Menhaden fish oil (Sigma-Aldrich) and food grade glycerol (J. T. Baker) were used for the reaction medium; tertbutylhydroxyquinone (TBHQ, Química Dresen) as an antioxidant, and dimelsulfoxide (DMSO) as an emulsifier. The NV-435 lipase from Candida antarctica (Novo Nordisk Denmark) immobilized in macroporous acrylic resin was used as catalyst. For high-performance thin layer chromatography (HPTLC) analysis, 10x10 cm plates (Uniplate) were used with analytical grade silica gel, hexane and ethyl ether (Sigma-Aldrich), glacial acetic acid (Fermont) and reagent grade iodine (Fermont).

To isolate and concentrate PUFAs from Menhaden oil, the procedure described by Correa et al. (2017) was used. One hundred g of oil were weighed and TBHQ (0.02 % w/w) was added as an antioxidant. The mixture was saponified with 200 mL of 7 M KOH in 70 % ethanol by heating at reflux for one hour. A liquid extraction was carried out with distilled water (240 mL) and hexane (200 mL), the aqueous phase was then acidified to pH 1.0 with HCl to subsequently extract the free fatty acids (FFA) with hexane (200 mL). The hexane was removed by evaporation in a rotovapor (40 rpm, 40°C, vacuum 25 mmHg), and the remaining moisture was removed with anhydrous sodium sulfate. In a flask, each 15 g of FFA were mixed with 25 g of urea in 100 mL of 95 % ethanol and heated until the mixture turned a homogeneous yellow color. The solution was transferred to centrifuge tubes and quickly cooled in an ice water bath for subsequent cooling (4°C, 8 h). The crystals formed were separated by centrifugation at 6000xg (15 min, 0°C). The supernatant was acidified to pH=4 and transferred to a separatory funnel to extract the PUFA with a 1:1 hexane:water mixture, stirring for 30 min. Finally, the organic phase was evaporated in a rotary evaporator (40 rpm, 40°C, vacuum 25 mmHg) to obtain the PUFA concentrate.

The experimental runs to study the enzymatic esterification reaction were carried out based on a RCCD. Mixtures of 5 g with different molar proportions of n-3 PUFA and G (0.48, 1.5, 3, 4.5, 5.52 mol/mol), TBHQ (0.02 % w/ p) as antioxidant, and DMSO (1 % w/w) as emulsifier, were prepared in glass vials (20 mL) with magnetic stirrer. Each vial was placed inside a double arm cooling jacket reactor (Wheaton Celstir®) placed on a magnetic stirring plate at 800 rpm. After stirring the mixture for 30 minutes, a peristaltic pump was used to recirculate the medium at a flow rate of 1.65 mL/min through an PBR (length = 5.104 cm and diameter = 0.717 cm) with 0.9115 g of immobilized lipase NV- 435 of Candida antarctica. A water recirculator bath was used to maintain an isothermal process, experimental runs were carried out at different operating temperatures (28.14, 37, 50, 63, 71.86 °C) and 35 µL samples were taken at different reaction times (t = 0.24 , 0.75, 1.5, 2.25, 2.76 h). A representative diagram of the process is shown in Figure 1.

Figure 1

The analysis was carried out in a HPTLC equipment (CAMAG). Aliquots (5 µL) of the sample were diluted in a chloroform:methanol solution (2:1 v/v) to obtain a final concentration of 20 µg AG/ µL sample. One microliter of each diluted sample was applied on a silica gel plate 1 cm from the edge and with a separation of 1 cm between each sample. The plate was saturated with the mobile phase of hexane:ethyl ether (38:62 v/v) with 1 % glacial acetic acid in a vertical double-channel glass chamber for thin layer chromatography by 40 minutes. The plate was placed into the mobile phase for chromatographic development until a path of 9 cm was reached. The plate was then removed from the chamber and the solvent was evaporated into the air. The plate was developed by atomizing a solution of iodine in ethanol (0.5 % w/v), until achieving a light yellow color on the surface of the plate and the solvent was cold evaporated for 30 min. The developed plate was scanned, and shadow area analysis of the spot bands was performed using ImageJ 1.45 software (NIH; Rasband, 2018). For the quantification of MAG, DAG and TAG, a standard curve was generated using lipid standards (Sigma-Aldrich) containing a mixture of cis-9-monoolein, cis-9-1,2-diolein, cis-9-1,3-diolein and cis-9-triolein.

The G concentration was calculated using a stoichiometric mass balance considering the concentrations of MAG, DAG and TAG, determined by HPTLC analysis using the following equation:

C G = C o G - ( C M A G + C D A G + C T A G ) (1)

To determine the global esterification (GE), 30 µL aliquots were taken and diluted in ethanol. The samples were titrated with a 0.05 N NaOH solution using phenolphthalein as an indicator.

A RCCD was selected to evaluate the effect of the molar ratio between PUFA and G (x ₁ =M), temperature (x ₂ =T) and time (x ₃ =t), on GE (y ₁ ), MAG (y ₂ ); DAG (y ₃ ) and TAG (y ₄ ) production.

To predict the dependent variables (y_i) a third order polynomial regression model was considered:

y i = β 0 + ∑ i = 1 k β i x i + ∑ i = 1 k β i i x i 2 + ∑ i = 1 k β i i i x i 3 + ∑ i = 1 k - 1 ∑ i < j k β i j x i x j + ε (2)

Where β ₀ , β _i , β _ii , β _iii , and β _ij represent the interaction coefficients in the regression, the linear effect, quadratic effect, cubic effect and the combined effects, respectively.

A simplified model was considered for each response according to the significative variables (α = 0.05) resulted from the analysis of variance. Statistical analysis was performed in a JMPpro v17.0 ® platform.

The parameters with the greatest effect on the GE are the linear and quadratic terms of the molar ratio and reaction time (p < 0.005). In the analysis of variance, the effect of temperature did not have a representative effect on GE (p = 0.3954), but it was included in the equation to have a hierarchical model because the interaction term between temperature and time had a significant effect (p < 0.1). The model to describe the GE obtained through second-order polynomial regression is the following:

G E   =   3.019   M 2   -   6.157   t 2   -   2.647   M   t   +   0.223   T   t   -   15.369   M

+   21.593   t   -   0.180   T   +   48.124 (3)

The RSM graphs show that a GE close to 80 % can be achieved by operating with a molar ratio of 0.5 mol PUFA/mol G and a temperature of 70°C for 3 h (Fig. 2a). It is important to clarify that by decreasing the molar ratio (M), n-3 PUFAs have more binding sites in the G structure due to the surplus of this substrate considering that the stoichiometric ratio is 3:1 mol PUFA/mol G, therefore, it is normal to obtain higher degrees of esterification. Linder et al. (2005) studied the enzymatic esterification reaction of PUFA and G in a batch reactor, and obtained a higher GE at a molar ratio of 0.83 mol PUFA/mol G at 46°C. It is important to highlight the high reaction rate that occurs in the PBR, since a high GE was obtained at relatively short operating times, unlike different studies in intermittent reactors where long reaction times are required to achieve a significant GE (Noriega et al., 2013; Wang et al., 2011). The RSM of temperature versus molar ratio provides very interesting information (Fig. 2b). Firstly, the model suggests that at higher molar ratios (M > 4) the GE can increase a little, however, it is not advisable to operate under these conditions because PUFA are the limiting reagent in esterification, and these are the most expensive substrate of the process due to its laborious obtaining process (Torres et al., 2014). Secondly, it can be seen that temperature is not as significant a variable in the GE as time and molar ratio, obtaining similar GE at different operating temperatures in relation to the proportions of the substrates in the medium. Correa et al. (in press) studied the kinetics of enzymatic esterification of n-3 PUFA and G at different temperatures in a batch reactor, and it was observed that an increase in temperature raised the GE reached during the reaction relatively little.

Figure 2

MAGs are the first product formed in the enzymatic esterification reaction, and as the reaction proceeds, the formation of DAG consumes the MAGs present in the reaction medium. The parameters with the greatest effect on the percentage molar fraction of MAG (% MAG) are the linear term of the molar ratio, and the quadratic terms of the molar ratio and temperature (p < 0.005). The statistical model that describes the MAG % is:

M A G   %   =   1.876     M 2   +   0.018     T 2   +   0.093   M   T   -   0.299   T   t   -   19.099   M

- 1.585   T   +   14.251   t   +   107.631 (4)

Figure 3 shows the RSM graphs of the MAG %. The model suggests that the MAG % can be optimized, reaching 80 %, by operating with a molar ratio of 0.5 mol PUFA/mol G and 30°C for 2.75 h. Byun et al. (2007) studied the formation of MAG in the enzymatic esterification of fish oil, and determined that, at low operating temperatures, greater MAG formation occurs. Similarly, Zhao et al. (2011) sought to optimize the formation of MAG in an enzymatic esterification reaction without solvents, and obtained conversions of 49.6 % at a molar ratio of 0.16 mol AGL/mol G and a temperature of 50°C. Figure 3 b shows the response surface of the molar ratio against temperature (t = 0.5 h), observing that operating with higher molar ratios can slightly increase the MAG %, this result is similar to that obtained by Noriega et al. (2013), in which the enzymatic esterification reaction of n-3 PUFA and G was studied in an intermittent reactor.

Figure 3

DAGs are an intermediate product in the enzymatic esterification reaction, however, these are lipids with high added value due to their multiple benefits. The parameters with the greatest effect in the model are the quadratic effects of time, molar ratio and temperature, presenting values p < 0.001. The model that determines the percentage mole fraction of DAG (% DAG) is the following:

D A G   %   =   - 1.265   M 2 -   0.016   T 2   -   5.297   t 2 +   0.799   M   t   +   0.126   T   t

+   7.045   M   +   1.497   T   +   7.166   t   -   18.672 (5)

The RSM analysis performed on temperature versus time (Fig. 4a) shows that at a temperature of 50.33°C, molar ratio of 3.26 mol PUFA/mol G and reaction time of 1.52 h, the production of DAG can be maximized in the enzymatic esterification reaction (DAG %=36). This behavior is due to the fact that, in the reaction mechanism, at one of the first stages of esterification, the MAG react with PUFA to form DAG, but in the next stage these are consumed for the formation of TAG. Different investigations have published kinetic profiles of DAG formation similar to those obtained in this investigation (Correa et al., 2017; Bornadel et al., 2013). Duan et al. (2010) sought to optimize the formation of 1,3-DAG from an enzymatic esterification reaction in a batch reactor. They obtained optimal formation conditions at 60°C and a molar ratio of 2.5 mol FFA/ mol G, reaching a concentration of 40 %. Lo et al. (2007) studied the enzymatic esterification of palmitic and oleic acid catalyzed by the lipozyme lipase RM IM, obtaining optimal conditions for the formation of DAG at 66.29°C, operating with a molar ratio of 2.14 mol FFA/mol G. Although the results obtained differ slightly from those reported by other researchers, the enzymatic esterification reaction presents many factors that can influence the performance and behavior of the process, such as the composition of the substrate (Rivero et al., 2020; Kahveci and Xu, 2011), the lipase used and the water present in the reaction medium.

Figure 4

Due to the low goodness of fit of a second-order polynomial model (R²=0.74) to describe the percentage mole fraction of TAG, it was decided to consider a third-order model that included the combined effect of all factors and their respective cubic terms. The terms with the greatest effect in the model are the quadratic and cubic terms of the molar ratio (p < 0.01). The equation that describes the % TAG is:

T A G   %   =   - 0.844   M 3 -   3.695   t 3 +   0.059   M T t +   7.550   M 2 -   0.009   T 2 +   17.97   t 2 -   0.153   M T -     4.139   M T -   8.384   M +   1.042   T -   22.073   t +   5.029 (6)

TAGs are the products generated at the last stage of the enzymatic esterification reaction; this is reflected in their low concentration obtained in the experimental results due to the reversibilities present during the reaction. The model obtained through RSM indicates that higher molar fractions of TAG are obtained by increasing the reaction time and with n molar ratio of approximately 4.5 mol PUFA/mol G, reaching a molar fraction close to 26 % (Fig. 5a). The low TAG conversions achieved in the experimental runs may also be due to the reversible effects caused by the water produced during the esterification reaction, for this reason, many investigations seek to eliminate or isolate the water present in the medium to optimize the process and obtain better performances in the formation of TAG (Rodrigues and Fernandez, 2010). Noriega et al. (2013) studied the yield of TAG formation in the enzymatic esterification of PUFA catalyzed by the NV-435 lipase. After a 24 h reaction they only reached yields close to 12 %, so they added molecular meshes in order to isolate the water in the reaction medium and after another 24 h they achieved yields close to 80 %; this implies that the water caused strong reversible effects in the formation of TAG. The MSR graph of temperature versus molar ratio (Fig. 5b), at t=1.5 h, reflects that higher TAG conversions can be obtained at temperatures close to 50°C.

Figure 5

It is recommended that the model obtained be considered only within the operating range established in the experimental design because, although a good fit (R²=0.92) is observed, certain inconsistencies occur when extrapolating the study range. Studying a process through an experimental design is very effective to contemplate the effect of factors on the responses analyzed, but the models are usually limited to values established at the treatment levels (Montgomery, 2019).