Melatonin’s unique radical scavenging properties – roles of itsfunctional substituents as revealed by a comparison with itsstructural analogs

Introduction

Numerous studies have demonstrated that melatonin is a

potent scavenger of some of the most reactive anddangerous free radicals [1–4]. Its capacity of protectingcells from oxidative damage has been widely documented

and repeatedly reviewed [4–6]. The mechanisms by whichmelatonin interacts with free radicals are partially under-stood, and two major products have been identified. Under

general considerations, hydroxyl radicals (ÆOH) can reactwith melatonin in three different ways: (1) by abstraction ofan electron, (2) by abstraction of a hydrogen atom, or (3)by addition. Although hydrogen abstraction, leading to the

formation of a neutral melatonyl radical, has been observed[7], electron abstractions is, as far as we can see, thequantitatively most important pathway of oxidation. The

resulting melatonyl cation radical subsequently combineswith an O2

Æ) [2, 4], a reaction by which the electrons are

Abstract: Melatonin’s O-methyl and N-acetyl residues are not only the basis

of its amphilicity enabling the molecule to enter all organs and all subcellular

compartments, but are also decisive for its antioxidant properties. We have

compared melatonin’s redox chemistry with that of several structural analogs:

tryptamine, N-acetyltryptamine, serotonin, N-acetylserotonin,

5-methoxytryptamine, 6-chloromelatonin and 2-iodomelatonin. Scavenging

of hydroxyl radicals (ÆOH) was measured in a scavenger competition assay

based on ABTS cation radical (ABTSÆ+) formation. The capability of

undergoing single-electron transfer reactions was studied using an ABTSÆ+

reduction assay, reflecting the more general property of scavenging organic

cation radicals. Direct scavenging of superoxide anions (O2Æ)), under non-

catalyzed conditions, was investigated in a hematoxylin autoxidation assay.

Measurements of chemiluminescence were used for studying scavenging of

O2Æ) under catalyzed conditions, either by hemin-mediated interaction or by

combination with the respective indolyl cation radicals. Light emission was

determined in the absence or presence of the ÆOH scavenger dimethylsulfoxide

and the O2Æ) scavenger Tiron. Products formed by oxidation of the respective

indoles in a moderately alkaline, hemin-catalyzed H2O2 system were analyzed

by thin-layer chromatography and fluorometry. Absence of either the

O-methyl or the N-acetyl residue causes marked diminutions in the capacities

of scavenging ÆOH and ABTSÆ+ as well as in chemiluminescence emitted

during oxidation. The importance of theN-acetyl group is insofar remarkable

as it seems, at first glance, to be isolated from the indolic moiety; interactions

between side chain and indolic moiety are therefore decisive for melatonin’s

redox properties. The 5-hydroxylated compounds are not generally more

efficient scavengers, but particularly better reducers of ABTSÆ+; in the

alkaline H2O2 system generating ÆOH and O2Æ), melatonin was much more

rapidly oxidized than the 5-hydroxylated and non-substituted analogs.

Oxidative products formed from any of the compounds studied contained

much less of substituted kynuramines as in the case of melatonin, indicating

that radical chain termination by O2Æ) is considerably more efficient with

melatonin. These findings are supported by measurements of

chemiluminescence, which largely reflects pyrrole ring cleavage as a result of

combination with superoxide anions. In this regard, only 6-chloromelatonin

equalled melatonin, whereas the efficiency of 2-iodomelatonin was much

lower, another indication for the importance of 2,3-dioxygenation.

Burkhard Poeggeler1,2, SandraThuermann1, Andreas Dose1,Markus Schoenke1, SusanneBurkhardt1 and RudigerHardeland1

1Institute of Zoology and Anthropology,

University of Goettingen, Goettingen,

Germany; 2Department of Pathology,

University of South Alabama Medical Center,

Mobile, AL, USA

Key words: antioxidants, free radicals, indolic

compounds, kynuramines, melatonin

Address reprint requests to Dr Burkhard

Poeggeler, Institute of Zoology and

Anthropology, University of Goettingen,

Berliner Str. 28, D-37073 Goettingen,

Germany.

E-mail: rhardel@gwdg.de

Received September 6, 2001;

accepted December 19, 2001.

J. Pineal Res. 2002; 33:20–30 Copyright � Blackwell Munksgaard, 2002

Journal of Pineal ResearchISSN 0742-3098

20

paired and electrical charges are neutralized, therebyterminating the radical reaction chain. The melatonylcation radical has meanwhile been detected by two groups

[8, 9]. Moreover, melatonin’s capability of easily under-going single electron donation reactions, e.g. with ABTScation radicals [10, 11], favors this possibility. Finally,formation of the product resulting from the mechanism

described, N1-acetyl-N2-formyl-5-methoxykynuramine(AFMK), could not be as easily explained on the basis ofa neutral melatonyl radical. A corresponding mechanism,

initiated by electron abstraction and followed by combina-tion with an O2

Æ), was described for photo-oxidation bylight-induced protoporphyrinyl cation radicals; again the

product is AFMK [4, 12].The addition reaction of an ÆOH can lead to an

intermediate radical which combines with a second ÆOH;an identified product is cyclic 3-hydroxymelatonin

(3-OHM) [13]; again, the radical reaction chain is termin-ated by this mechanism.The products mentioned are not the only oxidative

metabolites formed (S. Burkhardt, B. Poeggeler, R. Harde-land, unpublished data). This implies that various concur-rent reactions can take place during oxidation of indolic

compounds. In fact, other indoleamines such as tryptophanor serotonin were shown to give additional products, e.g. ab-carboline or dimers [14, 15]. Contrary to the quantita-

tively substantial rate of serotonin dimerization, a dimerfrom melatonin has been found to date only as a very smallfraction contaminating commercial preparations [16], notin the biological material. With regard to the presumed

toxicity of dimers, this difference in oxidation chemistrymight be of considerable significance.The differences clearly indicate that radical scavenging

by indolic compounds is not solely a question of thepresence of an indole moiety, but demonstrate that thereactivity of these molecules is profoundly modulated by

their functional residues. All indoleamines share a het-eroaromatic ring system of high electroreactivity and theyonly differ in carrying functional groups in their side

chains. These substituents determine to a great extent thereactivity, potency and efficiency of radical scavenging.Variations in redox behavior of the particular indolescould be of fundamental significance. This was already

indicated by the initial findings by Tan et al. [1] revealingpro-oxidative properties of serotonin. The role of themethoxy group in melatonin might therefore be seen as a

means of avoiding formation of O-centered indolylradicals. But it has still remained unclear as to whetherthe methoxy residue also contributes to melatonin’s

reactivity. The importance of the N-acetyl group becameparticularly obvious when Tan et al. [13] demonstratedthe formation of 3-OHM, a reaction requiring such aterminal residue at the side chain. Moreover, data of the

earlier paper [1] had already indicated that 5-methoxy-tryptamine differed from melatonin in its capacity forscavenging ÆOH, despite its similar affinity, a finding

difficult to interpret at that time.For these reasons, we have conducted a comparative

study on the oxidation chemistry of melatonin and its

analogs, using various assays describing different aspects ofradical scavenging and interactions with different reactive

oxygens, and looking at the metabolites formed from theindolic educts. Results will demonstrate how the functionalsubstituents convey unique properties to the molecule

melatonin, and a major conclusion will be that it is notjust general radical reactivity that makes melatonin sovaluable and distinguishes this molecule from other indole-amines of similar structure.

Materials and methods

All assays were carried out at 20�C, either in a tempera-ture-controlled photometer or, for purposes of chemilu-minescence measurements, in the temperature-controlled

detection chamber of a scintillation spectrometer. Most ofthe assays were carried out in triplicate. Luminescence wasmeasured at least three times until the timepoint ofaddition of competing inhibitors, for some indoleamines

more often (melatonin n ¼ 7; 5-methoxytryptamine andserotonin n ¼ 6); runs in the presence of competitorsvaried between 1 and 3. Representative runs were selected

for presentation.Scavenging of ÆOH was investigated by means of a

scavenger competition assay based on ABTS cation radical

(ABTSÆ+) formation initiated by a Fenton reaction andoptimized according to the principles outlined by Harde-land et al. [17]. The reaction system consisted of 0.9 mL

distilled water, adjusted with HCl to pH 5.0, 0.15 mLABTS [2,2¢-azino-bis-(3-ethylbenzthiazoline-6-sulfonicacid)] 1 mM, 0.15 mL FeSO4 0.5 mM, 0.15 mL of therespective indolic compound 10, 5 or 2.5 mM, where

applicable also 0.5 mM (pre-dissolved in ethanol at con-centrations of 100 mM, diluted with distilled water adjustedto pH 5.0), and 0.15 mL H2O2 10 mM. Controls received

the respective ethanol/water pH 5.0 mixture instead of theindolic compound solution. Measurements of ABTSÆ+ weremade at 420 nm.

ABTSÆ+ reduction was assayed according to Re et al.[18], with the following modifications: final concentrationof ABTSÆ+ 105 lM, start of reaction after 5 min ofequilibration; final concentrations of indolic compounds

25 lM or 12.5 lM [1 mM or 0.5 mM in 0.025 mL ethanol per1 mL of reaction mixture]. Controls received the respectiveamount of ethanol. Ethanol does not reduce ABTSÆ+.

Measurements were made at 420 nm.Direct, non-catalyzed scavenging of O2

Æ) was determinedby a variant of a method originally designed for measuring

superoxide dismutase (SOD) activity [19]; the assay, basedon hematoxylin autoxidation, was carried out as describedin detail by Burkhardt et al. [20]. Indolic compounds were

pre-dissolved at a concentration of 100 mM in ethanol,diluted with distilled water and used in final concentrationsof 0.33 mM.Scavenging of O2

Æ) under catalysis, e.g. by indolyl cation

radicals formed by previous electron abstraction or byhemin-mediated transfer [4, 21], was followed by measur-ing chemiluminescence in a moderately alkaline (pH 8.0),

hemin-containing H2O2 system [21], using a total volumeof 1 mL of reaction mixture. In these reactions, lumines-cence reflects the decay of emitters (‘active carbonyls’ at a

resonance-stabilized moiety); these excited intermediatesare formed by oxygen addition in the course of pyrrole

Melatonin’s functional substituents

21

ring cleavage and represent a typical luminescent reactionof dioxetan-type structures [22, 23]. Light emission wasdetermined using a temperature-controlled scintillation

spectrometer operated under settings optimized for lumin-escence [24]. Indolic compounds were pre-dissolved in avery small volume of dimethylsulfoxide (DMSO) (c. 1 mg/10 lL) and further diluted with water (final concentrationin the reaction mixture: 0.4 mM). For reasons of solubility,6-chloromelatonin was diluted directly in the buffer,whereas 2-iodomelatonin (4 mM) had been dissolved in

0.1 mL DMSO. The reaction mixture contained, therefore10% DMSO and a corresponding experiment had to becarried out using melatonin in the presence of DMSO at

the same concentration. Studies on the requirements ofÆOH and O2

Æ) in the formation of excited intermediateswere carried out by adding either 100 lL of DMSO as anÆOH scavenger or 100 lL of 22 mM of Tiron (¼ 4,5-

dihydroxy-1,3-benzenedisulfonic acid), representing a par-ticularly suitable and fairly specific O2

Æ) scavenger avoidingSOD’s disadvantages of pro-oxidant behavior in the

presence of H2O2 [25]; Tiron had already been appliedwith success in comparable studies on other indolic,quinaldic [21, 26] and phenolic [27, 28] radical scavengers.

The same reaction mixture as in the measurements ofchemiluminescence was used for studies on products formedduring oxidation of the respective indoles. This procedure of

oxidation is more preservative for products than, e.g. H2O2/UV light systems in which primary products are easilydestroyed (unpublished data of this laboratory). Mixtures ofproducts and residual educts were applied to the thin-layer

plates both directly and after previous extraction with the10-fold volume of ethylacetate. Substances were separatedby thin-layer chromatography using the following solvent

mixtures: 90% ethylacetate/10% methanol; 70% ethylace-tate/30% methanol; 90% methanol/10% water. For purpo-ses of further characterization by fluorometry solvent

mixtures were selected according to the relative lipophilicityof educts and products; melatonin, N-acetyltryptamine andtheir products: 90% ethylacetate/10% methanol; 5-meth-

oxytryptamine, serotonin, N-acetylserotonin, tryptamineand their products: 90% methanol/10% water. Educts andfluorescent products were detected on the chromatogramsby fluorescence excited at 254 and 366 nm. Educts and

products, in particular, substituted kynuramines formedduring oxidation were eluted from chromatograms andsubjected to fluorometry, as described earlier for AFMK

[12, 22], using glycylglycine buffer pH 8.0, 0.1 M, as solvent.Fluorescence spectra were measured in an Aminco-Bowmanspectrofluorometer (model J4-8065-E, equipped with ellip-

soidal condensing system, American Instrument Co., SilverSpring, MD, USA), at ratio mode.

Results

All indolic compounds tested were capable of interactingwith ÆOH and ABTSÆ+, although with different effi-

ciency. However, scavenging of O2Æ) in the absence of

catalysts, as it is detected in the hematoxylin autoxida-tion system, was poor or negligible (not shown in

detail). At a final concentration of 0.33 mM, tryptamine,serotonin and N-acetylserotonin were practically inert,

whereas 5-methoxytyptamine was slightly pro-oxidative;N-acetyltryptamine scavenged O2

Æ) at a negligibly low,but demonstrable rate, while melatonin showed a small

(23%) reduction of hematoxylin autoxidation, an effectwhich was, however, inapparent at lower melatoninconcentrations.Radical scavenging activity by melatonin and several

other indoleamines was easily demonstrable and consider-able in the ABTS competition assay (Fig. 1). In this testsystem, the concentration of the remarkably stable ABTSÆ+

is measured, which are generated via electron abstractionby ÆOH. The scavenger assayed competes with ABTS forÆOH which is predominantly formed in a burst-like fashion

by a Fenton reaction. Therefore, the first minutes duringwhich ABTSÆ+ concentration is rising, a diminution of theincrease reflects mainly scavenging of ÆOH by the substancetested; this conclusion is, however, only justified if the

scavenger is not a much better electron donor than ÆOHscavenger. In the following section of the ABTSÆ+ absorb-ance curve, a plateau is approached; declines in the presence

of the scavenger indicate the capability of undergoingsingle-electron transfer reactions with ABTSÆ+. In thisassay, melatonin exhibited both strong ÆOH scavenging, as

reflected by inhibitions of the rise in ABTSÆ+, and ABTSÆ+

reduction, indicated by the subsequent decline (Fig. 1A).The comparison with other indoleamines revealed a crucial

importance of melatonin’s functional residues. Surprisingly,the absence of an N-acetyl residue, as in 5-methoxytrypta-mine, strongly diminished the capacity of radical scaveng-ing (Fig. 1B), despite the seemingly isolated position of this

substituent. The lack of the 5-methoxy group, as inN-acetyltryptamine, caused a substantial loss of reactivity,especially with regard to ÆOH scavenging, and only a minor

capability of reducing ABTSÆ+ was maintained (Fig. 1C).Tryptamine, lacking either of these residues, did not proveto be an even poorer scavenger of ÆOH, but was, in this

regard, comparable with melatonin, whereas ABTSÆ+

reduction was less expressed under these test conditions(Fig. 1F).

The two 5-hydroxylated indoleamines, N-acetylseroto-nin and serotonin, caused much stronger suppressions ofABTSÆ+ (Fig. 1D,E). However, these effects, measured atsame concentrations as with the other indoleamines, were

too strong for distinguishing properly between ÆOHscavenging and ABTSÆ+ reduction. Using other testsystems, we shall demonstrate below that ÆOH scavenging

was not necessarily higher than with melatonin, butrather undeterminable because of a very pronouncedelectron donation capacity. Moreover, the increases of

ABTSÆ+ in the presence of the hydroxylated compoundsshould be noted, indicating pro-oxidant properties ofthese substances.We also investigated electron donation directly in an

ABTSÆ+ reduction assay (Table 1). In this system ÆOH isnot involved, and the assay can be carried out at approxi-mately physiological pH. Again melatonin proved to be a

good reducer of the cation radicals; 5-methoxytryptaminewas similar in this regard, tryptamine too, whereasN-acetyltryptamine was poorer. The previous assumption

of extremely strong electron donation by 5-hydroxylatedindoleamines was confirmed in this assay, serotonin being

Poeggeler et al.

22

superior to N-acetylserotonin, both of them more effectivethan the other indoleamines tested.Differences in oxidation chemistry became even more

apparent in the moderately alkaline (pH 8), hemin-cata-lyzed H2O2 system. This was used for generating productsto be analyzed by thin-layer chromatography as well as formeasurements of chemiluminescence. The decisive property

of this system is the generation of both ÆOH and O2Æ). The

former is produced in a Fenton reaction which is, however,not burst-like as with uncomplexed iron and proceeds for

many hours; O2Æ) is formed in high abundance from their

equilibrium with H2O2, driven by the alkaline pH [22]. O2Æ)

is required in this system for chemiluminescence. They can

either interact with indolyl cation radicals formed byinteraction with ÆOH, or be oxidized by hemin in conjunc-tion with oxygen transfer to the pyrrole ring [2]. In thissystem, oxidation of melatonin caused chemiluminescence

at a high rate (Fig. 2A). After a short initial peak, lightemission continued at a more or less constant level for morethan an hour, until the reaction ceased after about 2.5 hr.

N-Acetyltryptamine (Fig. 2B) showed a similar time course,but luminescence was by more than an order of magnitudelower. The difference to 5-methoxytryptamine (Fig. 2C)was remarkable: luminescence was not only even lower than

with N-acetyltryptamine, but the reaction was extended formore than 15 hr; the reaction started with a low rate, butincreased towards a maximum, which was attained, at the

concentration used in the run shown, after more than 9 hr.Tryptamine (Fig. 2D) exhibited a similar kinetics, but themaximum appeared after about 7 hr, under these condi-

tions. N-Acetylserotonin (Fig. 2E) gave a higher yield ofphoton emission and reached, only briefly, a level similar tothat measured during oxidation of melatonin. However, thereaction started with an intermediate rate, rose towards a

Fig. 1. Radical scavenging by melatoninand five structural analogs in the ABTScompetition assay. (A) melatonin; (B)5-methoxytryptamine; (C) N-acetyltryp-tamine; (D) N-acetylserotonin; (E)serotonin; (F) tryptamine. Dotted lines:without indoleamines, in the presence ofthe respective amounts of ethanol used incorresponding experiments; solid lines:with indoleamines. Symbols (final con-centrations of indoleamines, or respectivecontrol runs): squares 1 mM; lozenges0.5 mM; triangles: 0.25 mM; circles:0.1 mM; lines: 0.05 mM. Ordinates:absorbance of ABTS cation radicals (ODat 420 nm); abscissae: time (min).

Melatonin’s functional substituents

23

maximum after about 2.5 hr (i.e. when melatonin’s reactionwas almost terminated) and continued for many morehours. Chemiluminescence from the oxidation of serotonin

was again considerably lower (Fig. 2F). Again, the maxi-mum appeared very late, after more than 9 hr and wasextended for many other hours.In correspondence to the considerable differences in

turnover rates in this system, product formation was highlydivergent. After 4 hr of reaction, melatonin was quantita-tively oxidized, whereas this was not the case with the

majority of the other indoleamines. Only in the case ofmelatonin, high amounts of a kynuric product had beenformed, namely, AFMK (Table 2). The identity of this

compound was confirmed by fluorometry of the isolatedsubstance. An additional product was obviously non-kynuric, according to its fluorescence characteristics. Onlytraces were found of another known product from mela-

tonin oxidation, 3-OHM (cf. [13]), which was weaklydetectable, but not quantifiable. All the other indoleaminesincluded in Table 2 gave a much lower yield of fluorescent

products and residuals of educt were more or less discern-ible, differing in their amount from substance to substance(details not shown). In the case of serotonin, two products

were formed exhibiting fluorescence when excited with366 nm. Whether one of these is a kynuramine, the other ab-carboline [14] remains to be studied in detail. Neverthe-less, in no case amounts of products were formed equallingthat of AFMK from melatonin oxidation.We extended our studies in the hemin-catalyzed H2O2

system to 6-substituted analogs. The pattern and level of

chemiluminescence from 6-chloromelatonin oxidation re-sembled very much those of melatonin (Fig. 3A); lightemission was even slightly higher than in the mother

compound. However, 6-hydroxymelatonin showed muchlower rates of light emission and the conversion continuedfor many more hours, having a maximum around 8 hr

(Fig. 3B).Because of its low solubility in water, another halogen-

ated indoleamine, 2-iodomelatonin, had to be investigatedin the presence of higher amounts of DMSO, a substance

having properties of an ÆOH scavenger. For this reason,chemiluminenscent emitters are only deriving from amechanism independent of ÆOH. A comparison with mela-

tonin investigated in the presence of DMSO at sameconcentration revealed that light emission from 2-iodo-melatonin was by orders of magnitude lower (Fig. 4).

Percent ABTSÆ+ reduction

Compound Concentration (mM) After 1 min After 10 min

Melatonin 0.025 72.0 94.70.0125 52.3 86.1

5-Methoxytryptamine 0.025 79.2 98.00.0125 58.2 87.0

N-Acetyltryptamine 0.025 48.6 86.40.0125 33.6 75.0

Tryptamine 0.025 71.2 93.70.0125 52.2 86.4

N-Acetylserotonin 0.025 96.2 97.60.0125 65.7 86.7

Serotonin 0.025 98.6 98.70.0125 84.7 96.4

Table 1. Reduction of ABTSÆ+ by mela-tonin and five structural analogs

Fig. 2. Chemiluminescence emitted dur-ing oxidation of melatonin and fivestructural analogs in the hemin-catalyzedH2O2 system. (A) melatonin; (B) N-acet-yltryptamine; (C) 5-methoxytryptamine;(D) tryptamine; (E) N-acetylserotonin; (F)serotonin. Ordinates: luminescence (mil-lions of cpm); abscissae: time (hr). Finalconcentrations of indoleamines 0.4 mM.

Poeggeler et al.

24

The ÆOH scavenger DMSO was also added at, or shortlybefore, the luminescence maximum, in order to determinethe proportions between ÆOH-dependent and -independent

oxidation mechanisms [2, 21]. In each case studied here,elimination of ÆOH caused only a partial suppression ofluminescence, usually to about a half (Figs. 5 and 6). At the

Table 2. Main products formed from melatonin and its structural analogs in the hemin-catalyzed H2O2 system: fluorescence andchromatography characteristics

Fluorescence:maxima of

Rf values in thin-layer chromatography, separation by

Band intensity onchromatogram at

fluorescence excitationwith (arbitrary units)

Educt/product no.

excitation andemissiona

Ethylacetate/methanol 9:1

Ethylacetate/methanol 7:3

Methanol/water 9:1 366 nm 254 nm

Melatonin/1 (AFMK) 344 nm/478 nm 0.435 ± 0.041 0.759 – +++++ +++Melatonin/2 290 nm/352 nm 0.768 ± 0.012 – – ) +++N-Acetyltryptamine/1 316 nm/419 nm 0.762 ± 0.011 0.795 ± 0.077 0.775 ± 0.003 + )N-Acetyltryptamine/2 ND** 0.621 ± 0.011 0.715 ± 0.025 – + +++5-MeO-tryptamine/1 332 nm/405 nm 0.760 ± 0.008 0.825 ± 0.033 0.811 ± 0.023 ) ++5-MeO-tryptamine/2 342 nm/458 nm 0.231 ± 0.016 – 0.231 ± 0.010 ++ +Tryptamine/1 312 nm/400 nm – – 0.805 ± 0.044 ) +++Tryptamine/2 ND – – 0.277 ± 0.011 ++ ++N-Acetylserotonin/1 327 nm/430 nm – 0.760 ± 0.024 0.779 ± 0.020 + ++N-Acetylserotonin/2 ND – 0.341 ± 0.018 0.197 ± 0.005 ++ )Serotonin/1 337 nm/446 nm – 0.897 ± 0.006 0.753 ± 0.028 +++ ++Serotonin/2 338 nm/487 nm – 0.801 ± 0.010 0.438 ± 0.019 +++ )

aIn glycylglycine buffer pH 8; ND: not determined; statistics: S.D.

Fig. 3. Chemiluminescence emitted during oxidation of 6-substi-tuted melatonin analogs in the hemin-catalyzed H2O2 system. (A)6-chloromelatonin; (B) 6-hydroxymelatonin. Other details as inFig. 2.

Fig. 4. A 2-iodo substituent strongly decreases formation of aluminescent emitter: chemiluminescence measured during oxida-tion of 2-iodomelatonin and melatonin in the hemin-catalyzedH2O2 system, in the presence of 10.2% DMSO (DMSO was presentat start of experiments). (A) Iodomelatonin; (B) melatonin. Otherdetails as in Fig. 2.

Melatonin’s functional substituents

25

expense of the ÆOH-dependent reaction, the DMSO-insen-

sitive reaction was extended to a correspondingly longerduration, in all cases studied. Contrary to the limited effectof DMSO, addition of the O2

Æ) scavenger Tiron caused arapid and strong, often almost complete inhibition of light

emission (Table 3).

Discussion

Our findings clearly demonstrate that it is the combinationof the O-methyl and the N-acetyl residues which is crucial

for melatonin’s peculiarities in oxidation chemistry. Inparticular, this comprises of the reactivity towards ÆOH, theconversion rates demonstrated in the hemin-catalyzed

H2O2 system, and the formation of kynuric products, allof which are lower in N-acetyltryptamine and 5-methoxy-tryptamine. These data demonstrate that it is not simply theaddition of two effects resulting from these substituents, but

rather a kind of balance. This becomes obvious by the factthat tryptamine is a better ÆOH scavenger in the ABTScompetition assay than N-acetyltryptamine and 5-meth-

oxytryptamine.Therefore, these two functional groups are not only

important physically but also chemically. The physical

property of amphiphilicity enables the molecule to crossany membrane, but also to remain at a sufficiently highproportion in the aqueous phase [29, 30]. Amphiphilicityis biologically highly important and represents a funda-

mental difference to the majority of physiologicallyrelevant radical scavengers, which are frequently hydro-philic such as ascorbate, NAD(P)H and glutathione, or

lipophilic such as a-tocopherol and carotinoids. Not leastfor this reason, melatonin can be so effective in cellprotection. We now state that the combination of func-

tional residues has consequences for both the physical andchemical properties.Nevertheless, one can attribute individual properties to

the respective functional groups. The 5-methoxy residueshould be important for avoiding the formation of O-cen-tered radical intermediates, as occurring in serotonin [31,

32], and may also diminish C-centered radicals, two typesof intermediates which seem to be implicated inpro-oxidative behavior and formation of toxic productssuch as dimers [15]. The N-acetyl group, physiologically

also important with regard to preventing degradation bymonoamine oxidase, clearly participates in the oxidation

Fig. 5. Effects of ÆOH scavenging byDMSO on chemiluminescence emittedduring oxidation of melatonin and fivestructural analogs in the hemin-catalyzedH2O2 system. (A) melatonin; (B) N-acetyltryptamine; (C) 5-methoxytryptamine;(D) tryptamine; (E) N-acetylserotonin; (F)serotonin. DMSO of 100 lL were added(arrows) to 1 mL of reaction mixturebefore the respective maximum of lumin-escence. Other details as in Fig. 2.

Fig. 6. Effects of ÆOH scavenging by DMSO on chemiluminescenceemitted during oxidation of 6-chloromelatonin and 6-hydroxy-melatonin in the hemin-catalyzed H2O2 system. (A) 6-chloro-melatonin; (B) 6-hydroxymelatonin. DMSO of 100 lL were added(arrows) to 1 mL of reaction mixture before the respective maxi-mum of luminescence. Other details as in Fig. 2.

Poeggeler et al.

26

reactions, despite its distance from the indolic moiety. Thishad already been shown for the cyclization reaction in theformation of 3-OHM [13], but is now also demonstrated for

the pyrrole ring cleavage reaction leading to the formationof chemiluminescent emitters. When comparing the N-acet-ylated compounds with their N-unsubstituted analogs, it isobvious that the N-acetyl group enhances luminescence and

accelerates the oxidative turnover. This difference is, in fact,spectacular in the case of melatonin and 5-methoxytrypta-mine, in which N-acetylation causes an enhancement of

luminescence by a factor of about 23-fold and a shift of themaximum from more than 9 hr to less than 1 hr. Corres-ponding effects of the N-acetyl residue were found when

comparing N-acetylserotonin with serotonin and, to asmaller extent, N-acetyltryptamine with tryptamine.With regard to these data, chemiluminescence has proved

to be a valuable tool for studying some aspects of oxidation

chemistry. Light emission during radical-mediated oxida-tion of melatonin was first shown by Uemura and Kadota[33], but the xanthine/xanthine oxidase system was much

less efficient, with regard to both rates of emitter formationand duration of measurements, than the hemin-catalyzedH2O2 system we used for determining luminescence from

melatonin oxidation [22, 23, 34].The effects demonstrated at the level of chemilumines-

cence do not imply that rates of light emission are

proportional to radical reactivity, because (a) the propor-tion between light-emitting and concurrent ‘black’ reactionscan differ from substance to substance, and (b) as emitterstrength can considerably vary even among structural

analogs. Nevertheless, our data clearly show that theN-acetyl group does participate in the oxidation reactions,a conclusion also confirmed by the highly divergent

turnover rates seen in the luminescence measurements andin the analyses of products. Therefore, bending of the sidechain towards the indolic nucleus has to be involved in, at

least, some relevant types of indoleamine oxidation. It isnot just the indolic moiety which determines the oxidationchemistry, but we can demonstrate a profound, indispens-

able contribution of the N-acetyl group.The comparison with the halogenated melatonin deriv-

atives led to the following conclusions. Substitution atposition 2 with a bulky iodine atom strongly inhibits the

formation of an efficient luminescent emitter. Formation of

a dioxetane-type oxygen adduct is hindered both stericallyand by blockade of a bond required for oxygen addition.Substitution in position 6 did not substantially modify

emitter strength nor the rate of conversion to the emitter.Luminescence from the oxidation of 6-chloromelatonin waseven slightly higher than that generated from the unsubsti-tuted mother compound.

Substitution in position 6 with a hydroxyl group had aremarkably different result. Although the oxygen of thisresidue should also have an electronegative effect, emitter

formation proceeded at a much lower rate. Luminescencefrom 6-hydroxymelatonin was by a factor of about 22smaller than that from melatonin. This was not primarily

because of lower emitter strength, because the reaction with6-hydroxymelatonin continued for many more hours with amaximum around 8 hr, instead of less than 1 hr withmelatonin.

This influence of a hydroxyl group finds their parallels inthe cases of serotonin and N-acetylserotonin. With either ofthese substances, turnover in the hemin-catalyzed H2O2system is slower, as demonstrated by both luminescenceand product formation. Because in any of these substances,about half of the turnover unanimously depends on ÆOH, as

shown by the inhibition by DMSO, one cannot concludethat the 5-hydroxylated indoleamines possess a highergeneral radical reactivity. On the one hand, the slower turn-

over of the hydroxylated compounds is well in accordancewith the earlier conclusions on ÆOH scavenging in an H2O2/UV light system [1], in which, however, the superposition ofantioxidant and pro-oxidant effects has to be considered

(pro-oxidant reactions of 5-hydroxylated indoleamineswere also indicated by our ABTS competition data; cf.Fig. 1). On the other hand, our findings throw a light on the

meaning of radical reactivity determinations. Radicalreactivity is obviously strongly dependent on the radicalspecies representing the reaction partner. Interactions with

galvinoxyl radicals showing higher reactivities of serotoninand N-acetylserotonin than melatonin [35] do not allow todeduce the reactivity towards ÆOH nor efficiency of scav-

enging. This becomes obvious also in our data on ABTSÆ+

reduction. Serotonin and N-acetylserotonin proved to beconsiderably better ABTSÆ+ reductants than melatonin,and this effect was so strong that it obscured the ABTS

competition assays carried out with these substances. In

Table 3. The O2Æ) scavenger Tiron strongly

inhibits chemiluminescence emitted duringoxidation of various indoleamines

% Inhibition by Tiron

IndoleamineTime of addition

of Tiron10 min afteraddition

30 min afteraddition

Melatonin 1 hr 98.5 99.8N-Acetyltryptamine 40 min 73.8 92.35-Methoxytryptamine 7 hr 96.8 98.0Tryptamine 5 hr 93.2 95.1N-Acetylserotonin 2 hr 84.9 93.7Serotonin 8 hr 61.7 76.7a

6-Hydroxymelatonin 6 hr 72.9 82.9b

6-Chloromelatonin 30 min 93.2 99.42-Iodomelatoninc 1.5 hr 78.3 84.0

aAfter 2 hr: 90.4%; bafter 2 hr: 92.2%; cin the presence of 10.2% DMSO; Tiron was addedat times selected according to maxima of light emission.

Melatonin’s functional substituents

27

comparison to the much slower conversion in the hemin-catalyzed H2O2 system, the conclusion has to be thatserotonin and N-acetylserotonin are not superior to mela-

tonin with regard to ÆOH scavenging, but rather theircapability of undergoing single-electron transfer reactions ismuch more pronounced. From these findings, a particularlyimportant conclusion can be drawn, namely, that efficiency

of scavenging is not simply ruled by the redox potential.The long-lived, resonance-stabilized ABTSÆ+ is certainlymuch less reactive than ÆOH. Nevertheless, a substance can

be a better ABTSÆ+ scavenger than ÆOH scavenger. Thisobservation was also made with several other compounds,namely, 5-methoxytryptophol, which was a relatively poorÆOH scavenger in the ABTS competition assay, but a betterreductant for ABTSÆ+ [36], and quinaldic acid, whichpractically did not scavenge ÆOH but did reduce ABTSÆ+

[37].

All indoleamines tested for chemiluminescence in thehemin-catalyzed H2O2 system showed the same behavior inthe presence of Tiron and DMSO. The O2

Æ) scavenger Tiron

suppressed the luminescent reaction almost totally andmore or less instantaneously. This is plausible because ofthe requirement of O2

Æ) for dioxetane formation. However,

DMSO inhibited light emission only by about a half,regardless of conversion rates of the individual indoleam-ines. In the case of melatonin, we present data both for the

presence of DMSO from the beginning (Fig. 4) and foraddition after 1 hr (Fig. 5). The amount of DMSO added(100 lL to 1 mL of reaction mixture) was not too small forsufficiently eliminating ÆOH, because half the concentration

was sufficient for almost totally suppressing chemilumines-cence from kynurenic acid oxidation [21] and for preventingformation of the pink intermediate from caffeic acid

oxidation [27]. The only partial inhibition by DMSOseems to be a characteristic of many indolic compoundsand was also observed with indole-3-propionic acid [21].

The still considerable rates of luminescence measured in thepresence of DMSO indicates two concurrent pathways ofoxidative pyrrole ring cleavage, in the course of which

emitter intermediates are formed. This is not only valid formelatonin [2, 4] and indole-3-propionic acid [21], but alsofor the other indoleamines investigated here, except2-iodomelatonin, in which pyrrole ring cleavage is hin-

dered. The DMSO-sensitive pathway is initiated by ÆOHand the resulting indolyl cation radical combines with O2

Æ),the DMSO-insensitive mechanism represents hemin-cata-

lyzed oxygen transfer from O2Æ), whose unpaired electron

reduces hemin.With indoleamines as educts, either of these pathways

leads to kynuramines. Our analyses of products, evenwithout identifying any of them chemically, allow, how-ever, several relevant conclusions. First, luminescence ratesand duration of conversion in the hemin-catalyzed H2O2system find their expression, grosso modo, also in theamount of products with typical kynuramine fluorescence,although no exact proportionality can be expected, because

emitter strengths as well as fluorescence excitation maximashould not be identical, and because products may befurther converted. Secondly, none of the indoleamines

included in Table 1 led to the formation of kynuramineconcentrations as high as melatonin; the amounts of

AFMK produced exceeded by far any kynuric compoundderived from the other indoleamines. From our point ofview, this is insofar important, as the formation of

kynuramines represents termination of the radical reactionchain and therefore under physiological conditions, avoid-ance of propagating oxidative damage. This does notexclude that a kynuric product should be unable to

undergo other radical reactions itself, and, in fact ÆOHscavenging by AFMK has been recently demonstrated [38,39]. Nevertheless, the enormous difference between AFMK

formation and that of kynuramines deriving from otherindolic compounds sheds a light on melatonin’s outstand-ing capability of terminating radical chains, and of

avoiding pro-oxidative reactions as they are typical forthe hydroxylated compounds and, perhaps, for otherindoleamines generating non-kynuric products, too. Inthis context, we would also like to emphasize that the

hemin-catalyzed H2O2 system has characteristics whichseem to be closer to the physiological situation thanconventional Fenton reagents, H2O2/UV light systems or

other procedures attempting to generate predominantlyÆOH: this is the higher abundance of OÆ)

2 in comparisonwith ÆOH, which are generated as well. Therefore ÆOH can

initiate reactions at a substantial rate, but chain termin-ation by O2

Æ) is favored relative to other possibilities ofreaction taking place solely under the influence of ÆOH.

Therefore, AFMK formation in this system may bephysiologically meaningful.Our results have shown that the combination of

functional groups is, to a certain extent unexpectedly,

decisive for melatonin’s oxidation chemistry. Both groupsare not only important with regard to amphilicity, butalso for reactivity, capacity of chain termination, avoid-

ance of pro-oxidant reactions, and formation of a protect-ive oxidation product. Making use of both the physicaland chemical properties of melatonin, nature has selected

a molecule combining, in a remarkable way, severaladvantages presumed to be important for purposes ofprotection.

References

1. TAN D-X, CHEN L-D, POEGGELER B et al. Melatonin: a potent,

endogenous hydroxyl radical scavenger. Endocr J 1993; 1:57–

60.

2. HARDELAND R, REITER RJ, POEGGELER B et al. The signifi-

cance of the metabolism of the neurohormone melatonin:

antioxidative protection and formation of bioactive substances.

Neurosci Biobehav Rev 1993; 17:347–357.

3. REITER RJ, MELCHIORRI D, SEWERYNEK E et al. A review of

the evidence supporting melatonin’s role as an antioxidant.

J Pineal Res 1995; 18:1–11.

4. HARDELAND R, FUHRBERG B. Ubiquitous melatonin – presence

and effects in unicells, plants and animals. Trends Comp Bio-

chem Physiol 1996; 2:25–45.

5. HARDELAND R. Melatonin: multiple functions in signaling and

protection. In: Skin Cancer, UV Radiation. Altmeyer P,

Hoffmann K, Stucker M, eds., Springer, Berlin, 1997; pp. 186–

198.

6. REITER RJ. Oxidative damage in the central nervous system:

protection by melatonin. Prog Neurobiol 1998; 56:359–384.

Poeggeler et al.

28

7. STASICA P, ULANSKI P, ROSIAK JM. Melatonin as a hydroxyl

radical scavenger. J Pineal Res 1998; 25:65–66.

8. ZHANG H, SQUADRITO GL, PRYOR GA. The reaction of

melatonin with peroxynitrite: formation of melatonin radical

cation and absence of stable nitrated products. Biochem

Biophys Res Commun 1998; 251:83–87.

9. POEGGELER B, CHYAN Y-J, PAPPOLLA MA. Direct scavenging

of carbonate radicals by melatonin. Identification of a new

neuroprotective mechanism against Ab-mediated toxicity.

J Neuropathol Exp Neurol 1999; 58:511.

10. POEGGELER B, REITER RJ, HARDELAND R et al. Melatonin, a

mediator of electron transfer and repair reactions, acts syner-

gistically with the chain-breaking antioxidants ascorbate,

trolox and glutathione. Neuroendocrinol Lett 1995; 17:87–92.

11. POEGGELER B, REITER RJ, HARDELAND R et al. Melatonin and

structurally-related, endogenous indoles act as potent electron

donors and radical scavengers in vitro. Redox Rep 1996; 2:179–

184.

12. HARDELAND R, BALZER I, POEGGELER B et al. On the primary

functions of melatonin in evolution: mediation of photo-

periodic signals in a unicell, photooxidation and scavenging of

free radicals. J Pineal Res 1995; 18:104–111.

13. TAN D-X, MANCHESTER LC, REITER RJ et al. A novel mela-

tonin metabolite, cyclic 3-hydroxymelatonin: a biomarker of

in vivo hydroxyl radical generation. Biochem Biophys Res

Commun 1998; 253:614–620.

14. UEMURA T, KANASHIRO M, YAMANO T et al. Isolation, struc-

ture, and properties of the b-carboline formed from

5-hydroxytryptamine by the superoxide anion-generating sys-

tem. J Neurochem 1998; 51:710–717.

15. HUETHER G, REIMER A, SCHMIDT F et al. Oxidation of the

indole nucleus of 5-hydroxytryptamine and formation of

dimers in the presence of peroxidase and H2O2. J Neural

Transm Supplement 1990; 32:249–257.

16. NAYLOR S, JOHNSON KL, WILLIAMSON BL et al. Structural

characterization of contaminants in commercial preparations

of melatonin by on-line HPLC-electrospray ionization-tandem

mass spectrometry. Adv Exp Med Biol 1999; 467:769–777.

17. HARDELAND R, WENDLER J, HOLST S et al. On the use of 2,2¢-azino-bis-(3-ethylbenzthiazoline-6-sulfonic acid) [ABTS] in

studies on free radicals: limitations of test systems for inhibi-

tion of cation radical formation and for cation radical reduc-

tion. In: Studies on Antioxidants and Their Metabolites.

Hardeland R, ed., Cuvillier, Goettingen, 1999; pp. 35–37.

18. RE R, PELLEGRINI N, PROTEGGENTE A et al. Antioxidant

activity applying an improved ABTS cation radical decolor-

ization assay. Free Radic Biol Med 1999; 26:1231–1237.

19. ANTOLIN I, OBST B, BURKHARDT S et al. Antioxidative pro-

tection in a high-melatonin organism: the dinoflagellate

Gonyaulax polyedra is rescued from lethal oxidative stress by

strongly elevated, but physiologically possible concentrations

of melatonin. J Pineal Res 1997; 23:182–190.

20. BURKHARDT S, COTO-MONTES A, HARDELAND R. The catalase

inhibitor 3-amino-1,2,4-triazole has side effects as a scavenger

of superoxide anions. In: Studies on Antioxidants and Their

Metabolites. Hardeland R, ed., Cuvillier, Goettingen, 1999;

pp. 127–128.

21. HARDELAND R, ZSIZSIK BK, POEGGELER B et al. Indole-3-py-

ruvic and -propionic acids, kynurenic acid and related meta-

bolites as luminophores and free-radical scavengers. Adv Exp

Biol Med 1999; 467:389–395.

22. FUHRBERG B, HARDELAND R. Oxidation of melatonin by free

radicals: catalysis by hemin. In: Cell Biological Problems in

Chronobiology. Hardeland R, ed., University of Goettingen,

Goettingen, 1994; pp. 75–81.

23. HARDELAND R, FUHRBERG B, ZSIZSIK BK et al. Chemilumi-

nescence as a tool for monitoring the oxidation of indolic and

quinaldic radical scavengers. Eur J Clin Chem Clin Biochem

1997; 35:A108.

24. HARDELAND R. Effects of catecholamines on bioluminescence

in Gonyaulax polyedra (Dinoflagellata). Comp Biochem Phys-

iol 1980; 66C:53–58.

25. OOSTHUIZEN MM, GREYLING D. Antioxidants suitable for use

with chemiluminescence to identify oxyradical species. Redox

Rep 1999; 4:277–290.

26. HARDELAND R, ZSIZSIK BK. Kynurenic acid as a free

radical scavenger: measurements of educt and product

fluorescence and of light emission from an excited inter-

mediate state. In: Biological Rhythms and Antioxidative

Protection. Hardeland R, ed., Cuvillier, Goettingen, 1997;

pp. 153–160.

27. WENDLER J, HOLST S, HARDELAND R et al. Phenolic and

indolic radical scavengers as luminophores. In: Biolumines-

cence and Chemiluminescence 2000. Case JF, Herring PJ,

Robinson BH et al., eds., World Scientific Publishers, Singa-

pore, 2001; pp. 255–258.

28. WENDLER J, POEGGELER B, HARDELAND R. Chemilumines-

cence emitted during oxidation of resveratrol and other

phenolic free radical scavengers. In: Chemiluminescence at

the Turn of the Millenium – An Indispensible Tool in

Modern Chemistry, Biochemistry and Medicine. Albrecht S,

Zimmermann T, Brandl H, eds., Schweda, Dresden, 2001;

pp. 216–221.

29. COSTA EJX, LOPES RH, LAMY-FREUND MT. Permeability

of pure lipid bilayers to melatonin. J Pineal Res 1994; 16:123–

126.

30. SHIDA CS, CASTRUCCI AL, LAMY-FREUND MT. High mela-

tonin solubility in aqueous medium. J Pineal Res 1994; 16:198–

201.

31. PEREZ-REYES E, MASON RP. Characterization of the structure

and reactions of free radicals from serotonin and related

indoles. J Biol Chem 1981; 256:2427–2432.

32. POEGGELER B, REITER RJ, TAN D-X et al. Melatonin, hydroxyl

radical-mediated oxidative damage, and aging: a hypothesis.

J Pineal Res 1993; 14:151–168.

33. UEMURA T, KADOTA K. Serotonin- and melatonin-dependent

light emission induced by xanthine oxidase. In: Progress in

Tryptophan and Serotonin Research. Schlossberger HG,

Kochen W, Linzen B et al., eds., Walter de Gruyter, Berlin,

1984; pp. 673–676.

34. HARDELAND R, ZSIZSIK BK, FUHRBERG B. Chemiluminescence

during oxidation of indolic and quinaldic free-radical scaven-

gers. J Biolum Chemilum 1998; 13:204.

35. GODA K, HAMANE Y, KISHIMOTO R et al. Radical scavenging

properties of tryptophan metabolites. Estimation of their

radical reactivity. Adv Exp Med Biol 1999; 467:397–402.

36. HALDAR C, HARDELAND R, BOEKER J. Oxidation of 5-meth-

oxytryptophol by free radicals: scavenger competition and

chemiluminescence studies. In: Studies on Antioxidants and

Their Metabolites. Hardeland R, ed., Cuvillier, Goettingen,

1999; pp. 99–109.

37. ZSIZSIK BK, HARDELAND R. Comparative studies on kynure-

nic, xanthurenic and quinaldic acids as scavengers of hydroxyl

and ABTS cation radicals. In: Studies on Antioxidants and

their Metabolites. Hardeland R, ed., Cuvillier, Goettingen,

1999; pp. 82–91.

Melatonin’s functional substituents

29

38. BURKHARDT S, REITER RJ, TAN D-X et al. DNA oxidatively

damaged by chromium (III) and H2O2 is protected by mela-

tonin, N1-acetyl-N2-formyl-5-methoxykynuramine, resveratrol

and uric acid. Int J Biochem Cell Biol 2001; 33:775–783.

39. TAN D-X, MANCHESTER LC, BURKHARDT S et al. N1-Acetyl-

N2-formyl-5-methoxykynuramine, a biogenic amine and

melatonin metabolite, functions as a potent antioxidant.

FASEB J 2001; 15:2294–2296.

Poeggeler et al.

30