Novel interaction of antioxidant-1 with TRAF4: role in inflammatory responses in endothelial cells

Abstract

NADPH oxidase (NOX)-derived reactive oxygen species (ROS) and copper (Cu), an essential micronutrient, have been implicated in vascular inflammatory diseases. We reported that in proinflammatory cytokine TNF-α-stimulated endothelial cells (ECs), cytosolic Cu chaperone antioxidant-1 (Atox1) functions as a Cu-dependent transcription factor for the NOX organizer p47phox, thereby increasing ROS-dependent inflammatory gene expression. However, the role and mechanism of Atox1 nuclear translocation in inflamed ECs remain unclear. Using enface staining and nuclear fractionation, here we show that Atox1 was localized in the nucleus in inflamed aortas from ApoE^−/− mice with angiotensin II infusion on a high-fat diet, while it was found in cytosol in those from control mice. In cultured human ECs, TNF-α stimulation promoted Atox1 nuclear translocation within 15 min, which was associated with Atox1 binding to TNF-α receptor-associated factor 4 (TRAF4) in a Cu-dependent manner. TRAF4 depletion by siRNA significantly inhibited Atox1 nuclear translocation, p47phox expression, and ROS production as well as its downstream VCAM1/ICAM1 expression and monocyte adhesion to inflamed ECs, which were rescued by overexpression of nuclear targeted Atox1. Furthermore, Atox1 colocalized with TRAF4 at the nucleus in TNF-α-stimulated inflamed ECs and vessels. In summary, Cu-dependent Atox1 binding to TRAF4 plays an important role in Atox1 nuclear translocation and ROS-dependent inflammatory responses in TNF-α-stimulated ECs. Thus the Atox1-TRAF4 axis is a novel therapeutic target for vascular inflammatory disease such as atherosclerosis.

INTRODUCTION

Reactive oxygen species (ROS) derived from NADPH oxidase (NOX) activated by proinflammatory and proatherogenic cytokines including TNF-α induce redox-sensitive inflammatory gene expression in endothelial cells (ECs), which contributes to vascular inflammatory diseases such as atherosclerosis (1, 18, 21). In addition, copper (Cu), an essential micronutrient, plays an important role in inflammation-dependent pathophysiologies (27, 35, 36). Cu levels are increased in atherosclerotic lesions (33) while Cu chelators inhibit vascular inflammation and atherosclerotic lesion development in ApoE^−/− mice (34–36). The mechanisms by which Cu stimulates inflammatory responses and its linkage with NOX activation remain poorly understood.

Since excess Cu is toxic, bioavailability of intracellular Cu is tightly controlled by transport proteins and Cu chaperones (6, 16). Antioxidant-1 (Atox1) is classically known as a cytosolic metallochaperone that obtains Cu from the Cu importer CTR1 and delivers Cu to the Cu transporter ATP7A at trans-Golgi network to transport Cu to the secretory Cu-dependent enzymes such as lysyl oxidase (LOX) and extracellular SOD (SOD3) or exports excess Cu to maintain intracellular Cu level (6, 7, 16, 29). Atox1 has conserved the NH₂-terminal Cu-binding motif (MTCXXC) and COOH-terminal nuclear localization signal (NLS; 2 conserved Lys56 and Lys60) (7, 12, 31). We reported that Atox1 is increased in the intima of atherosclerotic lesions in ApoE^−/− mice and that Atox1 is localized in the nucleus in pathological conditions such as hypertension and atherosclerotic vessels (12–14). Furthermore, we demonstrated that Atox1 functions as a Cu-dependent transcription factor for cyclin D1 to promote Cu-induced cell growth by binding to DNA at the unique cis element (GAAAGA; Atox1-RE) and inducing transactivation through the Cu-binding domain (CBD) and NLS motif of Atox1 in fibroblast (12). Moreover, we reported that Atox1 is translocated from cytosol to the nucleus to increase NOX organizer p47phox transcription, resulting in ROS/NF-κB-dependent inflammatory gene expression in cultured ECs in response to TNF-α (3, 4). However, the molecular mechanism of Atox1 nuclear translocation in inflamed ECs is entirely unknown.

TNF receptor-associated factors (TRAFs) play important roles in TNF-α signaling by interacting with signaling molecules linked to the pathogenesis of atherosclerosis in experimental mouse model (22, 23, 28, 42). Six members of the TRAFs family have been identified. All TRAFs share a COOH-terminal homology region termed the TRAF domain that is capable of binding to the cytoplasmic domain of receptors and to other TRAF proteins. In addition, TRAF 2–6 have RING and Zinc finger motifs that are important for signaling downstream events. TRAF4 is a unique member of the TRAF family having the most limited sequence homology to the other TRAF family (15). Of note, TRAF4 acts as a cytoplasmic adaptor molecule to mediate the translocation of binding partners from cytoplasm to nucleus through a specific NLS sequence (39, 41). There are also reports about cytoplasmic/nuclear distribution of TRAF4 (41) and oncogenic function of TRAF4 in breast cancer (30). In ECs, TRAF4 is shown to be involved in TNF-α-induced ROS production (19, 38). However, the linkage between Atox1 and TRAF4 in inflamed ECs has never been reported.

In this present study, we demonstrate that TNF-α stimulation of human ECs rapidly induced Atox1 binding to TRAF4 in a Cu-dependent manner, which was associated with their translocation from the cytosol to the nucleus. We also found that Atox1 was colocalized with TRAF4 at the nucleus in inflamed aortas from ApoE^−/− mice with angiotensin II (ANG II) infusion on a high-fat diet (HFD), while Atox1 was found in cytosol in those from control mice. TRAF4 depletion in ECs inhibited Atox1 nuclear translocation, p47phox expression, ROS production, and its downstream VCAM1/ICAM1 expression as well as monocyte adhesion to inflamed ECs activated by TNF-α. Our study provides a novel insight into the Atox1-TRAF4 axis as a potential therapeutic target for oxidant stress-dependent inflammatory vascular diseases such as atherosclerosis.

MATERIALS AND METHODS

Animals.

Eight- to twelve-week-old male ApoE^−/− or C57/BL6 [wild type (WT)] mice (Jackson Laboratory, Bar Harbor, ME) were used for this study. All mice were maintained at the Augusta University animal facilities. All studies were carried out in accordance with the guidelines approved by Institutional Animal Care and Use Committee at Medical College of Georgia.

Experimental design and vascular inflammation model.

Vascular inflammation was induced by ANG II (1,000 ng·kg⁻¹·min⁻¹; cat. no. A2900; Sigma-Aldrich) infusion via Alzet osmatic minipump (Alzet model no. 2004; Durect; Cupertino, CA) and high-fat diet (HFD) feeding (diet no. TD.88137; Harlan Teklad) for 14 days, as described previously (10, 26).

Systolic blood pressure (SBP) was measured noninvasively on conscious mice by a computerized tail-cuff blood pressure system as previously described (25). The total cholesterol and HDL and LDL level in plasma were measured by using a kit from manufacturer (cat no. EHDL-100; BioAssay Systems).

Cell culture and transfection reagent.

Human umbilical vein endothelial cells (HUVECs) were grown on 0.1% gelatin coating in EndoGRO basal medium (Millipore) containing 5% fetal bovine serum (FBS) as previously described (3, 4). Passage number did not exceed P7. Human epithelial kidney cells (HEK293) were used in mammalian two-hybrid system and coimmunoprecipitation. HEK293 were grown in 10% FBS DMEM media. Human TNF-α (R&D Systems) was used in all in vitro experiments at 10 ng/ml. Protein knockdown by siRNA (Ambion) in HUVECs was done by Oligofectamine (Invitrogen) for 4 h and followed by 48-h growth. Transfection of HEK293 by plasmid DNA was accomplished through Polyfect (Qiagen) according to the manufacturer protocol. Transfection was allowed for 18 h before treatment with harvest or TNF-α treatment. THP-1 monocytes were grown in RPMI media containing 10% FBS.

Immunoblotting.

Snap-frozen aortic tissues were homogenized in RIPA buffer (5 mM Tris·HCl pH 7.6, 150 Mm NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) with protease inhibitor followed by brief sonication as described previously (3, 4). With the use of Bio-Rad protein assay dye (no. 5000006; Bio-Rad), equal amount of proteins are separated by SDS-PAGE followed by incubation with overnight primary antibody anti-Atox1 (12), anti-CCS (Santa Cruz Biotechnology), anti-COX17 (Proteintech), anti-TRAF1/TRAF2/TRAF3/TARF6 (Cell Signaling), anti-TRAF4 (Santa Cruz Biotechnology), anti-Flag (Sigma), anti-Myc (Sigma), anti-VCAM1(Santa Cruz Biotechnology), anti-lamin B1(Santa Cruz Biotechnology), anti-P84 (GeneTex), anti-GAPDH (GeneTex), and anti-tubulin (Santa Cruz Biotechnology). After incubation with secondary antibodies (goat anti-rabbit IgG-HRP conjugate and goat anti-mouse IgG-HRP conjugate; Bio-Rad), protein expression was visualized by ECL (Amersham). Band density was quantified by ImageJ. Nuclear/cytoplasmic fractionation of aorta was performed using an NEPER Nuclear and Cytoplasmic Extraction Reagents Kit (Pierce) according to the manufacturer’s protocol as described previously (3, 4, 17).

Immunohistochemistry and immunofluorescence.

Frozen sections were prepared by overnight 4% PFA incubation followed by sucrose dehydration and OCT embedding. Seven-micromolar-thick sections were stained with conventional hematoxylin-eosin and with antibodies against Mac3 (BD PharMingen) and monocyte chemoattractant protein 1 (Santa Cruz Biotechnology). This was followed by incubation with biotin-conjugated secondary antibody (Vector Laboratories). Next, we used R.T.U. Vectorstain Elite (Vector Laboratories) followed by DAB visualization (Vector Laboratories) as described before (3, 4). Counterstaining with hematoxylin was performed. Immunofluorescence staining was performed with primary antibodies against Atox1 (3, 4, 12), TRAF4 (Santa Cruz Biotechnology). Secondary antibodies were Alexa Fluor 488- or 546-conjugated goat anti-rabbit IgG and goat anti-mouse IgG (Invitrogen). In each experiment, DAPI (Vector Laboratories) was used for nuclear counterstaining. Images were captured by Axio scope microscope (Zeiss) or confocal microscopy (Zeiss) and processed by AxioVision 4.8 or LSM510 software (Zeiss). All positive stained cells were counted in at least three microscopic fields (×40 or ×63).

Enface staining.

Animals were euthanized by CO₂ inhalation. The aorta was perfused via the left ventricle with saline containing heparin followed by 4% paraformaldehyde in PBS for 10 min. After adipose tissues were removed, aortas were cut open longitudinally and permeabilized with PBS containing 0.1% Triton X-100 and blocked by TBS containing 10% goat serum and 2.5% Tween 20 for 30 min as described previously (8, 9, 40). Aorta were triple stained with anti-β-catenin (Santa Cruz Biotechnology; as an EC marker, Alexa Fluor-546, red), anti-Atox1 (Alexa Fluor-488, green), and anti-TRAF4 (Santa Cruz Biotechnology; Alexa Fluor-488, green), and nuclei were counterstained with DAPI (blue). Images were captured by confocal microscopy (Zeiss) and processed by AxioVision 4.8 or LSM510 software (Zeiss).

Monocyte adhesion assay.

HUVECs were grown in 24-well glass bottom plates and treated for 18 h with TNF-α. THP-1 cells were labeled with Cell Tracker Green CMFDA (Invitrogen) and were added to the EC monolayer. Cultures were incubated in a CO₂ incubator for 1 h. Nonadherent cells were removed from the plate by gentle washing with PBS, and the number of adherent cells was determined by confocal microscopy as previously described (3). Cells were counted in five random ×20 fields per well.

Mammalian two-hybrid system.

To map the binding sites we used CheckMate Mammalian Two-Hybrid System (Promega). We fused Atox1 several construct as before (12) in pBIND Vector that contains the yeast GAL4 DNA-binding domain. TRAF4 several deletion constructs (38) were fused in pACT Vector that contains the herpes simplex virus VP16 activation domain. Each Atox1 and TRAF4 mutant as described in Fig. 3C was amplified by PCR and subcloned. pBIND Vector expresses the Renilla reniformis luciferase, which allows the user to normalize the transfection efficiency. The pGAL4 and pVP16 fusion protein constructs are transfected along with pG5luc Vector into human epithelial kidney cells (HEK293) using Polyfect (Qiagen) according to manufacturer protocol. Two to three days after transfection, the cells are lysed, and the amounts of Renilla luciferase and firefly luciferase are quantitated using the Dual-Luciferase Reporter Assay System (Promega). Interaction between the two proteins, as GAL4 and VP16 fusion constructs, results in an increase in firefly luciferase expression over the negative controls.

Fig. 3.

TNF-α stimulation induces Atox1 binding to TNF-α receptor-associated factor 4 (TRAF4) in a Cu-dependent manner in endothelial cells (ECs). A: human umbilical vein endothelial cells (HUVECs) were stimulated with 10 ng/ml TNF-α for indicated times. Lysates were immunoprecipitated (IP) with anti-TRAF4 antibody or IgG (negative control), followed by immunoblotting (IB) with anti-antioxidant-1 (Atox1) and anti-TRAF4 Abs. B: HEK 239T cells transfected with Flag-tagged Atox1 and Myc-tagged TRAF4 were stimulated with 10 ng/ml TNF-α for 15 min, and lysates were IP with anti-Flag antibody or IgG (negative control), followed by IB with either anti-Flag or anti-Myc Ab. C: identification of domains of Atox1 and TRAF4 necessary for their interaction using mammalian two-hybrid assay. Schematic representation of the Atox1 and TRAF4 constructs showing the pBIND vectors expressing the GAL4-DBD (DNA-binding domain) and the Atox1 or its deletion mutants (pBIND-Atox1) as well as pACT vectors expressing the VP16 activation domain containing TRAF4 or its mutants (pACT-TRAF4). The pGAL4 and pVP16 fusion constructs were cotransfected along with pG5luc Vector into HEK cells and analyzed by Dual-Luciferase Reporter Assay System. NLS, nuclear localization signal. D: HUVECs transfected with Flag-tagged Atox1-wild-type (WT) or Atox1 mutated with Cu-binding domain (CBD; Flag-Atox1 C12,15S) were stimulated with 10 ng/ml TNF-α. Lysates were IP with anti-TRAF4 Ab followed by IB with Flag Ab. E: HUVECs transfected with Flag-Atox1-WT in the presence of siRNA for either CTR1 or control (Cont) were stimulated with 10 ng/ml TNF-α for 15 min. Data represent means ± SE. *P < 0.05. F: HUVECs treated with Cu chelators BCS (200 μM, 48 h) and TTM (20 nM, 24 h) were stimulated with 10 ng/ml TNF-α for 15 min. Lysates were IP with anti-TRAF4 antibody or IgG (negative control), followed by IB with anti-Atox1 and anti-TRAF4 Abs. Data represent mean ± SE. *P < 0.05.

Quantitative real-time PCR.

Total RNA of HUVECs was isolated by using phenol/chloroform and isolated using Tri Reagent (Molecular Research Center, Inc.). Reverse transcription was carried out using a high-capacity cDNA reverse transcription kit (Applied Biosystems) using 2 ug of total RNA as before (3, 4). The PCR was performed according to the manufacturer’s protocol using ABI PRISM 7000 Sequence Detection System 26 (Applied Biosystems, CA) and QuantiFast SYBR Green PCR Kit (Qiagen, Valenica, Foster City, CA). Samples were all run in triplicates to reduce variability. The expression of genes was normalized and expressed as fold changes relative to GAPDH or18S.

Statistical analysis.

Results are expressed as means ± SE. Statistical significance was assessed by Student’s paired two-tailed t test or ANOVA on untransformed data, followed by comparison of group averages by contrast analysis using the SuperANOVA statistical program (Abacus Concepts, Berkeley, CA). Statistical tests were performed using Prism v4 (GraphPad Software, San Diego, CA). Values of P < 0.05, P < 0.01, and P < 0.001 were considered statistically significant.

RESULTS

Atox1 is localized at the nucleus in endothelium of inflamed aorta.

To determine if Atox1 is localized in the nucleus of inflamed vessels, we examined the subcellular localization of Atox1 in aorta of ApoE^−/− mice with ANG II infusion (1,000 ng·kg⁻¹·mice⁻¹) and high-fat diet (HFD) for 2 wk (ApoE^−/−/ANG II + HFD), which is known to induce vascular inflammation in vivo (10, 26). We used C57Bl6 mice with a standard chow diet mice as a control (WT) aortas. En face immunofluorescence staining revealed that Atox1 was localized in the nucleus in endothelial layers of inflamed lesions of ApoE^−/−/ANG II + HFD mice, (Fig. 1A)while it was in the cytosol in WT mice. Nuclear-cytosol fractionation further confirmed the predominant Atox1 localization in the nucleus in the inflamed ApoE^−/−/ANG II + HFD mouse aorta compared with control aorta (Fig. 1B). Of note, total protein expression of Atox1 and other Cu transport proteins was not changed in aorta of these two age-matched groups (Fig. 1C).

Fig. 1.

Antioxidant-1 (Atox1) is localized at the nucleus in endothelium of inflamed aorta. A: representative images of en face immunofluorescence staining in the endothelium of inflamed aortas from ApoE^−/− mice with ANG II infusion and high-fat diet (HFD) for 14 days (ApoE^−/−/ANG II/HFD) and control aortas from C57Bl6 mice with regular chow diet [wild type (WT)]. Aorta was costained with anti-β-catenin (as an EC marker, Alexa Fluor-546, red), anti-Atox1 antibody (Ab; Alexa Fluor-488, green), and DAPI (blue, nuclear marker). Images were taken using confocal fluorescence microscopy; n = 10 mice per group. B: nuclear and cytoplasmic fractions of aorta lysates from ApoE^−/−/ANG II/HFD and WT mice were immunoblotted with anti-Atox1, GAPDH (cytoplasmic marker), and p84 (nuclear marker) Abs. C: Atox1 and other Cu chaperone protein expression in inflamed and control aorta. Aorta lysates from ApoE^−/−/ANG II/HFD mice and C57Bl6 WT mice were immunoblotted with anti-Atox1, CCS, and COX17 antibodies or tubulin antibody (for loading control) (n = 4, each group). Data represent means ± SE. *P < 0.05.

TRAF4 is required for TNF-α-induced Atox1 nuclear translocation in a Cu-dependent manner in ECs.

We then examined the mechanism by which Atox1 is translocated to the nucleus in response to TNF-α in cultured ECs. TRAF4 is a unique member of TRAFs family having a specific NLS sequence and mediates the translocation of binding partners from cytoplasm to nucleus (39, 41). It is also required for TNF-α-induced ROS production (19) and associated with vascular inflammation (11, 19, 37, 38). Thus we examined the role of TRAF4 in TNF-α-induced Atox1 nuclear translocation in ECs. Figure 2A shows that both Atox1 and TRAF4 were mainly colocalized at the cytoplasm in resting ECs. In response to TNF-α, not only Atox1 but also TRAF4 translocated to the nucleus where they colocalized within 15 min. This result was further confirmed by the subcellular fractionation analysis (Fig. 2B). Importantly, TRAF4 knockdown using siRNA almost completely inhibited TNF-α-induced Atox1 nuclear translocation while Atox1 depletion had no effects on TRAF4 nuclear translocation (Fig. 2C). However, knockdown of the Cu importer CTR1 inhibited TNF-α-stimulated nuclear translocation of Atox1 without affecting that of TRAF4 (Fig. 2C). These results suggest that TRAF4 nuclear translocation is Cu independent but is required for Cu-dependent Atox1 translocation to the nucleus in inflamed ECs. CTR1 and Atox1 knockdown efficiency is shown in Fig. 2, D and E.

Fig. 2.

TNF-α receptor-associated factor 4 (TRAF4) is required for antioxidant-1 (Atox1) nuclear translocation in a Cu-dependent manner in inflamed endothelial cells (ECs). A: human umbilical vein endothelial cells (HUVECs) stimulated with TNF-α (10 ng/ml) for indicated time were stained with anti-Atox 1 (green) and anti-TRAF4 (red) Abs. In each image, immunofluorescence staining of cytosolic and nuclear location of Atox1 and TRAF4 was calculated from 5 randomized views, and the cell images are representative of 3 different experiments. *P < 0.01 vs. untreated cells. B: nuclear and cytoplasmic fractions were immunoblotted with anti-Atox 1, α-tubulin (cytoplasmic marker), or laminin B1 (nuclear marker) Ab. Right: averaged data for nuclear and nonnuclear expression of Atox1 protein, expressed as fold increased from control (Cont) ECs (n = 3). *P < 0.01 vs. untreated cells. C: HUVECs transfected with siRNAs for Atox1, TRAF4, CTR1, or control were stimulated with 10 ng/ml TNF-α for 30 min. Immunofluorescence staining were performed, as described in A. White arrows indicate either nuclear Atox1 or nuclear TRAF4 predominant cells. D: qPCR analysis of CTR1 mRNAs. E: immunoblot analysis for Atox1 or tubulin (loading control in HUVEC lysates.). Data represent means ± SE. *P < 0.05.

TNF-α stimulation induces Atox1 binding to TRAF4 in Cu-dependent manner in ECs.

To address the relationship between Atox1 and TRAF4, we next examined the physical interaction of Atox1 and TRAF4 in cultured ECs. Coimmunoprecipitation analysis revealed that TNF-α stimulation rapidly induced Atox1 association with TRAF4 within 10 min and continued at least for 30 min, which was cotemporaneous with their nuclear translocation (Fig. 3A). Atox1 binding to TRAF4 was further confirmed using HEK cells cotransfected with Flag-Atox1 and Myc-TRAF4 stimulated with TNF-α (Fig. 3B). To determine the binding site of Atox1 with TRAF4, we utilized a mammalian two-hybrid assay using pBIND vectors expressing the GAL4-DBD (DNA-binding domain) and the Atox1 or its deletion mutants (pBIND-Atox1) as well as pACT vectors expressing the VP16 activation domain containing TRAF4 or its mutants (pACT-TRAF4). The pGAL4 and pVP16 fusion constructs were cotransfected along with pG5luc Vector into HEK cells and analyzed by Dual-Luciferase Reporter Assay System (Fig. 3C). We found that the TRAF4 (210 to 267) containing the zinc finger (ZF) domain interacted with the NH₂ terminus of Atox1-(1–55) containing the CXXC (CBD) (12) motif. The role of Cu for Atox1 binding to TRAF4 was further confirmed that TNF-α-induced Atox1-TRAF4 association was significantly reduced in HUVECs transfected with Flag-Atox1-CBD mutant (Atox1-C12,15S) compared with Flag-Atox1-WT (Fig. 3D) as well as in Cu importer CTR1-depleted ECs with siRNA (Fig. 3E). Moreover, the cell permeable Cu chelator TTM or cell impermeable Cu chelator BCS inhibited Atox1 interaction with TRAF4, indicating that TRAF4-Atox1 interaction was Cu dependent (Fig. 3F). These findings suggest that Cu is required for interaction of Atox1 with TRAF4 and subsequent to their nuclear translocation in inflamed ECs.

Nuclear Atox1 rescues decreased p47phox-dependent ROS production and inflammatory responses in TRAF4 depleted ECs stimulated with TNF-α.

We previously reported that nuclear translocation of Atox1 is required for NOX organizer p47phox transcription, resulting in ROS production, inflammatory gene expression, and monocyte adhesion to inflamed ECs in a Cu-dependent manner (3, 4). We thus examined the role of TRAF4 in these nuclear Atox1-dependent downstream responses in ECs stimulated with TNF-α. TRAF4 depletion with siRNA significantly inhibited TNF-α-induced p47phox mRNA expression (Fig. 4A), intracellular H₂O₂ production as measured by PEG-catalase-inhibitable DCF-DA fluorescence (Fig. 4B) (3), and ICAM-1 and VCAM-1 mRNA (Fig. 4D) and protein expression (Fig. 4E) in ECs. TRAF4 siRNA also blocked adhesion of monocyte THP1 cells to HUVECs activated by TNF-α (Fig. 4C). Of note, these siTRAF4-mediated inhibitory effects on ICAM1 and VCAM1 mRNA were rescued by overexpression of nuclear targeted Atox1 (Atox1-NLS) (Fig. 4, A–D). We further confirmed nuclear localization of Atox1 in TRAF4-depleted ECs transfected with Atox1-NLS (Fig. 4F). These results suggest that TRAF4 is upstream for the Atox1 nuclear translocation to promote p47phox-depedent ROS production and inflammatory responses in ECs stimulated with TNFα.

Fig. 4.

Nuclear antioxidant-1 (Atox1) rescues decreased p47phox-dependent reactive oxygen species (ROS) production and inflammatory responses in TNF-α receptor-associated factor 4 (TRAF4)-depleted ECs stimulated with TNF-α. A–D and F: human umbilical vein endothelial cells (HUVECs) were transfected with si-control (Cont) or siTRAF4 for 4 h and then infected with adenovirus expressing either Atox1 with nuclear-target sequence (Ad-Atox1-NLS) or control empty virus (Ad-Null) for 48 h. Cells were stimulated with 10 ng/ml TNF-α and used for each assay. A and D: real-time PCR analysis of mRNA levels of p47phox and Atox1 (A; n = 4) and ICAM1 and VCAM1 (D: n = 4) in ECs stimulated with TNF-α (10 ng/ml for 6 h). F: cells were stimulated with 10 ng/ml TNF-α for 30 min and used for immunofluorescence staining for Atox1 (green) and DAPI (blue, nuclear marker). NLS, nuclear localization signal. B: DCFDA fluorescence analysis in ECs stimulated with TNF-α (10 ng/ml for 6 h) is shown. Representative images for DCF fluorescence and DAPI staining (blue; top) and quantification of fluorescence intensity (bottom; n = 3). C: monocyte adhesion to ECs. The numbers of bound THP1 monocytes (fluorescently labeled) to ECs stimulated with TNF-α (10 ng/ml for 18 h) were measured using fluorescence microscope. Bottom: quantification (n = 15 regions per group from 3 independent experiments). HAEC, human aortic endothelial cells. E: lysates were immunoblotted with anti-TRAF4, anti-VCAM1, and anti-α-tubulin (loading control) antibodies. HUVECs transfected with control or TRAF4 siRNA were stimulated with 10 ng/ml TNF-α for 16 h. Data represent means ± SE. *P < 0.05; **P < 0.01; ***P < 0.001. NS, not significant.

TRAF4 and Atox1 were colocalized in the nucleus of inflamed aorta.

Since both TRAF4 and Atox1 were translocated to the nucleus in inflamed cultured ECs and Atox1 was found in the nucleus of inflamed aorta (Figs. 1 and 2), we next examined the subcellular localization of TRAF4 in the endothelium of these vessels using en face staining. Figure 5A showed that some TRAF4+ cells and a majority of Atox1+ cells were found in the nucleus in inflamed aorta while most of TRAF4+ and Atox1+ cells were in the cytosol in control aorta. Moreover, TRAF4 and Atox1 coimmunostaining revealed that they were colocalized in the nucleus of inflamed vessel but not in that of control vessel (Fig. 5B). Of note, Atox1/TRAF4 double-positive cells in the nucleus were found not only in the endothelium but also in the other cells in the intimal lesion (Fig. 5B). Furthermore, coimmunoprecipitation analysis confirmed that Atox1/TRAF4 complex was markedly increased in the inflamed vessels, as compared with control vessels (Fig. 5C). Thus these data suggest that Atox1 and TRAF4 were colocalized in the nucleus of the inflamed aorta in vivo.

Fig. 5.

TNF-α receptor-associated factor 4 (TRAF4) and antioxidant-1 (Atox1) were colocalized in the nucleus of inflamed aorta. A: representative images of en face immunofluorescence staining of Atox1 and TRAF4 in the arterial endothelium of wild-type (WT) and ApoE^−/− mice with ANG II infusion and high-fat diet (HFD) for 14 days. Aorta were costained with anti-β-catenin (as an EC marker, Alexa Fluor-546, red), anti-Atox1, or anti-TRAF4 Ab (Alexa Fluor-488, green), and DAPI (blue). Images were taken using confocal fluorescence microscopy; n = 10 mice per group. B: representative hematoxylin-eosin (H&E) and immunostaining for Atox1 (red), TRAF4 (green), DAPI (blue, nuclear marker), or their colocalization (merge) in inflamed or noninflamed aorta. Serial section stained with H&E. Arrow indicates the colocalization of Atox1(+)/TRAF(+) cells in the nucleus of endothelium, *Atox1(+)/TRAF(+) cells in the nucleus of other inflammatory cells. Scale bar = 50 μm. C: coimmunoprecipitation (IP) analysis of TRAF4 and Atox1 in inflamed (ApoE^−/−/ANG II/HFD) and noninflamed (WT) aortas. Right: quantitative analysis. Data represent means ± SE; n = 3; *P < 0.05.

DISCUSSION

Atox1 is classically known as a cytosolic metallochaperone, but we previously reported that in inflamed ECs Atox1 functions as a Cu-dependent transcription factor for the NOX organizer p47phox to promote ROS/NF-κB-dependent inflammatory gene expression in ECs (3, 4). However, the mechanisms of Atox1 nuclear translocation and its role in vascular inflammation remain unknown. Using coimmunoprecipitation, colocalization, and cotransfection assays, here we identified TRAF4 as a novel binding partner of Atox1 in ECs stimulated with TNF-α. The mammalian two-hybrid assays revealed that TRAF4 the containing zinc finger domain interacted with the Atox1 containing the Cu-binding CBD domain. Endogenous TRAF4 knockdown by siRNA with rescue by overexpression of nuclear targeted Atox1 showed that TRAF4 was required for Cu-dependent Atox1 nuclear translocation, p47phox expression, and ROS production as well as its downstream VCAM1/ICAM1 expression and monocyte adhesion to ECs. We also found that Atox1 was colocalized with TRAF4 at the nucleus in TNF-α-stimulated inflamed ECs and vessels, which was consistent with a role of Atox1 in vascular inflammation (Fig. 6).

Fig. 6.

Proposed model showing that Cu-dependent antioxidant-1 (Atox1)-TNF-α receptor-associated factor 4 (TRAF4) binding is required for Atox1 nuclear translocation to promote p47phox/reactive oxygen species (ROS)-dependent inflammatory responses in inflamed endothelial cells. NLS, nuclear localization signal.

TRAF4 has a specific NLS sequence and mediates the translocation of binding partners from cytoplasm to nucleus (39, 41). TRAF4 is shown to be involved in TNF-α-induced ROS production (19) and associated with human diseases and atherosclerosis (11, 19, 37, 38). We reported that Atox1 is translocated to the nucleus in response to TNF-α in a Cu-dependent manner, thereby increasing p47phox expression, ROS production, and inflammatory gene expression in ECs (3, 4). However, the linkage between TRAF4 and Atox1 in inflamed ECs remained unknown. The present study shows that TNF-α stimulation rapidly promotes Atox1 binding to TRAF4 and that knockdown of TRAF4 or the Cu importer CTR1 inhibits Atox1 nuclear translocation, while Atox1 or CTR1 depletion has no effect on TRAF4 nuclear translocation. We also found that the Atox1-CBD mutant (C12,15S) inhibits TNF-α-induced Atox1-TRAF4 binding as well as Atox1 nuclear translocation in ECs. Mammalian two-hybrid assay demonstrates that TRAF4 at the zinc finger domain interacts with the Atox1 at Cu-binding CBD domain. These results suggest that TRAF4 nuclear translocation is Cu or Atox1 independent and that Cu binding to Atox1 at the CBD domain is required for forming complex with TRAF4 at the zinc finger domain, which, in turn, promotes the Atox1 nuclear translocation in response to TNF-α. Other domains of TRAF4 including the TRAF domain (24) or the RING domain, which functions as an E3 ubiquitin ligase involved in cancer development, metastasis, and chemoresistance (20, 32), may mediate nuclear translocation of other transcription factors for enhancing NF-κB/Wnt signaling (39, 41), as reported in cancer cells.

A functional significance of Atox1 binding to TRAF4 is demonstrated by the finding that TRAF4 siRNA significantly inhibits Atox1-dependent p47phox expression, ROS production, VCAM1/ICAM1 expression, and monocyte adhesion to inflamed ECs, which are rescued by overexpression of nuclear-targeted Atox1. Thus these data support the notion that Cu-dependent Atox1/TRAF4 binding is required for TNF-α-induced Atox1 nuclear translocation, which contributes to p47phox-dependent ROS production and redox-sensitive inflammatory responses in ECs. Of note, it is reported that TRAF4 binds to phosphorylated p47phox to promote the rapid NOX activation and ROS production required for acute TNF-α signaling (19). Thus it is likely that TRAF4 is involved in the early phase of TNF-α-induced ROS elevation via binding to p-p47phox and in the later phase of ROS production via binding to Cu-bound Atox1, which contributes to nuclear Atox1-mediated p47phox expression in ECs. Given that TRAFs play an important role in regulating NF-κB-dependent inflammatory signaling pathways (5), our study provides a new mechanism by which TRAF4 promotes ROS-dependent inflammatory responses via binding to Atox1 in a Cu-dependent manner.

To assess the in vivo significance of interaction of Atox1 with TRAF4 in inflamed ECs, we demonstrate that Atox1 was colocalized with TRAF4 in the nucleus in the arterial endothelium and other cells such as inflammatory cells of vessels inflamed with ANG II/HFD treatment, while they were localized in the cytosol in control noninflamed vessels. Since recruited inflammatory cells following tissue injury secrete cytokines and chemokines at the injured sites (10, 26), they may promote Atox1 nuclear translocation in both inflammatory cells and ECs. Consistent with this finding, we previously reported that Atox1 colocalized with Mac3-positive macrophages in wound injury site (4). Of note, it is shown that hyperproliferative and inflammatory lesions in cancer and atherosclerosis have higher Cu levels in the nuclei than normal tissues (12, 29, 33), while Cu chelation prevents tumor growth, neointimal thickening after vascular injury, or atherosclerosis (2, 27, 35, 36). Thus it is tempting to speculate that Cu increase in the nucleus in vascular inflammatory and proliferative diseases may reflect the Cu-bound Atox1 nuclear accumulation. In addition, not only TNF-α but also Cu and inflammatory growth factor ANG II have been shown to promote Atox1 nuclear translocation (25). Thus we cannot eliminate the possibility that other adaptor proteins or posttranslational modification of Atox1 may be involved in the mechanism of its nuclear translocation in ECs and other cells including vascular smooth muscle cells and macrophages. Investigating the in vivo role of nuclear Atox1 in other vascular inflammation models and atherosclerosis models is the subject of future investigation.

In conclusion, our study suggests that Cu-dependent Atox1/TRAF4 binding is required for Atox1 nuclear translocation to promote p47phox/ROS-dependent adhesion molecule expression in ECs, thereby accelerating inflammatory cell recruitment. These mechanisms may contribute to development of atherosclerosis, providing insight into a central role of Atox1 in oxidant stress-dependent vascular inflammatory diseases.

GRANTS

This research was supported by National Heart, Lung, and Blood Institute Grant R01-HL-070187 (to T. Fukai), R01-HL-116976 and R01-HL-133613 (to T. Fukai and M. Ushio-Fukai), and R01-HL-135584 (to M. Ushio-Fukai), Department of Veterans Affairs Merit Review Grant I01BX001232 (to T. Fukai), and American Heart Association Grants 16POST27790038 (to A. Das) and 15SDG25700406 (to V. Sudhahar).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.D., M.U.-F., and T.F. conceived and designed research; A.D. and V.S. performed experiments; A.D. analyzed data; A.D., V.S., M.U.-F., and T.F. interpreted results of experiments; A.D. and T.F. prepared figures; A.D. drafted manuscript; A.D., M.U.-F., and T.F. edited and revised manuscript; A.D., V.S., M.U.-F., and T.F. approved final version of manuscript.