To isolate and purify astrocytes, mice were euthanized and their brains were rapidly removed. The meninges were carefully stripped, and the brain tissue was minced and digested with trypsin. The resulting cell suspension was passed through a cell strainer to remove debris and then centrifuged to pellet the cells. For magnetic bead sorting, the cell pellet was resuspended in buffer containing anti-GFAP conjugated magnetic beads. Following incubation, the cells were passed through a magnetic column, where astrocytes bound to the beads were retained while other cell types were washed away. The purified astrocytes were then eluted from the column, counted, and plated onto poly-L-lysine coated flasks in Dulbecco’s modified Eagle’s medium/Ham’s F12 (DMEM/F12, Gibco) with 10% fetal bovine serum for further culture and experiments.

To achieve targeted knockdown of PDPN expression in astrocytes in vivo, we utilized a GFAP-specific promoter-driven adeno-associated virus (AAV) vector. This vector was engineered to carry either a control interfering sequence (GFAP-AAV-con) or a Pdpn-interfering sequence (NM_010329, GFAP-AAV-pdpn^-), obtained from Genechem Co., Ltd, Shanghai, China. GFAP serves as a biomarker for astrocytes, guiding the specificity of this approach.

The sense and anti-sense strands of siRNAs were: Pdpn, sense 5’-CCACGAUCACAAAGAACAUTT; antisense 5’-AUGUUCUUUGUGAUCGUGGTT.

6-week-old male C57BL/6J mice were randomly allocated into 4 groups (n=8 per group). Following a 1-week acclimatization period, a low-dose STZ injection and a high-fat diet were administered to induce T2DM models, the following concise steps are employed: Feed the mice a high-fat diet for a specified duration (3 weeks) to induce insulin resistance and metabolic alterations characteristic of T2DM. After the high-fat diet period, administer a low dose of STZ (commonly in the range of 50 mg/kg body weight) via intraperitoneal injection. This dose is chosen to partially destroy the pancreatic beta cells, causing a moderate reduction in insulin secretion without causing full-blown diabetes as seen with higher doses. Following STZ injection, monitor blood glucose levels regularly to confirm the development of hyperglycemia, indicative of T2DM. Mice with persistently elevated blood glucose levels (typically >13.9 mmol/L or >250 mg/dL) are considered to have successfully developed the T2DM phenotype. Then, after 1 weeks, mice underwent intracranial stereotactic injection (hippocampus) with either GFAP-AAV-Pdpn^- virus or GFAP-AAV-con virus (1x10¹¹ v.g per mouse), yielding the following groups: CON, T2DM, T2DM+AAV-con, and T2DM+AAV-pdpn^-. All housing and experimental procedures adhered to the guidelines of the Yangzhou University Animal Care and Use Committee, ensuring compliance with institutional regulations and national animal welfare standards.

The step-down apparatus consisted of a test box, an energized electric grid floor (10×10 cm), and an insulated jumping platform (YLS-3TB, Shandong Academy of Medical Sciences, Jinan, China), strategically placed in a corner. The three-day testing regimen began with a 5-minute acclimation period on day one, allowing mice to freely explore the uncharged box. On day two, the grid was activated at 36V and 0.5mA, and mice were gently positioned on the platform. Upon descending, mice encountered an electric shock, prompting immediate retreat to the insulated safety. On the final day, mice were retested, with their initial jump latency (step-down latency) and cumulative jumps (number of errors) within 3 minutes recorded for analysis.

Total RNA was isolated using a rapid extraction method via SteadyPure Universal RNA Extraction Kit (AG21017, Accurate Biotechnology, China). For the cDNA synthesis, 1 mg of total RNA of each sample was reverse transcribed using HiScript II Q RT SuperMix for qPCR (R223- 01; Vazyme, China). Real-time PCR was performed on cDNA samples using the FastStart Universal SYBR Green Master (Roche) on Step One Plus system (Applied Biosystems, Foster City, CA, USA). Primers are described in Supplementary Table S1 . The PCR settings used included denaturation (95°C for 2 min) and amplification steps repeated 40 times (95°C for 15s, 55°C for 30s, 72°C for 30s, and acquisition temperature for 15s). Analysis was conducted using the sequence detection software supplied with the instrument. For each sample, the delta delta Cycle of Threshold (ΔΔCT) (crossing point) values were calculated as the Ct of the target gene minus the Ct of the b-actin, assuming PCR efficiency equals to 1. Gene expression was derived according to the equation 2^–ΔΔCt; changes in gene expression are expressed relative to levels of the other group.

Briefly, flash-frozen tissues or cells were homogenized via radioimmunoprecipitation assay (RIPA) buffer containing protease inhibitors. BCA protein quantification was performed for tissue or cell lysates. PVDF membranes containing electrophoretically separated proteins from cells were probed with rabbit antibodies against Podoplanin n (EPR7073; Abcam), β-actin (81115-1-RR; Proteintech) treated with peroxidase (HRP)-conjugated goat anti-mouse or anti-rabbit IgG secondary antibody (BL001A; BL003A; Biosharp), and then visualized by enhanced Chemiluminescent horseradish peroxidase Substrate (Millipore).

Paraffin sections of 4 μm thickness were cut. Antigen repairing was performed in paraffin sections of brain tissues after dewaxing. Cell slides were fixed in 4% formaldehyde for 15min and permeabilized in 0.1% Triton X-100 for 5min. Then the tissue sections and cell slides were blocked by 10% goat serum (Servicebio, Wuhan, China) and incubated with primary antibodies at 4°C overnight. The primary antibodies used in this experiment were: rabbit anti-IL-6 (1:200; 12912; Cell signaling technology), rabbit anti-IL-1β (1:200; 16806-1-AP; Proteintech), mouse anti-TNF-α (1:200; 60291-1-Ig; Proteintech), rabbit anti-Transforming growth factor β-1 (TGF-β1; 1:200; 21898-1-AP; Proteintech). The secondary antibodies (SA00013-3, SA00013-4) for immunostaining were all purchased from Proteintech Co., Ltd. 4′,6′-diamidino-2-phenylindole (DAPI; Servicebio) was used for nuclei staining. Olympus FV1000 confocal microscopy system was used for image capture.

Based on previous transcriptome sequencing of activated and quiescent stellate cells (13), we identified 10 up-regulated genes and 10 down-regulated genes to explore the transcriptome alterations in activated astrocytes. Our results indicated that the Pdpn gene was the most notably overexpressed in the T2DM group. Subsequent qPCR ( Figures 1A, B ) and Western blot ( Figures 1C, D ) validations further confirmed that the Pdpn gene was significantly overexpressed in astrocytes of T2DM mice. Given the crucial role of Pdpn in inflammatory response, as revealed by GO biological process analysis, it suggests that Pdpn is implicated in the inflammatory processes that contribute to T2DM-induced brain injury.

As shown in Figure 2A , all four groups of mice maintained consistent body weights before the 8th week. However, the group with a high-fat diet and induced with low-dose STZ for T2DM experienced a significant increase in body weight after the 8th week ( Figure 2A ). Furthermore, these T2DM groups exhibited significantly elevated fasting blood glucose levels compared to the CON group, with statistically significant differences as depicted in Figure 2B .

In comparison to the 12-week-old T2DM group, the T2DM+AAV-pdpn^- group demonstrated an extended latency period. Additionally, this group exhibited a reduction in errors during the step-down passive avoidance test. Conversely, no statistically significant differences were discernible between the T2DM group and the T2DM+AAV-con group ( Figures 2C, D ).

Analogous findings were replicated in 28-week-old mice, where the T2DM+AAV-pdpn^- group displayed a markedly longer latency period in comparison to the T2DM group ( Figure 2E ). Furthermore, the T2DM+AAV-pdpn^- group had significantly fewer errors ( Figure 2F ).

These findings revealed that T2DM mice displayed compromised performance in the step-down test as early as 12 weeks of age, progressing to severe memory deficits by 28 weeks. Notably, the knockdown of PDPN expression significantly ameliorated the memory impairment observed in 28-week-old T2DM mice.

Given the established association between the Pdpn gene and the biological process of inflammatory response, we hypothesized that knockdown of Pdpn in astrocytes would reduce inflammation levels in the hippocampus of T2DM mice, thereby ameliorating their memory impairment. To validate this hypothesis, we performed immunofluorescence staining for IL-1β, IL-6, TNF-α, and TGF-β1, inflammatory markers closely linked to endothelial cell injury, on hippocampal sections from four groups of mice. Our results demonstrated that knockdown of Pdpn significantly decreased the expression of these four inflammatory indicators, suggesting a reduction in inflammation within the hippocampus ( Figure 3 ). This finding supports our initial premise that modulating Pdpn expression in astrocytes can mitigate the inflammatory milieu contributing to memory deficits in T2DM mice.