2.1 Reagents

Sc and At were purchased from Shenyang Tianyitang (China) and identified by UPLC. Samples were stored in the author's laboratory. Aβ25-35 was obtained from Bioss (BJ10228890, China). TNF- α kit was purchased from Beijing Chenglin Biotechnology Co., ltd (E-30633, China). Aromatase kit was obtained from Shanghai MLBIO Biotechnology Co., ltd (ML734808, China). BCA protein quantification kit was from Beyotime (P0012, China). d-galactose, Standard 17β-Estradiol and chloramphenicol were acquired from Shanghai Yuanye Biotechnology Co., ltd (R24O11H128688, YN1124DA14, D23S10G98511, China). Standard testosterone was purchased from Dr. Ehrenstorfer GmbH (G1088065, Germany). Acetonitrile and methanol were procured from Merck (I1133629 105, I1143835 114, Germany). Ultrapure water was from Watsons (China). Formic acid and ammonium acetate were purchased from Thermo Fisher Scientific (202674,153073, USA). Leucine Enkephalin was from Waters Corporation (W12011813, USA). All other reagents were of HPLC grade.

2.2 Preparation of drugs extract

The drug extraction method was followed according to our group's previous component screening method. Sc-At extracts were prepared by ultrasonic ethanol extraction. Sc and At were weighed as 50 g (1:1, w/w), soaked them in 90 % ethanol for 1 h and extracted them twice with 90 % ethanol for 2 h. The extract was concentrated in vacuum and lyophilized into powder. It was formulated into 1.5 g /mL Sc-At solution in distilled water, put the mixture for ultrasonic dispersion to obtain a soluble solution and stored at 4 °C (Chen, 2022).

2.3 Animals and drug administration

The animal experiments were approved by the animal ethics committee of China Medical University (CMU2019257). Liaoning Changsheng Biotechnology Co. ltd provided 8-week-old male Wistar rats (weight 200 g ± 20 g) and were kept at 24 ± 2 °C with humidity of 55 ± 5 %. A 12 h light–dark cycle was set and rats were given access to standard diet and water. After one week of environmental adaptation, they were randomly divided into three groups: Blank, Model, and Sc-At. Model, and Sc-At groups were given 120 mg/kg/d of d-galactose and 40 mg/kg/d of aluminum trichloride (Chiroma, 2018). From 5th week, Sc-At group began to receive 2.5 g/kg/d Sc-At treatment. Until the 8th week, both Model and Sc-At groups were injected with 10 μg/side of Aβ 25–35 into bilateral ventricles (Zuo, 2018; McDonald et al., 2021). Sc-At group continued to get Sc-At for 1 week. The previous results showed significant difference in serum estrogen levels between ovariectomized APP mice and sham-operated APP mice, however there was no difference in brain estrogen levels (Yue, et al., 2005). Therefore, castration was not performed in this experimental design.

2.4 Morris water maze (MWM) test

MWM device consisted of circular pool with 50 cm depth and 160 cm diameter, circular platform with 10 cm diameter escape platform, tracking camera acquisition equipment, and computer analysis system. It was divided into four quadrants on average. The escape platform was fixed in the center of fourth quadrant, and various geometric shapes were set around the pool wall to assist the positioning memory of experimental rats. The day before the experiment, pool was cleaned, peculiar smell was removed, and pool was filled with water until the depth was about 1.5 cm above the escape platform. An appropriate amount of black ink was then added to the water, stirred to make it turbid for hiding the escape platform. Water temperature was adjusted to 25 ± 2 °C and the experimental environment was set to the best light. The original platform was removed on the 6th day, and rats were placed at any water entry point in the water. All rats were at the same water entry point and numbers of times the rats crossed original platform in 2 min were recorded.

2.5 Sample collection

Samples were collected after MWM test. Rats groups were decapitated and brain tissue samples were collected on day 73. Half were fixed in 10 % formalin and the other half stored at −80 °C.

2.6 Hematoxylin-Eosin (H & E) staining

Brain tissue fixed in 10 % formalin was removed, dehydrated with ethanol solution, embedded in paraffin, and sliced into 5 μm with section cutter. Staining was done with H & E, observed, and photographed under visible light microscope (Hyman, 1984).

2.7 TNF-α inflammatory factor expression assay

AD is a chronic inflammatory disease and studies have reported that TNF-α signaling exacerbates Aβ and tau protein pathology, and inhibition of TNF-α expression delays cognitive impairment in AD patients. The experiments were performed according to the commercial TNF-α Elisa kit.

2.8 Untargeted metabolomics

2.9 Network pharmacological analysis

The active components of Sc-At were screened from TCMSP database (https://tcmsp-e.com/) according to the oral bioavailability (OB) ≥ 30 % and drug likeness (DL) ≥ 0.18. Quantitative estimate of drug-likeness (QED) were evaluated through ETCM database (http://www.tcmip.cn/ETCM/index.php/Home/Index/), QED: >0.67 (Good), QED: 0.67–0.49 (Moderate), and QED < 0.49 (Weak) (Xu, 2020). The canonical smiles of active ingredients were obtained through PubChem database (https://pubchem.ncbi.nlm.nih.gov/). The corresponding target was predicted through TargetNet database (http://targetnet.scbdd.com/) using canonical smiles of active ingredients, and selected prob = 1.0. Prediction of disease targets was made using GeneCards database (https://www.genecards.org/) with chosen score of ≥ 30. The target of Sc-At in AD treatment was obtained by Venn diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/). These targets were imported into UniProtKB (http://www.uniprot.org/) to standardize the gene and protein names. A protein–protein interaction (PPI) network was established by STRING 11.0 (https://string-db.org/). Hub genes were identified based on degree. Enrichment of KEGG pathway was done through David database (https://david.ncifcrf.gov/tools.jsp). GO enrichment analysis was also performed. C-T-P-D networks were constructed by Cytoscape 3.7.2.

2.10 Bioinformatics analysis of Sc-At targets related to Aβ and tau pathology

The correlations between potential targets and AD disease were further explored. Screened targets were imported to CFG Rank module of AlzData database (http://www.alzdata.org/index.html) for the difference analysis, Pathology cor (Aβ): correlation of target gene expression with AD pathology in Aβ line AD mouse models, Pathology cor (tau): correlation of target gene expression with AD pathology in tau line AD mouse models, CFG: total CFG score of target gene, 1 CFG point was added if any of the above evidence was significant; CFG point ranged from 0 to 5 (Xu, 2018).

Sc-At's normalized expression targets against AD in control and AD groups of GEO dataset were analyzed by “Differential expression” module of AlzData database (Zeng, 2021). The brain regions were divided into four areas: hippocampus, frontal cortex, entorhinal cortex, and temporal cortex. The respective statistical analysis was carried out. Visualization was made using GraphPad Prism software. Values were presented as mean ± SD. A total of 11 datasets from GEO database were used.

2.11 Molecular docking

The 3D structure of Sc-At active ingredient was obtained from Chemical Book (https://www.chemicalbook.com/ProductIndex.aspx). The crystal structures of hub targets were acquired from PDB (https://www1.rcsb.org/). The protein was introduced into MOE (2015.10), and the structure was optimized by adding hydrogen atoms and removing water. A site finder was used to select the best docking pocket. Compounds were treated with minimum energy, the molecular docking parameter replacement was set to triangle matcher, rescoring 1 to ASE, refinement to the forcefield, rescoring 2 to ASE, and default for other parameters. Finally, the docking results were achieved.

2.12 Analysis of related metabolites and targets

The compound-reaction-enzyme-gene network was constructed to further reveal the relation between metabolites and potential targets (Li, 2021). KEGG IDs of 30 screened differential metabolites and the gene symbols of hub targets were imported into MetScape plugin of Cytoscape software to construct a pathway-based correlation network map.

2.13 Aromatase activity verified by ELISA

The aromatase activity in rat brain was detected through commercial ELISA kit by following the manufacturer’s instructions. The protein concentration was determined by BCA protein quantification kit and the actual protein activity was calculated for each sample.

2.14 Targeted metabolomics

2.15 Statistical analysis

The data was expressed as the mean ± SD. Statistical analysis was performed by one-way analysis of variance (ANOVA) using GraphPad Prism 9.0. A p-value of < 0.05 was statistically significant.

3.1 MWM test

In the space exploration experiment on 6th day, the number of crossing platform for model group decreased compared to the blank group. The number of crossing platform of Sc-At increased compared to the model group. The results showed that Sc-At had significant effect on increasing the number of crossing platform of AD rats (Fig. 2A-B).

Close

Fig. 2

(A) In MWM test, the motion trajectories of three groups of rats. (B) In MWM test, the times of crossing the platform of the three groups of rats were compared (n = 6, ***P < 0.001, *P < 0.05). (C) The results of H & E staining in the hippocampus of rat brain tissue. (D) Expression of TNF-α in rat brain tissue. (n = 3, *P < 0.05).

Export to PPT

3.2 H & E staining

The neuronal layer cells in hippocampal CA1 area were observed by H & E staining. The neurons in blank group's CA1 region of hippocampus were tightly arranged with large and regular nuclei. The neurons in CA1 area of hippocampus in model group were sparse and disordered. Compared with model group, the rats in Sc-At group showed that the neurons in CA1 area of hippocampus in rats were arranged more neatly, and the numbers of cells were also significantly greater (Fig. 2C).

3.3 Expression of TNF-α in rat brain tissue

In the model group, expression of TNF-α was significantly higher than that in blank group and Sc-At groups. (Fig. 2D).

3.4 Analysis of untargeted metabolomics results

3.4.1

3.4.1 Data analysis

The brain tissue samples were detected in positive and negative modes and QC was inserted to every-six samples to evaluate the instrument's stability. The representative total ion chromatography through ESI + and ESI- is shown in Fig. 3. The supervised PLS-DA diagram showed that QC group gathered together proved the high stability of instrument, the samples within the group were gathered together while samples between the groups were different. The model group was far from the blank, indicating difference between the two. Sc-At group was closer to the blank, revealing that the metabolic disorder caused by disease had been recovered after its administration and treatment, (ESI+, R²Y = 0.896, Q² = 0.655; ESI-, R²Y = 0.998, Q² = 0.757). The results of 200- Permutation test for PLS-DA analysis showed that PLS-DA model was not over-fitted, and the model results were reliable. OPLS-DA diagram showed that the model group was significantly separated from the blank, and differential metabolites were being tracked in this model. These results showed that Sc-At improved the metabolic disorder in the brain of AD rats (Fig. 4A-H). (OPLS-DA parameters of each group and Permutation test results are shown in Table S1 and Figure S1).

Close

Fig. 3

Representative total ion chromatography of (A) ESI+ (B) ESI-.

Export to PPT

Close

Fig. 4

Results of metabolomic analysis of brain tissue based on liquid mass spectrometry. ESI+ (A) PLS-DA score plots. (B) PLS-DA analysis of 200 permutation test (C) OPLS-DA score plots Model vS Blank. (D) OPLS-DA score plots Model vS Sc-At. ESI- (E) PLS-DA score plots. (F) PLS-DA analysis of 200 permutation test (G) OPLS-DA score plots Model vS Blank. (H) OPLS-DA score plots Model vS Sc-At.

Export to PPT

3.4.2

3.4.2 Screening differential metabolites and pathway enrichment

789 metabolites were identified in ESI + mode and 254 in ESI- mode. According to VIP > 1 and P < 0.05, 30 metabolites were differentially expressed between the model and blank groups, 18 differential metabolites between the Sc-At and model groups, 19 differential metabolites between Sc-At and blank groups. The differential metabolites between the Model and Blank groups are mentioned in Table 1. Compared with the blank, the content of 26 metabolites in model group decreased after administration, the content of 2 metabolites increased after administration, and the trend is reflected in Figure S2. Differential metabolites for model and Sc-At groups are in Table S2, differential metabolites for the blank and Sc-At groups are in Table S3. The unsupervised clustering heat map of differential metabolites between the model and blank groups, the blank and Sc-At groups was drawn, and correlation between differential metabolites was analyzed (Fig. 5A-C).

Table 1 Differential metabolites identified in brain tissue between the blank and model groups.

Description	Compound ID	Formula	Adducts	m/z	VIP	Anova (p)	Mass Error (ppm)	Fold change	Scan mode	Trend (M/B)
3-Dehydroxycarnitine	HMDB0006831	C7H15NO2	M + K	184.07	1.13	2.13E-03	4.63	1.71	ESI+	↓
N-Formyl-l-glutamic acid	HMDB0003470	C6H9NO5	M + Na, M + K	198.04	1.57	4.06E-09	2.83	1.53	ESI+	↓
Proflavine	HMDB0015255	C13H11N3	M + NH4	227.13	1.61	2.45E-06	−1.49	1.91	ESI+	↓
Eplerenone	HMDB0014838	C24H30O6	M + H, M + Na, M + K, M + NH4, M + H-H2O	415.21	3.14	6.76E-04	4.32	1.82	ESI+	↓
16-alpha-Hydroxyandrosterone	HMDB0013156	C25H47NO5	M + K	480.31	2.79	2.22E-03	2.56	3.95	ESI+	↓
LysoPE(18:0/0:0)	HMDB0011130	C23H48NO7P	M + H, M + Na, M + H-H2O	482.32	4.05	1.10E-02	1.13	2.24	ESI+	↓
LysoPC(16:0)	HMDB0010382	C24H50NO7P	M + H	496.34	2.43	1.77E-03	1.15	1.64	ESI+	↓
LysoPC(18:2(9Z,12Z))	HMDB0010386	C26H50NO7P	M + H	520.34	1.23	2.91E-03	4.53	3.62	ESI+	↓
LysoPC(18:1(11Z))	HMDB0010385	C26H52NO7P	M + H, M + Na	522.36	6.63	1.07E-05	0.51	2.33	ESI+	↓
LysoPC(18:0)	HMDB0010384	C26H54NO7P	M + H, M + Na	524.37	4.26	8.78E-03	0.18	2.85	ESI+	↓
LysoPC(20:4(5Z,8Z,11Z,14Z))	HMDB0010395	C28H50NO7P	M + H	544.34	4.49	2.09E-04	1.43	1.86	ESI+	↓
LysoPC(22:6(4Z,7Z,10Z,13Z,16Z,19Z))	HMDB0010404	C30H50NO7P	M + H	568.34	3.95	4.41E-04	1.46	1.93	ESI+	↓
Ceramide (d18:1/18:0)	HMDB0004950	C36H71NO3	M + Na	588.53	5.85	1.19E-03	3.44	4.89	ESI+	↑
PC(16:0/16:0)	HMDB0000564	C40H80NO8P	M + H	734.57	12.51	4.11E-02	3.07	1.53	ESI+	↓
PC(14:0/20:3(5Z,8Z,11Z))	HMDB0007881	C42H78NO8P	M + H	756.56	4.60	1.78E-03	4.30	5.23	ESI+	↓
PE(18:3(6Z,9Z,12Z)/20:5(5Z,8Z,11Z,14Z,17Z))	HMDB0009137	C43H70NO8P	M + K	798.45	3.66	2.32E-07	−1.99	2.07	ESI+	↓
PE(18:3(9Z,12Z,15Z)/22:0)	HMDB0009171	C45H84NO8P	M + K	836.55	1.12	2.58E-03	−2.72	1.41	ESI+	↓
PE(20:2(11Z,14Z)/24:1(15Z))	HMDB0009311	C49H92NO8P	M + H-H2O	836.66	1.45	3.70E-02	4.74	2.94	ESI+	↓
trans-2-Enoyl-OPC6-CoA	HMDB0011121	C37H58N7O18P3S	M + H-H2O	996.28	2.20	8.12E-07	2.59	1.43	ESI+	↓
l-Tryptophan	HMDB00929	C11H12N2O2	M−H	203.08	1.11	3.16E-03	−1.77	1.58	ESI-	↓
Palmitoleic acid	HMDB03229	C16H30O2	M−H	253.22	1.03	1.23E-03	1.50	1.52	ESI-	↓
Inosine	HMDB00195	C10H12N4O5	M−H	267.07	3.29	2.52E-03	2.18	1.22	ESI-	↑
Linoleic acid	HMDB00673	C18H32O2	M−H	279.23	2.34	4.33E-02	2.31	1.56	ESI-	↓
Elaidic acid	HMDB00573	C18H34O2	M−H	281.25	3.48	5.82E-04	2.83	1.42	ESI-	↓
Mesterolone	HMDB06036	C20H32O2	M−H	303.23	8.35	1.40E-04	2.84	1.37	ESI-	↓
8,11,14-Eicosatrienoic acid	HMDB02925	C20H34O2	M−H	305.25	1.02	6.10E-04	1.58	2.03	ESI-	↓
Docosahexaenoic acid	HMDB02183	C22H32O2	M−H	327.23	7.57	2.45E-05	3.24	1.71	ESI-	↓
Adrenic acid	HMDB02226	C22H36O2	M−H	331.27	1.67	6.36E-06	2.71	1.64	ESI-	↓
DG(18:1(9Z)/18:2(9Z,12Z)/0:0)	HMDB07219	C39H70O5	M + K-2H	655.47	1.38	3.25E-07	1.81	4.74	ESI-	↓
TG(16:0/16:1(9Z)/20:4(5Z,8Z,11Z,14Z))[iso6]	HMDB05380	C55H96O6	M + K-2H	889.67	1.71	4.27E-02	0.82	10.00	ESI-	↑

Close

Fig. 5

Unsupervised clustering heat map of the differential metabolites between (a) the model and the blank group, (b)the blank group and the sc-at group (c)correlation analysis between different metabolites (d) kegg pathway analysis of brain samples collected from ad rats.

Export to PPT

Further, the metabolic pathways affected by Sc-At treatment of AD were explored, and the obtained differential metabolites were enriched through MetaboAnalyst5.0, Linoleic acid metabolism, Biosynthesis of unsaturated fatty acids, Glycerophospholipid metabolism, and alpha-Linolenic acid metabolism pathway were significantly affected（P < 0.05） (Fig. 5D, Table S4).

3.5 Network pharmacological analysis results

According to OB ≥ 30 % and DL ≥ 0.18, 9 active compounds had been identified in Sc and 5 in At. TargetNet database predicted the target of active compounds, and 71 targets were predicted. AD-related targets finally determined 14 active compounds by regulating 54 AD related targets and constructed the component-target-pathway network diagram (Fig. 6A). PPI network was made by Cytoscape to further explore the key genes of Sc-At in AD treatment. Hub genes were calculated according to degree where ESR1, PTGS2, and CYP19A1 were the hub genes (Fig. 6B). The role of potential targets in AD treatment was further studied, KEGG pathway was enriched and the top 20 according to P-value were selected to construct the bubble diagram, including the steroid hormone biosynthesis and tyrosine metabolism (Fig. 6C). Moreover, GO enrichment analysis was carried out and top 20 were screened according to p-value. The response to lipid process in biological system was closely related to the development of AD (Fig. 6D-E).

Close

Fig. 6

Network pharmacology analysis of Sc-At treating AD. (A) The constructions and analysis of Compound-Target-Pathway-Disease (C-T-P-D) network of Sc-At. Candidate compounds (yellow diamond) and different pathological properties (good: red border, moderate: blue border, week: green border and N/A: grey border), putative targets (green ellipse) and AD-associated proteins (red ellipse), pathways (green rectangle) and diseases (purple V) formed the C-T-P-D network of Sc-At. (B) The PPI network of Sc-At treatment on AD. Node color reflects its degree. The nodes with red borders represent the hub genes. (C) The KEGG pathways enrichment analysis by David database. (D-E) GO enrichment analysis.

Export to PPT

3.6 Bioinformatics analysis of Sc-At targets related to Aβ and tau pathology

The relationship between target, and Aβ and tau protein was explored where 14 out of 54 potential targets were correlated with Aβ and 7 with tau protein. Higher the CFG, higher was the correlation with AD. The score of CYP19A1 was 3 (Table S5).

Compared to the controls, ADRA2A and TUBA1A were higher in AD patients while AR, MCL1, and RELA were lower. These results indicated that Sc-At related anti-AD targets were closely associated with the pathology of Aβ and tau (Fig. 7A).

Close

Fig. 7

(A) Targets of Sc-AT against AD in the control and AD groups of the GEO dataset. (Entorhinal cortex, n = 39 in each group. Hippocampus, n = 66 in the control group, n = 74 in the AD group. Frontal Cortex, n = 88 in the control group, n = 73 in the AD group. Temporal cortex, n = 39 in the control group, n = 52 in the AD group. Values are presented as the mean ± SD.) Molecular docking results. （B）The 3D interaction diagrams of aromatase. (C) The 2D interaction diagrams of aromatase.

Export to PPT

3.7 Molecular docking simulation

Molecular docking simulation prediction was carried out according to the key targets screened by network pharmacology, and the targets of strong correlation with AD achieved by bioinformatics analysis. S was scored according to Gibbs free energy. The docking score for CYP19A1 was optimal (Table S6), for example between CYP19A1 and Schizandrer B, hydrogen bond was formed through Cys at position 437. The results are shown in Fig. 7B-C.

3.8 Comprehensive analysis of metabolomics and network pharmacology

PTGS, XDH, and CYP19A1 were the key targets as shown in Fig. 8. Related metabolites were arachidonic acid, linoleic acid, inosine, estrogen, etc. They may play essential role in treating AD by Sc-At.

Close

Fig. 8

The compound-reaction-enzyme-gene networks of the key metabolites and targets. The red hexagons, grey diamonds, green round rectangle, and blue circles represent the differential metabolites, reactions, proteins, and genes, respectively, and the target determination metabolites with a blue border. The key metabolites, proteins, and genes were magnified.

Export to PPT

3.9 Detection of aromatase activity by ELISA

Based on the molecular docking results, the scores of CYP19A1 and active compounds in Sc-At were the highest, and AlzData database showed that CYP19A1 had correlation with AD pathology. However, the verification experiment of GEO database showed that compared with the control, the gene CYP19A1 of AD group had no significant changes in all regions of brain tissue (Prange-Kiel, 2016). The aromatase activity was tested as per the findings of relevant literature on therapeutic mechanism of effective components in Sc and At. The aromatase activity in model group decreased compared to the blank and increased after Sc-At treatment. The enzyme activities of three groups are shown in Fig. 9A.

Close

Fig. 9

(A)Effect of Sc-At extracts on aromatase activity in the brain of Rats. The chromatogram obtained through the final UPLC/MS/MS showed good resolution among the two compounds. (B)Testosterone and Chloramphenicol(C)17β-Estradiol and Chloramphenicol. Testosterone and 17β-Estradiol levels in rat brain. (D)Testosterone (E)17β-Estradiol. ALL data were expressed as means ± SD, (n = 3, *P < 0.05, **P < 0.01, ***P < 0.001).

Export to PPT

3.10 Analysis of targeted metabolomics results

Each compound's standard solution (2 μg/mL) was infused for MS/MS optimization (Table 2). The chromatograms of test substance and IS chloramphenicol are shown below. IS was well separated from testosterone (Fig. 9B) and 17β-Estradiol (Fig. 9C). The standard curves of testosterone and 17β-Estradiol had an excellent linear relationship. Results showed that the content of testosterone in model group increased compared with the blank, while content of 17β-Estradiol decreased. The testosterone content decreased in Sc-At group compared to the model group, while the content of 17β-Estradiol increased (Fig. 9D-E).