Biochemical characterization of the cyclooxygenase enzyme in penaeid shrimp

Shrimp trials and experiments in this study were approved by the Institutional Animal Care and Use Committee of the National Center for Genetic Engineering and Biotechnology, Thailand (BT-Animal 28/2560). All experiments were performed in accordance with Animal Research: Reporting of In Vivo Experiments (ARRIVE) and conformed with international and national legal and ethical requirements [28, 29].

Shrimp ovary samples were subjected to total RNA extraction using Trizol® reagent (Thermo Fisher Scientific, Massachusetts, USA). The RNA concentration was estimated using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Ovarian cDNA was synthesized using the RevertAid^TM First Strand cDNA Synthesis Kit with oligo (dT) 18 primer (Thermo Fisher Scientific) according to the manufacturer’s instructions.

The full-length Pmcox gene (GenBank: KF501342.1↗) was amplified from P. monodon ovarian cDNA using Pmcox-HindIII-F and Pmcox-XbaI-R primers. Primer sequences for Pmcox cloning are shown in Table 1. The full-length Pmcox gene was cloned into pcDNA3.1^TM (+)/myc-His B (Thermo Fisher Scientific), resulting in a plasmid construct encoding the full-length PmCOX protein with a Myc-His tag at the C-terminus. Sequencing was performed (1^st-BASE, Malaysia) to confirm the correct sequence in the plasmid [30].

Protein BLAST (BLASTP version 2.10.0⁺) analysis was performed to verify the identity of the protein [30]. AU-rich elements (AREs) and the AAUAAA motif in the 3´ untranslated region (3´ UTR) were analyzed using the PolyAPred algorithm [31]. PmCOX conserved domains were predicted using the Conserved Domain Architecture Retrieval Tool (CDART) program [32]. Multiple sequence alignment between PmCOX and its homologs was performed using the Clustal Omega program [33]. A phylogenetic tree was constructed using the Molecular Evolutionary Genetics Analysis (MEGA) version 10.1 program (https://www.megasoftware.net/↗) using the maximum likelihood method [34]. An N-glycosylation motif search was performed using NetNGlyc 1.0 prediction (www.cbs.dtu.dk/services/NetNGlyc/↗) [35]. An N-glycosylation potential of 0.5 was considered as the minimum cut-off value.

To remove potential glycosylation sites on PmCOX, site-directed mutagenesis was performed using KAPA HiFi PCR kit mutagenesis (Kapa Biosystems, Massachusetts, USA) for N-to-Q substitutions at amino acid residues 79, 170 and 424. Single, double and triple mutations at the glycosylation sites of PmCOX were generated. Primer sequences used for the PCR mutagenesis are shown in Table 1. The restriction enzyme DpnI (Thermo Fisher Scientific) was used to remove the template plasmid prior to transformation into E. coli DH5α. Transformants were selected for plasmid extraction using colony PCR. Recombinant plasmids harboring mutated Pmcox were verified by DNA sequencing (1^st-BASE) and Clustal Omega multiple sequence alignment against wild-type PmCOX [33].

Human embryonic kidney 293T cells (ATCC ® CRL-11268^TM; ATCC, Virginia, USA) were maintained in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin-streptomycin solution (Invitrogen, New York, USA). Cells were incubated at 37°C in 5% CO₂ and subcultured every three days. For transient protein expression, expression plasmids containing wild-type PmCOX or PmCOX glycosylation mutants, the empty vector (pcDNA3.1^TM(+)/myc-His B), or pEGFP-N1 (negative control) were transiently transfected into 293T cells by calcium phosphate precipitation [36]. Culture media were replaced 18 h post-transfection and the transfected cells were harvested at 48 h to assess protein expression by Western blotting.

Cells were transfected with expression plasmids encoding wild-type PmCOX or PmCOX glycosylation mutants. At 48 h post-transfection, fresh DMEM without FBS was added to the transfected cells. Cells were incubated with either 0.1% dimethyl sulfoxide (DMSO; vehicle) or 10–40 μM ARA (Cayman Chemical, Michigan, USA) for 30 min at 37°C. Cells were then harvested for Western blot analysis to verify protein expression. Cell culture media were collected to determine levels of secreted PGE₂ by enzyme immunoassay (EIA). The remaining media were subjected to C18 solid phase extraction (SPE) and analyzed using ultra-performance liquid chromatography high-resolution tandem mass spectrometry (UPLC-HRMS/MS). Five replicates were used for each experimental condition.

Levels of secreted PGE₂ were determined using the PGE₂ EIA kit (Cayman Chemical), according to the manufacturer’s instructions. Dose-response curves were generated using PGE₂ standards ranging from 15.63–250.00 pg/mL. The limit of quantification for PGE₂ was 15.63 pg/mL. Intra- and inter-day coefficients of variance (CV) were 6.41% and 7.57%, respectively.

Culture media harvested from PmCOX-expressing cells were adjusted to pH 4.0 using formic acid. Ten percent (w/v) butylated hydroxytoluene (Sigma-Aldrich, Missouri, USA) in ethanol and prostaglandin B₁ (PGB₁) were added to the media as an antioxidant and an internal standard, respectively. Ten milliliters of cell culture media were loaded onto Vertipak^TM C18 SPE cartridges (Vertical Chromatography, Bangkok, Thailand), which had previously been washed with 10 mL methanol and 10 mL water. Sample-loaded cartridges were washed twice with 5 mL water and eluted twice with 5 mL ethyl acetate. Eluates were evaporated and reconstituted in ethanol for the identification and quantification of prostaglandins by UPLC-HRMS/MS.

Five micrograms of pcDNA3.1^TM(+)Pmcox/myc-His B were transfected into 293T cells. Forty-eight hours after transfection, cells were washed with phosphate-buffered saline (PBS) and incubated with complete DMEM supplemented with 10% FBS spiked with 2 μg/ml of tunicamycin (Sigma-Aldrich, Missouri, USA) or 0.1% DMSO (vehicle). Cells were incubated at 37°C for 16 h in a CO₂ incubator before being harvested and stored at –20°C for Western blot analysis.

Transfected 293T cells were harvested and sonicated for 10 s in radioimmunoprecipitation assay (RIPA) buffer (1% Nonidet^TM P-40, 1% sodium deoxycholate, 1% sodium dodecyl sulfate, 150 mM sodium chloride, 0.01 M sodium phosphate buffer pH 7.2 and 2 mM EDTA) containing a cOmplete™ EDTA-free protease inhibitor cocktail (Sigma-Aldrich). Protein concentrations from cell lysates were determined using the Lowry assay [37]. Forty micrograms of protein samples were separated by electrophoresis on 10–12% polyacrylamide gels, transferred to Hybond® ECL nitrocellulose membrane (GE healthcare, Illinois, USA) and incubated with anti-His monoclonal IgG antibodies (Abcam, Massachusetts, USA) for 1 h. Proteins were visualized using horseradish peroxidase-conjugated secondary antibodies (Vector Laboratories, California, USA) and SuperSignal West Pico Chemiluminescent Substrate^TM (Thermo Fisher Scientific).

Three hundred micrograms of cell lysate containing PmCOX-Myc-His was incubated with 2 μL of anti-His antibodies (Abcam) with gentle rocking at 4°C for 1 h. Sixty microliters of protein A Sepharose beads CL-B4 (GE Healthcare, Illinois, USA) was added to RIPA buffer to prepare a 50% slurry before addition to the cell lysate mixture. The mixture was incubated with gentle rocking at 4°C for 1 h. Protein-bound Sepharose beads were spun down and washed twice in 1 mL RIPA buffer.

Immunoprecipitated PmCOX was digested with endoglycosidase H enzyme (Promega, Wisconsin, USA) according to the manufacturer’s instruction. Briefly, protein A Sepharose beads bound to PmCOX proteins were incubated in denaturing solution at 95°C for 5 min. Endoglycosidase H enzyme was added to the denatured protein and the mixture was incubated at 37°C for 18 h.

P. monodon haemolymph was drawn from the ventral sinus at the first pair of pleopods from 10 juvenile P. monodon with an average body weight of 12 g. The haemolymph was mixed with 10% ice-cold sodium citrate (anti-coagulant) at a 1:1 ratio (v/v) of haemolymph:anti-coagulant. The haemolymph was pooled, divided into 1-mL aliquots, and incubated with 15 or 150 ng/mL aspirin, 20 or 200 ng/mL indomethacin, and 20 or 200 ng/mL ibuprofen, based on effective dosages published in other crustaceans [38–40]. After a 30-min incubation, 0.1% DMSO or 10 μM ARA was added to the haemolymph and the mixture was incubated at 28°C, with shaking at 200 rpm for 30 min. Haemolymph was then spun down at 1,500 ×g for 5 min at 4°C to remove haemocytes and other cell debris. PGE₂ levels in the haemolymph were determined using the EIA kit. The experiment was performed in triplicate to determine percent inhibition of each COX inhibitor. Levels of PGE₂ in haemolymph samples were subtracted by the basal levels of PGE₂ obtained from the samples treated with 0.1% DMSO (negative control), with 100% PGE₂ production defined as levels of PGE₂ from haemolymph treated with 10 μM ARA (positive control). Percentages of PmCOX inhibition under inhibitor treatment were determined as follows: %inhibition=PGE2levelpositivecontrol−PGE2inhibitorPGE2levelpositivecontrolx100

Juvenile P. vannamei with a body weight range of 10–12 g (n = 12) were obtained from commercial culture farms in Prachuap Khiri Khan province, Thailand. The shrimp were tested to be free from specific pathogens (Taura syndrome virus, white spot syndrome virus, yellow head virus, and infectious hypodermal and hematopoietic necrosis virus) using the EZEEGENE® nested PCR test kit (BIOTEC, Pathum Thani, Thailand). Shrimp were acclimatized in a 600-L tank and maintained at a salinity of 30 ppm and temperature of 25–32°C at the Aquaculture Product Development and Services Laboratory, BIOTEC, Thailand, for the duration of the study. Shrimp were fed twice a day with StarFeed commercial pellets (CPF, Bangkok, Thailand). Water quality in the rearing tank was evaluated every three days by measuring the temperature, pH and dissolved oxygen levels. Ammonia–nitrogen, nitrite–nitrogen and alkalinity levels were monitored weekly.

To determine the effects of ibuprofen on prostaglandin levels in haemolymph, shrimp were intramuscularly injected with PBS (control), or 40 μg or 400 μg ibuprofen. At 48 h after the injection, haemolymph were drawn, mixed with 10% ice-cold sodium citrate (anti-coagulant) at 1:1 (v/v) and centrifuged at 1,500 ×g for 5 min at 4°C to remove haemocytes. Levels of PGE₂ in haemolymph were then estimated by EIA. Percentage of PvCOX inhibition under inhibitor treatment was determined using the same equation as shown for the in vitro assay.

The full-length Pmcox cDNA is 2,816 bp in length, containing a 1,842-bp coding region (NCBI accession number KF501342↗) that translates to a polypeptide 614 amino acids in length. The 5´ and 3´ UTRs are 205- and 766-bp long, respectively. TBLASTX analysis revealed that the closest PmCOX homolog is P. vannamei COX (PvCOX, GenBank accession no. XP027218437, E-value 0.0) with 94.95% sequence identity. CDART algorithm predicted that the putative PmCOX enzyme contains three conserved domains, which are the calcium-binding epidermal growth factor-like domain (residues 57–95), the prostaglandin endoperoxide synthase domain (residues 115–602) and the peroxidase domain (residues 172–600) (Fig 1A). Multiple sequence alignment revealed that PmCOX contains all seven essential catalytic residues, namely R146, H233, Y383, Y413, H414, H416 and S556, which were highly conserved with those in mammalian homologs (Fig 1B). Additionally, these conserved residues were also identified in COX sequences of P. vannamei, the kuruma prawn P. japonicus, the blue crab Callinectes sapidus, the American lobster Homarus americanus, the Hawaiian red shrimp Halocaridina rubra and the two amphipod crustaceans Gammarus sp. and Caprella sp. Interestingly, although mammalian COXs typically contain the KDEL/STEL-type endoplasmic reticulum retention/retrieval signal, the C-termini of PmCOX and crustacean COXs are shorter than the mammalian homologs and lack these signals.

AU-rich elements (AREs; AUUUA) and potential polyadenylation AAUAAA signals have previously been identified in the 3´ UTR region of COX from two amphipod crustaceans, Gammarus sp. and Caprella sp. In the Pmcox gene, the 3´ UTR also contains five AREs and one polyadenylation signal (Fig 1C). The presence of conserved domains and residues in both the coding region and the 3´ UTR suggested that PmCOX could likely regulate the production of PGH₂ similar to COXs in other organisms.

For mammalian COXs to become catalytically active, three asparagine residues in COX1 and up to four asparagine residues in COX2 must be glycosylated. Multiple sequence alignment matched two known glycosylation sites in ovine COX1, N68 and N144, with PmCOX N93 and N170, respectively (Fig 1B, blue highlight). However, the third glycosylation site in ovine COX1 at N410 did not align with any asparagine residues from PmCOX or COXs from other crustaceans. As a result, the PmCOX sequence was therefore scanned for glycosylation motifs (N-(X)-S/T), and three potential glycosylation sites were identified at N79, N170 and N424 (Fig 1B, pink highlight). The presence of these motifs suggested that the N-glycosylation pattern may be conserved in crustacean COXs.

A phylogenetic tree of COX sequences was constructed using the neighbour-joining method, revealing two distinct clusters of vertebrate and crustacean COXs. Vertebrate COXs were further separated into two major branches, corresponding to COX1 and COX2 isoforms. COX sequences from penaeid shrimp, including P. monodon, P. vannamei and P. japonicus, clustered together (Fig 2) and also formed part of a larger cluster that includes COX sequences from other decapod crustaceans, including C. sapidus, H. americanus and H. rubra. COXs from two amphipod crustaceans, Gammarus sp. and Caprella sp., were part of a separate arm from the decapod crustacean cluster. Lastly, the D. pulex COX sequence was the most evolutionarily distant from the rest of the group.

To determine PmCOX enzymatic function, 293T cells were untreated (control 1), transfected with pEGFP-N1 (control 2), or transfected with pcDNA3.1^TM(+)Pmcox/myc-His B. Forty-eight hours after transfection, cells were incubated with either 0.1% DMSO (vehicle) or 10 μM ARA for 10 min at 37°C. Transfected cells and cell culture media were collected separately for Western blot analysis and EIA, respectively. It was predicted that the PmCOX-Myc-His molecular mass would be approximately 70 kDa. However, Western blot analysis revealed that PmCOX expressed in 293T cells had a higher molecular weight than expected, at 77 kDa (Fig 3A), suggesting the presence of post-translational modifications on the PmCOX protein.

EIA was performed to determine the effects of ARA treatment on PmCOX-expressing cells. In the DMSO-treated group, levels of secreted PGE₂ were 0.23, 0.22 and 1.20 nM in cell culture media collected from control 1, control 2 and cells expressing PmCOX-Myc-His, respectively (Fig 3B). As PmCOX expression increased levels of secreted PGE₂ by 5.2- to 5.6-fold compared to the control groups, these findings indicated that PmCOX possessed similar enzymatic function as mammalian COX enzymes and was incorporated as part of the prostaglandin biosynthesis pathway in 293T cells.

In ARA-treated samples, levels of secreted PGE₂ were 1.19 and 1.21 nM for controls 1 and 2, respectively. For PmCOX-expressing cells treated with ARA, 72.54 nM PGE₂ was detected, which was approximately 60-fold higher than in both controls 1 and 2. These findings further confirmed the function of PmCOX. Additionally, the effects of ARA treatment were more pronounced in PmCOX-expressing samples, with PGE₂ levels from PmCOX-expressing cells increasing by 60-fold after ARA treatment compared to around 5-fold for controls 1 and 2.

To verify EIA results, 293T cells were transfected with either pcDNA3.1^TM (+)/myc-His B (empty vector) or pcDNA3.1^TM (+)Pmcox/myc-His B. Transfected cells were treated with 10, 20 or 40 μM ARA or 0.1% DMSO (vehicle) for 10 min at 37°C. Cell culture media were harvested, subjected to C18 SPE and analyzed by UPLC-HRMS/MS. The base peak chromatogram of cell culture extracts showed peaks at retention times 5.36, 5.84 and 16.61 min, which corresponded with the retention times of PGF_2α, PGE₂ and ARA standards, respectively (Fig 4A). The mass spectra and product ions of PGF_2α, PGE₂ and ARA were used to verify identities of these metabolites (Fig 4B–4D). Levels of PGF_2α and PGE₂ were determined using the standard curve method. On the other hand, levels of ARA in cell culture medium exceeded the linearity range and were not quantified in this analysis. To establish prostaglandin basal levels, non-transfected controls and cells expressing PmCOX were incubated with 0.1% DMSO as negative controls. UPLC-HRMS/MS analyses revealed that levels of PGE₂ and PGF_2α were below the detection limit when cells were treated with 0.1% DMSO, suggesting that basal levels of these prostaglandins were negligible in this assay (Fig 4E). When non-transfected controls were incubated with 40 μM ARA, 1.07 nM PGE₂ was detected in cell culture media while PGF_2α remained under the limit of detection. The production of PGE₂ and PGF_2α was more robust only in PmCOX-expressing cells that were treated with ARA. In fact, levels of secreted PGF_2α and PGE₂ corresponded with levels of exogenous ARA used in the treatment in a dose-dependent manner. Incubation with 10, 20 and 40 nM ARA resulted in 0.90, 1.08 and 1.65 nM PGE₂ and 0.21, 0.21 and 0.36 nM PGF_2α, respectively.

The presence of three N-glycosylation motifs in the PmCOX protein sequence (Fig 1B) suggested that N-glycosylation is required for PmCOX to be catalytically active. Pulled-down PmCOX-Myc-His was incubated with endoglycosidase H, which cleaves oligosaccharides from asparagine residues of glycoproteins, resulting in decreased protein molecular mass (Fig 5A). Interestingly, the sizes of glycosylated PmCOX and endoglycosidase H-treated PmCOX in this assay were slightly higher than our other Western blot analyses at approximately 90 and 80 kDa respectively, possibly due to the effects of the endoglycosidase H digestion mixture on protein migration.

To confirm the N-glycosylation of PmCOX, cells expressing PmCOX-Myc-His were treated with tunicamycin, which inhibits N-linked glycosylation by preventing the addition of oligosaccharide to nascent polypeptides. Western blot analysis revealed that the tunicamycin-treated sample resulted in two protein bands at high (approximately 77 kDa) and low (approximately 70 kDa) molecular weights, which corresponded with glycosylated and unglycosylated PmCOX, respectively (Fig 5B). Based on these results, it could be concluded that PmCOX was expressed in 293T cells as an N-glycosylated protein with a molecular mass of 77 kDa.

Site-directed mutagenesis was performed to disrupt PmCOX potential glycosylation sites at residues 79, 170 and 424 to create three PmCOX single glycosylation mutants (PmCOXN79Q, PmCOXN170Q and PmCOXN424Q), three double glycosylation mutants (PmCOXN79QN170Q, PmCOXN170QN424Q and PmCOXN79QN424Q), and one triple glycosylation mutant (PmCOXN79QN170QN424Q). Western blot analysis revealed that PmCOX molecular mass gradually decreased as the number of glycosylation mutations increased (Fig 5C), suggesting that all three asparagine residues were glycosylated at the same time when expressed in mammalian cells.

To determine whether N-glycosylation is required for PmCOX catalytic function, cells transfected with the empty vector or vectors encoding wild-type PmCOX, PmCOX single glycosylation mutants or the PmCOX triple glycosylation mutant were treated with either 0.1% DMSO or 40 μM ARA at 37°C for 30 min. Western blot analysis was performed to verify protein expression and to demonstrate the differences in molecular mass for wild-type PmCOX and various PmCOX glycosylation mutants (Fig 6A).

Due to the presence of endogenous COX in 293T cells, the basal level of PGE₂ was determined using cell culture media collected from 293T cells transfected with the empty vector. EIA revealed that levels of PGE₂ were 0.64 and 2.68 nM when treated with 0.1% DMSO and 40 μM ARA, respectively (Fig 6B). This indicated that endogenous COX enzyme in 293T cells was able to convert exogenous ARA into secreted PGE₂ in cell culture media. On the other hand, when wild-type PmCOX was expressed in 293T cells without ARA treatment, PGE₂ levels were only 0.62 nM, comparable to the basal level of PGE₂ in cells transfected with the empty vector. Therefore, the availability of ARA as a COX substrate had a larger impact on levels of secreted PGE₂ than the amount of COX enzyme present in 293T cells.

To determine whether mutations at the PmCOX glycosylation sites resulted in reduced PmCOX catalytic function, cells expressing wild-type PmCOX or PmCOX glycosylation mutants were treated with 40 μM ARA. Expression of wild-type PmCOX resulted in 7.40 nM PGE₂ (Fig 6B). Expression of PmCOXN79Q, PmCOX170Q, and PmCOXN424Q resulted in 8.65, 2.45 and 2.79 nM of secreted PGE₂, respectively. This suggests that N-glycosylation at residues 170 and 424, but not residue 79, is required for PmCOX catalytic function. Additionally, expression of the PmCOX triple mutant resulted in 2.23 nM secreted PGE₂. As levels of PGE₂ from PmCOXN170Q, PmCOXN424Q and the PmCOX triple mutant were comparable to those found in cells transfected with the empty vector, it was postulated that the removal of either N170 or N242 glycosylation site is sufficient to completely disrupt the PmCOX enzymatic activity.

UPLC-HRMS/MS analysis was performed on cell culture medium harvested from cells expressing wild-type or triple-mutant PmCOX in the presence or absence of exogenous ARA. In the absence of ARA, levels of PGE₂ and PGF_2α were below the detection limit regardless of whether cells were transfected with the empty vector or expressing wild-type or triple-mutant PmCOX (Fig 6C). In the presence of 40 μM ARA, however, levels of secreted PGE₂ were 1.1, 1.7 and 1.2 nM from cells transfected with the empty vector and cells expressing wild-type or triple-mutant PmCOX, respectively (Fig 6C, black bars). On the other hand, PGF_2α was detected only in cells expressing wild-type PmCOX but not in cells transfected with the empty vector or those expressing the PmCOX triple mutant (Fig 6C, white bars). These findings confirmed that the removal of N-glycosylation sites on PmCOX disrupted the production of PGE₂ and PGF_2α in 293T cells.

To investigate whether the effects of PmCOX N-glycosylation mutations were observed only when cells were treated with high concentrations of ARA, cells expressing wild-type or triple-mutant PmCOX were treated with 10, 20 or 40 nM ARA. UPLC-HRMS/MS analyses revealed that levels of PGE₂ steadily increased in a dose-dependent manner (Fig 6D). However, PGF_2α levels were mostly unaffected, with comparable amounts being secreted from cells expressing wild-type PmCOX while being undetectable from cells expressing the PmCOX triple mutant (Fig 6E).

Although our data indicated that PmCOX was catalytically active, PmCOX enzymatic function had yet to be demonstrated in P. monodon. Based on conserved catalytic residues and domains between PmCOX and its mammalian homologs, we hypothesized that commercially available COX inhibitors could reduce PmCOX catalytic function in shrimp. In vitro inhibition assays were performed using P. monodon haemolymph. Haemolymph treated with 10 μM ARA (positive control) was set as 0% inhibition of the PmCOX enzymatic activity (Fig 7A). Haemolymph samples were also pre-treated with varying concentrations of aspirin, ibuprofen or indomethacin for 30 min at 28°C followed by 10 μM ARA for an additional 1 h. Incubation with 150 ng/mL aspirin and 200 ng/mL ibuprofen resulted in the highest and second highest inhibitory effects at 62.3% and 57.0% inhibition, respectively. Based on the ability of aspirin and ibuprofen to reduce PGE₂ levels in shrimp haemolymph, we propose that a functional PmCOX is present in P. monodon.

Based on the in vitro inhibition assays where aspirin and ibuprofen effectively reduced PGE₂ levels in P. monodon haemolymph, the effects of COX inhibitors were subsequently verified in vivo. As aspirin is poorly soluble in water, ibuprofen was selected for intramuscular injection in shrimp. Although we intended to repeat the experiment in P. monodon, specific pathogen free (SPF) P. monodon of appropriate size could not be obtained for the experiment. P. vannamei was selected as a substitute based on the presence of prostaglandin biosynthesis genes (S2 Table) and the identification of PGE₂ and PGF_2α in P. vannamei post-larvae (S6 Data). Therefore, SPF P. vannamei was chosen for testing the effects of ibuprofen in vivo. P. vannamei were intramuscularly injected with 100 μL PBS (vehicle) or ibuprofen at a dose of 40 or 400 ng per gram shrimp. As P. vannamei body weights were approximately 12 g per shrimp, 0.48 or 4.8 μg ibuprofen were injected into each shrimp. Levels of PGE₂ in haemolymph of P. vannamei injected with PBS were 933.5 pM, which was set at 0% inhibition (Fig 7B). Levels of PGE₂ in haemolymph of P. vannamei injected with 0.48 and 4.8 μg of ibuprofen were 93.1 and 124.3 pM, which were equivalent to 90.0% and 86.7% inhibition, respectively. We speculated that PGE₂ levels in shrimp haemolymph were not reduced in a dose-dependent manner when treated with ibuprofen because the ibuprofen dosage at 40 ng/g shrimp already exceeded the highest concentration required to inhibit the function of P. vannamei COX (PvCOX) in vivo. Nevertheless, results from both in vitro and in vivo inhibitory assays confirmed the presence of enzymatically active COX enzyme in two penaeid shrimp.

One of the strategies used to identify candidate COX enzymes, such as those in the fruit fly Drosophila melanogaster, is the mapping of essential catalytic residues in the COX active site [45, 46]. These catalytic residues were identified by comparing with those previously characterized in ovine and human COXs [47, 48]. For example, residues R120 and Y355 in ovine COX1 (OvCOX1), which bind the carboxylic acid moiety of ARA, are equivalent to residues R146 and Y383 in PmCOX, respectively [49–51]. Similarly, OvCOX1 Y385 is responsible for extracting a hydrogen from ARA and is equivalent to PmCOX Y413 [47], a highly conserved residue in all crustacean COXs. Distal and proximal histidines, which are responsible for hydroperoxide reductase activities and coordinating the heme iron inside COX active site, are equivalent to residues H207 and H338 in OvCOX1 and residues H233 and H414 in PmCOX, respectively [16, 52, 53]. Lastly, OvCOX1 S530, which serves as the acetylation site for aspirin, aligns with S556 in PmCOX and COX enzymes in other crustaceans [47]. Our sequence alignment not only confirmed the identity of PmCOX by matching its essential catalytic residues to those in mammalian homologs, but also showed that these residues were well-conserved in crustacean COX sequences, suggesting the presence of functional COX enzymes in all analyzed crustacean species.

Various cell-based expression systems have previously been used to establish COX enzymatic activity and N-glycosylation requirements, including yeast, baculovirus-based expression in insect cells and mammalian cells [54–56]. Insect SF9 cells were used for expression of mouse COX2 to investigate its N-glycosylation sites [54]. Pichia pastoris has also been used to determine the N-glycosylation sites of human COX1 and COX2 [55]. Mammalian COS-7 cells have been used to express COX enzymes from zebrafish and amphipod crustaceans [23, 56]. Similarly, COS-1 cells were used to determine glycosylation sites and effects of COX inhibitors on human COX1 [18, 57]. In this study, SF9 and SF21cells had been considered as possible expression hosts for PmCOX as insect cells are more closely related to shrimp. However, Shimokawa and Smith (1992) reported that the insect cell system was not suitable for expressing large amounts of COX enzyme as it lacked the ability to completely glycosylate the expressed proteins [14]. Furthermore, the characterization of COXs from Gammarus sp. and Caprella sp. was performed in COS-7 cells, serving as an example of a successful mammalian cell expression system for crustacean COX expression and characterization [23].

Aside from conserved catalytic residues, another important characteristic of the COX enzyme is its N-glycosylation requirement. Site-directed mutagenesis was used to identify glycosylation residues of COX1 and COX2 [19], and revealed that COX1 is glycosylated at three asparagine residues (N68, N144 and N410) while COX2 may be glycosylated at up to four asparagine residues with a fourth glycosylation site at N580 [19, 53, 54]. It has been demonstrated that the first three glycosylation sites on COX1 are required for protein expression and enzymatic function [19]. On the other hand, COX2 glycosylation at N580 affects COX2 dimer formation, protein stability and efficacy of certain COX inhibitors [18, 57, 58].

The requirement for N-glycosylation was also conserved in invertebrate COXs. Four and six potential glycosylation motifs were identified in G. fruticosa COX-A and COX-B, respectively [21]. Expression of these COX isoforms in COS-7 cells treated with tunicamycin revealed that COX-A and COX-B were glycosylated at four and five glycosylation sites, respectively [21]. Similarly, tunicamycin treatment was also used to verify N-glycosylation on COX enzymes from two species of amphipod crustaceans Gammarus sp. and Caprella sp. [23]. Nevertheless, whether COX N-glycosylation is required for its enzymatic function in invertebrates has not previously been investigated. In this study, we demonstrated that PmCOX was N-glycosylated at three glycosylation sites and mutations at N170 or N424 completely disrupted PmCOX enzymatic function. Furthermore, Western blot analysis revealed decreased expression levels of the triple-mutant PmCOX compared to those of the wild type or single-glycosylation mutants, suggesting that removal of more than one glycosylation site may affect PmCOX protein folding or protein turnover rates. Further investigation is required to confirm the effects of N-glycosylation on PmCOX protein stability.

Effects of COX inhibitors on female reproductive maturation have often been used as evidence for the presence of a prostaglandin biosynthesis pathway in crustaceans [44, 45, 66]. For example, chronic ibuprofen treatment in the small planktonic crustacean Daphnia magna reduced the number of offspring in a dose-dependent manner [39]. In O. senex senex, injection with aspirin or indomethacin reduced the ovarian index, oocyte diameters and vitellogenin levels in crab ovaries [38]. Surprisingly, periodic injection of ibuprofen at 0.1 μg/g body weight into eyestalk-ablated P. vannamei resulted in higher rates of females with developing ovaries compared to the control [59]. Although these studies provide compelling evidence regarding the effects of COX inhibitors on crustacean ovarian development, changes in prostaglandin levels in these organisms after COX inhibitor treatment have not been examined. Our study is the first to report changes in levels of prostaglandins after ibuprofen injection in crustaceans. However, whether the COX inhibitor would also increase ovarian maturation rates in female P. monodon remains to be investigated.

COX substrate specificity is regulated by cavity space within the COX active site. In mammals, ARA is the preferred substrate for COX1 whereas COX2 is more flexible, binding ARA derivatives and other PUFAs, including EPA, α-linolenic acid, γ-linolenic acid and linoleic acid as well as ARA [10, 60, 61]. In fish, COX1 and COX2 enzymes have a more restrictive substrate specificity, with preference toward ARA, but not EPA and DHA [62]. As this study confirms that ARA serves as a substrate for PmCOX, the effects of dietary ARA on shrimp prostaglandin biosynthesis should be taken into consideration during the formulation of shrimp feed. Studies have shown that levels of PGE₂ increased during ovarian development in the Florida crayfish Procambarus paeninsulanus, O. senex senex and P. monodon [24, 41, 63, 64]. As studies in mammals indicated that abundant levels of EPA or DHA could have inhibitory effects on COX catalytic function in converting ARA to PGE₂ [65, 66], levels of EPA and DHA should be regulated in feed for broodstock to prevent the disruption of COX catalytic function during ovarian development.