The effect of transmembrane pressure and feed flow rate on transgalactosylation of lactose in an enzymatic membrane reactor

Abstract

Lactose hydrolysis and transgalactosylation catalyzed by β-galactosidase are important for the production of galacto-oligosaccharides (GOS). These reactions were performed in an enzymatic membrane reactor (EMR) with β-galactosidase immobilized on the inner lumen of the membrane. The effects of hydrodynamic parameters of transmembrane pressure (TMP) and nominal feed flow rate on the reaction were investigated. The rate of GOS formation increased as TMP was increased from 0.2 bar to 0.5 bar and then decreased as the TMP was further increased to 0.8 bar. In contrast to that of GOS, the formation rate of monosaccharides significantly decreased at high nominal feed flow rates. The separation factor between GOS and monosaccharides increased at high TMP, but the overall permeation flux decreased because of the high membrane resistant coefficient of membranes with immobilized β-galactosidase. Moreover, the specific productivity of trisaccharides was higher, whereas that of tetrasaccharides was lower in the EMR than that in batch reactor systems.

---

1. Introduction

An enzymatic membrane reactor (EMR) is a biocatalytic reactor used to produce fine chemicals. EMRs have been used to produce numerous types of products, such as optically pure drugs,¹ peptides,² esters,³ and oligosaccharides.⁴ An EMR contains two compartments separated by a permeable-selective membrane. The membrane also provides a structural support for the enzymes, that is, an enzyme can be loaded within the porous structure or on the surface of the membrane. This feature of EMRs has been manipulated for various chemical reactions, particularly for enzyme-catalyzed hydrolysis and esterification.^1,3,4

Membranes are usually utilized to enhance the performance of enzymatic reactions. For instance, EMR is used for β-galactosidase-catalyzed production of a prebiotic compound of galacto-oligosaccharides (GOS), with whey or lactose as substrate.^4–6 Lactose is converted by β-galactosidase (β-gal) through two-step reactions of hydrolysis and transgalactosylation. The reaction generates GOS as well as galactose and glucose as by-products. EMR can also be utilized to produce GOS through several techniques. A common method used is the initiation of the reaction in the retentate side of the membrane with free or immobilized enzymes; separation is then conducted to recover oligosaccharides, glucose, and galactose.^7–14 Enzyme can be immobilized on the surface or in the porous matrices of the membrane through adsorption, covalent bonding, or electrostatic interactions.^4,5,15,16 GOS production was higher in membrane reactors than that in conventional batch reactors, regardless of free or immobilized enzyme employed in the system.

Das et al. investigated GOS production using whey permeate in a recycle membrane reactor with free β-gal from Bacillus circulans.¹⁴ In this study, β-gal was retained with 5 kg mol⁻¹ polyethersulfone membrane, and the concentration of oligosaccharides was doubled compared with that in the batch reactor system. Moreover, the formation rate of monosaccharides was significantly lower in the membrane reactor system than that in the batch reactor system. Güleç et al. investigated lactose conversion into GOS in plasma-modified cellulose acetate membrane with immobilized β-gal from Aspergillus oryzae.⁶ A multi-layered catalytic layer was formed on the membrane surface because of the aggregation of β-gal with polyethyleneimine. A similar result was obtained when the GOS formation rate was increased from 21% to 27%. Galactose inhibits most types of β-gals (Aspergillus, Kluyveromyces, Bacillus, and etc.);^17–19 in some cases, glucose is also found to be a significant inhibitor of these enzymes.^20,21 Thus, the formation rate of GOS should be increased because of in situ removal of monosaccharides.^6,14 However, the formation of GOS is also affected by other factors, such as residence time and concentration of galactosyl acceptors in the feed or retentate side of the membrane.

A short residence time is appropriate when using free enzymes because the reaction is not significantly affected by diffusion of the substrate to the catalytic site.⁸ In these membrane systems, high permeation of the products is also required to minimize the hydrolysis side reaction of GOS and the inhibition effect of monosaccharides. Moreover, the depletion of the galactosyl acceptor (lactose) as a result of permeation and hydrolysis can be controlled with fresh feeding into the retentate side of the membrane. By contrast, a long residence time is required for systems with immobilized enzymes because of diffusion limitation. This finding could be due to immobility of the enzyme on the membrane surface, and the effect of diffusion of the substrate from the bulk solution to the catalytic site on the formation of transition state. The mass transport of the substrate is driven by chemical potential or pressure gradient between two compartments. As a result, a long reaction time is required to obtain high GOS yields.⁶ This finding is in agreement with those reported by Sen et al.; in this study, prolonged retention of lactose on the retentate side improves total substrate conversion.⁴ In this case, the performance of the reaction is influenced by hydrodynamic parameters such as transmembrane pressure (TMP) and feed flow rate. In the membrane field, these parameters are greatly affected by the mass transfer behavior of solutes from the bulk solution in permeate as it passed through the membrane.

Generally, the mass transport of solute through the membrane is increased with increase in TMP and feed flow rate. Thus, it is sensible to assume that the enzymatic reaction improves with enhanced substrate diffusion toward the catalytic site on the membrane surface. However, this is not only the case as the mass transport of solute also could decrease at relatively high TMP and feed flow rate. The concentration polarization layer forms on the membrane surface when the TMP is increased. In this event, the concentration of solutes gradually increases toward the membrane surface due to diffusive back transport. This phenomenon can leads to the formation of gel layer when the polarization layer is compacted at higher TMP. As a consequence, the mass transport of solute is decreased which in turn affects the enzymatic reaction on a membrane systems. In addition, the enzymatic reaction also could decrease with higher feed flow rate, whereby the solute is easily swept away from the catalytic membrane surface. To date, less attention has been given on the significance of these phenomenon attributed by TMP and feed flow rate on the enzymatic reaction in membrane reactor systems.

In this work, lactose conversion catalyzed by immobilized β-galactosidase was performed in an enzymatic hollow fiber membrane reactor. The enzyme was immobilized on the inner lumen of the membrane surface through simultaneous adsorption and fixation by polyethyleneimine and glutaraldehyde, respectively. This study aims to investigate the effect of hydrodynamic parameters of transmembrane pressure (TMP) and nominal feed flow rate on hydrolysis and transgalactosylation reactions. The investigation was carried out by evaluating solutes separation for the first time in addition to the formation of the products.

2. Materials and methods

2.1 Materials

Polysulfone hollow fiber membrane (10 kDa) was purchased from Pall (United States). A membrane with hollow-fiber configuration was used because of its high surface area to volume ratio to increase the total effective catalytic surface and provide maximum permeation flux. Moreover, 10 kDa polysulfone membrane was used to achieve high protein rejection and durability at a typical operating temperature of 55 °C for β-gal-catalyzed conversion of lactose. β-Gal originated from Aspergillus oryzae with minimum unit activity of 8 IU mg⁻¹ and was obtained from Sigma-Aldrich (United States). Lactose, glucose, and galactose (>99%) were obtained from Merck (Germany). GOS such as trisaccharides (globotriose) and tetrasaccharides (galactotetraose) were acquired from Sigma-Aldrich. Polymers such as polyethyleneimine (50% w/v H₂O) and glutaraldehyde (25% v/v H₂O) were obtained from Merck and used for immobilization. The other chemicals used in the study were also supplied by Merck.

2.2 Immobilization procedure

The EMR was prepared by immobilization of β-gal on the inner surface of the polysulfone hollow-fiber membrane.²² Immobilization was conducted at room temperature (28 °C) and atmospheric pressure. First, 5% w/v polyethyleneimine solution was prepared in deionized water. Moreover, β-gal–glutaraldehyde solution (5 mg mL⁻¹ β-gal, 7.5% v/v glutaraldehyde) was prepared by mixing an equal volume of 10 mg mL⁻¹ β-gal and 15% v/v glutaraldehyde solutions in deionized water. The solutions were sonicated for at least 2 minutes to create a homogeneous solution. The polyethyleneimine solution (5% w/v) was then circulated on the lumen side of the membrane by peristaltic pump at a flow rate 0.6 mL min⁻¹ for 30 min. The polyethyleneimine solution was collected from the membrane, and the lumen side of the membrane was dried by passing through air from an air pump for 10 min. Afterward, the coated membrane was circulated with β-gal–glutaraldehyde solution (5 mg mL⁻¹ β-gal, 7.5% v/v glutaraldehyde) at 0.6 mL min⁻¹ for 6 h. The β-gal–glutaraldehyde solution was then removed from the membrane. The membrane immobilized with β-gal was flushed with single pass of deionized water at 1 mL min⁻¹ for 30 min. The chemical composition change on the surface of the membrane was analyzed using Energy Dispersive X-ray (EDX) by Quanta FEG 450, FEI (United States). The membrane module was stored at 5 °C until further use for a reaction.

2.3 Determination of enzyme activity

The immobilized β-gal activity was determined by using colorimetric assay as described in former publication.²² The principle of the test procedure was based on the hydrolysis of o-nitrophenyl β-D-galactopyranoside (ONPG) to o-nitrophenol (ONP) and β-D-galactose. The reaction was carried out at 37 °C in 20 mM citrate–phosphate buffer solution at pH 5.5. In detail, 0.4 mL of 20 mM citrate–phosphate buffer solution (pH 5.5) was prepared in 2 mL tube. Then, 0.5 mL of 10 mM ONPG substrate and 0.1 mL of free β-gal in citrate–phosphate buffer solution (pH 5.5, 20 mM) were added into the tube. To determine the immobilized β-gal activity, the mixture was added with 1 cm hollow fiber membrane (immobilized with β-gal) cut into several pieces. Then, 0.1 mL of 20 mM citrate–phosphate buffer solution at pH 5.5 was added to a final volume of 1 mL. The resulting mixture was incubated in a water bath shaker at 37 °C for 5 min. The reaction was terminated by deactivation of β-gal with the addition of 1 mL of 200 mM sodium carbonate. The absorbance of the solution was measured at 410 nm by spectrophotometer, Carry UV 60 (Agilent Technologies, United States).

2.4 Transgalactosylation in EMR

β-Gal-catalyzed transgalactosylation of lactose was performed in the EMR rig (Fig. 1). The feed tanks were filled with substrate solution (T-2) and deionized water (T-1), then placed in an incubator and heated to the desired operating temperature of 55 °C. The solution in both tanks was agitated by a four-blade stirrer at 100 rpm. Afterward, the hollow fiber membrane immobilized with β-gal (R-1) was installed into the incubator (E-2). Deionized water from T-1 tank circulated into the shell side of the membrane module by a peristaltic pump (P-1) with a flow rate of 5 mL min⁻¹. The lumen side of membrane immobilized with β-gal was circulated with substrate solution by a peristaltic pump (P-2) with flow rate between 1 and 8 mL min⁻¹. The trans-membrane pressure between the lumen and shell side of the membrane was regulated at the desired pressure (0.2–0.8 bar) by manipulating the pressure regulator valve (V-8) and vacuum pump (P-3). The reaction was carried out for 24 h. Briefly, 1 mL of the sample was collected periodically from the lumen and shell sides of the membrane to examine the composition of saccharides. Separation factor (S_f,i) was determined during the reaction to determine the separation efficiency of the saccharide solution, which consists of glucose, galactose, lactose, and GOS. S_f,i of the product is calculated by eqn (1), which is normalized to the substrate concentration.where C_p,i and C_r,i are the permeate and retentate concentrations of the products, respectively. C_p,s and C_r,s are the permeate and retentate concentrations of the substrate, respectively. S_f,i can be expressed in the form of rejection coefficient as follows:withwhere R_p,i and R_s are the rejection coefficients of the products and substrate, respectively.

Fig. 1 Instrumental diagram for β-galactosidase-catalyzed lactose conversion via enzymatic hollow-fiber membrane reactor.

2.5 Sample analysis

The sample was directly heated in an oven at 90 °C for 10 min to deactivate any possible β-gal activity and terminate the reaction. The sample was filtrated through a 0.22 μm nylon membrane filter to remove any particulates. The composition of saccharides in the sample was analyzed by high performance liquid chromatography (HPLC) (Shimadzu, Japan). The composition of saccharides in the sample was fractionated by RCM monosaccharide 8% Ca²⁺ column (Phenomenex, United States). The column was incubated in an oven at 80 °C. The sample was then eluted with degassed reverse osmosis water under a feed flow rate of 1 mL min⁻¹. Saccharides were detected using a refractive index detector (RID) and cell temperature was maintained at 40 °C. The saccharides; galactose, glucose, lactose, trisaccharide and tetrasaccharide were eluted at 14.6 min, 13.2 min, 11.6 min, 10.5 min and 9.8 min, respectively. The chromatogram of the HPLC analysis is shown in Fig. 2.

Fig. 2 The chromatogram of saccharides analyzed by high performance liquid chromatography. The galactose, glucose, lactose, trisaccharide and tetrasaccharide were eluted at 14.6 min, 13.2 min, 11.6 min, 10.5 min and 9.8 min, respectively.

3. Results and discussion

3.1 Characterization of the immobilized β-gal

The β-gal was immobilized on the inner surface of polysulfone hollow-fiber membrane coated with polyethyleneimine as a polyelectrolyte layer. The immobilization of the β-gal was carried out in a continuous process with a constant feed flow rate of β-gal–glutaraldehyde solution. The adsorption and fixation of the β-gal occurred simultaneously during the immobilization process as described in details in previous investigation.²² The immobilized-β-gal was confirmed by determining chemical composition on the membrane surface and the result is tabulated in Table 1. The nitrogen atom is detected on sample S2 when the membrane surface was coated with polyethyleneimine. This result is expected since polyethyleneimine consists of significant amount of nitrogen atom. The presence of immobilized β-gal on the polyethyleneimine coated membrane (sample S3) was confirmed when the composition of oxygen atom is increased. This is due to the fact that β-gal is a protein which contains oxygen, nitrogen and carbon atoms.

Table 1 Energy dispersive X-ray analysis on (S1) untreated polysulfone membrane, (S2) PEI layered membrane, and (S3) β-gal immobilized membrane²²

Elements	Elements composition (%)
	S1	S2	S3
Carbon	71.21 ± 2.31	71.82 ± 2.63	66.37 ± 3.42
Nitrogen	0.00 ± 0.00	10.05 ± 0.93	7.68 ± 1.16
Oxygen	35.17 ± 1.42	16.50 ± 1.26	25.73 ± 2.83
Sulphur	0.93 ± 0.29	1.63 ± 0.34	0.22 ± 0.18

---

The activity of the immobilized β-gal was assayed with ortho-nitrophenyl-β-galactoside as a substrate with activity yield of 11.31 ± 0.60 IU. It was found that about 60% of unit activity was left in the β-gal–glutaraldehyde solution after 6 h of immobilization process. Further increase in activity yield is possible by increasing the total surface area of the membrane. However, maximum surface area has been reached for a bench scale hollow fiber membrane reactor used in the present study with total inner surface area of 15 cm². The β-gal activity on the membrane was analyzed for 14 days as shows in Fig. 3. It was found that the β-gal activity slowly decreased for 10 days and then become constant for the next following days. The immobilized β-gal was able to maintain 70% of its initial activity after 14 days.

Fig. 3 The activity of immobilized β-gal in 14 days.

3.2 Effect of catalytic layer on membrane performance

Polysulfone membrane with a nominal pore size of 10 kDa was used in the study. This membrane provides relatively high flux at low operating pressures for liquid phase separation. The chemical characteristic of monosaccharides and GOS are rather similar, and the fractionation of such mixture solely depends on molecular size. Hence, a chemically inert membrane is preferred to extract saccharide (monosaccharides) compounds of interest from the reaction mixture through molecular entrainment. In the present study, the membrane surface was prepared with a catalytic layer. The presence of the catalytic layer is predicted to change the micro-characteristics of the membrane for separation; these characteristics include permeation flux and membrane resistance coefficient. The implication of the catalytic layer on the water permeation flux and total membrane resistance is depicted in Fig. 4. As predicted, conversion of the membrane to a catalytic membrane decreases water permeation flux and increases total membrane resistance. Previous studies reported that the membrane resistance of polyethersulfone, polyimide, and cellulose triacetate membranes increases after β-gal immobilization.⁴

Fig. 4 Membrane performance before and after the preparation of catalytic layer. Effects of TMP on (a) permeation flux and (b) membrane resistant coefficient. Effects of feed flow rate on (c) permeation flux and (d) membrane resistant coefficient.

Permeation flux is linearly dependent on TMP for both membranes, as shown in Fig. 4(a). The permeation drag increases as TMP increases. In certain cases, the diffusive back transport may dominate, and permeation flux is reduced. Diffusive back transport is due to Brownian diffusion, shear-induced diffusion, and interaction-induced migration (attributed to electro-kinetic and surface forces).^23,24 However, the diffusive back transport of the solute is minimal at 0.4–0.8 bar of TMP because the permeation flux does not reach the plateau. This finding suggests that the pore channel of the membrane remains intact after the preparation of the catalytic layer to permit the diffusion of the solute. If the diffusive back transport from the surface of the membrane to the bulk solution dominates the system, the permeation flux will reduce and concentration polarization may occur. Subsequently, separation efficiency is affected, particularly when a gel layer is formed. This phenomenon normally occurs when the nominal size of the pore channel is smaller than the molecular cut-off of the solutes. Moreover, the total membrane resistance is slightly reduced as TMP increases from 0.4 bar to 0.8 bar [Fig. 4(b)]. This result implies that the diffusion of water (solvent) through the membrane pore channel improves with increasing pressure in the feed channel. The most distinct effect between the control and membrane with immobilized β-gal is the response of permeation rate against TMP [Fig. 4(a)]. Permeation rate in the control membrane increases by three orders of magnitude, whereas that of the membrane with immobilized β-gal increases by two orders of magnitude. This behavior implies that water retention is enhanced in the catalytic membrane. The resulting resistance remains acceptable with linear permeation flux of water at 0.4–0.8 bar TMP.

Feed flow rate does not significantly influence the control and immobilized membranes at flow rates of 1.27 mL min⁻¹ to 7.62 mL min⁻¹ [Fig. 4(c) and (d)]. Despite that the total permeation flux and membrane resistance are affected after membrane conversion into a catalytic membrane, the characteristic of the membrane against feed flow rate is retained. The membrane resistant coefficient in the catalytic membrane remains constant as the feed flow rate changes. This finding indicates that the diffusive back transport and permeation of water are at equilibrium; furthermore, solvent retention and membrane compaction on the catalytic layer have no issues.

3.3 Effect of transmembrane pressure

The effect of TMP was studied at 0.2–0.8 bar during the β-gal-catalyzed conversion of lactose to GOS. TMP alters the hydrolysis and transgalactosylation reaction, as depicted in Fig. 5. The optimum TMP was found at 0.5 bar, at which the maximum conversions of lactose and GOS yield are achieved. The equilibrium between hydrolysis and transgalactosylation is monitored by the ratio between GOS over glucose yield. At TMP of 0.5 bar, the hydrolysis reaction significantly dominates the catalytic site compared with the transgalactosylation reaction with a ratio of 0.23. Higher GOS yield is achieved compared with the reactions at lower or higher applied TMP. This behaviour is attributed to the diffusive back transport of the solutes from the membrane surface to the bulk solution. The back diffusion of the solutes increases with increasing diffusion drag (TMP increases). Thus, the concentration gradient on the surface of the catalytic layer is formed. The concentration of the solutes exponentially increases toward the catalytic membrane surface; this phenomenon leads to high lactose concentration and subsequently enhances hydrolysis and transgalactosylation reactions. The rate of hydrolysis and transgalactosylation reactions increases as lactose concentration increases.^10,25–27 As a result, high conversion of lactose and GOS yields are achieved at TMP of 0.5 bar. However, further increase in TMP (>0.5 bar) does not improve the catalytic reaction. This finding could be due to the formation of gel layer on the membrane surfaces and compaction of the catalytic layer at high TMP. Hence, the mechanism of substrate adsorption and product desorption at the active sites are affected by diffusion limitation in the compacted region.

Fig. 5 Effect of TMP on β-gal catalyzed reaction. (a) Effect on lactose conversion and GOS yield. (b) Effect on separation factor (S_f,i). Effect on the formation of (c) trisaccharides, (d) tetrasaccharides, (e) glucose, and (f) galactose.

Nath et al. investigated the effect of TMP (1–5 bar) on lactose (0.045 M) conversion by Bacillus circulans immobilized on polyethersulfone.²⁸ Similarly, lactose conversion is optimal at TMP = 3 bar. These studies suggested that the permeation of monosaccharide inhibitors enhances lactose conversion at low TMP (<3 bar); conversely, limited contact time of the substrate caused by fast diffusive flux at high TMP (>3 bar) reduces the transgalactosylation reaction. This justification is ambiguous given that the permeate flux remains constant as the TMP is increased from 3 bar to 5 bar.²⁸ In the present work, the reduction of the transgalactosylation reaction could be due to the formation of a gel layer. As a result, permeation flux does not improve even with changes in TMP. If the solute concentration reaches saturation, then the solution may solidify on the catalytic membrane surface, thereby influencing the adsorption and desorption mechanism at the active site of the immobilized β-gal.

Separation performance was also studied and represented by separation factor (S_f,i) of products with respect to the substrate (lactose). The separation factor increases as TMP increases from 0.4 bar to 0.8 bar [Fig. 5(b)]. As a result, the fractionation of GOS and monosaccharides is improved. The separation of galactose from the reaction environment is crucial in reducing the inhibition effect on the immobilized β-gal. High retention of GOS on the catalytic site of the membrane is preferable for further transgalactosylation synthesis of high degree of GOS. Despite that the fractionation of GOS and inhibitor (galactose) improves as TMP is increased to 0.8 bar, the reaction performance decreases. This result implies that the removal of monosaccharides is not the determining factor for β-gal-catalyzed conversion of lactose in the EMR. The formation of a gel layer as a result of concentration polarization on the catalytic layer surface is more likely to be the determining factor. Catalysis rate is highly affected in the compacted region of the gel layer because of diffusive back transport of solutes from the catalytic layer. Therefore, low substrate concentration should be employed to prioritize the separation factor and prevent the formation of gel layer on the catalytic membrane surface at high operating TMP.

3.4 Effect of feed flow rate

The effect of feed flow rate was studied using 1.3–6.4 mL min⁻¹ during β-gal-catalyzed conversion of lactose to GOS. Fig. 6 shows that the substrate feed flow rate in the lumen side significantly affects the hydrolysis and transgalactosylation of lactose. The formation rate of monosaccharides and GOS decreases as the feed flow rate increases. The formation of glucose and galactose decreases by about 40%, whereas the formation of trisaccharides and tetrasaccharides decreases by about 20% and 40%, respectively. This result is possibly due to the fact that the hydrolysis reaction requires longer residence time for the catalysis compared with the transgalactosylation reaction. Moreover, fast nominal velocity results in a short retention period of the substrate in the lumen side of the membrane. As a consequence, the ratio between GOS over glucose yield is improved at high feed flow rates [Fig. 6(a)].

Fig. 6 Effect of substrate flow rate on β-gal catalyzed reaction. (a) Effect on lactose conversion as well as GOS yield. (b) Effect on separation factor (S_f,i). Effect on the formation of (c) trisaccharides, (d) tetrasaccharides, (e) glucose, and (f) galactose.

From a different perspective, the results could be due to significant changes in the mass transfer rate of the substrate to the catalytic membrane. Generally, permeation flux increases at high operating feed flow rate, at which the shear rate imparted on the membrane surface increases.^29,30 However, in the present work, permeation flux is not affected by the initial feed flow rate because the applied TMP is low (0.5 bar) [Fig. 4(c)]. Therefore, the diffusion of solutes through the catalytic membrane becomes the controlling step.³⁰ At high shear rates, solute or substrate diffusion to the polarized layer reduces, thereby influencing the substrate binding. The mass transport of substrate normal to the membrane surface is significantly higher than the mass transport perpendicular to the membrane surface at relatively low operating TMP (0.5 bar) and high flow rate (>1.27 mL min⁻¹). Therefore, the substrates are easily swept away from the vicinity of the catalytic membrane surface, resulting in reduced reaction rates.

Meanwhile, the fractionation of the product is slightly affected as the feed flow rate increases [Fig. 6(b)]. The fractionation of saccharides initially improves as the feed flow rate increases from 1.3 mL min⁻¹ to 2.6 mL min⁻¹. This finding is probably due to increase in the permeation of monosaccharides (glucose and galactose) or decrease in GOS level. However, the separation factor of monosaccharides and GOS over lactose is nearly constant with an approximate ratio of 1.60 and 0.45, respectively, at feed flow rates faster than 2.6 mL min⁻¹. This finding indicates that the fractionation of the solutes through the catalytic membrane is less affected, and further increase in flow rate does not cause any significant changes.

3.5 Effect of lactose concentration

Fig. 7 shows the comparison between EMR and batch reactor systems on the specific productivity of glucose, galactose, trisaccharides, and tetrasaccharides as the substrate solution is increased from 0.05 M to 0.15 M. Generally, the rate of transgalactosylation reaction increases as the concentration of the substrate increases. The specific productivity of trisaccharides and tetrasaccharides significantly increases as lactose concentration increased in both types of reactors [Fig. 7(a) and (b)]. The performance in EMR is more significant compared with the batch reactor. As shown in Fig. 7(c) and (d), the specific productivity of glucose is constant, whereas that of galactose decreases, as lactose concentration increased from 0.05 M to 0.15 M in the EMR system. In comparison, the specific productivity of glucose increases, whereas that of galactose is almost constant in the batch reactor system.

Fig. 7 Effect of substrate concentration on the specific production (M IU⁻¹) of (a) trisaccharides (±0.0002–0.0294), (b) tetrasaccharides (±0.0001–0.0027), (c) glucose (±0.0300–0.1113), and (d) galactose (±0.0076–0.0695) during β-gal-catalyzed conversion of lactose in hollow-fiber membrane reactor. Clear bullets are for EMR, and filled bullets are for the batch reactor system. ◊ 0.05 M lactose, □ 0.10 M lactose, and △ 0.15 M lactose. Operating condition: TMP = 0.5 bar, feed flow rate = 1.27 mL min⁻¹.

This result suggests the unique features of EMR that favors the transgalactosylation reaction compared with hydrolysis reaction. The EMR was operated with applied TMP, which generates pressure gradient between the lumen (feed or retentate) and shell side (permeate) of the membrane. The concentration and pressure differences between these two compartments are the most dominant driving forces that affect solute diffusion through the membrane. The separation of saccharide solution with identical chemical characteristics is solely dependent on the nominal molecular size of the solute. Water and monosaccharides have the smallest molecules in media with highest permeation rate. As a result, the ability of water as galactosyl acceptor decreases and the formation of galactose decreases. Meanwhile, the inhibition effect toward both the immobilized β-gal and β-gal-galactosyl complex is reduced as a result of permeation of the monosaccharides. Glucose and galactose are well-known to be inhibitors toward β-galactosidases; thus, continuous removal is an added advantage for the reaction mechanism as a whole.²⁵

In this work, the biocatalytic reaction in the membrane reactor is only limited to a single plane because of the immobilization of enzyme on the membrane surface. Hence, the formation of transition state is mostly dependent on the diffusion of the solute or substrate to the active site of the enzyme. Generally, larger solutes are retained better on the surface of the membrane because of convection or back diffusion. Therefore, lactose is the most dominant galactosyl acceptor on the surface of the catalytic membrane layer compared with water, which leads to higher formation of trisaccharides. However, the specific productivity of tetrasaccharides is significantly lower for EMR compared with the batch reactor system. The formation of tetrasaccharides is dependent on its ability to compete with lactose as the galactosyl acceptor. A high ratio of lactose over trisaccharides on the surface of catalytic membrane layer results in the low formation of tetrasaccharides. Compared with the batch reactor system, the high mobility of enzyme and diffusion of trisaccharides increase the probability of tetrasaccharide formation.

4. Conclusions

The β-gal-catalyzed conversion of lactose was conducted in enzymatic reactor system to improve reaction performance. This present work highlights that the rate of hydrolysis and transgalactosylation reaction in EMR with immobilized β-gal is also affected by hydrodynamic parameters, such as TMP and nominal feed flow rate. Apparently, these two parameters influence the mass transfer of the solutes to the surface of the catalytic membrane layer. The rate of transgalactosylation reaction increases as TMP increases because of the localization of lactose on the surface of the catalytic membrane layer. Besides, the diffusion of water also increases and reduces its affinity for β-gal-galactosyl complex and consequently reduces the formation of galactose. The rate of hydrolysis decreases at faster nominal feed flow rate because of short retention period. Hydrolysis reaction is suggested to require longer retention period compared with transgalactosylation reaction. Meanwhile, separation factor is significantly affected by TMP compared with nominal feed flow rate. The fractionation between GOS and monosaccharides increases as TMP increases. A comparison between the batch and EMR systems demonstrates that the specific productivity of tetrasaccharides is lower for EMR, which is possibly due to lower trisaccharides over lactose ratio on the catalytic membrane surface at the applied TMP. However, the immobilization of β-gal on the membrane surface significantly increased the membrane resistant coefficient with reduced permeation flux. This drawback is a challenge in the development of EMR for fine chemical production.

Acknowledgements

The authors would like to acknowledge the Malaysian Ministry of Science, Technology and Innovation (MOSTI) for funding the current study through the Science Fund scheme (03-01-05-SF0568, PJKIMIA/6013387). The research facilities provided by Universiti Sains Malaysia (USM) are also duly acknowledged.

References

1. L. Sie Yon, F. N. Gonawan, A. H. Kamaruddin and M. H. Uzir, Ind. Eng. Chem. Res., 2013, 52, 9441–9453.
2. W.-H. Huang, J. Sun, H. He, H.-W. Dong and J.-T. Li, Food Chem., 2011, 128, 968–973.
3. A. Trusek-Holownia and A. Noworyta, J. Biotechnol., 2007, 130, 47–56.
4. D. Sen, A. Sarkar, A. Gosling, S. L. Gras, G. W. Stevens, S. E. Kentish, P. Bhattacharya, A. R. Barber and C. Bhattacharjee, J. Membr. Sci., 2011, 378, 471–478.
5. H. A. Güleç, Colloids Surf., B, 2013, 104, 83–90.
6. H. A. Güleç, S. Gürdaş, N. Albayrak and M. Mutlu, Biotechnol. Bioprocess Eng., 2010, 15, 1006–1015.
7. S. Chockchaisawasdee, V. I. Athanasopoulos, K. Niranjan and R. A. Rastall, Biotechnol. Bioeng., 2005, 89, 434–443.
8. P. Czermak, M. Ebrahimi, K. Grau, S. Netz, G. Sawatzki and P. H. Pfromm, J. Membr. Sci., 2004, 232, 85–91.
9. D. B. Sarney, C. Hale, G. Frankel and E. N. Vulfson, Biotechnol. Bioeng., 2000, 69, 461–467.
10. M. I. Foda and M. Lopez-Leiva, Process Biochem., 2000, 35, 581–587.
11. C.-S. Chen, C.-K. Hsu and B.-H. Chiang, Process Biochem., 2002, 38, 801–808.
12. B. Splechtna, T.-H. Nguyen and D. Haltrich, J. Agric. Food Chem., 2007, 55, 6772–6777.
13. K. Pocedičová, L. Čurda, D. Mišún, A. Dryáková and L. Diblíková, J. Food Eng., 2010, 99, 479–484.
14. R. Das, D. Sen, A. Sarkar, S. Bhattacharyya and C. Bhattacharjee, Ind. Eng. Chem. Res., 2010, 50, 806–816.
15. L. Engel, P. Schneider, M. Ebrahimi and P. Czermak, Open Food Sci. J., 2007, 1, 17–23.
16. P. Jochems, Y. Satyawali, S. Van Roy, W. Doyen, L. Diels and W. Dejonghe, Enzyme Microb. Technol., 2011, 49, 580–588.
17. M. Portaccio, S. Stellato, S. Rossi, U. Bencivenga, M. M. Eldin, F. Gaeta and D. Mita, Enzyme Microb. Technol., 1998, 23, 101–106.
18. M. Ladero, A. Santos and F. Garcıa-Ochoa, Enzyme Microb. Technol., 2000, 27, 583–592.
19. R. E. Goodman and D. M. Pederson, Can. J. Microbiol., 1976, 22, 817–825.
20. A.-R. Park and D.-K. Oh, Appl. Microbiol. Biotechnol., 2010, 85, 1427–1435.
21. B. C. C. Pessela, C. Mateo, M. Fuentes, A. Vian, J. L. García, A. V. Carrascosa, J. M. Guisán and R. Fernández-Lafuente, Enzyme Microb. Technol., 2003, 33, 199–205.
22. F. N. Gonawan, A. H. Kamaruddin, M. Z. Abu Bakar and K. Abd Karim, Ind. Eng. Chem. Res., 2016, 55, 21–29.
23. S. Chellam and M. R. Wiesner, Environ. Sci. Technol., 1997, 31, 819–824.
24. R. Singh, Membrane Technology and Engineering for Water Purification: Application, Systems Design and Operation, Elsevier Science, 2014.
25. A.-R. Park and D.-K. Oh, Appl. Microbiol. Biotechnol., 2010, 85, 1279–1286.
26. D. F. Neri, V. M. Balcão, R. S. Costa, I. C. Rocha, E. M. Ferreira, D. P. Torres, L. R. Rodrigues, L. B. Carvalho and J. A. Teixeira, Food Chem., 2009, 115, 92–99.
27. L. M. Huerta, C. Vera, C. Guerrero, L. Wilson and A. Illanes, Process Biochem., 2011, 46, 245–252.
28. A. Nath, C. Bhattacharjee and R. Chowdhury, Desalination, 2013, 316, 31–41.
29. M. Di Luccio, B. Smith, T. Kida, C. Borges and T. Alves, J. Membr. Sci., 2000, 174, 217–224.
30. X. Tan, S. Tan, W. Teo and K. Li, J. Membr. Sci., 2006, 271, 59–68.

---