High-performance enzymatic membrane bioreactor based on a radial gradient of pores in a PSF membrane via facile enzyme immobilization

Abstract

Enzymatic membrane bioreactors (EMBRs), endowed with synergistic catalysis-separation performances, have in recent decades shown enormous potential for practical applications. Conventionally, membrane properties and operating parameters play important roles in catalysis-separation processes of these complicated and large-scale systems. Therefore, to achieve higher catalytic and filtration efficiencies, hollow fiber polysulfone microfiltration membranes with a perfect radial gradient of pores were selected as substrates, and the subsequent enzyme-immobilization process was achieved in a facile way by pressure-driven filtration and crosslinking, to finally construct an enhanced EMBR system. Lipase from Candida rugosa was introduced as the functional enzyme and was crosslinked by glutaraldehyde (GA), with the catalytic hydrolysis of glycerol triacetate used as the model reaction. The performances of these designed EMBR systems were evaluated using the response surface methodology to optimize various operating parameters by testing substrate concentrations from 0.05–0.2 M, membrane fluxes from 102.6–287.4 L m⁻² h⁻¹, reaction pH values from 5.5–8.5, and temperatures from 25–55 °C. Corresponding association models were then established according to EMBR performances to obtain R² values of 96.69% and 97.27% respectively. The complete EMBR system showed an excellent performance of around 0.178 mmol min⁻¹ g⁻¹ under optimum operating conditions, and showed marked improvement in the stability and membrane activity of the EMBR after microfiltration and crosslinking. This simple and low-cost approach to fabricate a high-performance EMBR has great potential for applications in various industrial-scale lipase-catalyzed reactions.

---

1. Introduction

Enzymes are highly specific biocatalysts, and function at high conversion rates under mild environmental conditions in aqueous solutions, and with unique selectivity and specificity to substrates; these attributes thus guarantee their outstanding properties and prospects in applications as commercial catalysts in industrial processes.^1–3 Most enzymes have been immobilized onto insoluble support materials to achieve better stability during continuous operation, which is beneficial for the economic viability of the process by promoting enzyme recycling and re-use, and significantly improving overall productivity and robustness.⁴ There are various types of enzyme-immobilization supports, such as nanofibers,⁵ nano/microspheres,^6–9 beads,¹⁰ gels,¹¹ and capsules,¹² but polymer membranes, with their easily controlled properties and well-determined structures, have shown the most potential for industrial applications.¹³ The resulting enzyme-immobilized membranes have been assembled into reactors to construct enzymatic membrane bioreactors (EMBRs), integrating multiple functions of efficient biocatalysts such as facile purification, continuous operation, little inhibition, and enzyme re-use into a single EMBR system.^13–15 Meanwhile, scaling up these EMBR systems is relatively convenient, and critical operational parameters can be easily optimized.^15,16

Lipases, formally known as glycerol ester hydrolases, catalyze a series of diverse reactions, such as hydrolysis, alcoholysis, aminolysis, and transesterification, with various organic substrates. In recent decades, EMBRs that include lipases have shown a growing tendency to be used in mainstream markets worldwide in the food, pharmaceutical, and detergent industries, as their practical applications are of high value.^17,18 Lipases have been traditionally immobilized onto ultra/microfiltration membranes via physical adsorption, which is favorable for cost effectiveness, procedural simplicity, and retention of catalytic activity.^19,20 However, due to weak noncovalent interactions between lipases and supports, the risk of lipases leaching from the support material is high, resulting in enzyme loss.²¹ Preliminary attempts have been made to improve the operational stability of EMBRs by covalently bonding and hence immobilizing lipases onto membrane surfaces.^22,23 Nevertheless, such methods also lead to particularly complicated procedures in fabricating enzyme-modified supports and inevitably lower enzyme activities because of resulting restriction on the enzyme conformation.²⁴ Therefore, any method combining the advantages of physical adsorption and covalent binding is in desperate need of industrial-scale EMBRs, with their convenient manufacturing procedures, to attain high enzyme loading as well as stable catalytic activity. A facile filtration/crosslinking method, with pressure-driven filtration and crosslinking, has been applied to construct such an enhanced EMBR system.

In addition, membrane supports with limited mass transfer resistance are expected to become commercially available.²⁵ Ideal membrane supports are required not only to provide high porosity to reduce filtration resistance, but also to avoid excessively large pore structures for enhancing the overall mechanical properties of the membranes.²⁶ Therefore, membranes composed of many pores that are distributed in a perfect radial gradient are particularly advantageous, because this special structure may contribute to more enzyme molecules entering into membrane pores under low pressure while making them hard to escape, and hence lower the resistance to diffusion during the operating procedure.²⁷

In this study, an EMBR was prepared based on the hollow fiber polysulfone (PSF) microfiltration membrane with a perfect radial gradient of pores, high flux, and low resistance to permeation. Lipase from Candida rugosa was retained in the membrane by the combination of pressure-driven filtration and crosslinking. Glutaraldehyde (GA) was used as the crosslinking agent to form a stable enzyme aggregate in/on the membrane. In addition, the effect of the initial enzyme concentration and GA concentration on the EMBR activity was investigated using glycerol triacetate hydrolysis as the model reaction, and the corresponding produced acetic acid was linearly fit. To better understand the effects of various operational variables (i.e., membrane flux, substrate concentration, reaction temperature, and pH) on the performance of the EMBR and to obtain the optimum experimental conditions for realizing the highest activity, we applied response surface methodology (RSM), a method comprising a group of statistical techniques for empirical model building and exploitation.

2. Experimental

2.1 Materials

GA was purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China) and used without further purification. Candida rugosa lipase (CRL) powder (1150 units per mg solid), Bradford reagent, bovine serum albumin (BSA, molecular mass: 67 000 Da), and glycerol triacetate were obtained from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA) and used as received. All the other chemicals were of analytical grade and used without further purification.

A homemade hollow fiber PSF microfiltration membrane with an average pore size of 0.5316 μm was prepared by using the common phase inversion method.^28,29 The hollow fiber membrane comprised a perfect gradient of pores, whose sizes gradually decreased from the inner to outer surface. The prepared membrane was rinsed with deionized water followed by phosphate buffer solution (PBS 0.05 M, pH 7.0) thrice before use. Water permeability was then measured at 0.02 MPa for 2 min.

2.2 Preparation of enzyme-immobilized membranes

Lipase solutions were prepared by adding an appropriate amount of lipase powder in PBS. The insoluble impurities of the enzyme solution were removed by centrifugation at 4000 rpm for 15 min. The prepared hollow fiber PSF membrane module was placed in a dead-end filtration apparatus (Fig. 1), and then the enzyme solution was stirred continuously and filtrated under a flux of 102.6 L m⁻² h⁻¹. To crosslink the adsorbed enzymes, a specified concentration of 40 mL GA buffer solution was filtrated slowly through the membrane module within 0.5 h under the same membrane flux. Afterwards, the enzyme-immobilized membrane was washed for 30 min with 40 mL PBS under a higher flux of 197.7 L m⁻² h⁻¹ to remove the weakly adsorbed enzymes and the residual GA solution. The washings together with the reaction solution were collected for the determination of protein concentration. The amount of enzyme immobilized on the membrane was measured by the Bradford assay using the Coomassie brilliant blue reagent.³⁰ The enzyme loading was defined as the amount of enzyme (mg) per gram of the membrane. Each value was the mean of at least three parallel experiments, and the standard deviation was approximately 5%.

Fig. 1 Schematic representation of enzyme immobilization and catalysis process and the dead-end filtration apparatus.

A field-emission scanning electron microscope (FE-SEM, Hitachi, S4800, Japan) was used to observe the morphology of the membrane surface and of its cross section, and a liquid–liquid displacement porometer (LLP-1200A, Porous Materials Inc., USA) was used to determine the change of the pore size from before to after the enzyme immobilization. An energy-dispersive X-ray spectrometer (EDX, Hitachi, S4800, Japan) and a fluorescence microscope (Nikon, Ti-U, Japan) were used to detect the immobilized enzyme.

2.3 Activity assay for the enzyme-immobilized membrane reactor

The performance of the enzymatic membrane reactor was measured through the use of dead-end filtration. Briefly, to effect a continuous hydrolysis process in this reactor, a sample of 100 mL of aqueous glycerol triacetate was circulated using a peristaltic pump by using a magnetic stirrer. To keep the reaction temperature constant, the substrate reservoir was connected to a water bath at a set temperature. Moreover, the hydrolysis reaction was carried out under a set membrane flux. With the continuous hydrolysis of glycerol triacetate, the free acetic acid was dispersed into the substrate solution, hence lowering the pH. The substrate solution was neutralized by adding 0.1 M NaOH using an automatic titrator, and the pH of the solution was hence kept constant during the entire process. The performance of the EMBR was evaluated by periodically recording the apparent reaction rate based on the volume of the standard NaOH solution that was consumed. The apparent activity of the enzymatic membrane corresponded to the release of 1 mmol acetic acid per min per gram of the membrane under the assay conditions. Each data point was the average of at least three parallel experiments, and the standard deviation was ∼5%. Blank experiments were performed without immobilized lipase on the membrane to subtract the effect of self-hydrolysis at various pH values and temperatures.

2.4 Experimental optimization

RSM is a useful collection of statistical and mathematical techniques for developing, improving, and optimizing processes in which a response of interest is affected by several variables and for optimizing the response.³¹ RSM was applied to optimize the reaction conditions and in turn to maximize the performance of the enzymatic membrane reactor. We optimized four variables—substrate concentration, membrane flux, reaction temperature, and reaction pH—that significantly affect the activity of the immobilized enzyme. Two 2-factor-3-level central composite designs (CCDs) were used. Each design was constructed with Design Expert 8.0 software (Minneapolis, MN, USA), and a total of 13 unique experiments were performed (Table 1, ESI†). All of the experiments were carried out as simultaneously as possible to minimize the effect of experimental error in the observed response.

2.5 Desorption experiment

The lipase-immobilized membranes were continuously filtered by phosphate buffer solution (PBS 0.05 M, pH 7.0) under a membrane flux of 197.7 L m⁻² h⁻¹ (25 °C). The amount of lipase in the filtered PBS was analyzed every 40 minutes as described above. The desorption rate of lipase was calculated by the following expression:

2.6 Reusability test of the enzymatic membrane reactor

To evaluate their reusability, the tested lipase-immobilized membranes were washed with PBS to remove any residual substrate and product after the hydrolysis reaction, followed by filtering fresh substrate solution under the same experimental conditions. The hydrolysis reactions were conducted at 25 °C using 0.2 M glycerol triacetate solution as the substrate under an optimal membrane flux. The same procedure was repeated up to six times. The relative catalytic activities of the immobilized lipase were normalized to the highest activity.

3. Results and discussion

3.1 Characterization of PSF membrane before and after enzyme immobilization

Changes in the morphology and properties of the membrane would be expected to affect the performance of the EMBR. The SEM images in Fig. 2 indicated that the support PSF membrane displayed abundant pores arranged in a perfectly radial gradient in which the pores gradually became smaller from the inner surface to the outer surface. The pore structure of the membrane was observed to be more compact closer to the outer surface, and a series of irregular large pores were observed closer to the inner surface. After the immobilization process, no obvious change in the morphology of the membrane was observed. However, the average pore size decreased from 0.5316 to 0.4424 μm (ESI, Fig. 1†) leading to a decrease in the flux from 310 L m⁻² h⁻¹ to 286 L m⁻² h⁻¹ (0.02 MPa) for an enzyme loading of 7.46 mg g⁻¹. The adsorbed proteins in the membrane pores decreased the permeability of the enzyme-loaded membrane.³² To verify the adsorption of proteins on the membrane, FITC-labeled BSA was used as the model protein for the same immobilization study. Then, the outer surface and cross-section of the BSA-adsorbed membrane were observed under an emission wavelength of 488 nm using a fluorescence microscope. The element mapping of nitrogen shown in Fig. 3 indicated an even distribution on the outer surface and in the membrane pores after the immobilization of lipase. Furthermore, the fluorescence microscopic image also confirmed the successful retention of proteins in the membrane.

Fig. 2 Images of PSF hollow fiber membranes before (a, c, e and g) and after (b, d, f and h) lipase immobilization.

Fig. 3 Nitrogen distribution (measured by EDX) on the outer surface (a) and cross-section (b) of the membrane immobilized by CRL; fluorescence microscopic images of the outer surface (c) and cross-section (d) of the membrane immobilized with FITC-BSA.

3.2 Effect of initial lipase concentration on enzyme loading

To a certain degree, a high enzyme loading is desired to enhance the efficiency of biotransformation, which may be beneficial for large-scale applications. Fig. 4 shows the amount of lipase adsorbed under different initial lipase concentrations as indicated by the enzyme loading, indicating that the enzyme loading increased with increasing initial lipase concentration and then reached a plateau value of 7.46 mg g⁻¹. At first, the increasing enzyme loading was attributed to the increasing initial enzyme concentration from 0.5 to 2.0 mg mL⁻¹, and more lipase molecules could be adsorbed into the membrane pores and onto the surfaces. However, the adsorption sites on the membrane were limited, and the pores and surfaces were entirely occupied by lipase when the lipase concentration reached 2.0 mg mL⁻¹.

Fig. 4 Effect of the initial concentration of lipase on the adsorbed amount of protein on the membrane.

3.3 Effect of initial concentration of lipase on the activity of enzymatic membrane reactor

The amount of enzyme loaded on the membrane has a significant effect on the performance of the EMR. In this study, different amounts of enzyme were immobilized on the membrane by varying the initial concentration of the filtered enzyme solution. As shown in Fig. 5, the increase in the initial enzyme concentration from 0.5 to 1.0 mg mL⁻¹ doubled the apparent reaction rate from 0.024 to 0.049 mmol min⁻¹ g⁻¹. However, a further increase in the enzyme concentration dramatically decreased the reaction rate. This trend can be explained in two ways: (a) the mass transfer of the substrate and products through the membrane was affected by the high density of lipase in the porous area of the membrane, which would have inhibited the enzymatic reaction; and (b) the over-crowded lipase in the limited porous area of the membrane offered more chances for protein–protein interactions, which would have reduced the conformational flexibility of the enzyme and lower the accessibility of the catalytic site, finally affecting its activity. Thus, ∼4.85 mg g⁻¹ was chosen in the subsequent experiments, and the corresponding initial concentration of the lipase solution for immobilization was 1.0 mg mL⁻¹.

Fig. 5 Effect of the initial concentration of lipase on the activity of the enzymatic membrane reactor.

3.4 Effects of the cross-linking process on the enzyme loading and the activity of enzymatic membrane reactor

To ensure the stability of the enzyme adsorbed on the membrane, the GA solution was further filtrated to crosslink the enzymes, which aggregated the immobilized enzyme more durably under a high shear force condition. As shown in Fig. 6, an apparent increase in the membrane activity from 0.049 to 0.069 mmol min⁻¹ g⁻¹ was observed when the GA concentration increased from 0% (v/v) to 1.25% (v/v), followed by a sharp decrease with a further increase in the GA concentration. A membrane activity of only 0.026 mmol min⁻¹ g⁻¹ was obtained when the GA concentration was 4% (v/v). This decrease can be attributed to the addition of extra GA leading to excessive crosslinking between lipases, which may have gradually denatured the enzyme and led to an ultimate decrease in the performance of the enzymatic membrane. This catalytic behavior is in good agreement with the results observed from previous studies.³³ Furthermore, no change in the enzyme loading with increasing GA concentration was observed when using the same experimental process. Therefore, a GA concentration of 1.25% (v/v) was chosen as the optimal crosslinking condition.

Fig. 6 Effect of the concentration of GA on the adsorbed amount of protein on the membrane and the activity of the enzymatic membrane reactor.

3.5 Desorption experiments

The firmness of the immobilized lipase on the membrane is crucial to the long-term use of the EMBR. The plot of the desorption rate as a function of time is shown in Fig. 7. The desorption ratio of the EMBR without crosslinking agent reached 63.4% within 180 min, indicating a very weak interaction between the membrane and the purely physically adsorbed lipase. Therefore, the lipase can be easily flushed away under a high shear force condition. In contrast, the desorption rate remained at no greater than 10% after crosslinking, because crosslinking prevented leakage of the enzyme during operation of the reactor.

Fig. 7 Effect of rinsing time on the desorption rate of the lipase immobilized by filtration.

3.6 Optimization of the operating conditions of the EMBR

To obtain the best performance of the EMR, an appropriate operating condition should be selected. To optimize the operating conditions of the EMR, RSM was applied to model the apparent enzymatic membrane activity of the immobilized lipase using four parameters: substrate concentration (C), membrane flux (L m⁻² h⁻¹), reaction pH, and temperature (°C). These four variables were optimized using two 2-factor-3-level CCDs. The accuracy of the models was evaluated by the coefficient of determination (R²). The value of R² for designs 1 and 2 were 96.69 and 97.27%, respectively. The analysis of variance (ANOVA) (Table 2 in ESI†) showed that the models for both designs 1 and 2 were statistically significant, with a significance level of P < 0.0001, demonstrating the precision and reliability of these designs.

To evaluate the relationship between the substrate concentration and membrane flux, a response surface plot was constructed at a specified reaction temperature and reaction pH (Fig. 8). From the response surface plot, the membrane activity was found to increase rapidly with increasing concentration of substrate. In general, at a certain membrane flux, the immobilized lipase could easily contact the substrate at a high enough substrate concentration, resulting in an efficient mass transfer of the immobilized lipase and an improved apparent reaction rate. In contrast, at membrane fluxes >197.7 L m⁻² h⁻¹ and a certain substrate concentration, the membrane activities apparently decreased. Membrane flux is known to be one of the crucial factors determining the effectiveness of the transmission of the substrate and the product and determining the contact time between the substrate and the immobilized lipase. As a result, the membrane flux increased from 102.6 to 196.7 L m⁻² h⁻¹ (0.02 MPa) when the flow rate of the peristaltic pump increased, leading to the continuous removal of the product. This effect can shift the reaction equilibrium towards the product side, and hence increase the reaction productivity, and was hence found to be a notable advantage of the lipase-immobilized membrane bioreactors.³⁴ Moreover, a long reaction time was not required for the immobilized lipase to hydrolyze the substrate under a high membrane flux of 287.4 L m⁻² h⁻¹. The substrate was immediately washed away, decreasing the membrane activity.

Fig. 8 Response surface plot of the relationships between concentration, flux, and membrane activity.

The membrane activity as a function of reaction temperature and pH is shown in Fig. 9. The hydrolysis of glycerol triacetate was performed at the optimal concentration and flux of 0.2 M and 197.7 L m⁻² h⁻¹, respectively. The response surface plot showed an increase up to a maximum value of 0.178 mmol min⁻¹ g⁻¹, indicating that the membrane activity did not correspondingly increase at a temperature greater than 40 °C and pH greater than 7.0. Compared to the optimum reaction temperature and pH of free lipase, the immobilized lipase was more tolerant of high temperature and pH. Two possible explanations can be put forth to account for this phenomenon. In general, immobilizing an enzyme on a polysulfide support would result in a shift of the optimum pH to more alkaline levels. According to the allocation effects, the hollow fiber PSF membrane was negatively charged and would attract H⁺ from the solution to its surface, thus acidifying the microenvironment around the immobilized lipase. Nevertheless, the pH value of the bulk solution increased, so that the effect of the microenvironment could be counterbalanced. The increased reaction temperature influenced the membrane activity, because of the restraints on the conformational flexibility of the enzyme resulting from the interaction between the enzyme and the support or due to a low restriction of the substrate diffusion at higher temperatures. Moreover, the resistance to mass transfer decreased at higher temperatures, and such a decrease would benefit the occurrence of sufficient contact between the substrate and lipase.

Fig. 9 Response surface plot of the relationships between pH, temperature, and membrane activity.

3.7 Reusability of the enzymatic membrane reactor

The ability to reuse the immobilized lipase is significant for ensuring the economical use of the enzyme in repeated batches or continuous reactions. An EMBR with a relatively long lifetime would significantly reduce costs and increase its industrial applications. Enzyme immobilization with a dead-end filtration and followed by crosslinking is a simple and efficient method not only to enhance the performance of the EMR but also to improve its reusability. In the reusability study, the residual activity of the bioreactor in the first round was set at 100%, and the hydrolysis conversions in the subsequent reactions were compared. As shown in Fig. 10, 90% of the original activity of the crosslinked EMBR was retained after six reuses, whereas the membrane residual activity of non-crosslinked EMBR was approximately 65%. The activity loss was inevitable, relating to the inactivation of the immobilized lipase by continuous use, while the reusability of the crosslinked EMBR was much better probably because the aggregation of the lipase molecules was more tolerant of outside influences.¹³ Whereas the percentage of the activity of the immobilized lipase that was retained after several reuses for catalysis was previously reported to be only ∼50%,³⁵ the result in the current study revealed an outstanding recyclability.

Fig. 10 Reusability of the lipase-immobilized PSF membrane in the bioreactor.

4. Conclusions

A lipase-immobilized membrane bioreactor with enhanced performance was prepared by immobilizing lipase in/on a PSF hollow fiber microfiltration membrane with a radial gradient of pores through filtration and crosslinking. The properties of the enzymatic membrane reactor using the hydrolysis of glycerol triacetate as the model reaction were investigated in detail. Fabrication and catalytic condition parameters were optimized, by testing substrate concentrations ranging from 0.05–0.2 M, membrane fluxes from 102.6–287.4 L m⁻² h⁻¹, reaction pH values from 5.5–8.5, and temperatures from 25–55 °C, to achieve a membrane activity up to 0.178 mmol min⁻¹ g⁻¹. Moreover, the bioreactor showed excellent operational stability, with 90% of the original membrane activity remaining after six reuses. In conclusion, immobilization of lipase on/in a polysulfone microfiltration membrane by filtration and crosslinking with glutaraldehyde can achieve an increased membrane activity and a good operation. In addition, the reusability of the enzyme due to the stability of the set up would be beneficial for industrial-scale applications.

Acknowledgements

The authors are grateful for the financial support from the National Natural Science Foundation of China (Grant No. 51473143 and No. 21274126).

Notes and references

1. O. Barbosa, C. Ortiz, A. Berenguer-Murcia, R. Torres, R. C. Rodrigues and R. Fernandez-Lafuente, Biotechnol. Adv., 2015, 33, 435.
2. M. T. Reetz, J. Am. Chem. Soc., 2013, 135, 12480.
3. J. H. Schrittwieser and V. Resch, RSC Adv., 2013, 3, 17602.
4. N. Hilal, V. Kochkodan, R. Nigmatullin, V. Goncharuk and L. Al-Khatib, J. Membr. Sci., 2006, 268, 198.
5. P. C. Chen, X. J. Huang and Z. K. Xu, Cellulose, 2014, 21, 407.
6. T. Liu, L. Qu, K. Qian, J. Liu, Q. Zhang, L. Liu and S. Liu, Chem. Commun., 2016, 52, 1709.
7. Z. Y. Zhao, J. Liu, M. Hahn, S. Qiao, A. P. J. Middelberg and L. He, RSC Adv., 2013, 3, 22008.
8. J. Liu, C. Li, Q. Yang, J. Yang and C. Li, Langmuir, 2007, 23, 7255.
9. S. Ghasemi, M. Heidary, M. A. Faramarzi and Z. Habibi, J. Mol. Catal. B: Enzym., 2014, 100, 121.
10. C. Wu, N. Wang, W. Zhang, S. Zhang, Y. Meng and X. Yu, J. Biotechnol., 2015, 194, 12.
11. Y. C. Qian, P. C. Chen, X. Y. Zhu and X. J. Huang, RSC Adv., 2015, 5, 44031.
12. S. Zhang, Z. Jiang, X. Wang, C. Yang and J. Shi, ACS Appl. Mater. Interfaces, 2015, 7, 19570.
13. P. Jochems, Y. Satyawali, L. Dielsab and W. Dejonghea, Green Chem., 2011, 13, 1609.
14. P. Andrić, A. S. Meyer, P. A. Jensen and K. Dam-Johansen, Biotechnol. Adv., 2010, 3, 407.
15. L. Giorno and E. Drioli, Trends Biotechnol., 2000, 8, 339.
16. M. Pinelo, G. Jonsson and A. S. Meyer, Sep. Purif. Technol., 2009, 70, 1.
17. R. Sharmaa, Y. Chistib and U. C. Banerjee, Biotechnol. Adv., 2001, 19, 627.
18. H. T. Deng, Z. K. Xu, Z. W. Dai, J. Wu and P. Seta, Enzyme Microb. Technol., 2005, 36, 996.
19. J. Luo, A. S. Meyer, G. Jonsson and M. Pinelo, Bioresour. Technol., 2013, 147, 260.
20. J. Luo, A. S. Meyer, R. V. Mateiu, D. Kalyani and M. Pinelo, ACS Appl. Mater. Interfaces, 2014, 6, 22894.
21. K. Sakaki, L. Giorno and E. Drioli, J. Membr. Sci., 2001, 184, 27.
22. R. Dayal and T. Godjevargova, Macromol. Biosci., 2005, 5, 222.
23. M. S. Eldin, H. A. El Enshasy, M. El Sayed, S. El Sayed, B. Haroun and E. A. Hassan, J. Appl. Polym. Sci., 2012, 124, E27.
24. H. Shi, Y. Wang and G. Luo, Ind. Eng. Chem. Res., 2014, 53, 1947.
25. J. Wöltinger, A. Karau, W. Leuchtenberger and K. Drauz, Adv. Biochem. Eng./Biotechnol., 2005, 92, 289.
26. J. K. Poppe, R. Fernandez-Lafuente, R. C. Rodrigues and M. A. Z. Ayub, Biotechnol. Adv., 2015, 33, 511.
27. A. Liese and L. Hilterhaus, Chem. Soc. Rev., 2013, 42, 6236.
28. Z. G. Wang, Z. K. Xu and L. S. Wan, J. Membr. Sci., 2006, 278, 447.
29. X. J. Huang, L. W. Wang, L. L. Zhang and Q. L. Gao, Super-hydrophilic and gradient hole structured hollow fiber membrane, ZL:201410081610:A, 2014.
30. M. M. Bradford, Anal. Biochem., 1976, 72, 248.
31. D. Baş and İ. H. Boyacı, J. Food Eng., 2007, 78, 836.
32. B. Zeuner, J. Luo, C. Nyffenegger, V. Aumala, J. D. Mikkelsen and A. S. Meyer, Enzyme Microb. Technol., 2014, 55, 85.
33. W. Yujun, X. Jian, L. Guangsheng and D. Yongyuan, Bioresour. Technol., 2008, 99, 2299.
34. C. C. Lee, H. P. Chiang, K. L. Li, F. H. Ko, C. Y. Su and Y. S. Yang, Anal. Chem., 2009, 81, 2737.
35. P. C. Chen, X. J. Huang, F. Huang, Y. Ou, M. R. Chen and Z. K. Xu, Cellulose, 2011, 18, 1563.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra25602j

---