The molecular weight of BDP-I was assessed using HPGPC. The analysis was conducted on an LC-10A HPLC system (Shimadzu, Tokyo, Japan) equipped with a BRT105-104-102 series column of 8 mm × 300 mm. In addition, a standard curve of dextran was drawn, with the retention time (x, min) of each standard on the horizontal axis and the logarithmic value of its molecular weight (y, Log MW) on the vertical axis. The molecular weight of the polysaccharide was then determined from the best fit of the regression equation.

After precisely weighing 2 mg of the dried sample and 200 mg of potassium bromide, the two substances were combined and applied to the surface of the diamond prism in the attenuated total reflection configuration, followed by compression. Tablets composed solely of potassium bromide powder served as the blank control. Subsequently, the samples were analyzed using a Fourier transform infrared spectrometer (FTIR 650, Tianjin Gangdong Co., Hebei, China). Data analysis was performed using SpectraGryph version 1.2 software (Oberstdorf, Germany).

Standard solutions of 16 monosaccharides were prepared using a concentration gradient of 0.1 to 50 mg/L. A sample of 5 mg was carefully deposited into an ampoule and combined with 10 mL of a 2 M trifluoroacetic acid solution. The mixture was then subjected to hydrolysis at a temperature of 120°C for 3 h and then transferred to a tube for drying under a stream of nitrogen. The resulting solution (100 μL) was combined with 900 μL of deionized water. After centrifugation at 12,000 rpm for 5 min, the supernatant was collected for monosaccharide analysis on a Dionex ICS-5000 system (Thermo Scientific Co., Waltham, MA, USA) equipped with a CarboPacTMPA-20 analytical column (3 mm × 150 mm).

Following the methylation, hydrolysis, and acetylation procedures, the GC–MS technique was employed to determine and compare the obtained results with the standard mass spectrum library. The polysaccharide sample was first weighed (50 mg) and dissolved in anhydrous DMSO before initiating the methylation reaction by adding the methylating reagent solution. For hydrolysis, TFA was added to the sample, followed by the addition of sodium borohydride for reduction. After cooling, acetic anhydride was added for acetylation, with the resulting mixture added to toluene. Vacuum concentration and drying were then performed, and this process was repeated 4–5 times to eliminate excess acetic anhydride. Finally, the acetylated product was dissolved in dichloromethane. All analyses were performed with the GC–MS instrument (GCMS-QP-2010 model, Shimadzu, Tokyo, Japan).

The polysaccharide sample was measured (50 mg) and dissolved in 0.5 mL of deuterated water (D₂O) with a purity of 99.9%. After thorough dissolution, freeze dry and repeat the exchange three times. Dissolve the processed sample in D₂O again, centrifuge and transfer the solution to a 5 mm NMR tube, then the resulting solution was then freeze-dried for spectral analysis with a 600 MHz spectrometer (Bruker Corp. Fallanden, Switzerland) equipped with a 5-mm probe. Both 1D (1H, 13C) and 2D (COSY, TOCSY, HSQC, HMBC) spectra were acquired, with the results subsequently analyzed using Mestre Nova 14 software.

A cell suspension (100 μL), at a density of 1 × 105 cells/mL, was seeded into each well of 96-well plates before a 24-h incubation to allow cell attachment. Different concentrations of H₂O₂ (0 ~ 500 μM) were added to the culture plates before incubation for 3, 6, 12, and 24 h, respectively, to induce oxidative damage. The culture medium was discarded, and 100 μL of CCK-8 solution (CCK-8 reagent: serum-free culture = 1:10) was added to each well. Six replicates were prepared for each experimental group to ensure the accuracy of the results. After adding the CCK-8 solution, the plates were incubated at a constant temperature for 0.5 h to allow the reagent to react with the cells. The cell survival rate was subsequently calculated as Equation 1:

Where A₁ represents the absorbance value of the experimental group, A₂ represents the absorbance value of the blank group, and A₃ represents the absorbance value of the normal control group (NC).

The 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) fluorescence probe method was employed to assess the levels of reactive oxygen species (ROS). In this experiment, the fluorescence intensity was quantified utilizing an enzyme marker analyzer with an excitation wavelength of 488 nm and an emission wavelength of 525 nm.

A cell suspension (100 μL), at a density of 1 × 10⁵ cells/mL, was added to each well of a 96-well plate and allowed to attach for 24 h. Different concentrations of BDP-I (0.0625, 0.125, 0.25, 0.50, 1, and 2 mg/mL) were then added, and after incubation for 24, 48, and 72 h, respectively, the CCK-8 method was used to detect cell survival rate to determine the optimal concentration range of BDP-I.

Based on the results of cell survival, the following groups could be identified: the NC, where RPMI-1640 medium containing 10% fetal bovine serum was used, the positive control group (PC) treated with 300 μmol/L of α-LA, the model group (MC) treated with 250 μmol/L of H₂O₂ as well as the BDP-I-treated groups which could be further subdivided into the BDP-I low-dose group (0.0625 mg/mL), BDP-I medium-dose group (0.125 mg/mL) and BDP-I high-dose group (0.25 mg/mL).

Following the successful establishment of the oxidative stress model, different concentrations of BDP-I (0.0625, 0.125, 0.25 mg/mL) were used for treatment to assess its effects. Each group had six replicates, and both the blank group and the MC were cultured under identical conditions as the BDP-I-treated groups. The cells were then incubated for 24 h after treatment, with the survival rate of RIN-m5F cells eventually assessed as described in section 2.5.1. In addition, the ROS level was determined, as mentioned in section 2.5.2.

In this experiment, the cells were grouped and treated as described in section 2.6.1. Following treatment, the cells were collected, rinsed with PBS, centrifuged, homogenized using an extraction solution, and eventually pulverized through electric grinding for 10 s per iteration, followed by a 10-s pause in between for eight cycles. Antioxidant activity was finally assessed according to the instructions provided in the SOD, CAT, and TBARS test kits, the TBARS values were then expressed as equivalent nanomoles of malondialdehyde (MDA).

The cells were incubated in 3-cm dishes as described in section 2.6.1. Following three washes with 0.9% saline solution, they were quenched in liquid nitrogen for 3 min, after which an ice-cold 80% methanol: water (pre-cooled) solution was added. The cells were finally scraped off with a cell scraper.

An appropriate amount of the sample was added to a methanol/acetonitrile/water solution (2:2:1, v/v). After ultrasounds at low temperatures for 30 min, the sample was then centrifuged and vacuum drying the resulting supernatant. The collected was re-dissolution and eventually injected for mass spectrometry analysis.

A liquid chromatography system with ultra-high performance (Agilent, USA) was used to separate the samples. An automatic injector was used to inject the sample at 4°C, along with QC samples added to the sample queue to monitor and assess the system’s stability and data reliability.

An AB Triple TOF 6600 mass spectrometer (Agilent, USA) was used to collect the first and second spectra of the sample, with the positive and negative ion modes of electrospray ionization (ESI) subsequently used for detection. Data-dependent acquisition mode (IDA) and peak intensity value screening mode were used for the second mass spectrometry.

To determine the purity and molecular weight of the purified homogeneous yellow thorn polysaccharide BDP-I, HPGPC was performed, with glucose standards of known molecular weights ranging from 5,000 Da to 3,693,000 Da also included for the experiment. As illustrated in Figure 1C, the elution curve exhibited a single absorption peak characterized by favorable peak symmetry, reflecting the BDP-I sample’s high purity. Based on the glucose standard curve, BDP-I’s peak molecular weight was subsequently established as 3.947 kDa, with its weight-average molecular weight and number-average molecular weight being 4.771 kDa and 3.264 kDa, respectively. Compared to the study by Yang et al. (24), a homogeneous acidic polysaccharide (RPP-2a) was isolated from Raspberry Pulp, with a low weight-average molecular weight (Mw) of 55.582 KDa, BDP-I could be classified as a low-molecular-weight polysaccharide.

As illustrated in Figure 1D, the infrared spectrum of BDP-I exhibited an absorption peak within the range of 3,600–3,200 cm⁻¹, indicative of the -OH stretching vibration, a characteristic feature of carbohydrates. Similarly, an absorption peak was observed at 3300–3500 cm⁻¹, corresponding to the O-H stretching vibration, another distinctive absorption feature of carbohydrates. A weaker absorption peak was detected near 2930.1 cm⁻¹, attributed to the C-H stretching vibration on the sugar chain, which is recognized as one of the characteristic peaks of polysaccharides. Furthermore, the absorption peaks at 2927.41 cm⁻¹ and 2927.90 cm⁻¹ indicated the presence of asymmetric bending vibrations of C-H bonds, also characteristic of polysaccharides. The stretching vibration of C-H was also visible as a distinct absorption peak at 2927 cm⁻¹, while a peak at 1740 cm⁻¹ indicated the stretching vibration of C=O, which reflected the presence of acetyl groups within the BDP-I structure. Maldonado demonstrated that high methoxylation (degree of methoxylation, DM > 70%) pectin standards exhibit signals at 1630 and 1736 cm⁻¹, whereas low esterification (DM < 30%) pectin standards only display a band at 1593 cm⁻¹. The BDP-I pectin exclusively presented the carboxylate ion signal near 1730 cm⁻¹, suggesting a high degree of methoxylation (25). The absorption peak at 1612 cm⁻¹ may serve as a characteristic peak of crystalline water. A weak absorption peaks were evident at 1540 cm⁻¹ in the IR spectrogram, indicating that the polysaccharides contained a few proteins component. In addition, a peak at 1421 cm⁻¹ could be attributed to the stretching vibrations of C-O. Furthermore, absorption peaks in the range of 1,200–1,000 cm⁻¹ indicated the presence of pyranose structures. In this region, the band −1 at 1011 cm⁻¹ corresponds to the vibrational characteristics of the α − 1-4-C-O-C bond of galacturonic acid, and its intensity is roportional to the degree of polymerization. Finally, absorption peaks at 1105 cm⁻¹ could be linked to O-H bending vibrations, while absorption at 1020 cm⁻¹ could be attributed to pyranose ring stretching vibrations (26).

As illustrated in Figure 1E, compared with the HPLC chromatogram of monosaccharide standards, the monosaccharide composition of the purified polysaccharide included arabinose (Ara), galactose (Gal), glucose (Glc), and galacturonic acid (GalA). The molar ratio of each monosaccharide was determined by comparing its retention time and response intensity with those of various standard samples, with the results indicating that BDP-I was made up of Ara: Gal: Glu: GalA = 108:99:89:704.

In the previous work of the research, the monosaccharide results showed that the main monosaccharide of BDP was mannose (4). Mannose, a significant monosaccharide, frequently assumes a pivotal role in the architecture of polysaccharides. Notably, mannose is integral to the composition of plant cell wall polysaccharides, particularly in the biosynthesis of specific legume starches and fungal polysaccharides (27). The study found no mannose in BDP-1, likely due to the selectivity of polysaccharide purification and the sensitivity of analytical methods. We suspect there are two main reasons for this: Firstly, mannose may have beewith other components, or it may not have been sufficiently enriched during selective elution. Secondly, widely employed analytical techniques, including high-performance liquid chromatography (HPLC) and gas chromatography (GC), often exhibit inadequate sensitivity for the detection of mannose. This limitation is particularly pronounced in complex polysaccharide mixtures, where the presence of other components can interfere with the mannose signal, leading to its potential non-detection.

Due to aldonic acid’s presence in BDP-I, aldonic acid reduction was performed before methylation. The monosaccharide composition of the resulting sample was checked to confirm the successful reduction of aldonic acid (28).

Following methylation, hydrolysis, and acetylation of BDP-I, the resulting products included partially methylated alditol acetate derivatives (PMAAs). The GC–MS total ion chromatogram of these PMAAs is depicted in Figure 1F. Identification of the sugar residues present in BDP-I was achieved by comparing the mass spectra of each PMAA chromatographic peak with the established PMAA spectra of sugar residues available in the online database of the Complex Carbohydrate Research Center. Quantitative analysis was subsequently conducted by assessing the peak areas, enabling the types and molar percentages of methylated sugar residues to be determined (Table 1).

The glycosidic linkage pattern of BDP-I was found to be complex, with 11 different glycosidic linkage forms. Specifically, arabino-furanose (Araf) exists in five forms: terminal sugar (1-Araf), the 1,5-Araf, the 1,3,5-Araf, the 1,2,5-Araf, and the 1,3-Arap. Similarly, Gal exists in five forms, namely the terminal sugar (1-Galp), the 1,4-GalAp, the 1,3-Galp, the 1,6-Galp, and the 1,3,6-Galp, while for Glc, only the 1,4,6-Glcp is present. Overall, GalA and Gal had the highest proportion, with 1,4-GalAp and 1,3,6-Galp accounting for 18.03 and 35.00% of the samples, respectively. Additionally, the amount of 1,5-linked Araf, 1,4-GalAp, and 1,3,6-Galp residues was 10.37, 18.03, and 35.00%, respectively, thereby suggesting that the main chain of BDP-I may be connected by 1,6-O or 1,4-O glycosidic linkages.

Galacturonic acid (GalA) is a crucial polyhydric acid predominantly found in plant pectin. Recent studies have demonstrated that GalA not only contributes significantly to the structural integrity of plant cell walls but also exhibits notable antioxidant properties and provides protection against cellular oxidative stress. Research indicates that polysaccharides abundant in GalA, such as acidic polysaccharides derived from various plant components, possess substantial antioxidant activity. These polysaccharides enhance cellular antioxidant capacity by augmenting total antioxidant capacity and the activities of superoxide dismutase and catalase, while concurrently reducing intracellular levels of reactive oxygen species and malondialdehyde (29). Furthermore, studies have revealed that GalA and its derivatives can stimulate the proliferation of human skin fibroblasts and offer a degree of protection against oxidative damage (30). Arabinose (Ara) is a significant plant-derived monosaccharide. Polysaccharides extracted from prickly pear leaves, which contain arabinose, demonstrate notable antioxidant and alpha-glucosidase inhibitory activities in vitro, suggesting their potential as natural hypoglycemic or antioxidant agents (31). The interaction of arabinose with arabinose A and L-ascorbic acid (vitamin C) results in a pronounced synergistic antioxidant effect. This combination not only enhances the survival rate and antioxidant enzyme activity of HEK293 cells exposed to hydrogen peroxide but also mitigates the release of lactate dehydrogenase (LDH) and the accumulation of lipid peroxidation products (32).

The structure and properties of glycosidic bonds have a significant impact on their biological activity. Two polysaccharides (DGS1 and DGS2) isolated from Dendrobium officinale contain glycosidic bonds of (1 → 5)—Alaf, (1 → 4)—Glcp, and (1 → 6)—Glcp. These polysaccharides activate the expression of TNF-α, IL-6, and iNOs genes, particularly DGS2, which significantly increases the levels of TNF-α, IL-6, and NO, demonstrating stronger antioxidant activity (33). Polysaccharides rich in 1,3,6-Galp residues exhibit various biological activities, such as a water-soluble neutral polysaccharide (CAPW-1) that was considered of 1,3-Galp and 1,3,6-Galp, could stimulate the proliferation of RAW264.7 cells and promote the secret of nitrogen oxide (NO), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) with no cytotoxicity (34).

To obtain the configuration and linkage information of the sugar residues, the 1D/2D NMR spectra of BDP-I were analyzed (Figure 2), and these included 1H NMR, 13C NMR, COSY, TOCSY, NOESY, HSQC, and HMBC spectra. Interestingly, some of the signals were consistent with those reported in studies, hence establishing them as β-D-Galp and α-L-Araf (24, 35).

The pancreatic is susceptible to damage caused by reactive oxygen species (ROS), which can lead to apoptosis in β-cells and the subsequent development of diabetes (41). Several studies have applied H₂O₂ to induce oxidative damage in cells, thereby triggering oxidative injury in β-cells (18, 42). As shown in Figures 3A–D, an increase in the concentration of H₂O₂ was associated with a decrease in cell viability, as previously reported by Luo et al.’s findings (18). Compared with the NC, incubating cells with H₂O₂ concentrations of 200–500 μmol/L for 3–6 h significantly reduced cell viability (*p < 0.05). Notably, treatment with 250 μmol/L H₂O₂ led to a marked reduction in cell viability (*p < 0.01), reaching the half-maximal inhibitory concentration (IC₅₀). Figure 3E further demonstrates that the induction of cells with H₂O₂ concentrations ranging from 10 to 500 μmol/L resulted in a progressive increase in the ROS levels over time, indicating the influence of H₂O₂ on cellular ROS levels. Specifically, the levels of ROS exhibited a rising trend after 3, 6, 12, and 24 h of H₂O₂ induction, with a significant elevation observed at a concentration of 100 μmol/L (*p < 0.05 vs NC). Taking into account cell viability, the induction of cells with 250 μmol/L H₂O₂ for 3 h was selected as the optimal condition for subsequent experiments.

Figure 3F depicts the impact of BDP-I on cell viability after a 24-h treatment. Cell viability gradually increased within the concentration range of 0.0625 to 0.125 mg/mL, followed by a decline beyond 0.125 mg/mL. Within the range of 0.0625 to 1 mg/mL, there were no statistically significant differences in cell viability between the BDP-I-treated cells at 24 and 72 h compared to the NC (Figures 3F,G). However, at a concentration of 2 mg/mL, cell viability was significantly decreased (*p < 0.05) compared to the NC. These findings confirm that the safe concentration of BDP-I for cells falls within the range of 0.0625 to 1 mg/mL.

Figure 3H shows a significant decrease in cell viability after treatment with 250 μmol/L of H₂O₂ for 3 h compared to the NC (*p < 0.05). However, when cells were treated with BDP-I at concentrations ranging from 0.0625 to 0.25 mg/mL, cell viability was significantly increased compared to the MC (*p < 0.05). Consequently, it was concluded that BDP-I within this concentration range could enhance cell viability that had been compromised by H₂O₂, thereby exerting a protective effect against H₂O₂-induced cellular damage.

Figure 3I shows that, following treatment with 250 μmol/L of H₂O₂, the MC had significantly higher intracellular ROS levels than the NC (**p < 0.01). In contrast, after 24 h of treatment with 0.0625 to 0.25 mg/mL of BDP-I, there was a significant decrease in the ROS levels of cells compared to the MC (*p < 0.05). These results suggested that H₂O₂ induction increased ROS levels in RIN-m5F cells, while BDP-I could reduce these levels after H₂O₂-induced damage, thereby inhibiting oxidative stress caused by H₂O₂ (18).

The results for SOD and CAT activities as well as the TBARS content in RIN-m5F cells treated with BDP-I, are presented in Figures 4A–C. Following exposure to H₂O₂, there was a significant decrease in the activities of SOD and CAT compared to the NC (*p < 0.05). Subsequent treatment with BDP-I resulted in a gradual increase in the activities of these antioxidant enzymes in a dose-dependent relationship. Specifically, at a concentration of 0.25 mg/mL, the SOD activity reached (19.29 ± 2.09) U/mg, while the CAT activity reached (17.45 ± 1.70) U/mg.

Thiobarbituric acid reactive substances (TBARS) assay is extensively utilized to examine the impact of oxidative stress on health and the protective effects of antioxidants. In this study, the influence of BDP-I on lipid peroxidation was assessed by quantifying the lipid peroxidation product, TBARS, as depicted in Figure 4C. The TBARS level in untreated cells was measured at 0.123 ± 0.01 nmol malondialdehyde (MDA), whereas exposure to hydrogen peroxide (H₂O₂) significantly elevated TBARS levels to 0.617 ± 0.02 nmol MDA. Conversely, treatment with BDP-I at concentrations of 0.0625, 0.125, and 0.25 mg/mL markedly reduced the elevated TBARS levels induced by H₂O₂ (p < 0.05), with measurements recorded at 0.456, 0.395, and 0.279 nmol MDA, respectively.

The total ion chromatogram (TIC) of QC samples were compared and analyzed. In order to examine the overall metabolic changes in the sample profiles, separate PCA analyses were performed for the MC, NC, and BDP-I-treated group using SIMCA-P software (Supplementary Table S2). The ellipses on the graph subsequently represented the 95% confidence interval. The distinct separation observed between the NC and MC indicates the successful establishment of the oxidative stress cell model (R²X = 0.658), with significant alterations in cellular metabolites. The distinction between the polysaccharide group and the MC was less pronounced, whereas other groups displayed notable differences but were not entirely clustered. A supplementary PLS-DA analysis was performed to delve deeper into the disparities and patterns in cellular metabolites between the various groups. Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) is an adapted analytical approach for PLS-DA.

This study employed the variable importance in the projection (VIP) values derived from the OPLS-DA model to identify potential biomarkers. Table 3 shows significant differences between the MC and NC groups in positive and negative ion modes, indicating cell metabolic disturbances after oxidative damage. The sample distribution of the BDP-I-treated group approaches that of the NC, suggesting a potential protective effect of BDP-I on oxidatively damaged cells.

A comprehensive analysis of metabolites was conducted, identifying 947 metabolites in both positive and negative ion modes. Specifically, 501 metabolites were identified in positive ion mode, while 446 metabolites were identified in the negative one. These metabolites were subsequently categorized and quantified based on information on their chemical classification. The distribution of metabolites within each class is visually represented in Figure 5A. Furthermore, the OPLS-DA model analysis identified significant differential metabolites, with VIP > 1 and p < 0.05 selected at thresholds (Supplementary Figure S1). In addition, the results were compared with the results from the HMDB and KEGG databases. In the negative ion mode, 169 differential metabolites were identified. Of these, lipids and lipid-like molecules accounted for the largest proportion at 32.946%. Organic acids and derivatives followed at 18.585%, while heterocyclic compounds comprised 7.92% of the identified metabolites. Additionally, nucleosides, nucleotides, and similar compounds constituted 7.497%, benzene compounds accounted for 6.969%, and organic oxygen compounds represented 6.547%. These metabolites were primarily concentrated in glycerophospholipids, nucleotides, carboxylic acids and derivatives, fatty acyls, and organic oxygen compounds.

Following the completion of univariate analysis, a differential analysis was conducted on all identified metabolites (including unidentified ones), encompassing both positive and negative ion modes. A further screening process was then undertaken to identify differential metabolites that met the criteria of a fold change (FC) of >1.5 or <0.67 and a p-value <0.05. Using a volcano plot allowed the expression patterns of differential metabolites in the negative ion mode to be visually presented (Figures 5B–D). Within this plot, each point corresponded to a specific metabolite, with the size of the point indicating the Variable Importance in Projection (VIP) value. Upregulated metabolites are depicted in red, while downregulated ones are resented in blue. However, metabolites that did not reach statistical significance are displayed in black. Compared to the NC, the MC exhibited an increase in 17 metabolites and a decrease in 44 metabolites. Similarly, the BDP-I group displayed 7 upregulated metabolites and 15 downregulated ones compared to the MC group (Figure 5E).

The results presented in Table 3 illustrate potential biomarkers and trends after BDP-I treatment in cells displaying oxidative damage. Notably, BDP-I intervention had a significant impact on metabolic pathways. As indicated in Table 3, a total of 46 metabolites were found to be associated with T2DM, including 9 carboxylic acids and derivatives, 8 organic oxides, 10 pyrimidine nucleotides, 3 fatty acyls as well as 2 keto acids and derivatives: Gdp-l-fucose, 3 pyrazines, 1 imidazole compound (xanthine), and other compound groups such as amino acid-lysine, adenosine triphosphate, isopentyl pyrophosphate, DL-tryptophan, and β-nicotinamide adenine dinucleotide. After the implementation of the BDP-I intervention, the concentrations of phenylalanine, L-glutamic acid, pyruvic acid, α-ketoisovaleric acid, methylhydroxyvalerate, isoleucine-valine, 5′-deoxythymidine monophosphate (dTMP) and guanosine-5′-monophosphate were reversed, suggesting their potential as biomarkers for BDP-I intervention in pancreatic cells affected by oxidative damage.

Cluster analysis can provide a comprehensive visualization of the associations between samples while highlighting the distinct expression patterns of metabolites across different samples. Hierarchical cluster analysis was performed on samples from each group, and the resulting cluster tree is shown in Figure 5F, where samples clustered in the same branch display higher similarity. The NC was clearly segregated from the other two groups in this case. At the same time, the NC samples demonstrated cohesive clustering, indicating a high degree of similarity between them. This observation suggested the effective establishment of the cellular oxidative stress model at the metabolomic level. Figure 5G presents a cluster heatmap, revealing that pyruvic acid was significantly upregulated in the MC group but downregulated in the BDP-I group. Conversely, α-ketoisovaleric acid was downregulated in the NC and BDP-I-treated groups while upregulated in the MC.

Figure 5H illustrates the 15 pathways found to be associated with T2DM (p < 0.05), namely the central carbon metabolism in cancer, insulin resistance, alanine, aspartate and glutamate metabolism, insulin secretion, tricarboxylic acid (TCA) cycle, cGMP-PKG signaling pathway, taurine, and hypotaurine metabolism, pantothenate and CoA biosynthesis, insulin signaling pathway, rust disease, AMPK signaling pathway, oxidative phosphorylation, amino sugar and nucleotide sugar metabolism and carbon metabolism. These results suggest that by modulating these pathways, BDP-I primarily affects the oxidative damage to pancreatic β-cells.

Pancreatic β-cells are essential in maintaining equilibrium in Glc metabolism by facilitating insulin secretion (43). However, oxidative stress triggers the apoptosis of β-cells, leading to diminished insulin secretion. Research has demonstrated that in a T2DM animal model, antioxidants could potentially alleviate hyperglycemia, thereby regulating the typical expression of the insulin promoter PDX-1 and transcription factor MafA (46). Consequently, exploring new diabetes treatment and pathogenesis through the lens of oxidative stress holds promise for identifying novel active constituents or therapeutic strategies for efficacious disease management.

Based on previous investigations on cellular phenotypes, it was observed that BDP-I exhibited a good antioxidant effect by enhancing the viability of cells damaged by oxidative stress and augmenting the performance of antioxidant enzymes. Metabolomic analyses provided additional evidence that BDP-I can counteract the metabolic disturbances induced by oxidative damage by influencing 46 potential biomarkers. Glc is the primary energy source of glycolysis and the TCA cycle, universally present metabolic pathways in aerobic organisms. Succinic acid plays a key role in sugar metabolism, promoting ATP generation, exhibiting antioxidant stress effects, increasing tissue levels of GSH and SOD, reducing MDA content, inhibiting the expression of inflammatory factors and protecting cells from excessive apoptosis (47). In the present study, a reduction in succinic acid content was observed in the MC. However, following BDP-I intervention, a significant increase in succinic acid levels was observed. Given the involvement of succinic acid in multiple metabolic pathways, including the TCA cycle, these findings suggest that BDP-I may exert its therapeutic effects on oxidative stress by intervening in the metabolism of oxidatively damaged cells through diverse pathways and targets.

Additionally, alanine, which involved in various metabolic processes and is a crucial energy source, plays a significant role in amino acid metabolism. Research has demonstrated that in diabetic patients, plasma alanine levels tend to be elevated (48). However, this which can effectively be reduced by BDP-I, improving amino acid metabolism disorders. The MC exhibited notable reductions in the levels of amino acids such as serine and alanine, while the levels of phenylalanine and L-glutamine displayed the opposite effects. Phenylalanine, an aromatic amino acid, can be a predictive factor for diabetes risk and is positively associated with the risk of T2DM (49). L-glutamine, a non-essential amino acid abundantly present in the body, acts as a substrate for bodily metabolism and can enhance insulin secretion (50). Therefore, BDP-I may improve T2DM by regulating phenylalanine and L-glutamic acid levels to stimulate insulin secretion.

The metabolism of nucleotides in cells experiencing oxidative damage display alterations, especially with regards to a notable reduction in the concentrations of the four pyrimidine nucleotides (hydroxyethyl glucoside, Udp-N-acetylglucosamine, Udp-galactose, and cytidine-N-acetylneuraminic acid) and one purine nucleotide (Gdp-l-fucose). Purine metabolism is vital in preserving the equilibrium of mitochondrial oxidative stress within the organism, as it plays a pivotal function in providing energy, metabolic control, and coenzyme composition (51). The high levels of pyrimidine nucleotides in the MC led to oxidative stress in renal cells. The findings mentioned above suggest a strong correlation between the antioxidant mechanism of BDP-I and various metabolic processes, including alanine, aspartate, and glutamate metabolism, insulin secretion, the TCA cycle, the cGMP-PKG signaling pathway, taurine and hypotaurine metabolism, pantothenate and coenzyme A biosynthesis, the glucagon signaling pathway, rust disease, the AMPK signaling pathway, oxidative phosphorylation, amino sugar and nucleotide sugar metabolism, and disruption of carbon metabolism.

To investigate the association between metabolites and cellular oxidative stress, Spearman correlation analysis was used to evaluate the relationship between crucial differential metabolites and indicators related to oxidative stress (Figure 5I). The heatmap illustrates positive correlations in red and negative correlations in blue, with the intensity of color indicating the magnitude of the correlation. Specifically, isoleucine-proline and pyruvic acid exhibited a positive correlation with reactive oxygen species (ROS) and malondialdehyde (MDA) while displaying a negative correlation with superoxide dismutase (SOD) and catalase (CAT). These findings suggest that an elevation in isoleucine-proline and pyruvic acid levels may potentially enhance cellular oxidative stress. In addition, methylhydroxybutyrate, alpha-D-mannose-1-phosphate, isocitric acid isobutyl ester, Gdp-l-fucose showed a positive correlation with SOD but a negative correlation with ROS, indicating that these metabolites could have been associated with enhanced intracellular antioxidant enzyme activity and inhibited oxidative stress.