Diabetic db/db mice and their control littermates, db/dm, were obtained from the GemPharmatech Co., Ltd. Before the commencement of the study, the mice were acclimated to their new environment for 4–6 days. They were housed at a temperature of 21°C ± 2°C and a relative humidity of 50% ± 15%, under a 12 h light-dark cycle. The fasting blood glucose concentrations of the mice were measured at the beginning of the study (Supplemental Table S1). AAV9 harbouring Nrf2 shRNA (AAV9-CDH5-sh-Nrf2) and control vector (AAV9-CDH5-Scrambled) were injected intravenously into tail veins of 7 weeks old male db/db or db/dm mice respectively. To ensure the effect of 1 × 10¹² vg of virus infection, we did the subsequent experiment 2 weeks later. ALZET^® Osmotic Pumps (Model 2004) containing Cel (100 μg/kg/day, C0869, Sigma-Aldrich) (Fang et al., 2019) was implanted intraperitoneally, and then were calibrated to release the drug for 28 days (Figure 4E). The body mass and food intake of the mice were recorded every 3 days during the treatment period. All the mice were fasted for 12 h and subjected to glucose tolerance testing and insulin tolerance testing. For the analysis of signaling, bafilomycin A1 (10 mg/kg/2d, i. p., S1413, Selleck) and compound C (10 mg/kg/2d, i. p., S7306, Selleck) were administered. The procedures used in the study complied with the animal ethics guidelines of the institution and were approved by the Institutional Animal Care and Use Committee of Ningbo University, China.

Human umbilical vein endothelial cells (HUVECs), widely used for the study of vascular function and repair (Guixé-Muntet et al., 2017; Cherubini et al., 2023), were purchased from Lonza (Basel, Switzerland) and cultured at 37°C in a 5% CO₂-containing humidified incubator using endothelial cell growth medium-2 (EGM-2, CC-3156, and CC-4176, Lonza) containing normal glucose (NG, 5.5 mM) or HG + PA (33 mM HG + 200 μM PA) (Alnahdi et al., 2019; Huang et al., 2021), with or without 100 nM Cel, for 48 h (Li et al., 2017; Ma et al., 2020). Fifth-to-seventh-generation subconfluent cells were used for the experiments. Mannitol (MAN, 33 mM: 5.5 mM of glucose +27.5 mM of D-mannitol; M4125, Sigma-Aldrich) was used as an osmolarity control for HG. For the analysis of the signaling pathways, MG132 (5 μM for 3 h), bafilomycin A1 (Baf A1, 10 nM for 12 h), or compound C (10 μM for 12 h) were added.

HUVECs were transfected overnight with Nrf2-targeting shRNA (sc-37030, Santa Cruz) or nonsense shRNA at an MOI of 100× PFU/cell, and then the culture medium was replaced after 24 h. After 48 h, the expression of Nrf2 was measured by Western blot analysis.

Briefly, samples containing 30 μg protein were separated by SDS-PAGE on Tris-glycine gels and then transferred to polyvinylidene difluoride membranes. The membranes were blocked and incubated the primary antibody or secondary antibody [HRP-goat-anti-mouse (115-035-003, Jackson) or HRP-goat-anti-rabbit (111-035-003, Jackson)]. Immunoreactive bands were visualized using Pierce ECL Western blotting substrate (WBKLS0500, Millipore).

The HUVECs were lysed and obtained cytoplasmic and nuclear lysates using the Keygen Nuclear-Cytosol Protein Extraction Kit from Nanjing KeyGen Biotech. Co., Ltd.

The primary antibodies used to probe the membranes were against p62 (sc-48402, 1:1,000), Lc3 (sc-271625, 1:1,000) (Santa Cruz Biotechnology), p-AMPK (2531S, 1:1,000), AMPK (2532S, 1:1,000) (Cell Signaling Technology), Nrf2 (66504-1, 1:1,000), Keap1 (10503-2, 1:1,000), β-actin (20536-1, 1:1,000), GAPDH (60004-1, 1:5,000), and Lamin b (66095-1, 1:1,000) (Proteintech). ImageQuant 5.2 software (Molecular Dynamics) was used to quantitatively analyze the expression of specific proteins, and β-actin was used as the loading control.

IGEPAL CA-630 buffer [150 mM NaCl, 50 mM Tris-HCl, 2 mM EDTA, 1 μM leukopeptin (L5793, Sigma-Aldrich), 50 mM NaF, and 0.1 μM aprotinin (SRE0050, Sigma-Aldrich), 1% IGEPAL CA-630 (I8896, Sigma-Aldrich), pH 7.4] was used to lyse the HUVECs. After co-immunoprecipitation, the precipitates were washed five times with TBS. They were then eluted with glycine-HCl (0.1 M, pH 3.5) and the immunoprecipitates were subjected to immunoblotting using specific primary antibodies.

RNA was extracted from HUVECs using TRIzol reagent (9,108, Takara Bio Inc.), according to the manufacturer’s instructions. Next, total RNA (2 µg) was reverse transcribed into cDNA by using GoScript Reverse Transcription Kit (Promega, A5001). Quantitative RT-PCR analysis was performed using PowerUp SYBR Green Master Mix (Thermo Fisher Scientific, A25918). The relative expression of each gene was quantified using the 2^−ΔΔCT method and normalized to the expression of Actb. The specific primer sequences used for qRT-PCR are listed in Supplementary Table S2.

HUVECs (1.0 × 10⁴ per well) were cultured in 96-well plates for 24 h and treated with 25, 50, 75, 100, 200, 500, 750, or 1,000 nM Cel for 24 or 48 h, and their viability was assessed using the CCK-8 method. The absorbance of each well was measured using a microplate reader at 450 nm.

HUVECs treated with HG + PA were stained with dihydroethidium (DHE, D7008, Sigma-Aldrich) probes to measure their ROS concentrations. DHE is cell permeable and able to react with superoxide to form ethidium, which in turn intercalates with DNA and produces nuclear fluorescence. HUVECs were seeded on 24-well plates and treated with HG in presence or absence Baicalin for 72 h and then incubated with 5 μM DHE in DMSO for 30 min at 37°C. Nuclear DHE positive staining indicates superoxide generation in cells. The fluorescence intensity was observed with a computer-assisted microscope (EVOS, Thermo Fisher Scientific).

HUVECs were stained using an In situ Cell Death Detection kit (11684795910, Roche), according to the manufacturer’s protocol. Cultured HUVECs were fixed (4% paraformaldehyde for 30 min) and permeabilized (0.1% Triton-X 100 for 10 min) in 6-well plates. After washing with PBS, cells were incubated with 50 µL tunel reaction mixture for 60 min at 37°C. After washing, the nuclei were stained with DAPI. The stained cells were then examined using a confocal laser scanning microscope (TCS SP8, Leica, Wetzlar, Germany).

The angiogenic activity of the HUVECs was assessed using a Matrigel tube formation assay. Briefly, HUVECs were scattered on the surface of Matrigel (Corning, 354234) and incubated in a cell incubator at 37°C for 24 h, followed by staining with the cell permeability dye Calcein AM (Corning, 354216) for 30 min. The formation of capillary-like tubes was identified using a computer-assisted microscope (EVOS, Thermo Fisher Scientific), the lengths of the tubes were calculated using ImageJ software (National Institutes of Health, Bethesda, MD, United States), and the mean values for the replicate wells were calculated.

To establish a direct action of Cel on vascular, thoracic aortae from db/db and db/dm mice infected with AAV9-CDH5-sh-Nrf2 or AAV9-CDH5-Scrambled after 4-week treatment. Mice were surgically isolated, cleaned, and sectioned to form 0.5 mm rings. The lentivirus-mediated gene transfer (Lv-CDH5-sh-Nrf2) and vector (Lv-sh-Scrambled) were also respectively transfected with aortic rings from db/db and db/dm mice infected with AAV9-CDH5-sh-Nrf2 or AAV9-CDH5-Scrambled. Next the rings were embedded and cultured in 96-well plates containing type I collagen (08-115, Millipore, 1 mg/mL) as previously described (Aplin et al., 2008; Baker et al., 2011). Then the medium was removed and replaced with either mannitol (33 mM: 5.5 mM of glucose +27.5 mM of D-mannitol) or HG + PA (33 mM HG + 200 μM PA) in the presence or absence of Cel (100 nM). Pre-treatment with Cel and VEGF were performed every day to analyze the role of the key signaling pathway. The number of endothelial microvessels that grew out of the rings was counted during the exponential growth phase to assess angiogenesis. Before the regression phase, the rings were fixed and immunofluorescence-stained for CD31 (ab281583, Abcam). Photomicrographs were obtained after 12 days, and the total number of branches was counted using ImageJ.

Cell migration was assessed using a scratch assay, as previously described (Das et al., 2018). Cells were seeded into six-well plates and cultured overnight until a confluent monolayer formed, which was then scratched using a 200 µL pipette tip. Before and 36 h after this injury, images of the damaged cell monolayers were captured using a microscope (EVOS, Thermo Fisher Scientific) and quantified using ImageJ software. All the experiments were performed in the presence of mitomycin-C (Jain et al., 2019) (10 μM, Selleck Chemicals, S8146), which inhibits cell proliferation.

General anesthesia was induced in mice by the inhalation of 2% isoflurane. Full-thickness wounds were created to the shaved dorsal skin of db/db and db/dm mice infected with AAV9-CDH5-sh-Nrf2 or AAV9-CDH5-Scrambled using 8 mm skin biopsy punches. Skin wound edge injection of the Lv-CDH5-sh-Nrf2 and control vector Lv-Scramble, and then each of the wounds were treated with DMSO, Cel (100 nM) with a diameter of 10 mm, and bandaged with sterile gauze. The subsequent wound closure was assessed daily (Li et al., 2015). The formula for calculating the remaining open wound area was as follows: wound area remaining open (%) = (open area on the indicated day/original wound area) × 100%. Five-micrometer-thick paraffin-embedded tissue sections were prepared, dewaxed, rehydrated, and stained with immunofluorescence of keratin 14 (ab119695, Abcam) and CD31 (ab281583, Abcam) respectively. After washing, the sections were incubated with secondary antibodies [AlexaFluor 647-conjugated anti-rabbit (ab150079, Abcam) in different tissue sections. The re-epithelialization ratio (leading edge ratio) was calculated as [(a + b)/c] × 100% (shown in Figure 4F; a and b are the lengths of the leading edges, and c is the initial length of the wound) (Safferling et al., 2013).

Immunofluorescence was used to identify areas of CD31, VCAM-1, and p65 expression in the aortic wall. Eight-micrometer-thick paraffin sections were prepared and incubated with anti-CD31 antibody (ab281583, Abcam), VCAM-1 antibody (ab134047, Abcam), or p65 antibody (66535-1, Proteintech). After washing, the sections were incubated with secondary antibodies [AlexaFluor 647-conjugated anti-rabbit (ab150079, Abcam) or AlexaFluor 488-conjugated anti-mouse (ab150113, Abcam)] at room temperature for 60 min. DAPI was used to label the nuclei. The CD31 or VCAM-1-stained area were examined using a confocal laser scanning microscope (TCS SP8, Leica, Wetzlar, Germany) and measured using ImageJ software. Data are presented as the percentages of the total area that was immunostained.

To quantitatively analyze the nuclear localization of Nrf2 in HUVECs. Cultured HUVECs were fixed (4% paraformaldehyde for 30 min), permeabilized (0.5% Triton-X 100 for 30 min) and blocked (5% BSA for 2 h) in 6-well plates. Then, cells were incubated with rabbit anti-Nrf2 primary antibody (1:200, 66504-1, Proteintech) overnight at 4°C. After washing, the cells were then incubated with secondary antibody (AlexaFluor 488-conjugated anti-mouse, ab150113, Abcam). The nuclei were stained with DAPI. The results were photographed via a confocal laser scanning microscope (TCS SP8, Leica, Wetzlar, Germany). The colocalization of Nrf2 and DAPI was assessed in randomly selected cells.

All the analyses were performed with the investigator being blinded to the groups of mice or cultured cell treatments. Statistical comparisons were made using the two-tailed Student’s t-test for two experimental groups or the one-way analysis of variance (ANOVA) for multiple groups. Statistical analyses were performed using GraphPad Prism.

To determine the effects of Cel on the HG + PA-induced defects in HUVECs, we first evaluated the cytotoxicity of Cel in vitro. We found that concentrations of Cel <400 nmol/L caused no obvious cytotoxicity, whereas there was significantly lower cell viability when the concentration of Cel exceeded 750 nmol/L (Figure 1A). This is consistent with the effects of Cel to impair vascular growth in tumors at high concentrations (Pang et al., 2010). Furthermore, we found that the deleterious effect of HG + PA on the viability of HUVECs was significantly ameliorated by 100 nM Cel (Figure 1B). It has been reported that HG + PA-induced endothelial dysfunction is mediated through multiple mechanisms, including oxidative stress and pro-inflammatory responses (Xie et al., 2008). We demonstrated that HG + PA-induced inflammatory activity, reflected in the upregulation of the pro-inflammatory cytokines IL-1β, IL-6, IL-8, and TNF-α, was inhibited by Cel treatment (Figure 1C). In addition, we assessed the effect of Cel on oxidative stress in the endothelial cells using DHE. HG + PA treatment of HUVECs resulted in a significant increase in superoxide production, and this was attenuated by co-treatment with Cel (Figures 1D,E). HG + PA also induced a high level of apoptosis in HUVECs, as indicated by a larger proportion of TUNEL-positive cells. However, this HG + PA-induced apoptosis was significantly attenuated by Cel (Figures 1F,G). Furthermore, the migration of HG + PA-exposed HUVECs was significantly impaired, but this defect was significantly ameliorated by Cel (Figures 1H,I).

Nrf2 plays a key role in the cellular response to oxidative stress (Jyrkkänen et al., 2008). Therefore, we speculated that Cel may ameliorate HG + PA-induced endothelial damage by activating Nrf2 and its downstream target genes. Nrf2 protein expression was downregulated by the HG + PA treatment, but consistent with previous findings (Li et al., 2017), the Nrf2 protein expression was significantly restored by Cel treatment (Figures 2A,B). Importantly, we found that the mRNA expression of the Nrf2 target genes NQO1, NQO2, HO1, SOD2, and CAT was significantly reduced by HG + PA treatment, but this was corrected by Cel treatment (Figures 2C–G). In addition, the nuclear localization of Nrf2 was significantly reduced by HG + PA treatment, but this did not occur in the presence of Cel (Figures 2H,I). The nuclear localization of Nrf2 in HUVECs was further characterized using immunofluorescence (Supplementary Figure S1A), which showed similar results. Overall, these results demonstrate that Cel restores Nrf2 protein expression and promotes the nuclear enrichment of Nrf2. Furthermore, consistent changes in Nrf2 target gene expression accompany the effects of Cel on Nrf2 nuclear localization, indicating that Cel activates the transcriptional function of Nrf2, ameliorates HG + PA-induced oxidative stress, and reduces the inflammatory response.

Cel is thought to improve skeletal muscle and bone function through upregulation of the AMPK signaling pathway (Abu et al., 2020; Li et al., 2020), and it also ameliorates angiotensin II-mediated vascular smooth muscle cell senescence (Xu et al., 2017) and osteoarthritis (Feng et al., 2020) through the induction of autophagy. In addition, the function of Nrf2 has been reported to be regulated by the p62/Keap1/Nrf2 signaling pathway, the activation of which liberates Nrf2 from Keap1 and permits it to translocate to the nucleus (Zhang et al., 2021).

Consistent with this, we found that the low Nrf2 protein expression was restored (Figures 6A,C), and the expression of Keap1 remained unaffected by Cel and MG132 (a proteasome inhibitor) co-treatment (Figures 6A,D). Moreover, immunoprecipitation revealed that MG132 promotes the binding of Nrf2 to Keap1 in cells treated with HG + PA, indicating that Keap1 accumulates in cells and binds to Nrf2, causing it to undergo ubiquitination degradation in response to HG + pA. However, MG132 did not significantly increase the association of Nrf2 and Keap1 in Cel-treated cells, indicating that Cel affects Keap1 protein levels through the regulation of the non-proteasomal degradation pathway, ultimately inhibiting the binding of Keap1 to Nrf2, causing the latter to undergo ubiquitination degradation (Figures 6A,B).

Given that the accumulation of Nrf2 is accompanied by abundant autophagosomal degradation (Komatsu et al., 2010; Ichimura et al., 2013; Zhang et al., 2021; Taguchi et al., 2012), we next analyzed whether Cel increases Nrf2 expression by modulating autophagy. Co-treatment with Baf-A1, an inhibitor of autophagic degradation, blocked the effect of Cel on Nrf2 accumulation (Figures 6E–G). A previous study showed that Keap1 interacts with p62 (an autophagic adapter) and frees Nrf2 translocate to the nucleus (Taguchi et al., 2012). Consistent with this, in the present study, we found that the interaction between Keap1 and p62 in HUVECs was increased by HG + PA treatment and restored to normal by Cel administration (Figures 6E,F). Interestingly, the interaction between Keap1 and Nrf2 was increased by co-treatment with Baf-A1, even in the presence of Cel (Figures 6E,F), indicating that Baf-A1 inhibits the autophagy pathway involving Keap1 and p62, but does not reduce the expression of Keap1 protein, and increases the probability of Keap1 and Nrf2 binding, thereby increasing the ubiquitination and degradation of Nrf2. Thus, Nrf2 protein expression is reduced by Baf-A1 (Figures 6E,F). We also found that the effects of Cel on the phosphorylation of AMPK and autophagy were prevented by treatment with the specific pharmacological inhibitor of AMPK, compound C (Figures 6H–K). In addition, Cel had protective effects on tube formation (Figures 7A,B) and cell migration (Figures 7C,D), but these were abolished by compound C or Baf-A1 co-treatment.

To determine whether the protective effect of Cel against hyperglycemia-induced endothelial impairment is mediated via AMPK/p62-dependent autophagy in diabetic mice, compound C (10 mg/kg/2d, i. p.) or Baf-A1 (10 mg/kg/2d, i. p.) was administered. As expected, inhibition of the AMPK pathway or autophagy largely prevented the endothelial protective effect of Cel, as shown by a substantial increase in VCAM-1 expression in aortic vascular endothelium cells from db/db mice (Figures 7G,H) and an impairment in aortic ring sprouting (Figure 7E,F). These data suggest that Cel ameliorates the HG + PA-induced endothelial dysfunction by promoting the degradation of Keap1 in HUVECs, secondary to an effect on AMPK/p62-dependent autophagy, thereby liberating Nrf2 to translocate to the nucleus.