Continuous ultrafiltration membrane reactor coupled with nanofiltration for the enzymatic synthesis and purification of galactosyl-oligosaccharides

Separation and Purification Technology 144 (2015) 70–79

Contents lists available at ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Continuous ultrafiltration membrane reactor coupled with nanofiltration for the enzymatic synthesis and purification of galactosyl-oligosaccharides Hengfei Ren a, Junjie Fei a, Xinchi Shi a, Ting Zhao a, Hao Cheng a, Nan Zhao a, Yong Chen a,b,⇑, Hanjie Ying a,b,⇑ a b

College of Life Science and Pharmaceutical Engineering, Nanjing University of Technology, Nanjing 210009, PR China State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing University of Technology, Nanjing 210009, PR China

a r t i c l e

i n f o

Article history: Received 25 December 2014 Received in revised form 4 February 2015 Accepted 5 February 2015 Available online 21 February 2015 Keywords: Galactosyl-oligosaccharides b-galactosidase Ultrafiltration membrane reactor Nanofiltration Continuous diafiltration

a b s t r a c t Continuous enzymatic production of galactosyl-oligosaccharides (GOS) from a lactose substrate in an ultrafiltration membrane reactor (UMR) coupled with a nanofiltration separation system (CPNSS) was investigated. The overall rate of production over 4 h of continuous production was 80–104 mg GOS formed/U with an average residence time of 66 min, an initial lactose concentration of 300 g/L, an inlet pressure of 2.0 bar inlet pressure, and an outlet pressure of 1.5 bar. In the continuous diafiltration (CD) process, the concentration of various sugars and the relationship between the yield and purity of oligosaccharides was well predicted by mathematical models, the increased rate of sugar rejections was less than 10%, and the decreased rate of concentrations in the tank was less than 15%. In the CPNSS, 33.4 wt.% GOS was obtained in the UMR, and 1.24 kg of high-purity GOS (specifically, GOS purity of 57.2% and lactose content less than 20%) was achieved. This final yield was 80.1% GOS, which meets industry standards. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction In recent years, some of the most significant developments in functional food science have been related to the development of dietary supplements that beneficially affect the microbial composition of the gut [1,2]. Modulation of the ecology of the gut to improve the well-being of the host, by administering probiotics and prebiotics, is attracting increasing attention [3,4]. Prebiotics are non-digestible oligosaccharides (NDOs) which have been found to reach the human colon without being hydrolyzed or absorbed in the upper part of the gastrointestinal tract [5,6]. The key aspect of a prebiotic is that it is selectively metabolized by benign or health-positive species, such as bifidobacteria and lactobacilli, at the expense of less desirable groups, such as clostridia [7,8]. Galactosyl-oligosaccharides (GOS) are NDOs, which are recognized as prebiotics [9]. GOS selectively stimulate the growth ⇑ Corresponding authors at: College of Life Science and Pharmaceutical Engineering, Nanjing University of Technology, Nanjing 210009, PR China. Tel./fax: +86 25 86990001. E-mail addresses: [email protected] (Y. Chen), yinghanjie@njtech. edu.cn (H. Ying). http://dx.doi.org/10.1016/j.seppur.2015.02.020 1383-5866/Ó 2015 Elsevier B.V. All rights reserved.

of bifidobacteria in the lower part of human intestine. Increase in the growth of bifidobacteria is usually accompanied by the suppression of potentially harmful bacteria, such as the Clostridium and Bacteroides genera, in the intestine [10]. GOS consist of galactosyl-galactose chains with a terminating glucose and are the desired by-products of the enzymatic catalysis [11,12]. The composition of the GOS fraction varies in chain length and in the interconnection of the monomer units with varying b-glycosidic linkages depending on the enzyme’s source [13]. In contrast to the hydrolyzation of lactose, GOS are very slowly hydrolyzed both in vivo and in vitro. The GOS produced by b-galactosidases are low in molecular weight, not viscous, and water-soluble liquid dietary fibers that are stable at elevated temperatures and at low pH [14]. There have been several investigations on the synthesis of GOS by b-galactosidases from various sources. Among these sources, b-galactosidase from Kluyveromyces lactis has been extensively studied [15]. The enzyme was reported to have stronger hydrolytic activity than transferase activity and produced a high proportion of trisaccharide in the synthetic GOS mixtures [16,17]. There have been many studies on the immobilization of the enzyme, and various substrates have been investigated [18–20]. Continuous synthesis of galacto-oligosaccharide from lactose using b-galactosidase

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

71

Nomenclature J V A t

s m PR P E LAC TMP R L P C y Pu Y

the permeate flux (L (m2 h)) volume (L) the membrane effective area (m2) time (h) the average residence time (h) the average permeate flow rate (L/h) productivity (g of GOS formed/U) the average product output (g/h) the enzyme concentration (U/L) the average conversion of lactose (%) the transmembrane pressure (bar) the rejection coefficient (%) the concentration of lactose (g/L) pressure (bar) concentration (g/L) the yield value purity (%) yield (%)

from K. lactis in an ultrafiltration membrane reactor (UMR) has been reported [15,21,22]. However, to our knowledge, the continuous synthesis of GOS from lactose using b-galactosidase and a nanofiltration membrane to produce high concentrations of GOS has not yet been investigated. In this study, continuous production of GOS is achieved using a cross-flow UMR. We also investigated the optimization of the yield by selectively lowering the concentration of the product from the reaction mixture. Experiments evaluating the dependence on pressure, temperature, and concentration were conducted to determine the performance of the nanofiltration membrane based on the permeate flux and the apparent rejection coefficient. Then, the purification of a commercial GOS mixture was performed with the nanofiltration membrane using a continuous diafiltration (CD) procedure. Finally, we investigated the continuous synthesis of GOS from lactose using b-galactosidase coupled with nanofiltration membrane to achieve a high yield of GOS.

2. Experimental 2.1. Chemicals and materials Deionized water, 5 mmol/L potassium phosphate (pH 6.5) containing 5 mmol/L MgSO4 as a buffer were used in all experiments. b-Lactose was kindly provided by the Lactose Company of NEW Zealand Limited (NEW Zealand). The enzyme Lactozym Pure 6500L from K. lactis with an activity of 6500 LAU/g was kindly provided by Novozymes (Denmark). Other chemicals were of analytical grade and purchased from Sinopharm Chemical Reagent Co., Ltd (China).

2.2. Enzyme The commercially available enzyme used was b-galactosidases by Lactozym Pure 6500L(Novozymes, Denmark). The enzyme was highly purified, liquid, and prepared from K. lactis with an optimum temperature of 25–40 °C and an activity of 6500 LAU/g, where one LAU is defined as the quantity of enzyme that liberates 1 lmol o-nitrophenol from o-nitrophenyl-b-D-galactopyranoside (ONPG) per minute under standard conditions. The enzyme had a molecular weight exceeding 100,000 Da [23].

Superscript A micro-molecular solute B macro-molecular solute Subscript up np p r in F out f i

the ultrafiltration membrane permeate the nanofiltration membrane permeate permeate reaction mixture inlet feed outlet final initial

2.3. Membranes Table 1 lists the membranes used in this study. The membranes include an ultrafiltration (UF) membrane and a nanofiltration (NF) membrane. The UF membrane module (UOF4, Tianjin Motian Membrane Technology Co., Ltd., Tianjin, China) is tubular and contains a shell of UPVC material and 5000 roots membrane silks (350 lm i.d.  450 lm o.d.) of composite regenerated cellulose material. The ultrafiltration membrane had a 50-kDa nominal molecular weight cut-off (NMWCO) and an effective area of 0.12 m2. Maximum operating pressure and temperature for the membrane were stated to be 3.0 bar and 45 °C, respectively. The NF membrane is the patented three-layer composite film membrane obtained from General Electric Company (United States). The membrane from the inside to the outside is divided into three layers: the first layer is a thin separation layer made of a polyethylene piperazine material with a thickness of about 0.2 lm and a pore diameter of about 0.1–0.2 nm; the second layer is a porous intermediate support layer made of polysulfone with a thickness of about 40 lm and a pore diameter of about 15 nm; the third layer is a non-woven fabric made of polyester material with a thickness of about 120 lm. The NF membrane has a total active filtration area of 0.32 m2. The maximum operating pressure and temperature for the membrane were stated to be 10.0 bar and 50 °C, respectively. 2.4. Membrane properties towards enzyme A volume of 800 mL of enzyme (1 U/mL) diluted in the synthesis buffer was placed in the reactor vessel, passed through the membrane, and recycled to the vessel. The experiment was run at room temperature for four hours maintaining a 2.0 bar inlet pressure and a 0.5 bar outlet pressure. Every 30 min, samples (1 mL) were taken from both the retentate and permeate streams. 2.5. Continuous synthesis of GOS in an UMR Continuous production of oligosaccharides from lactose was performed using the laboratory equipment shown schematically (Fig. 1). The UMR contains a 1.5 L stirred-tank, an ultrafiltration membrane, and an electronic diaphragm pump. The reactor volume was 800 mL. The flow rate of the feed was controlled by an

72

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

Table 1 Membrane materials and properties. Membrane

Membrane type

Test pressure (psi)

NMWCO

Effective area (m2)

Material

pH range

Max. Temp. (°C)

Max. Pres. (bar)

UF NF

UOF4 DL1812C-34D

30 100

50,000 400–600

0.12 0.32

Composite regenerated cellulose Three-layer composite film

2–10 2–11

50 50

4 10

Fig. 1. Schematic of the continuous reactor/membrane system with b-galactosidase and enzyme circulation.

electronic diaphragm pump, which could be adjusted to work under 4.0 bar pressure by means of the restriction valve controlling the outlet pressure. Before introducing the reaction mixture, the membrane was prepared by flushing it with water for 30 min and maintaining the inlet pressure at 2.0 bar and the outlet pressure at 0.8 bar. Water flux was then measured to ensure that the membrane was in the same working conditions as the previous runs. After the experimental run was completed, the membrane was cleaned by circulating 0.1 M NaOH solution through the system for 30 min and then flushing it with water for another 30 min. Pressures applied during cleaning were the same as those used in the preparation step. Water flux was measured to check the cleaning performance. Repetition of cleaning was performed if the flux was not restored to the initial value. The membrane was stored at 4 °C in

0.05% (w/v) sodium hydroxide solution to prevent microbial contamination. The synthesis was carried out by varying the mean residence time of the process. Inlet pressure was varied between 1.0 and 3.0 bar, and the outlet pressure ranged between 0.5 to 2.5 bar (Tables 2 and 3). Lactose solution was prepared at a concentration of 300 g/L in the synthesis buffer. A volume of 800 mL of filtered lactose solution was heated/cooled to 25 °C before the addition of 1 U/mL b-galactosidases to initiate the reaction. The reactor was stirred at 100 rpm to ensure adequate mixing. During the synthesis, fresh lactose substrate (300 g/L) was fed into the reactor to replace the volume of permeate stream removed. The volume of permeate was recorded and samples (1 mL) were collected at specific time intervals. All samples were stored at 18 °C for subsequent analysis. The permeate was collected in the outer com-

Table 2 Operating parameters, productivities, and conversions associated with the continuous system.a

a

Inlet pressure (bar)

Outlet pressure (bar)

Transmembrane Pressure (bar)

Average flux (L/m2/h)

Average res. time (min)

Average instant. prod. (mg/U/h)

Total cumul. prod. (mg/U)

Average lactose conv. (%)

1.0 2.0 2.0 2.0 3.0 3.0 3.0 3.0

0.5 0.5 1.0 1.5 1.0 1.5 2.0 2.5

0.75 1.25 1.5 1.75 2.0 2.25 2.5 2.75

5.45 5.81 6.35 6.68 7.87 8.56 9.24 10.13

84.23 79.12 72.34 66.15 58.36 53.67 49.72 45.39

20.08 21.6 24.2 25.1 25.05 25.29 25.52 25.85

80.29 86.41 96.78 100.2 100.17 101.18 102.09 103.39

77.36 72.28 73.49 75.23 71.72 70.17 68.54 68.21

Synthesis starting with 4 L of 300 mg/mL lactose containing 1 U/mL b-galactosidases, operating over 4 h at 25 °C and pH 6.5.

73

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79 Table 3 Permeate flux at each hour during continuous GOS synthesis.* Inlet pressure (bar)

Outlet pressure (bar)

Transmembrane pressure (bar)

Parameters flux (L/m2/h) First hour

Second hour

Third hour

Fourth hour

1.0 2.0 2.0 2.0 3.0 3.0 3.0 3.0

0.5 0.5 1.0 1.5 1.0 1.5 2.0 2.5

0.75 1.25 1.5 1.75 2.0 2.25 2.5 2.75

5.54 6.11 6.87 7.64 9.28 9.99 11.10 12.23

5.52 5.78 6.35 7.25 9.16 8.87 9.91 10.61

5.49 5.74 6.14 6.71 8.57 7.99 8.29 9.58

5.27 5.61 6.04 6.19 7.24 7.39 7.66 8.07

Flux reductiona (%) 5 8 12 19 22 26 31 34

*

Synthesis starting with 800 mL of 300 mg/mL lactose containing 1 U/mL b-galactosidases, operating over 4 h at 25 °C and pH 6.5. The percentages of flux reduction were calculated from the permeate fluxes measured after the fourth hour compared to those measured after the first hour of the synthesis. a

partment. No permeate was recycled. At steady state for which the feed and permeate flow rates are equal and constant, the UF permeate fluxes (Jup) (L/(m2 h)) were calculated as follows:

J up ¼

Vp At

ð1Þ

where Vp is the permeate volume (L), A is the membrane effective area (m2), and t is the time duration (h) over which the permeate flow was measured. The average residence time (h) was obtained from:

s¼

Vr

ð2Þ

m

where Vr is the volume of reaction mixture (L), and m is the average permeate flow rate over the course of the reaction at steady state (L/h). The productivity of the bioreactor was measured as g of GOS per unit of biocatalyst used (g of GOS formed/U) and was calculated as an ‘‘instantaneous productivity’’ over a 2 h period. The instantaneous productivity (PRi) (g of GOS formed/U) is expressed as the yield obtained hourly:

PRi ¼

P  t Vr  E

ð3Þ

where P is the average product output (g/h) in the time period t (which for this present study was 2 h), m is the permeate flow rate (L/h), Vr is the reaction volume (L), and E is the enzyme concentration (U/L). The cumulative productivity (PRc, g of GOS formed/U) is the total yield obtained over the whole course of the reaction. The average conversion of lactose (LAC, %) was estimated as:

LAC ¼

  Li  Lp  100 Li

ð4Þ

where Li is the inlet concentration of lactose (g/L) and Lp is the lactose concentration in the permeate (g/L). The transmembrane pressure (TMP) for cross-flow ultrafiltration process was estimated as:

TMP ¼ PF  Pp

ð5Þ

where PF is the pressure of the feed and PP is the pressure on the permeate side of the membrane. The cross-flow gives rise to a pressure drop from the inlet to the outlet of the module, and PF is expressed as

PF ¼

Pin þ Pout 2

ð6Þ

where Pin is the pressure at the membrane inlet (inlet pressure), and Pout is the pressure at the membrane outlet.

2.6. Experimental procedure and theory of nanofiltration The new NF membrane was flushed with demineralized water for 30 min to remove possible contaminants. Then, it was precompacted with water for 1 h at 4 bar and 25 °C to achieve a constant flux, and the initial water flux was measured to later compare with the data obtained from the experiment. In the experiments, the total recycle mode was used to avoid the change of feed concentration, in which all of the permeate and retentate were recycled to the feed tank. The initial feed concentration was measured after the solution was cycled for 10 min in the system. The NF membrane permeate flux, Jnp (L/(m2 h)), and the apparent rejection coefficient were determined right after the system was stabilized for 30 min. Before starting of the CD process, the system was operated in total recycle mode for more than 30 min to reach the stabilized state. The initial permeate flux was measured, and the feed and permeate were sampled for the initial component analysis. Then, the permeate was removed from the system while simultaneously an equal amount of dilution water from the measuring cylinder was pumped into the feed tank to maintain the constant volume of feed. During the process, the measurement of the permeate flux and the sampling of the tank and permeate solution were carried out after each collecting cycle (2 L of permeate removed) was finished. A GOS mixture with more than 50% purity (about 1.5 times of the initial purity) could be used as a functional food or food ingredient. Alternatively, this mixture is also amenable to downstream processing (ion exchange chromatography or gel chromatography). Therefore, the CD process was stopped when the purity of oligosaccharides exceeded 1.5 times of the initial purity. The NF membrane permeate flux (Jnp) was expressed in (L/ (m2 h)):

J np ¼

Vp At

ð7Þ

where VP is the permeate volume, A the membrane effective area, t the time necessary for the VP of permeate to be collected. The observed rejection coefficient for a given solute, based on the concentration determined from the sample analysis, was calculated from the equation

R¼1

Cp CF

ð8Þ

where CP and CF are the concentration of the solute in the permeate and the feed, respectively. The yield values for each individual component throughout the CD process were calculated as a fraction of the original solute concentration remaining in the feed:

74

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

C F;f  100% C F;i

y¼

ð9Þ

where CF,f and CF,i are the concentrations in the final and initial feed, respectively. CD is a membrane process where water, at the appropriate pH and temperature, is added to the feed tank at the same rate as the permeate flux keeping the feed volume constant during processing [24]. Mathematical models found in the literature (e.g. Pontalier et al.) [25], describe mass transfer in nanofiltration separation processes by taking into account the membrane characteristics (e.g., pore size, charge), the feed solute and solvent characteristics (e.g., solution viscosity, solute radius), and also the type of solute flux through the membrane (e.g., diffusive, convective). The purpose of this study was to estimate the highest degree of CD purification, therefore a simple mathematical analysis of the CD process described by Grandison and Lewis [26], was used. Specifically, when R = 0 for a given solute and⁄ Cp = C(1  R), the concentration of that solute in the feed can be calculated using the following equation:

ln



C F;i C F;f



¼ ð1  RÞ

Vp VF

ð10Þ

where VF and Vp are the feed volume and the cumulative permeate volume removed during the course of the process, respectively. Similarly, the variation in concentration of a specific solute in the permeate can be monitored from the equation:

ln

  C F;i ð1  RÞ Vp ¼ ð1  RÞ Cp VF

ð11Þ

when R = 0 for a given solute, Eqs. (10) and (11) are simplified as the following equation:

    CF;i Vp C F;i ¼ ln ¼ ln C F;f VF CF

ð12Þ

C F;f ¼ C F

ð13Þ

Both Eqs. (12) and (13) are applicable when the rejection rate of a solute remains constant throughout the CD process, which for this study requires the total sugar concentration to remain fairly constant throughout the process. A multi-solute system is used in this study. Hence, all the solutes can be classified into two types: micro-molecular solute A and macromolecular solute B (A denotes monosaccharide and disaccharide; B denotes oligosaccharides). The purity was the percentage of the oligosaccharides in all solutes; the yield was the percentage of the original oligosaccharides remaining in the feed solution at a constant volume diafiltration. Purity Pu and yield Y are defined by the following equations:

Pu ¼

Y¼

C BF C FA þ C BF

C BF;f C BF;i

ð14Þ

ð15Þ

where C FA , C BF are the concentrations of micro-molecular and macro-molecular solute in the feed tank, and C BF;f ; C BF;i are the concentrations of the macro-molecular in the final and initial feed solution, respectively. Based on the assumption that concentration polarization is negligible, that is, permeate factor Pc (the ratio of the distance (m) through the membrane to the time (h) of the period, m/h) is constant, and is not dependent on the change of feed concentration, the yield can be expressed by Eq. (16) [27], and the

relationship between the purity and yield can be expressed as Eq. (17) [28]:

  Vp Y ¼ exp PBp VF

Pu ¼

1þ

A C F;i

C BF;i

Y

DPp =P Bp

ð16Þ !1 ð17Þ

where DPp = PBp  PpA : The quantity Vp/VF is defined as the multiple of the dilution; P pA , PBp

are defined as the permeation factors of micro-molecular and

A macro-molecular solute, and C F;i , C BF;i are the concentrations of micro-molecular and macro-molecular solute in the initial feed solution, respectively.

2.7. Continuous Production in a UMR coupled with nanofiltration separation In this study, continuous enzymatic production of GOS from lactose as a substrate in a UMR coupled with nanofiltration separation system (CPNSS) was conducted. The system consisted of an ultrafiltration membrane reactor and a nanofiltration membrane separation device (Fig. 1). The nanofiltration membrane device contained an electric diaphragm pump, a feed tank with controlled temperature and a nanofiltration membrane (molecular weight cut-off of 400– 600 Da). The temperature of the system was measured with temperature gauges in the pot and the exit of the loop. An electric diaphragm pump was used to pump the feed solution to the nanofiltration membrane. The electric diaphragm pump had two symmetric chambers on the left and right sides, each of which had two ball valves installed. When the electromotor drives the diaphragm to oscillate to-and-fro, the four valves alternate in turning on and off. At the same time, the electric diaphragm pump continuously inhales and ejects sugar solution. The pressures, controlled with a retentate valve, were indicated by two pressure gauges above and below the membrane module. The average value of the two gauges was taken to be the applied pressure. For the CD process, the volume of the dilution water and permeate was measured. To keep the feed volume constant, the flow rate of the dilution water was adjusted to be equal to the permeate stream using the peristaltic pump. The permeate was collected in the outer compartment. No permeate was recycled. All reported experimental results where oligosaccharides were produced are for steady-state conditions. 2.8. Analytical methods Lactose and reaction products were determined using Agilent Technologies 1200 Series HPLC equipment, with a refractive index detector (G 1362A), an isocratic pump (G 1311A), and an autosampler (G 1329A) using AminexÒ HPX-42A Ag+ Column (300 mm  7.8 mm) for carbohydrate analysis (Bio-Rad, Hercules, California, USA). Samples were diluted and filtered through 0.22 lm Millipore Durapore membranes prior to assay by HPLC and were eluted with milli-Q water at a flow-rate of 0.4 mL/min. Column and detector temperatures were 80 and 40 °C, respectively. Enzyme activity was assayed using o-nitrophenyl-h-galactopyranoside (oNPG) as a substrate at 40 °C for 10 min; 2 mL of 0.1 M Na2CO3 was added to inactivate the enzyme, and the absorbance was read at 420 nm. Protein was determined utilizing a dye binding method (Bradford) and bovine serum albumin (BSA) as the standard protein [29]. Each sample was analyzed twice, and the average was used.

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

75

3. Results and discussion 3.1. Continuous synthesis of GOS Based on previous studies, a high initial concentration of lactose could reduce hydrolysis and favor the GOS synthesis [15,30]. The probability of the galactosyl-enzyme complex, formed in a first reaction step, reaching a sugar as an acceptor is increased when less water molecules are available in the system. When the acceptor is a water molecule, free galactose is formed by hydrolysis [31]. Thus, we chose 300 g/L lactose as the initial substrate. Before carrying out the synthesis experiment, the compatibility between the enzyme and the membrane was tested. The aim was to ensure that the membrane could retain the enzyme without any significant loss of activity over time. No protein was detected in the permeate stream indicating that the membrane could completely retain the enzyme. Lactose conversion and GOS productivity have no significant change, which indicates that no loss of activity was found in the retentate stream. After cleaning the membrane with 0.1 N NaOH, the water flux could be restored to the initial value indicating that the enzyme did not cause irreversible fouling. Eight different residence times were investigated in duplicate. Table 2 shows the average values of the relevant parameters at steady state. The results show that the flux increased with increasing transmembrane pressure in most experiments, except at the highest transmembrane pressure. This can be explained by the theory of pressure-driven membrane separations. Theoretically, for pure solvents, flux is directly proportional to the applied pressure and inversely proportional to viscosity. However, with respect to solutions, this is true only under low pressures, low feed concentrations, and high feed velocities, for which concentration polarization effects are minimal. Concentration polarization is the resistance layer generated on the membrane. This is caused by the retardation of the flow near the surface of the membrane due to frictional forces between the reaction mixture and the membrane. Applying high pressure results in increasing the thickness of the polarized layer, and therefore, the flux becomes pressure-independent [24,32,33]. The permeate fluxes of the reaction mixture (300 mg/mL of total sugars) after 4 h of operation at various operating pressures resulted in a reduction of the fluxes (up to 34%) compared to their initial values (Table 3). This decrease occurred at a greater magnitude when applying higher pressures (Table 3), presumably due to the greater impact of concentration polarization [15,33]. At steady state (after 120-min reaction time when the feed and permeate flow rates are equal and constant), no reduction in the percentage of lactose conversion was detected at any operating pressure, indicating that the retained enzyme was not inactivated under those applied pressures. Instantaneous productivities during continuous synthesis varied from 20 to 27 mg GOS formed/U at 70–80% lactose conversion. Overall productivities over 4 h of continuous operation were 80– 104 mg GOS formed/U. From the experimental results, the optimum applied pressure based on the conversion of lactose for continuous GOS synthesis was 2.0 bar for the inlet pressure and 1.5 bar for the outlet pressure. Under these conditions, the reaction mixture had a residence time of 66 min, allowing approximately 80% of lactose conversion. After 4 h of synthesis, the process gave an average productivity of 100.2 mg GOS/U (25.1 mg GOS/U/h).

3.2. Selection of nanofiltration operating conditions and CD 3.2.1. Effect of pressure With both retentate and permeate recycled to the feed bank, pressure dependence experiments were conducted to select the

Fig. 2. The effect of pressure on permeate flux and apparent rejection coefficients of the sugars with the NF membrane at 25 °C.

suitable pressure according to permeate flux and solute apparent rejection factors. After the system was stabilized for an hour, the permeate flux of GOS mixture (50 g/L) was measured from 2 to 8 bar at 25 °C (Fig. 2). The apparent rejection values of the membranes for various sugars are shown in Fig. 2. The NF membrane tested with sugar solutions showed a linear relationship between the permeate flux and the applied pressure. The NF membrane tested with sugar solutions showed a linear relationship between the permeate flux and the applied pressure. The difference in rejections between glucose (monosaccharide) and lactose (disaccharide) is much larger than the difference between sucrose and GOS (trisaccharide) (Fig. 2). Solute flux is directly dependent on the pressure when permeation of the solutes is convective controlled, while it becomes independent of the pressure when permeation is diffusive controlled. The rejection values for the three sugars were directly dependent on the total sugar concentration of the feed solution, and increased with pressure due to membrane compaction and also due to increased solvent flux. Compaction reduces the membrane thickness, and that normally would lead to an increased permeate flux. However, pore size reduction, caused also by compaction, is the predominant characteristic upon which the rejection of neutral solutes is dependent (sieving effect), hence causing an overall increase in the rejections observed. Increased pressure also caused the differences between the rejections of the three sugars to decrease, indicating a less effective separation. That is because the effect of pressure on rejection is less marked as the molecular weight of the sugar increases, with respect to the pore size of the membranes used. As pore size decreases, increasing pressure causes the convective flux of solutes to decrease to a much greater extent than the diffusive flux. Thus, with neutral solutes, the significance of convective solute flux becomes greater as the molecular size of the sugar decreases leading to greater changes in rejection. The membrane with high permeate flux of the solution, low rejection of micro-molecular solutes, and high rejection of macromolecular solutes were expected. Our findings show that the NF used in this study offers lower permeate fluxes and higher rejection of monosaccharides than those of nanofiltration membranes previously reported [33,34], a difference that may be due to the different materials used for the NF membranes. The NF membrane is an appropriate membrane due to its permeate flux and relatively large difference in rejections between macro-molecular and micromolecular sugars. The optimum pressure for a separation process should also be a compromise between the permeate flux and the apparent rejection.

76

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

Fig. 3. The effect of temperature on permeate flux and apparent rejection coefficients of the sugars with the NF membrane at 5 bar.

3.2.2. Effect of temperature The effect of temperature on the permeate flux and apparent rejection coefficients of the sugars was studied with the NF membrane at 5 bar. When the temperature increased from 20 to 50 °C (the operation temperature of NF is 5–50 °C), the permeate fluxes increased linearly, and the apparent rejection coefficients of the sugars decreased (Fig. 3). This is due to the reduced viscosity of the feed solution and an increase in effective pore diameter of the membranes with increasing temperature, which agrees with the results of other investigators [15,33,35,36]. The monosaccharide (glucose) showed the greatest change in rejection values, followed by the disaccharide (lactose), which showed a similar, but less marked, change in rejection than glucose. The rejection of GOS, which was totally rejected at 20 °C, remained constant at elevated temperatures (Fig. 3). The constant rejection of GOS shows that the actual pore size of the membrane was not affected by temperature, although the effective pore diameter changed due to the thinner layer of adsorbed water molecules on the pore walls. Therefore, it is concluded that high temperature is favorable for the separation of the GOS mixture. 3.2.3. Effect of feed concentration The effect of feed concentration on the separation of the GOS mixture with the NF membrane was studied under a pressure of

Fig. 4. The effect of feed concentration on permeate flux and apparent rejection coefficients of the sugars with the NF membrane at 5 bar and 25 °C.

5 bar and the highest permitted temperature in the total recycle mode of operation. The feed concentration refers to the total sugar concentration of the GOS mixture before entering the NF membrane. A low concentration of feed solution was prepared in demineralized water and placed into the feed tank. After each measurement, appropriate powder material was added into the initial solution to obtain a series of higher concentrations. The effect of feed concentration on the permeate flux is shown in Fig. 4. The apparent rejection values of various sugars at varying feed concentrations are also shown in Fig. 4. As shown in Fig. 4, the permeate flux decreased with the increase of feed concentration due to the accretion of concentration polarization. With the increase of feed concentration, the effect of concentration polarization on the permeate flux increases and the apparent rejection coefficients of the sugars decreases. The apparent rejection coefficients of the sugars decreased with the increase of feed concentration. However, the rejection of neutral species at constant pressure is independent of concentration [37]. Therefore, the main reason for the decrease of the sugar rejection is due to the decrease in the pressure driving force induced by the increase of osmotic pressure (around 0.03 bar) with a constant applied pressure. The rejection of oligosaccharides decreased slightly, whereas the rejection of glucose and lactose showed large decreases with the increase of feed concentration. This resulted in the larger rejection differences between oligosaccharides and micro-molecular sugars, which was favorable for the separation of the GOS mixture. Therefore, the optimum feed concentration must be a compromise between the permeate flux and rejection coefficients. Based on the above-mentioned analysis, the optimum operating conditions of nanofiltration separation are 45 °C temperature, 5 bar pressure, and a feed concentration of about 50 g/L.

3.2.4. Continuous Diafiltration The CD process was carried out at 45 °C, 5 bar pressure, and 50 g/L initial concentration of the GOS mixture and analyzed with HPLC after the feed solution cycled for 20 min. During the separation, we measured the permeate flux and sampled the feed tank and permeate at the end of each collecting cycle when 2 L permeate was removed. Trace fouling appeared in the experiments, which may stem from the almost same permeate fluxes of demineralized water before and after the nanofiltration process. As shown in Fig. 5, during the process of CD, the permeate flux increased continually due to the decrease of concentration

Fig. 5. The effect of cumulative permeate volume on permeate flux and tank concentration at 45 °C and 5 bar in the CD process.

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

Fig. 6. The effect of cumulative permeate volume on experimental and the predicted sugar concentration of the feed tank in CD process with NF membrane at 45 °C and 5 bar.

polarization resulting from lower concentration and decreased viscosity of feed sample. The concentration of various sugars in the permeate and feed tank solution are shown in Figs. 6 and 7. All concentrations, predicted by Eqs. (12) and (13), are presented in these figures. The average apparent rejection values were used in Eqs. (12) and (13). If R is allowed to vary with time, the predicted concentration would perfectly correlate with actual concentration. Grandison and Lewis proved that Eqs. (12) and (13) were applicable when the rejection rate of the solute remained constant throughout the CD process, which required the total sugar concentration to remain nearly constant during the whole CD process [38]. In this study, although the rejection of glucose increased from 0.56 to 0.75 (Fig. 8) and the whole sugar concentration decreased from 50 to 27.0 g/L (Fig. 5), the concentrations of various sugars can be still predicted. Therefore, Eqs. (12) and (13) can be used in a wider operating range. When the volume of every batch of permeate removed is equal to the volume of the feed tank, the increased rate of sugar rejection was less than 10%, and the decreased rate of tank

Fig. 7. The effect of cumulative permeate volume on experimental and predicted sugar concentration of the permeate in CD process with NF membrane at 45 °C and 5 bar.

77

Fig. 8. The effect of cumulative permeate volume on the apparent rejection coefficient and yield of the sugars in the CD process with NF membrane at 45 °C and 5 bar.

concentrations was less than 15%. In this case, the model equations properly predict the CD process in this study. Fig. 8 shows the relationship between the yield of various sugars and the cumulative permeate volume during the CD process at 45 °C, 5 bar. The yield of glucose decreased rapidly with an increase in the cumulative permeate volume. Only 9.5% glucose remained in the feed tank when the cumulative permeate volume reached 10 L, which was 5 times that of the feed solution. The yield of lactose and GOS also decreased, which was much lower than that of glucose. This results from lower glucose rejection and relatively higher lactose and GOS rejection during the CD process. From Fig. 8, the difference in rejections between GOS (MW > 504 Da) and lactose (MW 342 Da) is less than that between glucose (MW 180 Da) and lactose. Moreover, the rejections of GOS and lactose changed slightly with the decrease of the feed tank concentration in the CD process, whereas the rejection of glucose increased obviously. This indicates that the rejections are not only dependent on the solute molecules but also on other factors, such as the different spatial configurations of the monosaccharides, disaccharides, and GOS. Eqs. (16) and (17) show that a low permeation factor of macromolecular solute and a large permeation factor difference between micro-molecular and macro-molecular solute results in high purity and yield of the GOS. The NF membrane has low disaccharide rejections and relatively large differences in the rejection between disaccharide and GOS. However, NF presented relatively lower rejections of GOS resulting in a great loss of GOS during CD. Consequently, this requires more energy cost and operation complexity. In fact, as shown in Eq. (17), the permeation factors of solutes and the initial composition of mirco-molecular and macro-molecular solute are important factors affecting the relationship between purity and yield of GOS. Because the difference of rejections between GOS and lactose is less than the difference of rejections between GOS and glucose, the content of micro-molecular lactose is much more important. If the initial GOS mixture has a high concentration of GOS and a low concentration of lactose, the NF process can obtain satisfactory GOS yield and purity. However, many GOS mixture products contain 24–40 wt.% of GOS, more than 35 wt.% of lactose, and a small amount of monosaccharide [39– 41]. In this study, with the specifications of 30.2 wt.% monosaccharide, 36.4 wt.% lactose and 33.4 wt.% oligosaccharides, 89.3% monosaccharide and 51.7% lactose in the mixture were removed

78

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

the permeate factor Pc is constant), it could be used to predict the relationship between purity and yield as long as the solute permeate factor and feed tank concentration changed proportionally and slowly.

3.3. Continuous production in an UMR coupled with NF

Fig. 9. Relationships between experimental and predicted purity with yield.

Fig. 10. Kinetics of the CPNSS. (A): Cumulative sugars. (B) Sugars concentration in the UMR.

with the nanofiltration membrane. The GOS purity of 55.3% (1.65 times of the raw material) and yield of 77.4% were achieved. Fig. 9 shows the relationship between experimental purity and predicted purity with yield of oligosaccharides. The experimental purity of oligosaccharides was calculated with Eq. (14), and the predicted purity was calculated with Eq. (17). The results indicated that Eq. (17) predicted the purity values well. Specifically, the deviation between the experimental and the predicted purity was less than 6%. Although Eq. (17) was obtained based on the assumption that the concentration polarization is negligible (i.e.,

Based on the above theory models and the experimental results, continuous production in an ultrafiltration membrane reactor coupled with nanofiltration separation system is feasible. Based on the previously described experimental analysis, the optimum reactor conditions for continuous GOS synthesis are 25 °C, pH 6.5, 1 U/ mL, a lactose substrate of 300 g/L, a residence time of 66 min, an inlet pressure of 2.0 bar, and an outlet pressure of 1.5 bar; the optimum operating conditions of nanofiltration are 45 °C, 5 bar pressure, and a feed concentration of about 50 g/L. In the CPNSS, continuous production started after a 2-h reaction time (the GOS concentration was approximately 100 g/L) to continuously remove the sugar from the reaction mixture into the feed tank while the enzyme is intercepted by UF to catalyze the synthesis of oligosaccharides, and concentrated lactose is pulse-fed into the reaction tank. The demineralized water was also pulse-fed into the feed tank to maintain sugar concentrations at around 50 g/L. Nanofiltration was initiated to purify oligosaccharides from the reaction mixture when the volume of the reaction mixture can be water bath, most of monosaccharides and disaccharides through the nanofiltration membrane to the permeate storage tank, the majority of the GOS is intercepted by NF and returned back to the feed tank (Fig. 1). Before starting of the CD process, the nanofiltration was operated in total recycle mode for more than 30 min to reach the stabilized state. The kinetics of continuous production of integrated CD over 96 h are shown in Fig. 10A and B. Over this period, a total of 5.36 kg lactose was consumed, and 2.01 kg of GOS, 1.37 kg of glucose, and 0.75 kg of galactose were produced; 1.28 kg of lactose was not utilized (Fig. 10A). The feed tank and the permeate containing the mixed sugars were collected and analyzed after each collecting cycle of 2 L, and the total sugar production was estimated based on the sugars in the UMR, feed tank, and permeate. The continuous production in the UMR was maintained at relatively stable GOS production, even when the GOS concentration in the reactor mixture fluctuated between 80 and 100 g/L (Fig. 10B); this indicates a dynamic equilibrium between GOS synthesis and removal. By adding the demineralized water, the concentration of the total sugar at the beginning of each collecting cycle can be maintained between 40–60 g/L (the permeate fluxes of UF and NF are approximately 6–8 and 15–18 L/(m2 h), respectively). When the GOS purity is higher than 57% and the purity of lactose is less than 20%, the mixed sugars were removed from the feed tank yielding 1.24 kg of high-purity GOS (GOS purity is 57–59%, lactose content is less than 20%) at the end of the CPNSS. Consistent results were obtained at 96 h, indicating that the continuous production process was stable and sustainable for long-term operation without significant enzyme activity

Table 4 Comparison of continuous production without nanofiltration separation, nanofiltration separation without continuous production, and CPNSS. N/A

Average flux of UF (L/m2/h)

Continuous production without nanofiltration separation Nanofiltration separation without continuous production CPNSS

6.68

Average flux of NF (L/m2/h)

Average lactose conversion (%)

Achieved GOS purity (%)

Monosaccharide content (%)

75.23

33.4

30.2

55.3

57.2

16.1

6.67

15.3

75.68

GOS Yield (%)

Achieved high-purity GOS spend time (h)

19

77.4

3.7

17

80.1

3.2

H. Ren et al. / Separation and Purification Technology 144 (2015) 70–79

degradation. This was comparable to results observed using other enzymatic means [17,18,20]. In this study, 30.2 wt.% monosaccharide, 36.4 wt.% lactose, and 33.4 wt.% GOS were obtained in the UMF. Our results show that 92.3% monosaccharide and 53.6% lactose in the mixture were removed with the NF membrane, and the average GOS purity of 57.2% was achieved (1.71 times of the raw material), which correlates to a yield of 80.1%. Compared to continuous production without nanofiltration, CPNSS achieved higher purities of GOS while more monosaccharide and disaccharide were removed (Table 4). Compared to nanofiltration separation without continuous production, less GOS were loss, and less time and dilution water was spent to obtain high-purity GOS (Table 4). 4. Conclusions This study demonstrates the feasibility of continuous production of high-purity GOS in CPNSS. The CPNSS is an efficient method to continuously synthesis GOS and reduce the time of obtaining high-purity GOS. With periodic b-galactosidases supplementation, the CPNSS maintained a stable productivity and high GOS yield for an extended period. Hence, this process is attractive for industrial production of GOS. Moreover, using the CPNSS could produce high-purity GOS that can meet industry standards. Acknowledgments This work was supported by the National Outstanding Youth Foundation of China (21025625), the National High-Tech Research and Development Program of China (863) (2012AA021203), the National Basic Research Program of China (973) (2013CB733602), the Major Research Plan of the National Natural Science Foundation of China (21390204), The National Technology Support Program (2012BAI44G01), the National Natural Science Foundation of China, General Program (21376118), the National Natural Science Foundation of China, Youth Program (21106070), Jiangsu Provincial Natural Science Foundation of China (SBK 201150207), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). References [1] E. Betoret, N. Betoret, D. Vidal, P. Fito, Functional foods development: trends and technologies, Trends Food Sci. Technol. 22 (9) (2011) 498–508. [2] S. Salminen, C. Bouley, M.C. Boutron, J.H. Cummings, A. Franck, G.R. Gibson, E. Isolauri, M.C. Moreau, M. Roberfroid, I. Rowland, Functional food science and gastrointestinal physiology and function, Brit. J. Nutr. 80 (S1) (1998) S147– S171. [3] L.J. Fooks, R. Fuller, G.R. Gibson, Prebiotics, probiotics and human gut microbiology, Int. Dairy J. 9 (1) (1999) 53–61. [4] R.G. Gibson, M.B. Roberfroid, Dietary modulation of the human colonic microbiota: introducing the concept of prebiotics, J. Nutr. 125 (1995) 1401– 1412. [5] D. Granato, G.F. Branco, F. Nazzaro, A.G. Cruz, J.A. Faria, Functional foods and nondairy probiotic food development: trends, concepts, and products, Compr. Rev. Food Sci. F. 9 (3) (2010) 292–302. [6] G.R. Gibson, P.B. Ottaway, R.A. Rastall, Prebiotics: New Developments in Functional Foods, Chandos, Oxford, UK, 2000. [7] M.B. Roberfroid, Prebiotics and probiotics: are they functional foods?, Am J. Clin. Nutr. 71 (6) (2000) 1682s–1687s. [8] H.C. Schoterman, Galacto-oligosaccharides: properties and health aspects, Adv. Dietary Fibre Technol. 42 (2001) 494–502. [9] H. Tomomatsu, Health effects of oligosaccharides, Food Technol. 48 (10) (1994) 61–65. [10] T. Sako, K. Matsumoto, R. Tanaka, Recent progress on research and applications of nondigestible galacto-oligosaccharides, Int. Dairy J. 9 (1) (1999) 69–80. [11] M.A. Boon, A.E.M. Janssen, K. van’t Riet, Effect of temperature and enzyme origin on the enzymatic synthesis of oligosaccharides, Enzyme Microbial Technol. 26 (2) (2000) 271–281.

79

[12] T. Murata, T. Usui, Enzymatic synthesis of oligosaccharides and neoglycoconjugates, Biosci. Biotechnol. Biochem. 70 (5) (2006) 1049–1059. [13] B.A. Rabiu, A.J. Jay, G.R. Gibson, R.A. Rastall, Synthesis and fermentation properties of novel galacto-oligosaccharides by b-galactosidases from Bifidobacterium species, Appl. Environ. Microbiol. 67 (6) (2000) 2526–2530. [14] S.I. Mussatto, I.M. Mancilha, Non-digestible oligosaccharides: a review, Carbohyd. Polym. 68 (3) (2007) 587–597. [15] S. Chockchaisawasdee, V.I. Athanasopoulos, K. Niranjan, R.A. Rastall, Synthesis of galacto-oligosaccharide from lactose using b-galactosidase from Kluyveromyces lactis: studies on batch and continuous UF membrane-fitted bioreactors, Biotechnol. Bioeng. 89 (4) (2005) 434–443. [16] D.P. Torres, M.D.P.F. Gonçalves, J.A. Teixeira, L.R. Rodrigues, Galactooligosaccharides: production, properties, applications, and significance as prebiotics, Compr. Rev. Food Sci. F. 9 (5) (2010) 438–454. [17] J.E. Prenosil, E. Stuker, J.R. Bourne, Formation of oligosaccharides during enzymatic lactose hydrolysis and their importance in a whey hydrolysis process: Part II: experimental, Biotechnol. Bioeng. 30 (9) (1987) 1026–1031. [18] C.J. Zhou, X.D. Chen, Effects of temperature and pH on the catalytic activity of the immobilized b-galactosidase from Kluyveromyces lactis, Biochem. Eng. J. 9 (1) (2001) 33–40. [19] Q.Z. Zhou, X.D. Chen, X. Li, Kinetics of lactose hydrolysis by b-galactosidase from Kluyveromyces lactis immobilized on cotton fabric, Biotechnol. Bioeng. 81 (2) (2003) 127–133. [20] D.F. Neri, V.M. Balcão, M.G. Carneiro-da-Cunha, L.B. Carvalho Jr., J.A. Teixeira, Immobilization of b-galactosidase from Kluyveromyces lactis onto a polysiloxane–polyvinyl alcohol magnetic (mPOS–PVA) composite for lactose hydrolysis, Catal. Commun. 9 (14) (2008) 2334–2339. [21] M.I. Foda, M. Lo´pez-Leiva, Continuous production of oligosaccharides from whey using a membrane reactor, Process Biochem. 35 (6) (2000) 581–587. [22] L. Engel, M. Ebrahimi, P. Czermak, Membrane chromatography reactor system for the continuous synthesis of galactosyl-oligosaccharides, Desalination 224 (1) (2008) 46–51. [23] R.D. Appel, A. Bairoch, D.F. Hochstrasser, A new generation of information retrieval tools for biologists: the example of the ExPASy WWW server, Trends Biochem. Sci. 19 (6) (1994) 258–260. [24] M. Cheryan, Ultrafiltration and Microfiltration Handbook, CRC Press, Boca Raton, Florida, 1998. [25] P.Y. Pontalier, A. Ismail, M. Ghoul, Specific model for nanofiltration, J. Food Eng. 40 (3) (1999) 145–151. [26] A.S. Grandison, Separation processes in the food and biotechnology industries, CRC Press, 1996. [27] J. Kilduff, W.J. Weber Jr, Transport and separation of organic macromolecules in ultrafiltration processes, Environ. Sci. Technol. 26 (3) (1992) 569–577. [28] W.Y. Li, J.D. Li, T.Q. Chen, C.X. Chen, Study on nanofiltration for purifying fructo-oligosaccharides, I. Operation modes, J. Membr. Sci. 245 (1) (2004) 123– 129. [29] N.J. Kruger, The Bradford method for protein quantitation, Basic Protein and Peptide Protocols, vol. 32, Humana Press, Totowa, NJ, 1994, pp. 9–15. [30] A. Gosling, G.W. Stevens, A.R. Barber, S.E. Kentish, S.L. Gras, Effect of the substrate concentration and water activity on the yield and rate of the transfer reaction of b-galactosidase from Bacillus circulans, J. Agr. Food Chem. 59 (7) (2011) 3366–3372. [31] V.A. Botelho-Cunha, M. Mateus, J.C. Petrus, M.N. vde Pinho, Tailoring the enzymatic synthesis and nanofiltration fractionation of galactooligosaccharides, Biochem. Eng. J. 50 (1) (2010) 29–36. [32] S.S. Kulkarni, E.W. Funk, N.N. Li, Ultrafiltration: membranes, in: Membrane Handbook, Van Nostrand Reinhold, New York, 1992, pp. 408–431. [33] A.K. Goulas, P.G. Kapasakalidis, H.R. Sinclair, R.A. Rastall, A.S. Grandison, Purification of oligosaccharides by nanofiltration, J. Membr Sci. 209 (1) (2002) 321–335. [34] Y.M. Feng, X.L. Chang, W.H. Wang, R.Y. Ma, Separation of galactooligosaccharides mixture by nanofiltration, J. Taiwan Inst. Chem. Eng. 40 (3) (2009) 326–332. [35] N. Aydogan, T. Gurkan, L. Yilmaz, Effect of operating parameters on the separation of sugars by nanofiltration, Sep. Sci. Technol. 33 (1998) 1767–1785. [36] L. Xu, S.Y. Wang, X.Y. Zeng, The maltitol purification and concentration by nanofiltration, Desalination 184 (1) (2005) 295–303. [37] W.R. Bowen, J.S. Welfoot, Modelling of membrane nanofiltration—pore size distribution effects, Chem. Eng. Sci. 57 (8) (2002) 1393–1407. [38] A.S. Grandison, M.J. Lewis, Separation Processes in the Food and Biotechnology Industries, Woodhead Publishing Ltd, Cambridge, 1996. [39] A. Goulas, G. Tzortzis, G.R. Gibson, Development of a process for the production and purification of a- and b-galactooligosaccharides from Bifidobacterium bifidum NCIMB 41171, Int. Dairy J. 17 (6) (2007) 648–656. [40] A.K. Goulas, A.S. Grandison, R.A. Rastall, Fractionation of oligosaccharides by nanofiltration, J. Sci. Food. Agric. 83 (7) (2003) 675–680. [41] J.E. Prenosil, E. Stuker, J.R. Bourne, Formation of oligosaccharides during enzymatic lactose. Part I. State of art, Biotechnol. Bioeng. 30 (9) (1987) 1019– 1025.