Original Contribution

DIRECT, REAL-TIME SENSING OF FREE RADICAL PRODUCTION BYACTIVATED HUMAN GLIOBLASTOMA CELLS

PHILIP MANNING,* CALUM J. MCNEIL,* JONATHAN M. COOPER,† and EDWARD W. HILLHOUSE‡

*Department of Clinical Biochemistry, The Medical School, University of Newcastle upon Tyne, Newcastle upon Tyne NE2 4HH,United Kingdom,†Department of Electronics and Electrical Engineering, University of Glasgow, Glasgow G12 8LT, United

Kingdom, ‡Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, United Kingdom

(Received30 July 1997;Revised24 October1997;Re-revised26 November1997;Accepted26 November1997)

Abstract—Primary brain injury initiates a cascade of events which result in secondary brain damage. Although, atpresent, the biochemical and molecular mechanisms of nerve cell death are not well understood, sufficient evidence nowexists to implicate free radicals in this brain injury response. In the light of the current understanding on the role of freeradicals in cell mortality, we report on the use of two specific sensors, which we use to measure the direct, simultaneousand real time electrochemical detection of both superoxide (O2

•2) and nitric oxide (NO), produced by activatedglioblastoma cells. The development and application of these novel methods has enabled us to show that both thecytokine-mediated induction of the enzymes responsible for the generation of these radical species, and the metabolicrequirements of the cell can modulate cell messenger release. Importantly, the data collected provides dynamicinformation on the time course of free radical production, as well as their interactions and their involvement in theprocess of cell death. In particular, one of the major advances afforded by this technology is the demonstration thatsuppression of one of either of the two cellular generated radical species (NO and O2

•2) leads directly to a correspondingincrease in the species that was not being deliberately inhibited or scavenged. This finding may indicate a mechanisminvolving inter-enzyme regulation of free radical production in glial cells (a phenomenon which may, in future, also beshown to operate in other relevant cell models). © 1998 Elsevier Science Inc.

Keywords—Free radical, Superoxide, Nitric oxide, Electrode, Immobilized cytochromec, Amperometric detection

INTRODUCTION

A variety of excitotoxins and neurotoxic species,1 in-cluding free radicals such as superoxide, (O2

•2) andnitric oxide (NO),2–4 are produced in response to braininjury induced by ischemia, resulting in a cascade ofmessenger “events” leading to secondary cell and tissuedamage.5 In addition, it has been proposed that thesespecies may be responsible for irreversible damage tocell components and constituents, ultimately leading tocell death.6 As a result, it is now clear that therapeuticintervention in such processes is likely to lead to animproved outcome.

Not withstanding these advances in our understandingof ischemia, the biochemical and molecular mechanisms

underlying neuronal cell death are not well characterizedand despite the evidence that free radicals may be in-volved in brain injury, there have been no attempts toexplore the role of free radical inter-relationships in cellculture models. This is primarily because the provisionof direct biochemical evidence for the involvement offree radicals has been severely hampered by analyticaldifficulties resulting from an inability to measure freeradical production directly. As a consequence, the in-volvement of free radical species in neuropathology hasonly been inferred both in vitro and in vivo by studiesusing a number of indirect techniques7–10 which requirelengthy assay times and which can only provide limiteddiscrete information regarding the dynamics of cell sig-naling processes.

Previously, studies have shown, using indirect end-point measurements, that exogenous neurotrophic andneurotoxic factors as well as cytokines can modify neu-ronal injury responses, for instance, in cell culture stud-

Address correspondence to: Dr. C.J. McNeil, Department of ClinicalBiochemistry, The Medical School, University of Newcastle uponTyne, Newcastle upon Tyne, NE2 4HH, UK; Tel:144 191 222 8259;Fax: 144 191 222 6227/7991; E-Mail: calum.mcneil@ncl.ac.uk.

Free Radical Biology & Medicine, Vol. 24, Nos. 7/8, pp. 1304–1309, 1998Copyright © 1998 Elsevier Science Inc.Printed in the USA. All rights reserved

0891-5849/98 $19.001 .00

PII S0891-5849(97)00455-3

1304

ies, tumor necrosis factor-a (TNF) induces proliferationof microglia and can produce reactive changes, includingO2

•2 production and the release of cytotoxic factors,11

whereas, studies in mice have shown that TNF canprotect neurons by stimulating antioxidant pathways.12

Excitotoxic insult results in the accumulation of reactiveoxygen species, primarily from microglia, the residentCNS macrophage.13 As a consequence, the physiologicalrole of O2

•2, its relationship to neurotoxicity throughreaction with NO to form peroxynitrite (ONOO2)14 andthe factors governing the possible contribution of O2

•2 toNO physiology have all been of considerable interest tous, particularly as it has been proposed that the balancebetween these radicals may regulate the actions of theindividual species, as well as modulate their overalleffects at both cellular and molecular levels. In addition,the in vitro evidence of O2

•2 production by nitric oxidesynthase under specific conditions, such as argininedeprivation,15,16 indicate that free radical control andproduction is more complicated than originally thought.Indeed we can provide evidence that O2

•2 controls theavailability of NO to its cellular targets and thus maymodulate the neuroendocrine response.

In this paper, we now describe the use of a sensor forO2

•2, based on the covalent immobilization of cyto-chromec at the surface of a gold electrode, which en-ables the real time investigation of cellular O2

•2 gener-ation with no physical disturbance of the glial cell pop-ulation. The device has a number of obvious advantagesover previous spectrophotometric or chemiluminescentmeasurements, including that of the provision of direct,real time information on the dynamics of O2

•2 produc-tion, thus enabling, for the first time, the study of the timecourse and extent of free radical production by activatedglial cells. In addition, we have used a commercial NOsensor (World Precision Instruments) to allow simulta-neous direct investigation of the combined role of NOand O2

•2, thus providing us with complementary infor-mation regarding the relationships between these twofree radicals.

MATERIALS AND METHODS

Preparation and calibration of electrodes

In order to detect O2•2, cytochromec was covalently

coupled to a 2 mmO.D. gold working electrode, encasedin KEL-F plastic (Biotech Instruments Ltd., Kimpton,UK), using a thiol-linking molecule, 3,39-dithiobis(sul-fosuccinimidylpropionate) (DTSSP, Pierce and Warri-ner, Chester, UK). As a result, the protein was covalentlybound to the metal, thus constituting the (bio)sensingelement of the device. To achieve this, the Au workingelectrode was cleaned thoroughly using a Bioanalytical

Systems Inc. PK-4 electrode polishing kit (BAS Techni-col Ltd., Congleton, Cheshire, UK).

After polishing with 5mm diamond slurry the elec-trode was rinsed with absolute ethanol and dried with amedical tissue. Thereafter, the electrode was polished ona ‘microcloth’ polishing pad, supplied with the polishingkit, using a 0.015mm alumina slurry in distilled water.After alumina polishing the electrode was rinsed thor-oughly in distilled water, sonicated for 5 min and driedusing a medical tissue.

The optimum procedure for cytochromec immobili-zation was as follows: a 50 mM solution of DTSSP, wasprepared in 100ml of distilled water immediately prior touse.17 The surface of the working electrode was com-pletely immersed in this solution for 1 min at 22°C toensure monolayer coverage of DTSSP on the gold elec-trode by chemisorption via Au–S bond formation.17 Af-ter incubation the electrode surface was carefully dried,and further incubated in 250ml of a 2 mM cytochromec(Sigma) solution prepared in 100 mM phosphate buff-ered saline (pH 7.6) at 4°C for 20 h. The O2

•2 electrodewas found to respond optimally for a period of 6–8 h inuse. Therefore, in order to maximize their efficacy formeasurements in cell culture under different conditions,new electrodes were prepared for each set of cellularexperiments. The O2

•2 electrode was used in conjunctionwith a Ag/AgCl wire reference electrode encased in a20 mm glass shaft (Clark Electromedical InstrumentsLtd., Reading, UK) and an ‘in house’ low-noise poten-tiostat linked to a two-channel chart recorder. The super-oxide electrode was tested before use using O2

•2 gener-ated by xanthine and XOD (Sigma) as previouslydescribed.17,18

The NO electrode (ISO-NO meter, World PrecisionInstruments Ltd., Sarasota, USA) was calibrated19 in astirred solution of a 0.1 M KI, acidified with 0.1 MH2SO4. NO was generated in solution by adding a knownvolume of 1M KNO2. A peak current response of 350 pAat the NO electrode corresponded to approximately75 nM NO.

Cell culture

A glioblastoma multiform cell line (A172) suspendedin Dulbecco’s modified eagle’s medium (DMEM) con-taining 10% FBS, 2 mM L-glutamine, 500 IU ml21

penicillin/streptomycin solution and 25 ngml21 ampho-tericin-b was seeded into 24 well cell culture plates atbetween 13 102–13 106 cells per well and incubated at37°C with 5% CO2 for 24 h. This procedure was repeatedfor a meningioma cell line (S572) which were used as anegative control for superoxide generation.

1305Real-time free radical sensing

Real-time free radical measurement

Immediately prior to experimentation, the DMEMmedium was replaced with 1 ml per well of Hank’sbalanced salt solution (HBSS). A cytochromec modifiedelectrode was placed in a fixed position 1 mm above thecells in one of the prepared wells and a steady baselinecurrent response achieved. The cells were then activatedby the addition of phorbol 12-myristate 13-acetate(PMA) dissolved in DMSO and any subsequent responsefollowed. The current rate due to O2

•2 production as afunction of cell number was established. This procedurewas repeated using lipopolysaccharide (LPS) dissolvedin DMSO as a stimulus. As a further control PMA andLPS were added to wells containing 1 ml of HBSS in theabsence of cells.

The experimental procedure was repeated with theintroduction of 25 Uml21 g-interferon (g-INF) into theculture medium at the time of cell seeding. Both O2

•2

and NO electrodes were positioned in the well containingcells and steady baseline current responses achieved. Thestimulus was added and the relative time course of NOand O2

•2 production as a function of stimulus concen-tration recorded. Free radical scavenging experimentsinvolved the use of bovine Cu/Zn superoxide dismutase(SOD, specific activity 3,300 U mg21, a gift from Dr. D.Saunders, Gru¨nenthal GmbH, Stolberg, Germany) andcarboxyphenyl-4,4,5,5-tetramethyllimidazoline-1-oxyl-3-oxide (Carboxy-PTIO, Calbiochem-NovabiochemLtd., Beeston, Nottingham, UK).

RESULTS AND DISCUSSION

Electrode response and specificity

In the first instance, the O2•2 sensor response was

examined using XOD to show that, under conditions ofsubstrate saturation, when the reaction kinetics are en-zyme-concentration dependent, the measured rate of cur-rent generation (nA min21) was linearly dependent uponXOD concentration (nM), and hence upon the rate ofO2

•2 production (y5 0.1023 1 0.405,n 5 9, r 5 0.98).Under the same conditions, using spectrophotometricmeasurement, 0.5mM XOD gave a rate of cytochromecreduction in solution of approximately 15mM min21. Inorder to assess the specificity of the sensor, it was shownthat O2

•2 produced by 0.5mM XOD, was completelyscavenged by 640 Uml21 superoxide dismutase (SOD),demonstrating that the observed current rates were due toO2

•2 generation (Fig. 1). Importantly, no sensor responsewas seen in the presence of 100mM H2O2, eliminatingthe possibility of peroxide being detected as a result ofspontaneous O2

•2 dismutation. In addition, although it iswell recognized that cytochromec in solution can bereduced by a variety of species, the O2

•2 electrode did

not respond to the addition of ascorbic acid at concen-trations up to 100mM. This observation would suggeststrongly that covalent immobilization of cytochromec ata gold electrode surface-modified with a short-chain al-kane thiol in the manner described herein conferred alarge degree of specificity with respect to cytochromecreduction.

Although peroxynitrite may re-oxidize cytochromec,20 at pH 7.6 the rate is sufficiently low that no crossreactivity between the NO and the O2

•2 electrode wasobserved (data not shown). Similarly, no O2

•2 responsewas recorded at the NO electrode (due to the presence ofthe permselective membrane). Since there was no inter-electrode interference, it was apparent that we would beable to measure simultaneously both O2

•2 and NO freeradical production by cells in culture.

Production of O2•2 by activated glioblastoma cells

Control experiments showed no production of O2•2

from the cell culture vessels either before or after theaddition of stimulants. Furthermore, non-activated glio-blastoma cells did not produce any detectable O2

•2.Following activation with either PMA or LPS, however,the cultured human glioblastoma cells displayed signif-

Fig. 1. SOD dependent inhibition of the current response observed atthe superoxide electrode in the presence of xanthine and xanthineoxidase. A known concentration of SOD was added to phosphatebuffered saline (pH 7.6) containing 3 mM xanthine in a reactionvolume of 1 ml. After 1 min preincubation, O2

•2 generation wasinitiated by the addition of XOD (0.5mM in PBS, pH 7.6) and theresulting current response recorded. All experiments were carried out at22°C. The results shown are the mean (62 SD) of triplicate measure-ments.

1306 P. MANNING et al.

icant rates of O2•2 production. For example, addition of

PMA gave O2•2 responses which were seen within 1–2

min of cell activation and which were sustained for up to1 h after initiation of the reaction. The magnitude ofresponse was dependent upon the number of cells used(Fig. 2). In addition, the current generated (pA min21)was linearly related to the concentration of the stimuliPMA (nM) and LPS (ng ml21). These relationships aredescribed by the equations: y5 0.423 1 8.6 (n 5 5, r 50.96) for PMA; and y5 0.213 1 5.0 (n 5 5, r 5 0.96)for LPS. Previous studies using primary rat microglialcell cultures13 have reported similar findings using longincubation periods and indirect spectroscopic measure-ment of cytochromec reduction. When glioblastomacells were pre-incubated with 640 Uml21 SOD prior tostimulation with either PMA or LPS, no response wasobserved at the superoxide electrode under the condi-tions described in Fig. 2. Thus, the O2

•2 specificity of theelectrode was confirmed.

Activation of human glioblastoma cells with LPSproduced similar results, although it can be seen fromFig. 2 that the responses were appreciably smaller thanthose obtained using PMA. This finding is in contrastwith those reported for rat primary glial cell cultures,which were reported to be insensitive to LPS.13 This maybe explained by the improved sensitivity of our tech-nique for measuring O2

•2 or by differences betweenhuman glioblastoma cells and rat microglial cells in theirresponses to LPS. Pre-incubation of the glioblastoma

cells with g-INF prior to activation with LPS signifi-cantly enhanced the responses obtained (e.g., from 50616 pA min21 to 906 19 pA min21, n 5 5, p , .01 using1 3 106 cells). As expected, a meningioma cell line didnot produce O2

•2 activation with either PMA or LPS.

Simultaneous Measurement of O2•2 and NO by

activated glioblastoma cells

Under the experimental conditions in which there wasno induction of iNOS and no added L-arginine, LPSactivation resulted in O2

•2 detection alone (Fig. 3a).When the enzyme was induced and provided with L-arginine, NO production was seen to reach a steady statelevel within 1 min and was sustained for at least 7 min(Fig. 3b). Superoxide production was not observed overthis time period under these experimental conditions.When D-arginine was used as the enzyme substrate, noNO production was observed, which reflects the knownstereospecificity of the enzyme.21 In contrast, however,O2

•2 production was detected, suggesting that a complexrelationship between the production of these free radicalsmay exist. Indeed, we have observed that when theproduction of NO in response to LPS has decayed, thereis a subsequent increase in the response at the O2

•2

electrode (Fig. 3c). The most likely explanation is thatwhen NO production is maximal this radical reacts quan-titatively with O2

•2 to produce ONOO2 before it can bedetected at the cytochromec modified electrode. The rateconstant for the reaction of NO and O2

•2 has beenestimated22 to be of the order of 4.3–6.73 109 M21s21,while the heterogeneous rate constant for the reactionbetween O2

•2 and the modified electrode surface is be-tween 0.18 and 3.4 s21.17,18,23

Selective scavenging of either radical was found toresult in a concomitant increase in detection of the spe-cies that was not being suppressed. For example, theaddition of the NO scavenger, carboxy-PTIO to the cel-lular system immediately before activation with LPSresulted in an inhibition of the NO response in a scav-enger concentration-dependent manner (Fig. 4a) and asimultaneous increase in the response at theO2

•2electrode. On repeating this procedure with variousconcentrations of SOD a similar dose dependent inhibi-tion of the O2

•2 response was observed with a corre-sponding increase in NO detection (Fig. 4b). This is thefirst time this phenomenon has been demonstrated di-rectly and in real-time in a biological system.

In conclusion, the ability of human glioblastoma cellsto produce both O2

•2 and NO has been demonstrated inreal time using direct amperometric detection systems. Inthe context of real-time measurement, the O2

•2 detectionsystem showed a lag of the order of 1 min. This may bedue to the electrode being positioned 1 mm above the

Fig. 2. Electrochemical measurement of O2•2 generation from varying

numbers of human glioblastoma or meningioma cells. Immediatelyprior to measurement, the culture medium was removed and replacedwith 1 ml of HBSS. Activation of the cells was achieved with 25 nMPMA (closed symbols) or 25 ngml21 LPS (open symbols). Each pointrepresents the mean (6 SD) of triplicate measurements. The volume ofDMSO used at each stimulus concentration was constant (15ml). TheO2

•2 electrode was held at a fixed distance approximately 1 mm abovethe cells. All experiments were carried out at 22°C.

1307Real-time free radical sensing

cells, however, a lag associated with stimulation of cel-lular processes leading to free radical production is alsolikely to contribute. This delay in measurement is stillextremely short compared with previous studies whichhave relied on the indirect interpretation, up to 48 h aftercellular activation, of end products which may have littlerelevance to short lived and important cellular events.Direct electrochemical detection of free radicals has dis-

Fig. 3. (a) Time course of O2•2 and NO production by 13 106

LPS-activated human glioblastoma cells in the absence of extracel-lular L-arginine andg-INF induction. (b) Time course of O2

•2 andNO production by 13 106 LPS-activated human glioblastoma cellsafter preincubation withg-INF for 24 h and in the presence ofextracellularL-arginine (10 mM). (c) Chart trace showing the decayin the NO response followed by a concomitant rise in O2

•2. In thiscase 13 106 human glioblastoma cells were preincubated withg-INF for 24 h, prior to activation with LPS.L-Arginine was notadded to the extracellular medium. All experiments were carried outat 22°C and the final concentration of LPS used in each case was25 ngml21. Both electrodes were held at a fixed distance approxi-mately 1mm above the cells. These traces are a representative seriesof chart recordings from numerous experiments (n . 10). In allcases the same qualitative behaviour was observed under the con-ditions described.

Fig. 4. Simultaneous measurement of NO and O2•2 during: (a) car-

boxy-PTIO dependent scavenging of NO produced by 13 106 glio-blastoma cells activated using 25 ngml21 LPS. Scavenging of NO ledto an increase in O2

•2 detection at the superoxide electrode; (b) SODdependent inhibition of electrochemical O2

•2 detection from 13 106

glioblastoma cells activated using 25 ngml21 LPS. Selective scaveng-ing of O2

•2 by SOD led to a concomitant increase in the current due toNO detection. Each point represents the mean (6 SD) of triplicatemeasurements. All experiments were carried out at 22°C.

1308 P. MANNING et al.

tinct advantages for the neurobiologist. These include therapid, simple, interference-free detection of free radicalsinstead of indirect measurement of presumed end prod-ucts. The measurement time scale afforded by this ap-proach is particularly important given the chemical re-activity and reported messenger function of free radicals.

Acknowledgements —Philip Manning is grateful to the Sir WilliamLeech Charitable Trust for a studentship. The authors are pleased toacknowledge The Wellcome Trust (046357/Z/95/Z), EPSRC (GR/J30189) and Action Research for financial support.

REFERENCES

1. Lees, G. Contributory mechanisms in the causation of neurode-generative disorders.Neurosci.54:287–322; 1993.

2. Schmidley, J. W. Free radicals in central nervous system ischemia.Stroke21:7–12; 1990.

3. Phillis, J. W. A radical view of cerebral ischemic injury.Prog.Neurobiol.42:441–448; 1994.

4. Dawson, V. L.; Dawson, T. M. Nitric oxide neurotoxicity.J. Chem. Neuroanat.10:179–190; 1996.

5. Gentleman, D. Preventing secondary brain damage after headinjury: a multi-disciplinary challenge.Injury 21:305–308; 1990.

6. Patel, M.; Day, B. J.; Crapo, J. D.; Fridovich, I.; McNamara, J. O.Requirement for superoxide in excitotoxic cell death.Neuron16:245–355; 1996.

7. Tsuchihashi, Y.; Kitamura, T.; Fujita, S. Immunofluorescencestudies of the monocytes in the injured rat brain.Acta Neuropathol.53:213–219; 1981.

8. Zini, I.; Tomasi, A.; Grimaldi, D.; Vannini, V.; Agnati, L. F.Detection of free radicals during brain ischemia and reperfusion byspin trapping and microdialysis.Neurosci. Lett.138:279–282;1992.

9. Archer, S. Measurement of nitric oxide in biological models.FASEB J.7:349–360 1993.

10. Meda, L.; Cassatella, M. A.; Szendrel, G. I.; Otvos Jr., L.; Baron,P.; Villalba, M.; Ferrari, D.; Rossi, F. Activation of microglial cellsby b-amyloid protein and interferon-g. Nature 374:647–650;1995.

11. Lipton, S. A. HIV-related neuronal injury. Potential therapeuticintervention with calcium channel antagonists and NMDA antag-onists.Mol. Neurobiol.8:181–196; 1994.

12. Bruce, A.. J.; Boling, W.; Kindy, M. S.; Peschon, J.; Kraemer,P. J.; Carpenter, M. K.; Holtsberg, F. W.; Mattson, M. P. Alteredneuronal and microglial responses to excitotoxic and ischemicbrain injury in mice lacking TNF receptors.Nature Med.2:788–794; 1996.

13. Colton, C. A.; Snell, J.; Chernyshev, O.; Gilbert, D. L. Induction ofsuperoxide anion and nitric oxide production in cultured microglia.Ann. N.Y. Acad. Sci.738:54–63; 1994.

14. Radi, R.; Beckman, J. S.; Bush, K. M.; Freeman, B. A. Peroxyni-trite-induced membrane lipid peroxidation. The cytotoxic potentialof superoxide and nitric oxide.Arch. Biochem. Biophys.288:481–487; 1991.

15. Pou S.; Pou S. W.; Bredt D. S.; Snyder S. H.; Rosen G. M.Generation of superoxide by purified brain nitric oxide synthase.J. Biol. Chem.267:24173–24176; 1992.

16. Culcasi M.; Lafon–Cazal M.; Pietri S.; Bockaert J. Glutamatereceptors induce a burst of superoxide via activation of nitric oxidesynthase in arginine-depleted neurons.J. Biol. Chem.269:12589–12593; 1995.

17. Manning, P. Investigation of the role of free radicals in biologicalsystems using direct, real-time electrochemical measurement.PhDThesis, University of Newcastle upon Tyne; 1997.

18. Cooper, J. M.; Greenough, K. R.; McNeil, C. J. Direct electrontransfer reactions of immobilized cytochromec at modified goldelectrodes.J. Electroanal. Chem.347:267–275; 1993.

19. Tsukahara, H. Gordienko, D. V.; Gologorsky, M. S. Continuousmonitoring of nitric oxide release from human umbilical veinendothelial cells.Biochem. Biophys. Res. Commun.193:722–729;1993.

20. Thomson, L.; Trujillo, M.; Telleri, R.; Radi, R. Kinetics of cyto-chromec21 oxidation by peroxynitrite: implications for superoxidemeasurement in nitric oxide-producing systems.Arch Biochem.Biophys.319:491–497; 1995.

21. Hibbs Jr, J. B.; Vavrin, Z.; Taintor, R. R.L-Arginine is required forthe expression of the macrophage effector mechanism causingselective metabolic inhibition in target cells.J. Immunol.138:550–565; 1987.

22. Czapski, G.; Goldstein, S. The role of the reactions of NO withsuperoxide and oxygen in biological systems: A kinetic approach.Free Rad. Biol. Med.19:785–794; 1995.

23. Jiang, L.; McNeil, C. J.; Cooper, J. M. Electrochemical coupling ofcomponents of the biological electron transport chain to modifiedsurfaces: molecular recognition between cytochromec peroxidaseand cytochromec. Angew. Chem.107:2610–2612; 1995.

1309Real-time free radical sensing