Six-week-old male Sprague Dawley (SD) rats were purchased from SiPeiFu (Beijing) Biotechnology Co., Ltd. (SCXK (Beijing) 2019–0010). The rats were housed in cages at a temperature of 22°C ± 2°C and humidity of 55% ± 5%, with a 12/12-h light/dark cycle in an SPF animal facility (containing sterilized water and feed). The animal experiment was approved by the Animal Ethics Committee of Xiyuan Hospital of China Academy of Chinese Medical Sciences (approval number: 2024XLC014-2) and adhered to the Guide for the Care and Use of Laboratory Animals from the National Institute of Health (NIH Publication No. 85–23, revised 1996).

We employed an STZ combined with VitD3 overload to induce the DVC model. STZ is a β-cell toxin and is the most commonly used method to induce diabetes in rodent models (Furman, 2021), and it is widely applied in the study of diabetic vascular complications (Stabley and Towler, 2017). STZ not only destroys pancreatic β-cells but also accelerates vascular injury and biomineralization reactions through DNA alkylation (Bennett and Pegg, 1981). This results in aggravated vascular endothelial cells dysfunction (Mao et al., 2024), along with increased mineral deposition and stiffness (Heath et al., 2014), which is more closely mimic the clinical conditions observed in patients with diabetic vascular complications. After 1 week of acclimatization, intraperitoneal injections of STZ (Sigma-Aldrich, United States) at a dose of 30 mg/kg were administered to the rats for three consecutive days. One week later, fasting blood glucose levels were measured. Rats with blood glucose levels ≥11.1 mmol/L and pronounced symptoms of “polyuria, polydipsia, and weight loss” were considered to have successfully developed diabetes to accelerate VC development and mitigate the effects of aging on VC formation, a previously reported method (Yang et al., 2023) was adopted and improved by using VitD₃ (Sigma-Aldrich, United States) overload to induce DVC. Specifically, diabetic rats received intraperitoneal injections of high-dose VitD₃ (5.5 × 10⁵ IU/kg) for four consecutive days. 14 days after the first injection is considered as the end of calcification modeling period, and its modeling success rate has been proved to be over 80% (Lu et al., 2024). Another group of SD rats were used as a normal control group (CON group) and received equivalent volumes of vehicle (pure water) instead of the STZ and VitD₃ injections. All groups of rats were granted free access to food and water throughout the study period.

The DLTM comprises of Salvia miltiorrhiza Bunge. (20 g), Coptis chinensis Franch. (10 g), Polygonum cuspidatum Siebold & Zucc. (10 g), Prunus mume (Siebold) Siebold & Zucc. (10 g), and Rhodiola crenulata (Hook.f. & Thomson) H. Ohba. (10 g), and mixed at a ratio of 2:1:1:1:1. In brief, 600 g of Salvia miltiorrhiza Bunge., 300 g of Coptis chinensis Franch., 300 g of Polygonum cuspidatum Siebold & Zucc., 300 g of Rhodiola crenulata (Hook.f. & Thomson) H. Ohba., and 300 g of Prunus mume (Siebold) Siebold & Zucc. were weighed. These botanical drugs were provided by Institute of Traditional Chinese Medicine, Chinese Academy of Traditional Chinese Medicine. The botanical drugs decocted twice with eight and six times the amount of pure water, respectively, for 1 h each time. The decoctions were subsequently filtered, combined, and left to stand overnight, after which the supernatant was collected. The supernatant was passed through a 6L SP825 macroporous resin column at a flow rate of 4.2 L/h, washed with 4 bed volumes (BV) of water, dried, and eluted with 3 BV of 65% ethanol. The ethanol was then recovered, and the solution was concentrated into a thick paste. This process was repeated, and the pastes were combined, vacuum-dried at 80°C, ground into a fine powder, and mixed with dextrin to a final weight of 153.2 g. Each gram of dry extract contained 23.5 g of crude drug.

After model induction, the DVC rats were randomly divided into a model (DVC group), low-dose DLTM (DLTM-L group), medium-dose DLTM (DLTM-M group), high-dose DLTM (DLTM-H group), or positive control Dapagliflozin (DAPA group) group (n = 12 for each group). The groups received different doses of DLTM and Dapagliflozin (AstraZeneca Pharmaceuticals, United States) by gavage for 4 weeks. The dosage of DLTM is determined with reference to the clinically equivalent dose, and the dose gradients are set at 1, 2, and 4 times the equivalent dose based on previous studies (Liu et al., 2024). Specifically, The human equivalent oral dose of DLTM dry extract (0.04 g/kg/day for a 60 kg person) was converted to the rat equivalent dose based on body surface area (BSA) normalization (Reagan-Shaw et al., 2008), resulting in doses of 0.24 g/kg dry extract for the DLTM-L group, 0.48 g/kg dry extract for the DLTM-M group, and 0.96 g/kg dry extract for the DLTM-H group. Alternatively, the DAPA group received oral Dapagliflozin at 5 mg/kg/day based on a previous study (Zhang et al., 2021), while the DVC and CON groups received equivalent volumes of the vehicle by gavage. The survival status, body weight, and fasting blood glucose of the rats were recorded weekly. After the drug intervention period, the rats were fasted for 12 h, anesthetized with an intraperitoneal injection of sodium pentobarbital (30 mg/kg), and blood and aortic samples were collected for further analysis.

UHPLC-MS was conducted using a Vanquish UHPLC system (Thermo Fisher Scientific, United States) coupled with a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific, United States). The chromatographic column used was a HYPERSIL GOLD C18 (100 × 2.1 mm, 1.9 μm, Thermo Fisher Scientific, United States). The flow rate was set to 0.3 mL min⁻¹, the column temperature to 40°C, and the injection volume to 5 μL. The mobile phases consisted of 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). The gradient elution program was as follows: 0–3.0 min, 5% B; 3.0–8.0 min, 5%–14% B; 8.0–16.0 min, 14%–40% B; 16.0–21.0 min, 40%–45% B; 21.0–26.0 min, 45%–65% B; 26.0–29.0 min, 65%–100% B; 29.0–34.0 min, 100% B; 34.0–35.0 min, 100%–5% B; and 35.0–40.0 min, 5% B.

Metabolite analysis was performed using an electrospray ionization (ESI) source with the following parameters: sheath gas flow rate, 40 arb; auxiliary gas flow rate, 10 arb; ion transfer tube temperature, 320°C; auxiliary gas heater temperature, 350°C; S-Lens RF level, 60 V; full MS resolution, 120,000; MS/MS resolution, 60,000; and scan range, 100–1,500 m/z. The spray voltage was 3.8 kV for positive ion mode and 3.0 kV for negative ion mode.

As previously described (Lu et al., 2024), alizarin red staining was employed to assess arterial calcification. For whole aorta staining, the aorta was fixed in 95% ethanol for 24 h and stained overnight with 1% alizarin red solution (Beyotime, C0140). The arterial tissue was then rinsed with 2% potassium hydroxide and photographed using an inverted microscope.

For aortic ring staining, freshly isolated aortic arches were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned at a thickness of 5 μm, and deparaffinized. The aortic sections were subsequently stained with 1% alizarin red solution for 5 min and photographed using an inverted microscope. Quantitative analysis of positively stained areas was performed using Image J software.

Aortic tissues were homogenized and centrifuged to separate the supernatant. ALP activity in the supernatant was then measured using an ALP assay kit (Beyotime, P0321S). Calcium content was determined using a commercial calcium assay kit (Beyotime, S1063S) according to the manufacturer’s instructions. Briefly, the supernatant was mixed with methylthymol blue (MTB), an alkaline solution, and a protein clearing reagent, and incubated at room temperature for 5 min. The absorbance was measured at 610 nm using a microplate reader, and the relative calcium content was normalized to the protein content.

Fasting blood glucose levels were measured weekly using a portable glucometer (ROCHE, Germany) after STZ injections. Serum calcium and phosphorus levels were measured using a Mindray BS-420 automatic biochemical analyzer (Mindray, China).

Serum levels of IL-6, IL-1β, and TNF-α were measured using enzyme-linked immunosorbent assay (ELISA) kits (MEIMIAN, China) with catalog numbers MM-0190R2, MM-0047R2, and MM-70625R2, respectively. The assay procedures adhered to the manufacturer’s instructions.

Total RNA was extracted from the rat aorta using Trizol reagent (CapitalBio Technology, Beijing) following the manufacturer’s protocol. RNA integrity was assessed using an Agilent 2,100 Bioanalyzer (Agilent Technologies, United States). Libraries were prepared using the TruSeq Stranded Total RNA and Ribo-Zero Gold kits (Illumina) and sequenced on an Illumina NovaSeq 6,000 platform (Illumina, United States). Raw data in fastq format were filtered using fastp software, and clean reads were mapped to the rat genome (Rnor_6.0 Ensembl 104) using Hisat2. Gene and transcript expression levels were quantified using featureCounts and StringTie. Differential expression analysis was performed using the limma R package, with thresholds set at p < 0.05 and |log2 (Fold Change)| ≥ 1.2 for significant differentially expressed genes (DEGs). DEG enrichment analysis was conducted using DAVID 6.8 and further analyzed using R packages.

Total RNA was extracted from the aorta using Trizol reagent (Invitrogen, United States) according to the manufacturer’s instructions, the concentration of which was measured using a NanoDrop spectrophotometer (Thermo Fisher Scientific, United States). RNA was reverse-transcribed into complementary DNA (cDNA) using the M-MLV Reverse Transcriptase kit (Promega, United States). PCR amplification was performed on the obtained cDNA using the QuantiTect Multiplex RT-PCR Kit (Qiagen, Germany). Real-time qPCR was conducted using SYBR Green Master Mix (Takara, Japan) on a 7500 FAST Real-Time PCR System (Applied Biosystems, United States). With β-actin as the internal control, relative mRNA expression levels were standardized and analyzed using the 2^−ΔΔCT method. Primer sequences specific to target genes are listed in Table 1.

Aortic tissues were lysed in RIPA buffer containing protease and phosphatase inhibitors. Protein concentration subsequently was determined using a BCA protein assay kit (Solarbio, China). Next, proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, United States). These membranes were blocked with 5% non-fat milk for 1 h at room temperature and incubated with primary antibodies overnight at 4°C. The next day, the membranes were further incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Bands were visualized using an imaging system (Amersham Imager 600) and quantified with ImageJ software. The primary antibodies used in Western blot are listed in Supplementary Table S1.

The acquired DLTM metabolites were screened based on Lipinski’s rule of five followed by using SwissADME (http://www.swissadme.ch/) to identify those with high gastrointestinal (GI) absorption and at least three out of five Lipinski, Ghose, Veber, Egan, and Muegge criteria (Lipinski et al., 2001). A total of 59 metabolites were identified for molecular docking. The protein 3D structures of these metabolites were obtained from the PDB database (http://www.pdb.org/) and saved as PDB files. The only exception was DLL3, whose structure was not available in the PDB dataset and was therefore predicted using the AlphaFold Protein Structure Database. Subsequently, PyMOL 2.4.1 was employed to remove water and ligands from proteins, AutoDock Vina 1.1.2 for hydrogen addition and charge calculation, and ChemOffice 22 for minimizing the energy of small molecule metabolites. Molecular docking was performed using AutoDock Vina 1.1.2. The binding affinity between targets and metabolites was evaluated using the docking score.

The protein-ligand complexes with the lowest binding energy was subjected to a 100 ns molecular dynamics simulation using Gromacs 2023 (Abraham et al., 2015). The topology of the protein was constructed using the CHARMM36 (Jo et al., 2008) force field parameters, while the topology of the ligand was generated based on the GAFF2 force field parameters. The protein-ligand complex was placed within a cubic box, the TIP3P water model was then used to fill the box with water molecules (Harrach and Drossel, 2014). Electrostatic interactions were handled using the Particle Mesh Ewald (PME) method and the Verlet algorithm. Subsequently, the system underwent 100,000 steps of isothermal-isobaric ensemble equilibration and isothermal-isochoric ensemble equilibration, with a coupling constant of 0.1 picoseconds, lasting for a duration of 100 picoseconds. Both van der Waals and Coulomb interactions were calculated using a cutoff distance of 1.0 nm. Finally, the system was subjected to molecular dynamics simulation under constant temperature (300 K) and constant pressure (1 bar) using Gromacs 2023, totaling 5,000,000 steps with a step size of 2 femtoseconds, resulting in a total simulation time of 100 nanoseconds. Using GROMACS internal tools to evaluate root-mean-square deviation (RMSD), radius of gyration (Rg), solvent accessible surface area (SASA), root-mean-square fluctuation (RMSF) and hydrogen bond formation (HBond). We used MM/PBSA to calculate the binding free energies of all complexes.

Statistical analysis and data visualization were conducted using IBM SPSS Statistics (V26.0), GraphPad Prism (V9.5.0), and ImageJ software. Normally distributed continuous variables are presented as mean ± SEM, while non-normally distributed variables are presented as median and interquartile range. For normally distributed data, one-way ANOVA or t-tests were used to determine statistical significance of differences among groups. For non-normally distributed data, the Kruskal–Wallis or Mann-Whitney U test was instead utilized. A p-value <0.05 was considered statistically significant.

UPLC-MS/MS was employed to establish the quality control of DLTM and to quantify its key metabolites. A total of 108 metabolites were identified (Figure 1), derived from the five botanical drugs, namely Salvia miltiorrhiza Bunge., Coptis chinensis Franch., Polygonum cuspidatum Siebold & Zucc., Prunus mume (Siebold) Siebold & Zucc., and Rhodiola crenulata (Hook.f. & Thomson) H. Ohba. Specifically, 28 metabolites were identified from Salvia miltiorrhiza Bunge., 28 from Coptis chinensis Franch., 21 from Polygonum cuspidatum Siebold & Zucc., 25 from Prunus mume (Siebold) Siebold & Zucc., and 27 from Rhodiola crenulata (Hook.f. & Thomson) H. Ohba. Notably, some metabolites were present in multiple botanical drugs. For instance, ferulic acid was found in all five botanical drugs. Detailed information on all identified metabolites in DLTM is provided in Supplementary Table S2.

To elucidate the effect of DLTM on DVC, a DVC rat model was established using STZ combined with VitD₃, with Dapagliflozin as a positive control. Previous studies have confirmed the efficacy of SGLT2 inhibitors in treating DVC (Chen et al., 2023; Li X. et al., 2024). For example, Dapagliflozin has been shown to significantly alleviate arterial calcification by suppressing endoplasmic reticulum stress and the osteogenic transdifferentiation of VSMCs (Li L. et al., 2024; Wu et al., 2024). In this study, after 4 weeks of DLTM or Dapagliflozin administration, the overall condition and vascular damage of the rats were assessed (Figure 2A). As shown, the DVC group exhibited typical diabetic symptoms such as emaciation and elevated blood glucose compared to the CON group (Figures 2B, C; Supplementary Figure S1; Supplementary Table S3, P < 0.05). VitD₃ injection significantly increased serum calcium levels in rats (Figure 2D; Supplementary Table S3, P < 0.05) but exhibited no effect on serum phosphorus levels (Figure 2E; Supplementary Table S3, P > 0.05). Compared to the DVC group, different doses of DLTM effectively reduced serum calcium levels in rats (Figure 2D, P < 0.05, P < 0.01), but there was no significant improvement in body weight, FBG and serum phosphorus levels (Figures 2B, C, E, P > 0.05). In contrast, Dapagliflozin effectively lowered blood glucose levels in rats (Figure 2C, P < 0.05) but had no significant effect on body weight and calcium-phosphorus metabolism (Figures 2B, D, E, P > 0.05). Regarding vascular injury, Gross and aortic alizarin red staining revealed extensive calcium deposition in the aortas of DVC rats (Figures 2F, I, J, P < 0.05), consistent with significant increases observed in vascular calcium content and ALP levels (Figures 2G, H, P < 0.001). Alizarin red staining showed a dose-dependent reduction in aortic calcium deposition in DVC rats treated with DLTM, with high-dose DLTM significantly outperformed low and medium doses of DLTM in improving aortic calcification in rats, reducing vascular calcium content and ALP activity, and exhibited effects comparable to those of Dapagliflozin (Figures 2F–J, P < 0.05). These results collectively indicate that high-dose DLTM and Dapagliflozin effectively inhibit VitD3-induced VC in diabetic rats, with DLTM offering vascular protection independent of glucose lowering.

VSMCs are the primary sites of VC. Their transformation from a contractile to an osteogenic phenotype is a hallmark of DVC, characterized by upregulation of osteogenic proteins such as RUNX2 and downregulation of contractile proteins such α-SMA (Lee et al., 2020). In this study, the extent of VSMC transdifferentiation in the rat aorta was assessed. The results showed that, compared to the CON group, STZ combined with VitD₃ injection significantly increased the expressions of BMP2 and RUNX2 in the rat aorta while decreasing the expressions of contractile proteins α-SMA and SM22α (Figures 3A–E, P < 0.001, P < 0.05). In contrast, oral administration of high-dose DLTM or Dapagliflozin effectively reversed the upregulation of osteogenic markers RUNX2 and BMP2, while increasing α-SMA and SM22α levels in the aorta of DVC rats (Figures 3A–E, P < 0.001, P < 0.01, P < 0.05). In contrast, the effects of medium-dose and low-dose DLTM on these osteogenic and contraction markers were limited, with some markers showing no significant improvement (Figures 3A–E, P> 0.05). In summary, these findings suggest that DLTM can mitigate the osteogenic transformation of VSMCs in vivo, thereby preventing further progression of VC.

To explore the specific molecular mechanisms of DLTM against VC, transcriptome sequencing analysis was conducted to identify differentially expressed transcripts. Considering that high-dose DLTM exhibited an optimal therapeutic effect comparable to the positive drug Dapagliflozin, transcriptome sequencing was performed on aorta samples from the DLTM-H group. Post DLTM treatment, significant changes occurred in the transcriptome profile of rat aorta, revealing a total of 813 DEGs. Among them, 494 DEGs were identified between the DVC and CON groups, including 137 upregulated and 357 downregulated genes (Figure 4A). Alternatively, a total of 319 DEGs were found between the DLTM and DVC groups, featuring 175 upregulated and 144 downregulated genes (Figure 4B). Notably, 56 DEGs overlapped between the two groups (Figure 4C). Further enrichment analysis of the overlapping DEGs unveiled that DLTM-affected DEGs were predominantly enriched in negative regulation of inflammatory response, regulation of inflammatory response, and bone morphogenesis (Figure 4D). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis highlighted close associations with pathways such as glycolipid metabolism (Figure 4E). To prevent the enrichment results from being biased by considerable upregulation or downregulation of genes and potentially missing out on DEGs of significant biological importance, gene set enrichment analysis (GSEA) was performed on the entire transcriptome (Figure 4F). The results revealed a significant upregulation of genes related to vascular smooth muscle contraction in the DLTM group, consistent with phenotypic observations. This finding suggests that DLTM may promote vascular smooth muscle contraction by upregulating VSMC contractile proteins and restoring the contractile phenotype. Notably, significant enrichment of the NOTCH signaling pathway, a calcium-related pathway, was observed among different groups, which was notably downregulated following DLTM treatment. NOTCH signaling is a highly conserved pathway crucial for immune cell differentiation and homeostasis (Gallenstein et al., 2023). Previous research indicates that activation of the NOTCH signaling pathway can stimulate pro-inflammatory macrophages, thereby initiating subsequent inflammatory responses and facilitating the osteogenic differentiation and calcification of VSMCs (Kang et al., 2024). Furthermore, among the calcium-related hub differentially expressed genes (hubDEGs), CCN3 garnered attention (Figure 4G), with PCR results confirming it as the most significantly affected calcium-related hubDEG by DLTM (Figure 4H). CCN3/NOV (nephroblastoma overexpressed), part of the cellular communication network (CCN) protein family, serves as an essential regulator in bone remodeling, with high expression observed in endothelial cells, smooth muscle cells, fibroblasts, and chondrocytes (Ouellet and Siegel, 2012; Peng et al., 2021). Previous research has demonstrated that CCN3 can play an important regulatory role across various cell types by modulating the NOTCH signaling pathway, including inhibiting cellular osteogenic differentiation (Minamizato et al., 2007; Ouellet and Siegel, 2012; Luan et al., 2023). Combining these findings with the results of Gene Ontology (GO) enrichment analysis, it is speculated that DLTM may suppress the progression of DVC by regulating the CCN3/NOTCH signaling axis to modulate inflammation.

The above results suggested that the mechanism by which DLTM treated DVC was related to the regulation of the CCN3/NOTCH signaling pathway. However, the interactions between the metabolites of DLTM and the molecules of the CCN3/NOTCH signaling pathway remained unclear. Therefore, molecular docking was performed using AutoDock to investigate the interactions between the potentially effective metabolites of DLTM and CCN3 as well as the classic NOTCH signaling molecules. The metabolites of DLTM were screened using Lipinski’s rule of five and SwissADME, resulting in potentially effective 59 metabolites for molecular docking (Supplementary Tables S4, S5). It is generally believed that when the binding energy is less than zero, the metabolite can spontaneously bind to the protein, with lower binding energy indicating a higher likelihood of interaction. The results showed that the average binding energy of the metabolites of DLTM with CCN3 and NOTCH signaling pathway proteins was −6.06 kcal/mol, with 69.30% of the molecules having binding energies < −5.25 kcal/mol and 26.93% having binding energies < −7 kcal/mol. The average binding energies for CCN3, NOTCH1, DLL4, DLL3, NOTCH3, DLL1, hes1, and hey1 were consistently below −5.25 kcal/mol (Figure 5A), suggesting superior binding activity with most DLTM metabolites (detailed in Supplementary Table S6). The 3D structure results further indicated that several main metabolites of DLTM, including worenine and tormentic acid, could well occupy the binding sites of CCN3 and classic NOTCH signaling molecules such as DLL1, DLL4, NOTCH1, hes1, and hey1 (Figures 5B–G). Conversely, the NOTCH ligand JAGGED1 exhibited lower binding energies with most DLTM metabolites, and NOTCH3 showed lower average binding energy than NOTCH1. Hence, it is hypothesized that CCN3, DLL1, DLL4, NOTCH1, and downstream transcription factors hes1 and hey1 may serve as key targets for DLTM in combating DVC.

To further evaluate the stability of the molecular docking results, we selected Worenine, the metabolite with the lowest average binding energy, and performed molecular dynamics simulations with the top two target proteins CCN3 and hes1, ranked by binding energy. The RMSD curve of the CCN3-Worenine complex remained relatively stable throughout the 0–100 ns simulation, fluctuating around 3.3 Å, and reached equilibrium after 80 ns (Supplementary Figure S2A). During the simulation, the Rg and SASA of the CCN3-Worenine complex remained relatively stable, indicating that the complex maintained a structurally stable and compact form (Supplementary Figures S2B, C). In contrast, the Rg of the hes1-Worenine complex exhibited noticeable fluctuations during the first 60 ns, and its SASA showed slight variations between 40 and 60 ns, suggesting that the binding of Worenine induced minor conformational changes in the protein, though these changes were within an acceptable range (Supplementary Figures S2B, C). Additionally, both the CCN3-Worenine and hes1-Worenine complexes maintained at least one hydrogen bond throughout the simulation, indicating stable binding between Worenine and the two target proteins (Supplementary Figure S2D). RMSF analysis suggested that the CCN3-Worenine complex exhibited higher stability, whereas the Hes1-Worenine complex was more flexible (Supplementary Figures S2E, F). Subsequently, we conducted MM/PBSA calculations to ascertain the binding energies of the two complexes and to pinpoint the key amino acids that substantially contribute to the binding affinity (Supplementary Figure S2G). These findings confirm the stable interaction of the DLTM metabolite Worenine with the target proteins CCN3 and hes1, further validating the molecular docking results. Combining the results of transcriptome, we speculated that the CCN3/NOTCH signaling pathway is a key pathway through which DLTM exerts its anti-DVC effects.

To verify the effect of DLTM on the CCN3/NOTCH signaling axis in vivo, the expression levels of CCN3 and the NOTCH signaling proteins with high binding activity in the aortas of rats from different groups were examined. The results showed that, compared to the CON group, the DVC group exhibited substantially decreased expressions of CCN3 and NOTCH1, while the expressions of NOTCH signaling ligands DLL1, DLL4, and downstream transcription factors hey1 and hes1 were significantly elevated. DLTM intervention effectively upregulated the expressions of CCN3 and NOTCH1 and inhibited the activation of DLL1 and downstream transcription factors hey1 and hes1. Nevertheless, it exerted no significant effect on DLL4 (Figures 6A, B, P < 0.001, P < 0.01, P < 0.05). Furthermore, ELISA was used to detect the levels of inflammatory factors in serum. The results showed that, compared to the CON group, the DVC group had significantly increased serum levels of TNFα, IL1β, and IL6. Meanwhile, DLTM intervention reversed these trends, reducing the release of the inflammatory factors TNFα, IL1β, and IL6 (Figures 6C–E, P < 0.001). These results collectively indicate that DLTM can exert anti-inflammatory effects by activating CCN3, inhibiting the expressions of NOTCH signaling ligand DLL1 and downstream transcription factors hey1 and hes1, and suppressing the activation of the NOTCH signaling pathway.