Design of a novel multi-layer enzyme membrane reactor for low-fouling, tailored production of oligodextran

Enzymatic conversion processes face challenges in controlling oligosaccharide molecular weight (Mw). Enzymatic membrane reactors (EMRs) with immobilized enzymes address this, but direct enzyme immobilization on the membrane surface can lead to deactivation and reduced hydrolysis efficiency. This study proposes a novel EMR configuration: a three-layer structure. An electrospun porous fibrous layer, modified with PDA, TA


Introduction
Oligosaccharides are bioactive molecules whose properties are highly related to their molecular weight (Mw).Oligodextran, a specific class of oligosaccharide, holds significant relevance as it serves as a precursor in the production of iron-dextran-a crucial therapeutic agent for the treatment of anemia.Extensive research, as exemplified by Critchley and Dundar (2007), consistently highlights the lower medical risks associated with low molecular weight (Mw) oligodextran.Consequently, the precise control of oligosaccharide Mw emerges as a critical imperative.To synthesize the oligosaccharides, the manufacturing routes are bottom-up synthesis or depolymerization of the polysaccharides (Kruschitz and Nidetzky, 2020).To reduce the generation of hazardous side products, enzymes can work as green catalysts that convert large polysaccharide molecules to short-chain oligosaccharides under mild conditions (Shoda et al., 2016).The produced oligosaccharides are then fractionated in a separation process.Enzymatic membrane reactor (EMR) is a green technology that combines enzyme conversion and membrane separation in one setup, where the membrane retains the enzymes and polysaccharides for the bioconversion, and the products with lower Mw can pass through the membrane and be collected in the permeate (Fan et al., 2020).An EMR offers great opportunities for enzyme hydrolysis on polysaccharides and to conduct simultaneous bioconversion and product separation (Popkov et al., 2023).
Due to the high efficiency of enzyme catalysis, the polysaccharides can be rapidly depolymerized into low Mw molecules and possibly be over-depolymerized into monosaccharide units.Such low Mw saccharides are less valuable compared to the intermediate-sized oligosaccharides, thus, controlling the enzymatic reaction is the key to ensuring high product quality.For example, dextranase can efficiently degrade the high Mw dextran from over 100 kDa to maltose (342 Da) (Larsson et al., 2003).The dextranase is a stable enzyme that retains high activity for days when being immobilized (Ding et al., 2021), thus the enzyme is commercially feasible for production of oligodextran.The use of an EMR can selectively permeate the oligodextran that has the most desired Mw (5-8 kDa) and avoid over-degradation of products (Su et al., 2018).On this basis, we immobilized dextranase on a membrane surface to better tailor the enzyme reaction over the membrane surface.Polydopamine (PDA) (Yang et al., 2018), tannic acid (TA) and 3-amino-propyltriethoxysilane (APTES) were used to assemble functionalized nanoparticles on the membrane surface for dextranase attachment (K.Jankowska et al., 2021;Dong et al., 2022).The catalytic behavior of the immobilized enzymes was studied.The hydrolysis pattern of the dextranase shifted from endo-hydrolysis to exo-hydrolysis when they were cross-linked on a PDA-modified membrane surface.Because of the static compaction among the cross-linked dextranase, the substrates could not easily diffuse into the active site of the immobilized enzymes thus causing undesired end-product accumulation in the EMR.However, when the non-cross-linked enzymes were immobilized on the TA/APTES assembled nanoparticles, that had a bigger size (200 nm) (Cheng et al., 2019), more active sides were exposed to the bulk solution and thus reduced the mass transfer limitation of the substrates.Consequently, the immobilized dextranase exhibited desired endo-hydrolysis with high enzyme activity and produced desired intermediate-sized oligodextran (Su et al., 2021).These results indicate that the configuration of the immobilized enzymes greatly impacts the catalytic behavior of the enzyme.Immobilized dextranase needs increased space to conduct efficient endo-hydrolysis to produce intermediate-sized oligodextran.The use of threedimensional (3D) scaffolds for enzyme immobilization stands as an ingenious approach, enabling the maximization of exposed active sites, thereby augmenting the overall activity of immobilized enzymes.Notably, 3D structured materials, including carbon nanotubes (CNTs) (Ji et al., 2016) and magnetic nanoparticles (Gebreyohannes et al., 2015), have been successfully employed for enzyme immobilization, leading to the achievement of remarkable catalytic efficiency.When integrated with a separation membrane, these Electro-Membrane Reactors (EMRs) manifest outstanding performance in both bioconversion processes and product separation.Furthermore, the incorporation of an additional layer within EMRs serves a dual purpose.Not only does it facilitate the prevention of direct contact between foulants and the separation membrane, thus curbing membrane fouling, but it also contributes to the overall enhanced performance of the system.
Electrospinning stands out as an efficient and cost-effective technology, lauded for its capacity to fabricate three-dimensional, open porous structures (Ahmed et al., 2015).It has found widespread application in the creation of matrices tailored for enzyme immobilization (K.Jankowska et al., 2021;Lu et al., 2019).A key advantage of electrospun fibers lies in their versatility, as they can be crafted from a diverse array of polymers.Moreover, the properties of the resulting fibrous matrix, including pore size and fiber thickness, can be finely tuned by adjusting polymer concentrations or manipulating electrospinning parameters such as voltage and injection rates.Consequently, electrospinning emerges as a highly promising technology for generating porous matrices replete with diverse functional groups, rendering them exceptionally well-suited for enzyme immobilization.The porous structure is of great benefit to retaining high enzyme activity (Z.-G.Wang et al., 2009), especially for enzymes like dextranase that need space for the hydrolysis reaction.Polystyrene (PS) is commonly used in electrospinning because it is an inexpensive commodity polymer that is used in different applications, such as packaging and filtration, and is easy to process.Martrou et al. produced surface-modified electrospun PS microfibers for the immobilization of horseradish peroxidase (HRP).They post-functionalized the pure PS microfibers by introducing amine groups to the fiber surface and successfully immobilized the HRP.The modified microfibers achieved high enzyme activity recovery and improved enzyme storability (Martrou et al., 2017).The coating with PDA and TA/APTES nanoparticles can improve the hydrophilicity of the materials surface (H.Zhang et al., 2018) and meanwhile introduce functional groups such as quinone and amine, that provide binding sites for the enzymes (Zhou et al., 2020;Markiton et al., 2017).Moreover, the strong adhesive property of PDA and TA allows the self-assembled nanoparticles to attach to a variety of substrate materials, thus they are considered outstanding modifiers to improve the enzyme loading efficiency on the electrospun fibers.
Jankowska et al. developed a novel EMR that combines the surfacemodified electrospun fiber for the immobilization of glucose oxidase (GO X ) and cholesterol oxidase (CO X ) (Katarzyna Jankowska et al., 2022).Such a biosystem compartmentalized different enzymes and endowed efficient cascade enzymatic reactions.In this study, we designed a novel EMR configuration that consists of surface-modified electrospun fibers and a commercial ultrafiltration (UF) membrane to immobilize dextranase.Such an EMR represents a novel configuration to achieve efficient enzyme immobilization and tailors enzymatic degradation on polysaccharides.Inspired by membrane fouling, the enzymes were filtrated through the multilayer membrane which caused enzyme accumulation near the membrane surface (Luo et al., 2013;Morthensen et al., 2017).The TA/APTES surface-modified fibers were fixed on the selective side of the membrane; therefore, the accumulated enzymes could bind to the functionalized fiber surface to form a catalytic layer above the UF membrane.Furthermore, because the fibrous material is above the membrane surface, the enzymes tended to attach to the fibers instead of directly adhering to the UF membrane, which causes severe pore blocking.
To our knowledge, dextranase was immobilized on a 3D matrix confined near a membrane surface for the first time.This novel design enabled high enzyme loading capacity, and simultaneously reduced the influence of the enzymes to the separation membrane.The modified 3D matrix is rich in quinone groups which work as the covalent binding sites for enzyme immobilization.At the same time, because the polystyrene materials are hydrophobic, the enzymes could also attach to the fibrous matrix via hydrophobic adsorption.The EMR therefore was able to retain both high enzyme activity and membrane selectivity.In our previous work, the cross-linked enzymes were immobilized directly on the membrane surface, which led to exo-hydrolysis and the accumulation of the undesired end-products (iso-maltose) (Su et al., 2021).In this work, the production of the iso-maltose could be minimized because the porous electrospun fibers offered a large enough space for the substrate to access the active sites of the immobilized dextranase, hence the enzymes conducted endo-hydrolysis and produced intermediate-sized oligodextran.At the same time, a separation membrane which works as a selective barrier permeates the target oligodextran and the desired intermediate-sized products can be collected on the permeate side.Different membrane materials were used in the development of the multiple-layered EMRs, which were subsequently used for the production of oligodextran.By investigating the process of oligodextran production, the interactions between enzyme reaction and membrane separation were explored.The influences of membrane materials were also discussed.The results in this work indicate that the use of multiplelayer membranes offers a smart strategy for the design of a highperformance EMR.
Z. Su et al.

Experimental procedures 2.2.1. Electrospun fiber modification
A dopamine or TA/APTES mixture was applied for surface modification of electrospun fibers that have the same area as the commercial membrane (13.4cm 2 ) with an approximate weight of 10 mg.For dopamine modification, pristine nanofibers were incubated with 10 mL of 2 g/L dopamine hydrochloride solution (pH 8.5, 10 mM Tris-HCl buffer) at 100 rpm and 25 • C for 4 h.
Electrospun fiber modification by TA/APTES was carried out according to our previous work: 2.4 mL 10 g/L APTES in EtOH solution then 17.6 mL of 2 g/L TA solution in Tris-HCl buffer (pH 8.5) was mixed to make a volume ratio of TA/APTES = 8:1 (Zhou et al., 2020).The electrospun fibers were then incubated in the TA/APTES coating solution at 100 rpm and room temperature (22 • C) for 18 h.The TA/APTES modification introduced a layer of nanospheres on the membrane surface that is rich in quinone groups for enzyme immobilization by covalent bonding.After modification, the membranes were cleaned by running distilled water to remove the residual modifiers and then the modified membranes were installed into a 50 mL Amicon stirred cell (Amicon UFSC05001, Merck Millipore, USA) for enzyme immobilization.

Fouling-induced mode.
The TA/APTES modified nanofiber layer was placed between a pristine commercial membrane (PES30 or RC10, on the bottom) and a porous support layer (on top) to construct the 3layer membrane.Then 10 mL 2 g/L dextranase solution was added into the Amicon cell and incubated with the 3-layer membrane and the modified fibers at 100 rpm for 1 h, after which the enzyme solution was filtrated through the membrane at a constant pressure (2 bar) until all the solution was filtrated through the membrane.The filtrated solution was collected for protein concentration measurements.Then 10 mL Milli Q water was added into the Amicon cell to wash the membrane at 2 bar.The washing procedure was carried out 3 times so in total around 30 mL of washing solution was collected.As a control, 10 mL 2 g/L dextranase solution was filtrated through a pristine PES30 membrane in the Amicon cell followed by the same washing procedure mentioned above.
The protein concentration of the enzyme solutions was measured by the Bradford assay: 0-16 μg/mL of bovine serum albumin (BSA) solutions were used for the calibration.Samples were diluted to be within the range of the protein calibration curve, as required.The enzyme solutions were mixed with Bradford reagent in a 1:1 volumetric ratio.After 5 min of incubation, absorbance was measured at 595 nm.Enzyme loading mass was calculated from the equation: where c is the soluble protein concentration and V is the volume of the solution at the corresponding concentration.Subscripts i, r, f and w represent initial, recovered, filtrated and washing solutions.The enzyme loading is defined below, where mi and ms are the mass of immobilized dextranase and soluble dextranase respectively: The configurations and immobilization methods of different EMRs are summarized in Table 1.

Activity test
To measure the observed activity of the immobilized enzymes, 20 mL 4 g/L DXT70K solution was added to the Amicon cell with the enzymatic membrane at room temperature and 100 rpm.Samples were collected at specified time intervals.The reducing sugar content of all the collected samples was measured by using 3, 5-dinitro salicylic (DNS) acid reagent, Specifically, 1 mL hydrolyzed samples were mixed with 1 mL DNS reagent and heated in a boiling water bath for 5 min.the samples were diluted 5 times by ultrapure water and measured at 540 nm.The enzyme activity was defined as the amount of iso-maltose (measured in μmol maltose) generated after 1 min dextranase catalysis at 25 •C, from DXT70K.The unit of the enzyme activity is μmol-isomaltose/min.

Oligodextran production
2.2.4.1.Free enzyme system.The PES30 membrane was compacted at 2 bar for 30 min with pure water.50 mL 20 g/L DXT70K solution and around 330 μg dextranase were mixed in the cell which was continuously fed with pure water under 2 bar.
2.2.4.2.Immobilized enzyme system.After the enzyme immobilization and washing process, 50 mL 20 g/L DXT70K solution was added to the Amicon cell which was fed by pure water at a constant pressure.

Productivity test
The phenol-sulfuric acid method was used for testing the concentration of the total saccharide in the permeate for the different enzymatic systems.Glucose solutions with concentrations between 0.01 and 0.07 g/L were used for calibration.Each 1 mL diluted sample was added to a 10 mL glass tube.0.5 mL of 6 % (v/v) phenol and 2.5 mL of sulfuric acid were mixed with the samples.Then the mixtures were shaken for 30 s and let stand for another 30 min.The sample absorbance was tested in proper cuvettes under 490 nm in a UV-2100 Spectrophotometer.Glucose solutions with concentrations between 0.01 and 0.07 g/L were used for calibration.

Water permeability of the membrane
The water permeability of the membrane was calculated as: Where V p is the permeate water volume (L) corresponding to a certain time: t (h) during the testing, A m is the effective filtration area (m 2 ), and TMP is the transmembrane pressure (bar).

Characterization
Gel permeation chromatography (GPC) was used to test the average Mw of oligodextran generated in the different reaction systems.50 μL of each sample was eluted under 1 mL/min in ultrapure water at 40 • C. A refractive index detector coupled with the G4000PWXL column from Shimadzu was used for testing the samples.Scanning electron microscopy (SEM) was used to visualize the morphology of PDA-modified PES membranes with immobilized enzymes.Here, samples with a gold coating (Balzers PV205P, Switzerland) were investigated using an EVO40 microscope (Zeiss, Germany).
The fiber diameters were measured on the SEM image using Image J (National Institute of Health, USA).The scale of the SEM image (10 μm) was used as a standard and 15 random points of the fibers were selected in the image.The average diameter of the fibers was calculated from the 15 random points.

EMR with free and immobilized dextranase
Before constructing the 3-layer EMRs, two control experiments were conducted.Enzyme immobilization in the fouling-induced mode was conducted on a pristine PES30 membrane, resulting in immobilization of 396.6 ± 3.5 μg of dextranase, corresponding to 64.2 % of the used dextranase (Fig. 1A).The immobilized dextranase was likely linked to the membrane by hydrophobic interactions.Although the MWCO of the membrane was 30 kDa, which was half of the Mw of the enzymes (67 kDa), 16.8 % of the enzymes could still pass through the membrane.In addition, a control experiment using free enzymes was conducted to evaluate the performance of the enzyme-immobilized EMRs.In the free enzyme experiment, 330 μg of free dextranase (in terms of soluble proteins) were directly mixed with 50 mL 20 g/L DXT70K substrates and was filtrated at 2 bar with continuous water feeding.
Fig. 2 shows the oligodextran production in the two control experiments, i.e. with free and immobilized dextranase, in terms of the average Mw change in permeates, permeate flux and productivity.In general, the average product Mw in the permeate is a function of the mass transfer, the hydrolysis rate of the enzymes, and the membrane fouling (Su et al., 2018).In the EMR with free enzymes and PES30 membrane, high Mw oligodextran was obtained at the beginning of production (>20 000 Da at 10 min) while the product Mw declined with time (Fig. 2A).In this type of reactor, the permeate flux increased at the beginning (Fig. 2B) because the dextran substrates were depolymerized and led to reduced filtration resistance.When the reaction time was prolonged, more and more oligodextran was generated and accumulated near the membrane surface.Along with the enzymes, the oligodextran may block the membrane pores and thus lead to permeate decline after 60 min.Regarding the Mw change, in the initial phase of the production, oligodextrans with larger molecular sizes were removed from the reactor before sufficient hydrolysis occurred.However, when the filtration continued, the free enzymes had more time to convert the long-chain dextrans to oligodextrans and thus yielded a lower product Mw in the permeate (Su et al., 2018).Meanwhile, the increase of permeate flux allowed a large amount of solutes to pass through the membrane thus most of the saccharides were permeated through the membrane within the first 60 min (Fig. 2D).With increasing time, the strong convective flux transported the enzymes and saccharides towards the membrane surface and caused membrane fouling (pore blocking and narrowing) thus increasing the rejection of oligodextran products.The result is lower saccharide concentration obtained in the permeate.
In the EMR with a dextranase-loaded PES30 membrane, the immobilized enzymes partially blocked the membrane pores and narrowed down the pore size (F.Wang and Tarabara, 2008).The membrane is therefore expected to reject oligodextrans of higher Mw.Thus, at the beginning of the production, the oligodextrans in the permeate had lower Mw than in the free enzyme system.As the filtration continued, the immobilized dextranase had a longer time to hydrolyze the substrates (although the immobilized enzymes exhibited lower activity on the membrane surface due to the shield of their exposed active sites (J.Wang et al., 2020)), and thus lower product Mw was obtained in the permeate.A similar phenomenon was reported in our previous study, where the fouling-induced enzyme immobilization adjusted the membrane pores sizes and improved the uniformity of low Mw oligodextran production (Su et al., 2018).At the same time, however, enzyme immobilization introduced serious membrane fouling that limited the transmission of oligodextran during the production and therefore the saccharide concentrations were low in the permeate (Fig. 2C).
However, the immobilized enzymes experienced a reduction in activity since the active sites were likely shielded by the planar structure of the membrane surface.As a result, large dextran molecules had limited mass transfer to reach the active sites of the immobilized enzymes, leading to high Mw of the product in the permeate.To overcome these challenges, the enzymes should have been immobilized onto a porous structure positioned just above the membrane surface.This arrangement would provide ample reaction space (as depicted in Fig. 1B), enabling efficient degradation of the polysaccharides (Zhang et al., 2010).

Enzyme immobilization by incubation mode on the electrospun fibers
An electrospun fibrous matrix was employed for the immobilization of dextranase, as depicted in Fig. 3A.The average diameter of the fibers was approximately 2 µm, as shown in Fig. 3B.The porous structure of the fibrous matrix serves a dual purpose.Firstly, it offers a substantial surface area for the enzymes to adhere to.Secondly, it mitigates the mass transfer limitations encountered by the substrates when approaching the immobilized enzymes.Consequently, the larger dextran molecules can diffuse more readily towards the active sites of the cross-linked dextranase located on the fibers.Simultaneously, the hydrolyzed products can freely diffuse back into the bulk solution.The cross-linked and non-cross-linked dextranase were immobilized on the PDA or TA/ APTES-coated electrospun fibers.The activity of the immobilized enzymes was measured by adding 4 g/L DXT70K to the dextranaseimmobilized matrixes.In consequence, the intermediate-sized oligodextrans were generated over time, and no immediate end-products accumulation took place when using the electrospun fibers to immobilize dextranase (Fig. 3D).The result means that the cross-linked enzymes exhibit an endo-hydrolysis pattern when immobilized onto electrospun fibers.In our previous work, the GA formed covalent bonds between the enzymes and the PDA coating layer on the membrane and simultaneously functioned as the cross-linkers to form cross-linked enzyme aggregates (CLEAs).These CLEAs exerted exo-hydrolysis behavior and the process simultaneously suffered from the mass transfer limitations.Thus the cross-linked enzymes tended to produce end-products (single units of isomaltose) during the reaction (Su et al., 2021).In this study, the CLEA clusters on the PDA-modified fibers were observed in the SEM image and the FTIR spectra (Fig. S1).The Mw results in this work indicate that the electrospun fibers can help to maintain the desired endo-hydrolysis pattern of the CLEAS, making the fibrous matrix a promising scaffold of enzyme immobilization for the production of intermediate-size oligodextran.
The activity of the non-cross-linked enzymes was also evaluated in this work.On the TA/APTES modified fibers, dextranase was immobilized without the cross-linker (GA) and that resulted in a decrease of the Mw of dextran produced (Fig. 3C).The decrease in Mw is likely because the TA/APTES modified membrane offered abundant binding sites (quinone groups) for dextranase immobilization, and since the noncross-linked enzymes have more active sites exposed to the substrates as compared to the cross-linked enzymes, where many of the active sites are embedded deep within the CLEAs and thereby, not available to the substrates.Thus, a faster Mw decline of the dextran was observed with non-cross-linked dextranase than with cross-linked dextranase.Hence the TA/APTES modification was used for further study.

Enzyme immobilization by fouling-induced mode on a 3-layer EMR
The fouling-induced enzyme immobilization can significantly improve the enzyme loading on the modified surface (Su et al., 2021).This enzyme immobilization strategy utilizes pressure to increase the enzyme concentration near the membrane surface.Here, the modified fibrous matrix was applied on a UF membrane that had a smaller pore size (10 or 30 kDa) than the enzymes (67 kDa).A porous support layer was mounted on top of the fibrous matrix thus the fibers can be used under a higher shear rate, avoiding the unordered motion of the enzymeloaded fibers.The enzyme solution was filtrated through the 3-layer membrane structure to perform the enzyme immobilization.The configuration of the 3-layer EMR is shown in Fig. 4.
The change of chemistry on the TA/APTES coated fibers was observed in the FITR spectrum (Fig. 5A).Compared with the pristine PS fibers, the absorption peak at 1706 cm − 1 is related to quinone groups of (C=O bond) on the oxidized TA molecules; the occurrence of a wide peak at around 3300 cm − 1 indicates that -OH groups on TA molecules were successfully coated on the fibers.Moreover, the bands of N-H and C-N were observed at 1600 cm − 1 and 1200 cm − 1 , which confirms the coating of APTES (Z.Wang et al., 2019).In the 3-layer EMR, the separation membrane (PES30 or RC10) can retain around 50 % of the enzymes.Due to the high enzyme concentration near the electrospun fibers, 332.1 μg of dextranase were immobilized in the PES30-based EMR and 334.5 μg of dextranase was immobilized in the RC10-based EMR (Fig. 5B).The immobilized enzymes on the fibers had high activity, especially at the beginning of the reactionthey produced 11.3 μmol-isomaltose per minute.The production rate decreased over time due to decreased substrate concentration.After a 4 h reaction, the immobilized enzymes can produce 0.72 μmol-isomaltose per minute (Fig. S3).As a comparison, the enzymes immobilized in an incubation mode on the fibers only produced around 0.075 μmol-isomaltose per minute (Fig. S2).The results indicate that the fouling-induced enzyme immobilization significantly improved the immobilized activity on the fibers.On the images of CLSM (Fig. 5C), fluorescence was observed on the electrospun fibers after TA/APTES modification and dextranase immobilization.These results further confirm the success of the modification and enzyme immobilization.
The 3-layer EMRs produced short-chain oligodextrans with stable Mw.This result was due to the high enzyme activity near the membrane surface and stable permeate flux throughout the reaction-separation process (Fig. 6B).Because the shear-induced diffusion around the electrospun fibers was low, the filtration resistance caused by the concentration polarization (CP) layer in the 3-layer EMR was high and thus resulted in low permeate flux.It is noteworthy that as the evolution of CP is negligible, the permeate fluxes were relatively constant over the whole production period.Meanwhile, on the membrane surface, the high saccharide concentration in the CP layer promoted the transfer of oligodextran through the membrane, leading to high concentrations of oligodextran in permeates.At the beginning (the first 60 min) of the production, the oligodextran concentration in the permeates fluctuated especially for the RC membrane.It was because the production was conducted in a diafiltration mode that the Mw and substrate concentration changed quickly at the beginning.Such changes caused a fluctuation of the product concentration in the permeate.When the production was prolonged, the Mw change and concentration change of the substrates slowed down, thus the concentration of the oligodextran in the permeates became more stable after 120 min (Fig. 6C).Productivity was calculated by multiplying the permeate volume and the concentration of saccharides.Although the permeate flux was low, high saccharide concentrations over the production period allowed the 3layer EMRs to achieve constant productivities over the 4 h production period Fig. 6D.In contrast, the single-layer EMRs had high productivity for the first 60 min but only small amounts of saccharides were obtained with prolonged production time (Fig. 2D).It is noteworthy that the enzyme amount loaded on the 3-layer EMR was around 330 μg which was 16 % lower than the single-layer PES 30 membrane (396 μg).
However, the 3-layer EMR with the PES30 membrane finally produced 341.3 mg oligodextran, which was 19.6 % higher than the products obtained in the single-layer dextranase-loaded PES30 EMR (285.3 mg).The results imply that the enzymes retained higher activity when they were immobilized on the 3-D matrix structure, which led to higher productivity of oligodextran.
The membrane materials used in the 3-layer EMRs influenced the Mw of the oligodextran products due to the different interplay between oligodextran and the membranes.Lower product Mw was observed in the PES30-based EMR than in the RC10-based EMR (Fig. 5A) during the first 60 min of production, despite PES30 having bigger membrane pores than the RC10 membrane.It was because the dextranase on the 3-D matrix over the PES30 membrane retained high enzyme activity and therefore could efficiently hydrolyze the large dextrans to low Mw products before the large molecules migrated into the membrane pores, hence the product Mw in permeate was low.With increasing reaction time (after 120 min), more and more saccharides were permeated out from the reactor and fewer saccharides were accumulated over the membrane surface, which led to lower filtration resistance of the large dextran.Consequently, larger dextran molecules could pass through the membrane and were obtained in the permeate, meanwhile, increased permeate flux was observed at 120--240 min (Fig. 5B).In comparison, due to the better hydrophilicity of the RC membrane, although pore size was lower than the PES membrane, the hydrophilic linear-shape oligodextran could easily pass through the RC10 membrane and the product average Mw was higher than that obtained in the PES membrane reactor.The results show that membrane hydrophilicity and thickness have just as much influence on product separation as the pore size of the membranes.The good production performance of the 3-layer EMR indicates a promising strategy for a stable reaction and separation process.
Water permeability indicates the filtration resistance in the production of oligodextran, which influences the separation efficiency of the desired molecules.Fig. 7A shows the least water permeability loss in the 3-layer EMRs after the enzyme immobilization, which indicates the novel EMR configuration provided the best performance.It was because the TA/APTES-modified electrospun fibers reduced the enzyme blockage in the PES30 membrane and hence the membrane maintained a good separation selectivity for the substrate and low Mw products during the reaction-separation process.Moreover, a big portion of enzymes was immobilized on the porous electrospun fibers and the substrates had more free access to the enzyme active sites, thus resulting in rapid dextran hydrolysis.The short-chain oligodextran with lower Mw   produced caused less membrane fouling in an EMR with a PES membrane (Su et al., 2018), therefore, the PES30 membrane has higher water permeability after the production.
In our previous research, dextranase was directly immobilized on the TA/APTES-modified PES membrane.The TA/APTES surface modification caused around 97 % of the water permeability drop and the membrane maintained only 1.7 ± 0.4 LMH/bar after the fouling-induced enzyme immobilization.Although the activity of the immobilized dextranase was high, because of the high filtration resistance, it is difficult to use such an EMR to purify the oligodextran products (Su et al., 2021).The coupling of TA/APTES-modified electrospun fibers in this study, however significantly improved the immobilized enzyme activity on the membrane surface, meanwhile, it also mitigated the membrane fouling caused by enzymes and dextran molecules.This novel design allowed a simultaneous reaction-separation of the oligosaccharides.Nevertheless, the water permeability still decreased from 369.1 LMH/bar to 124.6 LMH/bar (Fig. 7A) after the saccharides production.It was due to unavoidable pore blocking caused by free enzymes, with enzyme fouling happening during the immobilization.Due to the intrinsic hydrophobicity of the PES materials, enzymes could deposit on the membrane surface via hydrophobic interactions.In addition, enzymes that have smaller sizes might enter the membrane pores and cause pore blocking of the membrane.Moreover, in the production of oligodextran, part of the immobilized enzymes leaked from the electrospun fibers and promoted membrane fouling.Therefore, only 33.8 % of the water permeability was recovered after oligodextran production.
The cellulose membrane (RC10) showed an impressive anti-fouling capacity after the enzyme immobilization and saccharides production.The membrane retained the same water permeability after usage, which indicates that no irreversible fouling happened throughout the process.This observation is further confirmed by the SEM images, which show that the RC10 membrane had a flat and smooth surface after the foulinginduced enzyme immobilization.Its surface morphology barely changed compared to the pristine membrane.Meanwhile, after the foulinginduced immobilization, the immobilized enzymes (scattered white spots in Fig. 7B) were observed on the surface of the electrospun fibers.These results are likely due to the good hydrophilicity of the cellulose material and the smaller pore size (10 kDa) of the membrane.The water molecule layer on the hydrophilic membrane can prevent the enzyme attachment on the membrane surface and the membrane pores (10 kDa) are much smaller than the dextranase enzyme molecules (reported as 67 kDa (Larsson et al., 2003)), thus the enzymes were not able to enter into the membrane pores and therewith causing severe membrane pore blockage.The results in the study indicate that the 3-layer structure helps mitigate the membrane fouling caused by enzymes and polysaccharides.Therefore, the 3-layer membrane can reduce the cost of membrane cleaning in the real production process.
In summary, the 3-layer EMR outperforms the other EMR configurations in terms of immobilized enzyme activity, Mw control of the oligodextran, productivity and antifouling ability (Table S2).

Conclusion
The results in this study indicate that the 3-layer structure with a nanofiber matrix in the middle is an excellent configuration for an EMR.The TA/APTES coating offered a plentiful supply of binding sites (i.e.quinone groups on the TA-APTES modified surface (J.Zhang et al., 2021)) for the enzymes, while the porous structure of the electrospun fibers provided ample reaction space.As a result, the immobilized enzymes retained their endo-hydrolysis patterns and maintained a high level of activity.The dextranase loaded via the fouling-induced method exhibited more than 10 times higher catalytic activity than the incubation-immobilized dextranase.Hence, the 3-layer EMR rapidly produced short-chain oligodextran even at the beginning of the process.Moreover, the electrospun fibers mitigated the fouling caused by the enzymes and the saccharides, thus the membrane maintained a high selectivity along the production period.
This study also investigated the effect of membrane materials in a 3layer EMR.The hydrophilic RC10 membrane is promising for the development of a 3-layer EMR because of its excellent separation selectivity towards the target oligodextran.Negligible membrane fouling was observed on the RC10 membrane while the PES30 membrane lost 66 % of water permeability after the oligodextran production.Such EMR design offers a great opportunity to tune the reactionseparation process and can be applied in large-scale production of low Mw oligosaccharides production.The economic viability of the proposed EMR system is underpinned by several key factors.Firstly, the cost-effectiveness of the electrospinning technology, coupled with the affordability of coating materials, paves the way for scalability, making it commercially feasible on a larger scale.Additionally, the 3-layer structure exhibits robust antifouling properties, further bolstering its commercial appeal.The strategic configuration of the 3-layer EMR offers a dual advantage.It effectively retains enzymes and substrates with substantial molecular dimensions, while concurrently permitting the passage of smaller molecules through the membrane.This feature enhances its versatility and utility in various applications.Moreover, the EMR system holds promise for the production of peptides.Here, proteolytic enzymes can facilitate the hydrolysis of proteins within the reactor.The resulting peptides, characterized by smaller molecular sizes, can readily traverse the membrane and be efficiently collected in the permeates, further expanding the system's potential applications.
In a large-scale EMR, the electrospun fibers can be used as spacers in the membrane module and thereby provide improved hydrodynamics on the membrane surface along with its function as enzyme

Fig. 1 .
Fig. 1. (A) Dextranase distribution after immobilization on a pristine PES30 membrane; (B) Illustration of oligodextran production in the EMRs with immobilized or free dextranase.

Fig. 3 .
Fig. 3. (A) SEM image of the PS electrospun fibers after dextranase immobilization (by incubation mode); (B) Average fiber diameter and diameter distribution of the PS electrospun fibers.(C) Dextran Mw variation in a PDA and TA/APTES modified PS electrospun fibers obtained under incubation mode; (D) GPC chromatograms of oligodextran products produced by the catalytic electrospun fibers with cross-linked dextranase;

Fig. 4 .
Fig. 4. Configuration of the 3-layer EMR, including the process of fouling-induced enzyme immobilization and the production of oligodextran.Z.Su et al.

Fig. 5 .
Fig. 5. (A) FTIR spectra of pristine and modified electrospun fibers; (B) Dextranase distribution on the 3-layer EMRs; (C) CLSM images of the electrospun fibers before and after modification and enzyme immobilization.
Z.Su et al.

Table 1
Configurations and immobilization methods of different EMRs.