Specificity of Lipoxygenase Pathways Supports Species Delineation in the Marine Diatom Genus Pseudo-nitzschia

Oxylipins are low-molecular weight secondary metabolites derived from the incorporation of oxygen into the carbon chains of polyunsaturated fatty acids (PUFAs). Oxylipins are produced in many prokaryotic and eukaryotic lineages where they are involved in a broad spectrum of actions spanning from stress and defense responses, regulation of growth and development, signaling, and innate immunity. We explored the diversity in oxylipin patterns in the marine planktonic diatom Pseudo-nitzschia. This genus includes several species only distinguishable with the aid of molecular markers. Oxylipin profiles of cultured strains were obtained by reverse phase column on a liquid chromatograph equipped with UV photodiode detector and q-ToF mass spectrometer. Lipoxygenase compounds were mapped on phylogenies of the genus Pseudo-nitzschia inferred from the nuclear encoded hyper-variable region of the LSU rDNA and the plastid encoded rbcL. Results showed that the genus Pseudo-nitzschia exhibits a rich and varied lipoxygenase metabolism of eicosapentaenoic acid (EPA), with a high level of specificity for oxylipin markers that generally corroborated the genotypic delineation, even among genetically closely related cryptic species. These results suggest that oxylipin profiles constitute additional identification tools for Pseudo-nitzschia species providing a functional support to species delineation obtained with molecular markers and morphological traits. The exploration of the diversity, patterns and plasticity of oxylipin production across diatom species and genera will also provide insights on the ecological functions of these secondary metabolites and on the selective pressures driving their diversification.

The planktonic pennate diatom genus Pseudo-nitzschia is encountered in marine environments all over the world and includes ca. 30 taxonomically described species [33]. Pseudonitzschia cells are needle-shaped and form stepped-chains by aligning tip-to-tip. Species identification is based on cell shape and width, cell overlapping pattern in the chain, and ultrastructural features of the siliceous frustule [34]. Several groups of morphologically highly similar or even identical species occur, which can only be told apart using rapidly evolving DNA-markers such as the internal transcribed spacer region (ITS) and the hyper-variable domain (D1-D3) of the large subunit (LSU) in the nuclear ribosomal RNA cistron [33]. For instance, the morpho-species P. delicatissima consists of at least four genetically distinct species: P. delicatissima 'sensu stricto', P. arenysensis, P. decipiens, and P. dolorosa [35], [36] [37]. Many of these species occur in sympatry in our study area, the Gulf of Naples in the Mediterranean Sea [37].
Recently we described a novel protocol for rapid characterization of oxylipins in lipid extracts of diatoms by reverse phase liquid chromatography and tandem quadrupole-ToF mass spectrometry (qToF LC-MS/MS) [21]. In addition to traditional metabolite profiling, the method permits to deduce the positional specificity of lipoxygenation by MS/MS fragmentation of hydroxy-epoxy PUFA derivatives, generically named epoxy-alcohols, which are among the major classes of oxylipins so far reported in diatoms. Thus, the qToF LC-MS/MS analysis provides a powerful tool to investigate distribution of LOX activities among different diatom samples. The aims of the present study were to assess: i) the spectrum of oxylipins produced by different Pseudo-nitzschia species, i.e., if one or more compounds are produced per monoclonal strain; ii) if oxylipin composition changes among strains within the same genetically defined species; and iii) if a match exists between phylogenetic relationship and LOX profiles, i.e., if closely related species produce the same compounds and more distantly related species different ones, and in addition, if phylogenetically closely related Pseudo-nitzschia species produce more similar oxylipins than more distantly related species. Strains were genetically identified based on the hypervariable region (ca. 700 base pairs) at the 5'-end of their nuclear-encoded LSU rDNA and on the plastid-encoded large subunit of their ribulose-1,5-bisphosphate carboxylase/ oxygenase (rbcL). These markers have been shown to differ among biologically distinct species in Pseudo-nitzschia [31], [38].

Diatom cultures
Pseudo-nitzschia strains were isolated at the Long Term Ecological Research Station MareChiara in the Gulf of Naples (Tyrrhenian Sea, Mediterranean Sea) (Table S1). Pseudonitzschia multiseries strain NWFSC 316 was isolated in Clam Bay, Puget Sound, USA (Table S1). Clonal cultures were established by isolating single cells or short chains from phytoplankton net samples collected in the surface layer of the water column. Cultures were grown in sterile filtered oligotrophic seawater amended with f/2 nutrients [39] and maintained at a temperature of 20°C, a 14: 10h light: dark cycle, and with a photon flux of 100 μmol photons m -2 s -1 .

Molecular characterization
A 25 ml subsample was collected from each culture in exponential growth phase and genomic DNA was extracted following [40]. DNA was amplified by PCR using the primers D1R and D3Ca [41], [42] for the hypervariable domains D1 and D3 of the LSU rDNA and two degenerated primers, rbcL1 and rbcL7 [37] for the rbcL. PCR conditions and sequence analyses were carried out as reported in [37]. The accession number of the sequences produced within this investigation is reported in Table S1. Additional LSU rDNA and rbcL sequences of other Pseudo-nitzschia species were downloaded from GenBank, aligned by eyeball and used to build Maximum Likelihood phylogenies. Strain codes and GenBank accession numbers are reported in Table S1. Maximum-likelihood trees were inferred based on the substitution models selected through the Akaike information criterion option implemented in jModeltest [43] using MEGA5 [44], with 100 bootstrap replicates. Best fit models were GTR + I + G (I= 0.59; γ = 0.2266) for rbcL and GTR + I + G (I= 0.69; γ = 0.2180) for LSU.

Chemical analyses
To obtain the biomass required for chemical analyses, cultures were grown in Erlenmeyer flasks filled with 1 L f/2 medium at the conditions illustrated above. Cell concentration was monitored over time by collecting a 2-ml subsample that was used for estimating cell numbers with a Sedgewick-Rafter counting chamber utilizing a Zeiss Axiophot microscope (Karl Zeiss, Oberkochen, Germany). Cells were harvested at the stationary phase by centrifugation at 1200 g for 10 min at 5°C. The cell pellets were collected in 50 ml Falcon tubes, immediately frozen in liquid nitrogen, and kept at -80°C until chemical analysis was performed. Each strain was treated as an independent entry in the chemical analysis, irrespective of its taxonomic assignment; some strains have been grown and analysed on different occasions to test for constancy of their oxylipin pattern (Table S1). When possible, replicates of the chemical analyses were also carried out.
Cell pellets were dissolved in distilled water (1 ml·g -1 of wet pellet) and sonicated for 30 sec at 4°C prior to extraction with methanol (equal volume with water). The internal standard, 16hydroxyhexadecanoic acid (15 μg·g -1 of pellet) was added before the extraction, and the resulting suspension was centrifuged at 2000 g for 5 min at 5°C. The organic extracts were dried under reduced pressure with a rotary evaporator (Buchi, Rotavapor R-200) and then methylated with ethereal diazomethane in diethyl ether (0.4 ml per 10 mg extract) for 1 hr at room temperature. After removing the organic solvent under nitrogen stream, the raw extract was used for analysis of non-volatile oxylipins by a Waters Alliance liquid chromatography equipped with UV photodiode detector (PDA) and Micromass microq-ToF (qToF) mass spectrometer (Waters, Milford, MA, USA) as described in [21]. Identification of individual products was supported by matching of primary analytical indicators (PAIs), (retention time, UV spectrum, molecular weight and mass/mass fragmentation) with those experimentally determined (standards) or calculated [21]. Determination of lipoxygenase positional specificity is based on mass/mass analysis of epoxy-alcohols. Oxylipin levels were established by normalizing the component peak area by the peak area of the internal standard (16-hidroxyhexadecaoic acid methyl ester).

Molecular phylogeny
Maximum likelihood inference of the rbcL sequences provided a well-resolved phylogeny ( Figure 1A). Pseudonitszchia turgiduloides was recovered at a basal position and the remainder Pseudo-nitzschia species grouped into three main clades. One of these included all P. fraudulenta strains, the second one contained a clade with strains of P. multistriata as sister to a clade with strains of P. multiseries and P. pungens, and a third one contained a clade with stains of P. galaxiae as sister to a polytomy comprising species specific clades with strains of P. pseudodelicatissima and all species of the P. delicatissima complex (P. delicatissima, P. dolorosa, P. arenysensis). Within this third clade, strains SZN-B507 and SZN-B509 grouped together. The two strains showed morphological and ultrastructural characters very similar to those of P. delicatissima (D. Sarno, personal communication) and their sequences did not match with any sequence deposited in GenBank. We thus considered these strains as a new species, and provisionally designated it as P. cf. delicatissima new genotype (D. Sarno, MV Ruggiero and collaborators, manuscript in preparation). The rbcL marker separated P. pseudodelicatissima from P. cuspidata. Intraspecific genetic differences were evident among strains of P. arenysensis and P. pseudodelicatissima in the form of single nucleotide differences. Remarkably large genetic diversity was instead evident within the P. galaxiae clade.
The LSU rDNA tree ( Figure 1B) revealed the same terminal clades as the rbcL tree, with the following exceptions; the marker did not discriminate P. pseudodelicatissima from P. cuspidata and did not show differences amongst P. arenysensis strains. In addition, Fragilariopsis cylindrus grouped with P. turgiduloides, though with a low bootstrap support. Relationships among the terminal clades were less well resolved than in the rbcL tree as shown by low or lacking bootstrap support and by the recovery of polytomies. Nonetheless, the topologies between the two trees showed no conflicts. We did not obtain rbcL sequences for the two P. galaxiae strains analysed for oxylipin composition. However, we included in the phylogenetic analysis the rbcL sequences of two other P. galaxiae strains whose LSU sequences were identical to LSU sequences of the analysed strains (see Table  S1). The rbcL sequences of those P. galaxiae strains showed a remarkably higher level of nucleotidic divergence with respect to LSU.

Species specificity of LOX activity
The results of the present study show that the genus Pseudo-nitzschia exhibits a rich and varied LOX metabolism of eicosapentaenoic acid (EPA). As shown in Figure 4 that reports occurrence and relative abundance of individual oxylipin in the diatom samples undergone to qToF LC-MS/MS analysis, LOX positional specificity is suggestive of at least four chemotypes corresponding to 15-and 12-LOX in P. arenysensis, 5-LOX in P. delicatissima, 8-LOX in P. pseudodelicatissima and 14-LOX in P. fraudulenta and P. multistriata. Other possible chemotypes are associated to the presence of the unidentified compounds of group 1 and 2 (Mk353#1 and Mk353#2).
The species-specificity of these chemotypes was generally in good agreement with the molecular phylogenies based on rbcl and LSU, though not in all cases. For instance, the strains belonging to the distinct species P. fraudulenta and P. multistriata form a grade in the rbcL tree. Yet, they share 14-LOX, rendering the production of this compound probably a shared ancestral feature. Moreover, the fact that compound Mk353#2 is shared amongst different, not particularly closely related, species, suggests that the ability to produce this compound has been acquired at least twice, and possibly three times independently, or it has been acquired once and, subsequently, has gone lost in several species. In addition, each species produces just one or a few oxylipins. A notable exception is P. galaxiae, for which only traces of undefined oxygenated derivatives of fatty acids were detected. Our results support those of previous studies [26,45] that Pseudonitzschia lacks a distinctive trait of other planktonic diatoms, namely the LOX oxidation of C16 polyunsaturated fatty acids [46], [47], [48], [49], [45], [25].
Two LOX pathways appear to co-exist in P. arenysensis. Grouping of strains according to the minor rbcL differences amongst them did not corroborate a grouping according to 12and 15-LOX products. Since LOX activity in diatoms seems to be modulated along the growth phase when maintained in culture [26], differences in LOX patterns could be due to physiological differences among the strains at the time of harvesting. Alternatively, this strain differentiation could be the result of genetic variation of LOX enzyme. In fact, single mutations of the peptide chain has been reported to induce modification of the fatty acid binding pocket, converting a 15lipoxygenating enzyme into a 12-lipoxygenating enzyme and vice versa [50] [51] [52] (for a general discussion see [53]). The observation that some strains possess both compounds and others just one of these could then be explained by the fact that diatoms are diploid and that some of the strains are 12-LOX or 15-LOX homozygous and others 12-LOX + 15-LOX heterozygous.
Also P. pseudodelicatissima strains show minor differences in the rbcL sequences. Notably, the strain SZN-B317 that lacks 8-HEPE and 10,8-HEpETE also differs in the rbcL signature from the three strains that possess 8-LOX activity. In addition to the products reported in Figure 2, some of the species show compounds that were not conclusively characterized, but correspond to two sets of undisclosed LOX products (MK353#1 and MK353#2) with identical molecular weight. These products characterize all species of the apical rbcL clade grouping P. pseudodelicatissima, P, cf. delicatissima -new genotype, and P. delicatissima, but appear to have gone lost in P. arenysensis. Interestingly, the compound MK353#2 also characterizes P. multiseries, which is -according to the rbcL tree -not the sister of the abovementioned species. Previous studies on P. multiseries have established the presence of C20-derived lipoxygenase products, namely bacillariolides, that can be structurally related to the uncharacterized molecules reported here ( [54] [55]).

Oxylipins as identification markers of Pseudo-nitzschia species
Traditionally, identification and delineation of unicellular microalgal species is based on phenotypic features such as cell size and shape and fine ultrastructural characters of their cell wall. Nucleotide markers have increased the precision of this identification process and have uncovered numerous cases of several genetically distinct lineages existing within single phenotypically delineated species (e.g. [56]). Also in diatoms, morphological differentiation has not held equal pace with differentiation into genetically and biologically distinct species, as shown by the discovery of numerous cases of genetically and biologically distinct species that are either morphologically indistinguishable (cryptic species) or can be separated only by minor ultrastructural details of their siliceous frustule (pseudo-cryptic species) (e.g. [57], [37] [58], [38],). Differences in lipoxygenase products among a series of strains within Pseudo-nitzschia corroborate their genotypic delineation, even among genetically closely related cryptic species within the P. delicatissima complex, suggesting that these products can aid identification of species more precisely than morphological characters. This is particularly relevant when the presence of different epoxy-alcohols allows the rational deduction of the enzymatic dioxygenation site of the EPA polyunsaturated chain and, consequently, the classification according to the positional specificity of the putative LOXs. The three species within the complex (P. arenysensis, P, delicatissima, P, cf. delicatissima new genotype) are indistinguishable in light microscopy and present only minor morphometric differences in the density of the poroids on their siliceous frustule visible in EM ( [37], D Sarno, MV Ruggiero and co-authors, manuscript in preparation for P, cf. delicatissima new genotype). The fact that closely related species show different LOX positional specificity seems to confirm that genetic changes accumulate rapidly in the enzymes of this pathway in Pseudo-nitzschia.
Several studies tested the Biological Species Concept sensu [59] in diatoms by carrying out mating experiments between sympatric strains of the same species as defined with morphological and molecular characters (e.g. [60], [37], [61]). The approach has been deployed also to delineate several of the Pseudo-nitzschia species included in the present investigation [37]. The corroboration of oxylipin distribution, LOX positional specificity and species assignment of the strains used in the present study suggests that LC-MS/MS profiling constitutes a promising support for molecular identification of Pseudo-nitzschia species. However, more data, including analyses of strains from other geographic regions, need to be screened for their oxylipin profiles to test if these profiles are species specific.
Secondary metabolites have been used successfully as markers for species delimitation in fungi [62] [63] or in ciliates [64]. Chemotaxonomic approaches in fungi are based both on the use of one specific metabolite or on metabolite profiles that allow the identification of distinct chemotypes. At times, chemotypes matched with groups based on sequence data, at times not, giving support to the hypothesis that chemical profiling is useful for species recognition, but not for inferring phylogenies [63]. Oxylipin profiling and related MS/MS expansion can assist in the species identification of Pseudonitzschia strains, but they cannot be used as sole taxonomic identification tool, as is illustrated by the different LOX activities among the P. arenysensis strains and among the P. pseudodelicatissima strains as well as by the sharing of the same LOX profiles among different species. Clearly, no single marker constitutes the univocal criterion for species recognition, as markers may not accurately reflect species diversity. Different nucleotide markers can provide conflicting evidence about species boundaries [65]. Moreover, the very process of speciation implies an interim phase during which species delineation is, by definition, impossible. A way forward to strive for accurate species delineation -and to identify cases where that is impossible -is to deploy different and complementary types of identifiers. Examples of this approach provide promising results. Sympatric pseudo-cryptic species have been detected in the benthic estuarine diatom Navicula phyllephyta based on morphological and molecular investigations; these species showed different growth responses along a salinity gradient, thus suggesting that they occupy distinct ecological niches [66]. Also the planktonic diatom morpho-species Chaetoceros socialis conceals pseudocryptic genotypes with apparently different geographic ranges [67]. Qualitative and quantitative differences of the metabolomic profiles of the two genotypes suggest that they evolved distinct adaptive capabilities to the contrasting temperature characteristics of their environment [68].
The fact that Pseudo-nitzschia species produce one or two isomeric forms of oxylipins and not a broad spectrum of these compounds suggests the existence of specific metabolic pathway to produce them. The distinct oxylipin profiles among closely related species indicate that the profiles change rapidly and suggest that selective pressure drives the change. The nature of such a selective pressure depends on the roles of these compounds and on their specific targets [69]. Diatom oxylipins may function as species-specific mediators of bloom control [24], [26]. Species-specific chemical communication requires species-specific compounds and species-specific receptors. Hence, the receptors need to evolve in concert with the compounds. In case the oxylipins function as signaling tools amongst specimens within the same Pseudo-nitzschia species then different species producing the same compounds should not co-occur, or they should produce additional compounds that, together, function as species-specific communication. Moreover, the observation that strains belonging to the same species produce different LOX-profiles suggests that some compounds could serve for purposes other than intraspecific communication. In fact, in addition or alternatively, LOX pathways may play a key role in producing defence chemicals in the arms race against grazers [70]. Such a function favours rapid changes in the types of oxylipins produced to outwit the copepods' detoxification systems. The ensuing arms race is then expected to result in differences in LOX products among closely related species, or even between geographical populations of the same species. If the compounds serve as grazer deterrent, then intraspecific differences and interspecific identity do not pose any problem. Table S1. Pseudo-nitzschia strains analysed for oxylipin production. Species name and strain code, date in which the chemical characterization was carried out, LSU rDNA GenBank accession number, rbcL GenBank accession number. When LSU and rbcL sequences were identical to sequences already deposited in GenBank, the reference number of the deposited sequences is provided. (DOCX)