Secreted APE1/Ref-1 inhibits TNF-α-stimulated endothelial inflammation via thiol-disulfide exchange in TNF receptor

Apurinic apyrimidinic endonuclease 1/Redox factor-1 (APE1/Ref-1) is a multifunctional protein with redox activity and is proved to be secreted from stimulated cells. The aim of this study was to evaluate the functions of extracellular APE1/Ref-1 with respect to leading anti-inflammatory signaling in TNF-α-stimulated endothelial cells in response to acetylation. Treatment of TNF-α-stimulated endothelial cells with an inhibitor of deacetylase that causes intracellular acetylation, considerably suppressed vascular cell adhesion molecule-1 (VCAM-1). During TSA-mediated acetylation in culture, a time-dependent increase in secreted APE1/Ref-1 was confirmed. The acetyl moiety of acetylated-APE1/Ref-1 was rapidly removed based on the removal kinetics. Additionally, recombinant human (rh) APE1/Ref-1 with reducing activity induced a conformational change in rh TNF-α receptor 1 (TNFR1) by thiol-disulfide exchange. Following treatment with the neutralizing anti-APE1/Ref-1 antibody, inflammatory signals via the binding of TNF-α to TNFR1 were remarkably recovered, leading to up-regulation of reactive oxygen species generation and VCAM-1, in accordance with the activation of p66shc and p38 MAPK. These results strongly indicate that anti-inflammatory effects in TNF-α-stimulated endothelial cells by acetylation are tightly linked to secreted APE1/Ref-1, which inhibits TNF-α binding to TNFR1 by reductive conformational change, with suggestion as an endogenous inhibitor of vascular inflammation.

Chronic vascular inflammation plays a key role in the pathogenesis of atherosclerosis and other vascular disease 1 . Accordingly, the regulation of inflammatory reactions in the vascular endothelium is a potential target for therapeutic intervention in the treatment of chronic inflammation, such as atherosclerotic disease. Inflammation is mainly mediated by monocyte adhesion to endothelial cells. The recruitment of monocytes to the affected tissue and accumulation of monocyte-derived phagocytes 2 are actively mediated and precisely controlled by cytokines, such as interleukin-1β (IL-1β ), IL-6, IL-8, and tumor necrosis factor (TNF)-α . The interaction between blood monocytes and the vascular endothelium involves a cytokine-mediated process that includes monocyte rolling, arrest, firm adhesion, and diapedesis 3 . During vascular inflammation, the adhesion cascade of monocytes is regulated by a combination of endothelial cell surface adhesion molecules including vascular cell adhesion molecule-1 (VCAM-1), intercellular cell adhesion molecule-1, and E-selectin 4 .
In vascular inflammatory responses, TNF-α which is released from macrophages exert direct effects on a multitude of secondary inflammatory mediators via binding with the TNF-α receptors (mainly TNFR1) 5 , resulting in the production of reactive oxygen species (ROS) and the activation of nuclear factor-κ B (NF-κ B) 6,7 . Activated NF-κ B in the nucleus regulates the transcription of genes involved in the pathogenesis of inflammatory lesions, including cytokines, chemokines and adhesion molecules 8 . Therefore, the treatment of vascular inflammation with agents that block initial TNF-α activity can be highly beneficial and can minimize side effects or the disruption of overlapping intracellular signaling. For example, three representative drugs, infliximab, adalimumab, and etanercept, are TNF-α antibodies or TNFR1-Fc chimeras and function to prevent TNF-α from binding to its receptor; all are currently used to treat inflammatory disease 9 . Although TNF inhibition fails to improve symptoms in severe late-stage infectious diseases, trials are necessary to evaluate its use in vascular inflammatory diseases. TNFR1 is a member of the TNF receptor superfamily, which is a group of cytokine receptors that have the ability to bind TNFs via an extracellular cysteine-rich domain (CRD) 10 . TNFR1 has six consensus cysteine residues forming three disulfide bonds in each of the four CRDs for recognition of its ligand, homotrimeric TNF-α 11,12 . Considering the structure of the TNF-α /TNFR1 complex, some studies have reported the development of TNF-α inhibitors based on the key sites of the TNF-α /TNFR1 interaction, peptide mimics of the TNFR1 loop, or small molecules that bind to TNF-α directly 13 .
Apurinic apyrimidinic endonuclease 1/Redox factor-1 (APE1/Ref-1, also known as Ref-1) is a multifunctional protein; its N-terminal region is involved in redox activity and regulates multiple transcription factors, and its C-terminus is involved in base excision DNA repair activity 14 . APE1/Ref-1 undergoes active shuttling between the cytoplasm and nucleus in response to oxidative stress [15][16][17] . Interestingly, previous studies, including ours, have reported the possibility for the extracellular secretion of APE1/Ref-1. Auto-antibodies against APE1/Ref-1 have been found in patients with systemic lupus erythematosus 18 and lung cancer 19 , suggesting the exposure of APE1/ Ref-1 to the host immune system. Elevated levels of APE1/Ref-1 were also observed in the blood of endotoxemic rats 20 and in bladder cancer 21 , implying that APE1/Ref-1 functions as a secreted protein.
Because the level of secreted APE1/Ref-1 is substantially increased in response to acetylation, we hypothesized that secreted APE1/Ref-1 could be an effective regulator in inflammatory reactions via its reduction. We tested this hypothesis using TNF-α -treated human umbilical vein endothelial cells (HUVECs) as a vascular inflammation model. We provide compelling experimental evidence to indicate that extracellular secreted APE1/Ref-1 in response to intracellular acetylation inhibits inflammatory signaling via a reduction in TNFR1, showing that treatment of anti-APE1/Ref-1 antibody in histone deacetylase inhibitor (HDACi), trichostatin A (TSA)-mediated modulation against TNF-α -stimulated endothelial activation recovers not only upregulation of adhesion molecule but also the generation of ROS.

Results
TSA treatment caused downregulation of VCAM-1 in TNF-α-stimulated HUVECs. The HDACi, TSA inhibits the expression of the cell adhesion molecule VCAM-1 in TNF-α -stimulated endothelial cells 22 , but the sequence of events leading to anti-inflammatory effects in the vascular system is still unclear. Accordingly, we examined the mechanism of VCAM-1 suppression in TNF-α -stimulated endothelial cells treated with TSA. As shown in Fig. 1A,B, TSA treatment resulted in a considerable decrease in VCAM-1 expression and an increase in intracellular acetylation. The level of VCAM-1 was almost completely downregulated unlike cells simulated with TNF-α only (Fig. 1A).
To test whether TSA-mediated acetylation causes downregulation of VCAM-1 in TNF-α -stimulated endothelial cells, we observed the effect of deacetylase on VCAM-1 levels using an adenovirus expressing HDAC3. The level of VCAM-1 in TNF-α -treated cells was substantially increased in HDAC3-overexpressed cells presenting no acetylation (Fig. 1B,C). As shown in the bar graph in Fig. 1C, the constitutive expression of VCAM-1 increased to ~130% following HDAC3 adenoviral infection compared with that of TNF-α -stimulated cells. In contrast, the upregulated VCAM-1 caused by TNF-α treatment or additional HDAC3 adenoviral infection was considerably decreased by the introduction of adenoviral APE1/Ref-1 (Fig. 1C). The VCAM-1 level decreased by ~38% in response to the expression of APE1/Ref-1 based on densitometric scanning of the immunoreactive bands after correcting for the actin loading control. Collectively, these results indicated that VCAM-1 expression in TNF-α stimulated endothelial cells is regulated by acetylation/deacetylation, implying a functional role of the APE1/ Ref To confirm whether APE1/Ref-1 secretion is regulated by acetylation, we examined the acetylation of secreted APE1/Ref-1 using anti acetyl-lysine antibody. TSA-mediated acetylation caused an initial increase in secreted Ac-APE1/Ref-1 compared with untreated control cells (Fig. 2C). Interestingly, the acetyl group was rapidly removed by 0.

Neutralization of secreted APE1/Ref-1 recovered ROS generation, which was temporarily inhibited by TSA treatment in TNF-α-stimulated endothelial cells. Intracellular ROS generation
is involved in inflammatory signaling in TNF-α -stimulated HUVECs 24 . Using a DCFDA fluorescent probe to examine intracellular ROS levels in TNF-α -stimulated endothelial cells, we detected the maximum intracellular ROS level at 6 h, at which point ROS levels were almost 2-fold higher in treated cells than in untreated control cells (Fig. 4A). The elevated ROS level in response to TNF-α decreased dramatically to basal levels after TSA-mediated acetylation, indicating an anti-inflammatory effect of TSA via inhibition of ROS production in TNF-α -stimulated HUVECs. To confirm whether secreted APE1/Ref-1 in response to TSA-mediated acetylation also affects ROS generation in TNF-α -stimulated HUVECs, we determined the effect of a neutralizing antibody on intracellular ROS levels. Similar to the observations for VCAM-1 expression, intracellular ROS generation increased after anti-APE1/Ref-1 antibody treatment, while TSA treatment inhibited ROS production in TNF-α -stimulated Additionally, a change in mitochondrial ROS generation in response to TSA-mediated acetylation was observed using the mitochondrial superoxide indicator, MitoSox in TNF-α -stimulated endothelial cells. As shown in Fig. 4B, the increase in mitochondrial ROS by TNF-α -stimulation was decreased by 60% after treatment with TSA. However, the inhibited mitochondrial ROS generation was almost fully recovered in the presence  inflammatory signaling, the involvement of p66 Shc and MAPKs was investigated. As shown in Fig. 6A,B, the phosphorylation level of p66 shc on serine 36 residue increased in TNF-α -treated endothelial cells and then considerably decreased by 50% after TSA treatment. Consistent with the results for ROS generation, suppressed p66 Shc was almost completely reactivated after treatment with a neutralizing APE1/Ref-1 antibody, to a level similar to that of TNF-α treated cells, although immunoglobulin did not affect the reactivation of p66 shc . Additionally, the contribution of MAPK activation to TNF-α -induced Ser-phosphorylation of p66 Shc was studied. When TNF-αand TSA-treated cells were incubated with neutralizing APE1/Ref-1 antibody, the suppressed phosphorylation of p66 shc increased significantly (p < 0.001 vs. TNF-α -/TSA-treated cells) as shown in Fig. 6A,B. By contrast, treatment with the neutralizing antibody in JNK or ERK involvement to phosphorylation p66 shc in TNF-α /TNFR1 inflammatory signaling did not observed (data not shown). Thus, these results suggest that p66 Shc and p38, but not JNK or ERK, are involved in anti-inflammatory signals via secreted APE1/Ref-1 in TNF-α -stimulated cells.

Discussion
An understanding of the function of extracellular APE1/Ref-1 is necessary for its application as an anti-inflammatory agent and to establish clinical trials to support a mechanism-based approach to treat vascular inflammatory disease. The present study provides novel evidence for the role of the auto-or paracrine secretory protein in inflammation and indicates that the anti-inflammatory regulation of secreted APE1/Ref-1 in TNF-α -stimulated endothelial cells after TSA treatment is tightly linked to the inactivation of TNF-α /TNFR1 signaling. This conclusion is based on the following observations: Acetylation, a type of post-translational modification, is considered an important mechanism in the regulation of gene expression via activation of nuclear histone proteins; it regulates the functions of multiple cytoplasmic proteins involved in cell survival, proliferation, and death, as well as disease-related intracellular signaling. Cellular acetylation increases rapidly via HDACi administration. HDAC inhibitors are potential chemotherapeutic agents for cancer and inflammatory diseases [26][27][28] . Interestingly, previous studies have shown that TSA treatment induces remarkable regulation of TNF-α -induced VCAM-1 expression and thereby inhibits monocyte adhesion both in vitro and in vivo, even though the molecular basis for this effect is unclear. Moreover, acetylation at a specific amino acid residue, i.e., lysine, and changes in the function of cellular proteins have also been reported, including changes in DNA-protein interactions, transcriptional activity, protein stability, and enzymatic activity 29 . Treatment with HDACi, which suppresses the removal of acetyl groups from lysine residues, causes intracellular acetylation and dramatic extracellular translocation of several proteins 30 . TSA-mediated acetylation induces the secretion of APE1/Ref-1, this effect is dependent on K6/K7 acetylation in HEK293 cells 31 . Acetylation of hsp90α by treatment with a pan-HDAC inhibitor causes hyperacetylation and extracellular secretion, increasing in vitro tumor cell invasiveness 32 . Inhibition of HDAC-1 and − 4 in hepatocytes causes an increase in HMGB-1 release, indicating that acetylation is critical for active HMGB-1 release 33  In TNF-α -stimulated inflammation, TNFR1 is a potentially selective substrate for secreted APE1/Ref-1. TNFR1 is characterized by the ability to bind TNF-α via extracellular CRDs; the receptor itself is comprised of a chain of four CRDs, each of which is stabilized by three pairs of disulfide-linked cysteine residues 25 . To initiate intracellular inflammatory signal transduction, homotrimeric TNF-α directly binds to CRDs of the receptor molecule, leading to signal-competent multimerization of unligated TNFRs 25 . Interestingly, there is some evidence showing that some disulfide bonds at the cell surface existing mainly in an oxidized redox state are labile, affecting activity 37 . For example, representative reducing agents, DTT or N-acetylcysteine can cleave disulfide bonds, resulting in a change in protein structure, and hence a change in function 37 . In addition, TRX can modulate the activity of cluster of differentiation 30, a member of the TNFR family, via the selective reduction of a disulfide Immunoblotting for phospho-p66 shc (B) Immunoblotting for phospho-p38 MAPK. The blots were stripped and reprobed with anti-p66 shc , -p38, or -β-actin antibodies to ensure equal protein loading. Immunoblotting for each protein was performed two or more times by using independently prepared lysates and similar results were obtained. Foldchanges in phosphorylated vs. total protein or the levels of apoptosis markers relative to the control are shown for each time point. Columns, mean (n = 2-3); bars, SE. **P < 0.01, significantly different from TSA/TNF-α -or IgG/ TSA/TNF-α -treated cells; # P < 0.01, significantly different from only untreated control cells by one-way ANOVA followed by Dunnett's tests. These experiments were replicated with similar results. Representative blots are shown.
bond, despite their high content of disulfide bonds 37 . Interferon gamma-inducible lysosomal thiol reductase, which is secreted from circulating macrophages after exposure to bacterial lipopolysaccharides (LPS), plays a role in major histocompatibility complex class II-restricted processing and the presentation of antigens that contain disulfide bonds. Because, rh APE1/Ref-1 causes the reduction of disulfide bonds in the extracellular domain of rh TNFR1, and treatment with neutralizing anti-APE1/Ref-1 antibody rapidly prevents acetylation-mediated anti-inflammatory signals in TNF-α -stimulated endothelial cells, it seems reasonable to postulate that down regulation of ROS, p66 shc /MAPK, and VCAM-1 expression are induced by secreted APE1/Ref-1, either auto-or paracrinally. However, further studies are needed to systematically explore this possibility, including the identification of molecules that promote the interaction with the substrate TNFR1.
It is plausible that multiple disulfide bonds in the extracellular domain of other transmembrane receptors can be affected by reducing proteins. Cytokines and biochemical mediators released during inflammation intensify and propagate the inflammatory response. These mediators can act systemically and activate other receptors simultaneously. Based on our studies, inflammatory reactions stimulated by cytokines through the IL-1 receptor or the Toll-like receptor (each has 5 disulfide bonds in the extracellular domain) were effectively attenuated by exposure to APE1/Ref-1 ( Supplementary Figures 4 and 5). In accordance with TNF-α /TNFR regulation, the IL-1 or LPS-stimulated inflammatory signal was also inhibited by reduction of disulfide bonds, implying a broad-spectrum reducing effect of APE1/Ref-1.
In conclusion, the present study offers novel insight into the mechanisms of hyperacetylation-mediated anti-inflammation signals in human TNF-α -endothelial cells and shows a strong relationship between secreted APE1/Ref-1, but not Ac-APE1/Ref-1, and TNF-α /TNFR1 and the inflammatory process. Based on these observations, we propose that rh APE1/Ref-1 with reducing activity may be a useful therapeutic agent for the treatment of cytokine responses involved in vascular inflammation. Clarifying the mechanisms of the post-translational modification of secreted APE1/Ref-1 may provide specific markers for vascular diseases and facilitate the development of a vascular inflammation inhibitor.

Methods
Cell culture. HUVECs were cultured in endothelial growth medium as previously reported 38 . Cells (passages [4][5][6][7][8] were maintained at 37 °C in a humidified atmosphere of 5% CO 2 . software. Statistical significance in the differences in measured variables between the control and treated groups was determined by a one-way analysis of variance (ANOVA) followed by Dunnett's or Bonferroni's multiple comparison tests. Differences were considered significant at P < 0.05.