Impacts of ionic liquids on enzymatic synthesis of glucose laurate and optimization with superior productivity by response surface methodology

Process Biochemistry 50 (2015) 1852–1858

Contents lists available at ScienceDirect

Process Biochemistry journal homepage: www.elsevier.com/locate/procbio

Impacts of ionic liquids on enzymatic synthesis of glucose laurate and optimization with superior productivity by response surface methodology Xiao-Sheng Lin a , Qing Wen b , Ze-Lin Huang a , Yu-Zheng Cai b , Peter J. Halling c , Zhen Yang a,∗ a

College of Life Sciences, Shenzhen Key Laboratory of Microbial Genetic Engineering, Shenzhen University, Shenzhen 518060, Guangdong, China College of Life Sciences, Shenzhen Key Laboratory of Marine Bioresources and Ecology, Shenzhen University, Shenzhen 518060, Guangdong, China c WestCHEM, Department of Pure & Applied Chemistry, University of Strathclyde, Glasgow, G1 1XL, UK b

a r t i c l e

i n f o

Article history: Received 20 June 2015 Received in revised form 21 July 2015 Accepted 29 July 2015 Available online 1 August 2015 Keywords: Sugar fatty acid esters (SFAEs) Glucose laurate Lipases Ionic liquids (ILs) Bisolvent system Response surface methodology (RSM)

a b s t r a c t As biosurfactants derived from naturally renewable resources, sugar fatty acid esters have broad applications in food, cosmetic and pharmaceutical industries. Glucose laurate was synthesized by a lipase-catalyzed acylation of glucose with vinyl laurate in ionic liquid (IL) systems. 16 ILs were screened for investigating the impact of the IL’s compositions on both the solvent properties of the IL and the synthetic process. The conversions obtained in ILs showed a bell-shaped relationship with their log P values. ILs that carry hydrophobic cations and hydrophilic anions seemed to favor the sugar ester synthesis. 1-Hexyl-3-methylimidazolium trifluoromethylsulfonate ([HMIm][TfO]) was selected both as a pure solvent and as a co-solvent with 2-methyl-2-butanol (2M2B) for this application, and the influence of the affecting factors (such as reaction temperature, enzyme amount, molar ratio of the two substrates, and reaction time) has been studied. Response surface methodology (RSM) was applied to optimize the [HMIm][TfO]/2M2B bisolvent system, and an optimal productivity of 14.2 mmol/L/h was achieved, which is superior to other literature results. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Sugar fatty acid esters (SFAEs) are nonionic biosurfactants produced from naturally renewable resources (i.e., carbohydrates and fatty acids), thus being nontoxic, biodegradable and biocompatible. SFAEs are also tasteless, odorless and nonirritant, and have been widely used in food, cosmetic and pharmaceutical industries [1,2]. Lipase-mediated enzymatic synthesis of SFAEs has proven to be superior to the currently dominating chemical synthesis, and Novozym 435 (Candida antarctica lipase B (CALB) immobilized on acrylic resins from Novozyme), or CALB in general, is the most studied enzyme used for this application [1,2]. But the requirement of using volatile organic solvents (VOSs) as the reaction medium is problematic, mainly due to their inability to solubilize the two substrates of opposite polarity as well as concerns regarding their environmental health risks and enzyme compatibility [3]. Ionic

∗ Corresponding author. Present address: College of life sciences, Shenzhen University, Shenzhen 518060, China. Fax: +86 755 2653 4277. E-mail address: [email protected] (Z. Yang). http://dx.doi.org/10.1016/j.procbio.2015.07.019 1359-5113/© 2015 Elsevier Ltd. All rights reserved.

liquids (ILs) have recently emerged as an attractive “green” alternative to VOSs for carbohydrate biotransformation [3,4], particularly because of their superior solubilizing power [5] and their excellent compatibility with enzymes in support of the enzyme activity, stability and selectivity [6,7]. A search for appropriate ILs for this application requires a good understanding of the impacts of ILs on the enzymatic performance. However, So far only a few popular ILs (such as [BMIm][BF4 ] and [BMIm][PF6 ]) have been tested on sugar ester synthesis [8–10]. Therefore, it is important to set up an investigation on how the structure of an IL determines its properties and in turn the enzymatic performance in SFAE synthesis. Kim and Koo [11] have reported a preliminary study about the effects of physicochemical properties of 4 TfO- -based imidazolium ILs on enzymatic synthesis of butyl acetate. A comprehensive survey of IL-enzyme/protein interactions has been given by Yang [12]. On the other hand, optimization of the synthetic process by response surface methodology (RSM) is worth an investigation. RSM has been recognized as a successful tool for experimental design and optimization because by running only a small number of experimental trials, it can enable the building of models and the evaluation of the significance of the different factors considered as

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858

well as their interactions [13]. Up till now only a few investigations have been reported dealing with applying RSM to optimization of lipase-catalyzed sugar ester synthesis in ILs [14–16]. In this context, the production of glucose laurate by lipasecatalyzed transesterification of glucose with vinyl laurate was taken as a model reaction, and a systematic investigation was carried out by screening 16 different ILs as the reaction medium. The use of vinyl ester is superior to that of a free fatty acid in improving the production yield by driving the reaction through the tautomerization of the enol product. The major goals of this study were: (1) to evaluate the structure-property relationship of ILs; (2) to explore the impact of ILs on the SFAE synthesis; and (3) to investigate the SFAE synthesis in both a pure IL system and an IL/VOS bisolvent system and to optimize the process by utilizing RSM. It is anticipated that the results obtained from this study can provide some insights not only to the enzymatic synthesis of SFAEs but, more generally, to all biocatalytic processes in ILs. 2. Materials and methods 2.1. Materials Novozym 435 (C. antarctica lipase B immobilized on acrylic resins) was purchased from Novozymes (China) Investment Co., Ltd. Ionic liquids (99%) were obtained from ShangHai Cheng Jie Chemical Co., Ltd. Vinyl laurate (VL) and the Reichardt’s dye (2,6-diphenyl-4-(2,4,6-triphenyl-1-pyridinio) phenolate) were from Sigma–Aldrich China Inc. ␣-d-Glucose (Glc), lauric acid (LA), molecular sieves (4 Å), and all other reagents used were of analytical grade from local manufacturers. 2.2. Dissolution and solubility of glucose in ILs Glucose (100 mg) was added to a test tube containing 2.0 mL of an IL, followed by vortex mixing for 5 min and then shaking in an incubator/shaker at 30 ◦ C and 220 rpm for 12 h. After centrifugation, the supernatant was obtained for determining the glucose solubility by using the dinitrosalicylic acid (DNS) method [17].

1853

dissolved) to a 5-mL capped test tube containing 0.3 M vinyl laurate (totally dissolved) and 100 mg 4 Å molecular sieves in 1-mL solvent (pure IL or IL/VOS mixture, dried over molecular sieves for over a week prior to use). After vortex mixing for 5 min to maximize the glucose dissolution in the solvent, 100 mg Novozym 435 was added, and the tube was placed in an incubator/shaker with agitation of 300 rpm at 40 ◦ C to start the reaction. Periodically, a 10 ␮L sample was taken and 4 times diluted with DMSO for HPLC analysis as indicated below. The conversion was calculated based on the total amount of glucose added to the reaction system. The product is 6-O-lauroyl-d-glucopyranose, which has been verified by using HPLC and structural analyses with NMR, IR and MS (Lin et al., unpublished results). 2.5. HPLC analysis This was performed on a Shimadzu LC-20AT HPLC system equipped with a refractive index detector (Shimadzu RID-10A) and a 150 × 4.6 mm, 5 ␮m inertsil ODS-SP column (GL Sciences Inc. Japan). A 10-␮L sample was injected, and a solvent mixture of methanol/water (85:15 v/v, pH adjusted to 2.3 with acetic acid prior to addition of methanol) was employed as the mobile phase with a flow rate of 1.0 mL/min, operated at 40 ◦ C. The retention times for glucose laurate (GL), lauric acid (LA) and vinyl laurate (VL) were 3.31, 5.30 and 10.90 min, respectively. 2.6. Experimental design for optimizing the IL/VOS bisolvent system by RSM Four factors (i.e., reaction temperature, enzyme amount, VL/Glc molar ratio and reaction time) were selected for optimization. Single-factor experiments were carried out first, i.e., to obtain the optimum for one parameter at a time while keeping others constant. Then based on the above results, a 4-factor-5-level central composite rotatable design (CCRD) of response surface methodology (RSM) was undertaken using Design-Expert 8.0.6, a DOE software developed by Stat-Ease, Inc. The model was evaluated by regression analysis of the experimental data by means of ANOVA (analysis of variance) technique implemented in the Design-Expert software.

2.3. Physical properties of ILs Viscosity was measured at 40 ◦ C by using an AR1000 rheometer (TA Instruments, USA). Electron pair acceptor strength was determined spectrophotometrically by means of the Reichardt’s dye [18], which was dissolved in an IL at 0.5 mM. After centrifugation, the visible spectrum of the solution was scanned at 25 ◦ C with a PerkinElmer Lambda 25 UV–vis spectrophotometer. The wavelength of the absorption peak (max , in nm) was recorded for determination of the IL’s acceptor strength (ET N ) following the equation: ET N = [28591/max – 30.7]/32.4, where ET N is set on a dimensionless normalized scale between 0.0 for Si(CH3 )4 and 1.0 for H2 O. Log P (P is the partition coefficient of a solvent in an octanol/H2 O biphasic system) [19], a commonly used parameter to represent the solvent hydrophobicity, was estimated with the aid of a software, KowWin (http://www.epa.gov/opptintr/exposure/ pubs/episuite.htm). It has been shown to give the best result in log P determination among the four software packages (HyperChem, Pallas, KowWin and TOPKAT) [20], and has been used for predicting log P values of ILs [11,21]. 2.4. Enzymatic synthesis of glucose laurate in pure IL or IL/VOS systems A typical reaction was carried out by adding 0.054 g glucose (corresponding to 0.3 mol/L of the reaction system, only partially

3. Results and discussion 3.1. Correlation between properties and structures of the ILs Table 1 summarizes the physical properties of the 16 ILs (all are liquid under ambient temperature) that were tested in our study, all varying according to a change in the IL’s cation or anion. For instance, the viscosity of the ILs increased as a result of an increase in the alkyl chain length of the cation (e.g., [EMIm][TfO] < [BMIm][TfO] < [HMIm][TfO]) a decrease in the size of the anion (e.g., or [BMIm][Tf2 N] < [BMIm][TfO] < [BMIm][MeSO4 ] < [BMIm][PF6 ]). The ET N of common immidazolium and ammonium ILs normally fall in the range of 0.50–0.75 and 0.37–1.0, respectively, corresponding to that of dipolar solvents (protic or non-protic) [18]. Our measurements agree well with these ranges, suggesting that all these tested ILs are highly polar. In agreement with Carmichael and Seddon [22], our data have also revealed that the IL polarity increased slightly with a decrease in either the chain length of the alkyl substituents on the cation (e.g., [EMIm][TfO] > [BMIm][TfO] > [HMIm][TfO]) or the size of the anion (e.g., [BMIm][PF6 ] > [BMIm][TfO] > [BMIm][Tf2 N]). Log P has been generally considered as a major descriptor for the hydrophobicity of an organic solvent [19], and was used here to assess the hydrophobicity/hydrophilicity of the ILs and their

1854

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858

Table 1 Physical properties of 16 ILs.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

ILa

Density (g/cm3 )b

m.p. (◦ C)b

Viscosity (cP)

Acceptor strength (EN T )

Log P

Glc solubility (mM)

VL solubility (mM)

[BMIm][PF6 ] [BMIm][BF4 ] [Me(OEt)3 -Me-Im][Tf2 N] [BTMA][Tf2 N] [HTMA][Tf2 N] [EMIm][Tf2 N] [BMIM][Tf2 N] [HMIm][Tf2 N] [EMIm][TfO] [BMIm][TfO] [HMIm][TfO] [MMIm][MeSO4 ] [EMIm][MeSO4 ] [BMIm][MeSO4 ] [TMA][Ac] [TBA][Ac]

1.29 1.18 n.d.c 1.34 1.31 1.47 1.42 1.35 1.35 1.28 n.d.c n.d.c n.d.c n.d.c n.d.c n.d.c

9 −82 n.d.c 7 25.5 −16 −5 7 −10 13 n.d.c 48 n.d.c n.d.c n.d.c n.d.c

110.75 45.67 32.47 45.98 63.01 18.02 25.11 30.46 22.32 41.91 74.26 28.54 33.7 82.02 40.84 549.3

0.68 0.67 0.37 n.d.c n.d.c 0.67 0.6 0.65 0.68 0.66 0.63 n.d.c n.d.c n.d.c 0.59 n.d.c

n.d.c n.d.c n.d.c 1.36 2.34 −0.26 0.72 1.7 −2.82 −1.84 −0.86 −5.8 -5.31 −4.33 −2.09 3.8

7.98 26.42 17.95 5.92 23.27 5.01 5.31 5.35 71.45 27.1 8.56 1602.26 701.53 176.92 1343.79 173.17

14.49 11.32 106.52 99.52 374.32 75.83 212.07 886.27 10.85 40.82 527.07 0 1.15 9.6 0 65.84

a Abbreviations for IL cations: EMIm = 1-ethyl-3-methylimidazolium, BMIm = 1-butyl-3-methylimidazolium, HMIm = 1-hexyl-3-methylimidazolium, Me(OEt)3 -MeIm = 1-methyl-3-(2-(2-methoxyethoxy) ethoxy) ethylimidazolium, BTMA = butyltrimethylammonium, HTMA = hexyltrimethylammonium, TMA = tetramethylammonium, abbreviations for IL anions: PF6 = hexafluorophosphate; BF4 = tetrafluoroborate; Tf2 N = bis(trifluoromethylsulfonyl) imide; TBA = tetrabutylammonium; TfO = trifluoromethylsulfonate; MeSO4 = methylsulfate; Ac = acetate. b Both density and melting point data were provided by the IL manufacturer. c n.d. = not determined.

ions. Among the 16 ILs that were tested in this study, 7 of them holding anions such as TfO− , MeSO4 − and Ac− are negative in their log P values, while all Tf2 N− -based ILs (except [EMIm][Tf2 N]) are positive, confirming the hydrophilic characteristics of the former 3 IL anions and the hydrophobic nature for the last Tf2 N− anion. Reasonably, the log P value gets higher when the IL possesses a more hydrophobic alkyl chain attached on its cation or a larger anion (e.g., [HMIm][Tf2 N] > [BMIm][Tf2 N] > [EMIm][Tf2 N]; [BMIm][Tf2 N] > [BMIm][TfO] > [BMIm][MeSO4 ]). The solubilities of the two substrates, glucose and vinyl laurate, are also dependent on the nature of the cation and anion of an IL. Glucose dissolved better in an IL with a shorter alkyl chain on the cation, especially when combined with a more polar and hydrophilic anion (e.g., [MMIm][MeSO4 ] > [EMIm][TfO] > [HMIm][Tf2 N]), presumably due to the presence of the multi-hydroxyl groups on the sugar ring which can interact strongly with the IL anion through H-bonding. The order of anions (MeSO4 − > TfO− > BF4 − > PF6 − > Tf2 N− ) is similar to that of decreasing basicity, which is consistent with a role for accepting hydrogen bonds from the glucose. Indeed, an inverse relationship was noticed when plotting the glucose solubility data against the log P values of the ILs (plot not shown). The specially designed IL, [Me(OEt)3 -Me-Im][Tf2 N], dissolved more glucose than other Tf2 N− -based imidazolium ILs, simply because of the polyether-like cation that has a higher affinity for the H-bond donating glucose [23]. For high solubility of vinyl laurate, in contrast, it is favorable to use an IL with a longer alkyl chain on the cation combined with a hydrophobic anion such as Tf2 N− (Table 1). It is clear that in some ILs there is a solvophobic effect analogous to the hydrophobic effect in water [24]. This may tend to suppress solubility of vinyl laurate and so explain some of the trends observed.

3.2. IL screening and important impacts of IL properties on SFAE synthesis 16 different ILs have been screened as reaction media for enzymatic synthesis of glucose laurate, and the conversions obtained in 48 h can be compared in Fig. 1. All ILs (except [TBA][Ac]) exhibited poorer conversions relative to that obtained in the organic solvent 2-methyl-2-butanol (2M2B), one of the most com-

Fig. 1. Conversions obtained at 48 h in 16 ILs after transesterification of d-glucose with vinyl laurate. A reaction mixture containing 0.3 M glucose, 0.3 M vinyl laurate, 100 mg 4 Å molecular sieves and 100 mg Novozym 435 in 1-mL solvent was placed in an incubator/shaker with agitation of 300 rpm at 40 ◦ C for 48 h.

monly used organic solvents for sugar ester synthesis [1,25]. For ILs holding the same anion (e.g., TfO− , Ac− , Tf2 N− ), a higher conversion was obtained when the cation of the IL was constituted with a longer alkyl chain (e.g., HMIm+ > BMIm+ > EMIm+ , TBA+ > TMA+ ); while for the ILs possessing the same cation (e.g., BMIm+ ), the presence of a more hydrophilic anion promoted the conversion, ranking in the order of TfO− > BF4 − > Tf2 N− > PF6 − . All the Tf2 N− -based ILs, whether imidazolium or ammonium, exhibited a rather low conversion. No product was detected in the presence of the three MeSO4 − -based ILs. The highest conversion (34.2%) was obtained in [TBA][Ac] and the second (19.7%) in [HMIm][TfO]. According to the limited reports on this aspect so far, the superior effect of the hydrophilic anions (such as TfO− and BF4 − ) to that of the hydrophobic ones (such as Tf2 N− and PF6 − ) is evident [8,10,26,27], whereas the role of the IL cations are sometimes contrary to each other [8,11,26–28]. The investigation by Klähn et al. [29] on the role of anions and cations in the solvation of CALB in ILs via molecular dynamics simulations have shown that the strength

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858

Fig. 2. Correlation of conversions obtained at 48 h with the log P values of the ILs (a) and the log P values of the IL anions (b). Reaction conditions were the same as for Fig. 1. Log P values were estimated with the software LowWin (http://www.epa. gov/opptintr/exposure/pubs/episuite.htm).

of the enzyme-IL interaction is primarily determined by Coulomb interactions with anions and to a smaller degree by van der Waals interactions with cations. It is necessary to examine the impact of the IL properties on the enzymatic conversions obtained in ILs. In this study no obvious correlation was found between the conversions and the solubilities of either of the two substrates. But instead, a higher conversion was sometimes noticed in the IL with a lower solubility for glucose (e.g., [EMIm][TfO] < [BMIm][TfO] < [HMIm][TfO]); and no sugar ester product was produced in the three MeSO4 − -based imidazolium ILs, although glucose was highly soluble in them. Supporting examples are also found in [23,28,30]. Therefore, the speculation that an IL affording a high sugar solubility may also be a good solvent for the carbohydrate biotransformations to take place [8] is questionable [3]. Our conversion data also did not correlate well with either the viscosity or the polarity of the IL. Interestingly, however, there seems to be a loose bell-shaped relationship between the conversion and the log P value of the IL (provided that [TBA][Ac] can be taken as an outlier for some unknown reasons), and the optimal conversion was obtained in [HMIm][TfO] (log P = −0.86) (Fig. 2a). A similar relationship was also observed by Zhao et al. [31] when they carried out a transesterification between ethyl butyrate and 1-butanol in 12 ILs, catalyzed by the same enzyme (Novozym 435), and the optimum enzyme activity was also achieved at a log P = −0.90. A closer look at our data seems to suggest that the above bell-shaped relationship was contributed more by the anions than by the cations (Fig. 2b). It is important to note that log P is a property of individual species in partitioning between octanol and water. In the case of ILs, electroneutrality requires that cations and anions partition together between phases. Hence in our study, log P may be relevant

1855

if the IL ions are moving between the bulk IL and interaction sites in the solid-phase enzyme catalyst. At this scale the cation and anion can move independently to some extent, and it might be speculated that low log P ions will be more likely to interact with polar sites on the enzyme molecules and reduce their catalytic activity. In addition, higher log P ions may also interact with non-polar sites on the protein, causing changes unfavorable for catalytic activity. The molecular dynamics simulations of CALB in various imidazolium ILs have already demonstrated the diffusion of the alkyl chain of the IL cation into the active site of the enzyme [29]. It must be accepted that many other properties might contribute to differences between the ILs as reaction media. They may differ in removing essential water from the enzyme, and supporting examples have been reported previously [6]. The nucleophilicity and H-bond accepting ability of the IL anion may also be important [12], with lower values minimizing both the nucleophilic and H-bonding interactions between the IL anion and the protein, thus facilitating the enzyme activity. A contribution to the lower conversion in most IL’s of higher log P (Fig. 2a) may come from the increased solubility of vinyl laurate (Table 1). This better solvation will lead to the ground-state stabilization of this substrate, leading to a higher activation energy and hence a lower reaction rate. In fact, an enzymatic reaction in an IL is governed by a balance of the mutual interactions among the three components involved in the reaction system: the IL, the enzyme, and the substrate; and the solvent properties (such as hydrophobicity and viscosity), which are highly responsible for this balance, usually play dual functions, as has been discussed previously [32]. In addition, an IL with a high solubility for the substrate is beneficial in maximizing the substrate concentration available for the enzyme, but is also disadvantageous in stabilizing the substrate ground state so as to lower the enzymatic reaction rate. Moreover, high polarity permits an IL to be powerful in solubilizing a broad range of polar substrates such as sugars, but it may also be harmful to an enzyme by seriously interacting with it and hence denaturing it. A discussion about why the IL polarity affects the enzymatic performance so differently has been given in [6]. Therefore, the impact of ILs on enzymatic performance is a complex interplay of many factors, and the selection of an appropriate IL for sugar ester synthesis necessitates a compromise between the opposing effects caused by them. For subsequent experiments, [HMIm][TfO] was selected as both a solvent and a co-solvent for the glucose laurate synthesis, in the hope of optimizing the process, especially with the help of the RSM strategy. Our later experiments have revealed that [Bu4 N][Ac] gave a good conversion (Fig. 1) because it works both as a solvent and as a catalyst (unpublished results). 3.3. Use of [HMIm][TfO] as solvent for enzymatic synthesis of glucose laurate An initial trial in the pure IL system yielded a conversion of 13.7% in 24 h (Fig. 3, line a). Then the impacts of the affecting factors (i.e., reaction temperature, water content, enzyme amount, concentrations of the two substrates and their molar ratio) were examined in order to work out the optimum for each condition. Based on these results, the reaction conditions were modified: The temperature was increased from 40 ◦ C to 60 ◦ C and the substrate concentrations were changed from 0.3 M for both substrates to 0.05 M for Glc and 0.15 M for VL. The modified conditions resulted in an improvement in both the initial synthetic rate (from 1.9 mM/h to 3.3 mM/h) and the conversion obtained at 24 h (from 13.7% to 93.5%) (Fig. 3, line b). It is worth mentioning that the high conversion of 93.5% was to a great extent due to the reduced amount of glucose input (from 0.3 mol/L to 0.05 mol/L), and that is why the productivity was not significantly enhanced (from 1.71 mmol/L/h to 1.95 mmol/L/h). In the following experiments, the input of glucose was fixed at

1856

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858 Table 2 Variables and levels for central composite rotatable design. Variable

Symbol

Reaction temperature (◦ C) Enzyme amount (mg) VL/Glc Molar ratio Reaction time (h)

X1 X2 X3 X4

Level −2 30 60 0 4

−1

0

+1

+2

35 80 0.5 8

40 100 1 12

45 120 1.5 16

50 140 2 20

Fig. 3. The time course of the synthetic reaction in [HMIm][TfO] before (a) and after optimization (b). The reaction condition for Line a was the same as for Fig. 1, and that for Line b was modified by increasing the temperature from 40 ◦ C to 60 ◦ C and changing the substrate concentrations from 0.3 M for both substrates to 0.05 M for glucose and 0.15 M for vinyl laurate.

Fig. 5. The linear relationship between the conversions modeled by RSM and those obtained experimentally. A 4-factor-5-level central composite rotatable design (CCRD) of response surface methodology (RSM) was undertaken by using DesignExpert 8.0.6, a DOE software developed by Stat-Ease, Inc. The conversions obtained from the model were subjected to a standard deviation of 1.30 and a C.V.% of 3.74.

3.5. Optimizing the [HMIm][TfO]/2M2B bisolvent system by RSM

Fig. 4. The conversions obtained at 24 h in the [HMIm][TfO]/2M2B bisolvent system as a function of the IL/VOS volumetric ratios. Please refer to Section 2.4 for experimental details.

0.3 mol/L of the reaction system, and two strategies were employed to further improve the synthetic process: utilizing an IL/VOS bisolvent system and optimization with RSM.

3.4. Use of [HMIm][TfO] as co-solvent for enzymatic synthesis of glucose laurate Above [HMIm][TfO] has shown its potential as a reaction medium for sugar ester synthesis, although the conversion obtained in it was rather low. Mixing the IL with 2M2B is expected to provide a solution for lowering both the viscosity (74.26 cP for [HMIm][TfO] vs. 3.5 cP for 2M2B), hence promoting the conversion, and also the cost for the synthetic process. Indeed, at an optimal ratio of 0.05/0.95 (v/v, IL/2M2B) the conversion was significantly enhanced to 61.6% relative to that obtained in either the pure IL (13.4%) or pure 2M2B system (34.1%) (Fig. 4). Other researchers have reported that use of an IL/VOS bisolvent system is superior as the reaction medium for sugar ester synthesis [9,10,28,33].

The initial single-factor experiments suggested the optimum for each of the four parameters: 40 ◦ C as the reaction temperature, 12 h as the reaction time, 100 mg enzyme, and 1:1 molar ratio for VL/Glc. Based on these results, an RSM with a 4-factor-5-level CCRD was employed for modeling and optimization of the enzymatic synthesis of glucose laurate (Table 2). The volumetric ratio of the two solvents (IL/2M2B) was set at 0.05/0.95. A total of 30 runs were carried out, among which six were at the central point. For all the 30 runs, the conversions obtained by prediction from the model were linearly correlated with the experimentally obtained values with a slope of 0.9991 and an R2 -value and R2 adj value of 0.99907 and 0.99660, respectively, indicating that the model is able to fit the experimental data very well (Fig. 5). The model F-value of 410.79 and the model P-value of <0.001 imply respectively that both the model and all the model terms are significant. The adequate precision of 60.53 indicates an adequate signal. No statistical evidence of multi-collinearity was found because the VIF (variance inflation factor) values calculated for all the terms included in the model were lower than 6.4. All these parameters have elucidated the validity of the model, which well reflects the influence of each variable and their interactions on the conversion in the following polynomial equation: Y

= −27036.06 + 1343.49X 1 + 484.91X 2 + 800.40X 3 + 61.76X 4 − 24.04X 1 X 2 − 42.39X 1 X 3 − 2.95X 1 X 4 + 0.92X 2 X3 − 0.01X 2 X 4 + 2.50X 3 X 4 − 16.69X 1 2 − 2.32X 2 2 − 17.55X 3 2 − 0.15X 4 2 − 0.02X 1 X 2 X 3 + 0.30X 1 2 X 2 + 0.56X 1 2 X 3 + 0.04X 1 2 X 4 + 0.11X 1 X 2 2 − 0.0014X 1 2 X 2 2

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858

1857

Table 3 A comparison of our results with literature data in the conversion and productivity obtained after optimization with RSM (all were transesterification reactions between a sugar and an ester, catalyzed by Novozym 435). Reference Substrates

Sugar Ester Solvent Solvent volume Sugar input Enzyme amount Reaction time Reaction temperature Conversion Productivity Specific productivity

Solvent Reaction conditions

Results

[14]

[15]

[16]

This study

Naringin Vinyl butyrate [BMIm][BF4 ] 0.5 mL 0.047 M 40 mg 100 h 60 ◦ C 87% 0.41 mmol/L/h 5.11 ␮mol/h/g

Mannose Vinyl myristate [Bmpyrr][TfO] 0.5 mL 0.047 Ma 25 mg 24 h 80 ◦ C 75.60% 0.76 mmol/L/h 15.12 ␮mol/h/g

Glucose Vinyl laurate [BMIm][Tf2 N]/[BMIm][TfO] 1.0 mL 0.067 Mb 73.33 mg 6h 66.86 ◦ C 96.40% 10.71 mmol/L/h 146.07 ␮mol/h/g

Glucose Vinyl laurate [HMIm][TfO]/2M2B 1.0 mL 0.3 M 101.3 mg 16 h 40 ◦ C 75.75% 14.20 mmol/L/h 140.21 ␮mol/h/g

a The initial sugar concentration was calculated by using the following conditions used in [13]: mannose/vinyl myristate molar ratio = 1:10, total substrate quantity = 0.26 mmol. b The initial sugar concentration was calculated by converting the supersaturated glucose concentration of 12 g/L used in 16].

C o n v e r s io n ( % )

70 60 50 40 30 20 10 0

2.0

140.0 1.7

120.0 1.4

100.0 1.1

VL/Glc Molar ratio

80.0 0.8

60.0 0.5

40.0

Enzyme amount (mg)

Fig. 6. Response surface and contour diagram of conversion as a function of enzyme amount and reaction time. The diagram was generated by applying RSM as stated in Fig. 5 and Section 2.6.

where Y is the predicted conversion (%), while X1 , X2 , X3 and X4 refer to reaction temperature (◦ C), enzyme amount (mg), VL/Glc molar ratio and reaction time (h), respectively. One of the 3D response surface with contour plots was depicted in Fig. 6. A maximal conversion of 78.47% (varying in the range of 74.64–82.31%) was predicted by the model with a set of reaction conditions suggested: 101.3 mg enzyme, 1.5:1 VL/Glc molar ratio, 45 ◦ C and 16 h. Under these conditions, 3 tests were conducted and an average conversion of 75.75% (±2.86%) was obtained, which is reasonably close to the predicted value. This further confirms the validity and adequacy of the model. Our optimization through the use of CCRD is obvious, which significantly enhanced the conversion from 61.59% (Fig. 4) to 75.75% by increasing the temperature (from 40 ◦ C to 45 ◦ C) and the VL concentration (from 0.3 M to 0.45 M) while reducing the reaction time (from 24 h to 16 h) so that the production efficiency was also improved. The optimal conversion we obtained is higher than most reported previously that were obtained in organic solvents (see the review article [2] for examples cited therein). So far there are very limited reports on employing RSM for optimization of enzymatic synthesis of sugar esters or their derivatives in ILs. Under optimal conditions suggested by RSM, the acylation of naringin [14] and mannose [15], both with vinyl butyrate, yielded a maximal conversion of 87% and 75.6%, respectively, whereas a high acylation yield of 96.4% was obtained for the synthesis of glucose laurate from glucose and vinyl laurate [16]. It has to be stated that the initial sugar concentrations used in these studies (0.047 M, 0.047 M and 0.067 M, respectively) were much lower than ours (0.3 mol/L). As has been mentioned earlier, if low sugar levels are

applied, a fairly low product concentration will correspond to a high conversion (calculated based on the initial input of the sugar). Indeed, while keeping other variables constant, the conversions obtained in our [HMIm][TfO]/2M2B (0.5/0.95, v/v) system at 24 h were 100%, 95.9%, 69.4%, 52.5% and 26.9% when 0.05, 0.1, 0.2, 0.3 and 0.5 mole of glucose was added per liter of the reaction system. Therefore, it can be expected that if a same amount of glucose were applied in the reaction, our IL/2M2B bisolvent system would offer a higher conversion than the above three reported IL systems optimized with RSM. Moreover, the productivity (mmol/L/h) obtained by us was also significantly much higher than those obtained in the above 3 studies (Table 3). In terms of the specific productivity (␮mol/h/g), our result was 26-fold and 8-fold higher than those obtained by Katsoura et al. [14] and Galonde et al. [15], respectively, while being comparable to the result obtained by Mai et al. [16]. 4. Conclusion Enzymatic synthesis of glucose laurate was optimized by performing the reaction in the [HMIm][TfO]/2M2B bisolvent system through the RSM-mediated experimental design. The conversions obtained in ILs are closely correlated with the IL properties, depending on the cation and anion of the IL. ILs composed of hydrophobic cations and hydrophilic anions (such as [HMIm][TfO]) are beneficial to improving the enzymatic synthesis of SFAEs. The biocatalytic process in an IL system is governed by a balance of the mutual interactions among the IL, the enzyme and the substrate. It has to be admitted that despite their unique solvent properties, the use of ionic liquids is very challenging, especially due to their high cost and difficult recovery and reuse, and also products are not easy to be recovered from this nonconventional reaction medium. More work has to be done before biocatalytic processes in the IL system can be really employed in an industrial scale. Acknowledgement The authors are grateful to the National Natural Science Foundation of China for its financial support (Grant No. 21276159). References [1] T. Kobayashi, Lipase-catalyzed syntheses of sugar esters in non-aqueous media, Biotechnol. Lett. 33 (2011) 1911–1919. [2] A.M. Gumel, M.S.M. Annuar, T. Heidelberg, Y. Chisti, Lipase mediated synthesis of sugar fatty acid esters, Process Biochem. 46 (2011) 2079–2090. [3] Z. Yang, Z.-L. Huang, Enzymatic synthesis of sugar fatty acid esters in ionic liquids, Catal. Sci. Technol. 2 (2012) 1767–1775. [4] N. Galonde, K. Nott, A. Debuigne, M. Deleu, C. Jeroˆme, M. Paquot, J.-P. Wathelet, Use of ionic liquids for biocatalytic synthesis of sugar derivatives, J. Chem. Technol. Biotechnol. 87 (2012) 451–471.

1858

X.-S. Lin et al. / Process Biochemistry 50 (2015) 1852–1858

[5] Q. Liu, M.H.A. Janssen, R. van, F. antwijk, R.A. Sheldon, Room-temperature ionic liquids that dissolve carbohydrates in high concentrations, Green Chem. 7 (2005) 39–42. [6] Z. Yang, W. Pan, Ionic liquids: green solvents for nonaqueous biocatalysis, Enzyme Microb. Technol. 37 (2005) 19–28. [7] R. van, F. antwijk, R.A. Sheldon, Biocatalysis in ionic liquids, Chem. Rev. 107 (2007) 2757–2785. [8] S. Park, R.J. Kazlauskas, Improved preparation and use of room-temperature ionic liquids in lipase-catalyzed enantio- and regioselective acylations, J. Org. Chem. 66 (2001) 8395–8401. [9] F. Ganske, U.T. Bornscheuer, Optimization of lipase-catalyzed glucose fatty acid ester synthesis in a two-phase system containing ionic liquids and t-BuOH, J. Mol. Catal. B: Enzym. 36 (2005) 40–42. [10] E. Abdulmalek, H.S.M. Saupi, B.A. Tejo, M. Basri, A.B. Salleh, R.N.Z.R.A. Rahman, M.B.A. Rahman, Improved enzymatic galactose oleate ester synthesis in ionic liquids, J. Mol. Catal. B: Enzym. 76 (2012) 37–43. [11] H.S. Kim, Y.-M. Koo, Effects of physicochemical properties of ionic liquids on butyl acetate synthesis using Candida antarctica lipase B, Korean J. Chem. Eng. 29 (2012) 1610–1614. [12] Z. Yang, Ionic liquids & proteins: academic and some practical interactions, in: P. Domínguez de María (Ed.), Ionic Liquids in Biotransformations & Organocatalysis: Solvents and Beyond, John Wiley & Sons, Inc., New Jersey, 2012, pp. 15–72, Chapter 2. [13] M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Response surface methodology (RSM) as a tool for optimization in analytical chemistry, Talanta 76 (2008) 965–977. [14] M.H. Katsoura, P. Katapodis, F.N. Kolisis, H. Stamatis, Effect of different reaction parameters on the lipase-catalyzed selective acylation of polyhydroxylated natural compounds in ionic liquids, Process Biochem. 42 (2007) 1326–1334. [15] N. Galonde, Y. Brostaux, G. Richard, K. Nott, C. Jeroˆme, C. Fauconnier, Use of response surface methodology for the optimization of the lipase-catalyzed synthesis of mannosyl myristate in pure ionic liquid, Process Biochem. 48 (2013) 1914–1920. [16] N.L. Mai, K. Ahn, S.W. Bae, D.W. Shin, V.K. Morya, Y.-M. Koo, Ionic liquids as novel solvents for the synthesis of sugar fatty acid ester, Biotechnol. J. 9 (2014) 1–8. [17] G.L. Miller, Use of dinitrosalicylic acid reagent for determination of reducing sugar, Anal. Chem. 31 (1959) 426–428. [18] C. Reichardt, Polarity of ionic liquids determined empirically by means of solvatochromic pyridinium N-phenolate betaine dyes, Green Chem. 7 (2005) 339–351. [19] C. Laane, S. Boeren, K. Vos, C. Veeger, Rules for optimization of biocatalysis in organic solvents, Biotechnol. Bioeng. 30 (1987) 81–87.

[20] E. Benfenati, G. Gini, N. Piclin, A. Roncaglioni, M.R. Varí, Predicting log P of pesticides using different software, Chemosphere 53 (2003) 1155–1164. [21] L. Chrzanowski, M. Stasiewicz, M. Owsianiak, A. Szulc, A. Piotrowska-Cyplik, A.K. Olejnik-Schmidt, B. Wyrwas, Biodegradation of diesel fuel by a microbial consortium in the presence of 1-alkoxymethyl-2-methyl-5-hydroxypyridinium chloride homologues, Biodegradation 20 (2009) 661–671. [22] A.J. Carmichael, K.R. Seddon, Polarity study of some 1-alkyl-3-methylimidazolium ambient-temperature ionic liquids with the solvatochromic dye, Nile red, J. Phys. Org. Chem. 13 (2000) 591–595. [23] H. Zhao, G.A. Baker, Z. Song, O. Olubajo, T. Crittlea, D. Peters, Designing enzyme-compatible ionic liquids that can dissolve carbohydrates, Green Chem. 10 (2008) 696–705. [24] T.L. Greaves, C.J. Drummond, Solvent nanostructure, the solvophobic effect and amphiphile self-assembly in ionic liquids, Chem. Soc. Rev. 42 (2013) 1096–1120. [25] M.V. Flores, P.J. Halling, Full model for reversible kinetics of lipase-catalyzed sugar ester synthesis in 2-methyl 2-butanol, Biotechnol. Bioeng. 78 (2002) 795–801. [26] S.H. Lee, T. Dang, S.H. Ha, W.-J. Chang, Y.-M. Koo, Lipase-catalyzed synthesis of fatty acid sugar ester using extremely supersaturated sugar solution in ionic liquids, Biotechnol. Bioeng. 99 (1) (2008) 1–8. [27] N. Galonde, K. Nott, G. Richard, A. Debuigne, F. Nicks, C. Jeroˆme, M.-L. Fauconnier, Study of the influence of pure ionic liquids on the lipase-catalyzed (trans) esterification of mannose based on their anion and cation nature, Curr. Org. Chem. 17 (2013) 763–770. [28] Y. Hu, G. Guo, B.-M. Lue, X. Xu, Enzymatic synthesis of esculin ester in ionic liquids buffered with organic solvents, J. Agric. Food Chem. 57 (2009) 3845–3852. [29] M. Klähn, G.S. Lim, A. Seduraman, P. Wu, On the different roles of anions and cations in the solvation of enzymes in ionic liquids, Phys. Chem. Chem. Phys. 13 (2011) 1649–1662. [30] R. van, F. antwijk, F. Secundo, R.A. Sheldon, Structure and activity of Candida antarctica lipase B in ionic liquids, Green Chem. 8 (2006) 282–286. [31] H. Zhao, G.A. Baker, Z. Song, O. Olubajo, L. Zanders, S.M. Campbell, Effect of ionic liquid properties on lipase stabilization under microwave irradiation, J. Mol. Catal. B: Enzym. 57 (2009) 149–157. [32] Z.-L. Huang, T.-X. Yang, J.-Z. Huang, Z. Yang, Enzymatic production of biodiesel from Millettia pinnata seed oil in ionic liquids, BioEnerg. Res. 7 (2014) 1519–1528. [33] Z.-G. Chen, M.-H. Zong, G.-J. Li, Lipase-catalyzed acylation of konjac glucomannan in ionic liquids, J. Chem. Technol. Biotechnol. 81 (2006) 1225–1231.