Degradation of okadaic acid in seawater by UV/TiO2 photocatalysis: proof of concept.

P. ﬂ ash in after 2 The consumption of contaminated shell ﬁ sh with marine toxins causes adverse socioeconomical, environmental and healthimpacts.The marine toxin okadaic acid(OA) provokesdiarrheticshell ﬁ sh poisoning(DSP) syndrome characterized by severe gastrointestinal symptoms. Therefore, there is increasing interest in removing these toxins from the marine environment to protect shell ﬁ sh harvesting sites. Photocatalysis is proposed as an ef ﬁ cient method to detoxify the marine environment. In this study, Prorocentrum lima was used to produce high purity DSP toxins, in particular OA, for degradation studies. The pro ﬁ ling, characterization and quanti ﬁ cation of DSP toxins in the culture of P. lima were achieved by ultrahigh performance liquid chromatography coupled to quadrupole-time of ﬂ ight mass spectrometry (UPLC-QTOF-MS E ) for accurate-mass full spectrum acquisition data. The effectiveness of UV/TiO 2 system to de- grade OA in seawater was assessed in lab-scale experiments and identi ﬁ cation of transformation products was proposed based on the data obtained during analysis by UPLC-QTOF-MS E . The detoxi ﬁ cation potential of the UV/TiO 2 system was investigated using the phosphatase inhibition assay. Suf ﬁ cientamountofhigh-purityOA(25mg, N 90%purity)wasproducedin-houseforuseinphotocatalysisexper-iments by simple reversed-phase ﬂ ash chromatography. Complete degradation of OA was observed in seawater after 30 min and 7.5 min in deionized water. The rate constants ﬁ tted with the pseudo- ﬁ rst order kinetic model (R 2 N 0.96). High-resolution mass spectrometry analysis of the photocatalyzed OA allowed tentative identi ﬁ ca- tionoffourtransformationproducts.Detoxi ﬁ cationwasachievedinparallelwiththedegradationofOAindeion-ized water and arti ﬁ cial ocean water ( ≤ 20 min) but not for seawater. Overall, results suggest that UV/TiO 2 photocatalysis can be an effective approach for degrading OA and their TPs in the marine environment. Tothebestof our knowledge, this isthe ﬁ rstreporton the use of photocatalysis to degrademarinetoxins and its promising potential to protect shell ﬁ sh harvesting sites.


Introduction
Marine toxins are produced by natural phytoplankton and can be accumulated by filter feeding shellfish, meaning that they may enter the food chain and cause toxic episodes in human health. Marine toxins are responsible for N60,000 intoxications per year, with overall mortality of around 1.5% (Kantiani et al., 2010). Based on their resultant illness, marine toxins can be classified in diarrheic shellfish poisoning (DSP) toxins, paralytic shellfish poisoning (PSP) toxins, amnesic shellfish poisoning (ASP) toxic, neurotoxic shellfish poisoning (NSP) toxins, ciguatera fish poisoning (CFP) and clupeotoxin fish poisoning (CLP) toxins (Farabegoli et al., 2018;Gerssen et al., 2010).
In order to minimize economical loss and to protect human health caused by annual DSP outbreaks, monitoring systems have been implemented by local governments. They include the analysis of toxin content in shellfish, the concentration/presence of toxic algae in the water body and the evaluation of hydrographic/meteorological data (Anderson et al., 2019;Campbell et al., 2011;Shen et al., 2012). A current European regulatory limit of 160 μg of OA equivalents per kg of edible shellfish must not be exceed and this limit is currently accepted in several other countries (e.g.: USA, China, Brazil) (EU, 2002/225/EC). Liquid chromatography coupled with mass spectrometry detection (LC-MS) is the technique of choice for the analysis of DSP at trace levels in complex matrices (Dell'Aversano and Tartaglione, 2017). Although major advances have been made in the development of multi-residue analytical methods, unavailability of suitable standards, the high cost of the commercial standards, the presence of several toxin analogues and unknown toxic compounds are an additional challenge.
OA has been detected in nearshore seawater across the globe at concentrations in the ng/L range (Bosch-Orea et al., 2017;Fux et al., 2009;Hattenrath-Lehmann et al., 2018;He et al., 2020;Reguera et al., 2014). Even high concentrations of OA up to 1.78 μg/L in seawater and up to 560 μg/g in particulate matter were reported (Bosch-Orea et al., 2017). OA proved to be very stable in seawater, interstitial water and sediment Pizarro et al., 2009), thereby seriously threatening aquaculture industry and human health. Because of this, it was hypothesized that high concentrations of OA in seawater would indicate recent or past blooms that could last even a year .
There has been an increased interest in finding suitable techniques to remove these toxins from the marine environment. Advanced oxidation processes (AOPs) are widely used to degrade organic pollutants through free radicals as reviewed by (Ibhadon and Fitzpatrick, 2013;Matafonova and Batoev, 2018). Of different AOPs, heterogeneous photocatalysis with TiO 2 has been demonstrated to be an efficient method to remove freshwater toxins (e.g. cyanotoxins) (Antoniou et al., 2018;Liu et al., 2009;Pestana et al., 2015;Pinho et al., 2015). Major improvements have overcome the barriers to the application of this technology to complex matrices. For example, the feasibility of UV-LEDs as an alternative low energy source in complex environments (Schneider et al., 2019), the use of solar light activated materials to reduce application costs (Kinley et al., 2018), the addition of oxidants to reduce energy requirements (Antoniou et al., 2018) or the form of the catalytic material to recover it after treatment (e.g. fixed to a film, coated surfaces) (Kinley et al., 2018;Pestana et al., 2015;Shephard et al., 2002). However, this is still an understudied area regarding marine toxins (Djaoued et al., 2008;Khan et al., 2010).
This work aimed i) to produce cost effective amounts of high-purity DSP toxins from P. lima for photocatalysis experiments; ii) to evaluate the performance of a UV/TiO 2 system to degrade/mineralize OA in seawater; iii) to identify possible transformation products (TPs) generated during photocatalysis; iv) to evaluate the detoxification efficiency of UV/TiO 2 system for OA in seawater.
2.2. Extraction and purification of DSP compounds from P. lima P. lima CCAP 1136/11 (Culture Collection for algae and Protozoa, Oban, Scotland) was grown on seawater enriched with f2-Si medium (10 × 4 L medium in 10 L round Pyrex flasks; Guillard and Ryther, 1962) at 22 ± 1°C with continuous light at 20 μmol/m 2 /s. Cultures were sparged with sterile air at 2.3 L/min and grown for 35 days (Praptiwi, 2014). Harvesting was achieved by sedimentation of cells and removal of the supernatant, concentrated cells were stored at −20°C. Concentrated frozen cells obtained from 240 L cultures (approx. 5 × 104 cells/mL) were defrosted and combined to give a concentrated cell slurry (2 L). This cell slurry was combined with methanol and water at a ratio 1:1:3 (v/v/v) (culture-methanol-water) and loaded onto a SNAP C18 120 g cartridge (Biotage) to concentrate the toxins after tangential flow filtration (Biotage, Hengoed, UK). This process was repeated twice with the cell residue. Toxins were fractionated using a Biotage Selekt Flash Purification System (Biotage, Hengoed, UK). The mobile phase had a flow rate of 30 mL/min and a linear gradient from 10 to 100% MeOH over 90 min. Fractions (100 mL) were automatically collected using UV detection (220-254 nm). DSP content in each fraction was analysed by ultrahigh performance liquid chromatography coupled to quadrupole-time of flight mass spectrometry (UPLC-QTOF-MS E ). OA containing fractions were pooled, dried down in a centrifugal evaporator (Genevac, Ipswich, UK) and stored at −20°C.

Photocatalysis study
Photocatalysis experiments were conducted in a lab-scale photoreactor ( Fig. 1) equipped with 4 UV lamps (Philips PL-L 36W/09/ 4P UVA; wavelength 315-380 nm). 10 mL of filtered deionized water (DW), artificial ocean water (AOW) and seawater (SW) (collected in May 2019 from Stonehaven, Aberdeenshire, UK) spiked with OA (10 μg/mL prepared in DW), isolated from the P. lima culture, were maintained under magnetic stirring at 20-22°C for 60 min. Experiments were run in triplicate in the dark and under UV irradiation in the presence and absence of the catalyst TiO 2 (0.1%, w/v). The experiments were performed at pH 6.9 in DW, 7.3 in AOW and 7.8 in SW. The suspension was kept in the dark under stirring during 5 min (adsorption-desorption phase) and afterwards the UV lamps were switched on. At selected time intervals (5 min before UV,0,2.5,5,7.5,10,12.5,15,20,30 and 60 min) samples (0.5 mL) were collected and filtered through a 0.45 μm filter before UPLC-QTOF-MS E analysis.

Analysis by UPLC-QTOF-MS E
OA and related compounds were quantified by UPLC-QTOF-MS E (Waters, UK) equipped with an electrospray ionization (ESI) source. Separation was carried out on a C18 BEH column (100 × 2.1 mm, 1.7 μm particle size) at 40°C. Mobile phase was acetonitrile with 0.1% FA (B) and water with 0.1% FA (A) at a flow rate of 0.2 mL/min. Gradient elution was as follows: 20% B initial conditions rising to 70% B at 9.50 min, increasing further to 100% B at 10 min, holding until 11 min, dropping back to 20% B at 12 min and holding until 15 min. Injection volume was 5 μL. Samples were kept at 5°C.
The Q-TOF was operated in positive ESI mode using MS E acquisition, that allows both precursor and product ions to be acquired simultaneously. Parameters were: 3.2 kV capillary voltage, 40 V cone voltage, 80°C source temperature, 350°C desolvation temperature, 50 L/h cone gas flow, 600 L/h desolvation gas flow. Argon was used as the collision gas. MS E consisted in 3 functions: the first function used low collision energy 6 V, the second function used a collision energy ramp of 20-50 eV and the third function acquired the lock mass data for online mass calibration. Sodium formate was used as calibration solution over a mass range of m/z 50-2000. Leucine enkephalin (m/z 556.2771 for positive electrospray mode) was infused at a flow rate of 20 μL/min at 10 s intervals to correct for mass drifts. Acquisition and processing of MS data was done using MassLynx v 4.1 software (Waters).

Protein phosphatase inhibition assay
Phosphatase enzymes are able to hydrolyze pNPP into pnitrophenol that can be measured by absorbance detection. The colorimetric assay was performed as described in Sotoud et al. (2013) with some modifications. Briefly, a stock solution of 5 mM pNPP was prepared freshly in a buffer (pH 8.0) containing 50 mM Tris-HCl, 20 mM MgC 2 and 0.2 mM MnCl 2 . The PP1c from rabbit (specific activity 5000-15,000 U/mg protein) was prepared in a buffer (pH 7.4) containing 50 mM Tris-HCl, 1 mg/mL bovine serum albumin (BSA), 1 mM MnCl 2 and 2 mM dithiothreitol. One unit hydrolyses 1 nmol of pNPP per minute at pH 7.4 at 30°C according to the manufacturer's instructions (Sigma-Aldrich, UK). The assay was conducted in flat-bottom 96-well plates. Ten μL of OA (0.05-50,000 nM) or 10 μL of each sample was placed in each well in triplicate and 10 μL of PP1 solution was added. After 4 min incubation time at room temperature the reaction was initiated by addition of 180 uL of pNPP stock solution. The plate was incubated at 37°C immediately. PP1 activity was determined by measuring the absorbance at 405 nm at 4 min intervals for 40 min using a microplate reader (Epoch, BioTek, UK) and the Gen5 software for data acquisition (BioTek, UK). IC 50 determinations were calculated using fourparameter, variable slope, non-linear regression analysis (GraphPad Prism, USA).

Toxicity prediction
The toxicity of OA and its TPs was predicted by using the Toxicity Estimation software Tool (TEST, v. 4.2.1) from the US Environmental Protection Agency (USEPA, 2012). This software predicts toxicity values using mathematical models of Quantitative Structure Activity Relationship (QSAR) (Martin, 2016). In this work, the 96 h fathead minnow 50% lethal concentration (LC 50 ), the 48 h Daphnia magna LC50, the bioaccumulation factor and the developmental toxicity were chosen as toxicological endpoints. Consensus method was selected as toxicity is estimated by taking an average of the predicted toxicities from five different QSAR methodologies (Hierarchical clustering, Single Model, Group Contribution, the Food and Drug Administration (FDA) and Nearest Neighbor).

Screening and identification of DSP compounds in Prorocentrum lima culture
Many marine toxins have very complex structures that cannot or are prohibitively expensive to be synthesised. On the other hand, the high cost of commercially available toxin standards (naturally produced) make lab-scale studies requiring high amounts (≈3-5 mg) of toxin unfeasible. Several benthic species of the genus Prorocentrum have been confirmed to produce DSPs. Therefore, the culture P. lima was assessed as a viable source of DSP toxins, in particular OA, for the photocatalysis experiments.
As a preliminary screening, extract of P. lima culture was analysed using high resolution mass spectrometry in positive mode (Fig. 2). Positive mode MS E fragmentation showed the presence of OA (R t 9.20 min) and DTX1 (R t 11.15 min) in the culture of P. lima. Identification was done by comparison of retention time and fragmentation patterns of commercially available standards of OA and DTX1 (Fig. S2) with the samples. The low energy (LE) spectrum (R t 9.20 min) of OA in the P. lima culture (Fig. 2) (Paz et al., 2007;Quilliam, 1995). The LE spectrum of DTX1 (R t 11.15 min) shows the protonated ion m/z 819.4935, the Na + adduct at m/z 841.4717, the first loss of H 2 O at m/z 801.4843 and the multimer Na + adduct at m/z 1660.9813. The HE (R t 11.16 min) spectrum shows the Na + adduct the consecutives losses of H 2 O at m/z 801. 4865, 783.4685, 765.4592 and 747.4487, and the fragments 267.1255, 301.2088, 401.2854, 445.3135, 489.3397, 533.3653, 577.3853, 621.4148. In addition, a peak coeluted 0.3 min later than DTX1. The LE spectrum (R t 11.20 min) of the coeluting peak showed a signal at m/z 1057.7203 and 819.4976. No standard was available for comparison but the mass spectra suggested that the coeluting compound is an analogue derived from DTX-1 and maybe a novel diol ester.  (Carey et al., 2012;Paz et al., 2007;Quilliam, 1995).
Overall, 2 L of concentrated cells contained ≈84 mg of OA, ≈23 mg of DTX1 and ≈17 mg of the analogue derived from DTX1. Toxins were fractionated using a Biotage Selekt Flash Purification System and 28 fractions were obtained. OA was identified by UPLC-QTOF-MS E in fractions 16-19 (60-70% MeOH content), DTX1 in fractions 18-21 (67-77% MeOH content) and DTX1 analogue in fractions 21-23 (77-83% MeOH content). Fraction with the highest purity and content of OA was 17 (N90%, 25.3 mg), for DTX1 fraction 19 (N35%, 11.6 mg) and for DTX1 analogue fraction 22 (N45%, 13.2 mg). Prorocentrum species produce different DSP content depending on the species and/or strains Aquino-Cruz et al., 2018;Ben-Gharbia et al., 2016;Lee et al., 2016;Luo et al., 2017;Nishimura et al., 2019;Nishimura et al., 2020;Pan et al., 2017). Previous studies on largescale culture of P. lima and separation and purification of OA and DTX1 were published. For example, Wang et al. (2015) developed a vertical flat plate photobioreactor in which they obtained 15.2 mg/g of OA and 21.6 mg/g of DTX1 from a 60 L culture (3.2 g dry weight) of P. lima. Next, Chen et al. (2017) purified these toxins to obtain milligrams of OA and DTX1 of N80% purity using a macroporous resin column and high-speed counter-current chromatography-MS. Further purification with semi-preparative HPLC-MS achieved N98% purity. Although improved purity is obtainable as described by Chen et al. (2017), in our study, the culture of P. lima fit the purpose of producing sufficient amounts of OA (≈84 mg) that can be easily isolated and purified (N90%) to be used in photocatalysis studies using a more simple extraction and purification process.    (Fig. S3). Heating to 45°C also resulted in no losses of the toxin. OA was found to be very stable. A similar outcome was reported in previous studies. Blanco et al. (2018) reported the high stability of OA in seawater, interstitial water and sediments over a 23-day period. They said that even assuming the highest degradation rate, half of the toxin would remain in water after more than one year after a Dinophysis bloom.

Close-up
As a result of the persistence of the free OA toxin in water bodies, lab-scale photocatalysis (UV-TiO 2 ) experiments were performed to explore the effectiveness of this technique to degrade the toxin and subsequently its potential application in the marine environment.
Changes in concentration of OA with different exposure times were plotted in Fig. 3A-B. Results for dark (Fig. 3A) and photolysis (UV irradiation only) (Fig. 3B) indicate that there was little to no reductions in concentrations under experimental conditions after 60 min, which indicates that OA was not adsorbed onto the catalyst or the reactor vessel and that OA is resistant to UV degradation (OA removal: ˂5% in AOW and ˂15% in DW and SW). pH greatly affects the ionization of OA and the surface charge of TiO 2 which result in a higher or lower adsorption capacity. Within the pH range of the samples here used (6.9-7.8) both OA (pKa 3.87) and the TiO 2 surface (point of zero charge is pH 6.5) carry negative charges and adsorption on the catalyst surface is suppressed. On the other hand, it was found that OA was completely removed after 7.5 min, 12.5 min and 30 min in DW, AOW and SW, respectively using UV/TiO 2 0.1% w/v (Fig. 3C).
The kinetics of the OA degradation in AOW and SW fitted well to the pseudo-first-order model (R 2 N 0.96) (Fig. 3D): ln(C o /C t ) = k t; where C o and C t are the concentrations of OA in the solution at time 0 and t (min), respectively, k is the rate constant (min −1 ). The rate constant was 0.437 min −1 in AOW and five times lower in SW (0.088 min −1 ). In the case of DW, due to the fast degradation of OA, there was only three time points to calculate the rate constant so the uncertainty of the results is high. The rate constant in DW was 0.498 min −1 at with this pseudo-first-order model but the R 2 was 0.70 (Fig. 3D). Similar results were obtained when the initial concentration of OA was 10 times lower (1 μg/mL) in DW (0.568 min −1 , R 2 N 0.72). Interestingly DW and AOW showed similar rate constants, despite the amount of salts present in AOW. In photocatalytic processes the presence of ions has been reported as a negative effect, at concentrations as low as 10 −3 M, due to competitive adsorption at the active sites of the catalyst (Mills and Le Hunte, 1997) or acting as radical scavengers (Bennedsen et al., 2012). However, this effect was not observed in this study. There are no previous reports about the photocatalysis effectiveness of UV/TiO 2 on OA in seawater for comparison. The slower rate constant in SW compared to AOW might be due to the presence of organic matter in the matrix. As it has been reported previously, the nature and concentration of organic matter in the matrix can inhibit or promote photocatalytic reactions (Li and Hu, 2016). In the study of Muff et al. (2017) the influence of the seawater matrix on the photolytic and photocatalytic degradation of an organotin pesticide (tributyltin) was evaluated. They observed a reduction of 41% in seawater compared to demineralized water. Moreover, the photocatalytic TiO 2 surface was inactivated and produced radicals were scavenged by the relatively high salinity and content of organic matter in the water.
Barriers to real-life application of this technology remain but recent improvements are overcoming them (e.g. UV-LED and solar irradiation as energy sources, new catalyst supporters, modifications of the catalyst to increase the photocatalytic activity). Pestana et al. (2014Pestana et al. ( , 2020 demonstrated the feasibility of the technology in the field. A packed-bed flow-through UV photocatalytic reactor with pelletised TiO 2 successfully removed freshwater cyanotoxins and improved overall quality of waste lagoon water (Pestana et al., 2020) as well as it removed trace water contaminants (geosmin and 2-methylisoborneol) from water in a fish-farm raceway (Pestana et al., 2014).

Formation of TPs by UV/TiO 2
Transformation products (TPs) were formed and removed during the photocatalytic degradation of OA using UV/TiO 2 . Overall, four peaks were observed at t R 7.2, 7.5, 8.1 and 10.2 min. Time-dependent evolution profiles of TPs during UV/TiO 2 processes are shown in Fig. 4. The peak area of each TP was normalized to the initial peak area of OA for semi-quantification due to the lack of analytical standards for their quantitation. TPs arised at the beginning of irradiation and they were found at 2.5 min in all three matrices but were removed after 10 min of irradiation in DW, 20 min in AOW and they were present after 60 min in SW. The peak at 10.2 min was found the most abundant during the photocatalytic process, especially in AOW and SW. This peak was found at its maximum level at 2.5 min in DW and AOW and was removed after 20 and 10 min, respectively. It took longer to reach its maximum level in SW (10-20 min) and there was still 35% present after 60 min of UV irradiation.
Identification of TPs was based on the use of accurate mass measurements and MS E fragmentation (Fig. S4). The peak at 10.2 min with a protonated m/z of 759.4781 differs in 45 Da with OA (m/z 805.4775). The tentative identification points to norokadanone generated by the decarboxylation of OA. The spectrum (Fig. S4)  that indicates that they could be isomers of DTX1 (Fig. S4) Paz et al., 2007).

Toxicity evaluationphosphatase inhibition assay
Finally, it is important to determine if the photocatalytic system was able to detoxify the seawater. OA and DTxs are potent serine/threonine protein phosphatase inhibitor, for that purpose, the phosphatase inhibition assay was carried out.
The dose-dependent kinetic activity of PP1 in the presence of pNPP (5 mM) is shown in Fig. S5. The optical density (OD) increased linearly with increasing phosphatase concentrations (Fig. S5A) and with time at a fixed concentration of PP1 (5 μg/mL) (Fig. S5B). The concentrationinhibition curve of OA (0-50,000 nM) in the presence of PP1 (5 μg/mL) is shown in Fig. S5C. Under these conditions, the IC 50 value of OA was 1183 nM, which is higher than the ones reported earlier (IC 50 3.6-315 nM) (Bialojan and Takai, 1988;Huhn et al., 2009;Twiner et al., 2016). IC 50 values are difficult to compare because they depend, among others, on the amount of enzyme and the substrate concentration.
The potential toxicity of the samples before and during the heterogeneous photocatalytic degradation of OA was evaluated by the phosphatase inhibition assay. Results are presented as percentage of activity of PP1 in Fig. 5. A 10 μg/mL OA solution exhibited an activity value of 28% in DW, 13% in AOW and 34% in SW. The activity rose as the concentration of OA decreased. It is noteworthy that the highest toxicity was observed at the beginning of the experiment where OA was present at its maximum concentration. According to the above, the higher toxicity at t 0 could not be ascribed to the formation of toxic intermediates which appeared afterwards. At 10 and 20 min, complete detoxification of the irradiated solution was achieved in DW and AOW, respectively, demonstrating the efficiency of heterogeneous photocatalytic oxidation in the elimination of acute toxicity of OA under the investigated conditions. On the other hand, at 60 min, PP1 activity was 64% in SW when the investigated compounds were removed after 30 min. Findings from Twiner et al. (2016) demonstrated that small variations at the head region (i.e., C1/C2) of the OA/DTX structures results in significant changes in the phosphatase inhibitory potency. On the other hand, changes in methylation at the tail region (C31 and C35) only have mild effects in toxicological potency. In the present study, toxicity was also predicted by using the Toxicity Estimation Software Tool (TEST, version 4.2.1) from the US Environmental Protection Agency (USEPA, 2012). Compared to OA, the toxicity of DTX1 and TPs was lower.
It is important to note that not all the DSP toxins, even when they are structurally related, show the same biological negative effects and the mode of action is still not clear. For example, it was reported that phosphatase inhibition was responsible for the diarrhetic events (Cohen et al., 1990). However, other reports showed that other actions, such as changes in cytoskeletal elements (Creppy et al., 2002) and or neurotransmitters (Louzao et al., 2015) might be associated to gastrointestinal effects. Methyl okadaate has a higher potency than OA to disrupt the cytoskeleton and it is a non-phosphatase inhibitor (Espiña et al., 2010). Therefore, mechanism of toxicity of DSP toxins must be reevaluated (Abal et al., 2018;Botana et al., 2016;Valdiglesias et al., 2013).

Conclusions
Prorocentrum lima proved to be a viable source of DSP toxins, in particular OA. Two liters of concentrated cells contained 84 mg of OA, 23 mg of DTX1 and 17 mg of the analogue derived from DTX1. OA could be easily isolated and purified. This is the first study that assessed the effectiveness of UV/TiO 2 system to degrade OA. The marine toxin was completely degraded after 30 min in seawater and even faster in deionized water (7.5 min). The degradation follows a pseudo-first order kinetic type. Four TPs, norokadanone and three possible isomers of DTX1, were identified in the photocatalytic UV/TiO 2 system. Detoxification was parallel to OA degradation in deionized and artificial ocean water but not for seawater.
Overall, results suggest that UV/TiO 2 photocatalysis can be an effective approach for degrading OA and their TPs in the marine environment. Further research should be conducted to look for less energy demanding system that could be easily scaled up. It would be worthy to explore: (i) UV-LED as they can provide energy saving, long lifetime, and environment-friendliness (Hg free); (ii) doped TiO 2 ; (iii) different supporters for the catalyst; (iv) solar light activated materials; (v) degradation of intracellular and extracellular toxins.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.   Figure Figure S5. Dose dependent kinetic activity of PP1 in the presence of pNPP (5 mM) (A, B).
Optical density (OD) was measured at 405 nm. Effect of OA concentration on PP1 (5 µg/mL) (C). Three-parameter, variable slope, non-linear dose response analysis was performed and calculated concentration of inhibition at 50% (IC50). Data is expressed as mean values and SD of n=3.