By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
I. Feussner, Georg-August-University Goettingen, Albrecht-von-Haller-Institute for Plant Sciences, Ernst-Caspari-Building, Department of Plant Biochemistry, Justus-von-Liebig-Weg 11, D-37077 Goettingen, Germany Fax: +49 551395749 Tel: +49 551 395743 E-mail: firstname.lastname@example.org
In nearly every living organism, metabolites derived from lipid peroxidation, the so-called oxylipins, are involved in regulating developmental processes as well as environmental responses. Among these bioactive lipids, the mammalian and plant oxylipins are the best characterized, and much information about their physiological role and biosynthetic pathways has accumulated during recent years. Although the occurrence of oxylipins and enzymes involved in their biosynthesis has been studied for nearly three decades, knowledge about fungal oxylipins is still scarce as compared with the situation in plants and mammals. However, the research performed so far has shown that the structural diversity of oxylipins produced by fungi is high and, furthermore, that the enzymes involved in oxylipin metabolism are diverse and often exhibit unusual catalytic activities. The aim of this review is to present a synopsis of the oxylipins identified so far in fungi and the enzymes involved in their biosynthesis.
Oxylipins constitute a large family of oxidized fatty acids and metabolites derived therefrom. These bioactive lipids are abundant in mammals  as well as in nonmammals, including flowering plants , mosses, algae, bacteria and fungi . In plants, they serve as signal molecules regulating developmental processes such as pollen formation, and mediate responses to biotic and abiotic stresses such as herbivore or pathogen attack and desiccation .
In plants, oxylipins include fatty acid hydroperoxides, hydroxyl, epoxy, keto and oxo fatty acids, and epoxy alcohols, divinyl ethers, volatile alcohols or aldehydes, and jasmonic acid (JA) and its corresponding derivatives . These compounds are enzymatically formed by an initial peroxidation reaction of a polyunsaturated fatty acid (PUFA) that is catalysed by lipoxygenases (LOXs), thus starting the so-called LOX pathways. LOXs are nonhaem iron-containing enzymes that catalyse the stereospecific and regiospecific oxidation of a PUFA, containing a 1Z,4Z-pentadiene system  (Table 1). The hydroperoxy fatty acid formed is converted by further enzymes [i.e. allene oxide synthase (AOS), divinyl ether synthase (DES), hydroperoxide lyase (HPL), and peroxygenase), yielding a large variety of structurally different products, as shown in Fig. 1 . The majority of the hydroperoxide-transforming enzymes, namely AOS, HPL and DES, belong to the family of unusual cytochrome P450 enzymes – the Cyp74 enzymes [8,9]. In contrast to classical cytochrome P450s, these enzymes do not require molecular oxygen or NAD(P)H-dependent reductases, as they use the acyl hydroperoxide as their natural substrate and oxygen source . One specific branch of this LOX pathway, the 13-AOS branch with α-linolenic acid [18:3(n-3)] as substrate, leads to the formation of JA and its derivatives; this plant hormone is perceived via the SCFCOI1 complex [11,12], and is involved in plant stress and defence responses as well as in developmental processes .
Table 1. Basic biochemical properties of LOX, PGHS and Ppo/LDS (adapted from ).
Ppo/LDS DOX domain
Cytochrome P450 domain
PPIX, protoporphyrin IX.
Animals, plants, some prokaryotes
Animals: 75–80 kDa Plants: 94–105 kDa
68 kDa (active as a dimer)
110–130 kDa (active as a tetramer)
110–130 kDa (active as a tetramer)
Metal in the active centre
Haem (PPIX-Fe) (iron coordinated via the fifth His ligand)
Haem (PPIX-Fe) (iron coordinated via the fifth His ligand
Haem (PPIX-Fe) (iron coordinated via the fifth Cys ligand)
None (some require Ca2+)
Reducing equivalents for peroxidase activity
Reducing equivalents for peroxidase activity
Dioxygenation of C18-PUFA, C20-PUFA and C22-PUFA
Dioxygenation of C18-PUFA and C20-PUFA Cyclooxygenation of AA Peroxidation of hydroperoxides
Dioxygenation of C18-monounsaturated fatty acids as well as PUFAs
Isomerization of a fatty acid hydroperoxide to a dihydroxy fatty acid
PPIX*+-Fe(IV)=O (compound I)?
In contrast to plants, in which C18 fatty acids are prevalent and therefore most abundantly used for oxylipin biosynthesis, mammals use predominantly C20 fatty acids. The eicosanoids [i.e. prostanoids and leukotrienes (LTs)] constitute the best-characterized group of mammalian oxylipins, and are formed from either arachidonic acid [AA, 20:4(n-6)] or eicosapentanoic acid [EPA, 20:5(n-3)] . However, it should be noted that evidence is accumulating that signalling molecules derived from docosahexaenoic acid [22:6(n-3)] also play an important role, at least in mammals [15,16]. As in the LOX pathway in plants, the biosynthesis of eicosanoids in mammals starts with the peroxidation of the fatty acid substrate. One important LOX pathway is catalysed by 5-LOX, leading to the formation of a 5-hydroperoxide derivative – the common precursor for all LTs. This compound is converted by a cascade of other enzymes to further products, as illustrated in Fig. 2 . Alternatively, AA can be oxidized by the haem-containing prostaglandin (PG) endoperoxide H synthases (PGHSs), which introduce two equivalents of oxygen to the fatty acid backbone, yielding PGH2 (Fig. 2). These haem enzymes are bifunctional, as they catalyse the oxidation of AA to PGG2 (the so-called cyclooxygenase reaction), as well as the further reduction of this compound to PGH2 (the so-called peroxidase reaction) (Table 1) . The latter metabolite can be converted by further enzymes, yielding a large variety of different products – the so-called prostanoids (i.e. PGs and thromboxanes) (Fig. 2). Notably, two of these antagonistically acting enzymes are also unusual cytochrome P450s [i.e. prostacyclin synthase and thromboxane synthase (TXAS)], and have catalytic properties in common with the Cyp74 enzymes . The biological activity of mammalian oxylipins is often mediated through G-protein-coupled receptors (GPCRs), which mediate and coordinate immune responses .
Although the occurrence of oxylipins in fungi was reported nearly three decades ago [20–22], knowledge about their physiological function, except for their role in Aspergillus, and their biosynthetic pathways is still scarce. In this review, we will summarize which oxylipin species have been identified so far in which fungal organism, and focus on potential biosynthetic pathways. Regarding their potential functional role as cross-kingdom communication molecules, we will discuss, in the last section, possible interactions between the fungus and the host.
Formation of eicosanoids
In recent years, a number of fungal genomes have been analysed and sequenced (for further information, see http://www.broadinstitute.org/scientific-community/science/projects/fungal-genome-initiative/fungal-genome-initiative), facilitating the prediction of potential oxylipin-biosynthetic enzymes and routes. Interestingly, in many cases, the in silico identification of enzymes responsible for catalysing an experimentally observed reaction has been proven to be difficult – a fact that might indicate that fungal biosynthetic routes differ significantly from mammalian and plant pathways. For example, in the late 1990s and early 2000, it was reported that several pathogenic and nonpathogenic fungi (i.e. Candida albicans, Cryptococcus neoformans, Epidermophyton floccosum, Fusarium dimerum, Microsporum audiouinii, Microsporum canis, Trichophyton rubrum, Sporotrix schenkii, Absidia corymbifera, Aspergillus fumigatus, Histoplasma capsulatum, Blastomyces dermatitis, Penicillium spp., Rhizopus spp, and Rhizomucor pusillus) are able to form PGs (e.g. PGE2, PGF2α and/or PGD2) and LTs (i.e. CysLT and LTB4), either de novo or from exogenously added precursors such as AA [14,23–25]. Additionally, it was observed that Mortierella and Cunninghamella are able to generate PGE2 and PGF2α , and several yeasts of the Lipomycetaceae family and Saccharomyces cerevisiae have also been reported to produce PGF2 and PGF2-lactone [14,27]. Furthermore, it was recently shown that the fungus Paracoccidioides brasiliensis also utilizes both exogenous and endogenous AA for the synthesis of PGEX [28,29]. As both basidiomycetes and ascomycetes generally contain high amounts of C18-PUFAs but only very minor amounts of C20-PUFAs , which are needed for PG formation in mammals, this raises the question of the substrate availability for fungal PG biosynthesis. For C. neoformans, it was even shown that AA is not produced endogenously . Hence, it was speculated that host-derived AA serves as a substrate for the synthesis of fungal PGs . In this respect, it has been shown that Ca. albicans can cause release of AA from host tissues  and is also able to take it up when it is added exogenously  – a finding that supports this hypothesis.
As the early analysis of fungal PGs was based on ELISA, a technique that is prone to the identification of false positives, there is an ongoing debate about the authenticity of the observed compounds. Recent studies, however, have identified the PG derivative from C. neoformans  and Ca. albicans  as a PGE2 subtype by using an MS approach. The low amount of PGE2 formed in these experiments (ng·mL−1) and the slow kinetics of product formation (hours) as reported in these studies are nevertheless in conflict with an enzymatic reaction, and might also point to a nonspecific lipid peroxidation mechanism. In order to identify putative enzyme(s) responsible for PG synthesis, several studies were performed with different chemical inhibitors. These studies showed, curiously, that the use of PGHS inhibitors caused diverse effects on PG synthesis. In Ca. albicans, for example, PGE2 biosynthesis was inhibited not only by different nonselective PGHS inhibitors but also by the LOX inhibitor nordihydroguaiaretic acid . A similar effect was observed in P. brasiliensis, where the use of PGHS inhibitors led to a reduction in PGE2 production . On the other hand, the inhibitor CAY10404 [3-(4-methylsulphonylphenyl)-4-phenyl-5-trifluoromethylisoxaszole], which specifically inhibits PGHS-2, did not affect the formation of PGE2 in Ca. albicans . A similar result was obtained for C. neoformans when indomethacin or aspirin was used .
On the basis of these findings, it was suggested that the enzyme(s) responsible for the synthesis of PGE2 have general catalytic/structural features that are common with but also distinct from, in some regards, to mammalian LOXs and PGHS [34,35].
Thus, the identification of PGHS-like enzymes on the basis of these experimental data turned out to be problematic, and did not lead to clear candidate enzyme(s). This situation is further complicated by the fact that extensive database searches of different fungal genomes have not revealed sequences that show significant similarity to PGHS, except for precocious sexual inducer (Psi) factor-producing oxygenase (Ppo) enzymes . These enzymes have been identified by the Keller group, and will be discussed in more detail later. Instead, two non-PGHS/LOX-related enzymes have been shown to be involved in PG biosynthesis: the desaturase homolog Ole2, and the multicopper oxidase Fet3, which is a homologue of laccases. Strains in which either the Fet3 or the Ole3 allele was deleted showed no or a highly reduced ability to form PGE2. However, whereas the laccase alone was unable to convert either AA or PGH2 to PGE2, it transformed PGG2 into PGE2 .
In conclusion, the contradictory and often ambiguous results presented in the studies mentioned above, as well as the fact that no PGHS-related enzymes seem to be encoded by the genomes of several fungal species, makes the existence of a PG biosynthetic pathway in fungi very unlikely. The observed reactions resemble more an isoprostane type of nonspecific lipid peroxidation reaction that might be catalysed by any protein that harbours iron as a cofactor.
A further compound that has been thought to be formed in a PGHS-mediated reaction is 3-hydroxyeicosatetraenoic acid (3-HETE). This metabolite was first discovered in the South African fungus Dipodascus uninucleata , and was specifically produced from exogenous AA in a reaction that could be inhibited by aspirin. Later, when the stereochemistry of this metabolite was determined as the R-configuration, it became clear that 3R-HETE resulted from an incomplete β-oxidation cycle . Further studies showed that other 3-hydroxy oxylipins with different chain lengths were also formed when the fungus was incubated with different fatty acid substrates . These metabolites were mainly associated with the sexual phase during development, and were found to be localized at the outer surface of the sexual ascospores – a finding that might indicate that 3-hydroxy oxylipins serve as growth factors during sexual development . Other studies have shown that 3-hydroxy oxylipins are also formed by the baker’s yeast Saccharomyces cerevisiae  and also by Ca. albicans , where, additionally, 3,18-diHETE was detected . The latter is most likely formed by a classical cytochrome P450 reaction, which was reviewed recently in an accompanying article . Further biochemical studies have demonstrated that 3R-HETE can mimic AA, and consequently can be converted by PGHS, yielding novel 3-hydroxyeicosanoid compounds that are thought to harbour biological activity, as has been demonstrated . In a presumably similar reaction, D. uninucleata was also able to transform anandamide, an endocannabinoid and endovanilloid, into its corresponding 3-hydroxy derivative. The resulting compound showed dramatically reduced binding activity towards both cannabinoid receptor types (CB1 and CB2), whereas it retained activity at the TRVP1 receptor. On the basis of this finding, it was suggested that 3-HETE-producing yeasts contribute to inflammatory responses associated with fungal infection by converting host-derived anandamide to its 3-hydroxy derivative .
LOXs and LOX-derived products
Recently, another class of lipids derived from EPA have been observed in Ca. albicans – the so-called resolvins . Together with lipoxins and PGs, the resolvins promote inflammation in mammals . Here, these compounds are formed in a biosynthetic pathway that involves the reaction with aspirin-acetylated PGHS-2 and 5-LOX . It was demonstrated that Ca. albicans is also able to convert EPA to resolvin-E1 (RvE1); notably, this compound was chemically identical to that produced by humans. Moreover, it was shown that the biosynthesis of RvE1 is sensitive to specific chemical LOX inhibitors (i.e. esculetin, a 12/15-LOX inhibitor, and zileuton, a 5-LOX inhibitor) and cytochrome P450 inhibitors (i.e. 17-octadecynoic acid), indicating that enzymes of these classes might also be involved in forming RvE1. Although these inhibitors are thought to be specific for the respective enzymes, it is still unknown whether these compounds can also inhibit other enzymatic reactions and influence other pathways. Furthermore, in silico analysis of the Ca. albicans genome again failed to identify ORFs with significant homology to known LOX sequences . This raises again the question of whether the observed reactions resemble more isoprostane types of unspecific lipid peroxidation reactions that might be catalysed by any protein that harbours iron as a cofactor.
An example indicating that fungal biosynthetic routes might differ significantly from known mammalian or plant pathways is the following observation, reported recently by Jerneren et al. : it was demonstrated in this study that a cell-free extract of Aspergillus terreus is able to sequentially transform linoleic acid [18:2(n-6)] (LA), among other things, to the corresponding 9R-hydroperoxy derivative (9R-HPODE), and then to α-ketol and γ-ketol derivatives [i.e. (12Z)-9-hydroxy-10-oxo-octadec-12-enoic acid and (11E)-13-hydroxy-10-oxo-octadec-11-enoic acid]. From the plant field, it is already known that these products are formed nonenzymatically from an unstable allene oxide by chemical hydrolysis . In planta, allene oxide is formed from either 9-hydroperoxy or 13-hydroperoxy PUFAs by the action of an AOS, and, in the case of 13-HPOTE, the allene oxide is immediately further converted by allene oxide cyclase (AOC), yielding a cyclopentenone derivative . This product can be further transformed to the plant hormone JA. Thus, the finding that A. terreus seems to be able to express an AOS-like enzyme ultimately raises the question of whether fungi are also able to synthesize JA. Indeed, several fungal species have been shown to form and to secrete JA and numerous different JA derivatives e.g. Botrydiplodia theobromae , Lasiodiplodia theobromae , Aspergillus niger  and Fusarium oxysporum . For L. theobromae, it was shown recently that JA originates from the fatty acid synthetic pathway, and is presumably formed via the intermediary formation of a cyclopentenone, similar to the pathway in plants . However, Tsukada et al. reported that the facial selectivity of the cyclopentenone reduction leading to the formation of JA differs between fungi and plants (α versus β hydrogen attack), emphasizing the possibility of different biosynthetic routes in the two organisms. An additional argument for two different biosynthetic JA pathways lies in the observation of Jerneren et al. that the in silico analysis of the A. terreus genome did not reveal an ORF that showed significant homology to any LOX sequence. Therefore, it was suggested that the synthesis of the 9R-hydroperoxide is catalysed by a novel 9R-DOX that shows catalytic but no structural homology to LOXs. In line with these findings are those of previous studies by Matsuda et al., who, in 1976, isolated, characterized and crystallized a haem protein from F. oxysporum that showed LOX activity. This strikingly small enzyme, with a molecular mass of 12 kDa, converted LA at an alkaline pH to a mixture of 9-hydroperoxides and 13-hydroperoxides . As these early studies did not follow the stereochemistry of the formed oxylipins, and similar reactions in cyanobacteria, plants and mammals are catalysed by LOXs that contain nonhaem iron as a cofactor and are significantly larger (65–105 kDa), it will be interesting to see which enzyme is, indeed, catalysing the observed reaction (Table 1).
It should be emphasized at this point that the fungal AOS activity seems to differ from the those known so far from solanaceous plants. These are specific for 9S-hydroperoxides , whereas the putative fungal AOS specifically converted the 9R-hydroperoxy derivative . In addition, it should be stressed that only 13-AOSs, which are essential for JA biosynthesis, are found ubiquitously in plants . Whether similar enzymes also occur in some fungal species remains unknown.
On the other hand, several LOXs have been identified in different fungal species. The first bifunctional LOX was identified in the oomycete Saprolegnia parasitica [53,54]. In a first reaction step, the enzyme oxidized AA, yielding the 15-hydroperoxy derivative, which was further isomerized in a second reaction to two epoxy alcohols, as outlined in Fig. 1: (5Z,8Z,11S,12R,13E,15S)-11,12-epoxy-15-hydroperoxy-5,8,13-eicosatrienoic acid and (5Z,8Z,11S,13R,14R,15S)-13,14-epoxy-15-hydroperoxy-5,8,11-eicosatrienoic acid. The only other LOX displaying bifunctional activity that has been identified so far is from the moss Physcomitrella patens, which possesses a classical LOX and an additional hydroperoxide lyase activity .
More recent studies reported in addition that classical LOX activities were also detectable in extracts of several different fungal species, such as Geotrichum candidum , Penicillium camemberti, Penicillium roqueforti , Morchella esculenta  and various Mortierella strains . Furthermore, several LOXs have been purified from Terfezia claveryi [60,61], Thermomyces langinosus  and Pleurotus ostereatus , and have been partially characterized.
The most unusual and fascinating fungal LOX is the so-called manganese LOX (MnLOX). It was identified in 1998, and since then has been studied by Oliw et al. [64–70]. In contrast to all classical LOXs, this enzyme contains catalytic manganese instead of a mononuclear iron centre. EPR studies demonstrated that, in analogy to iron-containing LOXs, MnLOX exists in two different oxidation states. The resting enzyme contains Mn2+, and is inactive with any fatty acid substrate. Upon reaction with hydroperoxides, the manganese is oxidized, yielding Mn3+-OH, and the enzyme becomes activated. Thus, the manganese redox state is thought to cycle during catalysis between the inactive Mn2+ and the active Mn3+ in a comparable way as it is known for iron (Fe2+ and Fe3+, respectively) in classical LOXs . Despite these catalytic similarities, the presence of manganese in the mononuclear active site instead of iron is thought to have a remarkable effect on the reaction mechanism, as well as on the oxygenation products: The stereospecific abstraction of the proS hydrogen from the bis-allylic carbon in the (1Z-4Z)-pentadiene system of LA (i.e. C-11) represents, for iron-containing as well as manganese-containing LOXs, the first reaction step, and leads to the formation of a delocalized carbon-centred radical [6,72]. Notably, at this early stage of the reaction process, the mechanisms of LOX and MnLOX diverge: in the case of LOX, dioxygen attacks the alkyl radical in an antarafacial way, either at the [+2]-position or the [–2]-position with regard to the bis-allylic carbon atom, thus forming either, for example, the 9-hydroperoxy or 13-hydroperoxy derivative . Oxygen attack mediated by MnLOX, in contrast, occurs reversibly at the bis-allylic C-11 in a suprafacial way, forming the 11S-hydroperoxy radical, which is subsequently converted to the corresponding bis-allylic hydroperoxyl derivative. Unlike classical LOXs, MnLOX also isomerizes the 11-hydroperoxy derivative, yielding the respective 13-hydroperoxy fatty acid. This reaction proceeds via a mechanism that involves oxidation of this substrate to the peroxyl radical, followed by β-fragmentation and irreversible suprafacial oxygen attack at the newly formed delocalized alkyl radical at C-13 (outlined in Fig. 3) [64,69,73]. Recently, it was shown that the oxidation step is likely to occur via proton-coupled electron transfer, in which the H+ is transferred to the Mn3+-OH (catalytic base), and the electron is simulatneously transferred from the peroxide anion to Mn3+ .
It should be emphasized here that the substitution of the iron by manganese may not be the only factor facilitating the formation of the bis-allylic hydroperoxide derivative. Recently, a bis-allylic mini-LOX has been identified from Cyanothece sp. that contains a catalytic iron centre .
As with classical LOXs, the oxidation of fatty acids by MnLOX is characterized by a kinetic lag phase, in which the metal centre is oxidized to the active form and thus can be shortened/abolished by the addition of hydroperoxides [73,75]. In addition, it was shown recently that the isomerization of the 11-hydroperoxide to the respective 13-hydroperoxide also proceeds with a kinetic lag phase that, in contrast, is augmented by increasing substrate concentrations .
Initially, native MnLOX was isolated and purified from Gaeoumannomyces graminis ; in later studies, however, the enzyme was heterologously expressed in Pichia pastoris . With this system, site-directed mutagenesis studies identified not only amino acids that were responsible for ligating the manganese in the active site centre, but also Gly316 as an important residue for the reaction . This residue is conserved in almost all R-LOXs, whereas S-LOXs usually contain an Ala at this position . Moreover, this residue is also involved in directing oxygen attack either at the [+2]-position or [–2]-position with regard to the bis-allylic carbon atom. The replacement of this particular Gly by Ala in MnLOX not only shifted the position of oxygenation from C-13 to C-11 and C-9 when LA was used as a substrate, but also led to isomerization of the 13-hydroperoxide, yielding epoxy alcohols , akin to the reaction products of the LOX from S. parasitica  (se above). At present, it is still unclear what might be the chemical advantages of manganese over iron or vice versa. In a recent study by Goldsmith et al., a biomimetic Mn3+-OH model complex ([Mn3+ (2,6-bis(bis(2-pyridyl)methoxymethane)-pyridine)(OH)](CF3SO3)2) was synthesized and analysed . Although greater structural constraints were observed upon the redox change from Mn3+ to Mn2+ than with similar changes in iron homologues (because of Jahn–Teller distortions in the Mn3+ complex), the results of this study indicated that manganese and iron with similar coordination spheres can oxidize substrates via a similar mechanism .
Although the biological function of MnLOX is not known, it has been suggested that MnLOX-derived oxylipins are involved in causing oxidative damage to wheat root cells when G. graminis mycelia penetrate this tissue . In this respect, MnLOX has recently been applied in a paper production process, where it was used for delignification and to remove lipophilic extractives from flax and eucalypt pulps .
A possible function of fungal LOX-derived oxylipins might be to induce programmed cell death, as is also proposed for plant oxylipins . Furthermore, evidence is accumulating that these metabolites also influence population density-dependent transitions of dimorphic fungi: the selective inhibition of LOX in Ceratocystis ulmi by chemical inhibitors such as nordihydroguaiaretic acid led to a shift from fungal mycelia to the yeast form . In addition, it was shown that deletion of the lox allele in Aspergillus flavus affected the cell density-dependent sclerotial-to-conidial transition, suggesting that LOX-derived oxylipins might also be involved in quorum sensing – a process by which single cells coordinate their activities and act as a multicellular organism . Moreover, deletion of the homolog lox allele in Aspergillus ochraceus indicated that LOX products play a role in the developmental process as well as in the metabolic pathway that leads to mycotoxin formation by the fungus .
In the plant field, it has been discussed whether exogenously applied oxylipins lead to a different response than that to the same endogenously formed compounds . Several lines of evidence have indicated that this may also be the case in fungi, and may even have relevance for the interaction between a plant and a fungus: besides the influence of endogenously formed oxylipins, it was additionally shown that applying 9-HPODE or 13-HPODE to Aspergillus nidulans led to sporogenic effects and influenced the conidium-to-cleisthotetium shift [84,85]. Furthermore, it was reported that 9-HPODE and 13-HPODE inversely affected mycotoxin production [85,86]. On the basis of these findings, it was suggested that plant oxylipins are able to mimic fungal oxylipins, and thus it was proposed that reciprocal crosstalk between the plant and the fungus occurs  (see below). This hypothesis was supported by a study demonstrating that LOX expression was altered in maize and peanut seeds during infection with Aspergillus, and further strengthened when the expression of a maize lox gene in an oxylipin mutant of Aspergillus was shown to partially restore the phenotype . In this respect, it was additionally shown that a maize 9-LOX mutant was more susceptible to Aspergillus infections than the wild type .
In 1986, the first fungal non-LOX-derived or non-PGHS-derived product was observed to be secreted by Laetisaria arvalis, and was identified as an 8-hydroxy derivative of LA (8-HODE) . Later studies showed that this compound, which was also referred to as laetiaric acid, is also abundant in A. nidulans [22,90], Leptomituslacteus  and G. graminis . In addition, G. graminis was capable of converting LA to 7S,8S-diHODE . In subsequent studies, the responsible enzyme was identified [94,95], and was termed 7,8-linoleate diol synthase (7,8-LDS). During the last decade, 7,8-LDS was purified , cloned [96–98] and intensively studied [95,97–100]. It turned out that this enzyme is a haem protein with bifunctional activity: In a first reaction step, LA is oxidized to the corresponding 8R-hydroperoxy derivative by abstraction of the proS hydrogen atom from C-8 and antarafacial oxygen insertion. This intermediate product is further isomerized in a second reaction step, by a suprafacial mechanism, yielding 7S,8S-diHODE as an end-product. Biochemical and biophysical studies demonstrated that 7,8-LDS might use a similar reaction mechanism as PGHS for oxidizing a fatty acid substrate . Indeed, isolation and analysis of 7,8-LDS cDNA revealed that the enzyme shares 23–24% identity with mammalian PGHS-2 . On the basis of the sequence homology, potential haem ligands and the essential catalytic Tyr could be identified. Subsequent site-directed mutagenesis studies confirmed the structural and catalytic similarities between the enzymes [97,98].
Recently, 7,8-LDS has also been detected in Magnaporthe grisea  and Magnaporthe oryzae . The genome of M. oryzae contains two genes with homology to 7,8-LDS, and the contribution of one gene to the pathogenicity has been studied recently by using a knockout mutant. Notably, whereas this genetically modified strain showed no capacity for synthesizing any oxylipin metabolite associated with 7,8-LDS activity, the pathogenicity appeared to be unaffected. Furthermore, no effect upon sporulation was observed , hence leaving the physiological relevance of this enzyme for M. oryzae open. In Cercospora zea-maydis, however, an additional lds gene has been reported that is expressed under conditions promoting the production of the mycotoxin cercosporin. Thus, it was speculated that this gene might be involved in regulating mycotoxin production and pathogenesis . An additional enzyme with significant homology to LDS and PGHS was isolated recently from Ustilago maydis, and was termed spore-specific protein-1. This enzyme is abundantly expressed in mature teliospores, and is localized to spheric organelles that might be lipid bodies .
Other studies have shown that homogenates of A. nidulans, A. fumigatus and Aspergillus clavatus are able to covert LA sequentially into 8-H(P)ODE and 5S,8R-diHODE as major products, indicating the activity of an additional LDS-like enzyme [105,106]. In addition, 10R-HPODE and 10R-HODE were observed as minor products. Moreover, A. fumigatus , A. niger  and A. clavatus  were also capable of forming 8R,11S-diHODE. When the genome of A. nidulans was analysed in 2005 , three different genes with significant homology to the gene encoding 7,8-LDS were identified, and were termed Ppo genes (ppoA, ppoB and ppoC) . Analysis of strains in which each particular allele was deleted indicated that PpoA is responsible for the formation of 8R-HPODE (which is partially reduced to the corresponding hydroxide) and 5S,8R-diHODE . Although the biochemical properties of PpoB remained unclear in this approach (as the deletion mutant formed the same set of oxylipins as the wild-type enzyme), ppoC could be identified as coding for a 10R-dioxygenase (DOX) . Studies with stereospecific deuterated LA showed that 8R-H(P)ODE and 10R-H(P)ODE were formed via initial abstraction of the proS hydrogen from C-8 and antarafacial oxygen insertion either at C-8 or, after double bond migration from position Δ9 to Δ8, at C-10. 5S,8R-DiHODE was formed from 8R-HPODE via a suprafacial mechanism. Subsequent studies in which ppoA and ppoC from A. nidulans [110,111] and ppoC from A. fumigatus , respectively, were expressed heterologously confirmed the proposed catalytic activity of both enzymes (Fig. 4A).
Interestingly, the domain structure of Ppo enzymes appeared to be unique, as two different haem domains are predicted: a fatty acid haem peroxidase/DOX domain in the N-terminal region, and a cytochrome P450 domain in the C-terminal region (Fig. 4B) [109,112]. Indeed, on the basis of sequence alignments and spectroscopic analysis, it was demonstrated that PpoA consists of those different haem domains. Further studies on the reaction mechanism indicated that both domains are essential for two different enzymatic activities: In the N-terminal peroxidase/DOX domain of PpoA, the DOX reaction is catalysed and LA is oxidized to 8R-HPODE, presumably via a tyrosyl radical mechanism. This intermediate product is isomerized in the following reaction in the C-terminal cytochrome P450 domain to 5S,8R-diHODE as the end-product . On the basis of these results, PpoA was assigned as CYP6001A1, the first member of a new cytochrome P450 subfamily.
A re-evaluation of the 7,8-LDS sequence of G. graminis revealed that the full-length protein also contains the cytochrome P450 domain and that the previously annotated sequence based on determination of the amino acid sequence lacked this domain. In hindsight, this might also explain why recombinant 7,8-LDS showed only prominent 8-DOX activity. Consequently, it can be proposed that 7,8-LDS and PpoA belong to the same cytochrome P450 subfamily and share a similar reaction mechanism.
Although a similar cytochrome P450 domain has been predicted for PpoC, an analysis of the amino acid sequence showed that the crucial Cys, which serves as the fifth haem iron ligand in all cytochrome P450 enzymes, is conserved neither in PpoC from A. nidulans nor in that from A. fumigatus. Thus, it was proposed that this domain has no functional activity. Consistently, only a prominent 10-DOX activity was observed that, in analogy to PpoA and 7,8-LDS, takes place in the peroxidase/DOX domain . However, as well as 10R-H(P)ODE as the main product, several further downstream products have been observed to be formed from 10R-HPODE as side products, i.e. 10-oxo-LA, 9Z,10R,12S,13R)-12,13-epoxy-10-hydroxy-9-octadecenoic acid (12,13-epoxy-10-HOME), 10-oxo-decenoic acid, and volatile compounds such as 2-octenal and 1-octen-3-ol [105,111]. The last of these has also been reported to be formed from LA by homogenates of Lentinus decadetes , Psaliotta bispora [114–116] and Trichloma matsutake . However, in contrast to the formation of volatile compounds in Aspergillus strains, it was reported that, in T. matsutake and L. decadetes, the 10-HPODE formed was specifically (99%) in the S-configuration .
PpoA-derived 8-hydroxy and 5,8-dihydroxy fatty acid derivatives were also reported nearly 30 years ago, when Champe et al. demonstrated that these metabolites repress conidiation and induce premature sexual sporulation in A. nidulans [20,21]. Because of their suggested function, these substances were collectively called Psi factors, and were classified according to the fatty acids from which they are derived (LA, Psiα; oleic acid [18:1(n-9)], Psiβ; and α-linolenic acid [18:3(n-3)], Psiγ) and by the number and position of hydroxyl groups at the fatty acid backbone: PsiA, hydroxyl group at C-8 and δ-lactone ring; PsiB, hydroxyl group at C-8 (e.g. 8R-HODE); and PsiC, hydroxyl groups at C-5 and C-8 (e.g. 5S,8R-diHODE). The relative amounts of PsiA, PsiB and PsiC were reported to influence the ratio of sexual to asexual spore formation during development. Whereas the addition of PsiAα (8-HODE-δ-lactone), for instance, stimulated asexual and inhibited sexual sporulation, PsiBα (8R-HODE) and PsiCα (5S,8R-diHODE) showed the opposite effect . Other studies have shown additionally that PsiBα (8R-HODE) has biocontrol activity against the two soil pathogens Rhizoctonia solani and Pythium ultimum, as well as other plant pathogens . The physiological importance of Ppo enzymes for Psi factor production has been evaluated by analysis of the respective knockout mutants: deletion of the ppoA allele led to a significant reduction in the PsiBα (8-HODE) level and, consistent with the results of Champe et al. [20,21], an increased ratio of asexual to sexual spores . Although PpoC was shown not to form Psi factors by its own enzymatic activity [99,111], studies on the deletion mutant indicated an indirect role of this enzyme in the Psi factor production pathway and regulation of fungal development. Deletion of the ppoC allele in A. nidulans decreased the ratio of asexual to sexual spore development, and led to almost complete elimination of psiBβ (8R-HOME) . The respective A. fumigatus mutant, moreover, showed changes in morphology and development as well as increased oxidative stress tolerance and higher susceptibility to phagocytosis and killing by alveolar macrophages . PpoB, for which the enzymatic activity is still unknown, was also reported to be involved in Psi factor production. The deletion of ppoB led, by analogy to ΔppoA strains, to an increased ratio of asexual to sexual spore spores and, by analogy to the ΔppoC strain, to decreased production of PsiBβ (8-HOME) .
As well as the role of Ppo enzymes and their associated products in regulating fungal development, they have also been suggested to influence secondary metabolism . This is mainly illustrated by the overproduction of the antibiotic penicillin and simultaneous elimination of the mycotoxin sterigmatocystin in the ΔppoA/ΔppoC double mutant and the ΔppoA/ΔppoB/ΔppoC triple mutant of A. nidulans . The corresponding A. fumigatus triple mutant showed increased virulence in a murine model of pulmonary aspergillosis . Recently, these functions of Ppo enzymes have been extended to A. flavus, which contains, in addition to ppoA, ppoB and ppoC, a further gene termed ppoD. Studies have shown that these genes are necessary for density-dependent phenomena, and regulate aflatoxin production as well as seed colonization .
Oxylipin-mediated host–fungus crosstalk
As also mentioned in the previous sections, there are several lines of evidence showing that, on the one hand, plant oxylipins can partly substitute for fungal oxylipins [84,86,87], and on the other that fungal oxylipins are involved in influencing processes in infected host tissues, presumably by mimicking endogenous signal molecules [87,124]. On the basis of these findings, it was proposed that oxylipins act as host–fungus communication signals [36,125]. Notably, mimicry of a plant oxylipin by a pathogen metabolite binding to a plant receptor has been observed in a related context before: the plant pathogen Pseudomonas syringae secretes a compound upon infection that shows high structural similarity to the bioactive form of JA, the JA–Ile conjugate . This metabolite, known as coronatin (Fig. 5B), is able to bind to the plant receptor COI1 in a similar way as JA–Ile, and thus, by mimicking the plant hormone, it can manipulate the plant’s defence response . As no other oxylipin receptor has been identified to date, either in plants (except for COI1) or in fungi, the molecular basis of an interaction is still hypothetical and speculative (Fig. 5A, right). Considering that the chemical structures of PsiA and JA–Ile may be regarded as slightly similar (Fig. 5B), one may also assume that fungal Psi factors or the corresponding downstream products are recognized by plants via a coronatin-insensitive (COI)1-like receptor (Fig. 5A, right). Given the fact, however, that the perception by COI1 is highly specific for JA–Ile, and that even the JA and its precursor OPDA are very weak ligands for this receptor , it is immediately clear that this putative receptor should be different from COI1. Further evidence for fungal oxylipins hijacking plant receptors comes from a recent study by Thatcher et al., who showed that coi1 mutants are more resistant to infections with F. oxysporum, whereas JA biosynthesis mutants are more susceptible. Furthermore, the authors proposed that F. oxysporum is able to synthesize and to secrete one or multiple metabolites that induce a COI1–JAZ interaction and thus alter the plant response in a way that benefits pathogen development . Notably, as mentioned in one of the previous sections, several fungal species, including F. oxysporum, have been shown to harbour or secrete JA and its derivatives [48,50,51] – a fact that strengthens the hypothesis of cross-kingdom or host–pathogen-signalling.
On the other hand, it is known from the mammalian field that most eicosanoids are sensed by receptors that belong to the superfamily of GPCR-type receptors . Interestingly, it was observed that the LOX-derived 9S-HODE can bind to the mammalian GPCR G2A , indicating that mammalian cells might also be able to perceive fungal and plant oxylipins via this receptor type (Fig. 5A, left). As 9S-HODE is structurally closely related to its corresponding hydroperoxide derivative, 9S-HPODE, which has been shown to affect fungal sporulation and mycotoxin production [84,86], this finding indicates that GPCRs might also be involved in oxylipin perception on the fungal side. Reports of G-protein/protein kinase A-mediated signal transduction pathways regulating fungal sporulation and mycotoxin production are in line with this observation, and further strengthen this assumption . Consistently, the genomes of filamentous fungi contain a large number of genes with homology to those encoding GPCRs. In Sa. cerevisiae, in addition, three GPCRs are known to play a critical role in mating and filamentous growth, and are important for pheromone and carbon source perception .
We apologize to scientists whose work we may have overlooked. Our work on fungal oxylipins was supported by the German Research Foundation (International Research Training Group 1422, Metal Sites in Biomolecules: Structures, Regulation and Mechanisms). Critical reading of the manuscript by E. Hornung is gratefully acknowledged. We also wish to thank the three anonymous referees for detailed and very constructive suggestions on scientific as well as grammatical aspects of our review.