaks with missing rate > 50% are filtered. The area normalization, logarithmic transformation, and pareto scaling mode are used to transform the data from biased distribution to normal distribution. By establishing the models of unsupervised principal component analysis (PCA) and supervised orthogonal partial least square analysis (OPLS-DA), and drawing the S-plot diagram, the metabolites with variable importance in the projection (VIP) > 1 were screened and analyzed by t-test. And the difference compounds with VIP ≥ 1 and p ≤ 0.01 were selected as potential biomarkers, combined with the public database HMDB (http://www.hmdb.ca/), MassBank databases (https://www.massbank.jp/), and METLIN databases. The selected biomarkers were enriched and analyzed by the MetaboAnalyst database. Under the condition of p < 0.05, the metabolic pathways with significant enrichment were screened, which were considered to be the potential key pathways for the treatment of T2DM. And the related target proteins of the key pathways were searched by the KEGG database (https://www.genome.jp/kegg/).

3.1 Qualitative and quantitative of CT water extract

To characterize the constituents in CT water extract, UPLC-IMS-QTOF-MS was used to detect the samples in negative ion mode. The total ion chromatogram (TIC) was shown in Fig. 1. The obtained chromatographic mass spectrometry information includes the retention time of each peak, the quasi-molecular ion peak and its fragment ion information obtained after adducting ions, and the fragmentation rule of mass spectrometry. A total of 23 compounds were obtained, including 18 flavonoids. See Table 1 for details. Among them, the content of 5 components was measured (Figure A1), namely flavanomarein, isookanin, marein, coreopsin, okanin, which were 8.350 mg/g, 1.693 mg/g, 16.785 mg/g, 1.083 mg/g, 5.813 mg/g, respectively. The investigation of the linear relationship, LOQ, and LOD is shown in Table A1 of the Supplementary Material, and the methodological investigations are detailed in Table A2 of the Supplementary Material.

Close

Fig. 1

TIC of Coreopsis tinctoria Nutt. water extract in negative ion mode.

Export to PPT

3.2 Network pharmacology

3.2.2

3.2.2 Protein target database construction

The 408 target genes were input into the STRING database for PPI network construction and analysis. Then the PPI network of CT for the treatment of diabetes target proteins was obtained. The topology of the PPI network was studied by the “Network Analyzer” on Cytoscape, and the sub-network was constructed according to the obtained parameters including Degree, Betweenness Centrality, and Closeness Centrality. The medians of Degree, Betweenness Centrality, and Closeness Centrality were 5, 0.001535, and 0.31043, respectively. Select the target whose node degree value is greater than or equal to 5 times of the median value of all points, and meet the requirements of the betweenness centrality and closeness centrality that are greater than the median value of all points as the potential target for the treatment of T2DM with CT. A total of 25 key targets were obtained. The higher the node degree value, the larger the graph, and the corresponding color varies from green to orange. As in Fig. 2A, 25 key targets are SRC, TP53, HSP90AA1, STAT3, MAPK1, PIK3R1, AKT1, PIK3CA, HRAS, JUN, RELA, ESR1, EGFR, PTPN11, JAK2, CDK1, RHOA, MAPK8, RXRA, STAT1, LCK, CDKN1A, HDAC1, MAPK14, IGF1.

Close

Fig. 2

Network pharmacological prediction of CT in the treatment of T2DM. (A) Core targets and interaction relationship of CT in the treatment of T2DM. (B) Top 20 GO enrichment analysis. (C) Bubble diagram of KEGG signaling pathway. (D) “Active Ingredient-Core Target-Key Pathway” Network.

Export to PPT

3.3 Determination of biochemical indicators

The content of blood lipids such as TC and TG in the serum of rats after dosing was detected. Compared with the NC group as shown in Fig. 3, the serum TC and TG levels of the Mod group significantly increased (p < 0.05), and the body weight significantly reduce. The TG and TC values of the T2DM model rats were higher than those of the healthy rats with normal blood sugar, which indicated that the abnormal blood sugar level of the glucose and lipid metabolism mice was often accompanied by the symptoms of dyslipidemia. Compared with the Mod group, the serum total cholesterol of the rats in the CT low-dose group was significantly decreased (p < 0.05). It shows that the administration group can effectively reduce the level of TC in serum, thereby correcting the disorder of blood lipid metabolism and preventing T2DM.

Close

Fig. 3

Physiological index changes of each group in the eighth week after administration. (A) Changes of blood glucose in each group for 8 weeks of intervention. (B) Eight-week body weight change trend of intervention in each group. (C) The body weight at the 8th week. (D) Blood glucose values at the 8th week. (E) The triglyceride level at the 8th week. (F) The total cholesterol level at the 8th week. Values are presented as mean ± SD. (NC) normal group; (Mod) model group; (Met) positive group; (JH) high dose group; (JL) low dose group.

Export to PPT

3.4 Gut microbiome analysis

3.4.1

3.4.1 Species distribution annotation

The OTUs were classified and analyzed at the phylum, class, order, family, genus, and species classification level. All detected samples mainly include 44 phylum, 96 class, 191 order, 247 family, 319 genus, and 111 species. According to the species annotation results, select the top 10 species with the largest abundance at the taxonomic level of phylum and genus, and generate a column accumulation chart of the relative abundance of species to visually check the relative abundance of each sample at the taxonomic level of phylum, genus, and their proportions. The main phyla detected at the phylum classification level are Firmicutes, Bacteroidota, Proteobacteria, Actinobacteriota, Spirochaetota, unidentified_Bacteria, Acidobacteriota, Campilobacterota, Verrucomicrobiota, Desulfobacterota (Fig. 4A). The main ones detected at the genus taxonomic level were Lactobacillus, Prevotella, Succinivibrio, Bifidobacterium, and Treponema(Fig. 4B).

Close

Fig. 4

Gut microbiota species annotation analysis. (A) Histogram of relative abundance of species at the phylum level. (B) Histogram of relative abundance of species at the genus level. (C) Histogram of LDA value distribution in LEfSe analysis.

Export to PPT

3.4.2

3.4.2 Differential genus analysis

LEfSe analysis of all samples found a total of 37 species with significantly different species information (p < 0.05) at the level of phylum, class, order, family, genus, and species (Fig. 4C). The bacterial flora of the tract was also significantly affected. Lactobacillus showed a downward trend in the model group, and Prevotella, Succinivibrio, unidentified Ruminococcaceae, and Treponema showed an upward trend in the diabetes model group. CT can regulate the relative abundance of the above intestinal flora in the model group to varying degrees as shown in Fig. 5.

Close

Fig. 5

Effects of CT on intestinal microbial composition in T2DM. (A) Lactobacillus. (B) Succinivibrio. (C) Prevotella. (D) unidentified Ruminococcaceae. (E) Treponema. (NC) normal group; (Mod) model group; (Met) positive group; (JH) high dose group; (JL) low dose group.

Export to PPT

3.4.3

3.4.3 Pathway prediction analysis

Based on the function prediction analysis of Tax4Fun the pathway involved in KEGG is shown in Fig. 6. The pathway is mostly enriched in metabolic pathways, which mainly include Carbohydrate metabolism, Amino acid, Nucleotide metabolism, Energy metabolism, and Glycan biosynthesis and metabolism.

Close

Fig. 6

Functional predictive analysis.

Export to PPT

3.5 Metabolomics analysis

3.5.1

3.5.1 Multivariate statistical analysis

Unsupervised principal component analysis (PCA) was performed on each group, and the results found that the normal group and the model group were significantly separated, indicating that the urinary endogenous metabolites in T2DM dose undergo significant change. Meanwhile, the dosing group was far away from the model group, and there was no obvious difference between clusters, indicating that CT or metformin also changed the endogenous metabolites of diabetic rats to a certain extent (Fig. 7A).

Close

Fig. 7

Multivariate analyses of urine UHPLC-QTOF-MS/MS spectra data. (A) Principal Component Analysis (PCA) scores plot. (B, C) OPLS-DA score plot and S-plot for high-dose and model group. (D, E) OPLS-DA score plot and S-plot for low-dose and model group. (NC) normal group; (Mod) model group; (Met) positive group; (JH) high dose group; (JL) low dose group.

Export to PPT

To further screen out the differential metabolites with significant changes, a supervised orthogonal partial least squares discriminant analysis (OPLS-DA) model was constructed. The model group (Mod) and the CT high-dose group (JH) were clearly distinguished, with R2X, R2Y, and Q2 being 0.664, 0.999, and 0.943, respectively (Fig. 7B). The model group (Mod) and the CT low-dose group (JL) were distinguished, with R2X, R2Y, and Q2 being 0.371, 0.993 and 0.937, respectively(Fig. 7D). The more the parameters are inclined to 1, the more the model is reliable. Q2 > 0.5 indicated the good predictive ability of the OPLS-DA model. Moreover, the results of the random permutation test indicate that the model has a good explanatory and predictive ability.

3.5.2

3.5.2 Potential marker analysis

An S-plot graph was constructed to screen the differential metabolites between the model group and the two administration groups, as shown in Fig. 7C and E. The higher the VIP value, the greater the contribution to distinguishing the two groups. The screening conditions for the metabolites marker were as follows: VIP ≥ 1, fold change (FC) ≥ 2 or ≤ 0.5, and the difference was statistically significant between the two groups (t-test p-Value < 0.05). According to the molecular weights and secondary fragments corresponding to the variables, the potential markers were identified by searching the public databases HMDB, METLIN, and MetDNA, and comparing them with relevant literature. The 31 marker metabolites related to the treatment of diabetes progression are shown in Table 2, mainly including amino acids. To facilitate the observation of the changes in the relative contents of metabolites, the study standardized the original relative contents of the metabolites identified by the screening criteria (Unit Variance Scaling) by row and drew a heat map through the online platform. The results are shown in Fig. 8. It can be found that the contents of 8 metabolites decreased in the diabetic rat model group, while the contents of 23 metabolites increased. All 31 metabolites showed callback effects after administration of CT.

Table 2 Marker metabolites information in urine.

NO.	RT(min)	Mass(g/mol)	Formula	Compounds	VIP	p_value	Log2FC	Type
1	1.79	162.0528	C6H10O5	1,6-Anhydro-β-D-glucose	1.478542	0.040053862	−1.1986	down
2	1.97	268.0736	C16H12O4	formononetin	2.065681	0.009658918	−3.7877	down
3	6.34	569.3173	C25H43N7O8	Asn-Leu-Pro-Ile-Asn	1.817375	9.60589E-05	1.038247	up
4	6.35	546.3166	C27H42N6O6	Westiellamide	2.491242	1.30855E-06	−4.65889	down
5	6.50	612.3384	C30H44N8O6	Phe-Tyr-Lys-Arg	2.338088	0.003422727	1.580587	up
6	6.50	590.3540	C28H46N8O6	Gly-Leu-Arg-Val-Phe	2.453694	6.88578E-06	−3.99582	down
7	7.08	626.3388	C27H46N8O9	Glu-Lys-Leu-Thr-His	2.275839	0.018560722	1.711438	up
8	12.16	598.3915	C26H50N10O6	Ala-Pro-Arg-Lys-Lys	1.943954	0.00254572	−2.33234	down
9	13.82	337.3345	C22H43NO	13Z-Docosenamide	1.86724	0.002593778	−4.04777	down
10	14.31	594.3853	C28H50N8O6	His-Lys-Leu-Val-Val	2.266149	2.50169E-05	2.741213	up
11	14.47	731.5465	C40H78NO8P	1-Oleoyl-2-myristoyl-sn-glycero-3-phosphocholine(PC(18:1(9Z)/14:0))	2.434706	3.3567E-07	−1.73161	down
12	14.62	576.3635	C29H48N6O6	Ile-Gly-Lys-Ile-Phe	2.088116	0.013848311	1.078988	up
13	15.07	722.4758	C44H66O8	1,26-Dicaffeoylhexacosanediol	2.042758	0.017579569	1.264828	up
14	15.08	700.4914	C42H68O8	MycolactoneF	1.640435	0.009788668	1.1777	up
15	15.75	637.3071	C28H43N7O10	Lys-Tyr-Gln-Glu-Ala	2.068704	0.000803063	−1.23474	down
16	15.75	636.3119	C29H44N6O10	Asp-Tyr-Asn-Leu-Leu	1.862101	0.002836002	−1.15739	down
17	16.55	757.5622	C42H80NO8P	1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-N-methyl	1.972926	0.049881676	−2.5886	down
18	16.66	550.4961	C35H66O4	ArtemoinA	1.239144	0.01353093	−2.87506	down
19	16.67	761.5935	C42H84NO8P	1-Palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine(PC(16:0/18:0))	2.125899	0.001607988	−1.81388	down
20	16.72	713.5442	C40H75NO9	Glucocerebrosides	2.216238	0.000215953	−1.26903	down
21	16.82	590.4335	C39H58O4	CoenzymeQ6	2.269752	8.99233E-05	−1.03892	down
22	16.84	568.5067	C35H68O5	1,2-Dihexadecanoyl-sn-glycerol	1.986973	0.004734363	−4.7641	down
23	16.89	618.5223	C39H70O5	DG(16:0/0:0/20:3n9)	1.480091	0.022021349	2.83719	up
24	17.15	672.4449	C36H64O11	VinaginsenosideR12	1.149489	0.041819694	−1.24196	down
25	17.16	309.3032	C20H39NO	OleoylEthylAmide	1.669653	0.029553385	−1.60311	down
26	17.16	182.1307	C11H18O2	Decahydronaphthalene-2-carboxylic acid	1.752376	0.04116878	−1.39749	down
27	17.22	568.5067	C35H68O5	DG(16:0/16:0/0:0)	1.706035	0.008536144	−2.49324	down
28	17.36	296.3079	C20H40O	Phytol	1.662572	0.03975343	−1.50364	down
29	17.77	112.0888	C7H12O	xi-3-Methyl-3-cyclohexen-1-ol	2.013088	0.022770733	−1.3543	down
30	17.84	157.9625	C2H7O2PS2	Dimethyldithiophosphate	1.819238	0.039248913	−1.90517	down
31	18.24	297.3032	C19H39NO	Tridemorph	1.665878	0.023567774	−1.03526	down

Close

Fig. 8

Heatmap of 31 potential biomarkers.

Export to PPT

3.5.3

3.5.3 Metabolic marker-related pathway analysis

The marker metabolites were entered into the MetaboAnalyst database, and the most relevant metabolic pathways were found through the results of metabolic pathway enrichment analysis and topology analysis. Four metabolic pathways were obtained through enrichment, among which the important metabolic pathway was identified as Glycerophospholipid metabolism (Fig. 9).

Close

Fig. 9

Enrichment of pathway analysis with MetaboAnalyst.

Export to PPT

3.6 Correlation analysis

The potential targets obtained from the network pharmacology were combined with the targets involved in the significant pathways obtained by metabolomics enrichment, and two targets, PIK3R1 and PIK3CA, were obtained. To determine the potential link between the gut microbiota and the endogenous metabolites, correlation analysis was performed between the differential microbes obtained by the gut microbiota and the differential metabolites obtained by metabolomics. The Pearson correlation coefficient of the compounds was used to display the correlation between the microorganisms and the metabolites at the genus level through a cluster heat map. Heatmaps were drawn using the Complex Heatmap package of the R software (Fig. 10).

Close

Fig. 10

Heatmap of Pearson correlation analysis between bacteria and metabolites. (A) Heatmap of model and high-dose group. (B) Heatmap of model and low-dose group.

Export to PPT

The CT water extract can regulate the intestinal flora and metabolites to varying degrees, and the intestinal microbes and small molecule metabolites are closely related. Most of the metabolites are significantly correlated (p < 0.05) to Treponema, Prevotella, and Succinivibrio. Metabolites including 3, 5, 7, 10, 12–14, and 23 were negatively correlated with harmful bacteria Treponema, Prevotella, Succinivibrio, and Ruminobacteria, but positively correlated with the beneficial bacteria Lactobacillus.

4.1 Effects of CT on intestinal microbiota diversity

GLU, TC, and TG in the model group were all increased while the TC of the low-dose in the treatment group of CT was significantly decreased, indicating that CT can improve cholesterol metabolism of type 2 diabetes in rats. The results of 16S rDNA showed that the CT's overall performance is to increase the abundance of beneficial bacteria Lactobacillus and reduce the pathogenic bacteria abundance of Prevotella, Succinivibrio, unidentified_Ruminococcaceae, and Treponema.

Numerous studies have found that tea and herb play a critical role in the treatment of diabetes by adjusting the imbalance of intestinal microflora, promoting the abundance of beneficial bacteria, and inhibiting the reproduction of harmful bacteria. Tea supplementation was effective in reversing these changes (Liu et al., 2019; Liu et al., 2020). Lactobacilli are Gram-positive bacteria present in probiotics that restore the balance of the gut microflora (Usman et al., 2019; Panwar et al., 2014; Piano et al., 2012). The laboratory Lactobacillus used in the study assists in the normal expression of metabolic genes, suppresses inflammatory responses and the stress response to obesity and insulin resistance (Youn et al., 2021). Prevotella belongs to a class of anaerobic gram-negative bacilli and is closely related to low-grade inflammation and insulin resistance (Larsen, 2017). In our study, the Prevotella were significantly increased in T2DM mice and were significantly decreased after the CT intervention. Prevotella is a key factor in regulating inflammation and is mediated by caspase-8, which can induce more inflammatory factors to trigger inflammation. So CT may further improve T2DM by inhibiting inflammation. Succinivibrio can ferment a variety of sugars, and the main metabolic products are acetic and succinic acids while small amounts of formic and lactic acids. Ruminococcus was positively associated with T2DM according to most studies, and one of which showed that Ruminococcus bromii abundance increased and Ruminococcus torques decreased after bariatric surgery and diabetes remission (Yao et al., 2020). In the present study, the abundance of Ruminococcus, Treponema, and Succinivibrio significantly decreased after the intervention of CT in diabetic rats, which was partly contrary to the reported results that the abundance of Treponema sp. increased, and the abundance of Ruminococcus sp. decreased significantly in patients or animals with diabetes (Arangannal et al., 2013; Qu et al., 2018).

4.2 Effects of CT on urine metabolites

The 31 metabolites obtained in this research involved changes in carbohydrate metabolism, amino acid metabolism, and lipid metabolism. The level of 1,6-Anhydro-β-D-glucose in the model group increased and decreased significantly after administration. Dysregulation of glucose and impaired carbohydrate consumption resulting from defective or defective insulin secretion responses are key pathogenic factors in diabetes.

In this experiment, changes of nine amino acid metabolites were detected. Among them, Asn-Leu-Pro-Ile-Asn, Phe-Tyr-Lys-Arg, Glu-Lys-Leu-Thr-His, His-Lys-Leu-Val-Val, Ile-Gly-Lys-Ile-Phe of CT group significantly increased, while Gly-Leu-Arg-Val-Phe, Ala-Pro-Arg-Lys-Lys, Lys-Tyr-Gln-Glu-Ala, Asp-Tyr-Asn-Leu-Leu significantly decreased compared to the model. Studies have reported that impaired amino acid metabolism may lead to an increased risk of diabetes, which are closely related to the occurrence of T2DM in the population (Vanweert et al., 2021; Ramzan et al., 2020; Neinast et al., 2019).

Phosphatidylcholine PC(18:1(9Z)/14:0) and PC(16:0/18:0) were up-regulated in the T2DM model group, suggesting the disorder of phospholipid metabolism in T2DM rats. After the intervention of CT, they were significantly down-regulated, suggesting that CT mainly regulates the disorder of lipid metabolism to improve the process of T2DM. The relevant metabolites of the glycerophospholipid metabolic pathway are a class of phosphatidylcholines. Research analysis shows that the intermediate products of the two lipids are closely related to oxidative stress and inflammation complicated by diabetes (Gu et al., 2010). So the content of PC(18:1(9Z)/14:0) and PC(16:0/18:0) could be used as the basis for judging the therapeutic effect of CT on diabetes.

4.3 Common targets and pathways of urine metabolomics and network pharmacology

PIK3R1 and PIK3CA are important targets obtained from the correlation analysis. They are important mediators of insulin metabolism (Winnay et al., 2016; Sopasakis et al., 2010). Correlation analysis of gut microbiota and metabolomic showed that both high-dose and low-dose of CT groups could modulate gut microbiota abundance, and further adjusted the levels of the corresponding metabolites. And it is further found that gut microbiota and metabolites both act on carbohydrate metabolism, amino acid metabolism, and lipid metabolism. A large number of studies have shown that T2DM is a group of common metabolic endocrine diseases with glucose and fat metabolism disorders. Long-term carbohydrate, fat, and protein metabolism disorders can cause multi-system damage. The central link and pathogenesis of various metabolic problems together lead to the occurrence and development of diabetes. The results demonstrate that CT tea can treat type 2 diabetes rats by regulating the composition of intestinal flora, changing bacteria abundance and levels of endogenous metabolites.

4.4 Effects related chemical components of CT

Flavanomarein, luteolin, isookanin, marein, 7,3′,5′-trihydroxy-flavanone, okanin, and caffeic acid plays key role through SRC, TP53, HSP90AA1, STAT3, MAPK1, PIK3R1, AKT1, PIK3CA, HRAS and JUN. It is suggested that flavonoids are the key active compounds of CT in the treatment of T2DM.

The hypoglycemic effect of CT tea in T2DM is attributed to the multiple active components. Flavanomarein, luteolin, isookanin, marein, 7,3′,5′-trihydroxy-flavanone, and okanin belong to a large class of flavonoids. Studies have shown that they treat diabetes by regulating blood sugar. The mechanism is related to the restoration of pancreatic function, inhibition of α-glucosidase activity and regulation of enzymes in glucose metabolism. The monomer marein extracted from CT has been widely studied. Marein can reverse the high expression of SGLT2 and mRNA protein in HK-2 cells treated with high glucose, and activate p-AMPK/p-ACC protein to inhibit glucose uptake. Animal experiments in vivo show that marein could significantly reduce fasting blood glucose, serum triglyceride, cholesterol, and low-density lipoprotein in db/db mice. Marein can reduce gluconeogenesis through Akt/FoxO1, improve insulin resistance and lipid metabolism abnormalities induced by high glucose in HepG2 cells, and have a certain protective effect on mouse islet β cell line MIN6 damage model induced by high glucose and fat. In addition, it can also achieve the goal of reducing blood sugar by regulating the Bcl-2 protein family (Guo et al., 2020; Jiang et al., 2016a; Jiang et al., 2016b). Luteolin is also a kind of flavonoids. Wang et al. found that luteolin reduced cardiomyocyte apoptosis in diabetic mice by inhibiting RhoA/ROCK2 signal pathway, thus improving diabetic myocardial injury and insulin resistance, regulating disorders of glucose and lipid metabolism, and playing an anti-inflammatory role in the treatment of diabetes and its complications. In addition, the degree value of the caffeic acid is 57, ranking in the forefront. Studies show that caffeic acid has the same inhibitory intensity on α-glucosidase as acarbose in vitro, and can reduce fasting blood glucose in type 2 diabetic rats.

Chrysanthemum, with the Chinese name Juhua, has been used as a diet for>3000 years, especially for health care and scented tea. However, the unique geographical, climatic conditions and processing methods of different chrysanthemum producing areas have gradually led to the diversification of chrysanthemum varieties, including ‘Huaiju’, ‘Boju’, ‘Chuju’, ‘Gongju’, ‘Hangju’ and so on. Chrysanthemum contains phenolic acids, flavonoids, volatile oils, carotenoids, and other chemical components, as well as sugars, amino acids, non-volatile sesquiterpenes, triterpenes, and so on (Zhou et al., 2022; Li et al., 2019).

There are significant differences in flavonoid content and composition among different varieties of chrysanthemum. And these are the reason for their different flavor and function. Chrysanthemum, especially Huaiju and Boju, mainly contain flavonoid aglycone, such as apigenin, luteolin, acacetin, diosmetin, etc. While Gongju and Hangju, as traditional medicinal chrysanthemums, the main active components belong to chlorogenic acid, 3,5-dicaffeoylquinic acid, cynaroside, and quercitrin. Studies have shown that the anti-inflammatory, antibacterial, blood pressure, and blood lipid-lowering pharmacological activities of chrysanthemum are closely related to phenolic acids, flavonoids and volatile oil (Chen et al., 2021; Yang et al., 2016; Dong et al., 2014). Chrysanthemum flavonoids, luteolin, and cymaroside, could reduce the weight and level of total cholesterol, triglyceride, and low-density lipoprotein cholesterol, increase the level of high-density lipoprotein cholesterol in blood, reduce steatosis of liver, and improve the index of liver function (Wang et al., 2021; Salles et al., 2021). Also, chrysanthemum flavonoids have good effects on anti-oxidation and activities of enzymes related to lipid metabolism, and have a strong ability of antioxidation and regulation of lipid metabolism (Yamamoto et al., 2015). Therefore, flavonoids and organic acids derivatives in chrysanthemum demands more comparative study.