Address reprint requests to Leopoldo Ceraulo, via Archirafi, 32, 90123 Palermo Italy. E-mail: firstname.lastname@example.org
Abstract: Oxidation of melatonin by Fenton reagents as well as with hypochlorous acid or oxoferryl hemoglobin has been investigated. Analysis of products by low resolution/mass spectra (MS), high resolution/MS, 1H-nuclear magnetic resonance (NMR), 13C-NMR, correlated spectroscopy (COSY) and heterocorrelated spectroscopy (HETCOR) 2D NMR reveals the formation of a single mono-oxygenated product under all conditions and unequivocally assigns the N-[2-(5-methoxy-2-oxo-2,3-dihydro-1H-indol-3-yl)-ethyl]-acetamide structure, which had not been previously considered.
Melatonin (N-acetyl-5-methoxytryptamine) (1) (Fig. 1) is the main secretory product of the pineal gland of mammals and its structure was determined by Lerner et al. . It has been ascertained that, in mammals, melatonin plays an important role as a regulator of sleep and seasonal reproductive cycles . Further, the fact that melatonin (1) has been recovered also in bacteria , protista , and plants , strongly suggests its involvement in other biological activities in virtually all the living organisms. At present, indications for therapeutic applications for melatonin include sleep disorders [6–8], circadian rhythm disorders , insomnia in blind people , insomnia in elderly patients , aging , Alzheimer disease , and as an adjuvant in cancer therapy [14–17].
Recent investigations indicate that melatonin is a potent radical scavenger and may behave as a potential antioxidant . For instance, it has been determined that melatonin is five times more effective than glutathione as a scavenger of hydroxyl radicals , and 500 times more powerful than dimethyl sulfoxide in protecting DNA from radiation-induced chromosomal damage . Other studies indicate that melatonin acts a scavenger of lipoperoxyl radicals [21–22] and nitric oxide [23–24]. In addition to its present breadth of applications, these activities are associated with an extremely low toxicity ; this suggests application of melatonin as a valuable drug for prevention and cure of several diseases related to free radical-induced damage. In this context, determining the bio-transformation of melatonin, i.e., identifying compound(s) formed in biological models and/or in vivo, could be an useful tool to achieve information about its metabolic fate and mechanistic aspects related to its scavenging activity in biological systems . While the pathway that generates 6-hydroxy melatonin (2) (Fig. 1)  is well known, there have been a number of reports concerning formation, as well as the structure and occurrence in vivo of other non-enzymatic mono-oxygenated metabolites. Vakkuri et al.  isolated a metabolite from the urine of rats treated with melatonin and found it to be cyclic 2-hydroxy-melatonin structure (1-(8a-hydroxy-5-methoxy-3,3a,8,8a-tetrahydro-2H-pyrrolo[2,3-b]indol-1-yl)-ethanone) (3) (Fig. 1). Tan et al. obtained a mono-oxygenate metabolite by reaction of melatonin (1) with Fenton reagents, and the same molecule was successively isolated from both rat and human urine . Tan et al.  identified this metabolite as cyclic 3-hydroxy-melatonin (1 acetyl-1,2,3,3a,8,8a-hexahydro-3a hydroxy-5-methoxypyrrole [2.3-b] indole) (4) (Fig. 1). Vakkuri et al.  attributed the cyclic structure 3, on the basis of the non-equivalence of the α-methylene protons of the 3-alkyl side chain in the 1H-NMR spectrum, and Tan et al.  supported their suggestion for structure 4 by 2D-COSY-NMR experiments. Finally, Dellegar et al.  assigned the structure of 2-hydroxymelatonin (5) (Fig. 1) to a mono-oxygenated product of melatonin (1) after reaction with hypochlorous acid. More recently, while studying the oxidation of melatonin (1) with oxoferryl hemoglobin , we found evidence of a mono-oxygenated metabolite, the mass spectrum of which was quite comparable with that reported by Tan et al. .
Because of our recent interest in studies concerning melatonin [31–34], the aim of this work was to re-investigate the mono-oxygenate metabolite(s) of melatonin obtained after oxidation with Fenton reagents, or with HClO or oxoferryl hemoglobin, possibly to provide unequivocal evidence of the structure(s). This was performed by low resolution (LR)/mass spectra (MS), high resolution (HR)/MS, 1H-NMR, 13C-NMR, COSY and HETCOR 2D NMR.
Melatonin (99.5%, Aldrich, St Louis, MO, USA), hydrogen peroxide (30% v/v, Fluka, Buchs, Switzerland), water [Fluka, high performance liquid chromatography (HPLC) grade], sodium hypoclorite solution (≥ 4% Aldrich), Na2HPO4·12H2O (x99.0% Fluka) NaH2PO4·H2O (x99.0% Fluka), NaCl (99% Aldrich), Na2SO3 (98% Fluka), FeSO4·7H2O (99% Aldrich), ethyl acetate (99.5% A.C.S. reagent, Aldrich), ethyl alcohol (96% Fluka), CDCl3 (99.8 Aldrich) and D2O (99.9% Aldrich) were used as received; 6-hydroxymelatonin, bovine heart methemoglobin (met-HB) and catalase were from Sigma (St Louis, MO, USA), silica gel for chromatographic purification was 0.040–0.063 mm (Merck, Frankfurter, Germany), preparative chromatography was performed on 20 cm × 20 cm × 2 mm Kieselgel 60 F254 (Merck).
Oxidation of melatonin and product preparation and isolation
The oxidation of melatonin by Fenton reagents was carried out following the method described by Tan et al.  including modifications in order to improve the yield: 300 mg of melatonin (1) and 30 mL of H2O2 aqueous solution (30% v/v) were mixed in a flask under pH control (pH = 8) obtained using a saturated solution of NaHCO3 and the solution allowed to react for 1.5 hr. At this time a catalytic amount (only few crystals) of FeSO4· 7 H2O and a second 30 mL aliquot of H2O2 were added. The reaction was monitored and sampled every 2 hr for a total of 48 hr. However, the best yields in oxidized products were obtained with a total reaction time of 10 hr. The reaction mixture was extracted three times with 20 mL aliquots of CH2Cl2, desiccated with MgSO4, filtered and finally concentrated under vacuum in a rotary evaporator. The residual pale brown oil (70 mg) was purified by preparative TLC using ethyl acetate/methanol 85:15 as eluent phase. The final product (15 mg, brown wax) chromatographically pure was obtained.
The oxidation of melatonin by HClO acid was carried following the method described by Dellegar et al. , including modifications in order to improve the yield. A 500 mL beaker containing 200 mL of ethanol (95%) and 200 mg of melatonin was placed in a thermo stated bath (10∘C) and vigorously stirred with a magnetic stirrer. In another 500 mL beaker 2.8 mL of NaClO solution were added to 200 mL of neutral buffer solution prepared using Na2HPO4·2H2O (0.716 mg) and NaH2PO4·H2O (0.276 mg) pH variation was not observed. The buffered NaClO solution was added very slowly to the ethanol solution of melatonin maintained under constant stirring conditions. A rapid temperature increment was registered at this point and the solution reached the temperature of 26∘C. The solution was allowed to react for 12 s and Na2SO3 was added to quench the reaction, and the solution was allowed to react for 20 min under constant stirring to eliminate all residual chlorine (I); during this process the temperature slowly decreased to 12∘C. The ethanol was removed under reduced pressure in a rotary evaporator. The residual aqueous solution was extracted two times with two 100 mL aliquots of ethyl acetate, the extracts were mixed and washed with 50 mL of brine solution and the organic layer was desiccated using anhydrous Na2SO4 and filtered. The organic solution was concentrated in a rotary evaporator under low-pressure conditions. The microcrystalline residual (140 mg) had a brown color and consisted of a mixture of products. After purification by flash chromatography, using silica gel as stationary phase and a mixture of ethyl acetate/methanol 85:15 as mobile phase, 75 mg (37.5% yield) of the mono-oxygenate derivative of melatonin under investigation was obtained, as a brown wax.
Reaction of melatonin with oxoferryl hemoglobin was carried out as previously reported .
Oxoferryl hemoglobin was prepared under spectral control using a Beckman (Fullerton, CA, USA) DU 640 UV/vis spectrometer, equipped with a temperature control. Bovine met-HB in phosphate-buffered saline was oxidized by a 10-fold excess H2O2 for 1 min, at 37∘C, followed by addition of 250 IU catalase to remove excess H2O2 . Melatonin was incubated for 5 min with a 25-fold molar excess of oxoferryl-Hb. Under these conditions total consumption of melatonin was observed after organic extraction of the sample using ethyl acetate and HPLC analysis. Extraction of the sample and gas-chromatographic (GC)/MS analysis revealed two major product metabolites, 5-methoxykynuramine (AMFK) and N-[2-(5-methoxy-2-oxo-2,3-dihydro-1H-indol-3-yl)-ethyl]-acetamide.
Melatonin oxidation product analysis
Low resolution and HR electron ionization (EI) MS were recorded by the Autospec Ultima o-TOF (Micromass, Altrincham, UK) mass spectrometer connected with a GC system HP 6890 series (Hewlett Packard, Palo Alto, CA, USA) of the ‘Rete di Spettrometria di Massa-CNR’ operating into the Dipartimento di Chimica e Tecnologie Farmaceutiche-Palermo. LR-EI-MS (resolution power 1500) and HR-EI-MS (resolution power 8000, 10% resolution valley definition) were performed under the following experimental conditions: electron beam energy 70 eV; source temperature 210∘C; source pressure 10−7 Torr; trap current 250 μA; emission current 2.3 μA; accelerating voltage 8.0 kV. Accurate mass measurements (±10 ppm) were determined by HR-EI-MS using perfluorokerosene (PFK) as internal standard. Gas-chromatographic conditions were: injector temperature 290∘C; column ATTM-5 (Alltech, Deerfield, Fl, USA), film thickness 0.25 μm, length 30 m, ID 0.25 mm, carrier gas (helium) flow 1.0 mL/min, isotherm at 120∘C (5 min), ramp 120–240∘C (20∘C/min), isotherm 240∘C (9 min).
1H- and 13C-NMR spectra were recorded on a Bruker (Karlsruhe, Germany) AC 250 spectrometer in deuterochloroform solutions. 1H-NMR chemical shifts are given in ppm from tetramethylsilane (TMS) used as internal reference and 13C-NMR chemical shifts (in ppm) are given from CDCl3 (77.00) and were taken from fully decoupled spectra. 1H-1H COSY, 1H-13C HETCOR and DEPT experiments were performed with the usual pulse-sequence and data processing was obtained with standard software.
The IR spectra were recorded in the wavenumber range 400–4000/cm with a resolving power of 0.5/cm on a Perkin Elmer Spectrum RX I FT-IR spectrophotometer (Perkin Elmer, Wellesley, MA, USA) from CH2Cl2 sample solution. The signals were acquired four times and the mean signals were taken as the best value of the FT-IR spectra. Before every measurement the blank spectrum was also recorded, and automatically subtracted from the sample spectrum by the instrument software using the signal background ratio.
Semi-empirical molecular orbital calculations were performed on a Windows/PC based system using the ‘Hyperchem 2.0’ software (Hypercube, Gainsville, FL, USA).
Results and discussion
The mono-oxygenated products obtained after treatment of melatonin with Fenton reagent, or with hypochlorous acid, or oxoferryl hemoglobin, were analyzed by GC-MS. The identical retention time (RT = 16.5 min) and MS (Fig. 2), unequivocally provide evidence that the same product is generated by all oxidation reactions. Our spectra are quite similar to the EI-MS reported by Tan et al.  and Vakkuri et al. . A comparison with an authentic sample of 6-hydroxy melatonin (2) ruled out the possibility of such a compound. The latter compound had a very different retention time (RT = 18.5 min) and MS spectrum. Indeed the fragmentation peaks are similar to those of the present product, but the relative amounts are very different (spectrum not shown).
The HR-EI-MS confirms that the molecular weight 16 Da more than melatonin is because of an additional oxygen atom in its structure (calculated for C13H16N2O3 248.1161, found 248.1170), as well as that, according to the fragmentation pathways of melatonin (1) and several related compounds , the main peaks are because of acetamide elimination and N-methylenacetamide radical ejection, which for our compound leads to the 189 Th (calculated for C11H11NO2 189.0790, found 189.0802) and 176 Th (calculated for C10H10NO2 176.0711, found 176.0706) ions, respectively (Fig. 2).
The 1H-NMR spectrum (CDCl3) (Table 1) is essentially identical to that reported by both Vakkuri et al.  and Tan et al. . However, the amidic NH proton (6.59 ppm), which appears as a broad triplet (Jvic = 5.50 Hz) unequivocally excludes the involvement of the amide nitrogen in the formation of a new cycle and consequently rules out both structures 3 and 4. Further, the fact that only two protons (6.59 and 8.90 ppm) exchange with D2O does not agree with the 2-hydroxy melatonin structure (5) proposed by Dellegar et al. . These authors  assigned the structure 5 mainly on the ground of an intense peak at 204 Th in the EI-MS, corresponding to the loss of the HNCOH group. However, such a fragment was in very small amount in the mass spectrum of such an oxidation product (Fig. 2). A careful GC-MS investigation lead to an identification of an almost overlapping peak because of another compound with MW 292 (the structure of which has not been investigated) the EI-MS of which is characterized by the base peak at 204 Th. Possibly under the applied GC conditions, Dellegar et al.  obtained a co-elution of the two compounds, and consequently the MS of the mixture.
Table 1. The 1H-nuclear magnetic resonance data of the oxidized metabolite of melatonin, chemical shifts in ppm relative to the internal standard tetramethylsilane
s = singlet, d = doublet, t = triplet, dd = double doublet.
β -CH2 plus an other proton
CH3CONH-; exchangeable with D2O Jvic = 5.50 Hz
H-6; Jo = 8.39 Hz; Jm = 1.87 Hz
H-7; Jo = 8.39 Hz
H-4; Jm = 1.87 Hz
NH; exchangeable with D2O
All these findings strongly suggest the N-[2-(5-methoxy-2-oxo-2,3-dihydro-1H-indole-3-yl)-ethyl]-acetamide structure (6) (Fig. 3), which was not previously established as melatonin metabolite. The 13C-NMR data, summarized in Fig. 3 and mainly the presence of the two signals at 171.26 and 180.92 ppm, because of the amidic and lactamic carbon atoms, respectively, are also in agreement with structure 6.
The additional carbonyl group with respect to the starting product is also confirmed by the presence in the FT IR spectrum of the bands at 1718 and 1675/cm because of the carbonyl lactamic and the carbonyl amidic stretch. Possibly, the non-equivalence of the α-CH2 (because of the fact that they are diastereotopic) led Vakkuri et al.  to a misinterpretation of the 1H-NMR data, which resulted in the suggestion of the cyclic structure 3.
We have also performed the 2D COSY NMR spectrum (Fig. 4), which shows the expected cross peaks according to structure 6. The spectrum is quite identical to that reported by Tan et al. , as a definite prove of structure 4. In fact, they attributed the cross peak between the proton at 6.3 ppm (6.59 ppm in our hands) and the multiplet near 3.5 ppm to the coupling of the indole NH and the H-8a of structure 4. We think this is better explained by the coupling of the β-CH2 protons, with a chemical shift very near to that of H-3 (H-8a of 4 for Tan et al. ), with the acetamide N-H (indole NH of 4 for Tan et al. ).
In light of the contrasting structures reported in the literature [29, 30], we searched for other evidence to support structure 6 and carried out 1H-13C HETCOR NMR experiments (Fig. 5). The observed cross peaks agree with the proposed structure 6. In particular, it is relevant to note that in correspondence of the at 3.30–3.50 ppm of the 1H-NMR spectrum two cross peaks with the carbon atoms at 37.79 and 45.50 ppm, respectively, are observed, that provides more evidence for structure 6. The lack of the cross peak between α-CH2 protons and the α-CH2 carbon was possibly because of its very small intensity, which prevents us from putting forward any other structural hypothesis.
Finally, semi-empirical molecular calculations on optimized geometries for the formally inter-converting structures 3, 5 and 6, lead the ΔHf,∘ values of -342, -408 and -430 kJ/mol, respectively, so providing evidence that compound 6 is the thermodynamically favored.
In spite of the small ΔHf,∘ difference for tautomers 5 and 6, the NMR spectra do not provide evidence of equilibrium occurring in solution being predominantly compound 6. This is not surprising as it is known that the oxindole tautomeric form largely predominates over the 2-hydroxyindole one . On the contrary, the EI-MS fragmentation pattern of the mono-oxygenated product is very similar to that of either melatonin (1) or 6-hydroxy melatonin (2), which suggests the presence of compound 5, as the 2-hydroxy melatonin radical cation (5+ḃ), in the gas-phase. On this basis, semi-empirical molecular orbital calculations on optimized geometries of radical cations 5+ḃ and 6+ḃ were performed. The analysis showed that, in the gas phase, the 2-hydroxy melatonin radical cation (5+ḃ) (ΔHf,∘ 273 kJ) is more stable than the oxindole radical cation (6+ḃ) (ΔHf,∘ 295 kJ) which may account for the mass spectrum pattern.
For relevant biological properties associated with an extremely low toxicity, melatonin (1) can act as a valuable drug for a number of diseases [38–40]. In this light, determining the structures of melatonin metabolites may be important in order to distinguish between either metabolic pathways or antioxidant mechanisms.
The present study on the oxidation products of melatonin by Fenton reagents, HClO and oxoferrylhemoglobin which were analyzed by LR/MS, HR/MS, 1H-NMR,13C-NMR, COSY and HETCOR 2D NMR experiments led to the conclusion that only one mono-oxygenated product is obtained, the structure of which is N-[2-(5-methoxy-2-oxo-2,3-dihydro-1H-indole-3-yl)-ethyl]-acetamide structure (6), not previously considered. This should reconcile previous observations on the structure of the mono-oxygenate metabolite of melatonin [28–30].
The fact that the metabolite 6 is also obtained by oxidation of melatonin by HClO , a reaction that does not involve free radical species, strongly contrasts with the suggested use of such a metabolite as a biomarker of in vivo hydroxyl radical generation . Of course this does not rule out co-participation of hydroxyl radical to the formation of 6.
The authors thank University of Palermo (Fondi Ricerca Scientifica ex 60%) for financial support.