Free Radical Biology & Medicine 45 (2008) 521–529

Contents lists available at ScienceDirect

Free Radical Biology & Medicine

j ourna l homepage: www.e lsev ie r.com/ locate / f reeradb iomed

Original Contribution

Susceptibility of rat astrocytes to DNA strand scission induced by activation of NADPHoxidase and collateral resistance to the effects of peroxynitrite

Andrea Guidarelli a, Letizia Palomba a, Mara Fiorani b, Orazio Cantoni a,⁎a Istituto di Farmacologia e Farmacognosia, Università degli Studi di Urbino Carlo Bo, 61029 Urbino, Italyb Istituto di Chimica Biologica Giorgio Fornaini, Università degli Studi di Urbino Carlo Bo, 61029 Urbino, Italy

Abbreviations: ATZ, 3-amino-1,2,4-triazole; BHT, budihydrorhodamine 123; DPPD, N,N'-diphenyl-1,4-phenyneiodonium; HQNO, 2-hepthyl-4-hydroxyquinoline; IFNnitro-L-arginine methyl ester; LPS, lipopolysaccharide; N12-myristate-13-acetate; Rhod 2-AM, Rhod 2-acetoxym⁎ Corresponding author. Istituto di Farmacologia e Fa

Studi di Urbino Carlo Bo, Via S. Chiara, 27+IBM-61029 U303521.

E-mail address: orazio.cantoni@uniurb.it (O. Cantoni

0891-5849/$ – see front matter © 2008 Elsevier Inc. Aldoi:10.1016/j.freeradbiomed.2008.05.005

a b s t r a c t

a r t i c l e i n f o

Article history:

Rat astrocytes accumulate e Received 21 December 2007Revised 24 April 2008Accepted 2 May 2008Available online 17 May 2008

Keywords:AstrocytesNADPH oxidasePeroxynitriteH2O2

DNA damageMitochondrial Ca2+

xtensive DNA single-strand breakage in response to agents promoting activationof NADPH oxidase. Proinflammatory stimuli, as bacterial lipopolysaccharide associated with interferon-γ,caused a rapid/robust burst of superoxide radicals, sensitive to NADPH oxidase inhibition, followed bydismutation to H2O2, the species resulting in DNA damage via a Fenton-type reaction. There was nocontribution of superoxide radical/H2O2 of mitochondrial origin and there was no evidence for theformation/involvement of peroxynitrite. On the other hand, astrocytes were virtually invulnerable to theDNA-damaging effects of exogenous peroxynitrite, an agent causing DNA strand scission in other cell types,via the Ca2+-dependent mitochondrial formation of superoxide radical/H2O2. Resistance was not dependenton scavenging of peroxynitrite but, rather, on insufficient mitochondrial Ca2+ accumulation. Hence, differentmanipulations resulting in an increase of the mitochondrial Ca2+ pool were invariably associated with theformation of DNA-damaging levels of H2O2. In conclusion, it appears that the strategy adopted by astrocytesto avoid inflammation-dependent genotoxic events, in particular those mediated by peroxynitrite, is toprevent mitochondrial Ca2+ accumulation, critical for the formation of secondary species largely responsiblefor DNA damage induced by peroxynitrite.

© 2008 Elsevier Inc. All rights reserved.

Introduction

Astrocytes, the most abundant cell type in the brain, regulate anarray of brain functions [1–4] and generally promote neuroprotection[5,6], but nevertheless respond to proinflammatory stimuli producinglarge amounts of toxic species as peroxynitrite, the coupling productof nitric oxide (NO) and superoxide radical. The latter may begenerated by an array of sources, as the mitochondrial respiratorychain or NADPH oxidase, largely expressed in the microglia and tosome extent also in the astrocytes themselves [7]. Finally, enzyme-catalyzed dismutase reactions will also compete with NO to convertthe superoxide anion to an additional toxic species, H2O2 [8].

The above information is consistent with the notion that, underneuroinflammatory conditions, activated astrocytes release reactivespecies, thereby potentially contributing to neuronal damage andmost likely causing damage in the astrocytes themselves. Indeed, it

tylated hydroxytoluene; DHR,lenediamine; DPI, diphenyle--γ, interferon-γ; L-NAME, Nω-O, nitric oxide; PMA, phorbol-ethyl ester.rmacognosia, Università deglirbino (PU), Italy. Fax: +39 0722

).

l rights reserved.

makes sense to predict that these cells are a direct target of reactiveoxygen/nitrogen species they produce on stimulation, either intra-cellularly or in their close vicinity.

Astrocytes, however, are virtually invulnerable to the lethal effectsof peroxynitrite or H2O2 [9,10] and this defensivemechanism allowingsurvival even under harsh neuroinflammatory conditions also impliesthe existence of mechanisms preventing additional serious outcomes.Persistent inflammation is indeed heavily implicated in the develop-ment and progression of tumors [11] and both peroxynitrite [12,13]and H2O2 [13] are potent DNA-damaging species with the potential ofcausing somatic mutations and neoplastic transformation [13]. Thelack of a specific defensive strategy would promote these events,thereby leading to the development of glial tumors when neuroin-flammation ensues. Resistance to the toxic effects of peroxynitriteand/or other reactive species [9,10] should indeed dramaticallyenhance the rate of transformation, when mutations accumulate asa consequence of extensive DNA damage.

The present study was performed with the aim of gatheringinformation on the susceptibility of astrocytes to the DNA-damagingeffects of species largely available in the inflamed brain. Inparticular, weasked the question of whether proinflammatory stimuli produce DNAstrand scission in cellswith lowNADPHoxidase activity. The existenceofa functional NADPHoxidase complex in astrocytes has been indeed for along time controversial [14]. Recent work from Abramov et al. [7],however, demonstrates that astrocytes express the full complement of

522 A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

NADPH oxidase subunits and, consistently, we recently describedNADPH oxidase-derived superoxide radical formation in immunosti-mulated astrocytes [15]. We also found that suppression of theconstitutive NO synthase inhibitory signaling [16], functional to NF-kBactivation [17,18], promotes release of NO in parallel with NADPHoxidase-derived superoxide radicals, thereby leading to early formationof toxic levels of peroxynitrite [15].

A second aim of the present study was to investigate whetherastrocytes are susceptible to the DNA-damaging effects of peroxynitrite.In this direction, it is important to note that peroxynitrite only partiallymimics the oxidizing effects mediated by hydroxyl radicals, since theyare involved in other reactions, including one-electron oxidation andnitration, that fail to promote DNA strand scission [19,20]. Our work hasprovided experimental evidence indicating that most of the DNAcleavage accumulated by cultured cells in response to a bolus ofperoxynitrite is not the consequence of a direct damage. Rather, itwouldappear thatmost of theDNAdamage ismediated bymitochondrialH2O2

generated in a time-dependent fashion via a Ca2+-dependent mechan-ism, in a reaction in which ubisemiquinone serves as an electron donor[21,22]. In addition,mitochondrial H2O2 formation and the ensuingDNAstrand scission may also take place via a Ca2+-independent mechanism,under conditions of enforced inhibition of complex III [23].

Weherein report that astrocytes are sensitive to thegenotoxic effectsof H2O2, either added as a bolus or generated via NADPH oxidaseactivation. Indeed, the latter appears to be the onlymechanismwherebyproinflammatory stimuli generate DNA cleavage in these cells. Astro-cytes, however, display collateral resistance to DNA strand scissioninduced by peroxynitrite, entirely due to prevention of mitochondrialCa2+ accumulation, a key event for themitochondrial formation of H2O2.

Materials and methods

Materials

Lipopolysaccharide (LPS) (Serotype 0127:B8), phorbol-12-myristate-13-acetate (PMA), diphenyleneiodonium (DPI), Nω-nitro-L-argininemethyl ester (L-NAME), L-methionine, Trolox, antimycin A, 2-hepthyl-4-hydroxyquinoline (HQNO), rotenone, myxothiazol, CaCl2, rutheniumred, LaCl3, catalase, 3-amino-1,2,4,-triazole (ATZ), butylated hydroxyto-luene (BHT), N,N'-diphenyl-1,4-phenylenediamine (DPPD), o-phenan-throline, H2O2, and the remaining chemicals were from Sigma-Aldrich(Milan, Italy). Interferon-γ (IFN-γ) (2×107 IU/mg of protein) andapocynin were purchased from R&D Systems (SPACE, Milan, Italy) andCalbiochem (Milan, Italy), respectively. Dihydrorhodamine 123 (DHR),Rhod 2-acetoxymethyl ester (AM), and MitoTracker Green were fromMolecular Probes (Leiden, The Netherlands).

Cell culture and treatment conditions

Primary cultures of cortical astrocytes, derived from neonatal 6- or7-day-old Sprague-Dawley rats (Charles River, Calco, Italy), wereprepared as described in [24]. A pure astrocyte preparation was con-firmed by immunocytochemistry using antibodies directed to glialfibril-lary acidic protein. Primary astrocytes were cultured in Eagle's minimalessential medium supplemented with 10% horse serum, 10% foetalbovine serum (HyClone Laboratories, Logan, UT), penicillin (100 U/ml),and streptomycin (100 µg/ml) (HyClone) in 35-mm Primaria dishes(Falcon, Becton Dickinson Labware, Franklin Lakes, NJ) gassed with anatmosphere of 95% air 5% CO2. For experiments, the cells were used2weeks after plating into 35-mm culture dishes. At the treatment stage,total cell number was between 2.0 and 2.5×105 cells/dish.

Stock solutions of IFN-γ, LPS, L-NAME, L-methionine, CaCl2, ruthe-nium red, LaCl3, catalase, and ATZ were freshly prepared in distilledwater. Trolox was dissolved in 1 M NaHCO3. Apocynin, DPI, PMA,antimycin A, HQNO, rotenone, myxothiazol, BHT, DPPD, and o-phenan-throline were dissolved in dimethyl sulfoxide or 95% ethanol. At the

treatment stage, the final concentration of dimethyl sulfoxide or ethanolwas never higher than 0.05%.

Astrocytes were exposed to either LPS/IFN-γ or PMA, in 2 ml ofEagle's minimal essential medium supplemented as detailed above.Treatments with H2O2 or peroxynitrite were performed in 2 ml ofsaline A (8.182 g/L NaCl, 0.372 g/L KCl, 0.336 g/L NaHCO3, and 0.9 g/Lglucose). Details on the treatment procedure employed with perox-ynitrite to prevent changes in pH, and/or additional nonspecific effects,as well as appropriate controls are reported in [22]. In the samereference a detailed procedure for treatment of permeabilized cellsand the protocol employed to achieve cell permeabilization are alsoillustrated. Peroxynitrite was synthesized as indicated in [25].

Measurement of DNA single-strand breakage by the alkaline halo assay

DNA single-strand breakage was determined using the alkaline haloassay developed in our laboratory [26], with the modificationsillustrated in [22]. Details on the processing of fluorescence imagesand on the calculation of the experimental results are also given in [22].DNA single-strand breakage was quantified by calculating the nuclearspreading factor value, representing the ratio between the area of thehalo (obtained by subtracting the area of the nucleus from the total area,nucleus halo) and that of the nucleus, from 50 to 75 randomly selectedcells/experiment/treatment condition. Results are expressed as relativenuclear spreading factor values calculated by subtracting the nuclearspreading factor values of control cells from those of treated cells.

Measurement of mitochondrial Ca2+

The cells were first exposed for 30 min (4°C) to 10 µM Rhod 2-AM,washed three times with saline A, and finally incubated for 5 h inEagle's minimal essential medium supplemented as detailed above(37°C). This two-step cold loading/warm incubation protocol achievesloading of Rhod 2 into themitochondria [27]. In some experiments, thecells were exposed for 20min to 0.5 µMMitoTracker Green prior to theend of the 5-h incubation period. After treatments, the cells werewashed three times, and finally analyzed with a fluorescence micro-scope (BX-51, Olympus Italia, Milan, Italy) equipped with a SPOT-RTcamera unit (Diagnostic Instruments, Delta Sistemi, Rome, Italy). Theexcitation and emission wavelengths were 540 and 590 nm (Rhod 2),and 488 and 515 nm (MitoTracker Green) with a 55-nm slit width forboth emission and excitation. Images were collected with exposuretimes of 100–400 ms, digitally acquired, and processed for fluores-cence determination at the single-cell level on a personal computerusing Scion Image software (Scion Corp., Frederick, MD). Meanfluorescence values were determined by averaging the fluorescencevalues of at least 50 cells/treatment condition/experiment.

DHR oxidation

The cells were first exposed for 3 min to peroxynitrite and thenpostincubated for 27 min in the presence of DHR (10 µM). In otherexperiments, the cells were exposed for 5 min to LPS/IFN-γ and thenpostincubated for 25min in the presence of DHR. After treatments, thecells were analyzed for DHR fluorescence with a BX-51 microscopeand the resulting images processed as described above. The excitationand emission wavelengths were 488 and 515 nm, respectively.

Catalase activity

Cells were washed twice with saline A, resuspended in the samemediumat a density of 5 × 106 cells/ml, andfinally sonicated three timeson ice with a Branson sonifier operating at 20 W for 15 s. The resultinghomogenates were centrifuged for 5 min at 18,000 ×g at 4°C. Catalaseactivity was determined spectrophotometrically in the supernatant asdescribed in [28].

Fig. 1. LPS/IFN-γ and reagent H2O2, unlike peroxynitrite, promote strand scission of genomic DNA. Representative fluorescent photomicrographs of cells exposed for 10 (B), 20 (C), or30 (D) min to LPS (1µg/ml)/IFN-γ (1000U/ml) and subsequently processed with the alkaline halo assay (controls in A). In other experiments the cells were exposed for 30min to H2O2

(50 µM) (E) or peroxynitrite (200 µM) (F) prior to analysis of DNA single-strand breakage. Panel G shows the time dependence for DNA single-strand breakage induced by LPS/IFN-γ,calculated from three different experiments in which the nuclear spreading factor values were estimated from 50 to 75 cells/experiment. The same approach was employed todetermine the extent of DNA strand scission 30min after addition of LPS, IFN-γ, or the two agents combined (H). Panel I illustrates the DNA-damaging responsemediated by a 30-minexposure to increasing concentrations of peroxynitrite (open circles) or H2O2 either with (closed squares) or without (closed circles) previous exposure to ATZ (10 mM). Finally, panelL shows the kinetics of repair of DNA single-strand breaks/alkali-labile sites estimated after a 30-min exposure to LPS/IFN-γ, H2O2 (50 µM), or peroxynitrite (100 µM)/antimycin A(1 µM). Results represent the means±SD calculated from 3 to 5 separate experiments. #Pb0.01; ##Pb0.001 compared to untreated cells (two-way ANOVA followed by Bonferroni'stest); ⁎Pb0.01; ⁎⁎Pb0.001 compared to untreated cells (one-way ANOVA followed by Dunnett's test). The results illustrated in panel L represent the means of two separateexperiments. The repair kinetics are first order with respect to time; linear regression analysis of the data points leads to t1/2 values indicated in the figure.

523A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

524 A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

Statistical analysis

The results are expressed asmeans±SD. Statistical differenceswereanalyzed by one-way ANOVA followed by Dunnett's test for multiplecomparison or two-way ANOVA followed by Bonferroni's test formultiple comparison. A value of Pb0.05 was considered significant.

Results and discussion

LPS/IFN-γ and reagent H2O2, unlike peroxynitrite, cause early strandscission of genomic DNA

Rat astrocytes in primary culture were exposed to LPS/IFN-γ forincreasing time intervals and immediately processed with the alkalinehalo assay, a sensitive technique that measures DNA strand breaks/

Fig. 2. DNA strand scission induced by LPS/IFN-γ, or PMA, is mediated by NADPH oxidase-derrotenone (0.5 µM),myxothiazol (5 µM), L-NAME (1mM), Trolox (1mM), or L-methionine (20min the presence of DHR (10 µM). The DHR fluorescence responsewas then quantified as detaileall of the above agents as well as with catalase (10 U/ml) (either enzymatically active or heat-i30min to LPS/IFN-γ. The level of DNA single-strand breaks/alkali-labile siteswasmeasured imin complete culture medium for 2 h in the absence (open circles) or presence of ATZ (closedactivitywas determined in cells with orwithout exposure to ATZ (inset). ND; not detectable. Dof LPS/IFN- γ. Results represent the means±SD calculated from 3 to 5 separate experiments.IFN-γ (A,B) or PMA (C) (one-way ANOVA followed byDunnett's test); #Pb0.01; ##Pb0.001 co

alkali-labile sites at the single-cell level [26]. As illustrated in Figs.1A–D,visual inspection of typical images obtained after ethidium bromidestaining reveals a time-dependent increase in the size of the halos,associated with a parallel reduction of the size of the nuclear remnants.This observation is consistent with the notion that LPS/IFN-γ triggerssome early event leading to time-dependent formation of DNA-damaging species. The extent of this response, calculated from 50 to70 cells/treatment condition in three separate experiments, is shown inFig.1G. Interestingly, both LPS and IFN-γwere required formaximalDNAstrand scission, since a remarkably lower effect was mediated by LPSalone and the sole IFN-γwas devoid of intrinsic DNA-damaging activity(Fig. 1H).

It is important to note that astrocytes were sensitive to the DNA-damaging effects of H2O2 (Figs. 1E and I) and virtually invulnerable tolevels of peroxynitrite (Figs. 1F and I) producing extensive damage in

ived H2O2. (A) The cells were first treated for 5 minwith DPI (0.5 µM), apocynin (10 µM),M), subsequentlyexposed for 5min to LPS/IFN-γ, and incubated for an additional 25mind underMaterials andmethods section. (B) The cells were treated as detailed in (A), withnactivated), DPPD (10 µM), BHT (200 µM), or o-phenanthroline (25 µM), and exposed formediately after the treatments using the alkaline halo assay. (C) The cells were incubatedcircles) and subsequently treated for increasing time intervals with LPS/IFN-γ. Catalase. The cells were treated as indicated in B, except that PMA (1 µg/ml) was used in the place⁎Pb0.001 compared to untreated cells; (⁎)Pb0.001 compared with cells exposed to LPS/mparedwith cells exposed to LPS/IFN-γ (two-wayANOVA followed by Bonferroni's test).

Fig. 3. Peroxynitrite promotes DNA cleavage in astrocytes supplemented with inhibitorsof complex III. Astrocytes were first treated for 5 min with rotenone or myxothiazol,subsequently exposed for 3 min to peroxynitrite (100 µM), and finally incubated for afurther 27 min in the absence or presence of antimycin A, or HQNO (10 µM). The cellswere subsequently analyzed for DNA damage (A) or DHR oxidation (B). Resultsrepresent the means±SD calculated from 3 to 5 separate experiments.⁎Pb0.001 or(⁎)Pb0.001, compared to untreated cells or cells treated with peroxynitrite/antimycinA, or HQNO, respectively (one-way ANOVA followed by Dunnett's test).

Table 1The effect of scavengers of peroxynitrite, antioxidants, or iron chelators on the DNAstrand scission induced by peroxynitrite/antimycin A or H2O2

Treatmenta Relative nuclear spreading factorb

100 µM peroxynitrite/antimycin A 5.46±0.53+Trolox 0.69±0.30⁎+L-Methionine 0.59±0.26⁎+DPPD 5.25±0.64+BHT 5.39±0.85+o-Phenanthroline 0.42±0.20⁎

50 µM H2O2 6.49±0.78+Trolox 6.72±0.95+L-Methionine 6.30±0.64+DPPD 6.57±0.82+BHT 6.15±0.93+o-Phenanthroline 0.55±0.31⁎

aThe cells were treated (30 min) with peroxynitrite/antimycin A or H2O2, in the absenceor presence of Trolox, L-methionine, DPPD, BHT, or o-phenanthroline, and immediatelyanalyzed with the alkaline halo assay.bThe relative nuclear spreading factor values represent themeans±SD calculated from 3to 5 separate experiments.⁎ Pb0.01 compared with cells exposed to peroxynitrite/antimycin A (one-way ANOVAfollowed by Dunnett's test).

525A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

other cell types [21–23]. In addition, DNA single-strand breaksdetected under alkaline conditions after exposure to LPS/IFN-γ orH2O2 were repaired with identical kinetics (Fig. 1L).

These results indicate that the proinflammatory cocktail LPS/IFN-γtriggers a rapid DNA-damaging response in rat astrocytes, also sensitiveto reagent H2O2 but displaying collateral resistance to peroxynitrite.

DNA strand scission induced by LPS/IFN-γ, or PMA, is mediated byNADPH oxidase-derived superoxide radicals/H2O2

Fig. 2A provides evidence of extensive DHR oxidation after a 30-minexposure to LPS/IFN-γ, sensitive to two different NADPH oxidase inhib-itors, DPI andapocynin [29], but insensitive to rotenoneormyxothiazol. Itis important to note that rotenone, which prevents the entry of electronsin the respiratory chain through complex I, would abolish superoxideradical formation induced by peroxynitrite, taking place via amechanismrequiring mitochondrial Ca2 accumulation and electron transport in therespiratory chain [22]. The same is to be expected with myxothiazol,preventing ubisemiquinone formation via inhibition of the electron flowfrom the reduced coenzyme Q to cytochrome c1 [30].

These results suggest that the early formation of superoxideradicals induced by LPS/IFN-γ is not mediated by peroxynitrite, aconclusion further supported by the lack of effect of L-NAME (a NOsynthase inhibitor), or either of two different peroxynitrite scavengers(Trolox or L-methionine, Fig. 2A). Finally, the above effects weremeasured under the same conditions in which there was no evidenceof nitrotyrosine immunodetection (not shown).

Measurement of DNA strand scission in cells treated as detailed inFig. 2A demonstrated that the DNA-damaging response evoked byLPS/IFN-γ is sensitive to DPI and apocynin but insensitive to rotenone,myxothiazol, L-NAME, Trolox, and L-methionine (Fig. 2B). The speci-ficity of the effects mediated by these agents is emphasized by theresults of experiments described below.

Taken together, these results indicate that LPS/IFN-γ promotesDNA single-strand breakage via a mechanism requiring superoxideradical formation, largely resulting from activation of the enzymeNADPH oxidase with hardly any contribution from the mitochondrialrespiratory chain.

An additional important conclusion from the above results is thatperoxynitrite is not involved in the observed DNA strand scission.Hence, superoxide anion requirement for DNA strand scission isnot functional to the formation of peroxynitrite but most likely to itsdismutation to H2O2. The sensitivity to exogenous catalase, but not tothe boiled enzyme (Fig. 2B), is consistent with the involvement of

H2O2, a conclusion further supported by the observation that previouscatalase depletion with ATZ remarkably accelerates the rate of DNAcleavage induced by LPS/IFN-γ (Fig. 2C). Under the same conditions ofcatalase inhibition a leftward shift in the dose–response curve for DNAstrand scission mediated by H2O2 was also detected (Fig. 1I). Asindicated in the inset to Fig. 2C, astrocytes display low catalaseactivity, virtually undetectable after exposure to ATZ.

It is also important to note that the DNA-damaging response evokedby reagent H2O2 (Table 1), as previously observed with LPS/IFN-γ(Fig. 2B), was insensitive to Trolox or L-methionine. Finally, the intra-cellular iron chelator, o-phenanthroline, suppressedDNA strand scissioninduced by either reagent H2O2 (Table 1), or LPS/IFN-γ (Fig. 2B), therebyimplying the involvement of radical species resulting from the inter-action between H2O2 and divalent transition metals (e.g., iron).

These results indicate that the DNA damage induced by LPS/IFN-γis in fact mediated by H2O2, a conclusion also consistent with thesuperimposable repair kinetics previously measured in immunosti-mulated cells or after exposure to reagent H2O2 (Fig. 1L).

526 A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

Additional experiments revealed that PMA promotes a DNA-damaging response suppressed by NADPH oxidase inhibitors, ironchelators, and enzymatically active catalase, but insensitive torotenone, myxothiazol, L-NAME, Trolox, and L-methionine (Fig. 2D).Hence, an additional stimulus for NADPH oxidase [7] also promotesH2O2-dependent DNA single-strand breakage.

Taken together, the results presented in this section demonstratethat H2O2 is critically involved in the formation of DNA lesions in

cultured astrocytes exposed to LPS/IFN-γ, or PMA. H2O2, however, isnot the final DNA-damaging species but, rather, some reactive species,possibly the hydroxyl radical, generated by the reaction of the oxidantwith divalent transition metals, as Fe2+. As a final note, formation ofH2O2 ensues on dismutation of superoxide radicals, produced viaactivation of NADPH oxidase, with hardly any contribution of mito-chondria in this process.

Astrocytes are sensitive to DNA cleavage induced by peroxynitrite via theCa2+-independent mechanism

The results thus far presented indicate that rat astrocytes accu-mulate extensive DNA single-strand breakage in response to proin-flammatory stimuli leading to NADPH oxidase activation. The lack ofsensitivity to DNA damage induced by peroxynitrite was nextinvestigated in greater detail. Resistance was not dependent on thehigh peroxynitrite-scavenging capacity of these cells, since exposure toperoxynitrite was accompanied by the onset of an intense nitrotyr-osine immunostaining, similar to that observed in other cell typessusceptible to peroxynitrite-induced DNA strand scission (not shown).In addition, and most importantly, astrocytes acquired sensitivity toDNA strand scission in the presence of the complex III inhibitorantimycin A (Fig. 3A). Similar results were obtained by replacingantimycin A with HQNO, a structurally unrelated complex III inhibitor[31]. Under these conditions, therewas also evidence of DHR oxidation(Fig. 3B), given after peroxynitrite in order to prevent direct oxidation.Peroxynitrite alone, as expected, failed to promote oxidation of DHR(Fig. 3B). Finally, both the DNA damage (Fig. 3A) and the DHRoxidation(Fig. 3B) observed in the presence of antimycin A or HQNO weresensitive to rotenone or myxothiazol, thereby emphasizing the mito-chondrial origin of the radical species mediating DNA strand scission.

The above results also provide a positive control for the experimentsshowing that rotenone, or myxothiazol, fails to prevent superoxideradical formation (Fig. 2A) and DNA damage (Fig. 2B) induced by LPS/IFN-γ. In addition, as observedwith either LPS/IFN-γ (Fig. 2B) or reagentH2O2 (Table 1), DNA damage induced by peroxynitrite/antimycin Awassuppressed by o-phenanthroline (Table 1). Only DNA strand scissioninduced by peroxynitrite/antimycin A was, however, sensitive to Troloxor L-methionine, two established scavengers of peroxynitrite (Table 1).Hence, a positive control is also provided for the experiments showingthat Trolox and L-methionine fail to prevent superoxide radicalformation (Fig. 2A) and DNA damage (Fig. 2B) induced by LPS/IFN-γ.The specificity of the effects of Trolox and L-methionine in preventingthe DNA damage mediated by peroxynitrite/antimycin A, which mayalso act as antioxidants, is emphasized by the lack of effect of twoestablished antioxidants, namely DPPD and BHT (Table 1). DPPD andBHT also failed to prevent DNA damage induced by reagent H2O2

(Table 1), LPS/IFN-γ (Fig. 2B), and PMA (Fig. 2D).Collectively, the above results indicate that astrocytes, resistant

to the DNA-damaging effects of peroxynitrite mediated by the Ca2+-dependent mechanism, display collateral sensitivity to the Ca2+-independent mechanism leading to DNA strand scission underconditions of enforced inhibition of complex III [23].

Fig. 4. Resistance to peroxynitrite is circumvented by manipulations increasingmitochondrial Ca2+. (A) Permeabilized cells were exposed for 10 min to peroxynitrite(100 µM), or peroxynitrite/antimycin A, in the absence or presence of rotenone,myxothiazol, catalase (enzymatically active or heat-inactivated), ruthenium red (200nM), or LaCl3 (100 µM). The level of DNA single-strand breaks/alkali-labile sites wasmeasured immediately after the treatments. (B) Permeabilized cells were exposed for10min to increasing concentrations of CaCl2 in the absence or presence of peroxynitrite.(C) Permeabilized cells were exposed for 10 min to peroxynitrite, or peroxynitrite/CaCl2(3 µM), in the absence or presence of various additions, as indicated in the figure.Results represent the means±SD calculated from 3 to 5 separate experiments. ⁎Pb0.001compared to untreated cells; (⁎)Pb0.001 compared to peroxynitrite/antimycin A (A) orperoxynitrite/CaCl2 (C)-treated cells (one-way ANOVA followed by Dunnett's test).#Pb0.01; ##Pb0.001 compared with cells exposed to CaCl2 alone (two-way ANOVAfollowed by Bonferroni's test).

527A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

We therefore performed experiments in permeabilized cells tovalidate the notion that DNA strand scission mediated by peroxyni-trite in astrocytes supplemented with antimycin A indeed takes placevia a Ca2+-independent mechanism.

As indicated in Fig. 4A, antimycin A promoted extensive DNAcleavage after exposure to an otherwise ineffective concentration ofperoxynitrite also inpermeabilized cells. DNA cleavagewas sensitive torotenone, myxothiazol, and enzymatically active catalase, but insensi-tive to ruthenium red, used at levels specifically preventingmitochon-drial Ca2+ uptake [32] and lanthanium ions, known to competitivelyinhibit Ca2+ uptake [33]. Similar results were obtained in experimentsin which antimycin A was replaced with HQNO (not shown). It is im-portant to note that, as it will be discussed below (Fig. 4C), rutheniumred, or lanthanium ions, abolishes the DNA-damaging responseevoked by peroxynitrite via the electron transport/mitochondrialCa2+-dependent mechanism.

Fig. 5. Poor mitochondrial Ca2+ accumulation accounts for resistance to DNA strand scissionRhod-2 AM and MitoTracker Green and then treated for 10 min with either peroxynitrite (10MitoTracker Green fluorescence appear yellow. (B) Rhod 2-AM preloaded cells were exposedDPI. Fluorescence was then quantified as detailed under Materials and methods. (C) The cellsfinally incubated for a further 30 min with increasing concentrations of peroxynitrite (closeordinate axis indicates the DNA damage caused by LPS/IFN-γ. Results illustrated in the insetDNA-damaging response evocked by peroxynitrite in cells preexposed to LPS/IFN-γ/DPI. Recompared to untreated cells (B) or LPS/IFN-γ/DPI/peroxynitrite treated cells (C) (one-waperoxynitrite alone (two-way ANOVA followed by Bonferroni's test).

It therefore appears that astrocytes are sensitive to the indirectDNA-damaging effects of peroxynitrite mediated by the Ca2+-independent mechanism occurring under conditions of enforcedinhibition of complex III. H2O2, as previously established [21,23],appears to be involved in the formation of DNA lesions and indeedtheir repair kinetics were superimposable on those observed afterexposure to either reagent H2O2 or LPS/IFN-γ.

Resistance to peroxynitrite is circumvented by manipulations increasingmitochondrial Ca2+ content

The results illustrated in Fig. 4B indicate that exposure to 1–30 µMCaCl2 fails to promote DNA cleavage in permeabilized astrocytes, aspreviously observed with the low dose of peroxynitrite employed inthe experiments shown in Fig. 4A. A remarkable accumulation ofDNA lesions was, however, observed when the two treatments were

induced by peroxynitrite. (A) Representative microscope images of cells preloaded with0 µM) or LPS/IFN-γ. In the merged image (overlay) regions containing both Rhod-2 andfor 10 min to peroxynitrite, or to LPS/IFN-γ with or without prior exposure (5 min) towere first treated for 5 min with DPI, subsequently exposed for 5 min to LPS/IFN-γ and

d circles). The extent of DNA cleavage was immediately assessed. The open circle in thereport the effect of rotenone, myxothiazol, catalase, and heat inactivated catalase on thesults represent the means±SD calculated from 3 to 5 separate experiments. ⁎Pb0.001y ANOVA followed by Dunnett's test); #Pb0.01; ##Pb0.001 compared to cells with

528 A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

combined. This response was dependent on the concentration ofCa2+ added to the permeabilized cell system. Interestingly, the DNAcleavage induced by peroxynitrite and 3 µM CaCl2, the lowest Ca2+

concentration producing themaximal response (Fig. 4B), was sensitiveto rotenone, myxothiazol, enzymatically active catalase, rutheniumred, and lanthanium ions (Fig. 4C).

These results are therefore consistent with the notion thatperoxynitrite may induce DNA cleavage also in astrocytes, providedthat a parallel mitochondrial accumulation of Ca2+ also takes place.Under these conditions, peroxynitrite promotes the mitochondrialformation of H2O2 via a Ca2+-dependent mechanism also requiringelectron transport through complex I, as previously observed in U937cells [21–23]. It follows that resistance of astrocytes to the DNA-damaging effects of peroxynitrite is largely explained by the lack ofmitochondrial Ca2+ accumulation.

Poor mitochondrial Ca2+ accumulation accounts for resistance to DNAstrand scission induced by peroxynitrite

The fluorescent probe Rhod-2-AMwas used to assess the extent ofmitochondrial Ca2+ accumulation in astrocytes exposed to peroxyni-trite or LPS/IFN-γ. Fig. 5A provides evidence for a faint fluorescenceresponse in cells exposed to peroxynitrite in remarkable contrast withthe extensive fluorescence clearly detected in cells exposed to theproinflammatory mixture. Fig. 5B summarizes the results obtained inthree different experiments. Fig. 5A also provides evidence for amitochondrial localization of this fluorescent signal, emphasized bythe yellow color resulting from the overlay of the Ca2+ signal (Rhod-2,red) and the mitochondrial signal (MitoTracker Green, green).

These results are therefore consistent with the notion that poormitochondrial Ca2+ accumulation accounts for resistance to DNAstrand scission induced by peroxynitrite and provide the bases forinvestigating whether peroxynitrite would induce DNA single-strandbreakage in intact astrocytes with enhanced mitochondrial Ca2+

content. In these experiments, the cells were exposed for 5min to LPS/IFN-γ and DPI (or apocynin, not shown) to prevent NADPH oxidase-dependent DNA cleavage, and for an additional 30 min to concentra-tions of peroxynitrite failing to cause DNA damage in the absence ofadditional manipulations (Fig. 1I). It was found that cells acquiresusceptibility to DNA cleavage induced by peroxynitrite (Fig. 5C)under the same conditions inwhich DPI (or apocynin, not shown) failsto affect the extent of mitochondrial Ca2+ accumulation mediated byLPS/IFN-γ (Fig. 5B). The resulting DNA strand scission was sensitive torotenone, myxothiazol, and enzymatically active catalase (inset toFig. 5C), thereby implying a role for the mitochondrial respiratorychain in events leading to the mitochondrial formation of superoxideradical/H2O2 under conditions associated with the mitochondrialaccumulation of Ca2+.

Taken together, these results demonstrate that resistance ofastrocytes to DNA strand scission induced by peroxynitrite largelydepends on the ability of these cells to prevent the mitochondrialclearance of Ca2+.

Conclusions

The results presented in this study provide relevant informationas to the strategies adopted by astrocytes to cope with the DNA-damaging effects mediated by reactive species released on activationby proinflammatory stimuli. It appears that LPS/IFN-γ causes DNAcleavage entirely mediated by time-dependent release of superoxideradical/H2O2 via NADPH oxidase activation. More generally, we canstate that astrocytes are sensitive to the DNA-damaging effects of H2O2

and that H2O2-dependent DNA strand scission ensues on activation ofNADPH oxidase. Hence, these results indicate an important functionalconsequence of activation of astrocytic NADPH oxidase, therebyfurther emphasizing its existence and functionality [7]. In addition,

these results strongly suggest that astrocytes adopt no special strategyto cope with the DNA-damaging effects of H2O2. These cells, therefore,most likely proficiently repair H2O2-dependent DNA damage (Fig. 1L),as it can be expected from cells living in a tissue with high rates ofoxygen consumption, and thus with large basal formation of super-oxide radical/H2O2.

In remarkable contrast with these findings, astrocytes were foundhighly resistant to DNA strand scission induced by peroxynitrite,thereby indicating that the evolutionary mechanisms developed bythese cells to cope with this DNA-damaging agent are not proficientin repairing the induced DNA lesions. This makes sense sinceperoxynitrite is an inhibitor of DNA repair [34,35]. In this perspective,astrocytes select an ingenious strategy to preserve their DNA integrityand function in the presence of peroxynitrite. These cells activatemechanisms preventing mitochondrial Ca2+ accumulation and theensuing mitochondrial formation of superoxide radical/H2O2. As aconsequence, peroxynitrite is not a DNA-damaging agent for astro-cytes, unless some parallel event enforces the process of mitochon-drial Ca2+ accumulation.

Acknowledgments

This work was supported by grants from Ministero dell'Universitàe della Ricerca Scientifica e Tecnologica, Progetti di Ricerca di InteresseNazionale, and from the Associazione Italiana per la Ricerca sulCancro, (O. Cantoni).

References

[1] Kim, J. H.; Park, J. A.; Lee, S. W.; Kim,W. J.; Yu, Y. S.; Kim, K.W. Blood-neural barrier:intercellular communication at glio-vascular interface. J. Biochem. Mol. Biol.39:339–345; 2006.

[2] Verkhratsky, A.; Toescu, E. C. Neuronal-glial networks as substrate for CNSintegration. J. Cell. Mol. Med. 10:826–836; 2006.

[3] Volterra, A.; Meldolesi, J. Astrocytes, from brain glue to communication elements:the revolution continues. Nat. Rev., Neurosci. 6:626–640; 2005.

[4] Wiesinger, H.; Hamprecht, B.; Dringen, R. Metabolic pathways for glucose inastrocytes. Glia 21:22–34; 1997.

[5] Heales, S. J.; Lam, A. A.; Duncan, A. J.; Land, J. M. Neurodegeneration orneuroprotection: the pivotal role of astrocytes. Neurochem. Res. 29:513–519; 2004.

[6] Trendelenburg, G.; Dirnagl, U. Neuroprotective role of astrocytes in cerebralischemia: focus on ischemic preconditioning. Glia 50:307–320; 2005.

[7] Abramov, A. Y.; Jacobson, J.; Wientjes, F.; Hothersall, J.; Canevari, L.; Duchen, M. R.Expression and modulation of an NADPH oxidase in mammalian astrocytes.J. Neurosci. 25:9176–9184; 2005.

[8] Liochev, S. I.; Fridovich, I. The effects of superoxide dismutase on H2O2 formation.Free Radic. Biol. Med. 42:1465–1469; 2007.

[9] Bolaños, J. P.; Almeida, A. Modulation of astroglial energy metabolism by nitricoxide. Antiox. Redox Signal. 8:955–965; 2006.

[10] Dringen, R.; Kussmaul, L.; Gutterer, J. M.; Hirrlinger, J.; Hamprecht, B. Theglutathione system of peroxide detoxification is less efficient in neurons than inastroglial cells. J. Neurochem. 72:2523–2530; 1999.

[11] Coussens, L. M.; Werb, Z. Inflammation and cancer. Nature 420:860–867; 2002.[12] Beckman, J. S.; Koppenol, W. H. Nitric oxide, superoxide, and peroxynitrite: the

good, the bad, and ugly. Am. J. Physiol. 271:C1424–C1437; 1996.[13] Ohshima, H.; Tatemichi, M.; Sawa, T. Chemical basis of inflammation-induced

carcinogenesis. Arch. Biochem. Biophys. 417:3–11; 2003.[14] Wilkinson, B. L.; Landreth, G. E. The microglial NADPH oxidase complex as a source

of oxidative stress in Alzheimer's disease. J. Neuro-inflamm. 3:30; 2006.[15] Palomba, L.; Amadori, A.; Cantoni, O. Early release of arachidonic acid prevents an

otherwise immediate formation of toxic levels of peroxynitrite in astrocytesstimulated with lipopolysaccharide/interferon-γ. J. Neurochem. 103:904–913;2007.

[16] Palomba, L.; Persichini, T.; Mazzone, V.; Colasanti, M.; Cantoni, O. Inhibition ofnitric-oxide synthase-I (NOS-I)-dependent nitric oxide production by lipopoly-saccharide plus interferon-γ is mediated by arachidonic acid. Effects on NF-kBactivation and late inducible NOS expression. J. Biol. Chem. 279:29895–29901;2004.

[17] Colasanti, M.; Persichini, T.; Menegazzi, M.; Mariotto, S.; Giordano, E.; Caldarera,C. M.; Sogos, V.; Lauro, G. M.; Suzuki, H. Induction of nitric oxide synthase mRNAexpression. Suppression by exogenous nitric oxide. J. Biol. Chem. 270:26731–26733;1995.

[18] Togashi, H.; Sasaki, M.; Frohman, E.; Taira, E.; Ratan, R. R.; Dawson, T. M.; Dawson,V. L. Neuronal (type I) nitric oxide synthase regulates nuclear factor kB activity andimmunologic (type II) nitric oxide synthase expression. Proc. Natl. Acad. Sci. U. S. A.94:2676–2680; 1997.

[19] Cadet, J.; Douki, T.; Ravanat, J. L. One-electron oxidation of DNA and inflammationprocesses. Nat. Chem. Biol. 2:348–349; 2006.

529A. Guidarelli et al. / Free Radical Biology & Medicine 45 (2008) 521–529

[20] Lee, Y. A.; Yun, B. H.; Kim, S. K.;Margolin, Y.; Dedon, P. C.; Geacintov, N. E.; Shafirovich,V. Mechanism of oxidation of guanine in DNA by carbonate radical anion, adecomposition product of nitrosoperoxycarbonate.Chem. Eur. J.13:4571–4581; 2007.

[21] Guidarelli, A.; Tommasini, I.; Fiorani, M.; Cantoni, O. Essential role of the mito-chondrial respiratory chain in peroxynitrite-induced strand scission of genomicDNA. IUBMB Life 50:195–201; 2000.

[22] Guidarelli, A.; Sciorati, C.; Clementi, E.; Cantoni, O. Peroxynitrite mobilizes calciumions from ryanodine-sensitive stores, a process associated with the mitochondrialaccumulation of the cation and the enforced formation of species mediatingcleavage of genomic DNA. Free Radic. Biol. Med. 41:154–164; 2006.

[23] Guidarelli, A.; Cerioni, L.; Cantoni, O. Inhibition of complex III promotes loss of Ca2+

dependence for mitochondrial superoxide formation and permeability transitionevoked by peroxynitrite. J. Cell Sci. 120:1908–1914; 2007.

[24] Rose, K.; Goldberg, M. P.; Choi, D. V. Cytotoxicity in murine neocortical cell culture.In: Tyson, C.A., Frazier, J.M. (Eds.), Methods in Toxicology, 1. Academic Press,New York, pp. 46–60; 1992.

[25] Radi, R.; Beckman, J. S.; Bush, K. M.; Freeman, B. A. Peroxynitrite oxidation ofsulfhydryls. The cytotoxic potential of superoxide and nitric oxide. J. Biol. Chem.266:4244–4250; 1991.

[26] Sestili, P.; Cantoni, O. Osmotically driven radial diffusion of single-stranded DNAfragments on an agarose bed as a convenient measure of DNA strand scission. FreeRadic. Biol. Med. 26:1019–1026; 1999.

[27] Trollinger, D. R.; Cascio, W. E.; Lemasters, J. J. Mitochondrial calcium transients inadult rabbit cardiac myocytes: inhibition by ruthenium red and artifacts caused bylysosomal loading of Ca2+-indicating fluorophores. Biophys. J. 79:39–50; 2000.

[28] Beutler, E. Red cell metabolism. In: Grune, Stratton (Eds.), A manual of biochemicalmethods. New York, pp. 131–134; 1984.

[29] Irani, K.; Xia, Y.; Zweier, J. L.; Sollott, S. J.; Der, C. J.; Fearon, E. R.; Sundaresan, M.;Finkel, T.; Goldschmidt-Clermont, P. J. Mitogenic signaling mediated by oxidants inRas-transformed fibroblasts. Science 275:1649–1652; 1997.

[30] Brand, K. A.; Hermfisse, U. Aerobic glycolysis by proliferating cells: a protectivestrategy against reactive oxygen species. FASEB J. 11:388–395; 1997.

[31] van Ark, G.; Berden, J. A. Binding of HQNO to beef-heart sub-mitochondrialparticles. Biochim. Biophys. Acta 459:119–127; 1977.

[32] Carafoli, E. Intracellular calcium homeostasis. Annu. Rev. Biochem. 56:395–433;1987.

[33] Thomas, C. E.; Reed, D. J. Effect of extracellular Ca++ omission on isolatedhepatocytes. II. Loss of mitochondrial membrane potential and protection byinhibitors of uniport Ca++ transduction. J. Pharmacol. Exp. Ther. 245:501–507; 1988.

[34] Jaiswal, M.; LaRusso, N. F.; Nishioka, N.; Nakabeppu, Y.; Gores, G. J. Human Ogg1, aprotein involved in the repair of 8-oxoguanine, is inhibited by nitric oxide. CancerRes. 61:6388–6393; 2001.

[35] Chien, Y-H.; Bau, D-T.; Jan, K-Y. Nitric oxide inhibits DNA-adduct excision innucleotide excision repair. Free Radic. Biol. Med. 36:1011–1017; 2004.