Enzymatic Hydrolysis of Salmon Oil by Native Lipases : Optimization of Process Parameters

Três lipases microbianas nativas (Aspergillus niger, Rhizopus javanicus e Penicillium solitum) foram utilizadas na hidrólise do óleo de salmão (teor de AGPI n-3 de 30,1%) com o objetivo de concentrar o conteúdo de ácidos graxos poliinsaturados n-3 (AGPI n-3) nos acilgliceróis residuais. A metodologia de planejamento experimental e análise de superfície de resposta foi usada para se chegar às condições otimizadas de cada reação enzimática, utilizando as seguintes variáveis; temperatura (X 1 ), quantidade de lipase (X 2 ) e taxa de água/óleo (X 3 ). Com base nos resultados do planejamento, a lipase de Aspergillus niger foi a mais eficiente na concentração dos AGPI n-3, sendo que as condições ótimas de reação foram: concentração de enzima de 500 U g óleo, temperatura 45 °C e taxa de água/óleo de 2:1 m/m após 24 h de reação. O grau de hidrólise (60%) conduziu a um aumento do conteúdo de ácido docosahexaenóico (DHA) de 14,4% para 34,0% (enriquecimento de 2,4 vezes) nos acilgliceróis residuais após a hidrólise do óleo de salmão.


Introduction
The n-3 polyunsaturated fatty acids (PUFA) are commonly known to play an important role in human health.2][3][4][5] During the last decade, several studies have shown positive effects of fish oils on cognitive development and vision enhancement in newborns, as well as in young children. 6For medical or dietetic purposes, n-3 PUFA may be administered in different forms; as free fatty acids (FFA), as ethyl esters, or as acylglycerols.Some studies indicate that PUFA are most promptly absorbed from the intestines when FFA is given orally, are moderately absorbed as acylglycerols and are poorly absorbed as PUFA ethyl ester.However, free n-3 PUFA is oxidized most easily and, moreover, free n-3 PUFA is unacceptable as a food.Therefore, acylglycerols are considered to be the most desirable chemical form as food. 7][10][11] Lipase-catalyzed enzymatic production of EPA and DHA concentrates from fish oil has shown potential in producing a high quality product, due to the mild conditions (e.g., neutral pH and low temperatures) of the process. 12ommercial lipases from Candida rugosa (cylindracea),

G e o t r i ch u m c a n d i d u m , H u m i c o l a l a n u g i n o s e,
][17] In this paper, we evaluate the catalytic performance of three native lipases produced in our laboratory from Aspergillus niger, Rhizopus javanicus and Penicillium solitum for production of a salmon oil concentrate.In addition, a preliminary optimization to determine the optimal processing parameters for the hydrolysis degree of selected lipases was carried out.Three-factor response surface were employed to optimize the reactive conditions of lipase-catalyzed hydrolysis in order to obtain a maximum hydrolysis degree and acylglycerols rich in n-3 PUFA.
The method of process optimization by the response surface methodology is a faster and more economical method for gathering research results than the classical one-variable-at-a-time or full-factorial methods. 18,19

Chemicals and oil
Peptone and yeast extract were purchased from Difco Laboratories, (Detroit, MI, USA).The components for culture media, chemical reagents and other solvents were obtained from Merck (Darmstadt, Germany) and from Sigma-Aldrich Chemical Co. in the highest purity available.Low acidity olive oil was purchased at a local market.The salmon oil (saponification value, 185; acid value, 0.05) was purchased from Campestre Ltda (São Bernardo do Campo, São Paulo, Brazil).

Production and determination of lipases activities
This study was performed using Aspergillus niger, Rhizopus javanicus and Penicillium solitum lipases produced in our laboratory; these lipases achieved better kinetic properties in previous experiments. 15,17The lipases were produced in a basal medium with an initial pH value of 6.0 and consisted of 2% (m/v) peptone, 0.5% (m/v) yeast extract, 0.1% (m/v) NaNO 3 , 0.1% (m/v) KH 2 PO 4, 0.05% (m/v), MgSO 4 .7H 2 O and 2% (m/v) olive oil.Cultures were grown in Erlenmeyer flasks (500 mL) containing 120 mL of the growth medium.The cultures were inoculated with 1mL of spore suspension (10 5 -10 6 spores mL -1 ) and the flasks were stirred on a rotary shaker (130 rpm) at 35 o C for 72 h.After this period, the cultures were filtered and the supernatants were treated with ammonium sulphate (80% saturation).The precipitates were dialyzed in water and lyophilized for use as extracellular crude lipase preparations in powder form.The residual water in the lyophilized lipases was 0.2-0.3%(m/m).Lipase activity were quantified by triolein using olive oil as substrate employing a standard oleic acid curve. 20One unit (U) is defined as 1 µmoL of oleic acid released per minute of reaction.Lipases from A. niger with 7.5 U, R. javanicus with 6.8 U and P. solitum with 4.8 U were used.These lipases were used without further purification.

Hydrolysis reaction and experimental design
Three experimental designs, based on the 2 3 full factorial, were performed under the same experimental conditions for each lipase produced (A.niger, R. javanicus and P. solitum) in order to verify the power of hydrolysis of each native lipase using salmon oil.Reaction mixtures for enzymatic hydrolysis were carried out according to the experimental conditions in Table 1.Conical flasks (250 mL) were blanketed with nitrogen and kept sealed throughout the reaction for 48 h.The vessels were stirred in an incubator orbital shaker at 400 rpm at different temperatures.Different ratios of water (0.1 mol L -1 phosphate buffers) and salmon oil were mixed in the flasks and different amounts of lipase preparations (U of the enzyme g -1 of oil) were added.The initial water/ oil amount was fixed at 30 g for all experiments.Statistical optimization experiments for all lipases were carried out according to the 2 3 full factorial designs with 4 center points.The independent variables or factors studied were temperature (°C; X 1 ), amount of lipases (U; X 2 ) and water/ oil mass ratio (m/m; X 3 ).Response for the dependent variable studied was the degree of hydrolysis (DH) (%).The design matrix for A. niger, R. javanicus and P. solitum lipases shown in Table 2 was obtained by means of the Statistica 7.0 software.Significance of data was tested using the ANOVA statistical test.The experiments for all lipases were accomplished at four different times (6, 12, 24 and 48 h) in order to observe the kinetic behavior of each reaction using the different enzymes.Enzymes were deactivated by heating the reaction to approximately 90 ºC for 15 min in order to interrupt the reaction at a determined time.

Separation of reaction products
At the end of each enzymatic reaction, partially hydrolyzed oil containing residual acylglycerols (mono-, di-and triacylglycerols) and released fatty acids fractions were separated by the method described by Shimada et al. 8 After hydrolysis by Aspergillus niger lipase, the yield of fatty acids isolated (g) per oil (g) was 0.5-0.7.

Determination of degree of hydrolysis (DH)
Degree of hydrolysis (DH) was determined by measuring the acid value of both unhydrolyzed (original oil) and hydrolyzed oil at different treatment times (6, 12, 24 and 48 h) and the saponification value of unhydrolyzed oil, according to the American Oil Chemists' Society methods. 21lanks (no enzyme) were determined for each treatment.DH was calculated according to the following equation: Acid values indicate relative amount of free fatty acids released by the hydrolysis reaction, as measured by the mg of potassium hydroxide necessary to neutralize fatty acids in 1 g of salmon oil.

Determination of fatty acid composition
Fatty acids in unhydrolyzed salmon oil, residual acylglycerols and liberated fatty acid fractions were converted after hydrolysis into fatty acid methyl esters (FAME), according to the method of Hartman et al., 22 and their compositions were determined by gas chromatography (GC).A Chrompack gas chromatograph (CP-9001) equipped with a flame ionization detector and a CP-Sil 88 capillary column (Chrompak, WCOT Fused Silica 50 m × 0.25 mm i.d.; split ratio, 100:1) were used for analyzing FAME.The initial temperature of the column was 50 °C for 2 min and then the temperature was increased at a rate of 10 °C min -1 up to 210 °C; hydrogen was used as a carrier gas.The temperature of the injector was 220 °C and the detector's temperature was 250 °C.Fatty acids were identified by comparing the times of retention of the components present in each sample with the times of retention of the components present in the standard FAME Mix C4-C24 (Supelco).

Degree of hydrolysis
A three-factor response surface methodology (RSM) with 4 center points was employed to study the degree of hydrolysis (DH) by lipase-catalyzed synthesis of salmon oil for each native lipase.Experimental results of DH for catalyzed-hydrolysis reactions with salmon oil for each RSM using A. niger, R. javanicus and P. solitum lipases are given in Table 2.All the reactions were accomplished at four different times (6, 12, 24 and 48 h), although only results for the 48 h reaction were analyzed statistically.
DH´s of salmon oil from different lipases, conditions of run and reaction times are shown in Figure 1.In general, salmon oil was hydrolyzed rapidly during the initial 12 h.Afterwards, hydrolysis was more gradual, and only minor increases were shown throughout the 24 and 48 h time periods.A. niger was the lipase that most effectively hydrolyzed salmon oil (ca.60% of hydrolysis after 24 h under experimental conditions).
Table 3 depicts the analyses of the reaction parameter effects for the DH when the A. niger lipase was used.The concentration of the A. niger lipase, the water/oil mass ratio and the interaction between these parameters significantly affected the DH.Moreover, temperature did not affect the degree of hydrolysis, nor the interactions that involved this parameter.Therefore, these results suggest a linear model whose parameters are mainly X 2 , X 3 and X 2 X 3 .The coefficients of independent variables determined for the linear model for the DH, when the A. niger lipase was used, are given in equation 1.
DH (AN) = 18.33 + 8.12 × X 2 -14.87 × X 3 -6.92× X 2 X 3 (1)   The analysis of variance for the DH model is given in Table 4. Based on the F-test, the model (equation 1) is predictive, since the percent of explained variance is high (93.24%)and the calculated F-value (regression/residual) is more than 4 times higher than the critical F-value at 95% of confidence (F > F 0.95,4.7 of 6.09).In addition, there is little evidence of a lack of fit for the fitted model, since the calculated F-values (lack of fit/pure error) are lower than the critical F-value (F < F 0.95,3.2 of 19.16) at 95% confidence. 23he same analysis of effects of parameters and analysis of variance for R. javanicus and P. solitum lipases were studied (Tables 5 and 6).Observing the percent of explained variance (R 2 ) and F-test, the results demonstrate that only a first-order polynomial model does not represent the behavior of lipase-catalyzed reactions for these lipases, since the high lack of fit of the models and the regression of the parameters are not significant statistically.Whilst the results of the lipases from R. javanicus and P. solitum were not statistically significant enough to predict a model, they might be used to observe the behavior of the reactions in function of the parameters studied (temperature, amount of enzyme and water/oil mass ratio) when these native lipases were employed and compared with others in relation to the hydrolysis performance of each.
The next results show the hydrolysis performances of the lipases and their effects on the enzymatic reactions when submitted to different experimental conditions.Figure 2 illustrates the variation of the DH when, simultaneously, the concentration of lipase and the temperature are analyzed.For A. niger lipase (Figure 2 (a)), practically no influence of the temperature on the response was observed.On the other hand, the increase in temperature decreases the DH for the native lipases from R. javanicus and P. solitum, as can be seen in Figures 2 (b) and 2 (c).
Figure 3 shows that, although the three native lipases have differences in the DH (%), the same behavior was observed in relation to the concentration of lipases and ratio of water and oil in order to increase the DH.Higher DH is reached, increasing the concentration of lipases and increasing the amounts of water in the reaction, simultaneously.

Fatty acid composition of the reaction products
The composition of the main fatty acids in the residual acylglycerol and liberated fatty acid fractions after hydrolysis of salmon oil with A. niger lipase are shown in Table 7.
After 45% hydrolysis (6 h of reaction), the total n-3 PUFA content of the residual acylglycerol fraction increased by approximately 1.5-fold, compared to the original oil, giving a final concentration of approximately 45%.Most of the enrichment was due to DHA, which increased twofold, whereas the EPA content remained relatively constant with no significant differences, compared to the feed oil.This difference in selectivity is reflected in the increased DHA/EPA ratio.
The enrichment of n-3 PUFA was compensated by a reduction in the content of C16-C18 saturated and monounsaturated fatty acids.The liberated fatty acid fractions were correspondingly increased in the C16-C18 saturated and monounsaturated fatty acids, with a great depletion in DHA, compared to the feed oil.
When the reaction reached 60% hydrolysis (24 h of reaction), the residual acylglycerol fraction were further enriched in DHA, however the EPA content was reduced.Total n-3 PUFA was similar to that found after 45% hydrolysis, although the DHA/EPA ratio reached approximately three at this stage.

Discussion
Salmon oil was submitted to hydrolysis by three 1,3-specific lipases from A. niger, R. javanicus and P. solitum to preserve the long chain n-3 PUFA, which are preferentially located at the 2 position on the glycerol backbone. 24Due to the several cis-cis unsaturations of PUFA, these molecules are strongly bent, thus enhancing the steric hindrance effect.Therefore, the ester moiety of PUFA-containing acylglycerols is hardly split by lipase, in contrast to saturated and monounsaturated fatty acids (Scheme 1).
From the experimental design using native lipase from A. niger, it was possible to obtain a first order model to optimize the parameters of reaction in function of the DH of salmon oil.When lipase from A. niger was used, about 60% of hydrolysis is required after 24 hours of reaction to obtain a high content of DHA and a high DHA/EPA ratio in the unhydrolyzed fraction.Under these optimized conditions, the total n-3 PUFA content of the residual acylglycerol fraction was 45% (34% DHA + 11% EPA), approximately 2.4-fold greater than the initial values obtained with the non-optimized system.In contrast, salmon oil was not hydrolyzed as much with the other lipases; the R. javanicus and P. solitum lipases achieved just 20% and 3% hydrolysis under the same conditions.The temperature significantly influenced the process, causing a 30-40% decrease in DH for both the R. javanicus and P. solitum lipases.A higher reaction temperature (45 o C) caused the denaturation of two lipases.In a previous report, we showed that lipase from A. niger was more thermostable than other native lipases, demonstrating stability at the range 30 and 50 o C and maintaining at least 60% of its hydrolytic activity at 60 o C after 1 h. 16A probable explanation for the low lipases activities of R. javanicus and P. solitum is their enzyme structures, which do not accommodate the esters of PUFA from salmon oil in their active sites.
To our knowledge, this is the first finding of an enzymatic hydrolysis of salmon oil by native lipases.The commercial lipases from Candida rugosa (cylindracea) and Geotrichum candidum are useful for the production of oil containing high concentrations of PUFA.Geotrichum candidum lipase can DHA and EPA and Candida rugosa can enrich DHA, but not EPA. 8In general, the DH is higher in oil hydrolyzed by non-specific lipases (Candida sp.), compared with 1,3-specific lipases (Mucor javanicus, Rhizopus delemar and Aspergillus niger). 10,11,25anaka et al. 25 reported that Candida cylindracea lipase shows low reactivity for DHA esters, and is successful in concentrating DHA in the residual acylglycerol fraction during the hydrolysis (ca.70% rate) of fish oil.DHA was concentrated from 8.9% to 30.5% and EPA was changed from 13.3% to 9.8%, resulting in 40.3% of total n-3 PUFA in the residual acylglycerol fraction of the fish oil.
Some authors have reported that hydrolysis exceeding 50 hours might increase hydrolysis to around 90%. 26 In this study, a DH of 60% and an increase of 50% in n-3 PUFA were reached after 24 h and no further increase was noted following longer reaction times (data not shown).
Okada and Morrisey (2007) 11 investigated the production of n-3 PUFA concentrate from Pacific sardines using several commercial lipases.Oil hydrolyzed by Aspergillus niger lipase exhibited only a minor increase in DHA content (13.6% to 18.5% with 250 U after 1.5 h and to 18.7% with 500 U after 3 h) and no significant changes in EPA content. 11ur results clearly demonstrate the superiority of the lipase from Aspergillus niger in the enrichment of DHA by selective hydrolysis of salmon oil, which demonstrated to be a feasible method for the development of salmon oil health products.The experimental design procedure provided a powerful tool to optimize the hydrolysis conditions that permit an important improvement in the degree of hydrolysis of this lipase.Under optimum conditions, a good enrichment of DHA (2.4-fold) was achieved, which is considerably higher than those results previously reported using this lipase.

Figure 2 .
Figure 2. Varying concentrations of lipases and temperature as a function of DH for lipases: a) Aspergillus niger; b) Penicillium solitum; c) Rhizopus javanicus.

Figure 3 .
Figure 3. Varying water/oil mass ratio and concentrations of lipases as a function of DH: a) Aspergillus niger; b) Penicillium solitum; c) Rhizopus javanicus.

Table 1 .
Experimental conditions

Table 2 .
Three-factor designs and responses for lipases from Aspergillus niger, Rhizopus javanicus and Penicillium solitum a DH = degree of hydrolysis after 48 hours of reaction.

Table 3 .
Statistical analysis of the operating variables effects for the Aspergillus niger lipase on DH

Table 4 .
Analysis of variance (ANOVA) to fit the model presented in equation 1

Table 5 .
Statistical analysis of the operating variables effects for the Rhizopus javanicus and Penicillium solitum lipase on the DH

Table 7 .
Composition of the main fatty acids of the reaction product after 45 and 60% hydrolysis of salmon oil by Aspergillus niger lipase Corresponding to ca. 45% hydrolysis degree after 6 hours of reaction (run 4).b Corresponding to ca. 60% hydrolysis degree after 24 hours of reaction (run 4).