The sn-1,3 specific lipase Lipozyme RM IM (immobilized Rhizomucor miehei lipase) was obtained from Sigma Aldrich. The water content of Lipozyme RM IM was 3.5±0.3% as determined by Karl-Fisher titration with a Mitsubishi CA-20 moisture meter in triplicate. Glycerol was also purchased from Sigma Aldrich with a purity of ≥99.5% and a water content of 0.04±0.01%. Tert-butanol and tert-pentanol were purchased from Merck with a purity of ≥99% and a water content of 0.298±0.033% and 0.065±0.023%, respectively. Refined sardine oil was kindly provided by Industrias Afines S.L. (Spain).

The standards of MAG, DAG, FFA and fatty acid methyl ester (FAME) for the chromatography analysis were purchased from Sigma. All other solvents and reagents were of analytical or chromatographic grades from VWR.

The mixture of sardine oil, glycerol, water and tert-alcohol (TB or TP) was incubated at 50 °C in a water bath with orbital agitation in different vials. The reaction was initiated by the addition of the lipase. The enzyme loading was 10 wt%, based on the total amount of substrates. At different time intervals, the corresponding vial was withdrawn and filtered through a microfilter (0.45 μm, Sartorius RC) to stop the reaction by removing the lipase. All samples were stored at −18 °C prior to analysis. The samples were analyzed at least in duplicate.

The conditions of the kinetic experiments performed in this work are presented in Table 1. The amount of water added to the system varied in the range of 4 to 30 wt%, based on glycerol weight, in the reaction medium. The amount of tert-alcohol added was calculated on the basis of previous studies on liquid liquid equilibrium for the glycerolysis systems of sardine oil with the two tert-alcohols as reaction medium (Solaesa et al., 2013). Table 2 shows the amount of solvent added in other glycerolysis systems found in the literature to avoid mass transfer limitation due to the immiscibility of oil and glycerol. Compared to the amount of solvent added in this work (Table 1), it can be observed that in most cases the amount of solvent added is not optimized due to an excess or deficiency from the beginning of the reaction.

The neutral lipid profile (MAG, DAG, TAG and FFA) was analyzed by a normal-phase high performance liquid chromatography (NP-HPLC). The chromatographic apparatus consisted of an HPLC system (Agilent 1200) formed by a quaternary pump and an auto-injector. The chromatographic separation of the compounds was carried out at room temperature with a Lichrospher Diol column (5 μm, 4 mm×250 mm) and detection was performed in an evaporative light scattering detector (Agilent 1200 series) at 35 °C and 0.35 MPa. Gradient elution was achieved by mobile phases A (isooctane) and B (methyl tert-butyl ether:acetic acid =99.9:0.1, v/v). The course of the gradient was as follows: first, solvent A was flowing for 1 min, after that, solvent B was added in three steps, at up to 10% in 10 min, to 44% in 22 min and to 100% in 30 min. Subsequently, solvent B was decreased to 0% in 17 min to return to the initial conditions. Finally, the stationary phase was rinsed with solvent A for 2 min. Injection volumes of 10 μl and the elution flow-rate of 1 ml/min were used in all experiments. The different family of compounds were identified and quantified by using calibration curves of the corresponding standard compounds of DAG, a mixture of dipalmitin (26657-95-4) and diolein (25637-84-7), MAG, a mixture of 1-monopalmitin (542-44-9), 1-monoolein (111-03-05), 2-monoolein (3443-84-3) and monodocosahexaenoin and FFA, a mixture of oleic acid (112-80-1) and palmitic acid (57-10-3) in tert-pentanol. In the case of TAG, the calibration curve used was made with the refined sardine oil (TAG ≥99%) because the response factor of the oil as a complex mixture of TAG was very different than the pure standards of TAG. The calibration method has been validated by analyzing two different samples formed by MAG, DAG and FFA standards and sardine oil in the concentration range of the corresponding calibration curves to obtain good reliability.

The analysis of remaining glycerol was performed using High-Temperature Gas Chromatography (HT-GC). The method and the calibration procedure were previously developed (Solaesa et al., 2013).

The MAG fraction separated by NP-HPLC under the optimal reaction conditions with both solvents was collected according to its retention time. Separations were repeated at least six times to obtain enough sample for the fatty acid profile analysis. The fractions were stored in special flasks to evaporate the solvent under vacuum using a rotary evaporator (Heibolph VV2000) at 40 °C. Then the samples were transferred to screw-capped tubes to carry out the derivatization for conversion to methyl esters by the AOAC method (AOAC Official Method 991.39, 1995) and subsequent analysis by Gas Chromatography with an Agilent gas chromatograph (6890N Network GC System) equipped with a flame ionization detector (FID). A fused silica capillary column (OmegawaxTM-320, 30 m×0.32 mm i.d.) was used. The separation was carried out with helium (1.8 ml·min⁻¹) as carrier gas. The column temperature was programmed starting at a constant value of 180 °C during 20 min, heated to 200 °C at 1 °C·min⁻¹, held at 200 °C during 1 min, heated again to 220 °C at 5 °C·min⁻¹ and finally held at 220 °C for 20 min. A split injector (50:1) at 250 °C was used. The FID was also heated at 250 °C. The injection volume was 1 μL. Most of the fatty acid methyl esters were identified by comparison of their retention times with those of chromatographic standards (Sigma Chemical Co.). The calibration procedure was previously developed and quantification was possible by relating the peak areas to the area of an internal standard (methyl tricosanoate) as indicated by the AOAC method (AOAC Official Method 991.39, 1995).

The separation of the different acylglycerols was performed by TLC on silica gel plates (Silica gel 60 F254, Merck). A similar method of Jin et al. (2011) was used to visualize sn-1(3) and sn-2-MAG isomers. The plates were impregnated by immersion in a hydro-ethanolic solution (50%, v/v) of boric acid (1.2 wt%). Afterwards, they were activated by heating at 100 °C for 15 min. 20 μL of the reaction mixture were directly spotted on the TLC plate. The plates were then developed in chloroform/acetone/methanol (95:4.5:0.5, v/v/v). Spots of each lipid were visualized with a UV lamp.

As has been already mentioned, the carrier support of Lipozyme RM IM is hydrophilic. In the presence of a hydrophilic support, the tert-alcohol and mainly glycerol (μ=2.617, log P=−1.84 and ε (23 °C) =41.01) (Riddick et al., 1986) can be preferably adsorbed onto the support. Due to the hydrophilic nature of the reaction medium, some essential water from the lipase can be removed. Therefore, it is important to determine the optimal water content in the reaction medium since it affects lipase activity and therefore the reaction rate and yield product. On the other hand, when the water content is too high it does not enhance the MAG yield resulting in an increase in by-product formation, such as FFA (Zhong et al., 2009). Therefore water control is important to ensure enzyme activity as well as to minimize hydrolytic side reactions.

First, some preliminary experiments were carried out to study the effect of the polarity medium in the retention of titrable water by the enzyme. The moisture of different reaction media (tert-alcohol+glycerol+sardine oil) at different reactant mole ratio glycerol:oil, from 1:1 to 6:1 was first determined. Then, Lipozyme RM IM with a known amount of titrable water was added at 10% w/w based on the weight of the reactants. After 24 hours of contact time at 50 °C (time enough to reach sorption equilibrium) the water content of the reaction medium was again determined. Total titrable water retention by Lipozyme RM IM was evaluated according to Equation 4: % water retention = ( 1 - increase of water in the retention medium amount of titrable water of the lipase ) .100

Figure 1 shows the titrable water retention by Lipozyme RM IM in both tert-alcohols studied in this work. By increasing the amount of glycerol in the reaction medium (higher glycerol: oil mole ratio) more water is removed from the lipase surrounding. This can be related to the higher polarity of the reaction medium when increasing the glycerol content. This behavior could explain previous results on glycerolysis kinetics where no reaction took place even in the presence of Lipozyme RM IM as catalyst, proving that this lipase is water-dependent. When comparing both tert-alcohols, tert-butanol is more hydrophilic than tert-pentanol (log P=0.35 and log P=0.89, respectively), therefore a tert-butanol medium deprives more water from the enzyme than a TP medium under the same conditions. Consequently, TB as reaction medium would need more added water than TP to obtain similar results under the same conditions.

The effect of water content in the reaction medium on MAG and FFA yield at three different initial reactant mole ratios (glycerol:oil), 1:1, 3:1 and 5:1, was studied at 50 °C by varying the amount of water added to the reactive system in the range of 4–30 wt% of water based on glycerol weight (Table 1). Figure 2 shows the MAG and FFA content obtained by using tert-butanol and tert-pentanol as reaction medium after 20 h reaction time. As a general trend, MAG yield increased with the increase of water content in the reaction medium up to a maximum. The highest yield of MAG was obtained by adding around 7–17 wt% of water, depending on the substrate mole ratio. For instance at a mole ratio of 1:1 the percentage of water added to yield a maximum in the MAG production was around 7 wt% in TP as reaction medium and 10 wt% in TB as reaction medium. This behavior agrees with the results presented in Figure 1. However at a mole ratio of 5:1 the optimum percentage of water added for both tert-alcohols was of the same order, around 17 wt%. This could be explained by taking into account that at such high mole ratio most of the essential water of the enzyme has been deprived (Figure 1) and the effect of solvent is less pronounced. An increase in the amount of water in the reaction medium also leads to an increase in the FFA content, caused by higher hydrolysis rates due to the higher content of water. At a mole ratio of 3:1 FFA content in the reaction medium increased from 5.3 to 32.5 wt% when increasing the amount of added water from 7 to 24 wt%, respectively. Therefore an optimal amount of water to be added to the reaction medium taking into account the MAG yield and the FFA formation at the different mole ratios glycerol:oil does exist. In addition, mass transfer limitations of the lipophilic substrate, the sardine oil, from the reaction medium to the vicinity of the enzyme can become important at high water content since, as described in the literature (Cheirsilp et al., 2007; Zhong et al., 2009), an excess of water can surround the particles instead of forming a thin mono-hydration layer for catalytically-active conformation. Zhong et al. (2009) found that the highest yield of MAG was obtained when 10 wt% of water was added to the glycerolysis system of soybean oil with Thermomyces lanuginosus lipase and TB/Isopropanol (80:20) as solvent system.

Substrate mole ratio, glycerol:oil, has different effects on lipase-catalyzed glycerolysis. According to the equilibrium law, an increase in the glycerol content will shift the equilibrium towards MAG production. In the literature, it has been also described that glycerol can act as an effective stabilizer against thermal and solvent de-activation (Zhong et al., 2009). However, in the case of water-dependent lipases, the glycerol content can also directly affect the lipase activity due to its influence on the polarity of the reaction medium.

The effect of glycerol:oil mole ratio on the reaction rate was studied at 50 °C, 10 wt% of lipase loading, by varying the mole ratio from 1:1 to 5:1. Figure 3 shows the MAG profile at the different glycerol:oil mole ratio and at a high content of water added (from 24 to 30 wt% based on the glycerol content) for both tert-alcohols. By increasing the initial substrate mole ratio, MAG equilibrium conversion also increases. For instance MAG content at a mole ratio of 1:1 is around 50% in tert-butanol, while it reaches a maximum of 75% at a mole ratio of 5:1. However, it must be highlighted that there is little difference in MAG content at 3:1 and 5:1 glycerol:oil mole ratio. Based on these results, a glycerol:oil mole ratio of 3:1 was selected as optimum in the present study. This result agrees with some previous studies on glycerolysis of different types of oils. Zhong et al. (2009) selected 3.5:1 as optimum glycerol:soybean oil mole ratio, in their study on the glycerolysis reaction in a mixture of TB/Isopropanol (80:20) as reaction medium. Pawongrat et al. (2007) used MTBE as solvent and showed that the optimum mole ratio of glycerol to tuna oil for MAG production was 3:1.

Solvents can have different effects on reaction systems. They help to create a homogeneous reaction system and improve mass transfer by reducing the viscosity of the reaction medium. When polar solvents are used, essential water from the lipase can be removed leading to a decrease in the lipase activity. In this work two tertiary alcohols, tert-butanol and tert-pentanol, have been considered as reaction media. Log P is one of the most widely used parameters to correlate the solvent properties to the enzyme activity. Taking into account the log P of TB and TP (0.35 and 0.89 respectively), TB is slightly more hydrophilic than TP; therefore water retention by the lipase will be lower when using TB as reaction medium and slower reaction rates can be expected.

Figure 4 shows the effect of both tert-alcohols on the time course of the glycerolysis of sardine oil under the same reaction conditions at different amounts of water added to the reaction medium. As expected, the reaction rates in TB were slower than in TP. Nevertheless, MAG equilibrium yield was of the same order for both tert-alcohols.

Based on the results presented in this work, the optimal conditions for MAG production with 10 wt% loading of Lipozyme RM IM, based on substrate weight, at 50 °C were the following: initial mole ratio glycerol:oil 3:1, water content in glycerol 12 wt% and 60 wt% of the corresponding tert-alcohol to provide a homogeneous medium. Mole ratio higher than 3:1 does not lead to an increase in MAG yield (Figure 3). At this mole ratio, a maximum in MAG yield (around 78 wt%) in TP is reached at 12 wt% of water added, based on glycerol weight, with less than 20 wt% of FFA (Figure 2). The time course of the glycerolysis of sardine oil for all the compounds in both tert-alcohols under the optimal conditions is presented in Figure 5.

Figure 5a and 5b represent the glycerolysis course in TB and TP respectively. From Figure 5a it can be observed that the MAG content in TB as reaction medium was higher than 65 wt% after 20 hours of reaction time. At this reaction time the TAG content in the reaction medium was still about 20 wt%. DAG production slightly increased during the first 4 hours and then it remained relatively constant at around 7 wt%. At the beginning of the reaction, FFA formation is nearly negligible and then it rose slowly up to 10 wt%.

From Figure 5b it can be observed that the reaction rate was faster in TP than in TB. TAG decreased sharply during the first 2 h and then the reaction rate became slower. MAG content also increased quickly during the first 2 h reaching a value of 60 wt%. From this time on, the reaction rate slowed down and MAG content reached a maximum value of 78 wt%. DAG production achieved a maximum at 2–3 h reaction time and then its content steadily decreased. FFA formation is also faster in TP than in TB.

The FA profile of the different MAG fractions obtained under the optimal conditions was analyzed by GC. MAG fractions at two different reaction times (4 h and 20 h) were separated by HPLC and collected to determine their fatty acid profiles (Table 3). The MAG content in TP as reaction medium at the selected reaction times, 4 h and 20 h, was 70 wt% and 77 wt%, respectively. On the other hand, in TB as reaction medium, the MAG content was 22 and 66 wt% at 4 and 20 hours, respectively. A much longer reaction time is needed in TB than in TP to reach the same MAG percentage in the reaction medium.

From Table 3 a similar fatty acid profile of the MAG fraction obtained in tert-butanol at 20 h reaction time was observed (66 wt% of MAG) as in tert-pentanol at 4 h of reaction time (70 wt% of MAG). Although the reaction rate is slower in TB, it could be concluded that Lipozyme RM IM would behave qualitatively in the same way according to MAG conversion and fatty acid profile in both tert-alcohols. Figure 6 represents the different types of FA in the MAG fractions as a function of the MAG content in the reaction medium for both tert-alcohols. The PUFA content in the MAG fraction increased when the conversion degree also increased; while monounsaturated FA (MUFA) and saturated FA (SFA) decreased.

TLC analyses have been performed at certain reaction times (from 0.5 to 20 h) to check the formation of 1(3)-MAG and 2-MAG isomers. The FA profile of 1(3)-MAG and 2-MAG fractions at 8 h of reaction time was known by extraction of the appropriate band and the subsequent methylation for the FAME analysis in GC (data not shown). The results showed that the majority of EPA is found as 1(3)-MAG, while DHA is found mainly in 2-MAG. These outcomes are in agreement with the FA at the sn-2 position in the TAG (Table 3). In any case, acyl-migration could take place due to the positive surface charge of the support material, Duolite 568, the water added and the solvent medium (Fureby et al., 1996).