Danshenol A inhibits TNF-α-induced expression of intercellular adhesion molecule-1 (ICAM-1) mediated by NOX4 in endothelial cells

ICAM-1 overexpression and subsequent adhesion of leukocytes to endothelial cells play critical roles in the early stage of atherosclerosis. Danshenol A (DA) is an abietane-type diterpenoid isolated from traditional Chinese herb Salvia miltiorrhiza Bunge. The mechanisms under its regulation of adhesion of molecular expression are explored. Here, the effect of DA on TNF-α-induced ICAM-1 expression was investigated in endothelial cells. TNF-α-induced ICAM-1 expression and subsequent adhesion of monocytes, as well as elevated reactive oxygen species (ROS) generation and NOX4 expression were all significantly reversed by DA, siNOX4 and NOX4 inhibitor GKT137831. Furthermore, TNF-α-induced ICAM-1 expression, which was increased via IKKβ/IκBα-mediated activation of NF-κB p65, was also inhibited by DA. Interestingly, NOX4 overexpression suppressed the ICAM-1 expression, and this finding may be ascribed to the activation of Nrf-2. Additionally, NF-κB inhibitor PDTC, siNOX4, or DA can decrease the TNF-α-induced ICAM-1 expression and suppress the adhesion of monocytes. In all, DA inhibited TNF-α-induced ICAM-1 expression and subsequent monocyte adhesion to endothelial cells through the NOX4-dependent IKKβ/NF-κB pathway. Besides, NOX4 played dual role in regulating ICAM-1 expression via diverse signal pathway. This novel bioactivity will make DA a good candidate to be further explored for therapeutic or preventive application for atherosclerosis.

findings suggested that NOX4 might also be vascular protective. NOX4 overexpression promotes proliferation and migration of endothelial cells but reduces serum-deprivation-induced apoptosis in human microvascular endothelial cells 18 . Furthermore, transgenic mice with endothelium-specific NOX4 overexpression exhibited enhanced vasodilation and reduced blood pressure 19 . More recently, endothelial NOX4 demonstrated protection of ApoE −/− mice from atherosclerotic lesions 20 .
Several lines of evidence have shown that the expression of TNF-α-induced molecules is regulated at the gene level by activation of transcription factors. These transcription factors, including Nrf-1, Nrf-2, and NF-κB p65, are redox sensitive and identified to be involved in the regulation of the gene expression of leukocyte-adhesion molecules and cytokines [21][22][23] .
Salvia miltiorrhiza Bunge (Danshen) is a traditional Chinese herb with a lot of properties that could improve our health. In pharmaceutical field, it has been a continuous source of therapeutic agents for cerebrovascular disorders while in food field, danshen is used as food additives to make herbal tea, soups, seasoner and beverages. Danshen is currently going through trials and may even become the first traditional Chinese remedy to gain approval by the U.S. Food and Drug Administration (FDA) 24 .
Danshenol A (Fig. 1A) is an abietane-type diterpenoid isolated from Danshen in 1997 25 . Compared with well-known tanshinones such as tanshinone IIA, tanshinone I, and cryptotanshinone, its biological effects have been limited reported. Recently, studies focused on DA are paid more and more attentions. One paper reported that DA has a strong inhibitory effect on aldose reductase (AR) 25 . Also some study showed that DA has anti-inflammatory properties superior to those of tanshinone IIA 26 hinting that DA has potential for anti-atherosclerosis. One early phase of atherosclerosis involves the recruitment of inflammatory cells from the circulation and their transendothelial migration. This process is predominantly mediated by cellular adhesion molecules 27 .
ICAM-1, one of the largest families of cell adhesion molecules plays an important role in the process of atherosclerosis 28 . In the present study, we investigated the effects and potential mechanisms of DA on the TNF-α-induced ICAM-1 expression in human umbilical vein endothelial cells (HUVECs). We found that DA inhibited the TNF-α-induced ICAM-1 expression by suppressing the ROS-generating NOX4. Furthermore, NOX4 played a dual role in regulating the ICAM-1 expression induced by TNF-α.

Results
DA inhibited TNF-α-induced ICAM-1 expression and subsequent monocyte adhesion. We first characterized the effects of TNF-α on ICAM-1 expression in endothelial cells. TNF-α significantly increased the ICAM-1 protein expression in a concentration-and time-dependent manner ( Fig. 1B and C). The mRNA expression of ICAM-1 was also upregulated by TNF-α (Fig. 1D). Immunofluorescene assay showed that significant red fluorescence was observed in the TNF-α-treated endothelial membrane (Fig. 1E). Moreover, TNF-α treatment considerably increased the adhesion of monocytes to endothelial cells, but this adhesion was significantly inhibited by the anti-ICAM-1 antibody (Fig. 1F). The cytotoxicity of DA to endothelial cells was then evaluated, and low concentration of DA (100 nM) showed no cytotoxicity ( Fig. 2A). To minimize the cytotoxicity and test the efficacy of DA, 10 nM DA was selected in the following study. DA (10 nM) alone showed no effect on the ICAM-1 expression at both mRNA and protein levels ( Fig. 2B and C). Furthermore, TNF-α induced all the alterations mentioned above (Fig. 1), which were significantly reversed by DA or anti-ICAM-1 antibody pretreatment (Fig. 2D-G). . Cells were treated with TNF-α, and the ICAM-1 protein expression was detected by Western blot (B and C). Cells were treated with TNF-α (10 ng/mL) for 24 h; ICAM-1 mRNA and protein expression were detected by real-time PCR (D) and immunofluorescence staining (E), respectively. Cells were treated with TNF-α (10 ng/mL) for 24 h, and the adhesion of monocytes to endothelial cells was determined with or without ICAM-1 antibody (F). Data were presented as mean ± SD of three independent experiments.
DA restored TNF-α-induced redox imbalance by inhibiting TNF-α-induced NADPH oxidase activity. Compared with the control, TNF-α induced approximately three fold ROS generation, which was almost completely inhibited by DA or NAC (Fig. 3A). Similar to NAC, DA significantly suppressed Figure 2. DA inhibited TNF-α-induced ICAM-1 expression and monocyte-HUVEC adhesion. Cells were treated with DA, and cytotoxicity was determined by MTT assay (A). Cells were treated with DA (10 nM) for 24 h; ICAM-1 expression was detected by Western blot (B) and RT-PCR (C). Cells were treated with TNF-α (10 ng/mL) for 24 h in the absence or presence of DA (10 nM) pretreatment for 1 h; ICAM-1 protein and mRNA expression was detected by Western blot (D), immunofluorescence staining (F), and RT-PCR (E). The effect of DA on the adhesion of monocytes to endothelial cells was determined (G). Data were presented as mean ± SD of three independent experiments. NS indicates no significance.
TNF-α-induced H 2 O 2 production whereas DA had almost no effect on TNF-α-induced O2 •− production ( Fig. 3B and C). Additionally, TNF-α-induced decrease of the GSH/GSSG ratio was significantly reversed by DA and NAC (Fig. 3D). However, DA showed no direct free radical scavenging capacity with DPPH (Fig. 3E). To explore the specific sources of ROS generation induced by TNF-α, several inhibitors were applied. As shown in Fig. 3F, TNF-α-induced ROS production was significantly inhibited by DPI, a NADPH oxidase inhibitor. By contrast, ALL, Rot, TTFA, NDGA, and AA showed no obvious effects.

NOX4 knockdown inhibited TNF-α-induced ICAM-1 expression.
To further confirm the role of NOX4 in the TNF-α-induced ICAM-1 expression, NOX4 expression was decreased by siNOX4 or inhibitor GKT137831 ( Fig. 5A and B). Then we found that ROS production ( Fig. 5C) in response to TNF-α was significantly reduced in both siNOX4 and GKT137831 groups. Furthermore, TNF-α-induced expression of ICAM-1 levels was also decreased accompanied with decreased NOX4 (Fig. 5D). To investigate the direct inhibition of DA with NOX4, docking simulation studies were carried out. The crystal structure of NOX4 is not available, the 3D structures of NOX4 was constructed by homology modeling and the NOX4 model was displayed in Fig. 5E and F based on the X-ray crystal structure of NADPH binding domain of gp91(PDB3A1F) and the amino acid sequence of NOX4 (NP_056575.1). The ramachandran plot of NOX4 model suggested the good quality of this model. Then, we docked DA to NOX4. The docking score is −5.77 kcal/mol. DA bound at the entrance of the active site and make contacts with the catalytic residues (Tyr497, Gln495, Lys511 and Leu529). DA also developed a H-π interaction with residue Tyr497 and a hydrogen bond with the residue Gln495 ( Fig. 5G and K). Docking simulation studies indicates that DA interact with NOX4 through hydrogen bonds interactions.

Discussion
In this study, we investigated the inhibitory effect of DA on TNF-α-induced ICAM-1 expression in endothelial cells. The results demonstrated the following: (1) DA inhibited TNF-α-induced ICAM-1 expression; (2) ROS-dependent NF-κB signaling played an important role in this process; and (3) NOX4 served a dual role in regulating the TNF-α-induced ICAM-1 expression.
Consistent with previous reports 34, 35 , our results showed that TNF-α was a potent ICAM-1 inducer in endothelial cells at both protein and transcriptional levels. Immunofluorescence results suggested that TNF-α induced the membrane expression of ICAM-1. The inhibitory effect of anti-ICAM-1 on the TNF-α-induced adhesion of monocytes to endothelial cells showed that ICAM-1 plays an important role in cell adhesion.
During ROS generation, a series of very small and highly reactive molecules, including hydroxyl radical (·OH), O 2 •− , H 2 O 2 , and peroxynitrite, act as signaling molecules to mediate various biological responses 36 . TNF-α-induced endothelial ROS production has been well established, but the detailed mechanisms remain unclear. In the present study, we found that TNF-α-induced ROS included both H 2 O 2 and O 2 •− , which can be inhibited by both NAC and DA. Furthermore, DA showed no effect on DPPH free radicals. GSH comprises important intracellular antioxidants and antioxidative enzymes that help maintain the cellular redox homeostasis and protect cells from oxidative stress by directly scavenging ROS 37 . Interestingly, DA partly restored the TNF-α-induced decrease of GSH/GSSG ratio. Collectively, these data suggest that DA inhibited endothelial ROS formation. Accumulated data also suggest several sources of ROS involved in TNF-α-treated endothelial cells: NADPH oxidase, mitochondria, XO, and eNOS 38 . Under our experimental conditions, increased ROS was observed after TNF-α treatment, which was significantly suppressed by ROS scavenger NAC, NADPH oxidase inhibitor DPI, and DA. However, the inhibitors for the mitochondrial respiratory chain complex I-III, XO, and LOX showed no effects, suggesting that NADPH oxidase played an important role in this process. Considering  GKT137831 (A and B); After NOX4 was silenced, cells were treated with TNF-α (10 ng/mL) and ROS generation was detected by flow cytometry (C) and ICAM-1 was determined by Western blot (D); The constructed model of NOX4 (E); Ramachandran plot for NOX4 (F); Dark green dots represent the residues in most favored regions, yellow dots represent the residues in allowed regions; The ligand interaction of DA with NOX4 (G); The binding model of DA with NOX4 (K). The ligand is colored in blue, and the surrounding residues in the binding pockets are colored in orange. Data were presented as mean ± SD of three independent experiments. that endothelial cells expressed several NOXs members 17,30,39 , the expression of NOX1, NOX2, and NOX4 was determined. The TNF-α-induced expression of both NOX2 and NOX4 contributed to the increased production of H 2 O 2 and O 2 •− . However, DA significantly inhibited the expression of NOX4 without affecting NOX2. Thus, the inhibitory effect of DA on H 2 O 2 and O 2 •− might be merely ascribed to its effect on NOX4. The small transmembrane protein p22phox serves a scaffold function for the maturation and folding of the catalytically active NOX protein through a proline-rich region 40 . TNF-α induced the membrane expression of p22phox, which was suppressed by DA. Collectively, these results suggest that TNF-α induced NOX4 activation in HUVECs.
The role of NOX4 was further investigated. Jung et al. 41 reported that blockade of TNF-α-induced ROS by NOX4 siRNA suppressed the TNF-α-induced ICAM-1 expression and subsequent monocyte adhesion to human aortic endothelial cells. Also our previous study showed that NOX4 plays an important role in the process of atherosclerosis 42 . In the current research, we found that the knockdown of NOX4 by siRNA significantly inhibited the TNF-α-induced ROS production and ICAM-1 expression at both protein and mRNA levels, as well as the THP-1-HUVEC adhesions. Interestingly, the inducible expression of ICAM-1 in response to TNF-α was abolished in NOX4-overexpressed endothelial cells, although the generation of ROS was enhanced. Furthermore, TNF-α showed no obvious effect on Nrf-2 nuclear translocation in normal HUVECs but increased its nuclear Figure 6. DA inhibited NF-κB pathways in endothelial cells. Cells were treated with TNF-α (10 ng/mL) for 1 h with or without DA (10 nM) pretreatment; the expression of Nrf-1, Nrf-2, and NF-κB p65 in the nucleus was detected by Western blot (A) and immunofluorescence staining (B). The expression of p-IκBα, IκBα, p-IKKβ, and IKKβ was also determined by Western blot (C). The interaction of IκBα and IKKβ was determined by coimmunoprecipitation assay (D). Cells were treated with TNF-α (10 ng/mL) for 1 h in the absence or presence of NAC (5 mM) for 1 h; the expression of IKKβ was detected by Western blot (E). Data were presented as mean ± SD of three independent experiments. NS indicates no significance. translocation in NOX4-overexpressed HUVECs. This finding suggests that TNF-α activated the Nrf-2 pathways when NOX4 was overexpressed. Along with the role of NF-κB (discussed below), these results suggest that NOX4 played dual roles in the TNF-α-induced ICAM-1 expression mediated by NF-κB and Nrf-2.
Transcription factors such as Nrf-1, Nrf-2, AP-1, and NF-κB have been reported to play important roles in TNF-α-induced ICAM-1 expression 43,44 . Most of these transcription factors are redox sensitive and can be activated by ROS 7,45 . Hence, the protein expression of NF-κB p65, Nrf-1, and Nrf-2 in the nucleus after TNF-α treatment was examined. Only the nuclear expression and nuclear translocation of NF-κB p65 were enhanced by TNF-α, and this upregulation was reversed by DA. IKKβ, whose activation results in IκBα phosphorylation, plays a critical role in the activation of NF-κB in the canonical pathway. The TNF-α-induced phosphorylation of both IKKβ and IκBα and the co-immunoprecipitation results suggest that their interactions were also enhanced. DA pretreatment partly reversed these alterations. Thus, these results suggest that TNF-α activated the NF-κB pathway by regulating IKKβ and IκBα. The inhibitory effect of DA suggests potential targets for this compound in the NF-κB pathway. The inhibitory effect of PDTC, an inhibitor of NF-κB, on the TNF-α-induced ICAM-1 expression and adhesion of THP-1 and HUVECs further confirmed the important role of NF-κB in the TNF-α-induced ICAM-1 expression.
In conclusion, as depicted in Fig. 9, our results showed that DA, a natural product, inhibited TNF-α-induced ICAM-1 expression in endothelial cells. This inhibition was mediated by regulating the NOX4-dependent ROS generation and NF-κB activation. Furthermore, the results suggest that NOX4 played dual roles in regulating TNF-α-induced endothelial ICAM-1 expression, and the different expression statuses or patterns demonstrated distinct responses to TNF-α insult.

Materials and Methods
Chemicals, reagents, and antibodies. DA   MTT assay. HUVECs were planted into 96-well plates (1 × 10 5 cells/well) and treated with DA in series of concentrations. Cell viability was measured after 24 h of incubation followed by addition of 20 µL of MTT (5 mg/ mL). After 4 h, the MTT-containing medium was slightly removed, and DMSO (100 μL) was added to solubilize the formazan. Absorbance at 570 nm was recorded using a Spectra Max M5 Microplate Reader.  Real-time PCR. Total RNA was extracted using TRIzol reagent. Quantitative real-time PCR (RT-PCR) was performed using SYBR Green PCR reagents (Applied Biosystems) with forward and reverse primers for ICAM-1 and GAPDH. The specific primers for ICAM-1 were 5′-GGCTGGAGCTGTTTGAGAAC-3′ (forward) and 5′-ACTGTGGG GTTCAACCTCT G-3′ (reverse). For GAPDH, the primers were 5′-CGAGATCCC TCCAAAATCAA-3′ (forward) and 5′-TTCACACCCATGGACGAACAT-3′ (reverse). Briefly, 1 μL of cDNA from the RT reaction was added to 10 μL of the RT-PCR mixture containing 5 μL of Master Mix and 0.2 μM forward and reverse primers. The samples were incubated at 95 °C for 2 min, followed by 40 cycles of reaction at 95 °C for 15 s and combined annealing step at 60 °C for 1 min. The samples were assessed by 2 −ΔΔCt relative quantitative analysis to determine the expression differences.
After TNF-α treatment (with and without DA pretreatment), the slides were fixed with 4% PFA for 30 min. The slides were then permeabilized with PBST (containing 0.1% Triton x-100 in PBS solution) and blocked with PBS (containing 4% BSA in PBS solution). After incubation with primary antibody (1:1000) and secondary antibody  Determination of GSH. HUVECs (5 × 10 6 cells/well) in six-well plates were incubated with TNF-α for 1 h after DA pretreatment. GSH and GSSG contents were determined using commercial kits in accordance with the manufacturers' protocols.
DPPH assay. To evaluate the direct free radical scavenging effect of DA in a cell-free system, DPPH assay was performed as previously described 46 . Briefly, a series of DA concentrations (0.01 μM to 1 μM) was incubated with methanol solution of DPPH (5.0 × 10 −4 M) and kept in the dark for 20 min. The absorbance of the samples was measured using a Spectra Max M5 Microplate Reader at 517 nm.
Western blot analysis. The total proteins of the treated cells were collected. After quantitative determination of protein content, 30 μg of proteins was subjected to SDS-PAGE and then transferred onto PVDF membranes. After blocking with 5% nonfat milk in TBST (20 mM Tris-HCl, 500 mM NaCl, and 0.1% Tween-20) at room temperature for 1 h, the membranes were incubated with specific primary antibodies and secondary antibodies. Protein-antibody complexes were detected using an ECL Advanced Western Blot detection kit. The intensity of the bands was quantitated using Quantity One software (Bio-Rad).
Diluted siRNA and Lipofectamine TM 3000 were mixed for 30 min at room temperature. Afterward, 200 μL of siRNA-lipofectamine complexes and 800 μL of Opti-MEM reduced serum medium were added to corresponding wells. The complexes were removed after incubation for 5 h, and cells were cultured with completed medium. The siRNA sequence for NOX4 was sense 5′-ACUAUGAUAUCUUCUGGUA-3′.
For each transfection, cDNA (3 μg) was dissolved in 82 μL of nucleofector solution and then added with 18 μL of supplemented P5 primary cell solution. Approximately 1 × 10 7 to 1 × 10 8 cells were resuspended in 100 μL of cDNA plus P5 primary cell solution and immediately electroporated in a 4D-Nucleofector X Kit L cuvette. Electroporated cells were transferred to six-well plates containing 2.5 mL of warm complete medium. Plates were cultured at 37 °C in a humidified atmosphere of 5% CO 2 . After 24 h, the medium was replaced with fresh medium at daily intervals.
Immunoprecipitation. After determination of the protein concentrations, the cell extract was incubated with anti-IKKβ antibody (2 μg) for 2 h at 4 °C and then incubated with 20 μL of protein A/G plus-agarose beads overnight with constant shaking. Afterward, the beads were washed thrice with ice-cold radio immunoprecipitation assay (RIPA) buffer. The bound protein was extracted by adding 40 μL of 2 × SDS sample buffer and boiling for 5 min. The complexes were subjected to SDS-PAGE followed by Western blot.