e

smdseiti

l

Archives of Biochemistry and BiophysicsVol. 373, No. 2, January 15, pp. 447–450, 2000doi:10.1006/abbi.1999.1579, available online at http://www.idealibrary.com on

Lucigenin: Redox Potential in Aqueous Media andRedox Cycling with O2

2 Production1

Ivan Spasojevic, Stefan I. Liochev, and Irwin Fridovich2

Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710

Received August 31, 1999, and in revised form October 19, 1999

The use of lucigenin luminescence as a measure of[O2

2] has been questioned because lucigenin has beenshown to be capable of mediating the production ofO2

2. This being the case, lucigenin can signal the pres-nce of O2

2 even in systems not producing it in theabsence of lucigenin. The reduction potential of lu-cigenin should be in accord with its ability to mediateO2

2 production; but it has not heretofore been mea-ured in aqueous media. The problems facing sucheasurement are the insolubility of the divalently re-

uced form, which deposits on the electrode, and thelow conformational transition that follows the secondlectron transfer and which interferes with reversibil-ty. We have now used rapid scan cyclic voltammetryo determine that the reduction potential for lucigenins 20.14 6 0.02 V versus the normal hydrogen elec-

trode. This value applies to both the first and the sec-ond electron transfers to lucigenin and it is in accordwith the facile mediation of O2

2 production by thiscompound. © 2000 Academic Press

Key Words: superoxide; lucigenin; cyclic voltamme-try; reduction potential.

The sequence of reactions leading to the chemilumi-nescence of lucigenin, Luc21,3 involves several steps asfollows: (a) univalent reduction of Luc21 to Luc•1; (b)coupling of Luc•1 with O2

2 yielding a dioxetane; (c)decomposition of the dioxetane into two molecules of

1 This work was supported by grants from the National Institutesof Health, the Council for Tobacco Research, USA, the AmyotrophicLateral Sclerosis Association, NC Biotech Collaborative FundingAssistance Program, and Aeolus Pharmaceuticals.

2 To whom correspondence should be addressed. Fax: (919) 684-8885.

3 Abbreviations used: NHE, normal hydrogen electrode; Luc21,ucigenin; Luc•1, univalently reduced lucigenin; Luc••, divalently

reduced lucigenin prior to conformation transition; Luc, divalentlyreduced lucigenin after conformational transition; SOD, superoxidedismutase.

0003-9861/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

the N-methyl acridone, one of which is in an electron-ically excited state; and finally (d) emission of a photonfrom the excited state acridone as it returns to groundstate. The relevant structures are shown in Scheme I.Since O2

2 is a required reactant, Luc21 luminescence isinhibitable by superoxide dismutase and has been, andcontinues to be, employed as a measure of O2

2. Theproblem with this use of Luc21 is that Luc•1 readilyautoxidizes. Luc21 can thus mediate O2

2 production,much as do quinones and viologens, leading to artifac-tual detection of O2

2. Luc21 is thus a detector whichmisleads because it can cause production of that whichis being detected (1–5).

Nevertheless, perhaps because of the convenienceand sensitivity of luminescence methods, the use ofLuc21 as a measure of O2

2 continues. Indeed, one recentreport indicates that Luc21 cannot redox cycle withproduction of O2

2 (6). This report is based upon aninvalid extrapolation of an incorrect redox potential.We now present cyclic voltammetry data that estab-lishes the redox potential of Luc21 in aqueous mediaand which is compatible with its previously demon-strated ability to redox cycle with production of O2

2.

MATERIALS AND METHODS

Cyclic voltammetry (CV) measurements were performed using acomputer supported CH Instruments Model 600 voltammetric ana-lyzer. A three-electrode setup was used with a 3-mm-diameter but-ton glassy carbon working electrode from Bioanalytical Systems(BAS), a Ag/AgCl reference electrode (BAS, 3 M NaCl filling solution,Vycor porous glass tip), and a 0.5-mm Pt wire as an auxiliary elec-trode. In preliminary experiments in nonaqueous media the refer-ence electrode was Ag/AgNO3, which was used in order to minimizethe liquid–liquid junction potential. Prior to each CV measurementthe working electrode was polished with 0.3 mm Al2O3, rinsed thor-oughly with deionized water, wiped with paper tissue (without rub-bing), and left for several minutes to air dry. Slow CV scans (below0.1 V/s) were performed in a 5-ml glass beaker with a Teflon cap. Allthree electrodes were placed in the center of the measured solution(typically 3 ml) as close to each other as possible and the solution waspurged with argon. Fast CV scans (up to 4000 V/s) were performed ina “micro-cell” setup in which working and reference electrodes wereplaced against each other in a face-to-face fashion separated by a

447

V

448 SPASOJEVIC, LIOCHEV, AND FRIDOVICH

small piece of thin paper tissue. A 0.25-mm Pt wire was coaxiallywound around the working electrode/reference electrode interfaceand served as an auxiliary electrode. Typically, the working elec-trode was placed half-way through a thick wall rubber tubing (1 inchlong) followed by addition of 0.2 ml of the solution to be analyzed intothe remaining space of the tubing; a small piece of paper tissue wasplaced at the bottom of the solution (thus laying on the tip of theworking electrode). The solution was then gently bubbled with hu-midified argon gas (ultra-high-purity grade, less than 1 ppm ofoxygen) for at least 15 min. In the stream of argon, the reference andauxiliary electrodes were then placed against the working electrodeand the CV measurements immediately taken. Aqueous solutionscontaining 0.05 M Tris buffer, pH 7.8, 0.1 M NaCl (slow scans), or 3M NaCl (fast scans) were typically used. The measured potentialswere standardized against potassium ferrocyanide/potassium ferri-cyanide couple (7), and the redox potential of the Ag/AgCl referenceelectrode used was found to be 10.22 V vs NHE.

Lucigenin (Luc21, Bis-N-methylacridinium nitrate) (97%) was pur-chased from Aldrich and used without further purification.

RESULTS

To obtain a reliable redox potential by cyclic voltam-metry one must have reversibility. Thus both the re-duction wave and the reoxidation wave must be observ-able and the standard redox potential is then taken tobe midway between the reduction and the reoxidationwaves. Luc21 usually gives irreversible cyclic voltam-mograms in aqueous media. The reason for this is theinsolubility of Luc in water and the relatively slowconformational change that follows the reduction of

SCHEME I. Behavior of lucigenin in sup

SCHEME II. Behavior of lucigeni

Luc1• to Luc. The two tricyclic acridine ring systemsare joined by a single bond in Luc21 and in Luc•1, butby a double bond in Luc. This allows Luc21 and Luc•1 tominimize steric hindrance by having the two joinedacridine rings at right angles to each other. In contrastthe double bond in Luc forces a more nearly coplanararrangement. This was first elucidated by Ahlberg etal. (8) and was recently considered by Rodriguez-Amaro et al. (9). This situation is represented by thereactions in Scheme II. In this scheme the successiveunivalent reductions of the two covalently bonded acri-dine rings of lucigenin is followed by the slow transi-tion of the Luc•• biradical into the conformationallyrestrained Luc.

This problem can be circumvented by using rapidscan rates so that reduction and reoxidation will occurbefore the conformational change can occur. Of courserapid scan rates impose stray capacitance and ohmicvoltage drop problems, but these can be circumventedby using high ionic strength solutions and by closeapposition of the electrodes, as described under Mate-rials and Methods.

Figure 1A presents the cyclic voltammogram of 1.0 mMLuc21 in deaerated aqueous solution at 0.1 V/s. At 20.38

one sees the reduction of Luc21 as a sharp peak whichindicates that the product (Luc) is being adsorbed ordeposited on the working electrode. Were this a revers-

xide measurement by chemiluminescence.

a cyclic voltammetry experiment.

ero

n in

etclrdtohtrvsrwalpa

0

449REDOX POTENTIAL OF LUCIGENIN

ible process one would expect the reoxidation peak toappear at the same potential, i.e., at 20.38 V, without theusual separation of anodic and cathodic waves since nodiffusion of the electroactive species is involved in thiscase. However, at 0.1 V/s (Fig. 1A) the reoxidation peakappears at a much more positive potential (10.45 V) thanxpected due to the already discussed slow conforma-ional change of the reduced lucigenin. This being thease one would anticipate that at high scan rates and atower temperatures it would be possible to observe theeoxidation wave of the reduced lucigenin before it un-ergoes the conformational transformation. Indeed whenhe scan rate was increased to 100 V/s (Fig. 1B), anxidation wave begins to appear at 20.23 V. That stilligher scan rates were needed was indicated by the facthat the oxidation wave at 20.23 V was smaller than theeduction wave at 20.49 V and by the persistence of aery positive oxidation wave at 10.62 V which corre-ponds to the oxidation of Luc. At 1000 V/s (Fig. 1C)eversibility is more nearly achieved. Thus the reductionave at 20.60 V is nearly equalled by the oxidation wavet 20.10 V and the positive oxidation wave at 10.69 V isess prominent. Figure 1D, still at 1000 V/s, but at 0°C,rovides further improvement and Fig. 1E, at 4000 V/snd at 0°C, finally achieves reversibility.If we take the average of the E1/2 values from volt-

ammograms in Figs. 1B, 1C, 1D, and 1E, which is20.36 V, and correct it for the potential of the Ag/AgClreference electrode (10.22 V), we obtain E° 5 20.14 6.02 V vs NHE for Luc21 in aqueous medium. This

single value is assumed to apply to both univalentreduction steps. The reason for this assumption residesin the effect of a high dielectric medium such as waterin minimizing the effect of charge on reduction poten-tial. In a low dielectric organic solvent two reductionwaves were seen (8), while we noted only one. Thus, inwater, the reduction potential of one acridinium ringsystem is not influenced by the presence or absence ofa charge on the conjoined ring system. The redox po-tential of Luc21/Luc•1 couple obtained by cyclic volta-mmetry is virtually identical to the E° reported for theO2/O2

2 couple, i.e., 20.16 V vs NHE (10).

DISCUSSION

The Afanasev et al. paper (6) used the cyclic voltam-metry data of Legg and Hercules (11) which was irre-versible because, at a scan rate of only 3 V/s, the slow

IG. 1. Cyclic voltammograms of lucigenin (Luc21) in aqueous so-utions at pH 7.8, 0.05 M Tris buffer. (A) 1 mM lucigenin in 0.1 MaCl at 25°C; scan rate 0.1 V/s. (B) 1 mM lucigenin in 3 M NaCl at5°C; scan rate 100 V/s. (C) 5 mM lucigenin in 3 M NaCl, at 25°C;can rate 1000 V/s. (D) 5 mM lucigenin in 3 M NaCl at 0°C; scan rate000 V/s. (E) 5 mM lucigenin in 3 M NaCl at 0°C; scan rate 4000 V/s.lassy carbon button electrode, Ag/AgCl reference electrode (10.22vs NHE).

FlN2s1GV

2

r

dH

watsu

so

F

o;O

11

11

1

1

1

450 SPASOJEVIC, LIOCHEV, AND FRIDOVICH

conformational transition later deduced by Ahlberg etal. (8) had imposed irreversibility and this was not yetappreciated. Ahlberg et al. (8) achieved reversibility inCH2Cl2, at a scan rate of only 86 mV/s, by working at

50°C and their data yield E1/2 5 20.35 V vs Ag/Ag1

for the Luc21/Luc•1 couple, in good agreement with ourresults. Ahlberg et al. (8) were able to observe twoeduction waves in CH2Cl2 corresponding to two suc-

cessive electron transfers. It is reasonable to expectthat a solvent of low dielectric constant, such asCH2Cl2, would separate the potentials of the two re-

uction steps; whereas a high dielectric solvent such as2O would bring them together. This is anticipated

because high dielectric media minimize the effects ofelectrostatic charge.

Another problem was applying a correction of 10.45V to deduce E1/2 in water from E1/2 in aprotic media. Theeffect of dielectric constant on E1/2 depends upon

hether reduction of the compound being studied addscharge or eliminates a charge and on the nature of

he electrode. Thus in Fig. 6 of Ref. 6 the effects ofolvent polarity on E1/2 for a variety of quinones weresed to deduce the effect on Luc21. Reduction of the

quinones imparts a negative charge, whereas in thecase of Luc21 it eliminates a positive charge. In ourhands the E1/2 for Luc21 was only moderately affected(;0.1 V) by going from water to DMSO, acetonitrile, ordichloromethane (data not shown). In addition onedoes not know how much of the solvent effect is exertedon the species reduced/oxidized at the working elec-trode and how much is due to liquid–liquid junctionpotentials. We have, for the first time, been able tomeasure the E° for the Luc21/Luc•1 couple in aqueousmedia by using fast scan rate cyclic voltammetry.

Afanasev et al. (6) also presented experimental re-ults which were interpreted as indicating the absencef redox cycling by Luc21. Adding Luc21 to the xanthine

oxidase/xanthine reaction increases O22 production.

This was shown by using 2.5 mM cytochrome c andmeasuring O2

2 production in terms of the SOD-inhibit-able reduction of the cytochrome (4). Since Luc•1 candirectly reduce cytochrome c, higher concentrations ofcytochrome c cause more of the Luc•1 to be reoxidizedby the cytochrome and less by O2. Hence the sensitivityto inhibition of cytochrome c reduction by SOD de-creases. We noted that at 25 mM cytochrome c theeffect of Luc21 on SOD inhibitable cytochrome c reduc-tion was barely detectable (4). Afanasev et al. (6) used25 mM cytochrome c in their Fig. 1 and 50 mM in their

ig. 2.It has long been known that xanthine oxidase, acting

n xanthine, in air equilibrated neutral buffers, causes5 times more divalent, than univalent, reduction of2 (12). That being the case, one cannot expect to

measurably increase net electron flux through xan-

thine oxidase by adding low concentrations of an addi-tional univalent electron acceptor, such as Luc21.Moreover, the rate-limiting step in the xanthine oxi-dase reaction is oxygen independent and is probablythe dissociation of the product, i.e., urate (13, 14). Thatis what Afanasev et al. (6) demonstrate in their Fig. 3and interpret to mean the absence of redox cycling byLuc21.

The artifactual interpretation of Luc21 luminescencecontinues. Thus Li et al. (15, 16) recently reported thatLuc21 luminescence is caused by mitochondria and in-terpret the intensity of this luminescence to be a mea-sure of intramitochondrial O2

2 concentration. Whatthey may have actually demonstrated is that Luc21 canbe reduced to Luc•1 by mitochondrial enzymes. Howmuch of the O2

2 was endogenously produced and howmuch as a consequence of the autoxidation of Luc•1

remains unknown.The reduction potential of Luc21 is compatible with

its enzymic reduction by biological reductants, such asNAD(P)H, and with the ability of Luc•1 to reduce O2 toO2

2. Hence Luc21 is capable of causing the production ofO2

2; even when that radical was not being produced inthe absence of Luc21. This being the case, the use ofLuc21 luminescence as a measure of O2

2 concentrationis inadvisable.

REFERENCES

1. Faulkner, K., and Fridovich, I. (1993) Free Radicals Biol. Med.15, 447–451.

2. Liochev, S. I., and Fridovich, I. (1997) Arch. Biochem. Biophys.337, 115–120.

3. Fridovich, I. (1997) J. Biol. Chem. 272, 18515–18517.4. Liochev, S. I., and Fridovich, I. (1998) Free Radic. Biol. Med. 25,

926–928.5. Tarpey, M. M., White, C. R., Suarez, E., Richardson, G., Radi, R.,

and Freeman, B. A. (1999) Circ. Res. 84, 1203–1211.6. Afanasev, I. B., Ostrachovitch, E. A., and Korkina, L. G. (1999)

Arch. Biochem. Biophys. 366, 267–274.7. Kolthof, I. M., and Tomsicek, W. J. (1935) J. Phys. Chem. 39,

945–954.8. Alberg, E., Hammerich, O., and Parker, V. D. (1981) J. Am.

Chem. Soc. 102, 844–849.9. Rodriguez-Amaro, R., Sanchez, M., Munoz, E., Ruiz, J. J., and

Camacho, L. (1996) J. Electrochem. Soc. 143, 2132–2136.0. Wood, P. M. (1988) Biochem. J. 253, 287–289.1. Legg, K. D., and Hercules, D. M. (1969) J. Am. Chem. Soc. 91,

1902–1907.2. Fridovich, I. (1970) J. Biol. Chem. 245, 4053–4057.3. Edmondson, D., Massey, V., Palmer, G., Beacham, L. W., III, and

Elion, G. B. (1972) J. Biol. Chem. 247, 1597–1604.4. Edmondson, D., Ballou, D., Van Heuvelen, A., Palmer, G., and

Massey, V. (1973) J. Biol. Chem. 248, 6135–6144.5. Li, Y., Zhu, H., and Trush, M. A. (1999) Biochem. Biophys. Acta

1428, 1–12.6. Li, Y., Stansbury, K. H., Zhu, H., and Trush, M. A. (1999)

Biochem. Biophys. Res. Commun. 262, 80–87.