- Email: [email protected]
Evidence against the nuclear in situ binding of arsenicals–oxidative stress theory of arsenic carcinogenesis
Toxicology and Applied Pharmacology 232 (2008) 252–257
Contents lists available at ScienceDirect
Toxicology and Applied Pharmacology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y t a a p
Evidence against the nuclear in situ binding of arsenicals–oxidative stress theory of arsenic carcinogenesis Kirk T. Kitchin ⁎, Kathleen Wallace Environmental Carcinogenesis Division, Mail Drop B143-06, National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA
a r t i c l e
i n f o
Article history: Received 26 February 2008 Revised 23 June 2008 Accepted 26 June 2008 Available online 12 July 2008 Keywords: Arsenic Arsenite Binding DNA Histone Ferritin Oxidative stress Fenton
a b s t r a c t A large amount of evidence suggests that arsenicals act via oxidative stress in causing cancer in humans and experimental animals. It is possible that arsenicals could bind in situ close to nuclear DNA followed by Haber–Weiss type oxidative DNA damage. Therefore, we tested this hypothesis by using radioactive 73As labeled arsenite and vacuum filtration methodology to determine the binding affinity and capacity of 73As arsenite to calf thymus DNA and Type 2A unfractionated histones, histone H3, H4 and horse spleen ferritin. Arsenicals are known to release redox active Fe from ferritin. At concentrations up to about 1 mM, neither DNA nor any of the three proteins studied, Type II-A histones, histone H3, H4 or ferritin, bound radioactive arsenite in a specific manner. Therefore, it appears highly unlikely that initial in situ binding of trivalent arsenicals, followed by in situ oxidative DNA damage, can account for arsenic's carcinogenicity. This experimental evidence (lack of arsenite binding to DNA, histone Type II-A and histone H3, H4) does not rule out other possible oxidative stress modes of action for arsenic such as (a) diffusion of longer lived oxidative stress molecules, such as H2O2 into the nucleus and ensuing oxidative damage, (b) redox chemistry by unbound arsenicals in the nucleus, or (c) arsenical-induced perturbations in Fe, Cu or other metals which are already known to oxidize DNA in vitro and in vivo. Published by Elsevier Inc.
Introduction Human exposure to inorganic arsenic can lead to carcinogenesis in urinary bladder, lung, skin, liver, kidney and to many other nonneoplastic health problems (e. g. dermatological, cardiovascular and neurological effects) (National Research Council, 1999). Both the chemical and biological mechanisms whereby arsenic produces these multiple and diverse health effects are relatively unknown. A recent review of arsenic carcinogenesis listed three of the more likely biological mechanisms of arsenic carcinogenesis (induced chromosomal abnormalities, oxidative stress, a continuum of altered growth factors → cell proliferation → promotion of carcinogenesis) (Kitchin, 2001). When more chemical mechanisms of arsenic's biological action are considered (Kitchin et al., 2003), four stronger possibilities are (a) oxidative stress/reactive oxygen species/free radicals formed from arsenic exposure, (b) binding of trivalent arsenicals and sulfhydryls, (c) nucleophilicity of trivalent arsenicals and (d) hypomethylation of DNA.
Abbreviations: Bmax, maximum binding capacity; DMA(V), dimethylarsinic acid; DMA(III), dimethylarsinous acid; 8-OHdG, 8-hydroxy-2′-deoxyguanosine; Kd, dissociation equilibrium constant; MMA(V), monomethylarsonic acid; MMA(III), monomethylarsonous acid; TMAO, trimethylarsine oxide. ⁎ Corresponding author. Fax: +1 919 682 3276. E-mail address: [email protected] (K.T. Kitchin). 0041-008X/$ – see front matter. Published by Elsevier Inc. doi:10.1016/j.taap.2008.06.021
In many organisms arsenic can be metabolized via a series of reductions and oxidative methylations (Thomas et al., 2001): arsenate → arsenite → monomethylarsonic acid (MMA(V)) → monomethylarsonous acid (MMA(III)) → dimethylarsinic acid (DMA (V)) → dimethylarsinous acid (DMA(III)) → trimethylarsine oxide (TMAO) → trimethylarsine. In most mammals, the methylation of arsenic stops at the dimethylated forms which are rapidly excreted in the urine. It is well known that in the presence of metal ions Fenton chemistry can produce hydroxyl radicals from H2O2 and damage DNA (Aust et al., 1985; Kehrer, 2000; Lloyd et al., 1997). If the Fenton metal catalyst is subsequently reduced and the Fenton catalytic cycle repeats then the name Haber–Weiss is usually applied to reductant driven Fenton type reactions (Aust et al., 1985; Kehrer, 1997). Hydroxy radicals are highly unstable, do not diffuse for long distances and interact quickly with other molecules they hit (Kehrer, 1997). Transition metals found in biological systems at high concentrations, such as Fe++ and Cu++, are frequent catalysts of this DNA damaging Fenton chemistry (Aust et al., 1985; Meneghini, 1997; Kehrer, 1997) which can lead to carcinogenesis (Klaunig and Kamendulis, 2004; Evans et al., 2004) and other disease states (Evans et al., 2004). Exposure to other metals such as Ni++, V+++ and Cr+6 (which occur at much lower concentrations in most biological systems) also are known to oxidize DNA (Lloyd et al., 1997).
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Histones are small positively charged proteins intimately bound to DNA that are rich in lysine and arginine and remarkably low in cysteine content. The high concentration of histones in the nucleus (about 3 mM) makes these proteins excellent candidates for binding a metal or metalloid and oxidizing nearby DNA (Bal and Kasprzak, 2002). Ferritin is a large spherical Fe storage protein with the capacity to store thousands of Fe atoms inside of it (Hempstead et al., 1997). The majority of ferritin is found in the cytoplasm, but some ferritin is found in the nucleus (Surguladze et al., 2005). Exposure to all six common arsenicals found in mammals, but particularly DMA(III), released Fe from horse spleen ferritin in vitro and damaged DNA (Ahmad et al., 2000). It is unknown if arsenite must first bind to ferritin before releasing Fe from ferritin. The combination of (a) specific metal binding in the nucleus and (b) oxidative damage via Fenton chemistry has already been demonstrated for Fe++ (Meneghini 1997; Rai et al., 2005), Cu++ (Chevion, 1988), Ni++ (Huang et al., 1995) and to a lesser extent Mg++ (Anastassopoulou and Theophanides, 2002). At least three of the trivalent forms of arsenic are known to bind to sulfhydryls groups [arsenite, MMA(III) and DMA(III)]. Arsenite binds to sulfhydryl groups, particularly dithiols with a Kd (dissociation equilibrium constant) of about 2–25 μM (Kitchin and Wallace, 2005, 2006). About 99% of the arsenite inside of cells is expected to be in the bound state and not chemically free (Kitchin and Wallace, 2005) because of the high concentrations of sulfhydryl containing molecules such as glutathione, proteins, peptides and lipoic acid found inside cells. It is possible that trivalent arsenicals might function as nuclear Fenton catalysts in a similar manner to Fe++ and Cu++ although this has never been directly demonstrated. For the in situ binding of arsenicals and redox active Fenton chemistry to be a key event in the mechanism of action of arsenic carcinogenesis, one requirement is that arsenic binding to major nuclear components such as DNA and histones first needs to be demonstrated. Therefore, in the present study, we utilized radioactive 73 As labeled arsenite and vacuum filtration methodology to determine if arsenite would bind to DNA, histones and ferritin in a specific manner. Methods Binding studies. Type XV calf thymus DNA (catalog # D4522), type IIA histone (catalog # 9250) and type 1 horse spleen ferritin (catalog # F4503) were obtained from Sigma Chem. Co. Horse ferritin has 2 cysteines and 6 histidines in 175 amino acids in the light chain and 3 cysteines and 11 histidines in 182 amino acids in the heavy chain. Bovine histone H3, H4 (catalog # H5110-04) was obtained from US Biological. Bovine histone H3 has 2 cysteines and 2 histidines in 135 amino acids exactly the same as human histone H3. Bovine histone H4 has 0 cysteines and 2 histidines in 102 amino acids, again the same as the human histone H4 sequence. Collectively, the five major bovine histones contain a total of 595 amino acids but only 2 cysteines (C96, C110 of histone H3) and 10 histidines. Positive controls for arsenite binding included rat globin, bovine brain tubulin (Cytoskeleton Cat # TL238), peptide 29 based on Keap-1 of amino acid sequence containing four cysteines KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY, peptide 10, LECAWQGKCVEGTEHLYSMKCKNV containing three cysteines, peptide 24 RYCAVCNDYASGYHYGVWSCEGGKA with three cysteines and peptide 34 KACKACKAKAKAKK with two cysteines. All synthetic peptides were synthesized by and mass spectroscopy and HPLC purity determinations (ranging from 90 to 100%) were run by a commercial laboratory (Alpha Diagnostics, San Antonio, TX). The NH2 terminal end was labeled with fluorescein isothiocyanate as a fluorescent tag. These five peptides and proteins were run just before, during and after the DNA, histones and ferritin experiments. Rat globin positive controls were run concurrently with all test samples. Binding was observed in all cases with rat globin (a Cys containing positive control protein) and
253
this demonstrated that the arsenite binding assay was working well. Other prior (Kitchin and Wallace, 2005) and sometimes concurrent binding experiments (with a peptide based on Keap 1, data not shown) with peptides and proteins containing monothiol or dithiol binding sites (Kd values in the range of 1 to 200 μM) also demonstrated the arsenite binding assay was working correctly. 73 As (arsenate) was obtained from the Brookhaven National Laboratories and used as arsenate in a limited number of experiments with DNA and histone 3,4. In all other experiments arsenate was reduced to arsenite by bubbling SO2 gas into the arsenate solution and then warming to 37 °C to remove the excess SO2 gas. The specific activity of 73As used ranged from 0.0833 to 27.7 μCi/pmol arsenite. Binding experiments used the test macromolecule (DNA, histones or ferritin) diluted in cold water, 73As arsenite diluted in cold 150 mM NaCl, pH 7.5 buffer containing 100 mM Tris–HCl, a Brandel Model M-24C Membrane Harvester (Brandel Inc., Gaithersburg, MD) and either nitrocellulose filters (0.45 μm) soaked in a pH 7.5 solution containing 150 mM NaCl, 0.3% polyethylenimine and 100 mM Tris–HCl for protein experiments or Nylon filters (0.45 μm, Whatman Nytran SPC) for DNA experiments soaked in a pH 7.5 solution containing 150 mM NaCl and 100 mM Tris–HCl. The concentration of macromolecular targets used in our binding studies was approximately 33 μM for peptides and proteins and 200 μg/ml for DNA. Solutions were deoxygenated by bubbling nitrogen gas through them. Macromolecules and arsenite were incubated for 60 min at 2–8 °C prior to vacuum filtration. To reduce nonspecific binding, filtered macromolecules were washed three times with 2 ml of cold 150 mM NaCl, 100 mM Tris–HCl, pH 7.5, solution. The wash solution was kept on ice. Nonspecific binding was routinely determined by the addition of 100-fold excess of unlabelled arsenite to 73As arsenite. Gamma counting was done in a Packard Minaxi Auto gamma 5000. Counting efficiencies ranged from 8.4 to 8.7%. Protein was stripped from the Brandel NC-100 0.45 pore size nitrocellulose filters in a pH 10 buffer of 62.5 mM Na2CO2 containing 5% sodium lauryl sulfate. Using a Wallac Victor 1420 multilabel counter, the concentration of protein was determined spectrophotometrically using the Pierce bicinchoninic acid protein determination kit (using weighed bovine serum albumin as the standard). Protein recoveries from nitrocellulose filters ranged from 29 to 48%. DNA was stripped from the Nytran filters at room temperature with 0.3% sodium lauryl sulfate in 50 mM sodium phosphate buffered to pH 3.0. Quantification of the released DNA was done by a fluorescence binding assay using Hoechst 33258 and measuring the emitted fluorescence at 460 nm after excitation at 355 nm. DNA recoveries ranged from 45 to 96%. Triplicate samples were normally used for binding measurements; duplicate determinations of nonspecific binding samples via the 100fold excess cold carrier method were performed. For the ferritin saturation experiment of Fig. 4, only duplicate samples were run. All saturation binding experiments were repeated at least twice on different days. Representative saturation binding curves and Kd and Bmax values are given in this paper. Measurement day effects did not produce difficulties in these binding studies. Nonspecific binding is often fairly low in our studies but use of higher concentrations (from 100 to 1000 μM) of arsenite caused increasing nonspecific binding particularly to protein samples on nitrocellulose filters (Figs. 1–4). The amount of specific binding is estimated by subtracting the estimate of nonspecific binding from the estimate of total binding. The nonspecific binding is estimated by using a portion of the x axis that is 100-fold to the right of the region used to estimate the total binding. Kd and Bmax (maximum binding capacity) values were estimated from saturation binding experiments by use of the program Graph Pad Prism 4. The determined degree of ligand depletion that occurred in a saturation binding experiment was subtracted from the estimate of the actual free ligand concentration.
254
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Efficiency of arsenate reduction to arsenite/stability of arsenite. The presence of oxygen during the reduction of arsenate to arsenite may prevent complete reduction of arsenate. In addition, once the SO2 is removed, the arsenite may oxidize back to arsenate via oxygen exposure. To ascertain the reduction efficiency of arsenite by SO2, we determined the ratio of arsenite to arsenate by separating the radioactive arsenate from the arsenite using a strong anion exchange cartridge (Yalcin and Le, 2001). Stability of peptide sulfhydryl groups during the binding experiments. The o-phthalaldehyde fluorescence technique was used for determining the free sulfhydryl group concentration of proteins (Hissin and Hilf, 1976) both before and after 1 h incubations at 2–8 °C. Results Arsenate reduction and the stability of peptide sulfhydryl groups during the binding experiments Using a strong anion exchange cartridge to separate arsenate from arsenite (Yalcin and Le, 2001), our results indicated that the average % reduction of arsenate to arsenite was 93% and ranged from 85 to 99% (over 23 experiments). Over 15 experiments there was an average loss of 3.5% of the available sulfhydryl groups ranging from no loss observed to a maximum of 15% loss. Binding studies with DNA The concentrations of arsenite used in these studies were 30, 60, 100, 200, 300, 600 and 1000 μM. At no concentration did the total binding clearly exceed the nonspecific binding curve based on either (a) the 100-fold carrier nonspecific binding method or (b) the binding to the filter alone without any added DNA (data not shown). Thus, there was no positive evidence for 73As labeled arsenite specifically binding to calf thymus DNA even at 1000 times the arsenite concentrations found in exposed animals in vivo (Fig. 1). In one experiment using between 2 and 1000 μM arsenate, pentavalent 73 arsenate did not specifically bind to DNA. Binding studies with histone Type II-A and H3, H4 Histone type II-A is an unfractionated mixture of histone proteins obtained from calf thymus. It is expected to have a low concentration of Cys. Histone type II-A was studied up to about 1000 μM. Again, there was no sign of specific binding of radioactive arsenite to type II-A histones (Fig. 2). In this and in all subsequent cases where up to
Fig. 2. Saturation binding results for
73
As arsenite and Type II-A histones.
1000 μM arsenite was used as the ligand with nitrocellulose filters, the nonspecific binding curves were based on using 100-fold excess unlabelled arsenite. Saturation binding experiments with histone H3, H4 also showed that there was no specific binding of arsenite to histone H3, H4, in spite of the two Cys present in the amino acid sequence of histone H3. In one experiment using between 2 and 1000 μM arsenate, pentavalent 73arsenate also did not specifically bind to histone 3,4. Binding studies with ferritin Saturation binding studies with ferritin in the range of up to 1000 μM also showed no positive evidence of arsenite specific binding (Fig. 4). Values of the total and nonspecific binding were remarkably close over the whole range studied. Summary of the experimental conditions and results Table 1 summarizes the data on DNA and proteins examined for arsenite binding. Multiple positive controls were used concurrently in these studies (rat globin, tubulin, a peptide based on Keap-1 of amino acid sequence containing four cysteines KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY and peptide 10 based on human estrogen receptor alpha LECAWQGKCVEGTEHLYSMKCKNV containing three cysteines). In addition, other peptides and proteins were used in binding experiments both just before (the peptide based on Keap-1 and tubulin) and after (peptide 34, KACKACKAKAKAKK with two cysteines) the experiments with DNA, histones and ferritin. In all cases, the proteins and peptides run just before, during and after the DNA, histone and ferritin experiments demonstrated that the arsenite binding assay was working and capable of detecting any specific binding that occurred. For bovine tubulin, peptide 10, peptide 34 and peptide 29, the amount of binding observed in this series of experiments was consistent with their known Bmax values ranging from 15 to 80 nmol arsenite binding/mg protein. For rat globin used as a positive control, the ratio of observed counts was about 3 to 1 for rat globin samples to blanks. Discussion Arsenite and possible DNA binding
Fig. 1. Saturation binding results for
73
As arsenite and calf thymus DNA.
In DNA there are negatively charged phosphate groups, sugar moieties and the 4 bases which contain many aromatic moieties, amine and keto groups as possible binding sites. Based on the data of Fig. 1, none of the major functional groups of DNA seems to bind
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
255
Table 1 Summary of the specific binding results of arsenite and the studied macromolecules or peptides Macromolecule Experimental series DNA from calf thymus
Histone II-A from calf thymus
Histone 3,4 from cow
Ferritin from horse spleen
Positive controls Globin from rat
Cysteines/polypeptide chain
Specific binding Comment/reference observed?
No cysteines, but many amine, keto, hydroxyl, aryl, deoxyribose and phosphate groups are available as possible binding sites A commercially available mixture of histones
No
No
Collectively, the five major bovine histones contain a total of 595 amino acids but only 2 cysteines (C96, C110 of histone H3) and 10 histidines
Bovine histone H3 has 2 cysteines in 135 amino No acids exactly the same as human histone H3. Bovine histone H4 has 0 cysteines in 102 amino acids again the same as the human histone H4 sequence No 2 cysteines and 6 histidines in 175 amino acids in the light chain and 3 cysteines and 11 histidines in 182 amino acids in the heavy chain
5 cysteines in 289 amino acids based on Swiss Protein's amino acid sequence entries P01946 and P11517 4 cysteines in 46 amino acids
Peptide 29 based on an amino acid sequence from Keap-1 KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY Tubulin from cow (whole protein, α and β units) 13 cysteines in 448 amino acids in the alpha chain (Q5E986), 8 cysteines in 445 amino acids in the beta chain (Q6B856 of Swisss Protein) Peptide 10 LECAWQGKCVEGTEHLYSMKC KNV 3 cysteines in 24 amino acids Peptide 24 RYCAVCNDYASGYHYGV WSCEGGKA 3 cysteines in 25 amino acids Peptide 34 KACKACKAKAKAKK 2 cysteines in 14 amino acids
Yes
Yes
Rat globin is a convenient inexpensive positive control for arsenite binding studies Kitchin and Wallace (2008)
Yes
Kitchin and Wallace (2008)
Yes Yes Yes
Kitchin and Wallace (2005) Kitchin and Wallace (2005)
arsenite well. Based on the negative results with 73As labeled arsenite, a tridentate binding ligand (Jiang et al., 2003), it seems unlikely at physiological concentrations that other trivalent arsenicals (e. g. MMA (III), DMA(III) or trivalent arsines) can bind to DNA directly.
virtually not at all to protein sites lacking even one exposed monothiol. This has been confirmed by direct measurements of the binding of 73As labeled arsenite to a series of peptides containing 0, 1, 2, 3 and 4 cysteines (Kitchin and Wallace, 2005).
Arsenite and possible binding to histones
Arsenite and possible binding to ferritin
From previous studies of peptides of known amino acid composition and sequence, it is clear that arsenite only binds well to one Cys or two Cys (dithiol) type of binding sites. No other single amino acid seems to be capable of binding arsenite with much affinity. Arsenite does not seem to bind to any significant extent to amino acid side chains such as amine, carboxyl, keto, hydroxyl and aryl. Three exceptions to this generalization are known. Arsenite binds to (Kd of 15 μM) and inhibits the enzyme acetylcholinesterase. In the protein acetylcholinesterase which possesses no Cys or lipoic acid moieties, the arsenite binds to two tyrosines in a diester manner (Page and Wilson, 1985). In a second case, arsenic has been demonstrated to bind to the molybdenum atom of xanthine oxidase (Stewart et al., 1984). Finally, following the injection of arsenite and selenate into rabbits, the complex seleno-bis (S-glutathionyl) arsinium ion, [(GS)(2)AsSe] (−), was found in the bile (Gailer et al., 2002). As shown in Fig. 2, type II-A histones showed no binding of arsenite, as might have been predicted based on its low concentration of Cys. Other alternative binding sites such as multiple hydroxyl or carboxyl or amine functional groups might make a metal binding site for other metals, but no good binding sites for arsenite were found in unfractionated type II-A histones. Histone H3, H4 also did not bind any arsenite in a specific manner, despite two Cys in the amino acid sequence of histone H3 (Fig. 3). As with the mixed histones found in type II-A histones, other noncysteine potential binding sites did not show any higher affinity binding sites for arsenite. Thus, nuclear histones do not seem to have any free reduced Cys available as good arsenite binding sites. Trivalent arsenic compounds such as arsenite bind more strongly to dithiol sites than to monothiol sites (Johnstone, 1963; Aposhian, 1989) and
Six different chemical forms of arsenic [arsenate, arsenite, MMA (V), MMA(III), DMA(V) and DMA(III)] release varying amounts of Fe from horse spleen ferritin (Ahmad et al., 2000). The light chain of horse ferritin contains 2 cysteines, 6 histidines and 175 amino acids. The heavy chain of horse ferritin contains 3 cysteines, 11 histidines and 182 amino acids. However as shown in Fig. 4, arsenite did not bind to the protein ferritin. Apparently the mechanism of arsenical action in releasing Fe from ferritin does not involve the binding of trivalent arsenic species to the protein part of the ferritin macromolecule and
Fig. 3. Saturation binding results for
73
As arsenite and histone H3, H4.
256
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Fig. 4. Saturation binding results for
73
As arsenite and horse spleen ferritin.
changing the protein conformation, function, or pore sizes or properties. Arsenicals are also capable of changing their valence and thus can act as reducing or oxidizing agents. The remaining attractive explanation of arsenic's Fe releasing effects on ferritin is that it simply acts as a redox agent and reduces Fe+++ to Fe++ (the releasable form of Fe). This is consistent with the finding that trivalent arsenicals released more Fe from ferritin than did the corresponding pentavalent arsenical (3.4 and 1.6 for arsenite and arsenate, respectively; 8.7 and 5.4 for MMA(III) and MMA(V), respectively; and 29.8 and 7.1 nM/min for DMA(III) and DMA(V), respectively (Ahmad et al., 2000). If oxidation of only 0.1% of an arsenical [e. g. DMA(III) to DMA(V)] was coupled to Fe reduction and release from ferritin, all the results observed in the Ahmad et al. (2000) study could be explained by arsenicals acting via redox properties (reduced arsenical → oxidized arsenical coupled to Fe+++ → Fe++). In the Fe release from ferritin experiments, the tested concentrations of the arsenicals were 10 mM, somewhat more than the 1.0 mM that was used in the binding studies in this paper. Arsenate and possible binding to DNA and histone 3,4 The pentavalent arsenate molecule has either one or two negative charges at physiological pH (the pKa values for H3AsO4 are 2.25, 6.77 and 11.6). Thus, it is not surprising that arsenate did not bind to negatively charged DNA. However, positively charged histones could be an excellent target for binding of arsenate but no such specific binding of arsenate to histone 3,4 was experimentally observed. In situ metal binding of other transition elements and possible sources of DNA oxidation Because this experimental study found no binding of arsenite to either DNA, type II-A histone and histone H3, H4, we must argue strongly that the possible binding of arsenicals and in situ oxidation of DNA is not a viable explanation of arsenic's carcinogenicity. Trivalent arsenicals are expected to be uncharged at physiological pH so are very electrically different from other positively charged transition elements. However, there is positive evidence that in situ binding and DNA oxidation occurs with other transition elements, for example Fe++, Cu++ and Ni++. There is about an 80/1 ratio of Fe to Cu in the human body (Chevion, 1988). With Fe++, the base sequence GGG binds Fe++ preferentially, resulting in structural distortion of DNA and enhanced DNA cleavage (Rai et al., 2005). With combined exposures to Cu++, H2O2 and a Cu++ binding histone peptide (CH3CO-AKRHRKCONH2), enhanced DNA fragmentation, formation of 8-hydroxy-2′-
deoxyguanosine (8-OHdG) and increased electron spin resonance signal were observed (Midorikawa et al., 2005). Chromatin can concentrate Ni++ more than over 5000-fold over serum, mostly via electrostatic attraction to the phosphate groups of DNA (Rokita and Burrows, 2001). In vitro, exposure to Ni++, nucleohistone and H2O2, enhances 8-OHdG formation (Bal and Kasprzak, 2002). There are known binding sites for Ni++ and Cu++ on histone H2A and histone H3 having the amino acid sequences of CAIH, ESHH and TESHHK (Bal and Kasprzak, 2002). Ni++ also binds to histone H1 and in the presence of H2O2 increases the formation of 8-OHdG in vitro (Huang et al., 1995). When DNA is irradiated, the presence of Mg++ causes an approximate doubling of the production of hydroxyl radicals (Anastassopoulou and Theophanides, 2002). Thus, with four different positively charged metals [Fe++, Cu++, Ni++ and Mg++], there is positive evidence demonstrating that in situ binding of biologically active metals close to DNA can lead to enhanced DNA oxidation and damage. Arsenic exposures have also been linked not only to the common reactive oxygen species, but also to arsenic containing species such as dimethylarsenic radical (Yamanaka et al., 1990), dimethylarsenic peroxyl radical (Yamanaka et al., 1990) and dimethylarsenic peroxide (Mizoi et al., 2005). Conclusion At concentrations up to approximately 1 mM, neither DNA nor any of the three proteins studied, type II-A histone, histone 3 or ferritin bound radioactive arsenite in a specific manner. Therefore, it appears highly unlikely that in situ binding of trivalent arsenicals, followed by in situ oxidative DNA damage, can account for arsenic's carcinogenicity. Arsenite, lacking a positive charge and being predominately bound to Cys moieties of peptides and proteins, does not seem to operate in a similar manner as Fe++, Cu++ and Ni++ in binding close to DNA and causing in situ oxidation of DNA via Haber–Weiss processes. The experimental evidence of this study (lack of arsenite binding to DNA and histones) does not rule out (a) diffusion of longer lived oxidative stress molecules, such as H2O2 into the nucleus and ensuing oxidative DNA damage, (b) redox chemistry by free (unbound) arsenicals in the nucleus or (c) arsenical-induced perturbations in Fe, Cu or other redox active metals which are already the known to oxidize DNA in vitro and in vivo. It is always difficult to rule out Feand Cu-mediated pathways in mammalian oxidative stress and carcinogenesis. Even in mammals with a substantial amount of arsenicals from environmental or experimental exposures, there are still orders of magnitude more Cu and Fe present in the mammalian organism than there is total arsenic. Acknowledgments We thank Drs. Russell Owen and Janice S Lee for reviewing this manuscript as part of EPA clearance procedures. References Ahmad, S., Kitchin, K.T., Cullen, W.R., 2000. Arsenic species that cause release of iron from ferritin and generation of activated oxygen. Arch. Biochem. Biophys. 382, 195–202. Anastassopoulou, J., Theophanides, T., 2002. Magnesium–DNA interactions and the possible relation of magnesium to carcinogenesis. Irradiation and free radicals. Crit. Rev. Oncol. Hematol. 42, 79–91. Aposhian, H.V., 1989. Biochemical toxicology of arsenic. In: Hodgson, E., Bend, J.R., Philpot, R.M. (Eds.), Reviews in Biochemical Toxicology, vol. 10. Elsevier Publishing Co., New York, pp. 265–299. Aust, S.D., Morehouse, L.A., Thomas, C.E., 1985. Role of metals in oxygen radical reactions. J. Free. Radic. Biol. Med. 1, 3–25. Bal, W., Kasprzak, K.S., 2002. Induction of oxidative DNA damage by carcinogenic metals. Toxicol. Lett. 127, 55–62. Chevion, M., 1988. A site-specific mechanism for free radical induced biological damage: the essential role of redox-active transition metals. Free Radic. Biol. Med. 5, 27–37.
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257 Evans, M.D., Dizdaroglu, M., Cooke, M.S., 2004. Oxidative DNA damage and disease: induction, repair and significance. Mutat. Res. 567, 1–61. Gailer, J., George, G.N., Pickering, I.J., Prince, R.C., Younis, H.S., Winzerling, J.J., 2002. Biliary excretion of [(GS)(2)AsSe](−) after intravenous injection of rabbits with arsenite and selenate. Chem. Res. Toxicol. 11, 1466–1471. Hempstead, P.D., Yewdall, S.J., Fernie, A.R., Lawson, D.M., Artymiuk, P.J., Rice, D.W., Ford, G.C., Harrison, P.M., 1997. Comparison of the three-dimensional structures of recombinant human H and horse L ferritins at high resolution. J. Mol. Biol., 268, 424–448. Hissin, P.J., Hilf, R., 1976. A fluorometric method for determination of oxidized and reduced glutathione in tissues. Anal. Biochem. 74, 214–226. Huang, X., Kitahara, J., Zhitkovich, A., Dowjat, K., Costa, M., 1995. Heterochromatic proteins specifically enhance nickel-induced 8-oxo-dG formation. Carcinogenesis. 16, 1753–1759. Jiang, G., Gong, Z., Li, X.F., Cullen, W.R., Le, X.C., 2003. Interaction of trivalent arsenicals and metallothionein. Chem. Res. Toxicol. 16, 873–880. Johnstone, R.M., 1963. Sulfhydryl agents: arsenicals. In: Hochster, R.M., Quastel, J.H. (Eds.), Metabolic Inhibitors: A Comprehensive Treatise, vol. 2. Academic Press, New York, pp. 99–118. Kehrer, J.P., 1997. Mechanisms: free radicals and reactive oxygen species. In: Bond, J.A. (Ed.), General Principles, Toxicokinetics, and Mechanisms of Toxicology Vol 1 of Comprehensive Toxicology. Pergamon, pp. 275–301. Kehrer, J.P., 2000. The Haber–Weiss reaction and mechanisms of toxicity. Toxicology 149, 43–50. Kitchin, K.T., 2001. Recent advances in arsenic carcinogenesis: modes of action, animal model systems and methylated arsenic metabolites. Toxicol. Appl. Pharmacol. 172, 249–261. Kitchin, K.T., Wallace, K., Andrewes, P., 2003. Some chemical properties underlying arsenic's biological activity. In: Chappell, W.R., Abernathy, C.O., Calderon, R.L., Thomas, D.J. (Eds.), Arsenic Exposure and Health Effects. V. Elsevier, San Diego, pp. 345–354. Kitchin, K.T., Wallace, K., 2005. Arsenite binding to synthetic peptides based on the zinc finger and the estrogen binding region of the human estrogen receptor-alpha. Toxicol. Appl. Pharmacol. 206, 666–672. Kitchin, K.T., Wallace, K., 2006. Arsenite binding to synthetic peptides: the effect of increasing length between two cysteines. Journ. Biochem. Mol. Toxicol. 20, 35–38.
257
Kitchin, K.T., Wallace, K., 2008. The role of protein binding of trivalent arsenicals in arsenic carcinogenesis and toxicity. J. Inorg. Biochem. 102, 532–539. Klaunig, J.E., Kamendulis, L.M., 2004. The role of oxidative stress in carcinogenesis. Annu. Rev. Pharmacol. Toxicol. 44, 239–267. Lloyd, D.R., Phillips, D.H., Carmichael, P.L., 1997. Generation of putative intrastrand cross-links and strand breaks in DNA by transition metal ion-mediated oxygen radical attack. Chem. Res. Toxicol. 10, 393–400. Meneghini, R., 1997. Iron homeostasis, oxidative stress, and DNA damage. Free Radic. Biol. Med. 23, 783–792. Midorikawa, K., Murata, M., Kawanishi, S., 2005. Histone peptide AKRHRK enhances H2O2-induced DNA damage and alters its site specificity. Biochem. Biophys. Res. Commun. 333, 1073–1077. Mizoi, M., Takabayashi, F., Nakano, M., An, Y., Sagesaka, Y., Kato, K., Okada, S., Yamanaka, K., 2005. The role of trivalent dimethylated arsenic in dimethylarsinic acidpromoted skin and lung tumorigenesis in mice: tumor-promoting action through the induction of oxidative stress. Toxicol. Lett. 158, 87–94. National Research Council, 1999. Health effects of arsenic. Arsenic in Drinking Water. National Academy Press, Washington, DC., pp. 83–149. Page, J.D., Wilson, I.B., 1985. Acetylcholinesterase: inhibition by tetranitromethane and arsenite. Binding of arsenite by tyrosine residues. J. Biol. Chem. 260, 1475–1478. Rai, P., Wemmer, D.E., Linn, S., 2005. Preferential binding and structural distortion by Fe2+ at RGGG-containing DNA sequences correlates with enhanced oxidative cleavage at such sequences. Nucleic Acids Res. 33, 497–510. Rokita, S.E., Burrows, C.J., 2001. Nickel- and cobalt-dependent oxidation and crosslinking of proteins. Met. Ions. Biol. Syst. 38, 289–311. Stewart, R.C., Hille, R., Massey, V., 1984. Characterization of arsenite-complexed xanthine oxidase at room temperature. J. Biol. Chem. 259, 14426–14436. Surguladze, N., Patton, S., Cozzi, A., Fried, M.G., Connor, J.R., 2005. Characterization of nuclear ferritin and mechanism of translocation. Biochem. J. 388, 731–740. Thomas, D.J., Styblo, M., Lin, S., 2001. The cellular metabolism and systemic toxicity of arsenic. Toxicol. Appl. Pharmacol. 176, 127–144. Yalcin, S., Le, X.C., 2001. Speciation of arsenic using solid phase extraction cartridges. J. Environ. Monit. 3, 81–85. Yamanaka, K., Hoshino, M., Okanoto, M., Sawamura, R., Hasegawa, A., Okada, S., 1990. Induction of DNA damage by dimethylarsine, a metabolite of inorganic arsenics, is for the major part likely due to its peroxyl radical. Biochem. Biophys. Res. Comm. 168, 58–64.
Contents lists available at ScienceDirect
Toxicology and Applied Pharmacology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y t a a p
Evidence against the nuclear in situ binding of arsenicals–oxidative stress theory of arsenic carcinogenesis Kirk T. Kitchin ⁎, Kathleen Wallace Environmental Carcinogenesis Division, Mail Drop B143-06, National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA
a r t i c l e
i n f o
Article history: Received 26 February 2008 Revised 23 June 2008 Accepted 26 June 2008 Available online 12 July 2008 Keywords: Arsenic Arsenite Binding DNA Histone Ferritin Oxidative stress Fenton
a b s t r a c t A large amount of evidence suggests that arsenicals act via oxidative stress in causing cancer in humans and experimental animals. It is possible that arsenicals could bind in situ close to nuclear DNA followed by Haber–Weiss type oxidative DNA damage. Therefore, we tested this hypothesis by using radioactive 73As labeled arsenite and vacuum filtration methodology to determine the binding affinity and capacity of 73As arsenite to calf thymus DNA and Type 2A unfractionated histones, histone H3, H4 and horse spleen ferritin. Arsenicals are known to release redox active Fe from ferritin. At concentrations up to about 1 mM, neither DNA nor any of the three proteins studied, Type II-A histones, histone H3, H4 or ferritin, bound radioactive arsenite in a specific manner. Therefore, it appears highly unlikely that initial in situ binding of trivalent arsenicals, followed by in situ oxidative DNA damage, can account for arsenic's carcinogenicity. This experimental evidence (lack of arsenite binding to DNA, histone Type II-A and histone H3, H4) does not rule out other possible oxidative stress modes of action for arsenic such as (a) diffusion of longer lived oxidative stress molecules, such as H2O2 into the nucleus and ensuing oxidative damage, (b) redox chemistry by unbound arsenicals in the nucleus, or (c) arsenical-induced perturbations in Fe, Cu or other metals which are already known to oxidize DNA in vitro and in vivo. Published by Elsevier Inc.
Introduction Human exposure to inorganic arsenic can lead to carcinogenesis in urinary bladder, lung, skin, liver, kidney and to many other nonneoplastic health problems (e. g. dermatological, cardiovascular and neurological effects) (National Research Council, 1999). Both the chemical and biological mechanisms whereby arsenic produces these multiple and diverse health effects are relatively unknown. A recent review of arsenic carcinogenesis listed three of the more likely biological mechanisms of arsenic carcinogenesis (induced chromosomal abnormalities, oxidative stress, a continuum of altered growth factors → cell proliferation → promotion of carcinogenesis) (Kitchin, 2001). When more chemical mechanisms of arsenic's biological action are considered (Kitchin et al., 2003), four stronger possibilities are (a) oxidative stress/reactive oxygen species/free radicals formed from arsenic exposure, (b) binding of trivalent arsenicals and sulfhydryls, (c) nucleophilicity of trivalent arsenicals and (d) hypomethylation of DNA.
Abbreviations: Bmax, maximum binding capacity; DMA(V), dimethylarsinic acid; DMA(III), dimethylarsinous acid; 8-OHdG, 8-hydroxy-2′-deoxyguanosine; Kd, dissociation equilibrium constant; MMA(V), monomethylarsonic acid; MMA(III), monomethylarsonous acid; TMAO, trimethylarsine oxide. ⁎ Corresponding author. Fax: +1 919 682 3276. E-mail address: [email protected] (K.T. Kitchin). 0041-008X/$ – see front matter. Published by Elsevier Inc. doi:10.1016/j.taap.2008.06.021
In many organisms arsenic can be metabolized via a series of reductions and oxidative methylations (Thomas et al., 2001): arsenate → arsenite → monomethylarsonic acid (MMA(V)) → monomethylarsonous acid (MMA(III)) → dimethylarsinic acid (DMA (V)) → dimethylarsinous acid (DMA(III)) → trimethylarsine oxide (TMAO) → trimethylarsine. In most mammals, the methylation of arsenic stops at the dimethylated forms which are rapidly excreted in the urine. It is well known that in the presence of metal ions Fenton chemistry can produce hydroxyl radicals from H2O2 and damage DNA (Aust et al., 1985; Kehrer, 2000; Lloyd et al., 1997). If the Fenton metal catalyst is subsequently reduced and the Fenton catalytic cycle repeats then the name Haber–Weiss is usually applied to reductant driven Fenton type reactions (Aust et al., 1985; Kehrer, 1997). Hydroxy radicals are highly unstable, do not diffuse for long distances and interact quickly with other molecules they hit (Kehrer, 1997). Transition metals found in biological systems at high concentrations, such as Fe++ and Cu++, are frequent catalysts of this DNA damaging Fenton chemistry (Aust et al., 1985; Meneghini, 1997; Kehrer, 1997) which can lead to carcinogenesis (Klaunig and Kamendulis, 2004; Evans et al., 2004) and other disease states (Evans et al., 2004). Exposure to other metals such as Ni++, V+++ and Cr+6 (which occur at much lower concentrations in most biological systems) also are known to oxidize DNA (Lloyd et al., 1997).
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Histones are small positively charged proteins intimately bound to DNA that are rich in lysine and arginine and remarkably low in cysteine content. The high concentration of histones in the nucleus (about 3 mM) makes these proteins excellent candidates for binding a metal or metalloid and oxidizing nearby DNA (Bal and Kasprzak, 2002). Ferritin is a large spherical Fe storage protein with the capacity to store thousands of Fe atoms inside of it (Hempstead et al., 1997). The majority of ferritin is found in the cytoplasm, but some ferritin is found in the nucleus (Surguladze et al., 2005). Exposure to all six common arsenicals found in mammals, but particularly DMA(III), released Fe from horse spleen ferritin in vitro and damaged DNA (Ahmad et al., 2000). It is unknown if arsenite must first bind to ferritin before releasing Fe from ferritin. The combination of (a) specific metal binding in the nucleus and (b) oxidative damage via Fenton chemistry has already been demonstrated for Fe++ (Meneghini 1997; Rai et al., 2005), Cu++ (Chevion, 1988), Ni++ (Huang et al., 1995) and to a lesser extent Mg++ (Anastassopoulou and Theophanides, 2002). At least three of the trivalent forms of arsenic are known to bind to sulfhydryls groups [arsenite, MMA(III) and DMA(III)]. Arsenite binds to sulfhydryl groups, particularly dithiols with a Kd (dissociation equilibrium constant) of about 2–25 μM (Kitchin and Wallace, 2005, 2006). About 99% of the arsenite inside of cells is expected to be in the bound state and not chemically free (Kitchin and Wallace, 2005) because of the high concentrations of sulfhydryl containing molecules such as glutathione, proteins, peptides and lipoic acid found inside cells. It is possible that trivalent arsenicals might function as nuclear Fenton catalysts in a similar manner to Fe++ and Cu++ although this has never been directly demonstrated. For the in situ binding of arsenicals and redox active Fenton chemistry to be a key event in the mechanism of action of arsenic carcinogenesis, one requirement is that arsenic binding to major nuclear components such as DNA and histones first needs to be demonstrated. Therefore, in the present study, we utilized radioactive 73 As labeled arsenite and vacuum filtration methodology to determine if arsenite would bind to DNA, histones and ferritin in a specific manner. Methods Binding studies. Type XV calf thymus DNA (catalog # D4522), type IIA histone (catalog # 9250) and type 1 horse spleen ferritin (catalog # F4503) were obtained from Sigma Chem. Co. Horse ferritin has 2 cysteines and 6 histidines in 175 amino acids in the light chain and 3 cysteines and 11 histidines in 182 amino acids in the heavy chain. Bovine histone H3, H4 (catalog # H5110-04) was obtained from US Biological. Bovine histone H3 has 2 cysteines and 2 histidines in 135 amino acids exactly the same as human histone H3. Bovine histone H4 has 0 cysteines and 2 histidines in 102 amino acids, again the same as the human histone H4 sequence. Collectively, the five major bovine histones contain a total of 595 amino acids but only 2 cysteines (C96, C110 of histone H3) and 10 histidines. Positive controls for arsenite binding included rat globin, bovine brain tubulin (Cytoskeleton Cat # TL238), peptide 29 based on Keap-1 of amino acid sequence containing four cysteines KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY, peptide 10, LECAWQGKCVEGTEHLYSMKCKNV containing three cysteines, peptide 24 RYCAVCNDYASGYHYGVWSCEGGKA with three cysteines and peptide 34 KACKACKAKAKAKK with two cysteines. All synthetic peptides were synthesized by and mass spectroscopy and HPLC purity determinations (ranging from 90 to 100%) were run by a commercial laboratory (Alpha Diagnostics, San Antonio, TX). The NH2 terminal end was labeled with fluorescein isothiocyanate as a fluorescent tag. These five peptides and proteins were run just before, during and after the DNA, histones and ferritin experiments. Rat globin positive controls were run concurrently with all test samples. Binding was observed in all cases with rat globin (a Cys containing positive control protein) and
253
this demonstrated that the arsenite binding assay was working well. Other prior (Kitchin and Wallace, 2005) and sometimes concurrent binding experiments (with a peptide based on Keap 1, data not shown) with peptides and proteins containing monothiol or dithiol binding sites (Kd values in the range of 1 to 200 μM) also demonstrated the arsenite binding assay was working correctly. 73 As (arsenate) was obtained from the Brookhaven National Laboratories and used as arsenate in a limited number of experiments with DNA and histone 3,4. In all other experiments arsenate was reduced to arsenite by bubbling SO2 gas into the arsenate solution and then warming to 37 °C to remove the excess SO2 gas. The specific activity of 73As used ranged from 0.0833 to 27.7 μCi/pmol arsenite. Binding experiments used the test macromolecule (DNA, histones or ferritin) diluted in cold water, 73As arsenite diluted in cold 150 mM NaCl, pH 7.5 buffer containing 100 mM Tris–HCl, a Brandel Model M-24C Membrane Harvester (Brandel Inc., Gaithersburg, MD) and either nitrocellulose filters (0.45 μm) soaked in a pH 7.5 solution containing 150 mM NaCl, 0.3% polyethylenimine and 100 mM Tris–HCl for protein experiments or Nylon filters (0.45 μm, Whatman Nytran SPC) for DNA experiments soaked in a pH 7.5 solution containing 150 mM NaCl and 100 mM Tris–HCl. The concentration of macromolecular targets used in our binding studies was approximately 33 μM for peptides and proteins and 200 μg/ml for DNA. Solutions were deoxygenated by bubbling nitrogen gas through them. Macromolecules and arsenite were incubated for 60 min at 2–8 °C prior to vacuum filtration. To reduce nonspecific binding, filtered macromolecules were washed three times with 2 ml of cold 150 mM NaCl, 100 mM Tris–HCl, pH 7.5, solution. The wash solution was kept on ice. Nonspecific binding was routinely determined by the addition of 100-fold excess of unlabelled arsenite to 73As arsenite. Gamma counting was done in a Packard Minaxi Auto gamma 5000. Counting efficiencies ranged from 8.4 to 8.7%. Protein was stripped from the Brandel NC-100 0.45 pore size nitrocellulose filters in a pH 10 buffer of 62.5 mM Na2CO2 containing 5% sodium lauryl sulfate. Using a Wallac Victor 1420 multilabel counter, the concentration of protein was determined spectrophotometrically using the Pierce bicinchoninic acid protein determination kit (using weighed bovine serum albumin as the standard). Protein recoveries from nitrocellulose filters ranged from 29 to 48%. DNA was stripped from the Nytran filters at room temperature with 0.3% sodium lauryl sulfate in 50 mM sodium phosphate buffered to pH 3.0. Quantification of the released DNA was done by a fluorescence binding assay using Hoechst 33258 and measuring the emitted fluorescence at 460 nm after excitation at 355 nm. DNA recoveries ranged from 45 to 96%. Triplicate samples were normally used for binding measurements; duplicate determinations of nonspecific binding samples via the 100fold excess cold carrier method were performed. For the ferritin saturation experiment of Fig. 4, only duplicate samples were run. All saturation binding experiments were repeated at least twice on different days. Representative saturation binding curves and Kd and Bmax values are given in this paper. Measurement day effects did not produce difficulties in these binding studies. Nonspecific binding is often fairly low in our studies but use of higher concentrations (from 100 to 1000 μM) of arsenite caused increasing nonspecific binding particularly to protein samples on nitrocellulose filters (Figs. 1–4). The amount of specific binding is estimated by subtracting the estimate of nonspecific binding from the estimate of total binding. The nonspecific binding is estimated by using a portion of the x axis that is 100-fold to the right of the region used to estimate the total binding. Kd and Bmax (maximum binding capacity) values were estimated from saturation binding experiments by use of the program Graph Pad Prism 4. The determined degree of ligand depletion that occurred in a saturation binding experiment was subtracted from the estimate of the actual free ligand concentration.
254
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Efficiency of arsenate reduction to arsenite/stability of arsenite. The presence of oxygen during the reduction of arsenate to arsenite may prevent complete reduction of arsenate. In addition, once the SO2 is removed, the arsenite may oxidize back to arsenate via oxygen exposure. To ascertain the reduction efficiency of arsenite by SO2, we determined the ratio of arsenite to arsenate by separating the radioactive arsenate from the arsenite using a strong anion exchange cartridge (Yalcin and Le, 2001). Stability of peptide sulfhydryl groups during the binding experiments. The o-phthalaldehyde fluorescence technique was used for determining the free sulfhydryl group concentration of proteins (Hissin and Hilf, 1976) both before and after 1 h incubations at 2–8 °C. Results Arsenate reduction and the stability of peptide sulfhydryl groups during the binding experiments Using a strong anion exchange cartridge to separate arsenate from arsenite (Yalcin and Le, 2001), our results indicated that the average % reduction of arsenate to arsenite was 93% and ranged from 85 to 99% (over 23 experiments). Over 15 experiments there was an average loss of 3.5% of the available sulfhydryl groups ranging from no loss observed to a maximum of 15% loss. Binding studies with DNA The concentrations of arsenite used in these studies were 30, 60, 100, 200, 300, 600 and 1000 μM. At no concentration did the total binding clearly exceed the nonspecific binding curve based on either (a) the 100-fold carrier nonspecific binding method or (b) the binding to the filter alone without any added DNA (data not shown). Thus, there was no positive evidence for 73As labeled arsenite specifically binding to calf thymus DNA even at 1000 times the arsenite concentrations found in exposed animals in vivo (Fig. 1). In one experiment using between 2 and 1000 μM arsenate, pentavalent 73 arsenate did not specifically bind to DNA. Binding studies with histone Type II-A and H3, H4 Histone type II-A is an unfractionated mixture of histone proteins obtained from calf thymus. It is expected to have a low concentration of Cys. Histone type II-A was studied up to about 1000 μM. Again, there was no sign of specific binding of radioactive arsenite to type II-A histones (Fig. 2). In this and in all subsequent cases where up to
Fig. 2. Saturation binding results for
73
As arsenite and Type II-A histones.
1000 μM arsenite was used as the ligand with nitrocellulose filters, the nonspecific binding curves were based on using 100-fold excess unlabelled arsenite. Saturation binding experiments with histone H3, H4 also showed that there was no specific binding of arsenite to histone H3, H4, in spite of the two Cys present in the amino acid sequence of histone H3. In one experiment using between 2 and 1000 μM arsenate, pentavalent 73arsenate also did not specifically bind to histone 3,4. Binding studies with ferritin Saturation binding studies with ferritin in the range of up to 1000 μM also showed no positive evidence of arsenite specific binding (Fig. 4). Values of the total and nonspecific binding were remarkably close over the whole range studied. Summary of the experimental conditions and results Table 1 summarizes the data on DNA and proteins examined for arsenite binding. Multiple positive controls were used concurrently in these studies (rat globin, tubulin, a peptide based on Keap-1 of amino acid sequence containing four cysteines KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY and peptide 10 based on human estrogen receptor alpha LECAWQGKCVEGTEHLYSMKCKNV containing three cysteines). In addition, other peptides and proteins were used in binding experiments both just before (the peptide based on Keap-1 and tubulin) and after (peptide 34, KACKACKAKAKAKK with two cysteines) the experiments with DNA, histones and ferritin. In all cases, the proteins and peptides run just before, during and after the DNA, histone and ferritin experiments demonstrated that the arsenite binding assay was working and capable of detecting any specific binding that occurred. For bovine tubulin, peptide 10, peptide 34 and peptide 29, the amount of binding observed in this series of experiments was consistent with their known Bmax values ranging from 15 to 80 nmol arsenite binding/mg protein. For rat globin used as a positive control, the ratio of observed counts was about 3 to 1 for rat globin samples to blanks. Discussion Arsenite and possible DNA binding
Fig. 1. Saturation binding results for
73
As arsenite and calf thymus DNA.
In DNA there are negatively charged phosphate groups, sugar moieties and the 4 bases which contain many aromatic moieties, amine and keto groups as possible binding sites. Based on the data of Fig. 1, none of the major functional groups of DNA seems to bind
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
255
Table 1 Summary of the specific binding results of arsenite and the studied macromolecules or peptides Macromolecule Experimental series DNA from calf thymus
Histone II-A from calf thymus
Histone 3,4 from cow
Ferritin from horse spleen
Positive controls Globin from rat
Cysteines/polypeptide chain
Specific binding Comment/reference observed?
No cysteines, but many amine, keto, hydroxyl, aryl, deoxyribose and phosphate groups are available as possible binding sites A commercially available mixture of histones
No
No
Collectively, the five major bovine histones contain a total of 595 amino acids but only 2 cysteines (C96, C110 of histone H3) and 10 histidines
Bovine histone H3 has 2 cysteines in 135 amino No acids exactly the same as human histone H3. Bovine histone H4 has 0 cysteines in 102 amino acids again the same as the human histone H4 sequence No 2 cysteines and 6 histidines in 175 amino acids in the light chain and 3 cysteines and 11 histidines in 182 amino acids in the heavy chain
5 cysteines in 289 amino acids based on Swiss Protein's amino acid sequence entries P01946 and P11517 4 cysteines in 46 amino acids
Peptide 29 based on an amino acid sequence from Keap-1 KYDCEQRRFYVQALLRAVRCHSLTNFLQMQLQKCEILQSDSRCKDY Tubulin from cow (whole protein, α and β units) 13 cysteines in 448 amino acids in the alpha chain (Q5E986), 8 cysteines in 445 amino acids in the beta chain (Q6B856 of Swisss Protein) Peptide 10 LECAWQGKCVEGTEHLYSMKC KNV 3 cysteines in 24 amino acids Peptide 24 RYCAVCNDYASGYHYGV WSCEGGKA 3 cysteines in 25 amino acids Peptide 34 KACKACKAKAKAKK 2 cysteines in 14 amino acids
Yes
Yes
Rat globin is a convenient inexpensive positive control for arsenite binding studies Kitchin and Wallace (2008)
Yes
Kitchin and Wallace (2008)
Yes Yes Yes
Kitchin and Wallace (2005) Kitchin and Wallace (2005)
arsenite well. Based on the negative results with 73As labeled arsenite, a tridentate binding ligand (Jiang et al., 2003), it seems unlikely at physiological concentrations that other trivalent arsenicals (e. g. MMA (III), DMA(III) or trivalent arsines) can bind to DNA directly.
virtually not at all to protein sites lacking even one exposed monothiol. This has been confirmed by direct measurements of the binding of 73As labeled arsenite to a series of peptides containing 0, 1, 2, 3 and 4 cysteines (Kitchin and Wallace, 2005).
Arsenite and possible binding to histones
Arsenite and possible binding to ferritin
From previous studies of peptides of known amino acid composition and sequence, it is clear that arsenite only binds well to one Cys or two Cys (dithiol) type of binding sites. No other single amino acid seems to be capable of binding arsenite with much affinity. Arsenite does not seem to bind to any significant extent to amino acid side chains such as amine, carboxyl, keto, hydroxyl and aryl. Three exceptions to this generalization are known. Arsenite binds to (Kd of 15 μM) and inhibits the enzyme acetylcholinesterase. In the protein acetylcholinesterase which possesses no Cys or lipoic acid moieties, the arsenite binds to two tyrosines in a diester manner (Page and Wilson, 1985). In a second case, arsenic has been demonstrated to bind to the molybdenum atom of xanthine oxidase (Stewart et al., 1984). Finally, following the injection of arsenite and selenate into rabbits, the complex seleno-bis (S-glutathionyl) arsinium ion, [(GS)(2)AsSe] (−), was found in the bile (Gailer et al., 2002). As shown in Fig. 2, type II-A histones showed no binding of arsenite, as might have been predicted based on its low concentration of Cys. Other alternative binding sites such as multiple hydroxyl or carboxyl or amine functional groups might make a metal binding site for other metals, but no good binding sites for arsenite were found in unfractionated type II-A histones. Histone H3, H4 also did not bind any arsenite in a specific manner, despite two Cys in the amino acid sequence of histone H3 (Fig. 3). As with the mixed histones found in type II-A histones, other noncysteine potential binding sites did not show any higher affinity binding sites for arsenite. Thus, nuclear histones do not seem to have any free reduced Cys available as good arsenite binding sites. Trivalent arsenic compounds such as arsenite bind more strongly to dithiol sites than to monothiol sites (Johnstone, 1963; Aposhian, 1989) and
Six different chemical forms of arsenic [arsenate, arsenite, MMA (V), MMA(III), DMA(V) and DMA(III)] release varying amounts of Fe from horse spleen ferritin (Ahmad et al., 2000). The light chain of horse ferritin contains 2 cysteines, 6 histidines and 175 amino acids. The heavy chain of horse ferritin contains 3 cysteines, 11 histidines and 182 amino acids. However as shown in Fig. 4, arsenite did not bind to the protein ferritin. Apparently the mechanism of arsenical action in releasing Fe from ferritin does not involve the binding of trivalent arsenic species to the protein part of the ferritin macromolecule and
Fig. 3. Saturation binding results for
73
As arsenite and histone H3, H4.
256
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257
Fig. 4. Saturation binding results for
73
As arsenite and horse spleen ferritin.
changing the protein conformation, function, or pore sizes or properties. Arsenicals are also capable of changing their valence and thus can act as reducing or oxidizing agents. The remaining attractive explanation of arsenic's Fe releasing effects on ferritin is that it simply acts as a redox agent and reduces Fe+++ to Fe++ (the releasable form of Fe). This is consistent with the finding that trivalent arsenicals released more Fe from ferritin than did the corresponding pentavalent arsenical (3.4 and 1.6 for arsenite and arsenate, respectively; 8.7 and 5.4 for MMA(III) and MMA(V), respectively; and 29.8 and 7.1 nM/min for DMA(III) and DMA(V), respectively (Ahmad et al., 2000). If oxidation of only 0.1% of an arsenical [e. g. DMA(III) to DMA(V)] was coupled to Fe reduction and release from ferritin, all the results observed in the Ahmad et al. (2000) study could be explained by arsenicals acting via redox properties (reduced arsenical → oxidized arsenical coupled to Fe+++ → Fe++). In the Fe release from ferritin experiments, the tested concentrations of the arsenicals were 10 mM, somewhat more than the 1.0 mM that was used in the binding studies in this paper. Arsenate and possible binding to DNA and histone 3,4 The pentavalent arsenate molecule has either one or two negative charges at physiological pH (the pKa values for H3AsO4 are 2.25, 6.77 and 11.6). Thus, it is not surprising that arsenate did not bind to negatively charged DNA. However, positively charged histones could be an excellent target for binding of arsenate but no such specific binding of arsenate to histone 3,4 was experimentally observed. In situ metal binding of other transition elements and possible sources of DNA oxidation Because this experimental study found no binding of arsenite to either DNA, type II-A histone and histone H3, H4, we must argue strongly that the possible binding of arsenicals and in situ oxidation of DNA is not a viable explanation of arsenic's carcinogenicity. Trivalent arsenicals are expected to be uncharged at physiological pH so are very electrically different from other positively charged transition elements. However, there is positive evidence that in situ binding and DNA oxidation occurs with other transition elements, for example Fe++, Cu++ and Ni++. There is about an 80/1 ratio of Fe to Cu in the human body (Chevion, 1988). With Fe++, the base sequence GGG binds Fe++ preferentially, resulting in structural distortion of DNA and enhanced DNA cleavage (Rai et al., 2005). With combined exposures to Cu++, H2O2 and a Cu++ binding histone peptide (CH3CO-AKRHRKCONH2), enhanced DNA fragmentation, formation of 8-hydroxy-2′-
deoxyguanosine (8-OHdG) and increased electron spin resonance signal were observed (Midorikawa et al., 2005). Chromatin can concentrate Ni++ more than over 5000-fold over serum, mostly via electrostatic attraction to the phosphate groups of DNA (Rokita and Burrows, 2001). In vitro, exposure to Ni++, nucleohistone and H2O2, enhances 8-OHdG formation (Bal and Kasprzak, 2002). There are known binding sites for Ni++ and Cu++ on histone H2A and histone H3 having the amino acid sequences of CAIH, ESHH and TESHHK (Bal and Kasprzak, 2002). Ni++ also binds to histone H1 and in the presence of H2O2 increases the formation of 8-OHdG in vitro (Huang et al., 1995). When DNA is irradiated, the presence of Mg++ causes an approximate doubling of the production of hydroxyl radicals (Anastassopoulou and Theophanides, 2002). Thus, with four different positively charged metals [Fe++, Cu++, Ni++ and Mg++], there is positive evidence demonstrating that in situ binding of biologically active metals close to DNA can lead to enhanced DNA oxidation and damage. Arsenic exposures have also been linked not only to the common reactive oxygen species, but also to arsenic containing species such as dimethylarsenic radical (Yamanaka et al., 1990), dimethylarsenic peroxyl radical (Yamanaka et al., 1990) and dimethylarsenic peroxide (Mizoi et al., 2005). Conclusion At concentrations up to approximately 1 mM, neither DNA nor any of the three proteins studied, type II-A histone, histone 3 or ferritin bound radioactive arsenite in a specific manner. Therefore, it appears highly unlikely that in situ binding of trivalent arsenicals, followed by in situ oxidative DNA damage, can account for arsenic's carcinogenicity. Arsenite, lacking a positive charge and being predominately bound to Cys moieties of peptides and proteins, does not seem to operate in a similar manner as Fe++, Cu++ and Ni++ in binding close to DNA and causing in situ oxidation of DNA via Haber–Weiss processes. The experimental evidence of this study (lack of arsenite binding to DNA and histones) does not rule out (a) diffusion of longer lived oxidative stress molecules, such as H2O2 into the nucleus and ensuing oxidative DNA damage, (b) redox chemistry by free (unbound) arsenicals in the nucleus or (c) arsenical-induced perturbations in Fe, Cu or other redox active metals which are already the known to oxidize DNA in vitro and in vivo. It is always difficult to rule out Feand Cu-mediated pathways in mammalian oxidative stress and carcinogenesis. Even in mammals with a substantial amount of arsenicals from environmental or experimental exposures, there are still orders of magnitude more Cu and Fe present in the mammalian organism than there is total arsenic. Acknowledgments We thank Drs. Russell Owen and Janice S Lee for reviewing this manuscript as part of EPA clearance procedures. References Ahmad, S., Kitchin, K.T., Cullen, W.R., 2000. Arsenic species that cause release of iron from ferritin and generation of activated oxygen. Arch. Biochem. Biophys. 382, 195–202. Anastassopoulou, J., Theophanides, T., 2002. Magnesium–DNA interactions and the possible relation of magnesium to carcinogenesis. Irradiation and free radicals. Crit. Rev. Oncol. Hematol. 42, 79–91. Aposhian, H.V., 1989. Biochemical toxicology of arsenic. In: Hodgson, E., Bend, J.R., Philpot, R.M. (Eds.), Reviews in Biochemical Toxicology, vol. 10. Elsevier Publishing Co., New York, pp. 265–299. Aust, S.D., Morehouse, L.A., Thomas, C.E., 1985. Role of metals in oxygen radical reactions. J. Free. Radic. Biol. Med. 1, 3–25. Bal, W., Kasprzak, K.S., 2002. Induction of oxidative DNA damage by carcinogenic metals. Toxicol. Lett. 127, 55–62. Chevion, M., 1988. A site-specific mechanism for free radical induced biological damage: the essential role of redox-active transition metals. Free Radic. Biol. Med. 5, 27–37.
K.T. Kitchin, K. Wallace / Toxicology and Applied Pharmacology 232 (2008) 252–257 Evans, M.D., Dizdaroglu, M., Cooke, M.S., 2004. Oxidative DNA damage and disease: induction, repair and significance. Mutat. Res. 567, 1–61. Gailer, J., George, G.N., Pickering, I.J., Prince, R.C., Younis, H.S., Winzerling, J.J., 2002. Biliary excretion of [(GS)(2)AsSe](−) after intravenous injection of rabbits with arsenite and selenate. Chem. Res. Toxicol. 11, 1466–1471. Hempstead, P.D., Yewdall, S.J., Fernie, A.R., Lawson, D.M., Artymiuk, P.J., Rice, D.W., Ford, G.C., Harrison, P.M., 1997. Comparison of the three-dimensional structures of recombinant human H and horse L ferritins at high resolution. J. Mol. Biol., 268, 424–448. Hissin, P.J., Hilf, R., 1976. A fluorometric method for determination of oxidized and reduced glutathione in tissues. Anal. Biochem. 74, 214–226. Huang, X., Kitahara, J., Zhitkovich, A., Dowjat, K., Costa, M., 1995. Heterochromatic proteins specifically enhance nickel-induced 8-oxo-dG formation. Carcinogenesis. 16, 1753–1759. Jiang, G., Gong, Z., Li, X.F., Cullen, W.R., Le, X.C., 2003. Interaction of trivalent arsenicals and metallothionein. Chem. Res. Toxicol. 16, 873–880. Johnstone, R.M., 1963. Sulfhydryl agents: arsenicals. In: Hochster, R.M., Quastel, J.H. (Eds.), Metabolic Inhibitors: A Comprehensive Treatise, vol. 2. Academic Press, New York, pp. 99–118. Kehrer, J.P., 1997. Mechanisms: free radicals and reactive oxygen species. In: Bond, J.A. (Ed.), General Principles, Toxicokinetics, and Mechanisms of Toxicology Vol 1 of Comprehensive Toxicology. Pergamon, pp. 275–301. Kehrer, J.P., 2000. The Haber–Weiss reaction and mechanisms of toxicity. Toxicology 149, 43–50. Kitchin, K.T., 2001. Recent advances in arsenic carcinogenesis: modes of action, animal model systems and methylated arsenic metabolites. Toxicol. Appl. Pharmacol. 172, 249–261. Kitchin, K.T., Wallace, K., Andrewes, P., 2003. Some chemical properties underlying arsenic's biological activity. In: Chappell, W.R., Abernathy, C.O., Calderon, R.L., Thomas, D.J. (Eds.), Arsenic Exposure and Health Effects. V. Elsevier, San Diego, pp. 345–354. Kitchin, K.T., Wallace, K., 2005. Arsenite binding to synthetic peptides based on the zinc finger and the estrogen binding region of the human estrogen receptor-alpha. Toxicol. Appl. Pharmacol. 206, 666–672. Kitchin, K.T., Wallace, K., 2006. Arsenite binding to synthetic peptides: the effect of increasing length between two cysteines. Journ. Biochem. Mol. Toxicol. 20, 35–38.
257
Kitchin, K.T., Wallace, K., 2008. The role of protein binding of trivalent arsenicals in arsenic carcinogenesis and toxicity. J. Inorg. Biochem. 102, 532–539. Klaunig, J.E., Kamendulis, L.M., 2004. The role of oxidative stress in carcinogenesis. Annu. Rev. Pharmacol. Toxicol. 44, 239–267. Lloyd, D.R., Phillips, D.H., Carmichael, P.L., 1997. Generation of putative intrastrand cross-links and strand breaks in DNA by transition metal ion-mediated oxygen radical attack. Chem. Res. Toxicol. 10, 393–400. Meneghini, R., 1997. Iron homeostasis, oxidative stress, and DNA damage. Free Radic. Biol. Med. 23, 783–792. Midorikawa, K., Murata, M., Kawanishi, S., 2005. Histone peptide AKRHRK enhances H2O2-induced DNA damage and alters its site specificity. Biochem. Biophys. Res. Commun. 333, 1073–1077. Mizoi, M., Takabayashi, F., Nakano, M., An, Y., Sagesaka, Y., Kato, K., Okada, S., Yamanaka, K., 2005. The role of trivalent dimethylated arsenic in dimethylarsinic acidpromoted skin and lung tumorigenesis in mice: tumor-promoting action through the induction of oxidative stress. Toxicol. Lett. 158, 87–94. National Research Council, 1999. Health effects of arsenic. Arsenic in Drinking Water. National Academy Press, Washington, DC., pp. 83–149. Page, J.D., Wilson, I.B., 1985. Acetylcholinesterase: inhibition by tetranitromethane and arsenite. Binding of arsenite by tyrosine residues. J. Biol. Chem. 260, 1475–1478. Rai, P., Wemmer, D.E., Linn, S., 2005. Preferential binding and structural distortion by Fe2+ at RGGG-containing DNA sequences correlates with enhanced oxidative cleavage at such sequences. Nucleic Acids Res. 33, 497–510. Rokita, S.E., Burrows, C.J., 2001. Nickel- and cobalt-dependent oxidation and crosslinking of proteins. Met. Ions. Biol. Syst. 38, 289–311. Stewart, R.C., Hille, R., Massey, V., 1984. Characterization of arsenite-complexed xanthine oxidase at room temperature. J. Biol. Chem. 259, 14426–14436. Surguladze, N., Patton, S., Cozzi, A., Fried, M.G., Connor, J.R., 2005. Characterization of nuclear ferritin and mechanism of translocation. Biochem. J. 388, 731–740. Thomas, D.J., Styblo, M., Lin, S., 2001. The cellular metabolism and systemic toxicity of arsenic. Toxicol. Appl. Pharmacol. 176, 127–144. Yalcin, S., Le, X.C., 2001. Speciation of arsenic using solid phase extraction cartridges. J. Environ. Monit. 3, 81–85. Yamanaka, K., Hoshino, M., Okanoto, M., Sawamura, R., Hasegawa, A., Okada, S., 1990. Induction of DNA damage by dimethylarsine, a metabolite of inorganic arsenics, is for the major part likely due to its peroxyl radical. Biochem. Biophys. Res. Comm. 168, 58–64.