Sixty BALB/c mice were housed at 25 $\pm$ 3 °C and 50–70% humidity under a light–dark cycle with free access to food and water. After three days, they were randomly assigned to the control group (C, n = 20), the diabetes model group (T0, n = 20), and the TP (Batch No. 040508, Wuhan Ya Fa Biotechnology Company, Wuhan, Hubei, China)-treated group (T1, n = 20). Except for group C, all mice received a high-fat diet (mouse feed mixed with lard) and high-sugar drinking water (50% glucose solution) for two weeks. They were then intraperitoneally injected with Streptozotocin (STZ, Product Number: S0130-100. Sigma-Aldrich, Burlington, MA, USA)–citric acid buffer (45 mg/kg) for two consecutive days. One week later, fasting blood glucose (FBG) was measured.

The criteria for successful establishment of a diabetic mouse model are as follows: (1) fasting blood glucose (FBG) concentration exceeding 11.1 mmol/L; and (2) manifestation of classic diabetic symptoms—including polydipsia, polyphagia, and polyuria. These criteria are uniformly evaluated on the 7th day after the intervention of high-sugar and high-fat diet feeding combined with STZ intraperitoneal injection. Prior to assessment, mice are fasted for 12–16 hours with unrestricted access to water. Model establishment is deemed successful if both criteria are concurrently fulfilled.

After successful modelling, all mice were provided normal feed and water. Mice in group T1 received TP solution by gavage (100 mg/kg/d, once daily), whereas groups C and T0 received the same dose of saline. Administration continued for 21 days. FBG was measured every 3–4 days. All animal procedures were approved by the Scientific Ethics Committee of Henan Institute of Science (License No. LLSC2021034).

The initial body weight of mice was recorded before modelling. Body weight in each group was measured on days 1, 7, 14 and 21, and average values were calculated to document weight changes.

Water intake on the first day of the experiment was recorded. Daily water intake was then measured at a fixed time. On days 7, 14 and 21, the average weekly water intake per mouse was recorded. The calculation formula was:

Average water intake (g/d) = [Total water intake in a week (g) – Remaining water weight on the recorded day (g)] / {[(Time) (d)] × [Number of mice per cage (only)]}.

During the final week of the experiment, bedding wetness was recorded after one day under identical bedding conditions and the same number of mice per cage.

On days 1, 7, 14 and 21, five mice were randomly selected from groups C, T0 and T1, batch execution without repeated measurement. Mice were euthanized with 1% pentobarbital sodium (Cat. No: P3761, Sigma-Aldrich, St. Louis, MO, USA) (40 mg/kg), and hippocampal CA1 region tissues were collected. Half of each sample was fixed in 4% paraformaldehyde (PFA) (Cat. No: MM1504, Shanghai Maokang Biotechnology Co., Ltd., Shanghai, China) for paraffin embedding and H&E, toluidine blue (Cat. No: 2883914, Shanghai Macklin Biochemical Co., Ltd., Shanghai, China), and Hoechst 33342 (Cat. No: AR0039, Wuhan Boster Biological Technology Co., Ltd., Wuhan, Hubei, China) staining. The remaining tissue was stored at –80 °C for PCR and ELISA analysis.

Hippocampal tissue fixed in 4% PFA for 48 h was rinsed in running water, dehydrated through graded ethanol, cleared in xylene, infiltrated with wax, embedded, and sectioned at 5 µm thickness. Sections were stained with H&E, toluidine blue, and Hoechst 33342 (Boster), and observed under a 400$\times$ optical microscope (ECLIPSE Ci-L, Nikon, Tokyo, Japan).

For each group, the number of neurons with nuclear swelling per 100 neurons in the CA1 region was counted. The neuronal swelling index (%) was calculated as: (number of swollen neurons/total number of neurons) $\times$ 100.

In the CA1 region, the number of congested vessels per 30 vessels examined was recorded. The congestive vascular index (%) was calculated as: (number of congestive vessels/total number of vessels) $\times$ 100.

Images of CA1 neurons were captured at 1000$\times$ magnification. Images were analysed in Image-Pro Plus7.0 (Media Cybernetics, Inc., Rockville, MD, USA). Regions from comparable locations were selected, and grey values were measured. Higher grey values indicated lower Nissl substance content.

The number of apoptotic cells per 100 cells in the CA1 region was counted, including ECs, neurons, glial cells, and other cell types. The apoptosis index (%) was calculated as: (number of apoptotic cells / total number of cells) $\times$ 100.

Total RNA was extracted from hippocampal tissues of groups C, T0 and T1 on Days 1, 7, 14 and 21, following the manufacturer’s instructions. RNA purity and concentration were measured, and 1 µg of total RNA was reverse transcribed into 20 µL cDNA. Real-time PCR was performed, and the mRNA expression levels of ZO-1, JAM-3, ZO-3, claudin-5, claudin-3, occludin, TNF-$\alpha$, IL-6, IL-1$\beta$ and Glut1 were quantified. Relative expression levels were calculated using the 2^{-Δ⁢Δ⁢Ct} method. Primer sequences are shown in Table 1.

Cryopreserved hippocampal tissue was homogenised in pre-cooled PBS and centrifuged at 1000 $\times$g for 10 min at 4 °C. The supernatant was collected. Phosphorylated P38 protein was measured using a mouse phosphorylated P38 ELISA kit (ZK-5246, Shanghai ZhenKe Biological Technology Co., Ltd., Shanghai, China). Phosphorylated P65 was measured using a phosphorylated P65 ELISA kit (EY-01M8419, Shanghai Yiyan Biotechnology Co., Ltd., Shanghai, China). MDA levels were assessed using a mouse MDA assay kit (ZK-4654, Shanghai ZhenKe Biological Technology Co., Ltd.). ROS was measured using a reactive oxygen species (ROS) ELISA kit (EY-(Ela)-0372, Shanghai Yiyan Biotechnology Co., Ltd.). AGEs and RAGE protein levels were measured using mouse AGEs (ZK-4500) and RAGE (ZK-5226) ELISA kits (Shanghai ZhenKe Biological Technology Co., Ltd. Shanghai, China). P-gp was measured using a mouse P-gp ELISA kit (EY-01M8338, Shanghai Yiyan Biotechnology Co., Ltd.). Absorbance was read at 450 nm using a microplate reader (Multiskan FC, Thermo Fisher Scientific, Waltham, MA, USA), and protein concentrations were calculated according to the manufacturer’s instructions.

Data were analysed using Prism 8.0 (GraphPad Software, Boston, MA, USA) and SPSS 23.0 (IBM Corp., Chicago, IL, USA). Results are expressed as mean $\pm$ SEM. The t-test was used for comparisons between two groups. Pearson correlation analysis (SPSS) assessed associations between changes in RAGE expression and corresponding indicators. A p value 0.05 was considered statistically significant.

Mice in group C remained active, maintained normal diet and water intake, and showed no clinical abnormalities.

Prior to model induction, no significant difference in mean body weight was observed among the three groups of mice (p $>$ 0.05). Model induction—achieved through high-sugar, high-fat diet feeding followed by intraperitoneal injection of STZ—was confirmed on day 0 of the experiment. On Day 1, mean body weight in the T0 and T1 groups was significantly higher than that in the control (C) group (p 0.05). Thereafter, mean body weight in the T0 and T1 groups gradually declined over time; by days 7 and 14, the reductions were evident but did not reach statistical significance relative to the C group (p $>$ 0.05). By Day 21, mean body weight in the T1 group remained comparable to that in the C group (p $>$ 0.05), whereas the T0 group exhibited a statistically significant reduction compared with both the C and T1 groups (p 0.05) (Table 2).

Water intake in the T0 and T1 groups was significantly elevated relative to the control (C) group on Days 1, 7, 14, and 21 of the experiment (p 0.05 or p 0.01). Moreover, water intake in the T1 group was significantly lower than that in the T0 group on Days 7, 14, and 21 (p 0.05) (Fig. 1).

Fig. 1.

Comparison of water intake among group C mice, group T0 mice, and group T1 mice (n = 20, the initial total sample size of each group). Note: Compared with group C, * indicates p 0.05, ** indicates p 0.01. Compared with group T0, # indicates p 0.05.

In addition, as the diabetic model progressed, mice in the T0 group exhibited a marked increase in urine output (Fig. 2) and progressive deterioration of coat condition, characterized by unkempt and damp fur (Fig. 3). In contrast, relative to the T0 group, the T1 group showed significantly reduced urine output (Fig. 2), restoration of fur luster and smoothness, and marked improvement in both physical condition and mental alertness (Fig. 3).

Fig. 2.

The effect of TP on bedding of diabetic mice. Note: Compared with group C, the wet area of bedding in group T0 expanded. Compared with group T0, the wet area of bedding in group T1 decreased. The red box indicates the wet area of the bedding.

Fig. 3.

Comparison of appearance among group C mice, group T0 mice, and group T1 mice at 21 d.

At 1, 4, 7, 10, 14, 17 and 21 days, the average blood glucose level of group T0 remained significantly higher than that of group C (p 0.01), and blood glucose in all mice in group T0 exceeded 11.1 mmol/L. With continued TP intervention, blood glucose in group T1 showed a gradual decline compared with group T0. At 10, 14, 17 and 21 days, blood glucose in group T1 was significantly lower than in group T0 (p 0.05) (Table 3).

Throughout the experiment, AGEs protein expression in groups T0 and T1 increased significantly compared with group C (p 0.05). At 7, 14 and 21 days, AGEs levels in group T1 were significantly lower than those in group T0 (p 0.05) (Fig. 4a).

Fig. 4.

The effect of TP on the expression of AGEs, RAGE, P38, and NF-$\kappa$B proteins in the hippocampal tissue of diabetic mice (n = 5). “a” represents the change in AGEs protein expression, “b” represents the change in RAGE protein expression, “c” represents the change in P38 protein expression, and “d” represents the change in NF-$\kappa$B protein expression. Note: Compared with group C, * indicates p 0.05, and ** indicates p 0.01. Compared with group T0, # indicates p 0.05. AGEs, advanced glycation end products; RAGE, receptor for advanced glycation end products.

RAGE protein expression in groups T0 and T1 increased significantly or highly significantly at 1, 7, 14 and 21 days compared with group C (p 0.05 or p 0.01). At 7, 14 and 21 days, RAGE expression was significantly lower in group T1 than in group T0 (p 0.05) (Fig. 4b).

P38 protein expression increased significantly in group T0 at 7, 14 and 21 days (p 0.05), and in group T1 at 7 and 14 days (p 0.05). At all three time points, P38 expression was significantly lower in group T1 than in group T0 (p 0.05) (Fig. 4c).

Activation of NF-$\kappa$B increased significantly in group T0 at 7, 14 and 21 days (p 0.05), and in group T1 at 14 and 21 days (p 0.05). At 7, 14 and 21 days, phosphorylation of NF-$\kappa$B was significantly lower in group T1 than in group T0 (p 0.05) (Fig. 4d).

At day 1, TNF-$\alpha$ mRNA expression did not differ significantly among groups (p $>$ 0.05). At 7, 14 and 21 days, TNF-$\alpha$ mRNA expression increased significantly in both T0 and T1 groups (p 0.05). At 14 and 21 days, expression in group T1 was significantly lower than in group T0 (p 0.05) (Fig. 5a).

Fig. 5.

The effect of TP on the expression changes of key genes in inflammation and oxidative stress in the hippocampal tissue of diabetic mice (n = 5). “a” represents the expression level of TNF-$\alpha$ mRNA, “b” represents the expression level of IL-6 mRNA, “c” represents the expression level of IL-1$\beta$ mRNA, “d” represents the change in ROS content, and “e” represents the change in MDA content. Note: Compared with group C, * indicates p 0.05. Compared with group T0, # indicates p 0.05.

Similarly, IL-6 mRNA expression showed no significant differences at day 1 (p $>$ 0.05). At 7, 14 and 21 days, IL-6 expression increased significantly in group T0 (p 0.05), and at 14 and 21 days increased significantly in group T1 (p 0.05). At 21 days, IL-6 expression in group T1 was significantly lower than in group T0 (p 0.05) (Fig. 5b).

For IL-1$\beta$ mRNA, no significant differences were observed at day 1. At 7, 14 and 21 days, IL-1$\beta$ expression increased significantly in group T0 (p 0.05). At 14 and 21 days, IL-1$\beta$ expression also increased significantly in group T1 (p 0.05). Moreover, at 14 and 21 days, expression in group T1 was significantly lower than in group T0 (p 0.05) (Fig. 5c).

ROS levels were significantly elevated in groups T0 and T1 compared with group C across all time points (p 0.05). At 7, 14 and 21 days, ROS levels in group T1 were significantly lower than in group T0 (p 0.05) (Fig. 5d).

At day 1, MDA levels did not differ significantly among groups. MDA levels increased significantly in group T0 at 7, 14 and 21 days (p 0.05), and in group T1 at 14 and 21 days (p 0.05). At 14 and 21 days, MDA levels were significantly lower in group T1 than in group T0 (p 0.05) (Fig. 5e).

P-gp protein expression in group T0 showed no significant differences from group C at days 1 and 21 (p $>$ 0.05), but increased significantly at 7 and 14 days (p 0.05). In group T1, P-gp levels increased significantly at 7, 14 and 21 days (p 0.05). At 14 and 21 days, P-gp expression was significantly higher in group T1 than in group T0 (p 0.05) (Fig. 6a).

Fig. 6.

The effect of TP on P-gp protein expression, and Glut1 mRNA expression in hippocampal tissue of diabetic mice (n = 5). “a” represents the change in P-gp protein expression, and “b” represents the expression level of Glut1 mRNA. Note: Compared with group C, * represents p 0.05. Compared with the T0 group, # represents p 0.05. P-gp, P-glycoprotein.

Glut1 mRNA expression showed no significant differences among groups at day 1 (p $>$ 0.05). Expression increased significantly in both T0 and T1 groups at 7, 14 and 21 days (p 0.05). At 14 and 21 days, Glut1 mRNA expression was significantly higher in group T1 than in group T0 (p 0.05) (Fig. 6b).

P38, NF-$\kappa$B, IL-6, TNF-$\alpha$, ROS and Glut1 are key downstream factors in the AGEs–RAGE signalling pathway. Correlation analysis showed that the expression patterns of these molecules were significantly or highly significantly positively correlated with RAGE protein levels (p 0.05 or p 0.01) (Table 4).

Compared with group C, Claudin-3 mRNA expression in group T0 increased significantly at 1 and 7 days (p 0.05), showed no significant difference at 14 days, and decreased significantly at 21 days (p 0.05). In group T1, Claudin-3 mRNA expression increased significantly at 1, 7 and 14 days (p 0.05), but did not differ significantly at 21 days. Compared with group T0, Claudin-3 expression in group T1 was significantly higher at 14 and 21 days (p 0.05) (Fig. 7a).

Fig. 7.

The effect of TP on the expression level of key molecules of TJ in BBB of diabetic mice (n = 5). “a” represents the expression level of Claudin-3 mRNA, “b” represents the expression level of Claudin-5 mRNA, “c” represents the expression level of ZO-1 mRNA, “d” represents the expression level of ZO-3 mRNA, “e” represents the expression level of JAM-3 mRNA, and “f” represents the expression level of Occludin mRNA. Note: Compared with group C, * indicates p 0.05. Compared with group T0, ** indicates p 0.01. Compared with group T0, # indicates p 0.05. TJ, tight junction; BBB, blood–brain barrier.

Claudin-5 mRNA expression in group T0 increased significantly at 1 day (p 0.05), showed no significant difference at 7 days, and decreased significantly at 14 and 21 days (p 0.05). In group T1, Claudin-5 mRNA expression increased significantly at 1 and 7 days (p 0.05), with no significant differences at 14 and 21 days. Compared with group T0, Claudin-5 expression in group T1 was significantly higher at 7, 14 and 21 days (p 0.05) (Fig. 7b).

ZO-1 mRNA expression in group T0 did not differ from group C at 1 day but decreased significantly or highly significantly at 7, 14 and 21 days (p 0.05 or p 0.01). Compared with group T0, ZO-1 expression in group T1 increased significantly at 7, 14 and 21 days (p 0.05) (Fig. 7c).

ZO-3 mRNA expression in group T0 increased significantly at 1 day (p 0.05), showed no significant difference at 7 days, and decreased significantly at 14 and 21 days (p 0.05). In group T1, ZO-3 expression increased significantly at 1 and 7 days (p 0.05). Compared with group T0, ZO-3 expression in group T1 was significantly higher at 14 and 21 days (p 0.05) (Fig. 7d).

JAM-3 mRNA expression in both T0 and T1 groups increased significantly at 1 day (p 0.05), showed no significant difference at 7 days. Moreover, JAM-3 mRNA expression in group T0 decreased significantly at 14 and 21 days (p 0.05). Compared with group T0, JAM-3 expression in group T1 increased significantly at 7, 14 and 21 days (p 0.05) (Fig. 7e).

Occludin mRNA expression in group T0 increased significantly at 1 day (p 0.05), showed no significant difference at 7 and 14 days, and decreased significantly at 21 days (p 0.05). In group T1, Occludin expression increased significantly at 1, 7 and 14 days (p 0.05). Compared with group T0, Occludin expression in group T1 was significantly higher at 7, 14 and 21 days (p 0.05) (Fig. 7f).

H&E staining showed pronounced histological changes in the hippocampus of group T0 compared with group C. Early lesions were observed from CA1 to CA4, with CA1 being most severely affected. At 7 and 14 days, capillary congestion and neuronal structural damage were evident in group T0, with swelling of pyramidal neuron nuclei and cytoplasm. At 21 days, neuronal arrangement became disordered, capillary congestion worsened, and the Virchow–Robin space widened. Some neurons showed lighter cytoplasmic staining and became rounded and swollen. In group T1, the Virchow–Robin space was narrower at 21 days, and neuronal swelling was significantly reduced compared with group T0 (Fig. 8A). As shown in Fig. 8B,C, the neuronal swelling index in group T0 increased significantly at 14 and 21 days (p 0.05), and that in group T1 increased significantly at 21 day (p 0.05). In group T1, the index was significantly lower than in group T0 at 21 day (p 0.05). In addition, the congestive vascular index in groups T0 and T1 increased significantly or highly significantly at 7, 14 and 21 days (p 0.05 or p 0.01). Moreover, compared with T0 group, TP treatment significantly reduced this index at 7, 14 and 21 days (p 0.05), indicating that TP reduced neuronal swelling and vascular congestion.

Fig. 8.

The effect of TP on the histological structure of the hippocampus in DM mice, n = 5. Note: (A) H&E staining were used to analyze the pathological changes of hippocampal tissue. “➞” indicates that the capillaries were congested and the spaces were enlarged. “➡” indicates the swelling of neurons. (B) Semi-quantitative analysis of neuronal swelling index. (C) Semi-quantitative analysis of vascular congestion index. Compared with group C, * indicates p 0.05, and ** indicates p 0.01. Compared with group T0, # indicates p 0.05. Scale bar = 25 µm. DM, diabetes mellitus.

Toluidine blue staining showed no obvious neuronal changes in the CA1 region at 1 and 7 days in group T0. At 14 and 21 days, neurons became rounded and swollen, and Nissl substance decreased significantly. Compared with group T0, Nissl substance in group T1 increased significantly at 14 and 21 days (Fig. 9A). As shown in Fig. 9B, grey values of Nissl substance were significantly higher in group T0 at 14 and 21 days (p 0.05). In group T1, grey values decreased significantly at 21 day compared with group T0 (p 0.05), indicating that TP reduced the loss of Nissl substance in hippocampal neurons.

Fig. 9.

The effect of TP on the content of Nissl substance in neurons of the CA1 region in the hippocampus of mice, n = 5. Note: (A) Toluidine blue staining. “➡” indicates neurons in which the Nissl substance in the cytoplasm has decreased or disappeared. (B) Semi-quantitative analysis of changes in Nissl substances in neurons. Compared with group C, * indicates p 0.05. Compared with group T0, # indicates p 0.05. Scale bar = 10 µm.

Hoechst 33342 staining showed no significant differences in apoptotic cells between groups at 1 and 7 days. At 14 and 21 days, apoptotic cells increased significantly in endothelial and glial cells in group T0. In group T1, apoptotic cells in the CA1 region were significantly reduced at both time points (Fig. 10A). As shown in Fig. 10B, the apoptotic index was significantly or highly significantly elevated in groups T0 and T1 at 14 and 21 days (p 0.05 or p 0.01). Moreover, compared with T0 group, TP treatment significantly lowered the apoptotic index at 14 and 21 days (p 0.05), indicating that TP reduced apoptosis in hippocampal tissue of diabetic mice.

Fig. 10.

The effect of TP on apoptosis of hippocampal cells in DM mice, n = 5. Note: (A) Hoechst 33342 staining was used to analyze the pathological changes of hippocampal tissue. “➡” indicates apoptotic cells. (B) Semi-quantitative analysis of apoptotic cells. Compared with group C, * indicates p 0.05, and ** indicates p 0.01. Compared with group T0, # indicates p 0.05. Scale bar = 25 µm.

Previous studies show that green tea significantly improves fasting blood glucose, glycosylated haemoglobin and insulin resistance in type II diabetes [21]. Green or black tea alleviates polyuria, polydipsia, hyperglycaemia and other diabetic symptoms in db/db mice by upregulating kidney water reabsorption proteins, including protein kinase C-$\alpha$ (PKC-$\alpha$), membrane PKC-$\alpha$ and glycosylated UT-A1 [22]. In addition, diabetic model rats exhibit markedly elevated FBG levels; administration of TP at 300 mg/kg significantly reduces FBG [23]. In this study, diabetic mice showed typical symptoms of weight loss, polyuria and polydipsia. After TP administration, these symptoms were substantially alleviated, and FBG was significantly lowered. These findings are consistent with earlier reports [21, 22, 23].

The interaction between AGEs and RAGE contributes to BBB dysfunction. Hyperglycaemia in DM elevates RAGE levels, increases ROS production, activates the NF-$\kappa$B pathway and promotes the expression of pro-inflammatory mediators, thereby enhancing inflammatory responses [12]. Studies also show that AGEs accumulate in the brains of diabetic rats and induce upregulation of RAGE, creating a cycle of inflammation and oxidative stress that intensifies vascular inflammation [24]. Combined treatment with insulin and idebenone inhibits the ROS/AGE/RAGE/NF-$\kappa$B signalling pathway, upregulate occludin, claudin-5 and ZO-1, and reduce BBB permeability in diabetic rats, indicating that elevated RAGE may contribute indirectly to BBB injury. Green tea extract reduces the inflammatory response in LPS-stimulated L2 lung epithelial cells during acute respiratory distress syndrome by suppressing NF-$\kappa$B, NOD‑like receptor family, pyrin domain containing 3 (NLRP3), TLR-4 and RAGE expression, as well as IL-12, TNF-$\alpha$ and CRP, while reducing apoptosis and increasing viable cell numbers [25]. EGCG decreases high mobility group box 1 protein (HMGB1) and RAGE expression, inhibits NF-$\kappa$B activation and reduces neointimal hyperplasia in a rat carotid balloon injury model [26]. In this study, AGEs and RAGE protein levels increased progressively in both diabetic groups, indicating their involvement in early BBB injury. TP administration significantly reduced both proteins. During early hyperglycaemia, changes in P38 and phosphorylated NF-$\kappa$B expression were not notable; however, both increased under prolonged hyperglycemic conditions. TP significantly reduced P38 and phosphorylated NF-$\kappa$B, likely through suppression of AGEs and RAGE and subsequent inhibition of downstream effectors [25, 26].

Oxidative stress is a physiological response induced by adverse internal or external stimuli. Excessive production of ROS and reactive nitrogen species (RNS) exceeds the body’s clearance capacity, disrupts redox homeostasis and leads to an imbalance between oxidative and antioxidative processes. Malondialdehyde (MDA), a final product of lipid peroxidation, is an indicator of lipid peroxidation severity and oxidative stress. Studies report that MDA levels in the hippocampus of type 2 diabetic mice are significantly higher than in normal mice, whereas SOD activity is significantly reduced and inflammatory cytokines such as IL-1$\beta$, IL-6, TNF-$\alpha$ and CRP are significantly elevated [11, 27]. In male Wistar rats at the prediabetes stage, regular intake of white tea increases lung antioxidant enzyme activity and total antioxidant capacity (TAC) and improves parameters related to protein nitration, lipid peroxidation and histone H2A [28]. In patients with type 2 diabetes, two months of EGCG supplementation (300 mg/d) significantly increases TAC [29]. EGCG also significantly reduces MDA levels in diabetic animal models [30]. In this study, ROS and MDA levels increased significantly in the hippocampal tissues of both diabetic groups throughout the experiment. TP treatment significantly reduced both indicators, suggesting that TP attenuates oxidative stress in diabetic brain tissue, consistent with previous reports [29, 30].

Chronic peripheral inflammation induced by persistent hyperglycaemia can activate inflammatory signalling within the CNS, disrupt neuronal structure and function in the hippocampus and contribute to cognitive impairment and behavioural abnormalities. IL-1$\beta$, IL-6 and TNF-$\alpha$ are common pro-inflammatory cytokines. Prior studies show that TNF-$\alpha$, IL-6, IL-1$\beta$ and MDA levels in the hippocampus of diabetic rats rise significantly compared with nondiabetic control rats (p 0.05) [31]. Other studies report that oxidative stress, inflammatory cytokines and the TXNIP/NLRP3/IL-1$\beta$ pathway are upregulated in podocytes exposed to high glucose and in diabetic nephropathy mouse kidneys, whereas EGCG reduces ROS and inflammatory factors such as TNF-$\alpha$, IL-6 and IL-18 [32]. In elderly T2DM model rats, TP (300 mg/kg) regulates intestinal microbiota and inhibits TLR4/NF-$\kappa$B signalling in the hippocampus, reducing neuroinflammation and improving synaptic plasticity [33]. Curcumin also protects hippocampal tissue in diabetic rats by reducing TNF-$\alpha$, IL-6, IL-1$\beta$ and MDA levels [31]. In this study, no significant differences in TNF-$\alpha$, IL-6 or IL-1$\beta$ mRNA expression were observed at the early stage. With prolonged hyperglycaemia, inflammatory cytokines increased markedly in the diabetes group. Although cytokines also rose in the TP group at later stages, their levels remained significantly lower than in the diabetes group. These results indicate that TP reduces neuroinflammation in the hippocampus, consistent with earlier findings [32, 33].

P-gp is an essential transmembrane transport protein that exports exogenous substances such as drugs and toxins out of cells, thereby influencing drug absorption, distribution and metabolism. In this study, P-gp expression initially increased and then declined during early diabetes. After TP treatment, P-gp expression increased significantly. Early diabetic stress may transiently stimulate P-gp expression as a protective response, consistent with earlier work [10]. As hyperglycaemic injury progressed, P-gp expression declined. TP upregulated P-gp expression, which may enhance the removal of harmful substances and protect brain tissue more effectively.

Glucose transporters (Glut proteins) regulate glucose uptake across membranes. Glut1 is expressed in vascular ECs and astrocytes [34], where it transports glucose across the BBB and into astrocytes for glycolysis. Glut1 has a high affinity for glucose and functions cooperatively with Glut3 to deliver glucose from the CNS interstitial fluid to neurons. Normal Glut1 function is essential for brain glucose uptake and utilisation [35]. Glut1 supports basal glucose uptake, whereas Glut4 mediates insulin-responsive glucose transport in peripheral tissues. Green Tea (GT) and EGCG increase Glut4 expression and reduce insulin resistance [36]. Reduced Glut1 expression impairs glucose transport and utilisation and contributes to brain energy deficiency. In vivo studies show that Glut1 levels decrease significantly in the hippocampus of mildly obese type 2 diabetic mice, accompanied by cognitive impairment [37]. In vitro, high-glucose exposure decreases Glut1 membrane expression in astrocytes, limiting glucose entry, reducing ATP production and impairing cognitive-related cellular functions [38]. In this study, Glut1 expression increased early in diabetes, then declined as injury progressed. TP treatment significantly increased Glut1 expression. Early diabetic stress may induce compensatory Glut1 expression to improve glucose uptake and mitigate initial metabolic disturbances. As hyperglycaemia persists, Glut1 expression decreases. TP enhanced Glut1 mRNA expression, reduced brain tissue damage and supported glucose transport and utilisation, consistent with its effects on Glut4 [35, 36].

In this study, RAGE protein expression exhibited a significant or highly significant positive correlation with P38, NF-$\kappa$B, IL-6, TNF-$\alpha$, ROS, and Glut1 expression. These findings suggest that TP may regulate key downstream components of the AGEs–RAGE signalling pathway by suppressing RAGE expression in hippocampal tissue.

Impaired BBB integrity is associated with reduced TJ proteins and inflammation in brain microvessels and neural tissue. Studies show that diabetes-induced PKC activation, overexpression of growth factors and intensified oxidative stress contribute to endothelial dysfunction [39]. EGCG, as a water-soluble compound, reduces A$\beta$1–42 plaque formation and decreases BBB permeability in ageing rats [40]. A combination of green tea, saffron, trans-resveratrol and phosphatidylcholine improves drug delivery across the BBB, preserves BBB integrity, reduces mitochondrial damage and enhances cell survival in CCF-STTG1 neurodegenerative models. This mixture also modulates cellular energy metabolism and apoptosis and reduces oxidative stress-driven inflammation [41]. GT polyphenols inhibit endothelial hyperpermeability [42]. Several studies also indicate significant reductions in claudin-5, claudin-3, occludin and ZO-1 expression in diabetic mice [43, 44, 45], although some T2DM models show increased Occludin and ZO-1 levels. In this study, compared with group C, ZO-1 mRNA expression showed no significant change at day 1, whereas other related molecules increased significantly. As modeling duration progressed, claudin-5, claudin-3, occludin, ZO-1, ZO-3 and JAM-3 mRNA levels decreased significantly, consistent with previous findings [43, 44, 45]. TP administration slowed the decline in these TJ molecules. By the end of the experiment, their expression levels in the TP group showed no significant difference from those of the normal group. These observations suggest that early hyperglycaemic stress may promote the production of TJ molecules to counteract BBB injury. However, prolonged hyperglycaemia suppresses TJ molecule production and gradually disrupts BBB structure. TP reduces the long-term effects of hyperglycaemia, promotes TJ molecule expression and alleviates BBB damage [40, 41, 42].

Hippocampal tissue consists mainly of neurons (pyramidal and granule cells) and glial cells, forming an extensive synaptic network. STZ-induced DM reduces hippocampal cell proliferation and survival [46]. H&E and Nissl staining reveal neuronal loss in diabetic mice [47]. Previous studies show that apoptosis increases in T2DM and db/db mice [48], with elevated pro-apoptotic proteins and reduced anti-apoptotic and synaptic proteins [49]. EGCG promotes proliferation, migration and angiogenesis of endothelial progenitor cells and inhibits apoptosis [50]. Green tea prevents hippocampal neuronal apoptosis in diabetic rats by inhibiting the JNK/MLCK pathway and improves cognitive function [15]. EGCG also reduces oxidative and nitrosative stress induced by long-term alcohol exposure, increases glutathione (GSH) and enhances antioxidant enzyme activity, improving pathological alterations in the cerebral cortex, cerebellum, and hippocampus [51]. In this study, hippocampal capillaries in group T0 were congested at 7 days. By day 14, cytoplasmic staining became lighter, neurons appeared rounded and swollen, and vascular spaces widened. By day 21, Nissl substance decreased markedly, and apoptosis was increased in glial cells and ECs, consistent with previous studies [48, 49]. TP treatment reduced neuronal swelling, narrowed perivascular gaps, lowered apoptosis and increased Nissl substance. These results indicate that hippocampal pathology initiates as early as day 7 after diabetes induction and worsens over time. TP markedly attenuated tissue injury, consistent with earlier findings [15, 50, 51].

This study has several limitations. First, this study was only conducted in early-stage diabetic mice, and the long-term effects of TP on hippocampal structure and BBB integrity remain to be further explored. Second, the protective mechanism was verified mainly at the mRNA and protein expression levels; further in-depth experiments such as gene intervention or pathway blockade are still needed to confirm the causal relationship. Third, this is a preclinical animal study, and the results cannot be directly extrapolated to human clinical applications without further verification.