The COX enzyme exists in at least two isoforms in mammals, COX-1 and COX-2, which are inhibited by conventional non-steroidal anti-inflammatory drugs (NSAID) such as ibuprofen and aspirin, but only one COX isoform is generally present in invertebrates and lower vertebrates [2]. Mammalian COX-1 catalyses the generation of prostaglandins (PG) involved in many basic physiological functions such as regulation of blood pressure, gastric mucosal protection, maintenance of homeostasis, and reproductive and nervous system function; while PGs metabolised by COX-2 are involved in inflammation, ovulation and mitogenesis [3,4]. COX-1 is a constitutive enzyme (i.e. it is constantly expressed) found in the cell membranes of most mammalian tissues, while COX-2 is an inducible enzyme, not found in all cell types, that is located in nuclear membranes [1]. In the mammalian LOX pathway AA is converted by different LOXs into hydroperoxyeicosatetraenoic acids (HPETE) that may be further metabolised into hydroxyeicosatetraenoic acids (HETE) by glutathione peroxidase. Leukotrienes are synthesised from 5-HPETE by 5-LOX, while 8-LOX, 12-LOX and 15-LOX catalyse the production of different lipoxins from 8-HPETE, 12-HPETE and 15-HPETE (Fig. (Fig.11).

Research on eicosanoids has mainly been mammalian-driven and has lately been aimed at designing NSAIDs that are COX-2 selective due to the potential negative side-effects of COX-1 inhibition, which may affect the gastrointestinal tract, heart and kidneys [3]. Eicosanoids act at both the extracellular and intracellular level by interacting with distinct transmembrane G-protein coupled receptors (extra- and intracellular) and nuclear peroxisomal proliferator-activated receptors (PPAR) [5]. Activated G-proteins may, depending on cell type, stimulate second messengers such as cyclic adenosine monophosphate (cAMP) and/or intracellular calcium release [5]. PPAR are transcription factors which also have a role in ligand binding (eicosanoids), so directly influencing the expression of target genes involved in e.g. controlling prenatal and postnatal development [6]. These examples emphasise the biological significance of eicosanoids.

Less is known about eicosanoids in non-mammalian species; however, during the last three decades considerable evidence has been gathered concerning their synthesis and action. Eicosanoids have now been identified in almost every major metazoan phyla including some plants [1]. There is general consensus that eicosanoids act as autocrine or paracrine signallers (also referred to as local hormones) in both vertebrates and invertebrates, where they mainly function as important mediators in reproduction, the immune system and ion transport [1]. It is clear from a number of reports that eicosanoid generation is subject to inhibition by NSAIDs in a wide range of invertebrates [1,2]. Moreover, a COX derived mechanism similar to the mammalian biosynthesis of PGs has been proposed in the coral Plexaura homomalla [7]. There is also evidence of a LOX derived pathway being present in invertebrates based on the work of Ragab and colleagues [8] on the primitive wingless insect, the firebrat Thermobia domestica; although little is known about the structure of the pathway. Overall, invertebrate eicosanoid biosynthesis seems to have a simpler structure than its mammalian counterpart, as seen with the COX pathway [7], but also appears to be split into two instead of three pathways. There is currently no proof of an epoxygenase pathway being present in invertebrates [1].

There is little doubt that disruption of eicosanoid biosynthesis may upset many important physiological functions in both invertebrates and vertebrates, which can have serious consequences for both the individual and the population. Current evidence suggests that the main mode of action of the NSAID ibuprofen in D. magna relates to interruption of eicosanoid biosynthesis which reduces fecundity [9-12]. Eicosanoids may therefore play a pivotal role in daphnid reproduction. A wealth of synthetic and natural chemicals may affect invertebrate reproduction through endocrine disruption with one of the best known examples being imposex (masculinisation) of female molluscs caused by exposure to tributyl tin (TBT) [13]. It is therefore important to understand eicosanoid biosynthesis in Daphnia, and invertebrates in general, to fully recognize the potential mode of action of endocrine disrupters and how they may affect natural invertebrate populations.

Invertebrate and vertebrate COX protein sequences were retrieved from GenBank [19] and Ensembl [20] for phylogenetic analysis. Since no arthropod COX protein sequences were available that could aid the phylogeny with respect to D. pulex, we searched the GenBank Expressed Sequence Tag (EST) division (est_others) using the BLAST algorithm (tblastn) to obtain additional arthropod COX sequences. Sequences that significantly resembled the sea squirt Ciona intestinalis COX amino acid sequence (E-value < 1e-20) were retrieved and included in the dataset. Six ESTs were obtained, representing two malacostracan species (Crustacea): Four ESTs were derived from Homarus americanus (accession numbers DV772953, DV774102, EH401871 and FD699680) and two were derived from Petrolisthes cinctipes (accession numbers FE773225, FE820815). After translating the nucleotide sequences to putative amino acids, it appeared that one H. americanus sequence (FD699680) did not overlap with the other three sequences. The three remaining sequences constituted two slightly deviating amino acid sequences; DV774102 differed from EH401871 and DV772953. The four H. americanus sequences were combined to one sequence with variable and missing positions specified as 'unknown'. Furthermore, the inferred amino acid sequences of the two P. cinctipes ESTs were similar, and P. cinctipes was thus included once in the dataset. Note that these additional crustacean entries were EST derived, single pass sequenced, and therefore not guaranteed to be free from sequencing errors.

Both the COX and LOX pathways in Daphnia appeared to have a simpler structure than their mammalian counterparts (for comparison, see Figs. Figs.11 and and2).2). For instance, there was no bioinformatic evidence of prostacyclin synthase, which converts PGG₂ into PGI₂, in the daphnid COX pathway. The gene encoding this enzyme was likewise not identified in the genome of the urochordate C. intestinalis [2]. Additionally, there was only bioinformatic evidence of one gene encoding COX in the D. pulex genome. A phylogenetic comparison of the predicted D. pulex COX with other protein sequences revealed that daphnid COX clusters with the invertebrates being most closely related to other crustaceans. The COX phylogeny likewise showed that COX-1 and COX-2 comprise two distinct clades amongst the vertebrates (Fig. (Fig.3).3). Generally, it is understood that invertebrates and lower vertebrates only have one non-specific type of COX [2], but recently two COX isoforms have been identified in the corals Plexaura homomalla [7] and Gersemia fruticosa [27]. Rowley et al. [2] suggest that the COX genes found in corals are an early version that predates the (supposed) vertebrate duplication into the typical constitutive COX-1 and inducible COX-2 isozymes found in vertebrates. Only one copy of the COX gene has been identified in the C. intestinalis genome [2] which, as a member of the Phylum Chordata, shares a more recent common ancestor with the vertebrates. This was supported by our phylogenetic analysis (Fig. (Fig.3)3) suggesting that a duplication of the COX gene occurred in the Chordata. Rowley and colleagues further speculate that the evolution of COX-1 and COX-2 probably predates the emergence of bony fish some 350 million years ago [2]. This is supported by the fact that only one constitutively expressed COX isoform has been identified in the shark Squalus acanthias (Spiny dogfish) [2]. In our phylogenetic analysis S. acanthias COX clustered with the vertebrate COX-1 clade.

Like the putative Daphnia COX pathway, the structure of the daphnid LOX pathway also appeared to be more simple than its mammalian counterpart, which confirms previous findings [1]. There was gene sequence evidence that indicated the presence of three leukotrienes (LT) in Daphnia: LTA₄, LTB₄ and 12-oxo LTB₄ (Table 1). This is, however, doubtful since no LTs, nor the precursor 5-HPETE, have been identified in invertebrates and lower vertebrates to date [28]. It is possible that the putative LTA₄ hydrolase (LTA4H), which is a bi-functional enzyme in mammals, having both LTA₄ hydrolase and aminopeptidase activity, only has aminopeptidase activity in daphnids as has been reported in Caenorhabditis elegans [29]. However, recent evidence shows that bi-functionality of LTA4H is only six point mutations away in yeast compared to the mammalian enzyme [30]. There is reason to believe that daphnid LTA4H may be bi-functional, and thus able to convert LTA₄ into LTB₄; a hypothesis that we will test in future experiments. Furthermore, transcriptomic evidence from D. magna shows that the expression of leukotriene B4 12-hydroxydehydrogenase (Ltb4dh) increases with increasing concentration of the eicosanoid-inhibiting drug ibuprofen [11]. This does not prove that LTs are present in daphnids, but yet again we cannot rule out the possibility. The enzyme (LTB4DH) encoded by Ltb4dh is also bi-functional in mammals regulating eicosanoids by rapidly degrading different E and F series PGs and LTB₄ [31] (see below). The former function of LTB4DH (PG catabolism) appears to be the most likely in daphnids until solid proof exits about the presence of LTs (Fig. (Fig.22).

The presence of lipoxins is, however, more likely, and bioinformatic information from D. pulex suggests that at least two LOX enzymes are present (Table 1). LOX enzymes have been found in all organisms studied, from bacteria to man. The two putative D. pulex LOXs are both composed by two domains; an N-terminal lipase domain belonging to the InterPro protein family 734 (IPR000734) and a C-terminal LOX LH2 domain (IPR001024). This resembles mammalian LOX enzymes that are also comprised of two domains; a regulatory N-terminal domain that is similar to mammalian lipases and a catalytic LOX domain (C-terminal) [32]. LOX LH2 is the only LOX related domain that has been identified in the D. pulex genome. Other known LOX domains include e.g. mammalian LOX (IPR001885) and LOX, C-terminal (IPR013819). There are 13 D. pulex gene models that contain the LOX LH2 domain [14], but only two genes contain both an N-terminal domain similar to mammalian lipases (IPR000724) and a C-terminal LOX domain (Table 1). Further investigations are needed to specify what type of LOX enzyme these two Daphnia sequences represent prior to analysing their phylogenetic relationship. However, it is likely that they could be 8-LOX and 12-LOX, which synthesise different lipoxins from 8-HPETE and 12-HPETE, as these LOXs have been identified in a range of invertebrate species [8,33,34].

Two enzymes, 15-hydroxyprostaglandin dehydrogenase (PGDH) and LTB4DH are known to irreversibly inactivate bioactive eicosanoids in mammals. Both enzymes are key in regulating the hormonal-like action of eicosanoids by rapidly degrading PGE₂, PGF_2α, and LTB₄, as overproduction of these potent mediators may have serious physiological effects such as initiating inflammation [31]. It appeared that LTB4DH fulfils this regulatory function single-handedly in daphnids, as there was no indication of PGDH being present (Table 1, Fig. Fig.22).

The bioinformatic and transcriptomic evidence from D. pulex and D. magna (Fig. (Fig.2)2) suggests that PGs (e.g. PGH₂, PGE₂ and PGF_2α), lipoxins and possibly LTs could be present in daphnids. Low similarity of TXA₂ synthase and PGD₂ synthase to ortholog proteins from other genomes (Fig. (Fig.2)2) render the presence of PGD₂ and TXA₂ to be less certain in daphnids, although this could merely be due to daphnids having more divergent versions of these proteins. Nevertheless, it seems most likely that daphnids do not produce TXA₂ since the gene encoding TXA₂ synthase has not been identified in the genome of C. intestinalis [2]. The LOX encoding genes identified in D. pulex were also slightly doubtful due to the same reasons, but it would be more probable that these enzymes are present in Daphnia as both 8-LOX and 12-LOX derived lipoxins are common in invertebrates [8,33,34]. PGA₂ may also be present in daphnids as it is non-enzymatically rearranged from PGE₂ and has been detected in several arthropods [1]; but until verified by mass spectrometry or the like it remains speculative what eicosanoids are present in Daphnia. Moreover, the annotation of genes from the daphnid eicosanoid biosynthesis (and other daphnid genes for that matter) should improve as more invertebrate genomes become sequenced and annotated.