2.1 Animal procedures

Diabetic db/db mice and their control littermates, db/m, were obtained from Jackson Laboratories (Strain: BKS. Cg-Dock 7^m +/+ Lepr^db/J). The mice were housed in an environmentally controlled room for 4–6 days to adapt to the environment before experimentation. After withholding food for 12 h, blood samples were obtained from the tail veins. ALZET® Osmotic Pumps (Model 2004) containing PBS (Gibco BRL, 10010), bFGF (30 μg/kg/day, Thermo Fisher Scientific, CTP0263) or GSNO (10 mg/kg/day, Sigma-Aldrich, N4148) [7] were implanted intraperitoneally and were calibrated to release the drug for 28 days. After a 4-week course of treatment, aortic ring assays were performed. Blood glucose level at different time points were measured by glucose analyzer (Lifescan Surestep).

Male C57BL/6 mice (6–8 weeks, 18–23g), were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). All animal were kept in a standard laboratory condition of temperature 21 ± 2 °C, relative humidity 50 ± 15%, 12 h light-darkness cycles, with water and food available ad libitum. All animal experiments and methods performed in this study followed ethical guidelines for animal studies and were approved by the Institutional Animal Care and Use Committee of Wenzhou Medical University, China.

2.2 Cell culture and treatments

HUVECs were purchased from Lonza and cultured in endothelial cell growth medium-2 (EGM-2; BulletKit, Lonza, CC-3156 and CC-4176) until the start of the experiment. Subconfluent cells (5–7 passages) were used in the experiments. Before starting the experimental procedures, the medium was removed and replaced with phenol red-free low-glucose D-MEM (Gibco BRL, 11054020) supplemented with 1% calf serum (Gibco BRL, 16010159) for 12 h, then HUVECs were placed in EGM-2 consisting of either NG (5.5 mM) or HG + PA (33 mM HG +10 nM PA) in the presence or absence of bFGF (20 ng/mL, Thermo Fisher Scientific, PHG0369) [21] for 72 h; MAN (33 mM: 5.5 mM of glucose + 27.5 mM of D-mannitol) was used as the osmotic control for HG. Media were changed every 24 h.

When analyzing the alteration of eNOS and iNOS, HUVECs were cultured in HG medium for 2 days, 3 days, 4 days and 5 days, respectively. For signaling pathway analysis, each pathway antagonist: compound C [22] (10 μM; Selleck Chemicals, S7306), LY294002 [23] (20 μM; Sigma-Aldrich, L9908), carboxy-PTIO [24] (100 μM; Sigma-Aldrich, 217386) or BJG398 [25] (10 nM; Selleck Chemicals, S2183) were given as pretreatment for 2 h every day before bFGF administration. In some experiments, HUVECs were exposed to HG + PA in combination with the NO donor GSNO [26] (100 μM) for 72 h.

2.3 Immunoblotting analysis

Proteins were extracted from cells and subjected to SDS–polyacrylamide gel electrophoresis, as previously described [27]. Primary antibodies included: p-eNOS^S1177 (CST, 9571), eNOS (CST, 32027), iNOS (Abcam, ab3523), p-AKT^S473 (CST, 4060), AKT (CST, 4691), p-GSK3β^S9 (CST, 9323), GSK3β (CST, 9315), p-AMPKα^T172 (CST, 2535), AMPK (CST, 2603), p65 (CST, 8242), IKKβ (Abcam, ab178870), Biotin (Abcam, ab1227), Trx (CST, 2429), TrxR (CST, 6925), Txnip (CST, 14715), GSNOR (Abcam, ab177932). The expression of β-Actin (Abcam, ab8226) was used as a loading control.

2.4 RNA isolation and quantitative real-time-PCR (qRT-PCR)

Total RNA was extracted from HUVECs by using TRIzol Reagent (Invitrogen, 15596018) 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). Data were analyzed and normalized with β-Actin. Gene-specific primer sequences used for qRT-PCR are listed in Supplementary Information Table S3.

2.5 Construction of adenovirus

Recombinant adenovirus vectors were constructed, propagated and tittered as previously described [28]. We made replication-defective human adenovirus type 5 (devoid of E1) harboring human p65^WT (Ad-Cdh5-p65^WT), p65^C38S (Ad-Cdh5-p65^C38S); IKKβ^WT (Ad-Cdh5-IKKβ^WT), IKKβ^C179S (Ad-Cdh5-IKKβ^C179S) and human Txnip (Ad-Txnip). The Cdh5 promoter ensures restriction of transgene expression in endothelium which was originally described [29], and an in-house-generated Ad-LacZ was used as a control. For adenovirus-mediated gene transfer, these constructed adenovirus vectors were transfected into HUVECs at a MOI of 100 × PFU/cell for 24 h.

2.6 Construction of shRNA adenoviral expression vectors

The pSilencer 2.1-U6 expression vector was purchased from Ambion (Ambion, AM5762). The RNU6-1 RNA polymerase III promoter and the polylinker region were subcloned into the adenoviral shuttle vector pDC311 (Microbix, PD-01-25). The human eNOS-shRNA targeting sequence was 5ʹ-GTGGCCAACGCCGTGAAGATC-3ʹ. The human iNOS-shRNA targeting sequence was 5ʹ-GGTAACAAAGGAGATAGAAAC-3ʹ. The human Trx-shRNA targeting sequence was 5ʹ-GCAGGTGATAAACTTGTAGTA-3ʹ. For Scrambled shRNA, an in-house-generated shRNA adenovirus that encodes a scrambled sequence was used as control. For adenovirus-mediated gene knockdown, these constructed adenovirus vectors were transfected into HUVECs at a MOI of 20 × PFU/cell for 24 h.

2.7 Lentiviral vector construction, virus production and titration

To construct a lentiviral vector overexpressing the human p65^WT (LV-EGFP-Neo-p65^WT-3Flag), p65^C38S (LV-EGFP-Neo-p65^C38S-3Flag); IKKβ^WT (LV-EGFP-Puro-IKKβ^WT-3Flag), IKKβ^C179S (LV-EGFP-Puro-IKKβ^C179S-3Flag), the PCR product was digested with XbaI and Xho I, and then ligated into the lentiviral backbone plasmid pLenti-EF1a-EGFP-P2A-Neo-CMV-3Flag and pLenti-EF1a-EGFP-P2A-Puro-CMV-3Flag respectively by T4 DNA ligase (TaKaRa, 2011B). An in-house-generated LV-LacZ was used as a control. For lentivirus-mediated gene transfer, these constructed lentiviral vectors were transfected into HUVECs at a MOI of 100 × PFU/cell for 48 h. After 48 h, a stable p65^WT, p65^C38S, IKKβ^WT, IKKβ^C179S overexpression cells was established and selected with puromycin (BD Biosciences, B7587) and G418 (Selleck Chemicals, S3028), respectively.

2.8 Measurement of trx activity

Trx activity was measured using the insulin disulfide reduction assay as described elsewhere [30]. Briefly, Total cellular protein was extracted with lysis buffer. Cellular protein (40 μg) extracts were incubated with activation buffer at 37 °C for 15 min before they were incubated with Trx reductase (Sigma-Aldrich, T7915) in the reaction buffer at 37 °C for 30 min. Distilled water was loaded as a control. Finally, the reaction was terminated by stopping buffer. The absorbance was assessed using a multifunctional microplate reader (SpectraMax M5; Molecular Devices) at a wavelength of 412 nm. Trx activity was expressed as nicotinamide adenine dinucleotide phosphate oxidized per minute per milligram of protein (μM/min/mg).

2.9 Measurement of GSNOR enzymatic activity

HUVECs were prepared on ice in a solution. The GSNOR activity was measured by GSNO-dependent NADH (reduced form of NAD⁺) consumption [31]. Briefly, cell lysate (0.3 mg/mL) was incubated with 75 μM NADH in reaction buffer containing 0 or 100 μM GSNO at room temperature, and NADH fluorescence (absorption at 340 nm and emission at 455 nm) was measured over time to determine the initial rate of GSNO-dependent NADH consumption.

2.10 Biotin-switch assay

S-nitrosylated proteins were detected by using the biotin-switch assay [32] with some modifications. In brief, lysates (50 mL) were incubated in HENS buffer and 20 mM methyl methanethiosulfonate (MMTS) at 50 °C for 20 min. Proteins were precipitated with cold acetone, washed twice, resuspended in HENS, and mixed with 0.2 mM biotin-HPDP (BD Biosciences, A8008) and 2.5 mM ascorbate at ambient temperature for 1 h. Finally, biotinylated proteins were purified by using streptavidin-agarose beads (Sigma-Aldrich, GE90100484), separated by SDS-PAGE, and detected by immunoblotting.

2.11 Fluorescent detection of S-Nitrosylation and immunofluorescence

Biotin-switch assay was performed on fixed and permeabilized HUVECs by first blocking free thiols using HENS buffer containing 20 mM MMTS at 50 °C for 20 min. Nitrosothiols were labeled in HENS buffer (1% SDS), 0.4 mM biotin-HPDP and 1 mM ascorbic acid at ambient temperature for 1 h. Biotinylated proteins were visualized by using avidin conjugated with Alexa 568 (Thermo Fisher Scientific, S11226). Finally, cells were analyzed using a confocal laser scanning microscope (Leica TCS SP8, Wetzlar, Germany).

2.12 Electrophoretic mobility shift assay

DNA binding of NFκB was measured by performing electrophoretic mobility shift assay (EMSA) as previously described [33]. Binding reactions were resolved on a 4% nondenaturing polyacrylamide gel at 22 mA for 30 min at 4 °C in 0.5 × TBE. Samples were then electrotransferred onto positively charged nylon membrane and subjected to antibody reaction. Finally, chemiluminescent detection following ChemiDoc MP device (Bio-Rad, Hercules, CA, USA) and densitometry was conducted with ImageQuant 5.2 software (Molecular Dynamics, Sunnyvale, CA).

2.13 Luciferase assay

For luciferase reporter assays, firefly luciferase constructs driven by NFκB binding sequences and Renilla luciferase pRL-TK (Int-) plasmid were used (Promega, E2271). Lysates were analyzed with the Dual-Luciferase Reporter Assay System (Promega, E1910). Firefly luciferase and Renilla luciferase were detected on a Veritas Microplate Luminometer (Promega, Madison, WI, USA).

2.14 In vitro angiogenesis (tube formation) assay

The in vitro angiogenic activity of HUVECs was determined by Matrigel tube formation assay. Capillary-like tube formation was observed with a computer-assisted microscope (EVOS, Thermo Fisher Scientific, MA, United States). The tube length in duplicate wells was counted and averaged using ImageJ software (National Institutes of Health, Bethesda, Maryland, USA).

2.15 Aortic ring assay

To establish a direct action of bFGF on vascular, thoracic aortae from db/db and db/m mice after four-week treatment or C57BL/6 mice were surgically isolated, cleaned and dissected into 0.5 mm rings, then rings were embedded in fibrin [34,35].

For lentivirus-mediated gene transfer, these constructed lentiviral vector were transfected aortic rings from db/db and db/m mice respectively at 30–50 rings per well of a 24-well plate by adding 0.5 mL of Opti-MEM (Gibco BRL, 51985–026) and 1 × 10⁸ vector genome copies to each well for 48 h as described previously [14]. Before the regression phase, rings were fixed for immunofluorescence staining of CD31 (Abcam, ab24590). Pictures were taken on day 12, and the total number of branches was counted using ImageJ (National Institutes of Health, Bethesda, Maryland, USA).

2.16 Wound healing scratch assay

Cell migration was determined using the wound healing scratch assay as previously described [36]. Cells were seeded on 6-well plates and grown overnight, until forming a confluent monolayer and a scratch was made using a 200 μL pipette tip. Images of the wounded cell monolayers were taken using a microscope (EVOS, Thermo Fisher Scientific, MA, United States) at 0 and 36 h after wounding and controlled by ImageJ software (National Institutes of Health, Bethesda, Maryland, USA). All experiments were performed in the presence of Mitomycin-C (10 μM, Selleck Chemicals, S8146) [37] to inhibit cell proliferation.

2.17 Wound-healing assay

General anesthesia was performed with 2% inhaled isoflurane and then injected subcutaneously with the analgesic. Two full-thickness wounds were created on the shaved dorsal skin of each of the 8-week-old db/db mice and db/m mice using 8 mm skin biopsy punches, then each wound covered hydrogel scaffolds respectively including PBS, bFGF (30 ng/mL) and GSNO (150 μM) with a diameter of 10 mm, and bandaged with sterile cotton-cloth. In addition, to evaluate rescue effects, adenoviruses expressing endothelium-specific p65^WT (Ad-Cdh5-p65^WT), p65^C38S (Ad-Cdh5-p65^C38S); IKKβ^WT (Ad-Cdh5-IKKβ^WT), IKKβ^C179S (Ad-Cdh5-IKKβ^C179S) and Ad-LacZ (control) were injected intradermally into the wound edges in the db/db mice and db/m mice the day after wounding, and the wound closure rate was measured [38]. The wound areas remaining open were calculated as follows: wound areas remaining open (%) = (open area on the indicated day/original wound area) × 100%. Paraffin-embedded tissue sections (5 μm in thickness) were deparaffinized and rehydrated, followed by H&E staining. Reepithelization ratios (leading edge ratios) were measured and calculated by [(a + b)/c] × 100% (shown in Supplementary Fig. 5d, where a and b are the length of the axes for the leading edges, and c indicates the axis of initial wound lengths) [39]. Blood fasting glucose levels were measured during the experiment (Supplementary Information Table S2).

2.18 Immunofluorescence for CD31 staining

Capillary density in wound skin tissues were determined as reported [40]. Briefly, Paraffin-embedded tissue sections (5 μm in thickness) were deparaffinized and rehydrated, then fixed with 4% (w/v) paraformaldehyde for 20 min and permeabilized with 0.5% (v/v) Triton X-100 for 20 min at 37 °C. After wash, the cells were blocked in 2% (v/v) BSA for 2 h and incubated with primary Rabbit antibody against CD31 (Abcam, ab24590) at 4 °C overnight. After incubation and wash, cells were blocked with Alexa fluor 647-conjugated anti-rabbit IgG secondary antibody for 1 h. After washes with PBS, the cell nuclei were stained with DAPI for 15 min. The stained wound skin tissues were imaged with a confocal laser scanning microscope (Leica TCS SP8, Wetzlar, Germany). The total tissue area and the CD31 stained positive area were measured with ImageJ software (National Institutes of Health, Bethesda, Maryland, USA). Data were expressed as percentage of positive staining area per analyzed area.

2.19 Immunoprecipitation

HUVECs were lysed with IGEPAL CA-630 buffer (50 mM Tris-HCl, pH 7.4, [Sigma-Aldrich, T5030], 1% IGEPAL CA-630 [Sigma-Aldrich, I8896], 10 mM EDTA, 150 mM NaCl, 50 mM NaF, 1 μM leupeptin [Sigma-Aldrich, L5793], 0.1 μM aprotinin [Sigma-Aldrich, SRE0050]). After immunoprecipitation, the samples were washed with TBS 5 times. They were then eluted with glycine-HCl (0.1 M, pH 3.5) and the immunoprecipitates were subjected to immunoblotting using specific primary antibodies.

2.20 Isolation of nuclear and cytosolic extracts

The isolation of nuclear and cytosolic extracts was performed in HUVECs with a Nuclear and Cytoplasmic Extraction kit (Thermo Fisher Scientific, 78833) according to the manufacturer's instructions.

2.21 Nitrite/nitrate assay

Nitrite/nitrate production in HUVECs were analyzed by using an Ultra-sensitive assay for nitrite/nitrate assay kit (Sigma-Aldrich, 23479), according to the manufacturer's instructions.

2.22 TUNEL staining

HUVECs were stained with an In Situ Cell Death Detection Kit (Roche, 11684795910) according to the manufacturer's protocol.

2.23 Statistical analysis

The results are presented as mean ± SEM from at least four independent experiments. All analyses were performed with the experimenter blinded to the groups of mice and cultured cells. Statistical comparisons were made with two-tailed Student's t-test for two experimental groups or one-way analysis of variance (ANOVA) for multiple groups with SPSS software. For enzymatic activity analyses, Wilcoxon rank-sum test were performed. In the TUNEL⁺ cells analyses, we performed the Mann-Whitney U test for the TUNEL-positive cells. If not stated otherwise, asterisks indicate significance: ***p < 0.001, **p < 0.01, *p < 0.05 and n. s. represents no significant. Statistical analyses were done using GraphPad Prism (GraphPad Software).

3.1 Effect of bFGF on hyperglycemic and hyperlipidemic impairment of S-nitrosylation

We examined the impact of hyperglycemia and hyperlipidemia (palmitic acid, PA) combined treatment (HG + PA) on S-nitrosylation in HUVECs over 5 days of exposure to HG + PA. 2–3 days of HG + PA treatment decreased protein S-nitrosylation, especially on day 3, but subsequently reversed on days 4–5 (Supplementary Fig. 1A and B). Interestingly, HG + PA exposure for these discrete time points biphasically affected different NOS isoforms. At 2–3 days, activity of endothelial nitric oxide synthase (eNOS) was significantly decreased, while at 4–5 days, expression of inducible nitric oxide synthase (iNOS) gradually increased (Supplementary Fig. 1C and D). Consistently, NO production mimicked both aspects of this phenotype (Supplementary Fig. 1E), suggesting that protein S-nitrosylation was mediated by NO, as was reflected by the corresponding biphasic regulation of eNOS/iNOS activity or expression.

To determine whether dysregulation of iNOS-induced protein S-nitrosylation altered endothelial function under HG + PA conditions, iNOS was knocked down using an adenovirus expressing iNOS-shRNA. In HUVECs, iNOS-shRNA did not affect S-nitrosylation on days 1–3 of HG + PA treatment, but it eliminated the observed increase of S-nitrosylation on day 5 (Supplementary Fig. 1F and G). These findings were corroborated by whole-cell quantitative nitrite/nitrate production (Supplementary Fig. 1H). The results suggested that iNOS induced endothelial cell protein S-nitrosylation during late-stage exposure to HG + PA, while decreased S-nitrosylation in early-stage exposure was iNOS-independent.

As previous studies have demonstrated that endothelium maintains vascular homeostasis in part by regulating S-nitrosylation homeostasis [13], we hypothesized that bFGF could alleviate HG + PA-induced endothelial dysfunction by correcting aberrations in protein S-nitrosylation. Considering that decreased S-nitrosylation on day 3 of HG + PA exposure was the most important and physiologically relevant aspect of this results, thereby, we sought to determine whether bFGF corrected deficits in S-nitrosylation during early exposure of HUVECs to HG + PA. Firstly, we measured S-nitrosylation of endogenous proteins in the presence or absence of GSNO using biotin-switch assay and immunofluorescence detection of S-nitrosylation in HUVECs (Supplementary Fig. 2). Secondly and consistently, HUVEC protein S-nitrosylation was markedly decreased following 3 days of HG + PA exposure, but was restored by treatment with bFGF or the physiological NO donor S-nitrosoglutathione (GSNO) (Fig. 1C and D). Furthermore, we performed a biotin-switch assay on fixed HUVEC monolayers to visualize S-nitrosylation levels using Alexa-568-labeled avidin [41]. bFGF or GSNO co-treatment markedly increased cytoplasmic and nuclear SNO levels compared with HUVECs exposure to HG + PA (Fig. 1A and B).

Fig. 1

Effect of bFGF on hyperglycemia and palmitic acid-induced reduction of protein S-nitrosylation. HUVECs were exposed to MAN (33 mM) and HG + PA (33 mM HG + 10 nM PA) in the presence or absence of bFGF (20 ng/mL) or GSNO (100 μM) for 3 days, MAN (33 mM: 5.5 mM of glucose + 27.5 mM of D-mannitol) was served as the osmotic control for the HG. (A) Fixed cells were subjected to a biotin-switch assay. S-nitrosylation was visualized using Alexa-568-conjugated avidin (red) and nuclei were stained with DAPI (blue). Nonspecific staining was determined in control cells incubated with Alexa-568 avidin. Scale bars = 115 μm. (B) The quantitative analysis of fluorescent intensity in at least 6 separate fields, values displayed are means ± SEM of 6 independent experiments. One-way ANOVA Bonferroni's test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA. (C) Whole-cell lysates were subjected to the biotin switch assay as described in “Materials and Methods”. (D) The quantitative analysis of SNO protein immunoblots, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To determine whether the effect of bFGF was receptor mediated, HUVECs were treated with the pan-FGFR antagonist BGJ398 (BGJ). BGJ exposure did not affect S-nitrosylation in normal HUVECs, but largely blocked bFGF-mediated restoration of protein S-nitrosylation in HG + PA-treated HUVECs (Supplementary Fig. 3A–D). Together, these data indicated that bFGF elicited receptor-mediated activity to modify protein S-nitrosylation during early exposure of HUVECs to HG + PA.

3.2 Impaired S-nitrosylation contributed to endothelial dysfunction in vitro and ex vivo

To determine whether impaired protein S-nitrosylation contributed to endothelial dysfunction in HG + PA conditions, and whether bFGF improved endothelial function by regulating protein S-nitrosylation in this context, HG + PA-treated HUVECs were cultured with bFGF in the presence or absence of the NO scavenger carboxy-PTIO (c-PTIO) for 72 h. Notably, exposure of HUVECs to c-PTIO decreased S-nitrosylated protein levels even in the presence of bFGF (Fig. 2A and B). In parallel, HG + PA impaired cell migration (Fig. 2E and F) and tube-forming capacity (Fig. 2G and H) of HUVECs were improved by bFGF, but counteracted by c-PTIO co-treatment. In addition, bFGF alleviated HG + PA-induced apoptosis in HUVECs, as demonstrated by decreased TUNEL-positive cells (Fig. 2C and D), which was significantly impaired by co-treatment with c-PTIO. These results suggested that bFGF-mediated S-nitrosylation could improve cellular proliferation and migration, and suppress cellular apoptosis.

Fig. 2

Decomposition of S-nitrosylation contributes to endothelial dysfunction both in vitro and ex vivo. HUVECs were cultured either in MAN (5.5 mM) and HG + PA (33 mM HG+10 nM PA) medium in the presence or absence of bFGF (20 ng/mL) or c-PTIO (100 μM) for 72 h. (A) Biotin switch assay of whole-cell lysates. (B) The quantitative analysis of SNO protein immunoblots, the results were normalized to HUVECs exposed to MAN, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used. (C) The apoptotic cells were labeled with green, and nuclei were stained with DAPI (blue). Scale bars = 90 μm. (D) The quantitative analysis of TUNEL⁺ cells in at least 6 separate fields, values displayed are means ± SEM of 10 independent experiments. Non-parametric Mann-Whitney U test was used. (E) A scratch wound healing assay was performed in the presence of Mitomycin-C (10 μM). Cell monolayers were imaged at 0 and 36 h after wounding. Red vertical lines indicate the wound area borders. Scale bar = 65 μm. Wound area is analyzed by ImageJ software. (F) Cell migration distances were measured based on the data; values displayed are means ± SEM of 6 independent experiments. One-way ANOVA with Tukey's multiple comparison tests was used. (G) Capillary-like tube formation was assessed by matrigel angiogenesis assay in HUVECs. Scale bar = 85 μm (H) Quantification of the tube length in (G), images of tube morphology were taken in 6 random microscopic fields per sample and values displayed are means ± SEM of 6 independent experiments. Sample means were statistically significant as determined by one-way ANOVA with Bonferroni's correction. (I) Representative images of aortic rings from male C57BL/6 mice cultured in different mediums containing MAN and HG + PA in the presence or absence of bFGF or c-PTIO for 72 h. Scale bars = 350 μm. (J) Quantification of the number of sprouts in (I), one-way ANOVA with Bonferroni's correction was used; values displayed are means ± SEM of 6 independent experiments. Data shown in graphs represent the statistical significance: *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA; & p < 0.05 vs. exposed to HG + PA in combination with bFGF. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Next, the ex vivo aortic ring sprouting assay was utilized to further assess the endothelial protective effects of bFGF. Aortic rings from male C57BL/6 mice were cultured in HG + PA medium or mannitol (MAN) medium as an osmotic control in the presence or absence of bFGF and/or c-PTIO. In MAN medium, a well-structured microvessel network with clearly defined tubules and regular branching was present. By contrast, aortic rings cultured in HG + PA medium exhibited dramatically impaired sprouting function, which was preserved by bFGF but abolished by c-PTIO co-treatment (Fig. 2I and J). These observations suggested that bFGF mediated regulation of S-nitrosylation under HG + PA conditions alleviated endothelial injury.

3.3 bFGF-mediated S-nitrosylation of IKKβ and p65 suppressed chronic inflammation

NFκB is a transcription factor that plays a pivotal role in inflammation, cell survival, and cell proliferation [42], and is maintained in a latent form in the cytoplasm via sequestration by inhibitory κB (IκB) proteins. However, S-nitrosylation basally inhibits NFκB activity. Both subunits of the p50^C62-p65^C38 NFκB heterodimer are targeted for S-nitrosylation at a conserved cysteine residue in the Rel DNA binding domain, which disrupts its DNA binding [43]. In addition, the IκB-phosphorylating component of the complex, IKKβ^C179, is S-nitrosylated, which suppresses kinase activity and subsequent IκB phosphorylation [44]. Therefore, decreased S-nitrosylation under HG + PA could potentially activate NFκB, and then result to inflammation.

Based on the above prior findings, we sought to determine whether endogenous S-nitrosylation of IKKβ or NFκB was altered by HG + PA and subsequently triggered the chronic inflammation. To test this hypothesis, we measured ascorbate-dependent basal S-nitrosylation of endogenous p65 and IKKβ in HUVECs subjected to HG + PA (Supplementary Fig. 4A). HG + PA decreased p65 and IKKβ S-nitrosylation, but largely restored by bFGF or GSNO co-treatment (Supplementary Fig. 4B and C). Meanwhile, bFGF suppressed HG + PA-triggered nuclear translocation of the p65 NFκB subunit (Supplementary Fig. 4D and E) were further confirmed by immunofluorescence (Supplementary Fig. 4F). In addition, NFκB DNA binding and transcriptional activity were evaluated using an electrophoretic mobility shift assay (EMSA) (Supplementary Fig. 4G, H) and a luciferase assay, respectively (Supplementary Fig. 4I). These observations suggested that bFGF increased S-nitrosylation of p65 and IKKβ could alleviate HG + PA-induced inflammation.

To assess the detailed mechanism of bFGF mediated SNO-IKKβ and subsequent endothelial protection, HUVECs were infected with a lentiviral vector encoding LacZ (control), wild-type IKKβ (LV-IKKβ^WT), or a mutant IKKβ in which Cys-179 was replaced with the non-nitrosylatable residue serine (LV-IKKβ^C179S), and the expression was confirmed by immunoblotting (Fig. 3A). Consistent with a prior report [44], mutation of IKKβ^C179S abrogated bFGF-mediated SNO-IKKβ^C179 (Fig. 3B and C). Additionally, with the immunofluorescence staining of protein S-nitrosylation and IKKβ, we found that bFGF induced the colocalization of S-nitrosated protein (red) with IKKβ (green), which was prevented by the mutant, suggesting that SNO-IKKβ^C179S attenuated bFGF-mediated protein S-nitrosylation including SNO-IKKβ at these residues (Fig. 3D). Besides, this mutation also led to the phosphorylation and degradation of the IκB protein, and subsequently triggered nuclear localization of the p65 NFκB subunit (Fig. 3E and F) and initiation of NFκB-dependent gene transcription, as assessed by EMSA (Fig. 3G and H) and luciferase assay (Fig. 3I). By contrast, these changes were not observed in MAN conditions (Fig. 3B–I). These observations demonstrated that the IKKβ^C179S mutation abrogated bFGF restrained NFκB signaling, indicating that bFGF suppress NFκB by regulating S-nitrosylation of IKKβ.

Fig. 3

bFGF mediated SNO-IKKβ contributes to the suppression of chronic inflammation. HUVECs were infected with a lentiviral vector encoding LacZ (control), wild-type IKKβ (LV-IKKβ^WT), or a mutant IKKβ in which Cys-179 was replaced with the non-nitrosylatable residue serine (LV-IKKβ^C179S). After transfection, overexpression cells were established and selected with puromycin and cultured either in MAN or HG + PA medium in the presence or absence of bFGF and GSNO for 72 h. (A) Representative immunoblotting images of Flag confirmed that the cells were transfected with lentivirus successfully. (B) The mutation of IKKβ^C179S abrogated bFGF mediated SNO-IKKβ^C179, IKKβ protein loading was detected by anti-Flag antibody. (C) Quantification of S-nitrosylated protein levels (percentage of S-nitrosylated protein to total) in HUVECs, values displayed are means ± SEM of 4 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used, n. s. = not significant, *p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-IKKβ^WT-transfected HUVECs cultured in bFGF. (D) Fixed cells were subjected to a biotin-switch assay. S-nitrosylation was visualized using Alexa-568-conjugated avidin (red) and IKKβ was determined by immunofluorescent staining (green) in HUVECs, the merged puncta (yellow) is the IKKβ closed proximity with S-nitrosated protein or SNO-IKKβ itself, nuclei were stained with DAPI (blue), Scale bars = 200 μm. One-way ANOVA with Bonferroni's correction was used, values displayed are means ± SEM of 3 independent experiments. *p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in HG + PA; & p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs treatment with bFGF. (E) Nuclear and cytosolic extracts made from HUVECs were isolated, and overexpression of non-nitrosylatable IKKβ^C179S mutant exposure to bFGF resulted in the nuclear localization of p65. (F) Quantification of p65 localization, values displayed are means ± SEM of 4 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used, *p < 0.05. (G) EMSA assays were performed as described in “Materials and Methods”. Overexpression of non-nitrosylatable IKKβ^C179S mutant increases the DNA binding ability of p65. (H) The quantitative analysis of each immunoblots, values displayed are means ± SEM of 4 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used, *p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-IKKβ^WT-transfected HUVECs cultured in bFGF. (I) Equal amounts of cell lysates were tested for luciferase activity as described in “Materials and Methods”, values displayed are means ± SEM of 6 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used, *p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-IKKβ^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-IKKβ^WT-transfected HUVECs cultured in bFGF. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

In parallel, to evaluate the importance of SNO-p65 in bFGF restrained NFκB, HUVECs were infected with a lentiviral vector encoding LacZ (control), wild-type p65 (LV-p65^WT), or mutant p65, in which Cys-38 was replaced with the non-nitrosylatable residue serine (LV-p65^C38S), and the expression was confirmed by immunoblotting (Fig. 4A). Similarly, the p65^C38S mutation also abrogated bFGF restrained NFκB (Fig. 4B and C). Meanwhile, bFGF-induced suppression of NFκB transcriptional activity was partially abolished in LV-p65^C38S-infected HUVECs (Fig. 4G–I), without affecting the nuclear localization of p65 (Fig. 4D and E). Additionally, bFGF mediated protein S-nitrosylation (red) including SNO-p65 (partially yellow puncta) were also largely attenuated in non-nitrosylatable p65^C38S mutant HUVECs (Fig. 4F). Therefore, we ascertained that, in parallel to IKKβ suppression, bFGF-induced suppression of NFκB activity was partly attributed to the suppression of transcriptional activity via S-nitrosylation of the p65 DNA binding domain. Taken together, these findings suggested that bFGF counteracted inflammation was partially attributed to the S-nitrosylation of IKKβ^C179 and p65^C38.

Fig. 4

bFGF mediated SNO-p65 contributes to the suppression of chronic inflammation. HUVECs were infected with a lentiviral vector encoding LacZ (control), wild-type p65 (LV-p65^WT), or mutant p65, in which Cys-38 was replaced with the non-nitrosylatable residue serine (LV-p65^C38S). After transfection, overexpression cells were established and selected with G418 and cultured either in MAN or HG + PA medium in the presence or absence of bFGF and GSNO for 72 h. (A) Representative immunoblotting images of Flag confirmed that the cells were transfected with entivirus successfully. (B) The mutation of p65^C38S abrogated bFGF mediated SNO-p65^C38, p65 protein loading was detected by anti-Flag antibody. (C) Quantification of S-nitrosylated protein levels (percentage of S-nitrosylated protein to total) in HUVECs, values displayed are means ± SEM of 4 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used. n. s. = not significant, *p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-p65^WT-transfected HUVECs cultured in bFGF. (D) Nuclear and cytosolic extracts made from HUVECs were isolated, and transfection of p65WT and non-nitrosylatable p65^C38S mutant exposure to HG + PA or bFGF respectively have none impact on the nuclear translocation of p65. (E) Quantification of p65 localization, values displayed are means ± SEM of 4 independent experiments, one-way ANOVA with the Student-Newman-Keuls test was used * p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in HG + PA, n. s. = not significant. (F) Fixed cells were subjected to a biotin-switch assay. S-nitrosylation was visualized using Alexa-568-conjugated avidin (red) and p65 nuclear translocation was determined by immunofluorescent staining (green) in HUVECs, the merged puncta (yellow) is the p65 closed proximity with S-nitrosated protein or SNO-p65 itself, nucleus was blue, Scale bars = 200 μm. One-way ANOVA with Bonferroni's correction was used, values displayed are means ± SEM of 3 independent experiments. *p < 0.05 vs. LV-p65^WT transfected HUVECs cultured in HG + PA; & p < 0.05 vs. LV-p65^WT-transfected HUVECs treatment with bFGF. (G) Transfection of non-nitrosylatable p65^C38S mutant increases the DNA binding ability of p65. EMSA assays were performed as described in “Materials and Methods”. (H) The quantitative analysis of each immunoblots, values displayed are means ± SEM of 4 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-p65^WT-transfected HUVECs cultured in bFGF. (I) Equal amounts of cell lysates were tested for luciferase activity as described in “Materials and Methods”, values displayed are means ± SEM of 6 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in MAN; #p < 0.05 vs. LV-p65^WT-transfected HUVECs cultured in HG + PA; &&& p < 0.001 vs. LV-p65^WT-transfected HUVECs cultured in bFGF. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

3.4 bFGF-mediated S-nitrosylation of IKKβ^C179 and p65^C38 contributed to in vitro and in vivo angiogenic function under HG + PA conditions

We next investigated the potential link between S-nitrosylation, NFκB activity, and angiogenic functions of endothelial cells in vitro and in vivo. Given the regulation effect of SNO-IKKβ^C179 and SNO-p65^C38 on NFκB transcriptional activity, loss of S-nitrosylation at these sites could trigger chronic inflammation under HG + PA conditions in HUVECs. We thus sought to explore whether SNO-IKKβ^C179 or SNO-p65^C38 affected the angiogenic functions of endothelial cells. IKKβ^WT + p65^WT and IKKβ^C179S + p65^C38S expression were stably established in cultured HUVECs. Expression of NFκB target genes were significantly increased in IKKβ^C179S + p65^C38S mutant-overexpressing HUVECs compared with wild-type-expressing HUVECs, consistent with the notion that loss of S-nitrosylation could trigger proinflammatory cytokines such as IL-1β, IL-6, IL-8, and TNFα (Supplementary Fig. 5A). Meanwhile, HG + PA impaired tube formation activity in IKKβ^WT + p65^WT overexpressed HUVECs were alleviated by bFGF, but largely abrogated in IKKβ^C179S + p65^C38S-overexpressed HUVECs (Fig. 5A and B). Besides, the similar results were also observed by cell migration assay (Supplementary Fig. 5B; Fig. 5C).

Fig. 5

Decomposition of S-nitrosylation contributes to cell death both in vitro and in vivo. (A–C) HUVECs were infected with a lentiviral vector encoding wild-type (LV-IKKβ^WT; LV-p65^WT), or a mutant was replaced with the non-nitrosylatable residue serine (LV-IKKβ^C179S; LV-p65^C38S). After transfection, overexpression cells were established and selected with G418 and puromycin respectively. (A) Capillary-like tube formation was assessed by matrigel angiogenesis assay in HUVECs. Scale bars = 75 μm. Overexpression of non-nitrosylatable p65^C38S-IKKβ^C179S reduce tube formation by ~50% in bFGF rescued cells (B) Quantification of the tube length in (A), images of tube morphology were taken in 6 random microscopic fields per sample and values displayed are means ± SEM of 6 independent experiments. Sample means were statistically significant as determined by a One-way ANOVA with the Student-Newman-Keuls, *p < 0.05; #p < 0.05. (C) Cell migration distances were measured based on the data (Supplementary Fig. 5b), with each data point representing the mean of multiple measures within a single well in a representative experiment, values displayed are means ± SEM of 6 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05. (D) Cultured aortic rings from db/m or db/db mice were overexpressed LV-p65^WT + IKKβ^WT or its LV-p65^C38S + IKKβ^C179S mutant respectively, then cultured in different mediums containing MAN and HG + PA in the presence and absence of bFGF. Scale bars = 350 μm. (E) Quantification of the number of sprouts in (D), values displayed are means ± SEM of 6 independent experiments. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05. (F–J) db/m and db/db mice were wounded on the back skin, and wound regions were intradermally injected with p65^WT (Ad-Cdh5-p65^WT) and IKKβ^WT (Ad-Cdh5-IKKβ^WT); p65^C38S (Ad-Cdh5-p65^C38S) and IKKβ^C179S (Ad-Cdh5-IKKβ^C179S). (F) CD31⁺ capillary density was imaged with a confocal laser scanning microscope at 4th day. Scale bars = 20 μm. (G) Quantification of the number of capillary density, n = 6/group, and images were taken in 10 random microscopy fields per sample, values displayed are means ± SEM. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05. (H) The wound area size (%) was measured for 6 days (Supplementary Fig. 5c), n = 6/group, values displayed are means ± SEM. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05. (I) The wound closing rate was measured for 6 days (Supplementary Fig. 5e) and quantified; n = 6/group, values displayed are means ± SEM. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05. (J) Measurement of reepithelization ratio (leading edge ratio) in wound area (Supplementary Fig. 5D and E), n = 6/group, and images were taken in 6 paraffin-embedded tissue sections per sample, values displayed are means ± SEM. One-way ANOVA with the Student-Newman-Keuls test was used. *p < 0.05; #p < 0.05.

To further assess the endothelial protective function of bFGF-mediated IKKβ^C179 and p65^C38 S-nitrosylation, KKβ^WT + p65^WT or IKKβ^C179S + p65^C38S mutant overexpression were induced by lentiviral infection of aortic rings from db/m or db/db mice. During the experiment, the blood glucose was measured by glucose analyzer (Supplementary Table S1). Then, aortic rings were cultured in MAN or HG + PA media in the presence or absence of bFGF. A well-structured microvessel network with clearly defined tubules and regular branching was present in both IKKβ^WT + p65^WT and IKKβ^C179S + p65^C38S mutant-overexpressing aortic rings from db/m mice cultured in MAN medium (Fig. 5D and E). However, IKKβ^C179S + p65^C38S mutant overexpression significantly impaired sprouting function in rings exposed to HG + PA as compared with rings maintained in MAN, meanwhile, overexpression of mutant proteins abrogated the protective effect of bFGF in this context (Fig. 5D and E).

To further assess the roles of SNO-IKKβ^C179 and SNO-p65^C38 in bFGF-mediated endothelial protection, we utilized an in vivo skin wound healing model in T2DM db/db mice, in which chronic inflammation suppresses endothelial function and subsequent wound healing [45]. Adenoviruses expressing Ikkβ^WT + p65^WT or Ikkβ^C179S + p65^C38S mutants with the EC-specific vascular endothelial cadherin promoter (Ad-Cdh5-Ikkβ^WT + Ad-Cdh5-p65^WT or Ad-Cdh5-Ikkβ^C179S + Ad-Cdh5-p65^C38S) were delivered to wounded tissue in db/m and db/db mice through subcutaneous injection. Meanwhile, the blood glucose levels were determined. (Supplementary Table S2). We found that CD31⁺ capillary density was decreased in the regenerative skin tissue of db/db mice subcutaneously injected with Ad-Cdh5-Ikkβ^C179S + p65^C38S compared with that of db/db mice injected with Ad-Cdh5-Ikkβ^WT + p65^WT. Consistent with in vitro findings, bFGF treatment did not ameliorate CD31⁺ capillary density in Ikkβ^C179S + p65^C38S-overexpressing db/db mice compared with Ikkβ^WT + p65^WT-overexpressing db/db mice (Fig. 5F and G). In parallel, overexpression of Ikkβ^C179S + p65^C38S in endothelial cells delayed closure of the wound opening and abrogated the endothelial protective effect of bFGF in db/db mice (Supplementary Fig. 5C; Fig. 5H).

Although the gross wound size became comparable between Cdh5-Ikkβ^WT + p65^WT and Cdh5-Ikkβ^C179S + p65^C38S-overexpressing db/db mice on day 4 after wounding (Supplementary Fig. 5C; Fig. 5H), re-epithelialization was significantly decreased in Ikkβ^C179S + p65^C38S-overexpressing db/db mice relative to Ikkβ^WT + p65^WT-overexpressing db/db mice on day 6 (Supplementary Fig. 5D and E; Fig. 5J). Importantly, bFGF treatment unable to improve wound re-epithelialization in Ikkβ^C179S + p65^C38S-overexpressing db/db mice (Supplementary Fig. 5E; Fig. 5J). Additionally, histology analysis of wounds on day 6 after wounding revealed delayed wound closure and compromised re-epithelization in wounds from db/db mice overexpressing Ikkβ^C179S + p65^C38S compared with those from the Ikkβ^WT + p65^WT-overexpressing mice, which was particularly evident in the bFGF-treated group (Supplementary Fig. 5E; Fig. 5I and J). Taken together, these findings indicated that bFGF-mediated SNO-IKKβ^C179 and SNO-p65^C38 suppressed NFκB activity and dramatically alleviated T2DM-associated endothelial dysfunction.

3.5 bFGF induced eNOS-dependent IKKβ and p65 S-nitrosylation through the Akt/AMPK axis

Our earlier finding demonstrated that HG + PA-induced iNOS regulated S-nitrosylation during late-stage HG + PA exposure, while decreased S-nitrosylation during early-stage HG + PA exposure was iNOS-independent (Fig. 1, Supplementary Fig. 1). Therefore, to investigate the contribution of eNOS-derived NO to bFGF-mediated IKKβ and p65 S-nitrosylation at early-stage of HG + PA exposure, eNOS gene expression was knocked down with an adenovirus harboring eNOS-shRNA in HUVECs. eNOS knockdown abolished bFGF-induced IKKβ and p65 S-nitrosylation in HUVECs (Fig. 6A–C). Furthermore, to determine whether the effect of bFGF was eNOS-mediated, HUVECs were treated with the classical eNOS upstream pathway antagonist LY294002 (LY) or AMPK inhibitor Compound C (CC), respectively. Consistently, LY and CC decreased Akt/GSK3β and AMPK phosphorylation, respectively. Meanwhile, both treatments decreased eNOS activity in HUVECs following HG + PA treatment (Fig. 6D–J). Moreover, the effects of bFGF on eNOS activity were significantly abolished by both LY and CC treatment, suggesting that bFGF enhanced eNOS activity was dependent on these pathways (Fig. 6D–J). Together, these results suggested that bFGF-mediated S-nitrosylation of IKKβ and p65 could be due in part to eNOS up-regulation and activation through the Akt and AMPK pathways.

Fig. 6

bFGF induces eNOS-dependently S-nitrosylation of IKKβ and p65 through the Akt and Ampk pathway. (A) HUVECs were transfected with adenoviruses harboring human sh-eNOS (Ad-sh-eNOS) and a scrambled sequence (Ad-scramble) respectively. After transfection, HUVECs were cultured either in MAN or HG + PA medium in the presence or absence of bFGF for 72 h. Whole-cell lysates were subjected to the biotin switch assay as described in “Materials and Methods”. In the absence of eNOS abolished bFGF induced S-nitrosylation of IKKβ and p65 in HUVECs. (B–C) Quantification of S-nitrosylated protein levels (percentage of S-nitrosylated protein to total) in HUVECs, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF. (D–E) HUVECs were pretreated with or without LY294002 (20 μM), compound C (10 μM) for 2 h and then exposed to MAN and HG + PA for 72 h in the presence or absence of bFGF. (F–J) Quantitative analysis of phosphorylation of Akt, GSK3β, AMPK and eNOS protein, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. HG + PA; #p < 0.05 vs. HG + PA coincubated with bFGF.

3.6 bFGF-mediated IKKβ and p65 S-nitrosylation suppressed inflammation through a txnip-dependent mechanism

The mechanism underlying suppression of inflammation via bFGF-mediated S-nitrosylation was further explored. In fact, S-nitrosylation has emerged as a dominant regulatory mechanism of GSNOR and Trx-associated signal transduction [7,46]. Therefore, we determined whether alteration of GSNOR was involved in the S-nitrosylation of IKKβ and p65, and subsequent suppression of inflammation and increased vasculogenesis. We found that bFGF treatment did not alter the expression of GSNOR in HUVECs exposed to HG + PA (Fig. 7A and B), and these findings were corroborated by whole-cell quantitative RT-PCR (qRT-PCR) analysis (Fig. 7I). In contrast, although the GSNOR activity was not significantly affected in early-stage of HG + PA exposure, it experienced a slowly decline activity in a time-dependent manner after day 3 (Supplementary Fig. 6A; Fig. 7C), indicating that bFGF-mediated IKKβ and p65 S-nitrosylation was independent of GSNOR.

Fig. 7

bFGF-mediated S-nitrosylation of IKKβ and p65 through the Trx rather than GSNOR. (A–I) HUVECs were exposed to NG and HG + PA in the presence or absence of bFGF for 72 h. (A) Anti-GSNOR immunoblot analysis from whole-cell lysates confirms that bFGF treatment could not impact the expression of GSNOR proteins. (B) The quantitative analysis of GSNOR immunoblots, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, n. s. = not significant. (C) The mean GSNOR enzymatic activity in the HUVECs was no significant change at 3rd day; values displayed are means ± SEM of 6 independent experiments. Wilcoxon rank-sum test was used, n. s. = not significant. (D) Anti-Trx, TrxR and Txnip immunoblot analysis from whole-cell lysates confirms that bFGF treatment also could not impact the expression of Trx and TrxR proteins but bFGF treatment decrease the protein level of the Txnip. (E–G) The quantitative analysis of Trx, TrxR and Txnip immunoblots, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, n. s. = not significant, *p < 0.05 vs. NG or MAN; #p < 0.05 vs. HG + PA. (H) The mean Trx enzymatic activity in the HUVECs. Values displayed are means ± SEM of 6 independent experiments. Wilcoxon rank-sum test was used, **p < 0.01 vs. NG or MAN; #p < 0.05 vs. HG + PA. (I) mRNA expression of GSNOR, Trx and TrxR genes. Values displayed are means ± SEM of 5 independent experiments. Two-tailed Student's t-test was used, n. s. = not significant, *p < 0.05 vs. NG or MAN; #p < 0.05 vs. HG + PA. (J) HUVECs were transfected with adenoviruses harboring human sh-Trx (Ad-sh-Trx) and a scrambled sequence (Ad-scramble) respectively. After transfection, HUVECs were cultured either in MAN or HG + PA medium in the presence or absence of bFGF for 72 h. Whole-cell lysates were subjected to the biotin switch assay as described in “Materials and Methods”. (K–L) Quantification of S-nitrosylated protein levels (percentage of S-nitrosylated protein to total) in HUVECs, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF.

It is noteworthy that Trx modulates protein S-nitrosylation homeostasis, and acting as a double edge sword for the regulation of the cell fate. Forrester et al. reported that Trx-mediated protein denitrosylation is augmented by endogenously derived NO via repression of Txnip, which in turn protects against iNOS-mediated cell death [8]. On the contrary, it has been recently reported that Trx is essential for maintaining S-nitrosylation in endothelial cells [9], and can transfers ⁺NO to caspase-3, which leads to inhibition of apoptosis. To address this discrepancy, we determined whether Trx was involved in bFGF-mediated IKKβ and p65 S-nitrosylation in HUVECs. Similarly, we found that bFGF treatment did not affect Trx protein and mRNA levels in HUVECs exposed to HG + PA (Fig. 7D–F and 7I). The activity of Trx also decreased in a time-dependent manner during early-stage HG + PA exposure, and subsequently recovered and increased by day 5 (Supplementary Fig. 6B). Interestingly, the declined activity of Trx at day 3 was significantly ameliorated by bFGF treatment (Fig. 7H). To investigate the contribution of Trx regulation to bFGF-mediated IKKβ and p65 S-nitrosylation, Trx gene expression was knocked down with an adenovirus harboring Trx-shRNA in HUVECs. Trx ablation partially abolished bFGF-regulated IKKβ and p65 S-nitrosylation in HUVECs (Fig. 7J–L). Additionally, protein levels of other major redox enzyme protein systems (Supplementary Fig. 6C–E) and major NOS isoforms (Supplementary Fig. 6C, F–G) were unaffected by silencing of Trx in HUVECs. Nevertheless, Trx deficiency abolished itself activity (Supplementary Fig. 6H), and partially weakening bFGF promotion of NO availability, demonstrating that the promotion effect of Trx on NO availability was mainly attribute to ROS neutralization and cellular redox homeostasis maintenance during the window of the initial of inflammation, which in turn contributed to bFGF-mediated IKKβ and p65 S-nitrosylation subsequently.

On the other hand, to eliminate the potential function exerts by the interaction of Trx with p65 and IKKβ, which possible to regulate the S-nitrosylation state of p65 and IKKβ contributed by bFGF treatment, immunoprecipitation was performed. We found that the interaction between Trx and P65 were slightly decreased by HG + PA treatment at day 3, even in the presence of bFGF. While, totally no interaction was detected between Trx and IKKβ (Supplementary Fig. 7A). Taken together, these data indicated that Trx was essential for maintaining the homeostasis of S-nitrosylated molecules in endothelial cells, and bFGF-mediated SNO-IKKβ^C179 and SNO-p65^C38 through a mechanism attribute to the alteration of Trx activity rather than GSNOR.

Meanwhile, thioredoxin-interacting protein (Txnip) has originally been described as a negative Trx regulator that represses Trx activity through protein–protein interaction [47]. Therefore, we investigated whether Txnip was involved in bFGF-mediated IKKβ and p65 S-nitrosylation. bFGF significantly decreased Txnip protein level in HUVECs treated with HG + PA (Fig. 7D, G). Thereafter, we evaluated whether bFGF-mediated Txnip alteration contributed to the interaction between Txnip and Trx. Immunoprecipitation analysis revealed that bFGF decreased the association of Txnip with Trx (Fig. 8A and B). However, forced overexpression of Txnip abolished bFGF-mediated dissociation of Txnip and Trx (Fig. 8A and B). In addition, Txnip overexpression weakened Trx activity (Fig. 8C) and bioactivity of NO, and thus partly abolished bFGF-mediated IKKβ and p65 S-nitrosylation (Fig. 8D–F). Overall, these results suggested that bFGF-mediated IKKβ, and p65 S-nitrosylation was attributed at least in part through the suppression of Txnip, which promoted disassociation of Txnip and Trx, subsequently maintaining the content of S-nitrosylated molecules in endothelial cells.

Fig. 8

bFGF mediated S-nitrosylation of IKKβ and p65 regulates anti-inflammation through a mechanism involving Txnip. (A–F) HUVECs were transfected with adenoviruses harboring human Txnip (Ad-Txnip) and Ad-LacZ (served as a control) respectively. After transfection, HUVECs were cultured either in MAN or HG + PA medium in the presence or absence of bFGF for 72 h. (A) The cell lysates were subjected to immunoprecipitation with Trx antibody, followed by immunoblotting with the indicated antibodies. Cell lysates were also subjected to immunoprecipitation with IgG as negative control. Quantification of interaction between Trx and Txnip in HUVECs (IP: Trx, IB: Txnip), values displayed are means ± SEM of 3 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF (right panel). (B) The cell lysates were subjected to immunoprecipitation with Txnip antibody, followed by immunoblotting with the indicated antibodies. Cell lysates were also subjected to immunoprecipitation with IgG as negative control. Quantification of interaction between Txnip and Trx in HUVECs (IP: Txnip, IB: Trx), values displayed are means ± SEM of 3 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF (left panel). (C) Overexpression of Txnip weaken the activity of Trx, values displayed are means ± SEM of 6 independent experiments. Wilcoxon rank-sum test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF. (D) Whole-cell lysates were subjected to the biotin switch assay as described in “Materials and Methods”. (E–F) Quantification of S-nitrosylated protein levels (percentage of S-nitrosylated protein to total) in HUVECs, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF. (G) HUVECs were transfected with adenoviruses harboring human sh-eNOS (Ad-sh-eNOS) and a scrambled sequence (Ad-scramble) respectively. After transfection, HUVECs were cultured either in MAN or HG + PA medium in the presence or absence of bFGF or GSNO for 72 h. Related protein immunoblot analysis from whole-cell lysates. (H–K) The quantitative analysis of related protein immunoblots, values displayed are means ± SEM of 4 independent experiments. Two-tailed Student's t-test was used, n. s. = not significant. (L) Deficiency of eNOS partly abolished the bFGF promoted Trx activity. Values displayed are means ± SEM of 6 independent experiments. Wilcoxon rank-sum test was used, *p < 0.05 vs. MAN; #p < 0.05 vs. HG + PA coincubated with bFGF.

Although endogenously derived NO repressed Txnip expression [8], and bFGF promoted eNOS activation via the Akt/AMPK axis (Fig. 6D–J), bFGF also decreased expression of Txnip in eNOS knockdown HUVECs (Fig. 8G and H). Further results showed that GSNO stimulation did not affect hyperglycemia enhanced promoter activity of Txnip (Fig. 8G and H), demonstrated that bFGF affected the expression of Txnip in a NO-independent manner. Besides, other major redox enzyme protein systems (Fig. 8G, I–K) were also not affected by bFGF treatment with silencing of eNOS in HUVECs, but partly abolished bFGF-promoted Trx activity (Fig. 8L). In addition, we found that HG + PA-mediated dissociation of Trx and eNOS were reversed by bFGF (Supplementary Fig. 7B), and bFGF mediated S-nitrosylation of Trx was eNOS dependent in HUVECs (Supplementary Fig. 7C and D). Taken together, these findings suggested that bFGF-mediated IKKβ and p65 S-nitrosylation were mainly attributed to synergy between eNOS and Trx activity.