Silencing of a Pseudo-nitzschia arenysensis lipoxygenase transcript leads to reduced oxylipin production and impaired growth
Summary
- Because of their importance as chemical mediators, the presence of a rich and varied family of lipoxygenase (LOX) products, collectively named oxylipins, has been investigated thoroughly in diatoms, and the involvement of these products in important processes such as bloom regulation has been postulated. Nevertheless, little information is available on the enzymes and pathways operating in these protists.
- Exploiting transcriptome data, we identified and characterized a LOX gene, PaLOX, in Pseudo-nitzschia arenysensis, a marine diatom known to produce different species of oxylipins by stereo- and regio-selective oxidation of eicosapentaenoic acid (EPA) at C12 and C15.
- PaLOX RNA interference correlated with a decrease of the lipid-peroxidizing activity and oxylipin synthesis, as well as with a reduction of growth of P. arenysensis. In addition, sequence analysis and structure models of the C-terminal part of the predicted protein closely fitted with the data for established LOXs from other organisms.
- The presence in the genome of a single LOX gene, whose downregulation impairs both 12- and 15-oxylipins synthesis, together with the in silico 3D protein modelling suggest that PaLOX encodes for a 12/15S-LOX with a dual specificity, and provides additional support to the correlation between cell growth and oxylipin biosynthesis in diatoms.
Introduction
The unicellular eukaryotic group of diatoms accounts for 20% of the primary production on Earth and is at the basis of aquatic food webs, factors that make it an ecologically very important group (Nelson et al., 1995; Ban et al., 1999). Growth, defense and death in these organisms are under the control of sophisticated cellular mechanisms that include synthesis of oxylipins (Miralto et al., 1999; Ianora et al., 2003, 2006, 2011, 2015; Fontana et al., 2007b; Watson et al., 2009; Ianora & Miralto, 2010; Barreiro et al., 2011; Ruocco et al., 2020, 2016; Lauritano et al., 2016; Russo et al., 2018, 2020), important signal transduction molecules regulating a series of events correlated to physiological and pathological processes in different organisms (Blée, 2002; Howe & Schilmiller, 2002; Pohnert & Boland, 2002; Farmer et al., 2003; Noverr et al., 2003; Andreou et al., 2009; Gao & Kolomiets, 2009; Dave & Graham, 2012; Romano et al., 2013; Tourdot et al., 2014; Martínez & Campos-Gómez, 2016; Lim et al., 2017; Yeung et al., 2017; Deboever et al., 2020).
Oxylipins are oxygenated derivatives of polyunsaturated fatty acid (PUFA). Their biosynthesis starts by action of lipoxygenase enzymes (LOX) that catalyze regio- and stereo-specific addition of oxygen with formation of hydroperoxide fatty acids (Brash, 1999; d'Ippolito et al., 2003, 2005; Lion et al., 2006; Andreou & Feussner, 2009; Andreou et al., 2009; Mosblech et al., 2009; Bonaventure, 2014). These latter products are further transformed into a large variety of chemical products, through alternative and subsequent reactions. Despite a specific signature across genera and species, this family of metabolites is ubiquitously distributed in nature, being found in animals, plants, bacteria, mosses and algae (Hamberg, 1999; Stumpe & Feussner, 2006; Lee et al., 2008). In mammals, eicosanoid-derived oxylipins were demonstrated to have critical roles in monocyte lineage recruitment and in the inflammatory resolution (Serhan & Chiang, 2008; Gilroy et al., 2016), in obesity (Klein-Wieringa et al., 2013), in homeostasis, thrombosis and diabetes (Tourdot et al., 2014), and in platelet function (Yeung et al., 2017). In plants, oxylipins of linoleic (C18:2 ω-6) and linolenic (C18:3 ω-3) acids are involved in resistance to environmental stress and defense against pathogens and herbivore attack; some of the LOX products, such as jasmonic acid and short-chain aldehydes, act as signalling molecules in plant–plant communication (Turner et al., 2002; Farmer et al., 2003; Dave & Graham, 2012; Dar et al., 2015). Likewise, by being continuously challenged by potentially pathogenic organisms and ecological changes, macroalgal oxylipins may help to control interactions with other organisms and with the environment, promoting algae survival (Bouarab et al., 2004; Lion et al., 2006; Weinberger et al., 2011; Barbosa et al., 2016).
Diatom oxylipins include linear oxygenated fatty acids (LOFAs), previously named as nonvolatile oxylipins, and poly-unsaturated aldehydes (PUAs), both deriving from eicosanoid, mostly eicosapentaenoic acid (EPA, C20:5 ω-3) (Cutignano et al., 2011; d’Ippolito et al., 2018). Both LOFAs and PUAs have never been detected in intact, healthy diatoms (Fontana et al., 2007a), as their biosynthesis starts within seconds after cell breakage and release of free PUFAs from membrane lipids by lipolytic enzymes (Pohnert, 2002; d'Ippolito et al., 2004, 2006; Cutignano et al., 2006, 2011; Adelfi et al., 2019). Thalassiosira rotula and a few species of Skeletonema catalyze the identical reaction in converting very unusual ω-4 and ω-1 C16-PUFAs to hydroxy-, keto- and epoxy-alcohol derivatives (d'Ippolito et al., 2003, 2005, 2006), whereas metabolism of docosahexaenoic acid (DHA) has been reported in Leptocylindraceae (Nanjappa et al., 2014).
Like PUAs that induce abortions or congenital malformations in copepods, the main herbivores of zooplankton (Miralto et al., 1999; Ianora et al., 2003), LOFAs of diatoms are supposed to carry out a number of ecophysiological functions including reduction of grazing pressure and chemical communication, that may contribute to regulate phytoplankton dynamics (Fontana et al., 2007b; d'Ippolito et al., 2009; Barreiro et al., 2011) and the interactions with bacteria (Meyer et al., 2018). Recently, the first qualitative and quantitative analysis of diatom oxylipins from natural phytoplankton communities in the Gulf of Naples has demonstrated that oxylipin-per-cell production is inversely related to diatom density, validating this trend at a global scale through exploitation of the Tara Oceans data (Russo et al., 2020).
LOXs are a group of nonheme iron-containing dioxygenases that are encoded by a multigenic family ubiquitous in plants and mammals but also described in fungi, invertebrates, fish, algae, bacteria and microalgae (Ivanov et al., 2010; Chen et al., 2015; Zhu et al., 2015; Teng et al., 2017; Sugio et al., 2018; Djian et al., 2019). LOXs with manganese instead of iron in the active site also have been described in fungi (Oliw et al., 2011). Although the reaction of dioxygenation is very specific in the synthesis of fatty acid hydroperoxides, there is a considerable variety of structures that can be produced by different LOXs according to the type of PUFA used as substrate, the positional specificity of oxygen (O2) addition and the stereochemistry of the new bond between O2 and the carbon (C) atom of the substrate (Brash, 1999). In diatoms, an additional level of complexity is added by the variability of the LOX substrates and the extreme species or even strain specificity (Wang et al., 1993; Zheng & Shimizu, 1997; Fontana et al., 2007a,b; Gerecht et al., 2011; Lamari et al., 2013; Nanjappa et al., 2014; d'Ippolito et al., 2018; Ruocco et al., 2020).
Among diatoms, the genus Pseudo-nitzschia comprises around 52 planktonic species with a cosmopolitan distribution, including 26 species capable of producing the toxin domoic acid and, therefore, responsible for harmful algal blooms (Bates et al., 2018). Here we aimed at the functional and in silico characterization of the LOX(s) of Pseudo-nitzschia arenysensis (formerly P. delicatissima), a diatom known to synthesize hydroxy-eicosapentaenoic acids (HEPEs) and hydroxy-epoxy eicosatetraenoic acids (HEpETEs) by 12- and 15S-LOX pathways (Fontana et al., 2007b; d'Ippolito et al., 2009; Lamari et al., 2013). Biochemical and molecular knowledge of secondary enzymes of diatoms is still very limited, and thus we took advantage of information from the Marine Microbial Eukaryote Transcriptome Sequencing Project (MMETSP) (Keeling et al., 2014) to attempt identification and characterization of putative LOX encoding genes. Together, the identification of a single PaLOX candidate, its RNA silencing and the modelling of the encoded protein, provide cues to its functional role. This work also represents the first RNA silencing in a nonmodel diatom species.
Materials and Methods
Lipoxygenase enzymes and RNAi machinery gene identification
The PaLOX sequence was identified in the transcriptome of Pseudo-nitzschia arenysensis strain B593 available in the Marine Microbial Eukaryote Transcriptome Sequencing Project database (MMETSP) (at iMicrobe, https://www.imicrobe.us/#/samples/1815) (Keeling et al., 2014). The term ‘lipoxygenase’ was used to query the swissprot annotation table of the transcriptome and the transcript ID of the LOX-related function (MMETSP0329-20121206|11370) was used to filter the corresponding nucleotide and peptide sequences from the CDS.fa and the pep.fa files, respectively. A tBlastn search against the CDS.fa using the LOX protein sequence obtained from the previous step as query retrieved no other significant similarities.
The RNAi machinery genes of P. arenysensis were identified with a tBlastn search using the Arabidopsis thaliana protein sequences of Dicer-like 1 (DCL1, ID: NP_001184881.1), Argonaute3 (AGO3, ID: NP_001322306.1) and DEAD-box ATP-dependent RNA helicase 20 (RH20, ID: NP_175911.1) as queries for the similarity search. These alignments led to the identification of transcripts encoding for a protein with two RNase III domains (sequence ID: MMETSP0329-20121206|1584; E-value: 1e−16; identity: 28.81%), for a putative Argonaute protein (sequence ID: MMETSP0329-20121206|10545; E-value: 6e−8; identity: 22.70%) and for a DEAD/HELICASEc domain (sequence ID: MMETSP0329-20121206|10979; E-value: 7.39e−174; identity: 58.1%).
Cell cultures and growth curve
The VS3, SV6 and SV4 strains of P. arenysensis were obtained from crosses performed in the laboratory from strains isolated at the Mare Chiara LTER station in the Gulf of Naples. Cultures were grown in seawater enriched with F/2 nutrient (Guillard, 1975) incubated at 18°C under white light at a c. 60 μmol mol−1 and 12 h : 12 h, dark : light photoperiod. The cultures always were collected during their exponential growth phase, unless indicated otherwise.
For the growth curves, fresh diatom cultures were inoculated at a start cell density of 1 × 104 cells ml−1, grown under the abovementioned conditions and counted daily by observation of a Malassez chamber under an inverted microscope. Each curve was performed in triplicate.
Full-length cDNA cloning and sequence analysis of PaLOX
The total RNA was extracted as described previously (Amato et al., 2018) and then reverse-transcribed with the QuantiTect Reverse Transcription Kit (Qiagen) using standard methods. For the amplification of the PaLOX cDNA, two specific primers, PaLOX-F1 and PaLOX-R5 (Supporting Information Table S1), were designed for PCR amplification with the Q5® High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA). The DNA fragment with the desired size (2220 bp) was obtained and, thereafter, subcloned into the vector pCR™ 2.1-TOPO® using the TOPO-TA cloning Kit (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced.
For characterization of the gene introns/exons organization, the primers were designed to amplify regions from 350 to 700 bp in size (Table S1) with MyTaq™ Mix (Bioline, London, UK). Amplified PCR products were analyzed by agarose gel electrophoresis, purified by GenElute Gel Extraction Kit (Sigma-Aldrich) and sequenced. For this analysis, the genomic DNA extractions were performed as described previously (Barra et al., 2014). The PaLOX sequence has been deposited in GenBank with accession no. MW752160.
Identification of transit peptide in the PaLOX protein
The identification of the PaLOX signal peptide was performed via the prediction programs SignalP (TAA-FL, SignalP 4.1 Server, http://www.cbs.dtu.dk/services/SignalP) (Emanuelsson et al., 2007), ASAFind (Gruber et al., 2015) (http://rocaplab.ocean.washington.edu/tools/asafind) and ChloroP (http://www.cbs.dtu.dk/services/ChloroP/) (Emanuelsson et al., 1999).
Quantitative polymerase chain reaction
PaLOX gene expression was evaluated by quantitative polymerase chain reaction (qPCR), performed as reported previously (Adelfi et al., 2014), using the primers listed in the Table S1.
In order to analyze the expression levels of PaLOX relative to the reference genes, the Relative Expression Software Tool (Rest) was used (Pfaffl et al., 2002). Relative expression ratios above two-fold were considered significant. Statistical analysis was performed using the Pair Wise Fixed Reallocation Randomization test by Rest and Prism v.4.00 statistical software (GraphPad Software Inc., San Diego, CA, USA).
FOX2 assay
The measurement of hydroperoxide production was performed on sonicated cell pellets using the FOX2 method as described previously (Orefice et al., 2015). The determination of hydroperoxide products was set at 10 min after sonication, when the kinetic of hydroperoxide production reached the plateau. Results then were expressed as product amount per minute and normalized for protein concentration as measured according to the method of Bradford (1976) following the manufacturer's instructions (Applichem, Glenview, IL, USA) with bovine serum albumin (BSA) used as standard. Statistical analyses were carried out with Prism v.4.00 (GraphPad Software).
Silencing vectors
The constructs for the PaLOX RNA silencing were obtained using PmH4pShBle (Sabatino et al., 2015) as backbone. A 180-bp fragment of the PaLOX transcript, corresponding to the PaLOX gene sequence from 1807 to 1986 bp, was amplified by PCR from the P. arenysensis cDNA using the primers LOXNotI_F and LOXStuI_R (Table S1); next, this fragment was digested with NotI and StuI and subsequently cloned in the antisense orientation, between the Sh ble gene and the FcpA terminator, into the PmH4pShble vector (Fig. S1).
Transformation of P. arenysensis and transformed cells screening
The transformation of P. arenysensis strains was performed by microparticle bombardment using the Biolistic PDS-1000/HE Particle Delivery System (Bio-Rad), as described previously (Sabatino et al., 2015). Putative transformants were obtained after c. 12 d incubation and were screened as described previously (Sabatino et al., 2015).
Protein extraction and Western blot analysis
Cell extracts were prepared by incubating the cell pellets in 30 μl of lysis buffer (Tris–HCl pH 6.8, 50 mM SDS 2%). Proteins were measured by BCA assay protocol (Thermo Fisher Scientific) and then denatured adding Laemmli buffer ×6 and boiling at 100°C for 5 min. Forty micrograms of protein extract from wild-type (WT) and transformed clones were resolved on 10% SDS-PAGE gels and transferred on Immobilon-P PVDF membrane (Millipore) at 100 V for 1 h and 15 min in a cold room. Membrane was blocked for 1 h in a solution of 1 × TBS, 0.1% Tween-20, 5% powdered milk at room temperature, and incubated with a polyclonal anti-LOX primary antibody (Primm), diluted 1 : 1500, overnight at 4°C. The primary antibody was detected by an anti-Rabbit secondary antibody (Promega) for 1 h, and then exposed to a chemiluminescent substrate. After the detection of the PaLOX protein, the membrane was washed with TBS-T and incubated in a stripping buffer solution (25 mM glycine pH 2.2, 0.1% SDS, 1% Tween-20) for 1 h, to be re-probed with the anti-AtpB control (Agrisera, Vännäs, Sweden), recognizing the beta subunit of the ATP synthase, used already as housekeeping in Phaeodactylum tricornutum (Fortunato et al., 2016).
Oxylipin analysis
Oxylipin synthesis is induced by damage to the diatom cells by cell treatments with ultrasound. Oxylipins were analyzed by LC-MS/MS as sodium adducts with molecular ions (Cutignano et al., 2011; d’Ippolito et al., 2018) and inversely eluted according to the number of double bonds and length of the alkyl chain. Epoxy-alcohols from EPA were revealed by typical ion mass at m/z 371, whereas hydroxy-acids (HEPEs) from EPA were determined by molecular ion at m/z 355. Biosynthesis of these metabolites from hydroperoxy precursors is specific and proceeds by preservation of linkage between C and O2 (Chang et al., 1996). Extraction and characterization of products of the lipoxygenase pathway was performed as described previously (d'Ippolito et al., 2009). The extracts were dissolved in methanol to a final concentration of 1 µg µl−1 and directly analyzed by LC-MS. Analyses were performed in triplicate.
Structural prediction of PaLOX
The prediction of the 3D structure of the PaLOX candidate protein was performed considering the peptide sequence available. The analysis was performed using four template structures, whose atomic coordinates were obtained by scanning the Protein Data Bank database (Rose et al., 2017) with the HHpred partition at the MPI Bioinformatics Toolkit (Zimmermann et al., 2018), and selecting templates among the best-scoring hits. To generate the best PaLOX model, two independent modelling approaches were performed. In one case, the top four hits of the HHpred analysis were used as template structures (Cyanothece sp. 15-LOX (PDB ID: 5MED), Plexaura homomalla 8R-LOX (PDB ID: 3FG1), Homo sapiens 5-LOX (PDB ID: 3V98) and Sus scrofa 12-LOX (PDB ID: 3RDE)); in the other case, considering that P. arenysensis produces 12- and 15-LOX derived oxylipins, only 12- and 15-LOXs with the higher HHpred score were selected (Cyanothece sp. (PDB ID: 5MED), S. scrofa (PDB ID: 3RDE) and from H. sapiens (PDB IDs: 4NRE and 3D3L)). The template of the N-terminal portion (180 aa (amino acids)) of PaLOX, in addition, was obtained by a de novo prediction performed by the QUARK web server (Xu & Zhang, 2012).
In order to predict 3D models, a multiple sequence alignment, performed by ClustalOmega (Sievers & Higgins, 2018), between the PaLOX and the templates sequences, was submitted to the homology modeling software Modeller 9 v.20 (release 29 October 2019) (Webb & Sali, 2016). Moreover, the input for the Modeller software also consisted of the atomic coordinates of the four templates plus the N-terminal portion. The Modeller algorithm was set to generate 100 structural models. In order to evaluate the stereochemical quality of the resulting structures and to select the best model, the generated structures were uploaded, in the standard PDB file format, to the PDBsum server (Laskowski et al., 2018). A full set of Procheck structural analyses (Laskowski et al., 1993) was performed on all the generated models of both the modelling approaches that we applied, including an evaluation of the backbone conformations of all residues compared to the known allowed areas in the Ramachandran plot. The 12/15-LOX based model, compared to the one generated using as templates the top four hits retrieved by the HHpred platform, presented higher stereochemical quality indexes (data not shown), and was selected for subsequent structural analyses. The obtained models were displayed by using the molecular graphics software Vmd (Humphrey et al., 1996). The comparisons between the selected modeled structure and the templates structures were carried out by mTM-align (Dong et al., 2018).
Results
Lipoxygenase identification in Pseudo-nitzschia arenysensis
Taking advantage of the P. arenysensis transcriptome, a LOX sequence (PaLOX) was identified (MMETSP0329-20121206|11370) and characterized. The transcript consists of a 2230 nucleotide (nt) sequence with a 38 nt 5′-untranslated region (5′-UTR), an open reading frame (ORF) of 2115 nt and a 77 nt 3′-untranslated region (3′-UTR) (Fig. S2a). In order to verify the endogenous occurrence of this transcript, PCR amplification of the predicted full-length sequence was performed on the cDNA of P. arenysensis. Both the size and the sequencing of the amplified DNA confirmed the presence of the expected transcript in P. arenysensis (Fig. S2b), without single nucleotide polymorphisms (SNPs) falling in strategic points of the enzyme.
In order to identify introns, PCRs were performed on the genomic DNA using couple of primers designed along the full-length transcript, so that adjacent amplicons were partially overlapping. This analysis allowed us to reconstruct the entire gene locus, composed of three exons separated by two introns of 123 nt and 94 nt, respectively (Fig. S2c). The ORF encoded for a protein of 704 aa, having a calculated molecular weight of 79.291 kDa, a theoretical pI of 5.07 and containing a LOX domain at the C-terminal, between aa 183 and 687. Moreover, the analysis of the primary structure revealed the presence of a signal peptide in the first 20 aa of the protein with a cleavage site between position 19 and 20, as identified via the program SignalP (Emanuelsson et al., 2007) (Fig. S3a). This finding was confirmed by the ASAFind software, which specifically recognizes plastid proteins in algae with secondary plastids of the red lineage (Gruber et al., 2015). The score for the transit peptide of P. arenysensis was > 2, which confirmed the identification with a high confidence (Fig. S3b). This result also was corroborated by the analysis with the software ChloroP (Emanuelsson et al., 1999) (Fig. S3c). Finally, to make sure that only one LOX transcript was present in the P. arenysensis transcriptome, the PaLOX sequence obtained from the previous step was used as the query in a tBlastn search against the whole transcriptome of this species; from this analysis no other significant similarities were retrieved.
Expression profile studies
In agreement with a presumed mechanism of signalling that mediates growth termination in field blooms and laboratory cultures (Ribalet et al., 2007; Vidoudez & Pohnert, 2008; d'Ippolito et al., 2009), the concentrations of LOX products of P. arenysensis increased along the growth curve. Thus, in order to test the role of PaLOX in the synthesis of these products along the growth curve of P. arenysensis, we measured both gene transcription levels by real time qPCR and rate of the LOX-dependent production of fatty acid hydroperoxides (FAHs) in lysates of cell diatoms by the colorimetric FOX2 assay (Orefice et al., 2015) (Fig. 1). The analysis indicated a decrease of PaLOX mRNA levels during the declining phase (–2.203-fold change), which found a metabolic correspondence with the kinetics of FAH production in cell lysates that decreased significantly in the senescence phase (0.83 ± 0.02 µmol FAH (mg protein min)–1) (Fig. 1b).
Identification of sequences for the RNAi machinery in P. arenysensis
In order to predict whether gene silencing via RNAi could work in P. arenysensis, an in silico analysis of the P. arenysensis transcriptome was performed to search for known components of the RNAi pathway, such as Dicer, Argonaute-Piwi and DEAD/HELICASE domains containing proteins, even distantly related. The analysis was performed by sequence similarity search by tBlastn of the A. thaliana sequences involved in the RNAi machinery vs the P. arenysensis transcriptome proteins, as detailed in Materials and Methods. As described previously (De Riso et al., 2009), in diatoms a canonical Dicer sequence is absent, and in P. arenysensis we found a transcript (sequence ID: MMETSP0329-20121206|1584) encoding for a protein with two RNase III domains, that is the universal feature of the Dicer family.
Regarding the Argonaute proteins, the conserved C-terminal PAZ, MID and PIWI functional domains distinguish them (Hutvagner & Simard, 2008), and a putative protein containing these domains was identified in the P. arenysensis transcriptome (sequence ID: MMETSP0329-20121206|10545). Finally, a P. arenysensis protein with a DEAD/HELICASE domain also was found (sequence ID: MMETSP0329-20121206|6839). Assuming that these distantly related proteins could accomplish their function in an RNAi-like machinery, we proceeded to design a gene knock-down strategy to abolish the function of the PaLOX gene, in order to study its function in P. arenysensis.
PaLOX knockdown
An anti-sense construct containing a PaLOX fragment of 180 bp, corresponding to the PaLOX gene sequence from 1807 bp to 1986 bp, was generated. Expression in this vector was driven by the histone H4 promoter, already successfully used in the biolistic transformation of Pseudo-nitzschia (Sabatino et al., 2015). To increase the transformation efficiency, we cloned the LOX antisense sequence immediately after the Sh ble gene, which confers resistance to the antibiotic zeocin (Fig. S1a). Three different strains of P. arenysensis, VS3, SV4 and SV6, were transformed via particle gun bombardment. Transformant strains, resistant to zeocin selection, were screened by PCR performed on the genomic DNA, using a couple of primers able to discern the presence of the exogenous DNA construct (Fig. S1b), and a total of 22 transformed clones were obtained (Table 1).
| Experiment | Wild-type strain | No. of PCR-positive clones |
|---|---|---|
| 1 | VS3 | 8 |
| 2 | VS3 | 5 |
| 3 | SV4 | 5 |
| 4 | SV6 | 4 |
Putative silenced clones were screened by immunoblot for decreased PaLOX content, using a custom-made polyclonal antibody directed against a mix of three synthetic peptides of the putative LOX of P. arenysensis.
From a total of 22 transformed clones, 17 were analyzed and 14 of them displayed reduced amounts of the PaLOX protein with respect to the control. Fig. 2(a) shows a representative Western blot of a subset of transformants, all deriving from the same strain (VS3), where a wide variability of PaLOX protein concentrations among the clones could be observed; overall, the 14 transformed clones displaying a reduced amount of LOX were grouped in the graph of Fig. 2(b), where the downregulation ranged between 6% (clone 7.3) and 81% (clone 7.1) of the corresponding WT concentrations.
Reduced PaLOX activity and products in silenced mutants
In order to confirm that the protein concentration reduction was associated to an impairment of the LOX function, we used the FOX2 assay for the assessment of LOX-derived FAHs production potential of WT and silenced mutants (Orefice et al., 2015). Figure 3 shows two independent experiments performed on the transformants Int5.5 and Int11, two of the clones that showed the lowest concentrations of LOX protein according to the Western blot analysis (Fig. 2b). With respect to the corresponding WT strains (VS3 and SV6), FAH production rate was reduced by 95% in Int5.5 and 79% in Int11. Additional transformant clones obtained from the VS3 strain were analyzed and confirmed the reduction of the LOX products following gene silencing (Fig. S4).
The transformants Int5.5 and Int11 were also analyzed by liquid chromatography-mass spectrometry (LC-MS/MS) (Cutignano et al., 2011; d’Ippolito et al., 2018). In agreement with the literature (Lamari et al., 2013), P. arenysensis biosynthesized four major oxylipins that have been identified previously by NMR, MS and chromatographic methods as two 15S-LOX products, namely 15S-hydroxy-5Z,8Z,11Z, 13E,17Z-eicosapentaenoic acid (15S-HEPE) and 13-hydroxy-14-epoxy-5Z,8Z,11Z,17Z-eicosatetraenoic acid (13,14-HEpETE) (d'Ippolito et al., 2009), and two 12-LOX products, specifically 12-hydroxy-5Z,8Z,10E,14Z,17Z-eicosapentaenoic acid (12-HEPE) and 10-hydroxy-11-epoxy-5Z,8Z,14Z,17Z-eicosatetraenoic acid (10,11-HEpETE). The LC-MS/MS analysis of Fig. 3(b) shows the reduction of oxylipins detected in the samples as the ratio of the total amount of the transformants with respect to that of the corresponding WT. In sample Int5.5, the 13,14-HEpETE and 15S-HEPE derived from 15S-LOX were reduced by 84% and 69%, whereas 10,11-HEpETE and 12-HEPE derived from 12-LOX were reduced by 64% and 74%, respectively. The analogous analysis on transformant Int11 revealed a milder reduction of the 15S-LOX derived oxylipins (21% for the 13,14-HEpETE and 18% for the 15S-HEPEs), whereas no 12-LOX oxylipin was detected, neither in the control nor in the transformant. Very interestingly, the reduction of the oxylipin concentrations was in good agreement with the impairment of the growth of the silenced strains. In fact, we noticed a general tendency of interfered diatoms to grow less than the WT (Fig. 3c,d), and the negative impact was much more pronounced in Int5.5 that showed the most severe loss of oxylipin synthesis.
Structural prediction of the putative LOX enzyme
In Table 2 we report the list of the best scoring hits obtained from the HHpred similarity search by scanning the PDB database. The analysis resulted in 15 structures, all representing possible valid templates with a probability of 100%. Interestingly, all sequences, although from distant species, share with the C-terminal region of the putative protein, from 121 to 704 aa, a low e-value (ranging from 3.5e−90 to 4.6e−60), a high score (ranging from 806 to 556) and that the majority of the hits include 12-LOX and 15-LOX similarities. However, the alignment showed a poor conservation of the N-terminal region in the polypeptide chain encoded by the PaLOX gene.
| PDB ID | Type | Species | Probability | E-value | Score | Secondary structure score |
|---|---|---|---|---|---|---|
| 5MED | 15-LOX | Cyanothece sp. | 100 | 3.5e-90 | 797 | 49.2 |
| 3FG1 | 8R-LOX | Plexaura homomalla | 100 | 3.4e-89 | 806 | 52.1 |
| 3V98 | 5-LOX | Homo sapiens | 100 | 1.4e-88 | 800 | 49.4 |
| 3RDE | 12-LOX | Sus scrofa | 100 | 5.6e-89 | 787 | 44.2 |
| 4NRE | 15-LOX | H. sapiens | 100 | 5.4e-88 | 796 | 47.2 |
| 3VF1 | 11-LOX | Gersemia fruticosa | 100 | 1.5e-86 | 783 | 51.3 |
| 2P0M | 15-LOX | Oryctolagus cuniculus | 100 | 4.4e-85 | 766 | 47.6 |
| 3DY5 | 8R-LOX | P. homomalla | 100 | 1.5e-84 | 799 | 51 |
| 5IR4 | 15-LOX | Pseudomonas aeruginosa | 100 | 3.2e-84 | 762 | 47.1 |
| 3D3L | 12-LOX | H. sapiens | 100 | 3.5e-84 | 743 | 45.3 |
| 5FNO | MnLOX | Pyricularia oryzae | 100 | 3.5e-82 | 735 | 46.9 |
| 4WFO | 13S-LOX | Glycine max | 100 | 9.5e-82 | 756 | 45 |
| 6NS4 | LOX | Fusarium graminearum | 100 | 1.1e-81 | 742 | 38.7 |
| 5FX8 | 13R-MnLOX | Gaeumannomyces graminis | 100 | 6.4e-81 | 726 | 43.7 |
| 5EK8 | 9R-LOX | Cyanothece sp. | 100 | 4.6e-60 | 556 | 38,7 |
- List of possible templates (100% probability) for Pseudo-nitzschia arenysensis lipoxygenase (PaLOX) homology modelling detected by HHpred suite. For each hit, the Protein Data Bank (PDB) ID, the LOX type, the species, the structural homology probability, the e-value, the score and the secondary structure score are reported.
An additional template was generated by de novo protein structure prediction based on the first 180 aa of the putative protein identified as PaLOX. Based on our knowledge of the P. arenysensis oxylipin products (Lamari et al., 2013), four templates, namely a 15-LOX from Cyanothece sp. (UniProt ID: B7JX99), a 12-LOX from Sus scrofa (UniProt ID: P16469), and a 15-LOX type 2 and a 12S-LOX type 2 from Homo sapiens (UniProt IDs: O15296 and P18054, respectively) were chosen among the best-scoring hits obtained by searching the Protein Data Bank with the candidate diatom sequence. Fig. 4 shows the best modelled structure in terms of energetic and stereochemical quality among a set of 100 all-atom models generated by the in silico analysis.
The best model of PaLOX revealed 81% of residues in the most favored regions of the Ramachandran plot and an overall average G-factor of −0.43, according to the PROCHECK program. The predicted 3D representation of these putative sequences showed mainly an alpha structure characterized by thirty α-helices, five 3–10 helices and seven β-strands, corresponding to 47.4%, 2.7% and 4.3% of the sequence, respectively (Fig. S5). No disulfide bonds were detected in the model, nor in the templates.
Like in plants (Andreou & Feussner, 2009), the iron in the active site of the diatom protein is known to be octaedrically coordinated by the aa side chains of three histidine (H391, H396 and H575; Figs 4(d), S5), one asparagine (N579) and the C-terminal residue that in eukaryotes often is an isoleucine (I703) (Figs 4d, S5).
The template modelling (TM)-scores from the pairwise comparison of the core sequences of PaLOX with the four templates and the Root Mean Square Deviation (RMSD) to measure the average distance between the backbone atoms in the superimposed protein cores are reported in Tables S2, S3.
We also performed an in silico structural comparison of PaLOX with the structures of the selected templates to identify the aa responsible for the positional specificity in the modelled protein. Unfortunately, we were not able to add the 12-LOX from H. sapiens (PDB ID: 3D3L) to the comparison, owing to the presence of few missing residues (i.e. residues with a flat and uninterpretable electron density in the crystal structure and in the related PDB file) in the region of the positional specificity triad.
As shown in Fig. 5, the comparison of PaLOX with the three LOXs revealed an interesting difference in the catalytic pocket. Although PaLOX (Fig. 5d) conserved two of the three residues that are present in the pig 12-LOX (Fig. 5b), the substitution of phenylalanine with serine 383 reduced the steric hindrance in the active site of the putative enzyme of P. arenysensis without a major conformational change in comparison to both animal proteins. PaLOX preserved a spatial conformation of the region responsible for the positional specificity that is in line with other LOXs. We can observe a similar reduced steric hindrance in the bacterial LOX, despite the presence of different residues in the positional specificity triad (Fig. 5c).
Discussion
Oxylipins are important signal transduction molecules taking part in the complex mechanisms that control growth, defense and death of diatoms in planktonic and benthic communities (Miralto et al., 1999; Pohnert & Boland, 2002; Paffenhöfer et al., 2005; Fink et al., 2006; Fontana et al., 2007a; Vardi, 2008; Caldwell, 2009; Watson et al., 2009; Ianora et al., 2011, 2015; Barreiro et al., 2011; Gallina et al., 2016; Meyer et al., 2018; Ruocco et al., 2020; Russo et al., 2020). Although the structure and biosynthesis of these products have been well investigated, functional information is still lacking mostly because of limited knowledge of enzymes and genes involved in their synthesis and regulation. This is a tricky research field because of the heterogeneous nature in lipoxygenase (LOX) enzyme primary structure, for which plants and animals sequences share only 25% of identity and cluster in two different clades, with several subgroups within each kingdom (Brash, 1999). A phylogenetic analysis performed on the diatom LOXs, retrieved from the Marine Microbial Eukaryote Transcriptome Sequencing Project (MMETSP) dataset (Keeling et al., 2014), shows that they are closely related to a heterogeneous group of organisms in which the red algae Porphyra purpurea, Pyropia haitanensis and Chondrus crispus, the marine bacterium Shewanella violacea and the chordate Branchiostoma floridae are included, together forming a basal clade in the LOX tree (I. Orefice et al., unpublished). Moreover, this group, including red algae, is more closely related to the prokaryotic LOXs and appears more distant from higher plants than animals, as demonstrated previously (Chen et al., 2015; Zhu et al., 2015).
The model diatom species Phaeodactylum tricornutum and Thalassiosira pseudonana do not produce oxylipins and lack the related enzymatic machinery (Collins et al., 2016; Lupette et al., 2018). However, Pseudo-nitzschia multistriata, one of the two species of the genus Pseudo-nitzschia with a sequenced genome (Basu et al., 2017), synthesizes unusual oxylipins of a putative 14-LOX but does not show any sequence homologous to known LOX genes.
The diatoms of the genus Pseudo-nitzschia biosynthesize different LOFAs, namely hydroxyacid and epoxyalcohol derivatives of eicosapentaenoic acid (EPA), whose regiochemistry is strictly species-specific (Fontana et al., 2007b; d'Ippolito et al., 2009; Lamari et al., 2013; Adelfi et al., 2019). A study of the genus Pseudo-nitzschia, including 25 different species and strains, showed clustering according to four chemotypes corresponding to six families of LOX products, including oxylipins of the 15S- and 12-LOX pathways in P. arenysensis (Lamari et al., 2013). The reason for this variability among closely related species is not clear, but it suggests a differentiation of LOX enzymes driven by function (Jenke-Kodama et al., 2008). In fact, in analogy with the reports across different phyla, these metabolites have been suggested as species-specific mediators in the communication of a plankton community (Lamari et al., 2013). Moreover, oxylipins in P. arenysensis (formerly P. delicatissima) were found to vary along the growth curve (d’Ippolito et al., 2009) in agreement with a suggested mechanism of signalling in development and termination of field blooms and laboratory cultures (Ribalet et al., 2007; Vidoudez & Pohnert, 2008; d'Ippolito et al., 2009). Recently, this hypothesis has found indirect confirmation in the correlation of LOX products to the cell density in natural communities (Ribalet et al., 2007; d'Ippolito et al., 2009; Russo et al., 2020).
Our molecular analysis of P. arenysensis revealed the presence of a unique transcript, whose full-length molecule showed no single nucleotide polymorphisms (SNPs) in strategic codons of the substrate pocket. In agreement with the cell density hypothesis, the transcriptional analysis of P. arenysensis revealed a decrease of the transcript levels along the growth curve reaching an important downregulation during the declining phase (Fig. 1a). This behavior was further supported by the FOX2 assay that showed a significant decrease of FAH production rate during the declining phase (Fig. 1b).
Previous structural studies based on NMR, MS and chiral chromatography showed that the oxylipins of P. arenysensis derive from EPA by 12- and 15S-LOX enzymes (d'Ippolito et al., 2009; Lamari et al., 2013; Adelfi et al., 2019), but it is not known whether these compounds are the products of two different enzymes or of a single isoform with a dual positional specificity. The current RNA results indicated the presence of only one type of LOX and suggested the possibility of a multifunctional enzyme. Mammalian 12- and 15-LOX often are 99% identical at the amino acid (aa) level (Schneider et al., 2007), as well as the human 12/15-LOX that is an enzyme producing both series of oxylipins involved in inflammation and various human diseases (Singh & Rao, 2019). Algae and less complex organisms harbor strikingly simple oxylipin pathways, sometimes showing LOXs with multifunctional properties (Senger et al., 2005; Chen et al., 2015; Zhu et al., 2018). Additional examples of biochemical LOX plasticity are offered by the red alga C. crispus, which produces both C20 (animal-like) and C18 (plant-like) oxylipins (Bouarab et al., 2004; Teng et al., 2017), or by the diatoms of the genus Leptocylindrus, which are able to carry out oxygenation of EPA and docosahexaenoic acid (DHA) (Nanjappa et al., 2014).
Genetic manipulations constituted a big challenge in diatoms for a long time, because of the lack of tools to generate genetic mutants. We reported successful use of the biolistic method for the genetic transformation of P. arenysensis, using the H4 promoter of P. multistriata (Sabatino et al., 2015). Moreover, traditional loss-of-function approaches by RNAi already were applied for functional genomic analysis in the diatoms P. tricornutum and T. pseudonana, despite the molecular players normally involved in RNA silencing being only poorly conserved in diatoms (De Riso et al., 2009; Lavaud et al., 2012; Trentacoste et al., 2013; Levitan et al., 2015). The analysis of the P. arenysensis transcriptome revealed the presence of a putative Dicer protein and of an Argonaute-Piwi protein, key components of the RNAi machinery suggesting the involvement of an RNA-mediated silencing as regulatory mechanism that may be used as a reverse genetic tool in P. arenysensis, in the absence of a sequenced genome which would be required for more advanced approaches such as the CRISPR/Cas9 methodology.
Therefore, to silence the PaLOX gene we generated an anti-sense construct (Fig. S1a). Western blot analysis showed a wide variability of downregulation among the transformants, ranging between 6% and 81% of the corresponding wild-type (WT) levels (Fig. 2b). By FOX2 assay, we demonstrated that this protein reduction corresponds to a decrease in FAH production rate (taken as a fold-change of LOX enzymatic activity in comparison to the WT strains) (Orefice et al., 2015), reaching up to 95% and 79% less than the controls, in two of the most silenced clones, Int5.5 and Int11 (Figs 3a, S4). The LC-MS profile to evaluate the amount of oxylipins produced in these two transformants revealed a reduction of both 12- and 15S-LOX derived products, compared to the normal concentrations. Int5.5, in which the FAH reduction was more evident, showed a reduction of 84% and 63% of the 15S-LOX and 12-LOX derived HEpETEs, respectively. However, both in the WT SV6 and in its transformant Int11 only the products of 15S-LOX were detected, with a weak reduction in the silenced clone (Fig. 3b). Presumably, SV6 and its engineered derivative Int11 naturally lost the ability to synthesize 12-LOX products. This may be the result of a change in the primary structure of the protein that is not possible to functionally characterize because of the very high homology among LOX sequences (Schneider et al., 2007). Furthermore, the alignment of the substrate in the enzymatic cavity is a dynamic process as fatty acids are flexible molecules with several thermodynamically favored conformations (Sloane et al., 1991). Only one of these conformers may constitute the catalytically productive substrate and a possible reduction in the concentration of this conformer may explain the absence of 12-LOX products in SV6 and the resulting transformant Int11. It is noteworthy that variability of oxylipin products among natural clones also has been reported in the diatom Skeletonema marinoi and has been proposed as another process to shape diatom population (Gerecht et al., 2011, 2013).
The degree of oxylipin reduction results in a commensurate level of growth impairment of the silenced strains as we observed a general tendency of the interfered cells to grow less than the WT. In mammalians, oxylipins derived from 12- and 15-LOX activity exhibit diverse and often opposite physiological effects. If one accepts that a similar difference occurs also in diatoms, our knock-down results suggest that 15S-LOX plays a major control on P. arenysensis growth. In fact, the effect was much more pronounced in the Int5.5 strain that showed the most severe inhibition of oxylipins from the 15S-LOX pathway (Fig. 3c,d), whereas the reduced effect on the growth of Int11 strain was in line with the weaker decrement of these metabolites in this transformant. These results also are in excellent agreement with the previous report on the linear correlation between the increase of the synthesis of oxylipins by 15S-LOX, namely 15S-HEPE and 13,14-HEpETE, and the steady expansion of the culture of P. arenysensis (d’Ippolito et al., 2009) (HEPEs, hydroxy-eicosapentaenoic acids; HEpETEs, hydroxy-epoxy eicosatetraenoic acids). Because we have never observed growth impairment in P. arenysensis transgenic strains in past studies (Sabatino et al., 2015), nor has it ever been reported for P. tricornutum silenced strains (De Riso et al., 2009; Lavaud et al., 2012; Levitan et al., 2015), we believe that our data provide genuine support to the correlation between diatom cell growth/density and oxylipin concentrations. In this view, 12-LOX activity may be related to unknown functions different from growth control because the absence of the resulting oxylipins does not affect the growth of both the WT SV6 strain and its transformant Int11.
Over the last few decades a number of critical aa have been identified as regiospecific determinants at the active site of selected LOX isozymes. Inspired by studies in plants (Egmond et al., 1972), the so-called substrate orientation hypothesis has become the preferred model to explain the regiospecificity for LOXs (Sloane et al., 1991; Schneider et al., 2007; Ivanov et al., 2010; Vogel et al., 2010; Newcomer & Brash, 2015). According to this model, a substrate fatty acid is accommodated in a hydrophobic cavity and the basis of LOX specificity depends on (1) the type of PUFA used as LOX substrate, (2) the orientation of the PUFA inside the catalytic pocket that changes the hydrogen presented to the catalytic iron, (3) the depth of substrate insertion (substrate frame-shift) that influences the selection of the pentadiene system being attacked. In particular, the analysis of the triad of aa that are responsible for the positional specificity of the arachidonic acid oxygenation of animal LOXs (Vogel et al., 2010) showed that these aa, reported in the literature as a F-I-I in rabbit and human 12/15-LOXs type 2, regulate the depth of the substrate penetration within the substrate-binding pocket. The substitution with smaller hydrophobic residues is reported to increase the depth of the pocket, thus enabling a deeper sliding of the substrate and changing the position of dioxygenation.
The in silico model of PaLOX enabled the identification of the catalytic and regiospecific residues including the six residues, namely H391, H396, H575, N579 and I703, that are required in the iron atom binding and the catalytic activity. The aa S383, V451 and I633 instead were located at the end of the hydrophobic pocket to form the U-shaped cavity that may affect the regiospecificity of PaLOX. This amino acid arrangement also was consistent with the substrate-binding side of mammalian 12- or 15-LOX but it featured a sterically less crowded pocket that can explain the lower enzymatic specificity putatively permitting the synthesis of oxylipins by oxygenation of EPA at both C12- and/or C15 whose structure has been determined previously by chemical methods (d'Ippolito et al., 2009; Lamari et al., 2013; Adelfi et al., 2019). However, to confirm this in silico analysis, a determination of the crystal structure of PaLOX complexed with PUFA substrates and mutational analysis of all active residues must be performed.
In conclusion, this study reports a functional and in silico characterization of a LOX gene in diatoms, establishing for the first time a direct relationship between a gene of secondary metabolic pathways and its final products in this lineage of microalgae. The effective silencing approach with P. arenysensis demonstrated that the downregulation of PaLOX led to a decrease of the LOX activity and a consequent reduction of the diatom growth. Further experiments for the depletion of PaLOX, the generation of knock-out strains, and a refined biochemical analysis of the enzyme are necessary to confirm the physiological role of oxylipins and substantiate the double function of PaLOX. Nevertheless, these results strengthen the idea that LOX products in diatoms are correlated to cell growth, such as in other organisms including humans (Hayashi et al., 2006; Barooni et al., 2019) and plants (Kolomiets et al., 2001).
Acknowledgements
This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement no. 654008. IO was supported by an OU SZN PhD fellowship. PM was partially supported by the Gordon and Betty Moore Foundation through grant no. 7978. AF and GdI thank Lucio Caso for the technical support in algal growth and MS analysis.
Author contributions
VS, IO, GR, GdI, AF and MIF conceived and designed the project; VS, IO, LA, GdI and PM performed the experiments; VS, IO, PM, LA, MLC, GR, GdI, AF and MIF analyzed the data; VS, IO, PM, LA wrote the manuscript with inputs from all the authors; and all authors have read and agreed to the published version of the manuscript. VS and IO contributed equally to this work.
Open Research
Data availability
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
