Kynuramines, metabolites of melatonin and other indoles: the resurrection of an almost forgotten class of biogenic amines


Address reprint requests to Rüdiger Hardeland, Johann Friedrich Blumenbach Institute of Zoology and Anthropology, University of Göttingen, Berliner Str. 28, D-37073 Göttingen, Germany.


Abstract:  Kynuramines represent their own class of biogenic amines. They are formed either by decarboxylation of kynurenines or pyrrole ring cleavage of indoleamines. N2-formylated compounds formed in this last reaction can be deformylated either enzymatically by arylamine formamidases or hemoperoxidases, or photochemically. The earlier literature mainly focussed on cardiovascular effects of kynuramine, 5-hydroxykynuramine and their N1,N1-dimethylated analogs, including indirect effects via release of catecholamines or acetylcholine and interference with serotonin receptors. After the discovery of N1-acetyl-N2-formyl-5-methoxykynuramine (AFMK) and N1-acetyl-5-methoxykynuramine (AMK) as major brain metabolites of melatonin, these compounds became of particular interest. They were shown to be produced enzymatically, pseudoenzymatically, by various free radical-mediated and via photochemical processes. In recent years, AFMK and AMK were shown to scavenge reactive oxygen and nitrogen species, thereby forming several newly discovered 3-indolinone, cinnolinone and quinazoline compounds, and to protect tissues from damage by reactive intermediates in various models. AMK is of special interest due to its properties as a potent cyclooxygenase inhibitor, NO scavenger forming a stable nitrosation product, inhibitor and/or downregulator of neuronal and inducible NO synthases, and a mitochondrial metabolism modulator. AMK easily interacts with aromates, forms adducts with tyrosyl and tryptophanyl residues, and may modify proteins.


Biogenic amines are decarboxylation products of amino acids and structurally homologous metabolites thereof. In life sciences, the average investigator would immediately recall catecholamines and tyramine, perhaps their monohydroxylated analogs such as octopamine, and certainly indoleamines and histamine. Both aliphatic and aromatic amino acids can be transformed into amines of biological interest, but this review will focus on a specific aspect of aromatic amino acid metabolism, the formation and properties of kynuramines (also referred to as kynurenamines), another class of biogenic amines, which is frequently overlooked or disregarded.

The simplest compound of this family of molecules is the unsubstituted kynuramine (Fig. 1), which may be formally described as the decarboxylation product of a tryptophan metabolite, the amino acid l-kynurenine. This reaction is, in principle, possible and has been demonstrated in mammalian liver [1]. A corresponding decarboxylation, presumably catalyzed by a different enzyme, was described for 3-hydroxykynurenine, and the product 3-hydroxykynuramine was assumed to be an intermediate in the formation of the urinary metabolite 2-amino-3-hydroxyacetophenone O-sulfate [1]. Little attention has been paid to these metabolites and their biological relevance remains uncertain. The unsubstituted kynuramine, which exerts pharmacological cardiovascular effects, was later preferentially used as a monoamine oxidase (MAO) A and B substrate that leads to the formation of 4-hydroxyquinoline [2, 3]. In a dinoflagellate (Lingulodinium polyedrum, syn. Gonyaulax polydra), i.e. an organism phylogenetically distant to vertebrates, a strong stimulation of bioluminescence was observed [4, 5], but this action was later attributed to MAO inhibition [6].

Figure 1.

 Formation of kynuramines from indoleamines: an overview.

An alternate type of kynuramine formation consists in pyrrole ring cleavage of indoleamines (Fig. 1). This reaction is possible with various, perhaps all 2-unsubstituted indoleamines because it is not exclusively dependent on enzymes and their varying substrate affinities, but can also take place on the basis of pseudoenzymatic, photocatalytic or free radical reactions (see next sections). Pyrrole ring cleavage leads to the formation of N2-formylated kynuramines (Fig. 1), which may be deformylated by enzymatic or photoreactions.

For many decades, kynuramines were sparingly investigated for two reasons. (i) Most of these compounds were not commercially available. This situation has changed recently but only for a few metabolites which are now more frequently studied. (ii) The body of knowledge on the physiological effects attributable to kynuramines remained remarkably small for a considerable period of time. In part, this paucity of data is explained by the attempts of some investigators to only test the respective kynuramines for certain effects that are typical of the indoleamine parent compounds. Because of this unsatisfactory situation, kynuramines were frequently regarded as somehow exotic products, which might represent nothing more than metabolic waste, being devoid of physiologically relevant effects and not worth extensive investigative effort.

A major step forward in the perception of kynuramines as potentially important molecules was a discovery in a pioneering work by the group of Osamu Hayaishi, who showed that two 5-methoxylated kynuramines, N1-acetyl-N2-formyl-5-methoxykynuramine and N1-acetyl-5-methoxykynuramine, currently also identified by the acronyms AFMK and AMK, respectively, represent major brain metabolites of melatonin [7]. Up to that moment, melatonin was usually believed to be almost exclusively metabolized to 6-hydroxymelatonin and its excretion product, 6-sulfatoxymelatonin, and studies on melatonin metabolites were mostly confined to blood plasma, urine, and liver. At this time, the relevance and the wide distribution of tissue melatonin were not yet known, so it required more than three additional decades before the relative importance of AFMK and AMK was revealed.

The new discovery prompted investigators in the following years to experimentally test the new compounds. These studies led to a few remarkable, but also to many disappointing results. The main reason for the limited success was the adherence to their comparison with indoleamines. As will be discussed later, the more exciting results were obtained in studies that differed from repetitions of experiments usually conducted with indoleamines. As a consequence of the uncertainties about the relevance of kynuramines, interest in these metabolites again waned.

Another stimulus for studying kynuramines resulted from the finding that these products can be formed nonenzymatically, e.g. by photocatalytic and free-radical reactions of melatonin [8–10]. The further discovery of melatonin’s potent radical scavenging capacity [11–13] augmented interest in these compounds considerably. In recent years, the number of pertinent publications has risen steadily with new effects, sometimes exerted at very low concentrations; as a result one can literally speak of a resurrection of this previously almost forgotten class of biogenic amines.

The kynuric pathway: multiple reactions and the question of relevance

In earlier investigations, it had been assumed that pyrrole ring cleavage could only be enzymatically catalyzed, and the first enzyme shown to be capable of converting indoleamines to N2-formylated kynuramines was indoleamine 2,3-dioxygenase (IDO) [14–16]. This was first demonstrated for tryptamine and serotonin and later also for melatonin. Nevertheless, the enzyme’s name, which was formally justified, should not be misunderstood as its main substrate is l-tryptophan with tryptophan degradation by IDO in macrophages and microglia being a key response to interferon-γ [17–19]. However, a peculiarity of IDO became important in the context of pyrrole ring cleavage under the influence of free radicals. Rather than molecular oxygen, which is used as a co-substrate by the classic, hepatic tryptophan-degrading enzyme, tryptophan 2,3-dioxygenase, IDO utilizes a free radical, the superoxide anion, O2 [16, 20, 21]. This finding gave rise to the initial studies on melatonin oxidation in photocatalytic [8, 10] and hemin-catalyzed superoxide-generating systems [9, 22]. Later it turned out that, in the absence of catalysts, the direct affinity the O2 has for melatonin and other indoleamines is rather moderate; this free radical species is, however, important in terminating radical chain reactions initiated by oxygen species of higher reactivity [23–25].

The initial finding by Hirata et al. [7] that AFMK and AMK are major brain metabolites of melatonin faced a specific problem, namely, that IDO activities in the noncompromised brain, without microglia activation, were rather low, contrary to much higher activities found, e.g. in lungs or the gastrointestinal tract [26]. If it was not an IDO activation caused by the intervention of melatonin, additional reactions of melatonin oxidation to AFMK should exist. This was, in fact, observed to occur and included enzymatic, pseudoenzymatic, nonradical photocatalytic, and numerous free radical mechanisms. In recent years, the remarkably large number of reactions leading to the formation of AFMK have been summarized [27–30]. In brief, the following categories of agents were shown to generate this kynuramine; enzymatic: myeloperoxidase + O2 [31–35] (for involvement of O2 instead of hydrogen peroxide see ref. [34]); hemoperoxidase (horseradish) compound III + O2 [31, 36] (AFMK formation by both myeloperoxidase and horseradish peroxidase can be enhanced by chlorpromazine [37, 38]); pseudoenzymatic: oxoferryl hemoglobin [39, 40]; cytochrome c + hydrogen peroxide (H2O2) [41]; hemin + H2O2 or O2 [9, 22–24]; nonradical photocatalytic: singlet oxygen [O2(1Δg)] [23, 42–44]; O2 + UV [45–47] (details of direct photochemical melatonin destruction by UV light [48]); combinations of radicals (including photocatalysis): protoporphyrinyl IX cation radicals + O2•− [23, 42, 43, 49, 50]; substituted anthranilyl radicals + O2 [27, 51, 52]; NAD radical (NAD) + O2 [53]; hydroxyl radicals (OH) + O2 [23, 24, 54]; carbonate radicals (CO3) + O2 [25, 27, 52, 55–57].

Ozonolysis of melatonin also leads to AFMK [23, 27, 28], but may involve OH. AFMK was also shown to be formed from cyclic 3-hydroxymelatonin [58], another oxidation product of melatonin, and, in the course of cytochrome c-catalyzed oxidation, from 2-hydroxymelatonin via 2,3-dihydroxymelatonin [41]. Moreover, photocatalytic AFMK formation occurs at remarkable rates under the influence of extracts from the dinoflagellate Lingulodinium polyedrum [10, 43, 50] and the pheophycean Pterygophora californica [50, 59]. In most of the reaction systems summarized, AFMK was the most abundant (end) product, whereas other products prevailed when melatonin was photocatalytically oxidized using chlorophyll a [60] or the pesticide metabolite, 2-hydroxyquinoxaline [61, 62].

The remarkable fact that AFMK is formed in so many entirely different systems has to be considered in three aspects. First, melatonin has an obvious molecular preference for kynuramine formation, as revealed by comparisons with other indoleamines [24, 63]. Even under conditions during which virtually no kynuramine was formed from serotonin or N-acetylserotonin, such as in the 2-hydroxyquinoxaline photocatalytic system, a substantial fraction of AFMK was obtained from melatonin [61, 62]. Second, AFMK is obviously more stable than many other oxidative metabolites or its secondary product, AMK. This finding may be explained by the preference for two-electron transfer reactions [47], which do not favor interactions with free radicals [64], except for the highly reactive OH [47, 65]. Third, AFMK is a typical termination product of radical chain reactions [23–25, 42, 43, 45, 52, 54, 64]. This is especially apparent in systems not only generating electron/hydrogen abstracting radicals, but also at considerable rates O2, as would be typically found under physiological conditions [24, 25, 64]. In chemical or toxicological systems designed to investigate interactions preferably with a single free radical species, in particular OH, hydroxylated melatonin metabolites prevail [66–71]. This includes formation of cyclic 3-hydroxymelatonin [67, 68] and transformation of 2-hydroxymelatonin to the more stable 2-oxindole tautomer (2-indolinone) [70–73].

The multiplicity of enzymatic and nonenzymatic catalysts capable of forming AFMK implies different modes of reactions. The addition of two oxygen atoms, as indicated in Fig. 1, can take place via multiple means. This can be the addition of O2 species, such as combination of O2 transfer from O2 with electron transfer to the catalyst, as in the case of IDO or hemin [16, 20–24]. It may also take place in an interaction of O2 with photoexcited melatonin [45–47], and it certainly occurs with singlet oxygen [23, 42–44]. Finally, O2 can combine after interactions of the indole with an electron-abstracting free radical [23–25, 42, 43, 45, 52, 54, 64]. However, multiple hydroxylations are additional routes, which obviously occur in the peroxidase and peroxidase-like reactions [31–41] and in the conversion of cyclic 3-hydroxymelatonin to AFMK [58]. This possibility may be also deduced from the appearance of AFMK as a side product in an otherwise monohydroxylating and dealkylating system of cytochrome P450 enzyme preparations [74].

AFMK formation by multiple hydroxylations, however, should not be misinterpreted as the only mode of kynuramine synthesis. For mechanistic reasons, this is excluded in the enzymatic action of IDO [14–21]. Moreover, chemiluminescence observed in various reactions of melatonin and other indolic compounds with oxidants indicates the involvement of a dioxetane intermediate, which decays to the kynuric product by transiently forming a so-called active carbonyl representing the luminescent emitter [22, 24, 25, 55, 56, 75–80]. This dioxetane structure is well known from numerous luciferins, and the earlier finding by Uemura and Kadota [81] of light emission during melatonin oxidation in an H2O2/O2-generating xanthine/xanthine oxidase system gave rise to the idea that a process corresponding to the formation of bioluminescent emitters might occur. This prompted the experiments with the hemin-catalyzed systems already mentioned.

Formation of other kynuramines by reactions different from those by IDO or aromatic amino acid decarboxylase have been studied only to a limited extent. Product analyses and chemiluminescence data indicate that the following indoleamines can be converted to kynuramines in hemin-catalyzed oxidation systems; however, this is always accompanied by large amounts of nonkynuric products: 5-methoxytryptamine (poorly), tryptamine (poorly), N-acetyltryptamine, serotonin (poorly), N-acetylserotonin, 6-hydroxymelatonin (poorly) [24, 63, 80]. The 5-hydroxylated indoleamines preferentially form dimers or oligomers in other oxidation systems [61, 62, 82–84], or, in the case of serotonin in a xanthine/xanthine oxidase system, a β-carboline [85]. The endogenous hallucinogen N,N-dimethyltryptamine was shown to be converted by red blood cells to N1,N1-dimethylkynuramine [86], a pathway that may reflect pseudoenzymatic pyrrole ring cleavage by hemoglobin and subsequent deformylation by the hemoperoxidase activity of catalase (see next section for corresponding AMK formation). This metabolic pathway was interpreted as a detoxification route of the hallucinogen [86]. Collectively, the comparative investigations show that both the 5-methoxy group and an N-substituent favor nonenzymatic kynuramine formation. However, specific pyrrole ring cleavage of nonmethoxylated, non-N-substituted indoleamines such as serotonin and tryptamine is possible by IDO [14–16, 20].

The multitude of AFMK-forming processes may be indicative of a certain quantitative relevance, and may let the kynuric pathway in the brain [7] appear more plausible; but it is per se not yet proof of substantial quantities or physiological significance. Without administration of exogenous melatonin, serum levels of AFMK were undetectably low [87, 88]. However, after melatonin injection, the metabolite was present in the blood [87] and in the third ventricle of the brain [88]. However, it would be a misinterpretation to therefore conclude a fundamental irrelevance. It is important to recall the high quantities of melatonin in some tissues and the necessarily different organ-specific contributions of concurrent pathways to melatonin metabolism [89–91]. Therefore, it seems important to note that AFMK was found in the retinas of untreated rats and that the metabolite underwent a rhythm with a nocturmal maximum [88]. More studies on AFMK in tissues are urgently needed.

HaCaT keratinocytes in culture were shown to contain relatively high amounts of AFMK, which dose-dependently increased under UV B radiation [92]. However, the physiological relevance of this finding remains uncertain, as long as comparably high quantities of melatonin are not demonstrated in the skin in vivo [71, 91]. Nevertheless, these findings are in accordance with the assumptions of a photoprotective role of melatonin and of AFMK as an important photoproduct [10, 45, 50, 91, 93].

This may be also relevant in photoautotrophs, perhaps even more so than in mammals. AFMK formation was demonstrated in extracts of dinoflagellates and kelps under irradiation by UV and visible light [50, 93]. In the water hyacinth, Eichhornia crassipes (Pontederiaceae), an organism naturally exposed to high intensities of both visible and UV light, AFMK was physiologically present in high quantities [94]. AFMK also exhibited a diurnal rhythm with a maximum at the end of photophase [94], i.e. a time at which the damage of photosystems has accumulated over the day and H2O2 formation is usually highest in photoautotrophs [93].

Finally, an observation should be mentioned that was made in various small aquatic organisms, which are phylogenetically as distant as the dinoflagellate Lingulodinium polyedrum, the chlorophycean Chlorogonium elongatum, the ciliates Paramecium caudatum and P. bursaria, the latter one containing Chlorella symbionts, and the rotifer Philodina acuticornis [95]. After incubation with exogenous melatonin, the only oxidation product detected was AFMK, and this was observed in the light and, in organisms tolerating this, also in darkness. A remarkable difference became apparent when melatonin was incubated, under the same illumination in the respective culture media, in the absence of organisms. In this case, other products appeared in addition to AFMK, including cyclic 3-hydroxymelatonin, AMK, and a brownish-yellow pigment reminiscent of oligomers detected in other studies (cf. refs [61, 62]). Formation of AFMK from exogenous melatonin was also measured in various stages of the malaria parasite, Plasmodium chabaudi [96]. In organisms not possessing a liver, a hepatopancreas or even excretory organs, melatonin conversion to 6-sulfatoxymelatonin should be functionally meaningless. A release of other metabolites such as 5-methoxyindole 3-acetic acid [97] or AFMK may be favorable in small aquatic species, whereas, in terrestrial plants, only the formation of degradable compounds devoid of auxin activity should be an advantage. Under these conditions and with regard to the chemically favored formation by pyrrole ring cleavage, AFMK may represent a phylogenetically old melatonin metabolite [98].

In conclusion, AFMK may be widely distributed, and when exclusively examining serum or urinary levels in vertebrates may be misleading. First of all, it seems necessary to focus the search for AFMK formation in vertebrates to the nonhepatic tissues, because of the sometimes high amounts of tissue melatonin and its divergent metabolism outside the liver. Tissue melatonin has often been neglected in the metabolic balance of this indoleamine. Although a certain fraction is released from the gut without metabolic conversion, high quantities have to enter pathways different from 6-hydroxylation [89, 91, 99–101]. Another argument in favor of the quantitative importance of AFMK may be deduced from its formation by peroxidases, in particular, myeloperoxidase. An estimation on this basis generated the conclusion that up to 30% of melatonin may enter the kynuric pathway [35]. A further indication for both quantitative and qualitative, immunological significance of AFMK may be seen in the observation that elevated levels of the kynuramine were detected in the cerebrospinal fluid of patients with viral meningitis [102]. In samples containing more than 50 nm AFMK, protein concentrations and levels of IL-8 and IL-1β in the CSF were much below those from persons with AFMK contents between 10 and 50 nm [102]. Whether this reflects an upregulation of microglial IDO and/or elevated oxidative stress because of the inflammation remains to be clarified. Nevertheless, these findings support the original observation by Hayaishi’s group [7] that the methoxylated kynuramines are major brain metabolites of melatonin.

Deformylation of the N2-formylkynuramines – formation of AMK, a metabolite of particular interest

In earlier studies, deformylation of N2-formylkynuramines was believed to be exclusively catalyzed by arylamine formamidases (=arylformamidases = aryl-formylamine amidohydrolases), a group of enzyme isoforms differing in substrate affinity and being characterized by a usually low substrate specificity [103, 104]. The first-discovered forms became known as kynurenine formamidases and were, in vertebrates, thoroughly investigated in liver, brain, and gastrointestinal system. In the brain, arylamine formamidase was also shown to hydrolyze AFMK to AMK and formic acid [105]. The same reaction may take place in other organs. Its relevance should, however, depend on the AFMK concentrations attained, which may be low in liver because of preferential 6-hydroxylation of melatonin in this organ.

In principle, many other N2-formylkynuramines may be likewise deformylated by one or more arylamine formamidase isoforms, based on low substrate specificity. Some of the isoforms even accept N-formylanthranilic acid as a substrate [103], indicating that the long aliphatic side chain of a substituted N2-formylkynuramine is not decisive. However, this assumption requires further experimental support.

While deformylation of kynuramine derivatives may depend on cell type-specific expression rates of the arylamine formamidase subforms, the discovery of another AFMK-deformylating activity that was catalyzed by a ubiquitous enzyme, hemoperoxidase (=catalase) [106], led to the conclusion that AFMK could be readily converted to AMK in many cells. Homologous reactions may also occur with other N2-formylkynuramines. Most hemoperoxidases are very unspecific with regard to their hydrogen-donating substrate and, thus, they also accept AFMK. Details of the reaction, which finally leads to the liberation of CO2 from a carbamate intermediate, are presented in Fig. 2.

Figure 2.

 Deformylation of AFMK to AMK by hemoperoxidase (catalase).

More recently, a third pathway of deformylation was discovered. Under irradiation by short wavelength UV light, the photon has sufficient energy to directly cleave the N-C bond under release of CO [107]. Whether or not this is relevant to the skin remains to be investigated and may depend on the melatonin concentrations definitely present in the skin in vivo. In those rodents which display some diurnal activity, the photochemical cleavage might contribute to AMK formation in the Harderian gland, which is exposed to light and in which melatonin is converted to AFMK by photocatalytic [108] and presumably also 5-aminolevulinic acid-dependent radical generating [109] processes. Direct photochemical deformylation of AFMK may, however, be of particular importance in plants exposed to intense UV light, especially those living in tropical, subtropical, and high-altitude biotopes. Sufficiently high amounts of AFMK can be present in such organisms, as shown for Eichhornia [94]. An eventually missing demonstration of AMK in UV-exposed plants should, however, not be misinterpreted because of the high reactivity that is characteristic for this deformylated product and which leads to its rapid disappearance, especially in the presence of singlet oxygen [110] (see next section).

Kynuramines as scavengers and the products of AFMK and AMK

Formation of kynuramines via indoleamine oxidation frequently involves reactive oxygen species. Therefore, they were also of interest because they might contribute to the overall antioxidant efficacy of their parent compounds. This has been investigated in the case of few kynuramines, primarily AFMK and AMK, and, for comparative purposes, in reference to N1-acetylkynuramine.

AFMK was shown to protect DNA [47, 65] and lipids [47] from the attack by hydroxyl radicals. This is not surprising considering the high reactivity of this oxygen species. In the case of DNA protection, the efficacy of AFMK is roughly one-fifth that of melatonin [47]. However, AFMK turned out to be much more resistant to other oxidants, such as carbonate radicals [27, 55, 64], protoporphyrin IX cation radicals [27], quinoxaline-2-oxyl radicals [62], and also singlet oxygen [110], whereas all these agents efficiently destroyed AMK. It was also oxidized at substantially lower rates than AMK in hemin-catalyzed H2O2-systems, in an ABTS-based OH scavenger competition system [ABTS = 2,2′-azino-bis-(3-ethylbenzthiazoline-6-sulfonic acid)], and by ABTS cation radicals [27, 111]. The comparably low reactivity of AFMK was also observed when the methoxylated kynuramines were incubated with activated neutrophils [33].

The lower reactivity of AFMK towards free radicals is associated with the preference of this molecule for two-electron transfer reactions [64], as demonstrated by cyclic voltammetry [47]. As radical reactions represent single-electron transfer reactions, potent radical scavengers exhibit a preference for one-electron exchange [64], as shown for melatonin by cyclic voltammetry [66], and which is evident for AMK based on its efficient interaction with the ABTS cation radical [111], a radical species of comparably low reactivity. Moreover, the N-formyl group affects electron distribution and accessibility of the decisive anilinic amino group to oxidants. This consideration should also be valid for an oxidation product of AFMK, dehydro-AFMK [112] which has a double bond in the long aliphatic chain; but this molecular change should give rise to tautomers which may behave differently in their oxidation chemistry. The relevance of dehydro-AFMK is not yet clear.

From this point of view, it is not suprising that another nonformylated aniline compound, N1-acetylkynuramine (AK), is also a more potent OH scavenger than AFMK [111]. However, AMK was clearly superior to AK, by virtue of the enhanced reactivity introduced by the 5-methoxy group [111].

Despite lower reactivity of AFMK towards free radicals more stable than OH, it is not equally inert towards all of them. This was particularly shown in experiments with the long-lived ABTS cation radical [113]. In this case, a series of substituted C2-substituted 3-indolinones was obtained (Fig. 3), not to be confused with C3-substituted 2-indolinones which represent the keto tautomers of 2-hydroxylated indoleamines, as outlined above. The 3-indolinones are members of a novel class of secondary melatonin metabolites, for which no specific functions have been identified to date, but which may be of interest because of structural similarities to some melatonin receptor ligands [113]. Notably, no AMK was detected in the ABTS cation radical system [113], despite the suspicion that free radicals might perhaps be capable of deformylating AFMK. In fact, this was also not observed in the other purely radical-based oxidation systems mentioned above.

Figure 3.

 Several 3-indolinones formed from AFMK under the influence of ABTS cation radicals. Numbering of indolinone compounds according to Rosen et al. [113]. Structures of compounds 2a, 2b and 3 have been fully identified, the deformylated products 4, 5a, and 5b are partially characterized.

AMK was not only shown to react with various oxidizing free radical species, as already discussed in this section, but also exhibited several other noteworthy properties. Its destruction by carbonate radicals was remarkably fast and complete [27, 111]. In comparison with melatonin oxidized in the same system [25], it interacted more rapidly, indicating an even higher reactivity towards the carbonate radical than the parent indoleamine. It was also a better singlet oxygen scavenger than melatonin [110], and also more potent than classic singlet oxygen quenchers such as histidine, imidazole or diazabicyclo-(2,2,2)-octane [110]. Moreover, AMK scavenged peroxyl radicals generated from 2,2′-azo-bis-(amidinopropane) [111].

The search for oxidation products from AMK turned out to be accompanied by unexpected difficulties resulting from either the decomposition of primary products or oligomerization. When exposed to a Fenton reagent, the AMK solution was decolorized; fluorescent or colored products were only transiently and faintly detectable in chromatographic analyses and finally disappeared. The loss of absorbance was not only apparent at the maximum of 378 nm which is responsible for the yellow color of AMK, but also short wavelengths between 210 and 260 nm, indicating a destruction of the aromate (S.I. Schmidt and R. Hardeland, unpublished data).

Product analyses from reactions with tert-butylhydroperoxide in ethanolic solution [114] revealed the presence of two products with intense light blue fluorescence. These compounds showed a higher mobility than AMK in ethyl acetate/methanol (19:1). Unfortunately, the two substances were so unstable in high performance chromatography–mass spectrometry standard techniques that they could not be structually characterized. Preliminary data from direct chemical ionization-MS of silyl derivatives indicated that these compounds may be monomers. Despite the missing characterization, they seem worthy of being mentioned here because of an unusual property: during their decay, they regenerated AMK. Whether this indicates a multiple action of AMK in scavenging oxidants could be an interesting future study. Over time, the two intensely fluorescent compounds and also AMK were converted to a dark blue fluorescent substance of lower mobility than kynuramine. This compound has not yet been characterized, again for reasons of instability.

In the subsequent attempts to identify oxidation products of AMK, the much less reactive ABTS cation radical was used. In fact, numerous compounds were obtained from this oxidation system, but most of them were presumed to be unphysiological. The prevailing products were dimers and oligomers [115], resulting from AMK concentrations higher than physiological, a necessity for subsequent product analyses, which cannot be conducted with an educt in the nanomolar range. Despite their artificial nature, the results were of interest, as they indicated the reaction modes of oxidized AMK intermediates. These reaction modes can, nevertheless, be of physiological or pharmacological relevance because they explain the interactions of AMK-derived intermediates with other aromates (cf. Binding of AMK to aromates and protein AMKylation). The anilinic amino group was involved in the formation of all the multimers identified, showing that the reactivity of the intermediates is centered at this nitrogen and may involve the transient formation of an N-centered AMK radical. Several interactions, as shown in Fig. 4, can be discriminated: formation of (i) azo dimers, (ii) NH-bridges between two aromates, eventually under elimination of one methoxy group, presumably via a quinone imine, and (iii) 5-atom rings connecting the two aromates.

Figure 4.

 Oligomers of AMK: several variants of bonding. Azo dimers (I) can exist as either cis or trans isomers.

Another type of oligomer, with a mass of 373 and a sum formula of C22H19N3O3 (by high resolution MS) was obtained when AMK was incubated with ABTS cation radicals in the presence of various amino acids or derivatives thereof [116 and A. Kuschnereit, A.K. Klein, A. Koehler and R. Hardeland, unpublished data]. Obviously, the formation of this product did not depend on a specific amino acid, but rather on the presence of another redox-active compound in the system. The remarkably low hydrogen content of the substance indicates numerous double bonds and can be only explained by assuming a trimer connected by nitrogens under truncation of the side chains. In one of these complex reaction systems a monomeric AMK derivative was also detected; the derivative was N2-acetyl-AMK (N1,N2-diacetyl-5-methoxykynuramine) [116]. This product was obviously formed by transacetylation and was also found in systems containing other acetylated compounds.

Although the oxidation chemistry of AMK was surprisingly complex and awaits further clarification, the interaction of AMK with reactive nitrogen species gave definitive results and led to several newly discovered compounds. Starting with the observation that AMK interacted with traces of reactive nitrogen species in the air when exposed as a dry solid on the large surface of silica gel, subsequent studies in liquid systems showed that AMK was a highly potent scavenger of all NO congeners, NO+, NO and HNO, the protonated NO subform present at physiological pH [117–119]. In all these cases, the same nitrosation product was obtained, 3-acetamidomethyl-6-methoxycinnolinone (AMMC; Fig. 5). Contrary to the unstable, NO re-donating N-nitrosated indoles such as N-nitrosomelatonin, AMMC is remarkably stable due to a second, resonance-stabilized ring [118]. AMK nitrosation by NO+ is most easily explained, as the nitrosonium cation can directly interact under release of a proton and, during cyclization, of water [119]. However, this reaction is, at physiological conditions, the least probable one, as NO+ is extremely short-lived in aqueous solution at physiological pH (half-life about 10−10 s, at pH 7.4). More likely are interactions with NO or HNO. In the case of NO, the additional abstraction of an electron is required [119]. This may be achieved by other radicals including NO2, which is easily formed when NO is generated in the presence of oxygen, or may take place during nitrosation by N2O3 [120], which can be interpreted as the NO/NO2 adduct and would likewise lead to HNO2 (Fig. 5). Nitrosation via HNO is believed to proceed via an N-hydroxylamine intermediate, followed by N-nitroso-AMK which cyclizes to AMMC (Fig. 5). Two pathways are possible for the formation of N-nitroso-AMK differing in the sequence of steps [118]. For all congeners, the nitrosation of AMK to AMMC is specific and rapid, and its rate is determined by the NO release from the respective donor [117, 119].

Figure 5.

 Reaction products of kynuramines with nitrogen species. (A) Four reactions of AMK nitrosation, which lead to AMMC. (B) Putative oxadiazole and o-quinone diazide products of 3-hydroxykynuramine, as deduced from corresponding reactions of 3-hydroxykynurenine. (C) The nitration product of AMK, N1-acetyl-5-methoxy-3-nitrokynuramine (AMNK = 3-nitro-AMK). (D) Another AMK condensation product; MQA, recently shown to be formed in biological material.

Comparisons with other kynuramines revealed analogous reactions, however, at much lower reaction rates. In the case of AFMK [119], AMMC was formed with the loss of the formyl group; this reaction was enhanced by H2O2 [119]. Kynuramine and N2-acetylkynuramine (AK) led, based on the spectral data, to homologous compounds, which are presumably 3-aminomethylcinnolinone and 3-acetamidomethylcinnolinone, respectively (A. Fadavi and R. Hardeland, unpublished data). An entirely different nitrosation product (Fig. 5) has to be expected from 3-hydroxykynuramine, as assumed from experiments with its homologs, 3-hydroxykynurenine and 3-hydroxyanthranilic acid. These potent NO scavengers, which were instantaneously nitrosated, formed oxadiazoles which were converted to the favored tautomers, o-quinone diazides [121]. However, the formation of an additional substituted cinnolinone could not be ruled out, and 3-hydroxykynurenine showed, in fact, a more complex product spectrum than 3-hydroxyanthranilic acid, which is missing a long C3-substituent and, therefore, cannot form a cinnolinone.

Nitration of AMK to 3-nitro-AMK (=N1-acetyl-5-methoxy-3-nitrokynuramine = AMNK) was observed with a mixture of peroxynitrite and hydrogencarbonate [117], a reagent that generates the peroxynitrite-CO2 adduct, ONOOCO2, which decomposes to the carbonate radical, CO3, and NO2. Corresponding experiments in the absence of hydrogen carbonate caused, however, destruction of AMK, most likely due to OH formation from ONOOH and the extreme lability of AMK in the presence of this highly reactive oxidant.

Another interaction product of AMK with a nitrogen compound was discovered in dry solid on a silica gel. This was identified as N-[2-(6-methoxyquinazoline-4-yl)-ethyl] acetamide (MQA; Fig. 5) [117]. Formally, this substance could be interpreted as a condensation product of AFMK with ammonia, but the reaction system was definitely free of AFMK, and later investigations showed that this process requires elevated, nonphysiological temperatures. Despite this fact, a recent study demonstrated MQA formation in yeast incubated with AFMK [122]. Obviously, the compound was generated inside the cells from AMK. Therefore, MQA may now be regarded as a biological metabolite. Incubations of yeast with AMK turned out to be unsuitable because these cells excreted AMK, in contrast to AFMK, which was sufficiently taken up [122]. Details on a biochemically possible pathway for MQA will be presented in near future.

Time patterns and the transitory nature of kynuramines

Although several publications have dealt with the determination of kynuramines, in particular, AFMK [87, 88, 123–125], physiological time patterns without prior administration of melatonin are surprisingly rare. This situation may be partially caused by unsuccessful attempts to determine AFMK in the serum of untreated volunteers [87], so that this approach did not appear promising. However, we would like to emphasize that AFMK is a typical metabolite of nonhepatic tissues, as outlined above, and it seems important to emphasize this fact. Hence, blood levels of AFMK below detection threshold should not be mistaken as a sign of irrelevance. Instead, AFMK should be investigated in tissues and, perhaps, in cultured cells. Accordingly, high levels of AFMK were obtained in HaCaT keratinocytes [92], and this kynuramine was also readily detectable in the rat retina [88]. Moreover, endogenous AFMK was detected in the CSF of patients with meningitis [102]. These findings should encourage a further search for AFMK in tissues.

An indication of AFMK rhythmicity in vertebrates was obtained in the rat retina, but only on the basis of two time points. These data indicated that retinal AFMK was higher during darkness than in the light phase [88]; however, this does not yet allow conclusions as to its phase-relationship with melatonin, especially as the time-related differences were not as large as those of the parent indoleamine. Convincing evidence for a rhythm of substantial amplitude has, to date, only been presented for the water hyacinth, Eichhornia [94]. At the present state of knowledge, it is uncertain whether this finding reflects an endogenous, circadian rhythm or a light-dependent exogenous rhythm mainly caused by increasing damage to photosystems and oxidant formation throughout the photophase. A combination of endogenous and exogenous components may also be possible.

Relatively little is known about the elimination kinetics of AFMK. On one hand, it is more stable than its product, AMK, as outlined above, but, on the other hand, its conversion to AMK by an abundant enzyme, hemoperoxidase (catalase), may be indicative of a more rapid metabolism. Data on this important point are simply missing.

Except for studies based on melatonin administration, such as in the pioneering study of Hirata et al. [7], the chances of finding easily detectable levels of AMK are affected by the high reactivity of this compound. Even if it is generated at a high rate, the high probability of it being metabolized by reactive oxygen and nitrogen species may be a limiting factor for attaining easily detectable concentrations. More recent data on AMK levels are essentially missing, especially with regard to an eventual physiological role. After the discovery of relatively stable AMK products, such as AMMC and MQA, these compound may serve in the future as indicators for the rapidly decaying AMK.

Information on levels and dynamics of other kynuramines is also limited. The formation of N1,N1-dimethylkynuramine from N,N-dimethyltryptamine in red blood cells [86] indicates that this dimethylated kynuramine should be detectable in the blood. The same may be valid for the respective 5-substituted analogs, N1,N1-dimethyl-5-hydroxykynuramine and N1,N1-dimethyl-5-methoxykynuramine, formed by pyrrole ring cleavage and deformylation from the two other structurally related, endogenous hallucinogens bufotenin and N,N-dimethyl-5-methoxytryptamine. The N2-formylated intermediates have never been studied. A few details on brain, blood, and urinary levels have been reported for the unsubstituted kynuramine and for 5-hydroxykynuramine. The latter molecule is also known by the name ‘mausamine’ (Maus, in German = mouse) [126, 127] because of its discovery in mouse urine and it was the first member of the kynuramine family to be detected in biological material. Kynuramine was shown to be physiologically present in the rat brain [128–130] and to attain an overall level of 0.42 nmol/g [128], i.e. more than one-third that of the kynurenine level, but almost twice that of the tryptamine concentration. At least in the brain, it seems to be predominantly formed from kynurenine. Similarly, decarboxylation of 5-hydroxykynurenine to 5-hydroxykynuramine was reported [131, 132]. Marked regional differences were found in the central nervous system [130]. Both kynuramine and 5-hydroxykynuramine were detected in blood and urine, but were in general not sufficiently quantified, or only measured after precursor administration. Plasma levels in rats were reported to be in the range of 0.06 μg/g [128, 130]. Dynamics or their physiological fluctuations have not been studied.

The two compounds, kynuramine and 5-hydroxykynuramine, however, are easily metabolized by MAO (see above) to 4-hydroxyquinoline and 4,6-dihydroxyquinoline, respectively [130, 132]. This indicates a relatively fast metabolism. The same should be valid for other N1-unsubstituted kynuramines, whereas all N1-acetylated analogs are not suitable as MAO substrates, equivalent to the same difference between nonacetylated and N-acetylated indoleamines [90]. The fact that, in brain samples, kynuramine and 5-hydroxykynuramine were detectable in the absence of MAO inhibitors is as an indication for a transient nonaccessibility of the compounds to the enzyme, e.g. because of their accumulation in vesicles [130]. Therefore, the physiological turnover rate of these kynuramines remains open.

Kynuramines as indicator molecules?

As kynuramines are formed under the influence of either reactive oxygen species or via the inflammation-related enzymes, myeloperoxidase and IDO, it may be assumed that their appearance in elevated concentrations could be indicative for oxidative stress and/or activation of neutrophils, macrophages, and related cells including microglia. Rises of AFMK in the cerebrospinal fluid of patients with meningitis [102] support this possibility. However, there is considerable variation between patients, but these were correlated with other immunological parameters such as interleukin levels. The eventual discriminatory potential of this finding regarding subtypes of infection or prognostic value in the individual case may be seen as an intriguing idea worthy of future investigation.

Unfortunately, systematic measurements of in situ AFMK under conditions of oxidative stress or inflammation, in the absence of exogenous melatonin, are not available. In the presence of indoleamine precursors, formation of kynuramines has to be expected, but is unsuitable for judging the physiological relevance. Nevertheless, the possible use of AFMK as a physiological marker of oxidative stress and inflammation should not be precluded. With regard to brain inflammatory diseases, this may extend to the levels of the unsubstituted kynuramine and, perhaps, 3-hydroxy- and 5-hydroxykynuramines. Upon activation of microglia, IDO is induced [17–19] and kynurenines are formed via the marked rise in tryptophan metabolism resulting in their appearance at elevated concentrations. The three last-mentioned compounds are physiologically formed, as outlined above and, at least, kynuramine and 5-hydroxykynuramine attain substantial concentrations in the brain [128–130].

The frequently observed photochemical or photocatalytic formation of AFMK from melatonin, as already discussed in detail, may lead to the assumption that this molecule could be an indicator of exposure to visible and UV light [93]. This would be in accordance with the findings on AFMK formation in UV-exposed HaCaT keratinocytes [92]. It would also conform to the rhythm of AFMK in Eichhornia [94]. However, the data on the water hyacinth also show that the time courses need to be analyzed in detail. In this plant, melatonin also exhibited a marked rhythm, so that the availability of the precursor cannot be neglected. Nevertheless, the assumed photoprotective role of melatonin, along with the usual rises of oxidants with increasing time of light exposure in photoautotrophs [93, 133, 134], is still in favor of AFMK as an indicator of radiation stress or even damage by light. With regard to the high melatonin levels found in plants naturally exposed to intense visible and UV light [93, 135], the precursor would be amply available for the generation of substantial amounts of AFMK. The idea that AFMK might be an indicator of light had already been proposed earlier, in the context of the dinoflagellate Lingulodinium [6]. However, later studies did not reveal a chronobiological potential of this compound, as no robust phase shifting was observed with AFMK (R. Hardeland, unpublished data), but this would not be a criterion for a marker molecule.

Biological and pharmacological actions

Members of the kynuramine family do exert biological effects. However, it is important to clarify as to whether these actions are of a physiological or only of a pharmacological nature. In this regard, many uncertainties remain. The earlier literature was mainly focussed on kynuramine and 5-hydroxykynuramine. Kynuramine was shown to compete with the tryptamine receptor (Ki = 28 nm) in the rat frontal cortex [130, 136], whereas the 5-hydroxylated, 5-methoxylated and N1,N1-dimethylated analogs exhibited much lower affinities to this binding site [130]. Both kynuramine and 5-hydroxykynuramine stimulated the release of norepinephrine from sympathetic nerve fibers, but only when micromolar levels of these molecules were used [130, 137, 138]. Conversely, kynuramine also acted as an α1-adrenergic receptor antagonist [130, 136, 139]. As a consequence, kynuramine was capable of either causing a decrease in blood pressure of relatively short duration or an increase explained by a cardiac, indirect sympathomimetic action resulting from norepinephrine release to enhanced heart rate [130]. Additional effects on α2-adrenergic receptors were observed at higher pharmacological concentrations [130]. Another presumably indirect pharmacological effect was observed in the dinoflagellate Lingulodinium, in which kynuramine strongly stimulated bioluminescence [4, 5, 140, 141] and, additionally, induced the formation of asexual cysts at a concentration of 5 μm, an escape response for avoiding adverse environmental conditions otherwise mediated by 5-methoxytryptamine (5-MT) [141, 142]. These actions, both of which are related to cytoplasmic acidification, can be mimicked by MAO inhibitors [5, 43, 141], which lead to 5-MT accumulation [143]. Accordingly, inhibition of the dinoflagellate MAO by kynuramine also leads to 5-MT responses [6].

A number of earlier studies has dealt with both vasomotor and serotoninergic effects of 5-hydroxykynuramine. Although this compound had no substantial affinity to α1-adrenergic receptors [130], it exerted α2-adrenergic, rauwolscine-sensitive pressor responses, which were clearly distinct from the other serotoninergic effects [130, 144]. However, potentiation of α1-adrenoreceptor-mediated responses to norepinephrine was also described [139, 145]. Nevertheless, serotoninergic actions of 5-hydroxykynuramine were of interest, because of the structural similarity to the parent indoleamine. In fact, 5-hydroxykynuramine displayed affinities for the serotonin receptors 5-HT1 [130, 136], 5-HT2 [130, 146] (5-HT2A and 5-HT2B [147]), and 5-HT3 [146]. In the earlier literature, subtype specificity has not yet been sufficiently clarified, and the pharmacology remains contradictory in certain aspects, especially with regard to 5-HT1 receptors (cf. refs [130] versus [148]). Nevertheless, the following actions of 5-hydroxykynuramine were consistently observed: vasoconstriction [130, 139, 146, 147, 149], release of cardiac catecholamines [130, 137], acetylcholine release in the myenteric plexus [148], inhibition of serotonin-induced platelet aggregation, including the serotonin-mediated potentiation of ADP-induced platelet aggregation [150, 151]. Thus, 5-hydroxykynuramine seems to act either as a serotoninergic agonist or, in other cases, as an antagonist. It may also exert effects independent of serotonin signaling mechanisms, as the release of prolactin by 5-hydroxykynuramine was not blunted by melatonin, contrary to the serotonin-stimulated release [152]; this is in contrast with AMK which did not elicit prolactin secretion [152]. Despite the affinity of 5-hydroxykynuramine to some serotonin receptor subtypes [130, 136], this has always been less than that of the parent indoleamine. Therefore, the actions observed may be considered as being largely or entirely pharmacological. In the case of platelet aggregation in chicken, a pharmacological response has been assumed because no substantial levels of 5-hydroxykynuramine were detected in the blood [151].

The earliest reports [153, 154] on members of the kynuramine family have dealt with N1,N1-dimethylated analogs, which correspond to the endogenous hallucinogens within the indoleamine class, N,N-dimethyltryptamine and bufotenin. Both N1,N1-dimethylkynuramine and N1,N1-dimethyl-5-hydroxykynuramine were shown to exert hypotensive effects in anesthesized rabbits [130, 153, 154]. Moreover, N1,N1-dimethylkynuramine was reported to antagonize the hypertensive action of epinephrine [130, 153]. However, a later study on strips of canine cerebral arteries described methysergide-sensitive stimulations of contraction by N1,N1-dimethyl-5-hydroxykynuramine [149]. On the other hand, serotonin-induced contractions were antagonized by N1,N1-dimethyl-5-hydroxykynuramine [149], so that the two analogs may differ profoundly in their actions, as soon as serotoninergic receptors play a dominant role in the system.

After the discovery of AFMK and AMK, a number of studies has analyzed the effects of these compounds, mainly but not exclusively with the intention of identifying actions related to the parent compound, melatonin. Binding of AFMK to melatonin receptors [155, 156] and to the benzodiazepine-binding sites of GABA receptor complexes [157, 158] revealed, however, only moderate affinities. AMK binding to melatonin receptors was also almost two orders of magnitude lower than that of melatonin [159, 160]; such an action had been speculated because of potent inhibition of retinal dopamine release (reported IC50 = 1 nm) [130]. Perhaps, this value has to be revised. Effects on the reproductive system were also described for both AFMK and AMK, but they remained weaker than those of melatonin [161]. However, contrary to melatonin, no substantial effects of AFMK on prolactin release were observed in ewes [162].

An effect of a chronobiotic nature and being, in this regard, also reminiscent of melatonin, was observed after circadian phase shifts of rats: AFMK was shown to accelerate the re-synchronization of the circadian melatonin rhythm [163]. Unfortunately, this line of research has not been followed by other chronobiologists. This aspect of entrainment re-emerged only recently in an entirely different context, i.e. the life cycle of rodent malaria parasites, which were shown to be synchronized by AFMK [96]. This action was blocked by the MT1/MT2 receptor antagonist luzindole [96]. This latter result may be interpreted either in terms of unspecificity of luzindole, if one assumes a separate AFMK receptor, for which no evidence exists to date, or as a cross-talk between melatonin and AFMK at the melatonin receptor level. In the latter case, this would have considerable implications for understanding the interplay between melatonin and its kynuric metabolite. Moreover, the same study showed AFMK-induced rises in cytosolic calcium, which may be crucial for timing of the cell cycle [96].

Other actions that are particularly worth future attention and may gain considerable importance relate to anti-inflammatory effects and immunomodulation. As early as in 1984, AMK was shown to be a cyclooxygenase (COX) inhibitor more potent (IC50 = 14 nm) than acetylsalicylic acid (aspirin; IC50 = 160 nm) [105]. At that time, COX-1 and -2 were not discriminated. Unpublished data from another laboratory (mentioned in ref. [64]), indicate that AMK may be specific for COX-2, a finding which requires confirmation. A more recent study revealed specific downregulation of COX-2 expression in macrophages, although at pharmacological concentrations [164]. These experiments should be repeated with lower doses and extended to prostaglandin synthesis by cells devoid of high myeloperoxidase activity. At low concentrations, the oxidant-sensitive scavenger AMK would be immediately and quantitatively destroyed during an oxidative burst by activated macrophages.

Anti-inflammatory actions by AFMK and AMK go beyond the interference with prostaglandins. AFMK was reported to inhibit the lipopolysaccharide-induced release of tumor necrosis factor-α and interleukin-8 by neutrophils and, to a lesser extent by mononuclear cells [165]. As AFMK was more potent than melatonin [165], this was not likely an indirect effect via scavenging of oxidants, as in this case, melatonin should have been more efficient. AFMK, also formed by myeloperoxidase-expressing cells, was presumed to mediate some immunological effects of melatonin and to participate in the immunological cross-talk of neutrophils and mononuclear cells [165]. Moreover, AMK, but not AFMK, partially inhibited HOCl formation by neutrophils, but this effect was much less than that of melatonin [166]. However, AMK and even more so AFMK interfered with the bactericidal activity of neutrophils [166], effects to be further elucidated in the future.

Another anti-inflammatory aspect, particularly of AMK, is that related to NO metabolism. AMK is not only a potent NO scavenger [117–119], as outlined above, but also a more effective inhibitor of NO formation than melatonin [167, 168]. Still, some uncertainties remain as to whether these effects are of a pharmacological or physiological nature. Neuronal NO synthase (nNOS) was reported to be inhibited with an IC50 of 70 μm, what would appear to be pharmacological, but the IC20 value was as low as 10−11 m [168]. This unusual dose dependence may be of relevance. Therefore, physiologically possible AMK concentrations might have an effect, which would, however, be overcome by strong NO-generating stimuli. The inhibition of nNOS by AMK was interpreted to be a result of its noncompetitive binding to calmodulin [168]. Moreover, the weaker inhibition elicited by melatonin was blocked by norharmane, a – not fully selective – IDO inhibitor, whereas the AMK effect was norharmane-insensitive [168]. More recently, AMK was also shown to prevent the upregulation of inducible NOS (iNOS), including the mitochondrial subform of this enzyme, in a mouse model of Parkinson’s disease [169].

The mitochondrial actions of AMK represent one of the most intriguing aspects of kynuramine effects. The suppression of NOS subforms may already suffice to explain the prevention of secondary bottlenecks of electron flux, avoidance of electron leakage and maintenance of mitochondrial integrity [120, 170]. In fact, the support of mitochondrial electron flux and oxidative phosphorylation by AMK has been shown at physiologically possible concentrations [171] and in models of sepsis and Parkinson’s disease [169]. Whether or not a hypothesis by B. Poeggeler of an additional, AMK-based electron shuttle [64, 172], developed in analogy to results with substituted nitrones, will receive direct experimental support remains to be studied. The findings on mitochondrial effects are strongly reminiscent of similar findings with melatonin; in this regard, AMK may contribute to the action spectrum of the parent compound. If so, melatonin may be called with some justification a prodrug of AMK [89], despite its numerous direct actions in which kynuramines are not involved.

Protective effects of kynuramines

Some of the properties of 5-methoxylated kynuramines, such as radical scavenging, attenuation of mitochondrial electron leakage and anti-inflammatory actions, are indicative of a neuro- and cell protective potential. In fact, AFMK was shown to protect against oxidative damage in various models, such as chromium(III)/H2O2-induced 8-hydroxy-2-deoxyguanosine formation [47, 65], DNA oxidation and strand breaks by 5-aminolevulinic/Fe(II)-generated hydroxyl radicals [173], hepatic lipid peroxidation by hydroxyl radicals formed in a conventional Fenton reaction [47], cell death of hippocampal neurons induced by H2O2, amyloid β25–35 or glutamate [47], oxidative modification or destruction of DNA, proteins and lipids induced by X-rays [174] or by high energy charged particle rays, in a model of space radiation [175], and it also attenuated KCN-induced superoxide formation [176, 177]. The study using charged particle rays [175] also demonstrated protection of immature and proliferating neurons in the hippocampal dentate gyrus and attenuation of spatial memory losses. As far as hydroxyl radicals are implicated in the respective models, the efficacy of AFMK appears to be plausible as it can directly interact with the highly reactive oxidant. In systems in which primarily other radicals are formed, an involvement of AMK may be assumed, which is a more potent radical scavenger. Additionally, AMK is capable of preventing damage to the mitochondrial electron transport chain and, thus, secondary oxidant formation by dysfunctional mitochondria. Further mechanisms of protection by AFMK may result from inhibition of inflammatory cytokine release [165]. This should extend to AMK formed in the respective systems, as it suppresses other inflammatory signaling molecules, such as prostaglandins [105, 164] and NO [117–119, 167–169].

Although AMK is a more potent antioxidant than AFMK, and considerably more effective as an anti-nitrosating agent, the deformylated metabolite has been sparcely tested with regard to its protective potential. It was shown to inhibit protein destruction by peroxyl radicals in vitro [111]. The presumably best evidence for a protective role comes from mitochondrial studies, in which AMK not only inhibited MPTP-induced mitochondrial iNOS, but also restored Complex I activity and, thus, eliminated a bottleneck of electron throughput and a source of radicals resulting from electron leakage [169]. Moreover, another study on the support of mitochondrial function [171] demonstrated effects at low, near-physiological AMK levels. Thus, the role of this kynuramine may be more than just pharmacology.

Binding of AMK to aromates and protein AMKylation

Oligomerization of AMK products under oxidative conditions [115] is, at physiological concentrations, an uncommon process. However, a reactive intermediate derived from AMK by interaction with another free radical would be assumed to have other partners for adduct formation in the biological environment; these should most likely be aromates. For this reason, amino acids were incubated in the presence of AMK with ABTS cation radicals. The main problems in such experiments resulted from the unavoidable AMK oligomerization at concentrations required in chemical reaction systems, and from additional reactions involving the carboxy and amino groups of the amino acids. To avoid the last difficulty, amino acid analogs with differently truncated side chains were used. In fact, AMK was shown to form an adduct with a tyrosine fragment, 4-ethylphenol, and this product was unambiguously characterized [178]. The structure of the deduced AMK-tyrosyl adduct, as it would appear in a modified protein, it depicted in Fig. 6. AMK also interacts easily with tryptophan and its analogs. However, the products obtained in vitro have to date not been suitable for structural analyses, as they represent mixed oligomers or polymers, which are different from those obtained from AMK alone or tryptophan and its analogs alone (A.K. Klein, C. Heer and R. Hardeland, unpublished data). Interactions with the two aromatic amino acids, tyrosine and tryptophan, may have biological relevance because this occurs at low AMK concentrations and may be likely with regard to ready electron donation by AMK.

Figure 6.

 Structure of the putative AMK-tyrosyl adduct, as deduced from the identified AMK-4-ethylphenol adduct. R1, R2: peptide chain residues.

At the moment, it seems difficult to foresee whether a process of protein AMKylation [178] would be beneficial or detrimental. Especially tyrosyl residues are abundant in proteins of premier regulatory relevance, in particular, tyrosine receptor kinases. In the usually long cytoplasmic domains of such receptors, designed by nature not only for mutual tyrosine phosphorylation but also for the attachment of other key regulator proteins to single or double phosphotyrosyls, an AMK-tyrosyl adduct would have considerable consequences. Many of the tyrosine receptor kinases are involved in stimulation of cell proliferation. Therefore, it might be worthy to investigate whether AMK is capable of preventing cell proliferation via tyrosyl AMKylation, e.g. under the unfavorable conditions of on-going inflammation, which would bear the risk of DNA damage. The enhanced availability of oxidants during inflammation might favor both the production of AMK and the formation of reactive AMK intermediates capable of attaching to tyrosyl residues. However, another scenario of protein AMKylation would appear unfavorable, namely, the possibility that AMKylated proteins might be a source of autoimmunological or allergenic responses. Perhaps, this latter possibility may not be likely at low physiological AMK concentrations and as long as the modified proteins remain within the cells. Nevertheless, this has to be regarded as a caveat if AMK is to be considered a drug, e.g. for treating inflammation.


The molecular family of kynuramines first received attention because of the cardiovascular effects of kynuramine, 5-hydroxykynuramine and their N1,N1-dimethylated analogs. Additionally, the interference of especially 5-hydroxykynuramine with serotonin receptors became of interest as well as a number of indirect effects on the secretion of other neurotransmitters. Unfortunately, these studies were not extended when modern techniques became available, perhaps because the pharmaceutical use of those kynuramines did not appear to be promising. Along with the increasing awareness of the importance of melatonin, the kynuric metabolites of this 5-methoxylated indoleamine came into the focus. AFMK and AMK are currently regarded as molecules of particular interest and publications on these molecules are steadily increasing in frequency. The amounts of the 5-methoxylated kynuramines formed from tissue melatonin and by actions of activated leukocytes seem to be substantial, apart from the long-known, but frequently disregarded finding that they represent major melatonin metabolites in the brain [7]. Moreover, remarkable amounts of AFMK have been found in a vascular plant, Eichhornia [94], results that may be the starting point for searches for kynuramines as secondary plant metabolites. Cytoprotective actions have been demonstrated for both AFMK and AMK. Especially AMK deserves future attention due to several remarkable properties: AMK is (i) a potent scavenger of reactive oxygen and nitrogen species, (ii) a source of new cinnolinones and quinazolines possibly having additional pharmacological properties [116], (iii) a potent cyclooxygenase inhibitor, (iv) a suppressor of neuronal and inducible NO synthases, and (v) a modifier of mitochondrial function that supports electron flux and attenuates electron leakage. Whether or not compounds like AMK or, as a prodrug, its precursor AFMK may be suitable for pharmaceutical application, requires further exploitation. Despite the highly attractive actions as an anti-inflammatory agent and modifier of mitochondrial metabolism, a crucial question will be the judgment on adduct formation by reactive AMK-derived intermediates, which may be beneficial or detrimental, especially if this should exceed AMKylation of proteins and extend to formation of DNA adducts.

A major gap in our understanding of the functions of AFMK and AMK is the lack of knowledge on the molecular mechanisms of action, as far as they exceed the purely chemical reactions with free radicals and other aromates. The possibility that kynuramines of the various types may be signaling molecules in their own right was addressed already two decades ago [130], but this line of research was not adequately followed. Investigators should not only study the interference of kynuramines with receptors for other ligands, but also identify binding sites of possible physiological relevance. In particular, a search for AFMK and AMK receptors might be worth the effort.