Vicinal Dithiol-binding Agent, Phenylarsine Oxide, Inhibits Inducible Nitric-oxide Synthase Gene Expression at a Step of Nuclear Factor-κB DNA Binding in Hepatocytes*

Inflammatory cytokine interleukin 1β induces inducible nitric-oxide synthase (iNOS) mRNA and its protein, which are followed by increasing the production of nitric oxide, in primary cultures of rat hepatocytes. Nuclear factor-κB (NF-κB), an important transcription factor for iNOS gene expression, is also activated and translocated to the nucleus. In the present study, we found that vicinal dithiol-binding agent, phenylarsine oxide (PAO), inhibited the induction of iNOS protein and mRNA as well as the release of nitrite (nitric oxide metabolite) into the culture medium. Simultaneous addition of a vicinal dithiol compound, 2,3-dimercaptopropanol, with PAO completely abolished these inhibitions. PAO could not prevent either degradation of an inhibitory protein, IκB, of NF-κB or translocation of NF-κB to the nucleus. However, electrophoretic mobility shift assay demonstrated that PAO decreased the interaction between NF-κB and its binding consensus oligonucleotide. Transfection experiments with iNOS promoter-luciferase construct revealed that PAO inhibited NF-κB binding to DNA. These results indicate that PAO inhibits iNOS gene expression at a step of NF-κB binding to DNA by modifying its vicinal dithiol moiety, which may play a crucial role for the iNOS regulation in hepatocytes.

Nitric-oxide synthase (NOS) 1 catalyzes a production of nitric oxide from L-arginine, which plays an important role in a variety of physiological functions including vascular tone regulation, neurotransmission, and immune response mediation (1). NOS can be classified into two groups, constitutive NOS and inducible NOS (iNOS). Constitutive NOS is present in tissues such as the endothelium and brain (2,3), but iNOS is present in negligible quantities under physiological conditions. During inflammation, injury, and infection, lipopolysaccharide and proinflammatory cytokines induce iNOS in some tissues and organs including vascular vessels, kidney, and liver (4 -6). Induction of iNOS by lipopolysaccharide contributes to the pathogenesis of septic shock (7), leading to organ destruction, including the liver. We (8) and others (9) have reported that a single cytokine interleukin 1␤ (IL-1␤) mimicked the in vivo induction of iNOS in cultured hepatocytes. Other cytokines such as IL-6, tumor necrosis factor ␣ (TNF␣), and interferon ␥ had no effect under the same conditions (8), although a combination of these cytokines and lipopolysaccharide stimulated the nitric oxide production (10,11). A high level of nitric oxide production by iNOS influences various hepatic metabolism and functions, where nitric oxide inhibits the mitochondrial Krebs cycle enzyme (12) and ATP synthesis (13).
IL-1␤ is a key mediator in the inflammatory response and has been shown to activate the transcription factor, nuclear factor-B (NF-B), that is critical for the inducible expression of many genes involved in inflammation (14). The promoters of murine and human genes encoding iNOS contain a consensus sequence for the binding of NF-B (15)(16)(17), which is necessary to confer inducibility by cytokines. To induce iNOS mRNA, activation of the transcription factor NF-B is essential although not sufficient by itself (18 -20). Nuclear translocation of NF-B is triggered by changes in the redox state (21). As mentioned before, IL-1␤ activated and translocated NF-B to the nucleus, resulting in the induction of iNOS gene expression, in primary cultures of rat hepatocytes (22). NF-B is typically found in the form of p50/p65 heterodimers attached to the inhibitory molecule (IB) in the cytoplasm of cells (23). Activation of NF-B involves (i) proteolytic degradation of IB following the phosphorylation by IB kinase (24,25), (ii) translocation of NF-B to the nucleus, and (iii) binding to the promoter B site of a target gene. The last step requires reduced cysteine residue(s) of NF-B (26).
However, the role of the thiol residues in NF-B molecule on the induction of iNOS in hepatocytes is obscure. deVera et al. (27) reported that sodium arsenite inhibited cytokine-inducible nitric-oxide synthase expression in rat hepatocytes. Recently, Shumilla et al. (28) reported that thiol-reactive metals such as chromium, cadmium, mercury, zinc, and arsenite inhibited NF-B binding to DNA in vitro. Phenylarsine oxide (PAO), arsine oxide derivative, interacts strongly with vicinal dithiolcontaining molecules including enzymes and transcription factors (29). PAO also interacts with small molecular weight dithiol compounds such as 2,3-dimercaptopropanol and 1,4dithiothreitol (DTT) but hardly interacts with monothiol compounds (30). These reports prompted us to investigate 1) whether arsine oxide derivative, PAO, inhibits the production of nitric oxide induced by IL-1␤ in hepatocytes and, if so, 2) which step of iNOS induction including NF-B activation is influenced by PAO. In the present study, we found that PAO inhibited IL-1␤-induced nitric oxide production by blocking a step of NF-B binding to DNA, but not the translocation of NF-B into the nucleus.
Cultures-Hepatocytes were isolated from male Wistar strain rats (200 -250 g) by collagenase perfusion as described previously (31). All animal experiments were approved by the Animal Care Committee of Kansai Medical University. The isolated hepatocytes were suspended in culture medium at 5.5-6.0 ϫ 10 5 cells/ml, seeded onto plastic dishes (2 ml/dish, 35 ϫ 10 mm; 9 cm 2 , Falcon Plastic, Oxnard, CA), and then cultured as monolayers in a CO 2 incubator (under a humidified atmosphere of 5% CO 2 in air) at 37°C. The culture medium used was Williams' medium E supplemented with 10% newborn calf serum, Hepes (5 mM), penicillin (100 units/ml), streptomycin (0.1 mg/ml), dexamethasone (10 nM), and insulin (10 nM). After 6 -7 h, the medium was replaced by fresh serum-and hormone-free medium (1.5 ml/dish), and the cells were cultured overnight and then used for the experiments. The purity of isolated hepatocytes was greater than 98% by microscopic observation (32). The number of cells attached to the dishes was calculated from the number of cell nuclei (33). The nucleus/cell ratio was 1.37 Ϯ 0.04 (mean Ϯ S.E., n ϭ 7).
Measurement of Nitric Oxide Production-On day 1, cultured hepatocytes were washed with Williams' medium E, and the same medium was added (1 ml/dish). Then, IL-1␤ (1 nM, 347 units/ml) was added to the medium in the absence and presence of PAO, and cells were incubated in a CO 2 incubator. Alternatively, cells were pretreated with PAO for 15 min, washed with the medium, and then incubated with IL-1␤. In our preliminary experiment, we measured the sum of NO 2 Ϫ and NO 3 Ϫ in culture medium; the conditioned medium was reduced with nitrate reductase and then was assayed (NO assay kit; Roche Molecular Biochemicals, Germany). IL-1␤ increased the release of NO 2 Ϫ ϩ NO 3 Ϫ into the medium, where the ratios of NO 2 Ϫ : NO 3 Ϫ are approximately 1:2 (34). We found that PAO inhibited the release of NO 2 Ϫ ϩ NO 3 Ϫ dosedependently but did not change the ratios. Thus, accumulation of nitrite (NO 2 Ϫ ) in the medium was used as a measure of nitric oxide production. Nitrite was determined with the Griess reagent method (35) and sodium nitrite was used as the standard. PAO (2 M) caused no interference at all in the nitrite assay used.
Western Blot Analysis-Cultured hepatocytes were incubated with IL-1␤ in the absence and presence of PAO. After the indicated times, dishes were placed on ice and were washed twice with cold phosphatebuffered saline before being solubilized in 125 mM Tris-HCl buffer, pH 6.8, containing 5% glycerol, 2% SDS, and 1% 2-mercaptoethanol. The lysates were boiled at 100°C for 3 min and centrifuged at 16,500 ϫ g for 5 min. The supernatant was subjected to SDS-polyacrylamide gel electrophoresis and electroblotted onto a polyvinylidene difluoride (Bio-Rad) membrane. Immunostaining was performed using an ECL blotting detection agent (Amersham Pharmacia Biotech, Amersham, Bucks, UK) and rabbit polyclonal antibodies against the COOH-terminal fragments of mouse macrophage iNOS (Affinity-Bio Reagent, Neshanic station, NJ), human IB␣ (C-21), or human IB␤ (C-20) (Santa Cruz Biotechnology, Santa Cruz, CA) as the primary antibody.
Northern Blot Analysis-Total RNA was extracted from cultured hepatocytes after IL-1␤ stimulation in the absence and presence of PAO using the acid guanidinium-phenol-chloroform method (36). Ten micrograms of total RNA were fractionated by 1% agarose-formaldehyde gel electrophoresis, transferred to nylon membranes (Nytran, Schleicher & Schuell, Dassel, Germany), and immobilized by baking (80°C, 2 h) for hybridization with DNA probes. The cDNA probe of iNOS (rat vascular smooth muscle cells; RT-PCR product (830 bp) amplified with YNO12 and YNO56 primers (37)) was provided by Dr. Y. Nunokawa (Suntory Institute for Biochemical Research, Osaka, Japan) and was radiolabeled with [␣-32 P]dCTP by the random-primed labeling method.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared according to Shreiber et al (38) at 4°C unless otherwise stated. Briefly, approximate 2 ϫ 10 6 cells (two dishes) were placed on ice, washed with Tris-buffered saline, harvested with the same buffer by a rubber policeman, and centrifuged at 1,840 ϫ g for 1 min. The cells in a centrifuge tube were resuspended in 400 l of lysis buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 500 units/ml aprotinin, and 0.5 mM phenylmethylsulfonyl fluoride) and allowed to swell on ice for 15 min, and then 25 l of 10% Nonidet P-40 in the lysis buffer was added. Then, the tube was vigorously vortexed for 1 min at room temperature and centrifuged at 15,000 ϫ g for 1 min. After removal of the supernatant, the nuclear pellet was resuspended in 75 l of nuclear extraction buffer (20 mM Hepes, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 500 units/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride, DTT-free nuclear extraction buffer) or the buffer with 1 mM DTT. The tube was incubated on ice for 20 min with continuous mixing and centrifuged at 15,000 ϫ g for 5 min. Aliquots of the supernatant were frozen with liquid nitrogen and stored at Ϫ80°C until use.
Binding reactions (15 l total) were performed by incubating 4 g of protein of the nuclear extract with reaction buffer (20 mM Hepes, pH 7.9, 1 mM EDTA, 60 mM KCl, 10% glycerol, 1 g of poly(dI-dC)) in the absence and presence of cold probe as competitor (500-fold excess) for 30 min, followed by a 20-min incubation at room temperature with the probe (approximately 20,000 cpm). Products were electrophoresed at 100 V on a 4.8% polyacrylamide gel in high ionic strength buffer (50 mM Tris-HCl, 380 mM glycine, 2 mM EDTA, pH 8.5), and dried gels were analyzed by autoradiography. NF-B consensus oligonucleotide (5Ј-AGTTGAGGGGACTTTCCCAGGC) from mouse immunoglobulin light chain was purchased (Promega, Madison, WI, USA) and labeled with [␥-32 P]ATP and T4 polynucleotide kinase. In experiments with the treatment of PAO in vitro, nuclear extracts were prepared with DTTfree nuclear extraction buffer and incubated with PAO in the absence and presence of DTT or thioglycerol for 30 min at room temperature before mixing with the probe. The protein concentration was measured by the method of Bradford (39) with a dye binding assay kit (Bio-Rad) using bovine serum albumin as a standard.
Transfection and Luciferase Assay-Hepatocytes were cultured at 4 ϫ 10 5 cells per dish (35 ϫ 10 mm) in Williams' medium E supplemented with 10% serum, dexamethasone (10 nM), and insulin (10 nM) for 1 day. Then, cultured cells were replaced by the medium without serum and hormones and were subjected to transfection with 16 g of LipofectAMINE (Life Technologies) and 2 g of plasmid DNA: 1.5 g of pRNOS-Luc; and 0.25 g of CMV enhancer/promoter-driven ␤-galactosidase expression plasmid pCMV-LacZ as an internal control and 0.25 g of plasmid expressing NF-B (0.125 g of each pCMV-p65 and pCMV-p105). After a 5-h incubation at 37°C, the medium was replaced by fresh medium supplemented with 10% serum, dexamethasone, and insulin. On day 2, the cells were treated with PAO or the vehicle (0.1% dimethylsulfoxide) for 15 min, washed with the medium, and incubated for 8 h. Activities of luciferase and ␤-galactosidase were measured as reported previously (20). The luciferase activity was normalized by dividing relative light units by ␤-galactosidase activity. Transfection was performed in triplicate.
Statistical Analysis of Data-Statistical significance was analyzed by Bonferroni/Dunn's test, and a p Ͻ 0.05 was considered to be statistically significant.

Inhibition of Nitric Oxide Production by Arsine
Oxide Derivative in Hepatocytes-As reported previously, an inflammatory cytokine IL-1␤ increased levels of nitrite (nitric oxide metabolite) released to the medium in primary cultures of rat hepatocytes, where concentrations for the maximal and half-maximal effects were 1 nM and 30 pM, respectively (8). Simultaneous addition of PAO, a trivalent arsenical compound, inhibited the nitritereleasestimulatedbyIL-1␤inthetime-andconcentrationdependent manner (Fig. 1A). Concentration for the maximal inhibition was 2 M (Fig. 1B). We also observed the similar effect when PAO was pretreated for 15 min and was removed from the medium before experiments instead of simultaneous addition (data not shown). Thus the latter condition was used in the following experiments. Both trypan blue exclusion and lactate dehydrogenase release tests implied that drug toxicity was not a cause of the decreased production of nitric oxide in response to IL-1␤.
Inhibition of the Induction of iNOS Protein and Its mRNA by PAO-IL-1␤ induced the formation of iNOS protein, which has a calculated molecular mass of 130 kDa (37), with a maximum at 8 -12 h as reported (22,34). PAO inhibited the induction of iNOS protein (Fig. 2). Furthermore, PAO inhibited the induction of iNOS mRNA, which appeared at 3 h after addition of IL-1␤ and increased to a maximum at 6 -8 h (Fig. 3). The results indicated that PAO inhibited IL-1␤-stimulated nitric oxide production at a transcription step.

Blockade of PAO-induced Inhibitory Effects on IL-1␤-induced iNOS Induction by Vicinal Dithiol Compound-PAO is known
to react with two thiol groups of closely spaced protein cystenyl residues to form stable dithioarsine rings (29). Interaction between PAO and vicinal dithiol-containing proteins cannot be competed by monothiols, but in the presence of low molecularweight dithiols such as 2,3-dimercaptopropanol and 1,4-dithiothreitol (DTT) the binding is competed. PAO (2 M) was treated in the presence of dithiol, 2,3-dimercaptopropanol (50 M), or monothiol, 3-mercapto-1,2-propanediol (thioglycerol, 50 M). The 2,3-dimercaptopropanol, but not thioglycerol, abolished PAO-induced inhibitions on the production of nitric oxide and inductions of iNOS protein and its mRNA (Fig. 4). In the case of monothiols, even larger excess of thioglycerol (250 M) had no effect at all (data not shown).
Effects of PAO on IB Degradation and NF-B Activation-IL-1␤ stimulated a rapid degradation of inhibitory subunits, IB␣ and IB␤ proteins, of NF-B which were recovered within 1.5 and 4 h, respectively, after addition of IL-1␤. PAO had no effect on both IB degradation (data not shown). However, EMSA revealed that PAO markedly decreased NF-B band in nuclear extract prepared without DTT (Fig. 5) although PAO had no effect on IL-1␤-induced NF-B translocation in nuclear extract prepared with DTT. In EMSA experiments for DNA binding activity of NF-B, we used to prepare nuclear extracts in the presence of DTT. Thus, NF-B band in the nuclear extract without DTT had lower intensity than that with DTT in control (IL-1␤ without PAO) (Fig. 5). In the case of nuclear extract with DTT, it could be possible that DTT reversed the binding of PAO to vicinal dithiol-containing residue of NF-B molecule during the nuclear extraction. These results indicated that PAO could not influence the translocation of NF-B from the cytoplasm into the nucleus following IB degradation but could prevent a DNA binding of NF-B, resulting in the blockade of NF-B activation induced by IL-1␤ in hepatocytes.
Evidence for Interaction between PAO and NF-B-In the next, after NF-B activation by IL-1␤ in the absence of PAO, nuclear extracts were prepared without DTT, and then PAO were incubated in nuclear extracts in the absence or presence of DTT or thioglycol in vitro. Addition of exogenous PAO again decreased NF-B band in the absence of DTT but not in the presence of DTT (Fig. 6). We also found the similar decrease of NF-B band in the nuclear extract with monothiol, thioglyc- erol. Transfection experiments with iNOS promoter construct and p50/p65, subunits of NF-B, revealed that PAO inhibited the transactivation of iNOS promoter concentration dependently, and the inhibition was significantly blocked in the presence of dithiol, 2,3-dimercaptopropanol (Fig. 7). DISCUSSION In the present study, we found that arsine oxide derivative PAO inhibited the production of nitric oxide stimulated by IL-1␤ both time and concentration dependently (Fig. 1). This inhibition is because of the suppression at a step of its transcription, as both inductions of iNOS mRNA and its protein were abolished by PAO (Figs. 2 and 3). Simultaneous addition of low molecular weight dithiol, 2,3-dimercaptopropanol, completely blocked these inhibitory effects (Fig. 4), indicating that PAO interacted with vicinal dithiol-containing molecule(s) and inactivated its function, which is presumably involved in the pathway of iNOS induction in hepatocytes. Among candidates, NF-B seems likely to be a predominant molecule having the interaction with PAO, which is supported by the observation as follows: (i) EMSA experiments demonstrated that PAO inhibited the DNA binding of NF-B, but did not inhibit the NF-B translocation to the nucleus (Fig. 5), which was consistent with the result that PAO had no effect on the IB degradation (data not shown); (ii) in vitro incubation of PAO with nuclear extracts revealed that PAO inhibited the interaction between NF-B and the B consensus oligonucleotide (Fig. 6); and (iii) in trans-fection experiments with iNOS promoter-luciferase, PAO inhibited the iNOS promoter activity stimulated by p50/p65, subunits of NF-B (Fig. 7).
Thus, our results suggested that PAO rather directly binds to NF-B through its vicinal dithiol moiety. In support of this observation, Shumilla et al. (28) reported that arsenite as well as other thiol reactive metals inhibited NF-B binding to DNA in vitro with nuclear extracts isolated from TNF␣-treated A549 cells, through interaction with critical protein sulfhydryls, which was reversed by addition of dithiols, DTT. Kim and Stadtman (41) reported the similar mechanism for inhibitions of NF-B DNA binding and nitric oxide induction by selenite in human Jurkat T cells and lung adenocarcinoma cells. Selenite inhibited NF-B activation via adduct formation with the essential thiols of this transcription factor. Selenite inhibition was also reversed by addtion of DTT. Inhibition because of formation of such selenotrisulfide type of adduct is established as the mechanism of selenite action in the case of rat brain prostaglandin D synthase (42).
It has been reported that PAO is an inhibitor of a specific class of tyrosine phosphatases characterized by two vicinal thiol groups in the active site (43)(44)(45). Singh and Aggarwal (46) reported that PAO (2.4 M) inhibited the degradation of IB protein and NF-B activation induced by TNF␣ in human myeloid ML-1a cells, which were reversed by DTT, suggesting a critical role of sulfhydryl group. They concluded that PAO presumably inhibited tyrosine phosphatase, which occurs in the cytoplasm and is involved in the pathway leading to the activation of NF-B. They had no mention about interaction between PAO and NF-B. We cannot exclude the possibility that PAO inhibited tyrosine phosphatase in our study, which occurs in the nucleus and is involved in NF-B DNA binding. However, it seems that different types of cells have alternative pathways for the activation of NF-B (47). PAO may attack different targets which have a vicinal dithiol moiety and are involved in NF-B activation. Tewes et al. (48) reported that PAO inhibited IL-1-induced activation of IL-1R1 associated protein kinase and impaired the activation of NF-B in a murine T cell line, EL4 cells.
In our primary cultured hepatocytes, we found that pretreatment of sodium arsenite (20 M for 15 min) also markedly inhibited the production of nitric oxide and the induction of iNOS protein and mRNA by IL-1␤ as similar as PAO did. However, addition of DTT (500 M) could not reverse the inhibition. Furthermore, sodium arsenite had no effect on iNOS promoter activation with p50/p65 at all, 2 implying that PAO and sodium arsenite inhibits iNOS induction via different mechanisms. de Vera et al. (27,49) reported that sodium arsenite inhibited iNOS expression induced by a mixture of cytokines and lipopolysaccharide in rat and human hepatocytes through the induction of Hsp-72, which possibly attenuated the transcription level via the inhibition of NF-B. Although a precise role of vicinal dithiol residue in NF-B activation is obscure on the induction of iNOS in hepatocytes, our study demonstrates that vicinal dithiol residue(s) which interacts with PAO may be an essential component for the binding to iNOS promoter region. Thus, maintenance of active thiol forms by the thioredoxin-thioredoxin reductase system specifically, in part, determine NF-B activation on the induction of iNOS in hepatocytes. In support of our observation, Vos et al. (50) reported that glutathione depletion prevented iNOS induction in hepatocytes but not in inflammatory cells including neutrophils. PAO may attack the multi-molecules which are regulated by changes in the redox state.