Sustainable synthesis of omega-3 fatty acid ethyl esters from monkfish liver oil

: The search for economical and sustainable sources of PUFAs within the framework of the circular economy is encouraged by their proven beneficial effects on health. The extraction of monkfish liver oil (MLO) for the synthesis of omega-3 ethyl esters was performed evaluating two blending systems and four green solvents. Moreover, the potential solubility of the MLO in green solvents was studied using the predictive simulation software COSMO-RS. The production of the ethyl esters was performed by one or two step reactions. Novozym 435, two resting cells ( Aspergillus flavus and Rhizopus oryzae ) obtained in our laboratory and mix of them were used as biocatalysts in a solvent-free system. The yields for Novozym 435, R. oryzae and A. flavus in the one-step esterification were 63%, 61% and 46%, respectively. The hydrolysis step in the two-step reaction led to 83%, 88% and 93% of free fatty acids (FFA) for Novozym 435, R. oryzae and A. flavus respectively. However, Novozym 435 showed the highest yield in the esterification step (85%) followed by R. oryzae (65%) and A. flavus (41%). Moreover, selectivity in front of polyunsaturated fatty acids of R. oryzae lipase was evidenced, since it did slightly esterified docosahexaenoic acid (DHA) in all the esterification reactions tested.

The fatty acid content of marine fish, whether oily or white fish, is high in omega-3 polyunsaturated fatty acids, as these are synthesized by microalgae and they reach the fish through the food chain [10]. According to the Apromar report on aquaculture and fisheries in Spain 2019 [11], the EU consumed 13 million tons of aquatic products in 2018. Depending on the type of product the edible part, range 40 to 80%, generating by-products that can reach up to 60% of the products consumed [12]. Therefore, these by-products or residues can be used as starting materials to prepare

Monkfish liver oil extraction
The solid-liquid extraction of oil from fresh monkfish liver using five different solvents was performed by maceration using roller mixer and ULTRA-TURRAX® systems. Folch reagent [19] was used until exhaustion (until no colour was observed in the solvent) to determine the maximum content of oil in the liver. A percentage of 39.0% w/w was the maximum content of oil in the fresh material, which contents 49.8%of moisture. This is the first determination of the oil content in monkfish liver as far as we know. This value is higher that the percentage reported for tuna liver with an oil yield of 17.5 % [32], or from salmon by-products (head, frame and viscera) with an oil content ranging 13.09 to 19.2 % [33]. Ciriminna et al. [34] reported 1.5% of oil content from anchovy heads.

Extraction in the two systems: Roller Mixer (RM) and ULTRA-TURRAX® (UTs)
The results presented in Table 2 showed that the extraction yields were higher when the RM was used. All green solvents showed extraction yields between 96 and 100% of the maximum oil content in monkfish liver using RM. A single extraction using FR solvent yielded 89% of oil using RM. Authors such as Fang et al. also reported the importance of this variable in their tuna oil extraction experiments [31]. The lower extraction yields obtained with some green solvents, as well as with the FR when UTs was used could be explained by the formation of emulsions due to the presence of water in the fresh monkfish liver and by the type of agitation and the high speed (4000 rpm) used in this system. 2-MeTHF and CPME extracted the highest quantity of fish oil in the two extraction systems evaluated, which coincide with the COSMO-RS prediction. In the theoretical study DCM showed the lower probability of oil solubility; however, in the experimental extraction it showed better results than expected, e.g. 99% in RM. In the case of LMN presented worst extraction yields than those predicted by COSMO-RS. Nevertheless, this experimental result it can be due to the harsh conditions used to eliminate this solvent (90⁰C/ 0.3 mbar). LMN was the most difficult to evaporate of all tested solvents. Considering the results obtained using the RM system, it can be said that the evaluated green solvents could be interesting alternatives to replace conventional solvents such as hexane, chloroform and methanol to develop more environmentally friendly extractive processes for oil samples [21,22,23].

Analysis of the extracted oil
In the MLO 1 H-NMR spectra, signals corresponding to omega-3 fatty acids between 0.99 and 1.1 ppm were observed, in addition to other signals corresponding to PUFA, MUFA and saturated fatty acid (SFAs) [35,36]. Specific signal of DHA were observed at 2.4 ppm corresponding to the hydrogen bonded to both the α-carbon (=C-C-CH2-COOR) and the allyl-carbon (=C-CH2-C-COOR) [37]. No differences were observed between the spectra of oils obtained with different solvents or extraction methods (RM or UT).
On the other hand, the fatty acid profile of the extracted oils was determined by GC-FID (Table  3). MLO resulted to have 32.5% of SFAs, 42.1% of mono-unsaturated fatty acids (MUFAs) and 27.3% of PUFAs. According to the fatty acid profile, oleic acid is the major fatty acid with 21.1% of the total, with beneficial effects on human health as the omega-3 fatty acids [38]. Furthermore, the presence of 5.4% of vaccenic acid, an omega-7 isomer of the oleic acid, is worth to consider due to its associated with a low risk of cardiovascular disease [39,40]. Other MUFAs present are gadoleic acid (C20:1n9), which is characteristic of fish oils, and erucic acid (C22:1n9), which has also been reported that is found in fish liver [32,41]. DHA and EPA are the main PUFAs in monkfish oil with 15.2 and 4.1 % of the total, respectively. The oil content 39% w/w regarding to fresh monkfish liver and its composition confirms the potential that this by-product has as raw material to obtain products with high added value. The high content of these was expected since white fish tend to store the omegas 3 and 6 in the liver and not in the muscles as blue fish does. Table 3. Fatty acid profile (%) of monkfish liver oil.

Enzymatic preparation of fatty acid ethyl esters (FAEEs).
Looking for the best reaction conditions for the production of FAEEs using MLO and ethanol, two different procedures were performed: a) a one-step procedure (a transesterification reaction), and b) a two-step procedure through hydrolysis and esterification reactions. To catalyse these reactions a commercial enzyme (Novozym 435), two own resting cells (R. oryzae and A. flavus) and two mixtures of these resting cells (1:1 and 7:3, R. oryzae-A. flavus) were used in a solvent-free medium.
The results of the one-step (transesterification) and two-step reactions (hydrolysis and esterification) carried out are shown in Table 4. In the transesterification and esterification reactions an increase in the yield was observed with increasing reaction time. The commercial enzyme showed the highest yield in the transesterification reaction. Yields were 44%, 61% and 63% for 24 h, 48 h and 72 h, respectively. Transesterification yield with R. oryzae was higher (53%) for 24 h although the yield for 72 h was slightly lower (61%). A. flavus showed the lowest yields (46 % for 72 h). Therefore, the resting cells from A. flavus cannot be considered an alternative to the commercial enzyme for this transesterification reaction. Moreover, two mixtures of the fungal resting cells (1:1 and 7:3, R. oryzae-A. flavus) were used for the transesterification reaction. Only the 7:3 mixture led to a 57% yield after 72 h of reaction, a lower yield than that achieved using only R. oryzae. Considering these low yields, the fatty acid ethyl esters profiles of these reactions were not determined. The main FAEEs obtained in the one-step reaction studied using the three biocatalysts is shown in Figure 2. The commercial enzyme and the fungal resting cells led to different polyunsaturated fatty acid ethyl esters (PUFAEEs) contents. The commercial enzyme esterified 90% of DHA and 100% of EPA regarding to the total content of the corresponding fatty acid in the MLO. This result showed that this enzyme does not discriminate between the different fatty acids to synthetize the FAEEs. Similar results have been already describer for this commercial enzyme, [28,29] which indicate that it is a suitable biocatalyst for the synthesis of these omega-3 EEs. In contrast, the resting cells showed lowest yields for the esterification of PUFAs, mainly DHA. A. flavus showed the highest yield of DHA-EE (38%) of the two resting cells studied. R. oryzae only esterified 22% of DHA present in the fish oil after 72 h reaction. No differences were observed for the esterification of EPA. Therefore, it suggests that lipases from R. oryzae discriminated between the different PUFAs present in the monkfish liver oil. The selectivity of enzymes has been reported by different authors who suggest that some lipases can be selective for some type of fatty acid depending on their chain length, the types of solvents used in its extraction and purification, its method of immobilization or the reaction conditions (temperature, time or substrate) [42,43,44]. The selective synthesis showed by the lipases from R. oryzae could facilitate the separation of the DHA from the mixture of FAEEs.

Two-step synthesis: hydrolysis and esterification reactions
Hydrolysis, the first step of this study, was carried out using the three previous enzymes and two mixtures (1:1 and 7:3) of the two resting cells for 24 h (Fig. 3). The resting cells of R. oryzae mixed with A. flavus in a 7:3 ratio showed the highest percentage of hydrolysis (> 95% free fatty acid (FFA)). The second biocatalyst presenting a high hydrolysis percentage was the resting cells of A. flavus (> 93% FFA). All resting cells studied showed a higher percentage of hydrolysis than the commercial lipase (83%), indicating that the resting cells can be a cheap alternative to immobilized commercial biocatalysts for these reactions. These hydrolysis percentages can be considered high when compared to those reported [45] for the hydrolysis of fish oil with the lipase of Cryptococcus sp., which yielded 25% and 66.5% of FFA for 24 h and 72 h, respectively. Furthermore, it is worth to point out that monkfish liver oil (MLO) contains more DHA than hydrolysed cod, sardine, salmon and shark liver oils (Fig. 4). A scale up of the hydrolytic process was carried out using the best reaction conditions achieved in the previous experiment. Starting from 25 g of fish oil, a mixture of R. oryzae and A. flavus (7:3) allowed the preparation of hydrolysed monkfish liver oil (HMLO) with 97.8% of FFA. The esterification reaction was performed using the hydrolysed monkfish liver oil (HMLO) obtained in the previous experiment. In these experiments, a positive correlation was found between reaction times and esterification performance as in the transesterification. The commercial enzyme showed the highest yield in the esterification reaction. Yields were 54%, 70% and 85% for 24 h, 48 h and 72 h, respectively. Esterification yield with R. oryzae was 65% for 72 h. A. flavus again showed the lowest yields (41 % for 72 h). These results confirm that A. flavus resting cells are not very active in the esterification of the fatty acids from monkfish oil. Regarding to the mixtures of the fungal resting cells, the yields were lower than the ones achieved with R. oryzae as was in the transesterification reaction (see Table 4).
In relation to the FAEEs profiles (Figure 4), these are similar to those shown in the one-step reactions. There has been an enrichment of MUFAs such as oleic acid, gadoleic acid and vacenic acid, which are beneficial to health [38]. Vacenic acid is a source of omega-7 that has been positively correlated with the presence of DHA and EPA [39,40]. The increase of palmitic acid and stearic acid observed in the assays could be due to the fatty acids present in the sunflower oil used in the production of the resting cells as lipase inductors. The microorganisms themselves can also synthesize different fatty acids, Aspergillus sp produces long chain fatty acids (C16:0, C16:1n7, C17:0, C18:0, C18:1n9, C18:2, C18:3 and C20:0). This fatty acid profile has been used to discriminate fungal species belonging to Aspergillus genus [46]. In addition, the presence of some fatty acids such as gadoleic acid (C20:1n9) may depend on the type of substrate used in the culture medium [47]. Nevertheless, the increase of these fatty acids in the final crude of the reaction cannot be higher than 4% considering the percentage of biocatalysts used (10%) and the percentage of fatty acids present in the resting cells (40%).
EPA behaved similarly to one-step reactions, in which the commercial enzyme esterified up to 100% of EPA contained in HMLO and R. oryzae esterified 70% and A. flavus esterified 60%. For DHA, Novozym 435 esterified up to 90% of the DHA contained in HMLO, while the fungal resting cells did not exceed 20%. A. flavus lipase esterified up to 19% of the DHA content, while R. oryzae lipase esterified up to 16%. These low yields seem to confirm the hypothesis of DHA selectivity shown by R. oryzae. Because the selectivity of R. oryzae lipase, DHA should be found as free fatty acid within the esterified material. The possibility of isolating these free fatty acid is also interesting as it has been shown that PUFAs ingested in the form of free fatty acids may be more bioavailable than FAEEs and therefore more assimilable in human metabolism [48].  To confirm the above-mentioned hypothesis and confirm that DHA is present as an acid in the esterified fraction, these samples were subjected to total esterification through chemical catalysis (derivatisation process) using H2SO4. Figure 5 compares the DHA-EE obtained in the one and two-step reactions at different times (24,48 and 72 h) using R. oryzae or applying chemical catalysis. As expected, chemical catalysis achieved more than 90% esterification of the DHA contained in both monkfish liver oil and the hydrolysed monkfish liver oil. In the three reaction periods studied it was evidenced that the lipases from R. oryzae were able to effectively discriminate between the DHA and the rest of the fatty acids present in MLO and HMLO. In the HMLO the DHA must be present as free acid since a 97.8 of hydrolysis was achieved.
The selectivity demonstrated by the resting cells of this strain of R. oryzae has not previously been reported to our knowledge. However, Ashjaria et al. [49] studied the selectivity in the hydrolysis of fish oil of lipases isolated from R. oryzae and immobilized by different methods. All immobilized biocataysts discriminated between EPA and DHA in favour of EPA.

Computational method
This study has been performed by a theoretical procedure using a computational predictive method (COSMO-RS), by considering the technical properties of the solvents and via experimentation. The comparison was made considering the amount of monkfish liver oil extracted and the technical parameters of the solvents used.

COSMO-RS procedure
The Conductor-like Screening Model for Real Solvents (COSMO-RS) developed by Klamt and co-workers [50] is as known as a powerful method for molecular description and solvent screening based on the result of quantum chemical calculations for an understanding of the dissolving mechanism. COSMO-RS combines quantum chemical considerations (COSMO) and statistical thermodynamics (RS) to determine and predict the thermodynamic properties without experimental data. In the first step, the molecule is embedded into a virtual conductor. In such an environment the molecule induced a polarization charge density on its surface (σ-surface). Calculation was performed for each molecule of interest (Fig. 1). The second step used the statistical thermodynamic calculation. Blue is used to represent strongly positive polar regions, and red represents very negative polar surfaces. Green and yellow correspond to lower polarity. The thermodynamics of the molecular interactions that were based on the obtained σ-profile are then used to calculate the chemical potential of the surface segment (σ-potential) [21,23]. According to the composition of the monkfish liver oil determined by 1 H-RMN and GC-FID, we decided to use as TAG models in the COSMO-RS analysis four different models of TAG using the four main fatty acids from the oil.
In this work, the prediction of the relative solubility of the main triacylglycerides from monkfish liver oil in four green solvents 2-MeTHF, CPME, DMC and LMN and FR, that is, a mixture of chloroform and methyl alcohol (2:1, v/v) was made by implementing this COSMO-RS model in COSMOtherm software (BIOVIA COSMOthermX19; Dassault Systèmes, France). The chemical structures of the solvents and solutes were mutually transformed into their simplified molecular input line entry syntax (SMILES) notations, which were subsequently used to calculate the solubility parameters of solvents and compounds. The relative solubility is calculated from the following equation [50,51]. Relative solubility is always calculated in infinite dilution. The logarithm of the best solubility is set to 0 and all other solvents are given relatively to the best solvent. A solvent with a log10 (x_j) value of -1.00 yields a solubility which is decreased by a factor 10 compared to the best solvent. Also, the logarithm was transformed into the probability of solubility and was expressed in percentage.

Extraction in a roller mixer
The monkfish liver oil was extracted by solid-liquid extraction with FR (chloroform-methanol 2:1 v/v) as a reference solvent and using four green solvents (2-MeTHF, DMC, CPME and LMN). An amount of 20 g of monkfish liver previously defrosted and grinded were extracted with 100 mL of each solvent at ratio 1:5 w/v. The mixture was stirred at 60 rpm for 30 min in a roller mixer. The sample was filtered through paper, and the solid residue was washed two times with fresh solvent. Solvents were joined. A solution of 1% v/v of NaCl was added to the extract obtained with FR, the mixture was shaken vigorously and left to stand for 2 h. The organic phase was recovered and dried with anhydrous sodium sulphate. It was filtered and the solid was washed with chloroform. Final solutions were evaporated under vacuum in a rotary evaporator. Samples were dried under vacuum for 2 h and finally were stored at -20°C until analysed. All experiments were carried out in triplicates.
To perform a better comparison of the amount of monkfish liver oil obtained by every solvent, we determined the maximum content of oil in the monkfish liver by extracting with the FR one sample five times (until no colour was observed in the solvent). Solvents were joined and processed as indicated above. 4.3.2. Extraction in an ULTRA-TURRAX® system.
An amount of 2 g of ground monkfish liver, 2 g of ceramics balls and 10 ml of the solvents (green solvents or FR) were added to the ULTRA-TURRAX system. The mixture was mechanically stirred at 4000 rpm for 15 min at room temperature. The mixture was then filtered and centrifuged at 5000 rpm for 5 min to separate the supernatant. The organic phase was recovered and dried with anhydrous sodium sulphate. The mixture was filtered and then the solvent was evaporated in a rotary evaporator and the oil was dried under vacuum. The samples were stored at -20°C until analysis. All experiments were carried out in triplicate.

Preparation of resting cells
R. oryzae and A. flavus cells were grown in a synthetic liquid medium containing 2 g of asparagine, 1 g of K2HPO4, 0.5 g of MgSO4, 5 mg of thiamine hydrochloride, 1.45 mg of Fe (NO3)3·9H2O, 0.88 mg of ZnSO4·7H2O, and 0.235 mg of MnSO4·H 2O per liter of distilled water. The initial pH of the medium was adjusted to pH 6.0. Next, 250-mL aliquots of the medium were sterilized at 121°C for 15 min, and 1% (v/v) of refined sunflower oil was added aseptically. The medium was inoculated with 2.5 mL of a R. oryzae and A. flavus spore suspension (1-4 × 10 6 spores/mL) and then incubated at 28°C for 5 d using an orbital shaker at 200 rpm. Mycelium was harvested from the culture medium using a Buchner funnel and washed with distilled water followed by acetone. It was then dried under vacuum for 18 h and ground to a powder. The R. oryzae (CECT20476) and A. flavus (CECT20475.2.1) strains are housed in the Spanish Type Culture Collection (CECT). The enzymatic units (U) were determined beforehand on the basis of the enzymatic esterification rate of ethyl oleate from oleic acid and ethanol. The specific activity of the resting cells of R. oryzae was 1.14 U and A. flavus was 0.57 U.

Enzymatic preparation of ethyl esters
4.4.1. One-step synthesis: transesterification reaction A 1:3.2 mixture of monkfish liver oil (0.453 g; 0.5 mmol) and ethanol (0.0736 g; 1.6 mmol) was added to a reaction vial (5 mL) fitted with a PTFE-lined cap that contained 0.045 g of each biocatalyst (10% w/w based on the weight of monkfish liver oil), a commercial enzyme (Novozym 435) or resting cells (R. oryzae and A. flavus) or their mixtures (1:1 and 7:3, R. oryzae-A. flavus), respectively (Fig. 6). The mixture was stirred (220 rpm) continuously at atmospheric pressure and 28C. Reaction progress was evaluated for 24, 48 and 74 h. Samples were collected, filtered and solvent evaporated. An aliquot of 20 mg of the crude material of the reaction was dissolved in deuterated chloroform and the resulting solution was analysed by NMR. Experiments were carried out in triplicate.

Two-step synthesis: hydrolysis and esterification reactions
In the first step a biocatalytic hydrolysis reactions (Fig. 7) were performed using the commercial enzyme, the two resting cells and their mixtures (1:1 and 7:3, R. oryzae / A. flavus). A mixture of monkfish liver oil (0.453 g; 0.5 mmol), biocatalyst (0.045 g; 10% w/w based on the weight of monkfish liver oil) and water (0.453 g; 25 mmol) was stirred (220 rpm) at 28°C for 24 h. The sample was filtrated, centrifuged and the supernatant was collected. The resulting hydrolysate material was analysed by NMR and GC-FID, with this determining which of these tests offered the best results in terms percentage of free fatty acids. With the better biocatalyst, the hydrolysis was scaled up to obtain a minimum of 30 ml of hydrolysated, with which the esterification studies were carried out. The hydrolysate material was analysed by NMR and GC-FID to determine the hydrolysis degree and the fatty acids profile. In the second step, the hydrolysed oil was used to perform the esterification reactions using the same biocatalysts indicated above. A mixture of hydrolysed monkfish liver oil (0.300 g; 1 mmol), biocatalyst (0.03 g, 10% w/w based on the weight of hydrolysed monkfish liver oil) and ethanol (0.147 g; 3.2 mmol) were stirred (220 rpm) at 28°C. The esterification reactions were conducted for 24, 48 or 72 h. The reaction products were analysed by NMR and GC-FID. All the experiments were carried out in triplicate.

Analytical Methods
Preliminary analysis of monkfish liver oil, hydrolysed oil and ethyl esters were carried out by 1 H NMR. Spectra were recorded with a MERCURY plus NMR Spectrometer Systems VARIAN AS 400 MHz magnet using deuterated chloroform (99.9 atom % D) as solvent. The fatty acid profiles of the initial oil and the esters after synthesis reactions were analysed using an Agilent HP6890 series gas chromatograph (Barcelona, Spain) coupled to a flame ionization detector (FID). The chromatographic column was a 30 m × 0.25 mm fused silica capillary coated with a 0.25 μm film thickness (50%-cyanopropyl)-methylpolysiloxane (DB-23; Agilent J&W, Madrid, Spain). The temperature program used was 180°C for 1 min, followed by an increase of 20°C min-1 until the final temperature of 270°C was reached, which was then maintained for 20 min. A split-less of 20 mL/ min for 0.15 min was applied. Hydrogen was used as carrier gas at a constant pressure. The injection volume was 1 μL. The injection system was maintained at 270°C and the FID at 280°C.

Conclusions
During this study, two extraction methods and four green solvents were studied for the extraction of MLO. The Roller Mixer was the best extraction system for all the solvents studied. All the green solvents tested (CPME, 2-MeTHF, DCM and LMN) allowed the extraction of 96 to 100% of the oil contained in this co-product. In addition, all the green solvents showed higher extraction yields than the common FR extractive method. The percentages of esters from the MLO in one-step esterification were 63%, 61% and 46% using Novozym 435, R. oryzae lipase and A. flavus lipase, respectively. In the two-step esterification reactions, the yields were 85%, 65% and only 41% using the commercial enzymeR. oryzae A. flavus, respectively. Consequently, the latter is definitely not a good biocatalyst for this type of reaction. Another important observation was that R. oryzae lipase (CECT20476), showed lowest yields for DHA-EE in both the one or two steps processed, which suggests selectivity towards this fatty acid. The resting cells of filamentous fungi used in this study unlike of commercial enzymes have not gone through any kind of extraction, purification and immobilization process. Therefore, they are biocatalysts that are considered very cheap compared to the commercial ones. Moreover, several authors report that the enzymatic activity and the selectivity of the lipases can be affected by the process of immobilization. Similarly, it is clear that the commercial lipase esterified omega-3 fatty acids optimally, with yields above 90% for DHA and 100% for EPA contained in MLO. The resting cells of R. oryzae and A. flavus tested showed good percentage of hydrolysis (88% and 93%, respectively), higher than those of the commercial enzyme (61%). These results open a way to study PUFA enrichments using these resting cells from different oil sources.