Regulation of the physiological effects of peroxidovanadium(V) complexes by the electronic nature of ligands

Journal of Inorganic Biochemistry 121 (2013) 66–76

Contents lists available at SciVerse ScienceDirect

Journal of Inorganic Biochemistry journal homepage: www.elsevier.com/locate/jinorgbio

Regulation of the physiological effects of peroxidovanadium(V) complexes by the electronic nature of ligands Hironori Sugiyama a, Seiichi Matsugo b, c, Hirofumi Misu d, Toshinari Takamura d, Shuichi Kaneko d, Youhei Kanatani e, Mikako Kaido e, Chie Mihara e, Nilka Abeywardana e, Ayana Sakai e, Kyouhei Sato a, Yoshitaro Miyashita f, Kan Kanamori a,⁎ a

Advanced Nanosciences and Biosciences, Graduate School of Innovative Life Science, University of Toyama, 3190 Gofuku, Toyama 930-8555, Japan Faculty of Natural System, Institute of Science and Engineering, Kanazawa University, Kakuma-machi, Kanazawa, Ishikawa 920-1192, Japan c CREST, Japan Science and Technology Agency (JST), Chiyoda-ku, Tokyo 102-0075, Japan d Department of Disease Control and Homeostasis, Kanazawa University Graduate School of Medical Sciences, 13-1 Takara-machi, Kanazawa, Ishikawa 920-8641, Japan e Department of Chemistry, Faculty of Science, University of Toyama, 3190 Gofuku, Toyama 930-8555, Japan f Department of Applied Chemistry, Kobe City College of Technology, 8–3 Gakuenhigashimachi, Nishi-ku, Kobe 651-2194, Japan b

a r t i c l e

i n f o

Article history: Received 3 September 2012 Received in revised form 18 December 2012 Accepted 18 December 2012 Available online 8 January 2013 Keywords: Vanadium Peroxido complex Cytotoxicity Insulin signaling Reactive oxygen species

a b s t r a c t Although the physiological effects of peroxidovanadium(V) complexes (pVs) have been extensively investigated both in vitro and in vivo with regard to their pharmacological activity, such as insulin-mimetic and antitumor activities, the relationship between the chemical and pharmacological properties of pVs is still unclear. Rational drug design with pVs depends on a full understanding of this relationship. Toward this end, the current report evaluates the physiological effects of 13 pVs were evaluated bound to a variety of ligand. Six of these ligands are tripodal tetradentate ligands, one is a linear tetradentate ligand, one boasts two pendant groups, three are tridentate ligands, and two are alkoxido-bridging, dinucleating ligands. The cytotoxicities of these pVs could be classified into three groups: significantly toxic, moderately toxic, and non- or negligibly toxic. Further, IC50 values could be related with the LMCT transition energies of the peroxido group, particularly among complexes with similar ligands. This relation indicates that the electronic properties of the peroxido group affected the physiological activity of the pV complex. We also investigated the insulin-signaling intensity of each pV. Phosphorylation of protein kinase B and extracellular signalregulated kinase 1/2, two major insulin-signaling proteins, was observed after treating cells with pV for 30 min. Phosphorylation was particularly remarkable for complexes that exhibited high cytotoxicity. The present results demonstrate that the toxicity and physiological effects of pVs can be controlled by selecting an appropriate ancillary ligand. These findings provide a guide for synthesis of new pVs that may be used as candidate therapeutic agents. © 2013 Elsevier Inc. All rights reserved.

1. Introduction In recent years, several metal complexes, including those of molybdenum [1], cobalt [2], chromium [3], tungsten [4], zinc [5], and vanadium [6–9], have been shown to exert insulin-mimetic effects. The vanadium complexes have been studied in particular detail both in vivo and in vitro. The vanadium(V) ion can form a variety of peroxidovanadium(V) complexes (pVs). The physiological properties of pVs have recently attracted a great deal of attention as potential therapeutic drugs with various physiological activities, including antitumor effects in higher animals such as mice [10], insulin-mimetic effects [11–14], and inhibitory effects on tyrosine phosphatase (PTPase) [15] and angiogenesis [16]. ⁎ Corresponding author. Tel./fax: +81 76 445 6609. E-mail address: [email protected] (K. Kanamori). 0162-0134/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jinorgbio.2012.12.014

The insulin-mimetic effect and its mechanism have been investigated intensively by several groups using various vanadium(V) complexes with one or two peroxido groups. Insulin-mimetic effects of several bisperoxidovanadium(V) (bpV) and monoperoxidovanadium(V) (mpV) complexes containing 1,10-phenanthroline (phen) [17], 2,2′-bipyridine (bpy) [17], picolinate (pic) [18,19], and oxalate (ox) [20], or their derivatives [11,12,17,19] as ancillary ligands have also been identified. These complexes activate the insulin receptor (IR) of hepatoma cells [11], stimulate lipogenesis in adipocytes [11], and inhibit the dephosphorylation of autophosphorylated IR and epidermal growth factor receptors (EGFR) [11,12]. The nature of the ligand was investigated in intravenous (i.v.) administrations of K[VO(O2)2(phen)]⋅3H2O (bpV(phen)) and K2 [VO(O2)2(pic)]⋅H2O (bpV(pic)) [12]. BpV(pic) was more effective than bpV(phen) in terms of IRK activation in skeletal muscle. The authors concluded that bpV(pic) was able to penetrate muscle tissue more easily than bpV(phen).

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

The antitumor activities, toxicities, and insulin-mimetics of pVs have been studied in a leukemia mouse model [10]. Antitumor activity and toxicity depended on the nature of the ancillary ligand in the complex. Since strong antitumor effects were observed only with peroxido complexes, it has been suggested that intramolecular electron transfer between the peroxido group and the metal center influences the degree of antitumor activity. The cytotoxicity of bpV(phen) was investigated in rat pheochromocytoma cell lines [21,22], rat insulinoma cells [23], human ovarian carcinoma cell lines [22], and human cervical carcinoma cells [22]. In rat pheochromocytoma [21,22] and rat insulinoma cells [23], bpV(phen) induced strong and sustained c-Jun N-terminal kinase (JNK) activation. In contrast, activation of JNK was not observed in human ovarian carcinoma cells [22] and human cervical carcinoma cells [22] following bpV(phen) treatments. While the level of expression of mitogen-activated protein kinase (MAPK) phosphatase-1 (MKP-1) was suppressed after bpV(phen) treatment, bpV(phen) did not stimulate proteolytic processing of procaspase-3 in rat insulinoma cells [23]. These results indicate that caspase-3 is not activated and that MAPK and MKP-1 have roles in bpV (phen)-induced apoptosis. The effects of bpVs (bpV(phen) and bpV(pic)) on angiogenesis inhibition have been investigated using human endothelial cells [16]. Angiogenesis is a complex process involving the activation of endothelial cells through the triggering of several intracellular signaling pathways, including tyrosine phosphorylation. A reaction mechanism has been proposed for the inhibition of PTPase by pVs [15,24]. It is likely that vanadate and peroxidovanadate inhibit PTPase differently. Although PTPase inhibition is required for insulin-mimetic action, compounds with PTPase activity do not always exhibit insulinmimetic effects [15]. The relationship between cell death and insulin signaling in PTPase inhibition was investigated using hydrogen peroxide [25,26]. PTPase is inactivated by reactive oxygen species (ROS), such as hydrogen peroxide, leading to cell death through activation of ERK 1/2. However, insulin signaling is regulated by the concentration of hydrogen peroxide. At low concentrations, insulin signaling was enhanced by PTPase inhibition. At higher concentrations, insulin signaling was attenuated by the activation of JNK. Although a variety of physiological effects of pVs and other related chemicals have been reported, the relationship between cytotoxicity and these physiological effects is not clear. pVs have shown considerable promise as therapeutic agents. To design and synthesize pVs suitable for use as drugs, it is necessary to clarify the relationship between their chemical properties and their pharmacological actions. To establish this relationship, the physiological properties of a variety of complexes bearing different functional groups and numbers of coordination sites must be determined. In our laboratory, we have synthesized a number of peroxidovanadium(V) complexes and have examined their structures and properties in aqueous solution [14,27–30]. In general, mpV with an organic ancillary ligand adopts a pentagonal bipyramidal structure with seven coordination bonds. The peroxido group is situated in a pentagonal plane with the oxido group occupying the axial position. The stability of pVs in solution depends very much on the ancillary ligand and the acidity of the solution. For example, [VO(O2)(nta)] 2− (nta: nitrilotriacetic acid) is extremely stable even in acidic solutions and has been kept intact for more than 4 years [31]. Conversely, [{VO(O2)}2(dpot)] 3 − (dpot: 1,3-diamino-2-propanol-N,N,N′,N′tetraacetate) decomposes gradually in acidic solutions (pH 4.51) to yield a vanadium(IV) complex, although the complex is moderately stable at neutral pH [30]. In the present study, we evaluated cytotoxicity and insulin signaling by treating H4IIEC3 rat hepatoma cells with 13 different pVs to clarify the relationship between their chemical properties and pharmacological effects. The possible role of ancillary organic ligands was of particular interest. Preliminary results have been published elsewhere [28].

67

2. Experimental 2.1. Materials Cell cultures were prepared using the following reagents. Fetal bovine serum (FBS) and Dulbecco's modification of Eagle's Medium (DMEM) were obtained from Gibco. Solutions of 100 U/mL penicillin, 0.1 μg/mL streptomycin, 0.05% (w/v) trypsin, 0.53 mM EDTA⋅4Na, 200 mM L-glutamine, and phosphate-buffered saline (PBS) were obtained from Wako Pure Chemicals. 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay kits were obtained from Roche. Pretreatment was performed in accordance with the manufacturer's instructions prior to making measurements. A Multiscan FC microplate reader (Thermo Scientific) was used to measure solution absorbance at 595 nm. Western blots were performed using the following reagents and antibodies: Radio-Immunoprecipitation Assay (RIPA) buffer was obtained from Millipore in Billerica, MA. Complete™ Protease Inhibitor Cocktail Tablet, EDTA-free and PhosSTOP Phosphatase Inhibitor Cocktail Tablets were obtained from Roche Diagnostics. Electrophoresis was performed in Super Sep™ Ace 5–20% polyacrylamide gel, obtained from Wako Pure Chemicals. Block ace was obtained from DS Pharma Biomedical Co., Ltd. GE ECL Prime Western blotting detection reagent was obtained from GE Healthcare Bio-Sciences, Piscataway, NJ, USA. Phospho-Akt (Ser 473), Phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204), Akt (total) and p44/42 MAPK (ERK1/2) (total) antibodies and anti-rabbit antibodies conjugated to horseradish peroxidase were obtained from Cell Signaling Technology. An iBlot Gel Transfer Device (Life Technologies Japan Ltd. Tokyo, Japan) was used for transfers to polyvinylidene difluoride (PVDF) membranes. A Lumi Cube CMOS SLR camera system (Liponics, Inc., Tokyo, Japan) was used to detect chemiluminescence. 2.2. Preparation of ligands H3 hpnbpda (2-hydroxypropane-1,3-diamino-N,N′-bis(2pyridylmethyl)-N,N′-diacetic acid) [27], Himala (N-imidazolylmethyl-Lalanine) [28], Himphe (N-imidazolylmethyl-L-phenylalanine) [28], Bapda (H2pda = N-pyridylmethyliminodiacetatic acid) [29], Hbpg (N,N-bis(pyridylmethyl)glycine) [29], and H4edds (S,S′-1,2ethanediamine-N,N′-disuccinic acid) [32] were prepared according to methods published previously. H3nta (nitrilotriacetic acid), H2edda (1,2-ethanediamine-N,N′-diacetic acid), H5dpot (1,3diamino-2-propanol-N,N,N′,N′-tetraacetic acid), and H3ceida (N-(2-carboxyethyl)iminodiacetic acid) were obtained commercially and used as received. 2.2.1. N,N-bis(2-imidazolylmethyl)glycine (Hbimgly⋅H2O⋅ 2HCl) 20% Et4NOH solution (4.6 g, 6 mmol) was added to a solution of glycine (0.38 g, 5.0 mmol) in 10 mL of water. The solution was evaporated to dryness under reduced pressure and the residue was dissolved in 20 mL of methanol. A methanolic solution of imidazole carboxyaldehyde (0.96 g, 10 mmol) was added and the solution was acidified to pH 4.5 by addition of glacial acetic acid. The solution was stirred at room temperature for 15 min. To this white muddy suspension, sodium cyanoborohydride (0.32 g, 5 mmol) in 20 mL of methanol was added dropwise for 30 min. After this addition, a clear yellow solution was obtained and stirred at room temperature for 5 days. The final pH of the solution was 5.6. The solution was evaporated to dryness under reduced pressure, and 20 mL of water was added to the residue. The solution was acidified to pH 0.5 by adding conc. HCl and washed three times with 20 mL of chloroform. The aqueous layer was concentrated to dryness with a rotary evaporator. The crude product was washed with acetone (10 mL) three times and 20 mL of methanol was added. A white precipitate was collected by filtration, washed with methanol, and air-dried. Yield 0.93 g (57%). Anal. Calcd for C10H17N5Cl2O3: C, 36.82; H, 5.26; N, 21.47%. Found: C, 36.95; H, 5.06; N, 21.21%. IR (KBr disk)/

68

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

cm−1: 1476, 1397, 1102, 833, 624. 1H NMR/ppm: 4.01: im-CH2-N (4H singlet), 3.44: N-CH2-COOH (2H singlet), 7.37: N-CimH-C (2H singlet), 8.57: N-CimH-N (2H singlet). 2.3. Preparation of complexes [VO(O2)(imala)(H2O)]⋅2H2O [28], [VO(O2)(imphe)]⋅0.5H2O [28], NH4[VO(O2)(ida)] [33], K2[VO(O2)(nta)]⋅2H2O [34], Cs[VO(O2)(pda)] ⋅2H2O [14], [VO(O2)(bpg)] [35], [VO(O2)(Hheida)] [35], and Cs3 [(VO)2(O2)2(dpot)]⋅3H2O [30] were prepared according to methods published previously. 2.3.1. Na[VO(O2)(edda)]⋅H2O Vanadium pentoxide (0.54 g, 3.0 mmol) was suspended in 7 mL of 1 M NaOH. The suspension was stirred while heating to yield a clear solution. H2edda (1.05 g, 6.0 mmol) and 10% hydrogen peroxide (2.04 g, 6.0 mmol) were added, producing a red solution. Ethanol was added until the solution became cloudy. The solution was kept in a refrigerator for 1 day, resulting in the precipitation of orange plate crystals. Yield: 1.27 g (67.4 %). Anal. Calcd for C6H12N2NaO8V: C, 22.94; H, 3.85; N, 8.92%. Found: C, 22.89; H, 3.70; N, 8.84%. IR (KBr disk)/cm−1: 1625 (νas(CO2)), 1378 (νs(CO2)), 962 (ν(V O)), 929 (ν(O―O).) 2.3.2. Cs2[VO(O2)(Hedds)]⋅3.5H2O To a suspension of vanadium pentoxide (0.36 g, 2.0 mmol) in 4 mL of water, 2.0 mL of 4 M cesium hydroxide (4 ml, 8 mmol) was added. The mixture was stirred while heating to yield a clear solution. H4edds (1.16 g, 4.0 mmol) was then added and the pH of the solution was adjusted to 9.8 by addition of 4 M cesium hydroxide. After stirring for 30 min, the pH of the solution was adjusted to 4.3 using 1 M hydrochloric acid. After the insoluble material had been filtered off, cesium chloride (3.4 g, 12 mmol) and then 10% hydrogen peroxide (1.36 g, 4 mmol) were added to the solution in an ice bath. The solution was then stirred for 1 h in an ice bath. The resulting red solution was evaporated to about half of its initial volume under reduced pressure. Ethanol was added until the solution became cloudy. After 3 weeks of storage in the refrigerator, orange crystals were collected by filtration, washed with ethanol, and air-dried. Yield: 1.07 g (36.9%). Anal. Calcd for C10H20N2Cs2O14.5 V: C, 16.75; H, 2.81; N, 3.91%. Found: C, 16.56; H, 3.48; N, 3.91%. IR (KBr disk)/cm −1: 1667 (νas(CO2)), 1372 (νs(CO2)), 937 (ν(V O)), 923 (ν(O―O).) 2.3.3. K2[VO(O2)(ceida)]⋅H2O Vanadium pentoxide (0.36 g, 2.0 mmol) was added to 1 M potassium hydroxide solution (8.0 ml, 8.0 mmol). The resulting suspension was stirred while heating to yield a clear solution. N-(2-carboxyethyl) iminodiacetic acid (0.82 g, 4.0 mmol) was added to the solution at room temperature, to give an orange solution. After the solution being cooled in an ice bath, 10% hydrogen peroxide (1.36 g, 4.0 mmol) was added to give a dark red solution. Ethanol was then added until the solution became cloudy. After storage in the refrigerator, orange crystals were collected by filtration, washed with ethanol, and air-dried. Yield: 0.64 g (40.3 %) Anal. Calcd for C7H10NK2O10V: C, 21.16; H, 2.54; N, 3.53%. Found: C, 21.62; H, 2.49; N, 3.47%. IR (KBr disk)/cm−1: 1646 (νas(CO2)), 1374 (νs(CO2)), 962 (ν(V O)), 924 (ν(O-O).) 2.3.4. Na[{VO(O2)}2 (hpnbpda)]⋅1.5H2O Vanadium pentoxide (0.36 g, 2.0 mmol) was suspended in 30 mL of water and solid sodium hydroxide (0.19 g, 4.8 mmol) was added. The suspension was stirred for 1 h with heating to yield a clear solution. H3hpnbpda⋅HBr⋅3H2O (1.25 g, 2.4 mmol) was added and the pH of the solution was adjusted to ca. 9 using aqueous NaOH. After cooling in an ice bath, 10% hydrogen peroxide (1.38 g, 4 mmol) was added while stirring to yield a red solution. The solution was stirred for a further 1 h. The pH was then adjusted to 3.0 by addition of 60% perchloric acid. Twenty milliliters of ethanol was added to the red solution and the

solution was kept in a refrigerator. A red, powdery precipitate was collected by filtration, washed with ethanol, and air-dried. Yield: 0.76 g (62.8%) Anal. Calcd for C19H24N4NaO12.5V2: C, 36.03; H, 3.82; N, 8.85%. Found: C, 35.96; H, 4.10; N, 8.87%. IR (KBr disk)/cm−1: 1611 (νas(CO2)), 1412 (νs(CO2)), 981 (ν(V O)). 2.3.5. [VO(O2)(bimgly)]⋅2H2O Vanadium pentoxide (0.18 g, 1.0 mmol) was dissolved in 20 mL of water by adding 2 mL of 1 M NaOH solution with heating and stirring. The resulting clear, yellow solution was stirred at 0 °C. Hbimgly⋅H2O⋅2HCl (1.30 g, 4.0 mmol) was added to the solution and the pH was adjusted to 3.5 using 1 M NaOH followed by the addition of 0.7 mL of 10% H2O2 (2 mmol) to yield an orange solution. The solution was stirred for 2 h an orange, powdery precipitate (impurity) was filtered off. Allowing the filtrate to sit in a refrigerator afforded orange crystals. These were harvested by filtration, washed with ethanol, and air-dried. Yield: 0.16 g (22%). Anal. Calcd for C10H16N5O7V: C, 32.53; H, 4.38; N, 18.97%. Found: C, 32.91; H, 4.22; N, 18.76%. IR (KBr disk)/cm−1: 1632 (νas(CO2)), 1373 (νs(CO2)), 954 (ν(V O)), 921 (ν(O―O).) Safety Note. Although no problems were encountered during concentration of the reaction solution, solutions containing peroxides and/ or perchloric acid are potentially explosive and should be handled in small quantities and with care. 2.4. Measurements UV-visible (UV-vis) spectra were recorded using a Shimadzu UV3100PC spectrophotometer. IR spectra (KBr disk) were measured using a JASCO FT-IR8000 spectrometer. 1H and 51V NMR spectra (D2O solution) were recorded on a JEOL JMN-ECX300 (300 MHz) spectrometer. 2.5. X-ray structure determination A single crystal of the complex being measured was mounted onto a glass fiber, coated with epoxy as a precaution against solvent loss, and centered on a Rigaku Mercury CCD area detector for Na [VO(O2)(edda)]⋅H2O and K2[VO(O2)(ceida)]⋅H2O, a Rigaku AFC7R for Cs2[VO(O2)(Hedds)]⋅4H2O, or a Rigaku variMax RAPID-DW diffractometer for [VO(O2)(bimgly)]⋅ 2H2O using graphitemonochromated Mo Kα radiation (λ = 0.71070 Å) or confocal mirror loaded Cu Kα radiation (λ = 1.54187 Å for the bimgly complex). Lorentz, polarization, and linear decays correction, and the empirical absorption corrections on a series of psi scans were carried out. Molecular structures were solved by a direct method SIR2004 [36] for the edda and ceida complexes or SHELX97 [37] for the edds and bimgly comples, and by conventional difference Fourier techniques. The structures were refined by full-matrix least-squares techniques on F2. All non-hydrogen atoms, except the oxygen atoms of disordered water molecules in the ceida complex, were refined anisotropically. Hydrogen atoms of the complexes were located at calculated positions and were refined using the riding model while those of crystal solvents were not included in the calculations. All calculations were performed using CrystalStructure software [38]. Other crystallographic data are summarized in Table 1. The crystallographic data have been deposited with the Cambridge Crystallographic Data Centre (Nos. CCDC-898451(edds), 898452(bimgly), 898453(edda), and 898454(ceida)). Copies of the data can be obtained free of charge on application to CCDC, 12, Union Road, Cambridge, CB2 1EZ, U.K. (fax: +44 1223 336033); e-mail: [email protected]. 2.6. Cell culture Rat hepatoma H4IIEC3 cells were obtained from the American Type Culture Collection (ATCC number: CRL-1600). H4IIEC3 cells were cultured in DMEM supplemented with 10% FBS, penicillin (100 U/mL), and streptomycin (0.1 mg/mL) at 37 °C in a humidified atmosphere containing 5% CO2.

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

69

Table 1 Crystallographic data.

Formula Formula wt. Crystal system Space group a/Å b/Å c/Å β/° V/Å3 Z Dcalcd/g cm−3 Crystal size/mm λ/Å μ/cm−1 Temp./K Reflections Total Unique Observed (I > 2.0σ(I)) F(000) No. of variables R (I > 2.0σ(I)) Rw (I > 2.0σ(I)) GOF a

Na[VO(O2)(edda)]⋅H2O

K2[VO(O2)(ceida)]⋅H2O

Cs2[VO(O2)(Hedds)]⋅3.5H2O

[VO(O2)(bimgly)]⋅2H2O

C6H12N2NaO8V 314.10 monoclinic P 21/c (#14) 11.2332 (4) 7.2594 (2) 13.8677 (5) 93.8805 (5) 1128.27 (7) 4 1.849 0.40 × 0.30 × 0.15 0.71070 (Mo Kα) 9.54 (Mo Kα) 296

C7H10K2NO10V 397.29 monoclinic P 21/c (#14) 11.8137 (4) 7.4667 (2) 16.0589 (5) 102.5021 (5) 1382.96 (7) 4 1.908 0.25 × 0.15 × 0.05 0.71070 (Mo Kα) 13.69 (Mo Kα) 296

C10H20Cs2N2O14.5V 717.02 monoclinic P 21/c (#14) 9.976 (4) 18.062 (5) 11.993 (4) 97.36 (3) 2143.2 (11) 4 2.222 0.40 × 0.30 × 0.10 0.71069 (Mo Kα) 38.82 (Mo Kα) 200

C10H16N5O7V 369.21 orthorhombic P 212121 (#19) 7.55612 (14) 13.0112 (3) 14.2999 (3)

8937 2582 2371 640 173 0.038 0.126 1.009

10974 3151 2712 800 197 0.040 0.125 1.004

6536 6215 4318 1372 271 0.093a 0.281a 1.050

16120 2541 2387 760 208 0.028 0.081 1.143

1405.88 (5) 4 1.744 0.15 × 0.15 × 0.11 1.54187 (Cu Kα) 63.84 (Cu Kα) 173

The high R indices are probably due to low quality of the crystal. The structure of complex anion was deemed satisfactory.

2.7. Cell treatments

were subjected to direct densitometric analyses using a CMOS SLR camera system and Image J software.

For cell treatments, each complex was prepared in 1 mM DMEM solution. 3. Results and discussion 2.7.1. Evaluation of cytotoxicity Cells were grown in 96-well plates. When 90–100% confluent, the cells were exposed to a starvation state for 24 h, followed by treatment with vanadium complexes for 48 h. Control cells were treated with an identical solution without the vanadium complex. Cell viability was calculated by normalizing the observed absorbance to that of the corresponding control. For additional controls, we also evaluated the cytotoxicity of sodium orthovanadate as an inorganic vanadium(V) salt and a 1:1 mixture of hydrogen peroxide and sodium orthovanadate as a model solution of a monoperoxido vanadium complex without an organic ligand. To quantitatively compare the cytotoxicity of the various chemical species, half-maximal inhibitory concentrations (IC50) were calculated by fitting the observed values to a sigmoidal curve using OriginPro 8.6J software (OriginLab Corporation). 2.7.2. Evaluation of intracellular signaling Cells cultured as above were grown in six-well plates to 90–100% confluence, then serum-starved for 24 h and exposed to the vanadium complexes, recombinant human insulin (as a positive control), or vehicle (as a negative control). After 30 min, the cells were harvested in RIPA buffer containing a Complete™ Protease Inhibitor Cocktail Tablet, and EDTA-free and PhosSTOP Phosphatase Inhibitor Cocktail Tablets. The lysates were then sonicated and insoluble material was removed by centrifugation. Equal amounts of total protein were resolved by sodium dodecyl sulfate–polyacrylamide-gel electrophoresis and transferred to polyvinylidene difluoride membranes for immunoblot analyses. The membranes were blocked for 24 h at 4 °C in a buffer containing 4% aqueous Block Ace and then incubated with a specific primary antibody followed by a horseradish peroxidase-linked secondary antibody. Blots were visualized with GE ECL Prime Western blotting detection reagent according to the manufacturer's instructions. The membranes

3.1. Structures of pVs We examined the physiological functions of 13 pVs, including 5 that were newly prepared. Although physiological studies have been performed on homoleptic or mixed-ligand bpy and phen pVs, [11,12] all of the ligands used herein are derivatives of glycine, a simple biogenic ligand. The purpose of the this study was to identify the possible role(s) of organic ancillary ligands with related structures. Therefore, inorganic pVs, as well as bpy or phen pVs were not evaluated. The X-ray structures of the four pVs evaluated in this work are shown in Fig. 1. Previously determined structures of pVs are shown schematically in Fig. 2, including proposed structures for [VO(O2)(pda)] −, [VO(O2)(imphe)], and [{VO(O2)}2(hpnbpda)] −. The vanadium atom in these complexes adapts a heptacoordinate, pentagonal– bipyramidal structure that is typical for an mpV with an organic ancillary ligand. The peroxido group sits in the pentagonal plane. Although the elemental analysis of the imphe complex indicates a pentagonal pyramid structure without an aqua ligand in the solid state, this complex can be reasonably assumed to adopt a normal pentagonal–bipyramidal structure with an aqua ligand at the sixth coordination site in aqueous solution [28]. In complexes with tridentate ligands (ida, imala, and imphe), there exists a so-called free (substitution labile) coordination site trans to the oxido group, which would be occupied by solvent molecules in solution, where a substrate could interact. We will show later that this free-coordination site affects the cytotoxicity of pVs. Note that the heterocyclic groups pryridine and imidazole in bpg and bimgly, respectively, occupy the pentagonal plane. As a result, the trans position of the oxido group is occupied by a carboxy oxygen or a alkoxido oxygen atom. Although the structures of the pda and hpnbpda complexes could not be determined by X-ray crystallography, we hypothesized that the pyridine ring(s) of these complexes occupy the

70

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

Fig. 1. X-ray structures of four pVs determined in this study. Ellipsoids represent a 50% probability as drawn by ORTEP.

pentagonal plane and adopt similar structures as those of the related complexes, [VO(O2)(bpg)] and [{VO(O2)}2(dpot)] 3−, respectively. Several dimeric complexes containing two monoperoxidovanadium(V) moieties have been prepared and structurally characterized: [V2O2(O2)2(cit)2] 2 − (cit = citrate) [39], [V2O2(O2)(α-hhip)2] 2 − (α-hhip = α-hydroxyhippurate) [40], [V2O2(O2)2(C8H6O3)2] 2 − (C8H6O3 = α-hydroxybenzeneacetate) [41]. In these complexes, the two peroxido groups are situated at opposite sides of the V2O2 core. Conversely, in pVs with an alkoxido-bridging dinucleating ligand (dpot or hpnbpda), the two peroxido groups are situated in vicinal positions where the two peroxido groups can interact with each other. 3.2. Cytotoxicity In a pioneering work, Djordjevic and Wampler [10] examined the toxicity and antitumor activity against L1210 murine leukemia of 11 pVs, including the ida and nta complexes used in the present study, and showed that physiological activity, as measured by increased life span (ILS) of the test animals, depends on the type of ancillary ligand.

The cytotoxicities of the pVs used in the present work are shown in Fig. 3. For comparison, the cytotoxicities of sodium orthovanadate and a 1:1 mixture of hydrogen peroxide with sodium orthovanadate are also shown in Fig. 3. Low-dose treatment of the cells with orthovanadate, with or without peroxide, stimulated cell proliferation and the number of cells reached a maximum in both systems at a concentration range of 3–30 μM. However, these systems exhibited markedly different responses at concentrations above 100 μM. Orthovanadate with peroxide showed an acute reduction in cell viability (ca. 5% at 100 μM). Without peroxide, cell viability decreased gradually in a dose-dependent manner. Acute increases in toxicity have also been observed with the inorganic peroxidovanadium(V) complex [{VO(O2)2}2(μ-O)]4− [10], indicating that a similar inorganic pV species would be formed in a solution containing vanadate and peroxide. Expressions of cytotoxicity were classified into three groups. The first group showed an acute cytotoxicity beyond a certain threshold concentration, as observed with a 1:1 mixture of hydrogen peroxide and sodium orthovanadate (Figs. 3a, b, c, d and e). Each compound exhibited a different cytotoxicity threshold. A 100 μM, 1:1 mixture of hydrogen peroxide and sodium orthovanadate reduced cell viability by ~5%, while

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

71

Fig. 2. Previously determined crystal structures or estimated structures of nine pVs.

treatment with 200 μM of the nta complex reduced cell viability by ~0.7%. Similar behaviors were observed with the ida, imala, imphe, ceida, heida, bimgly, dpot, and hpnbpda complexes. However, all of the IC50 values obtained with the above complexes were greater than that of a 1:1 mixture of hydrogen peroxide and sodium orthovanadate. This agrees with the results of Djordjevic [10] and indicates that incorporation of an organic ligand mitigates the cytotoxicity of pVs. The second group (the pda and bpg complexes) showed moderate cytotoxicity in a dose-dependent manner, similar to the effects of sodium orthovanadate (Figs. 3f, g and h). Similar reductions in cell viability were observed at concentrations greater than 200 μM and 100 μM for the pda and bpg complexes, respectively. The third group, which is comprised of pVs with a linear tetradentate ligand, such as edda and its derivative, edds, showed little to no cytotoxicity (Figs. 3i and j). More than 70% cell viability was observed even when the cells were treated with a concentration of 800 μM. In these complexes, the ancillary ligand completely suppresses the cytotoxicity of the peroxidovanadium(V) complex. Edds differs from edda by the presence of two pendant carboxylate groups. Therefore, it is fair to conclude, in the current case, that the addition of pendant groups has little effect on cytotoxicity. Differences in the structure of the tetradentate ligands (linear and tripodal), however, strongly affected cytotoxicity. To determine if differences in dissolved pV stability are related to the observed differences in cytotoxicity, time-dependent 51V NMR spectra were measured on three representative pVs (the nta complex, which exhibits an acute toxicity, the pda complex, which has a moderate toxicity, and the non-toxic edds complex) in PBS up to 48 h after sample preparation (p. 1 of the SI file). 51V NMR signals were observed at the same positions as observed in water and did not change after 48 h, indicating that the pVs used herein retained their integrity in the test medium.

3.3. Relationship between chemical properties of the pVs and their cytotoxicity IC50 values were compared among the present complexes to identify a relationship between the chemical properties of the pVs and their cytotoxicity. No definite relation was found with regard to formal charges and the nature of functional groups on the pVs. Nevertheless, some minor effects were evident. The 51 V NMR chemical shifts did not show any relation with cytotoxicity (p.2 of the SI file). A modest relationship was observed between IC50 values and the energy of the ligand-to-metal charge transfer (LMCT) band due to the peroxide group. The LMCT transition is formally due to the reduction of vanadium(V) by the peroxide group. Theoretically, the energy of an LMCT transition can be determined by the difference in optical electronegativities between the ligand and the metal atom, |χligand − χmetal|. Thus, lower-energy charge transfer (CT) transitions indicate that the peroxido group more easily releases an electron and/or the vanadium center more easily receives an electron. Fig. 4 shows the relationship between IC50 and the energy of the LMCT band, as determined from visible absorption spectra. The dinuclear complexes (dpot and hpnbpda) had lower energy CT bands and were highly toxic. This low-energy CT is consistent with the reduction of V(V) to V(IV), as observed in acidic solutions of the dpot complex [30]. Among the complexes containing a tetradentate ligand, with the exception of the bimgly complex, a tendency can be recognized between IC50 and CTmax. Lower CTmax values corresponded to stronger cytotoxicities. This result is consistent with the hypothesis proposed by Djordjevic [10], which states that intramolecular electron transfer between the peroxido group and the metal center would affect the antitumor activity of the pV. It is likely that compounds such as radical oxygen species (ROS), generated by the intramolecular electron transfer in pVs, may

72

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

Fig. 3. Cytotoxicities of pVs with tetradentate ligands, a 1:1 mixture of hydrogen peroxide and sodium orthovanadate, and a solution of sodium orthovanadate. “c” in the figure indicates the control. (a) to (e) correspond to systems that exhibited strong cytotoxicities above a threshold dose. (f), (g), and (h) correspond to systems with moderate cytotoxicities. (i) and (j) correspond to complexes with little to no toxicity. (a) 1:1 mixture of Na3VO4 and H2O2, (b) K2[VO(O2)(nta)], (c) K[VO(O2)(Hheida)], (d) K2[VO(O2)(ceida)], (e) [VO(O2)(bimgly)], (f) Na3VO4, (g) Cs [VO(O2)(pda)], (h) [VO(O2)(bpg)], (i) Cs2[VO(O2)(Hedds)], (j) Na[VO(O2)(edda)].

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

Fig. 4. The half-maximal inhibitory concentrations (IC50) of pVs are shown as a function of the CT transition energy. IC50 values were calculated by fitting a logistic function to the observed values. The IC50 value of the edds complex could not be determined because of its low cytotoxicity. The IC50 value of the edda complex was obtained from a linear extrapolation of the available data. PVs with tridentate (▲), quadridentate (■), and dinucleating (●) ligands.

be responsible for the observed cytotoxicity. Differences in cytotoxicity between pVs bearing tripodal tetradentate ligands and those bearing linear tetradentate ligands can be attributed to the different electronic nature of the peroxido group. The reason the bimgly complex exhibits a stronger toxicity than indicated by its CTmax is not clear, but may be due to additional factors, such as cell permeability [42]. Among the complexes with tridentate ligands (ida, imala, and imphe) a similar tendency was discerned between IC50 and CTmax. Complexes with a tridentate ligand were more toxic than those with a tetradentate ligand. The existence of a substitution-labile site in pVs seems to enhance cytotoxicity. Such a site likely plays a significant physiological role (for example, coordination of intracellular substrate) in the expression pathway of cytotoxicity. Although structurally similar, the imphe complex exhibits a higher toxicity than the imala complex. This is likely due to the higher lipophilicity of the phenyl group in the imphe complex, which would facilitate passage of the complex through the cell membrane [28]. In summary, while the trends observed with complexes containing tridentate ligands deviated from those containing tetradentate ligands, both exhibited a modest relationship between IC50 and CTmax. Namely, complexes bearing a peroxido group with an easily removable electron seem to be more toxic. The present results indicate that it is possible to regulate the cytotoxicity of pVs to some extent by selecting appropriate organic ligands. The choice of ligand can influence the position of the LMCT transition, thereby altering the toxicity of the pV complex. This strategy may be employed in the design of peroxidovanadium(V) complexes as therapeutic agents.

3.4. Intracellular signaling Fig. 5 shows the effects of three types of pVs (nta, pda, edda) on insulin-mimetic signaling using H4IIEC3 hepatocytes. We chose these complexes due to their strong cytotoxicity, moderate cytotoxicity, and no cytotoxicity, respectively. After treating cells with the pVs for 30 min, we evaluated the phosphorylation of protein kinase B (Akt) and extracellular signal-regulated kinase 1/2 (ERK1/2), which are major insulin-signaling proteins, at five doses between 3 and 500 μM. At concentrations above 30 μM, all of the complexes induced the phosphorylation of both Akt and ERK1/2 in a dose-dependent manner. This indicates that all of the complexes activate insulin signaling pathways in H4IIEC3 cells.

73

Fig. 5a and b show insulin signaling (a: Akt, b: ERK1/2) following treatment with pV(nta) as a typical complex with a strong cytotoxicity. Akt phosphorylation was observed at low doses, the extent of which approximated that of 500 μM vanadate. The intensity of ERK1/2 phosphorylation by pV(nta) was much stronger than that induced by insulin. Treatment with 500 μM pV(nta) was more than 60 times more effective than insulin. All pVs with significant cytotoxicity tended to strongly phosphorylate ERK1/2. Fig. 5c and d show insulin signaling (c: Akt, d: ERK1/2) after treatment with pV(pda) as an example of a pV with moderate cytotoxicity. At a concentration at which the number of cells decreased, Akt phosphorylation intensity exceeded that induced by insulin. ERK1/2 phosphorylation intensity exceeded that induced by insulin at concentrations greater than 30 μM. However, phosphorylation of ERK1/2 by pV(pda) was weaker than that of pV(nta). pV(pda) and pV(nta) (each 500 μM) treatments were 5.6- and 60-fold, respectively, more effective than insulin. Fig. 5e and f show insulin signaling (e: Akt, f: ERK1/2) after treatment with pV(edda) as a non-toxic complex. Although the intensity of Akt phosphorylation was enhanced in a dose-dependent manner, it was weak compared to insulin at low doses and at 500 μM was comparable to insulin. ERK1/2 phosphorylation intensity exceeded that induced by insulin at concentrations greater than 200 μM. Compared to other complexes, the intensity of ERK 1/2 phosphorylation by pV(edda) was low. At treatment concentrations that decreased cell numbers (nta complex: 100 μM, pda complex: 200 μM, edda complex: 500 μM, see Fig. 3), the phosphorylation level of Akt did not differ from that induced by insulin stimulation. The phosphorylation of ERK1/2, however, was notably greater relative to that induced by insulin. It is notable that this tendency was closely related to the cytotoxicity of the pVs. pVs strongly induced the phosphorylation of ERK1/2 upon treatment with concentrations that decreased the number of cells. Conversely, the phosphorylation level of Akt, which is important in glucose metabolism, was identical to that induced by insulin, even upon treatment with higher doses of pVs. Both vanadate and hydrogen peroxide induce insulin-mimetic signaling; the effect can be enhanced by combining the two [43]. Treatment with a mixture of vanadate and hydrogen peroxide resulted in signaling stronger than that induced by insulin. This is consistent with the fact that PTPase inhibition was enhanced by pV treatments. It is therefore likely that the strong phosphorylation of ERK 1/2 induced by highly toxic pVs is due to a synergistic effect between peroxide and vanadium. Conversely, for pVs with moderate or no cytotoxicity, this synergism would not be detected, because the effect of peroxide would be suppressed by the ancillary ligands. 3.5. Relationship between cytotoxicity and intracellular signaling Fig. 6 shows the degree of insulin signaling after treatment with 100 μM of each pV complex. The more toxic complexes are on the right side of the Figure. Phosphorylation of Akt and ERK1/2 increased with pV toxicity. For highly cytotoxic complexes, the ERK1/2 phosphorylation exceeded that induced by insulin, while the extent of phosphorylation of Akt was comparable to that by insulin. For the moderately or negligibly cytotoxic complexes, the phosphorylation of Akt and ERK1/2 was weak relative to that induced by insulin. From these results, it can be hypothesized that the CT energy of a pV, which is related to its cytotoxicity as shown in Section 3.3, affects intracellular signaling. pVs with lower-energy CT bands tended to more effectively phosphorylate ERK1/ 2. It has been reported that vanadate treatment strongly induces the activation of ERK1/2, leading to apoptosis [44]. In addition, ROS formed in the cell can inactivate PTPase, which would in turn induce apoptosis through activation of ERK1/2 [26]. Therefore, it is likely that the activation of ERK1/2 is closely related to the cytotoxicity of the pVs. Note that all of the pVs that strongly activated ERK1/2 had lower CTmax values, as shown in Fig. 4. This indicates that complexes in

74

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

Fig. 5. Effects of pVs on insulin-related signaling. The data are presented as mean-fold increases over the control±S.E.M. (a) and (b) Akt and ERK1/2 treated with the nta complex. (c) and (d) Akt and ERK1/2 treated with the pda complex. (e) and (f) Akt and ERK1/2 treated with the edda complex. Placement of the samples was as follows: (1): negative control (insulin-), (2): positive control (insulin +), (3): 3 μM, (4): 30 μM, (5): 100 μM, (6): 200 μM, (7): 500 μM. The prefix “p” indicates the phosphorylated form and the prefix “t” indicates the total.

which electron transfer from the peroxido group to the metal center occurs easily induce the activation of ERK1/2. ROS produced by the electron transfer is likely responsible for the high activation rate of ERK1/2. Fig. 7 shows a possible mechanism of pV cytotoxicity. The pVs induce an insulin-mimetic signal pathway in the cell by phosphorylation of a tyrosine kinase receptor. Some of the complexes can enter the cell through the cell membrane. It is likely that interactions between

intracellular substrates and the pVs trigger electron transfer in the pVs. Thus, the complexes in the cell would transfer an electron from the peroxido group to vanadium. As a result, ROS (e.g., a superoxide) would be generated in the cell followed by a decomposition of pV. The ROS would then inactivate PTPase, thereby inhibiting the dephosphorylation of Akt and ERK1/2. Cytotoxicity is then caused by a high degree of ERK1/2 activation. For pVs in which the peroxido group is relatively

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

Fig. 6. Effects of pVs at a dose of 100 μM on insulin-related signaling. Placement of the samples was as follows: 1: negative control (insulin −), 2: positive control (insulin +), 3: edds complex, 4: edda complex, 5: pda complex, 6: imala complex, 7: bpg complex, 8: imphe complex, 9: dpot complex, 10: nta complex, 11: ceida complex, 12: ida complex, 13: hpnbpda complex. Complexes with strong cytotoxicities are shown towards the right of the figure. The prefix “p” indicates the phosphorylated form and the prefix “t” indicates the total.

stable (high energy CT transition), no ROS would be formed in the cell. As a result, phosphorylation of Akt and ERK1/2 would be weak.

phen Hpic bpy LMCT IC50 MTT RIPA FBS PBS DMEM MAPK Akt ERK1/2 JNK PTPase ROS

75

1,10-phenanthroline picolinic acid 2,2′-bipyridine Ligand-to-Metal Charge Transfer Half-maximal Inhibitory Concentrations 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide Radio-immunoprecipitation assay Fetal bovine serum Phosphate-buffered saline Dulbecco's modification of Eagle's medium Mitogen-activated protein kinase Protein kinase B Extracellular signal-regulated kinase 1/2 c-Jun N-terminal kinase Protein tyrosine phosphatase Reactive oxygen species

4. Conclusions The cytotoxicity and intracellular signaling of a variety of pVs was evaluated to clarify the relationship between the chemical properties of pVs and their pharmacological activities. The degrees of cytotoxicity and intracellular signaling depended strongly on the nature of the ancillary ligand. Relationships were discovered between cytotoxicity and the LMCT transition of the pV, and between cytotoxicity and the intensity of intracellular signaling. These results demonstrate that ligand choice can be used to tailor vanadium complexes as therapeutic agents and provide guidelines for the synthesis of useful complexes. Abbreviations H3nta nitrilotriacetic acid H3ceida N-(2-carboxyethyl)iminodiacetic acid H3heida N-(2-hydroxyethyl)iminodiacetic acid Hbimgly N,N-bis(2-imidazolylmethyl)glycine H2pda N-pyrydylmethyliminodiacetatic acid Hbpg N,N-bis(pyridylmethyl)glycine H2edda 1,2-ethanediamine-N,N′-diacetic acid H4edds S,S′-1,2-ethanediamine-N,N′-disuccinic acid H2ida iminodiacetic acid Himala N-imidazolylmethyl-L-alanine Himphe N-imidazolylmethyl-L-phenylalanine H5dpot 1,3-diamino-2-propanol-N,N,N′,N′-tetraacetic acid H3hpnbpda 2-hydroxypropane-1,3-diamino-N,N ′-bis(2-pyridylmethyl)-N,N′-diacetic acid

Fig. 7. A possible mechanism of the induction of intracellular signaling in H4 cells by treatment with pVs.

Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.jinorgbio.2012.12.014. References [1] A.T. Ozcelikay, D.J. Becker, L.N. Ongemba, A.M. Pottier, J.C. Henquin, S.M. Brichard, Am. J. Physiol. 270 (1996) E344–E352. [2] J. Ybarra, A. Behrooza, A. Gabriel, M.H. Koseoglu, F. Ismail-Beig, Mol. Cell. Endocrinol. 133 (1997) 151–160. [3] R.A. Anderson, N. Cheng, N.A. Bryden, M.M. Polansky, N. Cheng, J. Chi, J. Feng, Diabetes 46 (1997) 1786–1791. [4] A. Barbera, R.R. Gomis, N. Prats, J.E. Rodriquez–Gil, M. Domingo, R. Gomis, J.J. Guinovart, Diabetologia 44 (2001) 507–513. [5] S.F. Simon, C.G. Taylor, Exp. Biol. Med. 226 (2001) 43–51. [6] M.C. Cam, R.W. Brownsey, J.H. McNeill, Can. J. Physiol. Pharmacol. 78 (2000) 829–847. [7] I. Goldwaser, D. Gefel, E. Gershonov, M. Fridkin, Y. Shechter, J. Inorg. Biochem. 80 (2000) 21–25. [8] L. Marzban, J.H. McNeill, J. Trace Elem. Exp. Med. 16 (2003) 253–267. [9] M.Z. Mehdi, S.K. Pandey, J.F. Theberge, A.K. Srivastava, Cell Biochem. Biophys. 44 (2006) 73–81. [10] C. Djordjevic, G.L. Wampler, J. Inorg. Biochem. 25 (1985) 51–55. [11] B.I. Posner, R. Faure, J.W. Burgess, A.P. Bevan, D. Lachance, G. Zhang-Sun, I.G. Fantus, J.B. Ng, D.A. Hall, B.S. Lum, A. Shaver, J. Biol. Chem. 269 (1994) 4596–4604. [12] A.P. Bevan, P.G. Drake, J.-F. Yale, A. Shaver, B.I. Posner, Mol. Cell. Biochem. 153 (1995) 49–58. [13] D. Rehder, G. Santoni, G.M. Licini, C. Schulzke, B. Meier, Coord. Chem. Rev. 237 (2003) 53–63. [14] K. Kanamori, K. Nishida, N. Miyata, K. Okamoto, Y. Miyoshi, A. Tamura, H. Sakurai, J. Inorg. Biochem. 86 (2001) 649–656. [15] A. Morinville, D. Maysinger, A. Shaver, Trends Pharmacol. Sci. 19 (1998) 452–460. [16] C.J. Doillon, R.L. Faure, B.I. Posner, P.E. Savard, Angiogenesis 3 (1999) 361–369. [17] N. Vuletic, C. Djordjevic, J. Chem. Soc. Dalton Trans. (1973) 1137–1141. [18] U. Quilitzsch, K. Wieghardt, Inorg. Chem. 18 (1979) 869–871. [19] A. Shaver, J.B. Ng, D.A. Hall, L.B. Soo, B.I. Posner, Inorg. Chem. 32 (1993) 3109–3113. [20] R. Stomberg, Acta Chem. Scand. A40 (1986) 168–176. [21] Z. Cerovac, J. Ban, A. Morinville, K. Yaccato, A. Shaver, D. Maysinger, Neurochem. Int. 34 (1999) 337–344. [22] L. Rumora, A. Shaver, T.Ž. Grubišić, D. Maysinger, Neurochem. Int. 38 (2001) 341–347. [23] L. Rumora, K. Barišić, D. Maysinger, T.Ž. Grubišić, Biochem. Biophys. Res. Commun. 300 (2003) 877–883. [24] A. Shaver, J.B. Ng, D.A. Hall, B.I. Posner, Mol. Cell. Biochem. 153 (1995) 5–15. [25] S. Iwakami, H. Misu, T. Takeda, M. Sugimori, S. Matsugo, S. Kaneko, T. Takamura, PLoS One 6 (2011) e27401. [26] Y. Ishihara, K. Takeuchi, F. Ito, N. Shimamoto, J. Cell. Physiol. 226 (2011) 1007–1016. [27] K. Kanamori, K. Yamamoto, T. Okayasu, N. Matsui, K. Okamoto, W. Mori, Bull. Chem. Soc. Jpn. 79 (1997) 3031–3040. [28] H. Sugiyama, S. Matsugo, T. Konishi, T. Takamura, S. Kaneko, Y. Kubo, K. Sato, K. Kanamori, Chem. Lett. 41 (2012) 377–379. [29] K. Kanamori, E. Kameda, T. Kabetani, T. Suemoto, K. Okamoto, S. Kaizaki, Bull. Chem. Soc. Jpn. 68 (1995) 2581–2589. [30] K. Kanamori, K. Nishida, N. Miyata, T. Shimoyama, K. Hata, C. Mihara, K. Okamoto, Y. Abe, S. Hayakawa, S. Matsugo, Inorg. Chem. 43 (2004) 7127–7140. [31] M. Sivak, D. Joniakova, P. Schwendt, Transit. Met. Chem. 18 (1993) 304–308. [32] J.A. Neal, N.J. Rose, Inorg. Chem. 7 (1968) 2405–2412. [33] C. Djordjevic, S.A. Craig, E. Sinn, Inorg. Chem. 24 (1985) 1281–1283. [34] C. Djordjevic, P.L. Wikins, E. Sinn, R.J. Butcher, Inorg. Chim. Acta 230 (1995) 241–244. [35] C.J. Colpas, B.J. Hamstra, J.W. Kampf, V.L. Pecoraro, J. Am. Chem. Soc. 118 (1996) 3469–3478.

76

H. Sugiyama et al. / Journal of Inorganic Biochemistry 121 (2013) 66–76

[36] M.C. Burla, R. Caliandro, M. Camalli, B. Carrozzini, G.L. Cascarano, L. de Caro, C. Giacovazzo, G. Polidori, R. Spagna, SIR2004, 2005. [37] G.M. Sheldrick, Acta Crystallogr. A64 (2008) 112–122. [38] Rigaku Coroporation, CrystalStructure 4.0: Crystal Structure Analysis Package, 2000–2010. (Tokyo 196–8666, Japan). [39] C. Djordjevic, M. Lee, E. Sinn, Inorg. Chem. 28 (1989) 719–723. [40] P. Schwendt, M. Ahmed, J. Marek, Inorg. Chem. Commun. 7 (2004) 631–633.

[41] I.K. Smatanova, J. Marek, P. Švančárek, P. Schwendt, Acta Crystallogr. C56 (2000) 154–155. [42] Y. Zhang, X.D. Yang, K. Wang, D.C. Crans, J. Inorg. Biochem. 100 (2006) 80–87. [43] D. Heffetz, I. Bushkin, R. Dror, Y. Zick, J. Biol. Chem. 265 (1990) 2896–2902. [44] Q. Wang, T.T. Liu, Y. Fu, K. Wang, X.G. Yang, J. Biol. Inorg. Chem. 15 (2010) 1087–1097.