Sunteți pe pagina 1din 7
Journal of Biotechnology 119 (2005) 212–218 A new innovative process to produce lactose-reduced skim milk
Journal of Biotechnology 119 (2005) 212–218 A new innovative process to produce lactose-reduced skim milk

Journal of Biotechnology 119 (2005) 212–218

Journal of Biotechnology 119 (2005) 212–218 A new innovative process to produce lactose-reduced skim milk Senad

A new innovative process to produce lactose-reduced skim milk

Senad Novalin a , Winfried Neuhaus b, , Klaus D. Kulbe a

a Division of Food Biotechnology, Department of Food Sciences and Technology, University of Natural Resources and Applied Life Sciences, Muthgasse 18, A-1190 Vienna, Austria b Department of Medicinal/Pharmaceutical Chemistry, University of Vienna, Pharmacy Centre, Althanstrasse 14, A-1090 Vienna, Austria

Received 17 June 2004; received in revised form 29 March 2005; accepted 31 March 2005


The research field for applications for lactose hydrolysis has been investigated for some decades. Lactose intolerance, improve- ment for technical processing of solutions containing lactose and utilisation of lactose in whey are main topics in development of biotechnological processes. In this article, the establishment of a hollow fiber membrane reactor process for enzymatic lactose hydrolysis is reported. Mesophilic -galactosidases were circulated abluminally during luminal flow of skim milk. The main problem, microorganisms growth in the enzyme solution, was minimised by sterile filtration and UV irradiation. In order to characterise the process parameters, such as skim milk concentration, enzyme activity and flow rates were varied. In comparison to a batch process, enzyme activity could be used longer and enzyme rest into the product should not occur. Furthermore, the three-dimensional separation of the substrate from the enzyme solution minimise blocking and washing out effects, which restrict processes with immobilised enzymes. A conversion rate of 78.11% was achieved at a skim milk flow rate of 9.9 l h 1 , enzyme activity of 120 U ml 1 and a temperature of 23 ± 2 C in a hollow fiber reactor with a membrane area of 4.9 m 2 . © 2005 Elsevier B.V. All rights reserved.

Keywords: Enzyme technology; Lactose hydrolysis; -Galactosidase; Diffusional reactor

1. Introduction

Every year 3.2 million tonnes of lactose, dissolved in whey, is accrued by the cheese production world-wide (Ruttloff, 1994). Almost half of this amount is used for human and animal nutrition. The rest is waste that is

Corresponding author. Tel.: +43 1 4277 55089; fax: +43 1 4277 9551. E-mail address: (W. Neuhaus).

0168-1656/$ – see front matter © 2005 Elsevier B.V. All rights reserved.


difficult to dispose of and adds to the environmental pollution. Therefore, there is a need for investigation about further utilisation possibilities of lactose from whey. One of these applications with a high technolog- ical and dietetic interest is the enzymatic hydrolysis of lactose, whose economic importance has been increas- ing ever since the 1960s. Next to the medical aspect of lactose intolerance, some very important technological advantages result from the lactose hydrolysis into glucose and galactose.

S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218


For example, the solubility increases from 18 to 55% (w/v) at 80% conversion and the sweetness rises up to 70% related to sucrose. Furthermore, there is a lower- ing of the freezing point, an increase of the probability for non-enzymatic browning reactions and a faster fermentation process in lactose hydrolysed medium. Thus, the production of self-sweetening products or products with less sucrose addition would be possible by using lactose hydrolysed milk or whey. Also, posi- tive effects on the crystallisation and process properties would be achieved after lactose hydrolysis (Zadow,


In general, there are several technologies for enzy- matic hydrolysis of lactose (Pivarnik et al., 1995; Mahoney, 1985; Gekas and Lopez-Leiva, 1985). The easiest way is the discontinuous batch-process. After reaching the aimed conversion, the reaction is stopped by heating, which causes enzyme denaturation and con- sequently the loss of enzymatic activity. Furthermore, the enzymes become product components after the pro- cess. The immobilisation can be employed to use the enzyme’s activity for as long as possible. It can be affected by physical or chemical binding on a solid matrix like glass surfaces, cellulose acetate or oxiran- gel (Richmond et al., 1981). High cost of the immo- bilising steps, the activity loss during immobilisation and the occurrence of hygienic problems because of the fat and protein content in milk and whey (Reimerdes, 1985) has a detrimental effect on the decision to choose this process. The third possibility is the “physical immobilisa- tion” by separating enzyme solution from the substrate flow via ultrafiltration membranes (Czermak et al., 1988). This system causes workable and cheap enzyme fixation with little loss of the catalytic activity of the enzyme. Further advantages of this process are the con- tinuous operation of the reactor at low pressure and the selectivity control by selection of suitable membranes. High enzyme concentration application, easy replace- ment and regeneration of spent enzyme solution and little loss of enzymes due to wash out effects are possi- ble. A drawback is, that the diffusion resistance of the hollow fiber membrane seems to be the limiting factor at high conversion rates. Ultrafiltration modules were constructed as steam sterilisable plastic membranes by Czermak et al. (1990) and Czermak (1992), whose development and produc-

tion are too expensive for reaching the economic aim to be at least as cheap as the batch process. Thus, it is necessary to find another solution to control the micro- biological growth in the system.

2. Material and methods

2.1. Functional principles of the bioreactor system

An ultrafiltration unit (called module) consists of a bundle of small hollow fibers. The outside (shell side) compartment is constructed as a closed circulation and is filled with the enzyme solution. A continuous flow of the substrate (skim milk) is applied to the inside (tube side) of the hollow fiber membranes. The spatial enzyme separation from the substrate solution is guar- anteed by the selected membrane cut-off value (Fig. 1). In this way, continuous, enzymatic lactose hydrol- ysis is possible without inherent problems of immo- bilised enzymes. The driving force for this system is the lactose diffusion, which mainly depends on the con- centration gradient, the temperature and the flow rates of the substrate and the enzyme solution. Fig. 2 shows the standard flow sheet of the newly developed plant. The reactor is a hollow fiber module (43 in. module, 10 kDa cut-off and 4.9 m 2 membrane area), which consists of a bundle of capillaries made of polysulfone membrane. The shell side volume is about 2.5 l and the tube side volume is about 0.65 l. Skim milk is pumped through the hollow fiber module. Before enzymatic conversion takes place in the hollow fiber module, skim milk passes a heat exchanger, a manome- ter and a thermometer. After the end of the module, the lactose-hydrolysed product can be collected. The enzyme solution is pumped in a closed circu- lation. Temperature of enzyme solution was adjusted with a waterbath located before the hollow fiber’s shell side. Czermak and Bauer (1990) have constructed ultra- filtration modules as steam sterilisable plastic mem- branes. Development and production costs seemed to be too high for commercial use compared to the batch process. Thus, it is necessary to find another solution to control the microbiological growth in the system. In order to sufficiently reduce germs presence a UV irradi- ation module and a sterile filtration unit were included in the enzyme circulation. To our best knowledge, the use of a commercially available UV-unit (Visa, Austria)


S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218

et al. / Journal of Biotechnology 119 (2005) 212–218 Fig. 1. Principle of lactose hydrolysis in

Fig. 1. Principle of lactose hydrolysis in the hollow fiber module by -galactosidase denoted as -gal in the scheme. While skim milk is pumped through the hollow fiber, molecules with lower size than the membrane cut-off value, such as small proteins, salts and lactose pass through into the shell side. There, lactose is enzymatically converted to glucose, galactose and a small amount of oligosaccharides by - galactosidase. The product transfer back into the tube side is also effected by concentration gradient driven diffusion.

is a novelty in the enzyme technology field. In order to investigate the effects of UV irradiation on enzyme activity and germ number, enzymes were dissolved in milk buffer and pumped into a closed circulation through the UV module at a flow rate of 25 l h 1 using different UV intensity levels. Enzyme activity and total germ number (carried out by Koch plating process with standard count agar from Merck, Germany) were deter- mined from samples taken at several time points.

2.2. Material

Apparatus: Hollow fiber module RomiPro 5 in. × 43 in. PM 10 was from Koch (MA, USA), sterile filter FG-50 and FG-30 were purchased from Millipore (MA, USA) and the sterile filtration

module type MD 020 GP 2N was obtained from Microdyn (Germany). Chemicals: o-Nitrophenyl- -d-galactopyranoside (ONPG), o-nitrophenole (ONP) and d-glucose were purchased from Sigma (MO, USA), lactose was from Merck (Germany). Milk buffer described by Novo Nordisk (1977) should imitate the milk’s salt system and was used in enzyme experiments, when milk had to be excluded. For skim milk “medium heat extra skim milk powder” was dissolved in bidistilled water. The powder was produced by Lactoprot (Austria) and consisted of following components: 35% protein, 1.25% fat, 51% lactose, 8% ashes and less than 4% water. β-Galactosidases: The systematic name for this enzyme is -d-galactoside-galactohydrolases and its EC-number is Maxilact L 2000 and Maxilact LX 5000 from Kluyveromyces lactis were obtained from Gist-brocades (Netherlands). They catalyse the lactose hydrolysis.

2.3. Analytics

2.3.1. Determination of enzyme activity

This method was described by Petzelbauer et al. (1999) for thermophilic -galactosidase and was used with slight modification for mesophilic variants. ONPG was utilised as a substrate and the measurements of enzyme activity were carried out at a constant concen- tration of ONPG. By adding 20 l of the enzyme solu- tion to 980 l of the substrate solution (860 l 50 mM sodium phosphate buffer, pH 6.5 and 120 l 25 mM ONPG diluted in 50 mM sodium phosphate buffer, pH 6.5) the hydrolysis was initiated. The release of ONP caused an increase in the absorbance at 405 nm, which was measured for 5 min at 25 C.

One unit of ONPG activity is defined as 1 mol released ONP per minute under the reaction condi- tions described above. Whenever it was necessary, the enzyme solution was diluted in milk buffer.

2.3.2. Sugar analysis

Quantitative glucose determination was accom- plished with an Ebio glucose analyser from Eppendorf (Germany). The measurement range lies between 0.5 and 50 mM glucose. Consequently, the milk samples were diluted with bidistilled water in order to prepare suitable concentrations.

S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218


et al. / Journal of Biotechnology 119 (2005) 212–218 215 Fig. 2. Scheme of hollow fiber

Fig. 2. Scheme of hollow fiber reactor for lactose hydrolysis by -galactosidase: (1) hollow fiber module, (2) UV module, (3) sterile filtration module, (4) heat exchanger, (5) peristaltic pump, (6) heating bath, (7) storage tank for the substrate, (8) tank for enzyme solution, (9) tank for sterile filtration circulation, (10) sampling port, (11) three-way cock, (12) waste line and (13) product collecting.

Samples from the enzyme solution in the reactor’s shell-side were incubated at 100 C for 2 min to pre- vent interference of -galactosidase. Afterwards, the hot solution was centrifuged at 10,000 rpm for 5 min. Twenty microlitres of appropriately diluted supernatant was mixed with the Ebio buffer solution before the mea- surement. For one-point-calibration a 12 mM standard glucose solution from Eppendorf was used. HPLC was used to quantify lactose, galactose and glucose concentrations. Samples had to be precipi- tated by Carrez clarification with 85 mM K 4 [Fe(CN) 6 ] and 250 mM ZnSO 4 ·7 H 2 O followed by centrifu- gation at 11,000 rpm for 10 min. Concentrations of the supernatants were measured with a HPLC-system from Merck Hitachi (HPX 87-C-column from Aminex, 85 C, 0.7 ml min 1 flow rate, bidistilled water as elu- ent, ERC 7512 IR-detector from Erma Cr Inc., USA). External standard solutions of 10 g l 1 lactose, galac- tose and glucose were used for calculation.

3. Results and discussion

The aim of the present work was to establish an alter- native membrane bioreactor process for hydrolysing

lactose in complex media like skim milk. We focused on the development of a cheap and easy way to enhance long-term stability of both the process and enzyme’s activity. The high costs for a steam sterilisable system would not allow to reach our aim to develop a process with at least comparable costs to a batch process. In order to find another way for sufficient germ reduction in the enzyme solution, a combination of sterile filtra- tion and UV irradiation was developed. Fig. 3 shows the results of an experiment, where an UV intensity of 45% of a 25 W lamp has led to both a satisfying germ reduction and enzyme stabilisation at the same time. After 90 min, germ number was sufficiently reduced from 10 4 to 10 2 germs ml 1 and only 10% less enzyme activity was detected compared to the blank solution. Blank solution was supplemented with 0.02% NaN 3 in order to suppress microbiological growth. Enzyme stability is one reason to favour packed-bed reactors instead of hollow fiber reactors (Pivarnik et al., 1995). One main cause for loss of enzyme activity in hollow fiber reactors is the absorption of the enzymes by the membranes. In order to avoid high enzyme activ- ity loss by this, pre-absorption processes with protein solutions in the shell side were carried out. The cheap-


S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218

et al. / Journal of Biotechnology 119 (2005) 212–218 Fig. 3. UV experiment at 45% irradiation

Fig. 3. UV experiment at 45% irradiation intensity of 25 W UV lamp situated in a UV module. Enzyme solution (120 U ml 1 Maxilact L 2000, milk puffer pH 6.5, 106 mM lactose) was circulated at 25 l h 1 . Relative enzyme activity ( ), relative enzyme activity of a blank solution ( ) which was supplemented with 0.02% (w/v) NaN 3 and total germ numbers ( ) are depicted vs. time.

est and most practical way to cover the membrane surface with proteins was to fill the shell side (enzyme circulation) by ultrafiltration of skim milk. This was followed by circulation of this whey for 1 h before enzyme solution was added into the shell side. For these experiments, commercially available enzymes, Maxi- lact (Gist Brocades) and Lactozym (Novo Nordirsk, Denmark) were chosen as suitable enzymes due to their optimum temperature between 35 and 45 C and pH value about 6.5–6.8. Several tests showed that Maxi- lact was more stable and did not lose as much activity like Lactozym. Thus, Maxilact was used for the further studies. In order to develop a cost-effective process we aimed at gaining 75% lactose hydrolysis in skim milk at 10 l h 1 (360 g l 1 h 1 productivity) flow rate. Czermak et al. (1988) showed that this kind of pro- cess is mainly limited by lactose diffusion and not by enzyme concentration. Accordingly, dependence of lactose hydrolysis on enzyme concentration was inves- tigated (Fig. 4). Skim milk was supplemented with 0.02% NaN 3 and was pumped with 10 ± 0.5 l h 1 . Enzyme activity was adjusted by dosages to the enzyme tank. It is very important to recognise that the enzyme activity of 100 U ml 1 was still a limiting factor for the conversion rate. Further studies resulted in deter- mining optimum enzyme concentration at 120 U ml 1 . Adding more enzyme led to a minimal increase of lac- tose hydrolyis. Therefore, all further experiments were carried out with 120 U ml 1 Maxilact.

were carried out with 120 U ml − 1 Maxilact. Fig. 4. Dependence of lactose conversion

Fig. 4. Dependence of lactose conversion (%) on enzyme activity (U ml 1 ) at 10 ± 0.5 l h 1 skim milk flow, 0.02% (w/v) NaN 3 , lactose conversion (%) means the formed glucose related to lactose, which is determined by HPLC, in skim milk ( ) and enzyme solution ( ).

In order to optimise and characterise the process, parameters like skim milk flow rate and concentration were varied. At the beginning of every experiment,

a steady-state of the diffusing substances had to be

set between the enzyme solution and the skim milk. After steady-state was reached, constant measure val- ues and production were possible. Using an experimen- tal design where skim milk (supplemented with 0.02% NaN 3 ) was pumped with a flow rate of 10.5 l h 1 and the enzyme solution (120 U ml 1 , Maxilact L 2000) circulated at 25 l h 1 , measured lactose conversion in skim milk levelled at 81% after 2 h. Steady-state in the enzyme solution was reached after 3 h at 96%. Inter- estingly, after 90 min measured lactose conversion in skim milk was still higher than in the enzyme solution. Experiments with constant enzyme activity and var- ied skim milk flow rates were also important for under- standing this process. Fig. 5 summarises the influence of varied skim milk flow rates and skim milk concen- trations on lactose conversion rate at same enzyme activity (120 U ml 1 ). Firstly, the correlation between skim milk flow rate and lactose conversion was investi- gated. Lactose conversion in skim milk decreased from

92.48% at 5.04 l h 1 to 82.95% at 7.83 l h 1 and further

to 78.11% at 9.9 l h 1 , conversion rates determined in

enzyme solution were always higher at the same flow rate. In order to improve process productivity, several experiments with 1.5:1 and 2:1 concentrated skim milk were accomplished. Although higher blocking effects

S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218


et al. / Journal of Biotechnology 119 (2005) 212–218 217 Fig. 5. Influence of skim milk

Fig. 5. Influence of skim milk flow rate and concentration on lac- tose conversion (%). All experiments were carried out at 23 ± 2 C, 120 U ml 1 and 25 l h 1 enzyme solution flow rate. Lactose conver- sion means the formed glucose related to lactose, which is determined by HPLC, in skim milk ( ) and enzyme solution ( ) Maxilaxt L 2000. Skim milk was supplemented with 0.02% (w/v) NaN 3 . Also flow rate dependent lactose conversion (%) in 1.5:1 concentrated skim milk ( ) and enzyme solution ( ) Maxilact LX 5000, instead of NaN 3 45% UV irradiation 10 min h 1 and continuous microfiltra- tion and in 2:1 concentrated skim milk ( ) and enzyme solution ( ) Maxilact LX 5000, instead of NaN 3 45% UV irradiation 10 min h 1 and continuous microfiltration are depicted.

on the membranes were expected, higher substance concentration should result in increased diffusion rates and consequently in enhanced productivity. For the first test, 1.5:1 concentrated skim milk was not supplemented with NaN 3 . The enzyme solution (120 U ml 1 Maxilact LX 5000) was treated for 10 min for every hour with UV irradiation at 45% intensity and was continuously sterile filtered by microfiltration. Lactose conversion in skim milk decreased as expected depending on flow rate of substrate from 80.09% at 6.5 l h 1 to 72.68% at 8.6 l h 1 and further to 68.54% at 10.4 l h 1 , conversion rates in enzyme solution were always higher at the same flow rate. For the next test, 2:1 concentrated skim milk was also not supple- mented with NaN 3 . The enzyme solution (120 U ml 1 Maxilact LX 5000) was treated for 10 min for every hour with UV irradiation at 45% intensity and was continuously sterile filtered by microfiltration. Lactose conversion rate in skim milk decreased also depending on substrate flow rate. In general, productivity (hydrol- ysed lactose in g l 1 h 1 ) increased at higher flow rates and higher skim milk concentrations. Five hundred and

nineteen grams per litre per hour lactose conversion can be achieved with two-fold concentrated skim milk at 8.46 l h 1 flow rate and 63.99% lactose hydrolysis in contrast to 311.76 g l 1 h 1 for 1:1 skim milk at a similar flow rate. If highest productivity is the aim for the process, two-fold concentrated skim milk should be preferred, if long-term stability under these conditions is proved. Using high concentrated skim milk includes the danger of blocking effects due to fat or protein adsorption. This might harm the cartridges leading to cost explosion. Also, higher concentration of skim milk means more nourishment for microorganisms in the enzyme circulation and consequently faster growth and loss of enzyme activity due to digestion. All experiments lasted between 6 and 8 h. Future work will carry out long-term experiments over several days. Finally, two different procedures were compared to each other in order to establish the most suitable one for maximum lactose conversion. Under the same con- ditions, it was investigated whether lactose conversion rate would be higher at a flow of 5 or at 10 l h 1 , if the same amount is pumped through the reactor a second time. The idea was that the procedure with the higher flow rate would cause a higher concentration gradient and consequently higher diffusion rates. At a flow rate of 5 l h 1 a lactose conversion rate of 92.48% in skim milk and 97.08% in enzyme solution was achieved. Whereas running the second procedure (at a flow rate of 10 l h 1 , collecting the product and pumping it a second time through the hollow fiber apparatus) it was possible to increase lactose conversion rate minimally up to 94.35% in skim milk and to 99.14% in enzyme solution. This procedure may lead to a small improve- ment of the economic efficiency.

4. Conclusions

The major conclusions of this study are as follows:

It is well known that a membrane-diffusion reactor is applicable for the hydrolysis of lactose in whey or milk using enzymes. Because of the relatively low mass transfer, large membrane areas are nec- essary, which therefore, makes it difficult to reach economic goals. In the actual case, a commercially membrane module was used overcoming this first


S. Novalin et al. / Journal of Biotechnology 119 (2005) 212–218

problem. Results show high conversion rates at high flow rates and consequently high productivity. The second problem lies in the growth of microor- ganism within the enzyme solution (the shell side of the module). Two different processes were applied in order to minimise this issue: microfiltration and UV irradiation of the enzyme solution. As the experi- ments have shown, this technological idea could lead to a definite solution of this second problem.

In any case, the applicability of membrane-diffusion reactors needs to be tested commercially. We were able to introduce a process for direct lactose hydrol- ysis in skim milk without any ultrafiltration step before enzymatic conversion. Thus, our system can easily be connected directly with milk storage tanks in dairy industry as an inline installation. Compared to other published data (Splechtna et al., 2002) productivity seems to be much higher. Also it should be possi- ble to use other substrates for this process like whey or high concentrated whey. Future investigations will focus on optimising this process considering temper- ature influence on long-term stability and conversion performance. Furthermore, we will try to develop a mathematical possibility to predict lactose conversion rate depending on lactose concentration and flow rate for this system.


We gratefully acknowledge financial support from the European Commission under Contract EU CT 96-



Czermak, P., 1992. Scale up von Enzym-Membranreaktoren am Beispiel der kontinuierlichen Lactosehydrolyse. Bioengineering 1/92 8.Jg., Lebensmitteltechnologie, 47–52.

Czermak, P., Bauer, W., 1990. Optimization of the continuous hydrolysis of lactose in the dialysis enzyme membrane reactor. DECHEMA Biotechnol. Conf. 4 (B), 763–766. Czermak, P., Bahr, D., Bauer, W., 1990. Verfahrenstechnische Opti- mierung der kontiniuerlichen enzymatischen Laktose-Hydrolyse im Dialyse-Membranreaktor. Chem.-Ing.-Tech. 62, 678–679. Czermak, P., Eberhard, G., Konig,¨ A., Tretzel, J., Reimerdes, E.H., Bauer, W., 1988. Dialysis membrane reactors for enzymatic conversions in biotechnical processes: functional principles and examples for application. In: Behrens, D. (Ed.), DECHEMA. Biotechnol. Conf. 2, VCH Verlagsgesellschaft, Weinheim, New York, pp. 133–145. Gekas, V., Lopez-Leiva, M., 1985. Hydrolysis of lactose: a literature review. Process Biochem. 20, 2–12. Mahoney, R.R., 1985. Modification of lactose and lactose-containing dairy products with -galactosidase. In: Fox, P.F. (Ed.), Develop- ments in Dairy Chemistry-3. Elsevier Applied Science Publishers Ltd., Amsterdam, pp. 69–108. Novo Nordisk, 1977. Determination of K. fragilis lactase. Novo Industri AS, Analytical Method Nr. AF 171/1-GB-a. Petzelbauer, I., Nidetzky, B., Haltrich, D., Kulbe, K.D., 1999. Devel- opment of an ultra-high-temperature process for the enzymatic hydrolysis of lactose. I. The properties of two thermostable beta- glycosidases. Biotechnol. Bioeng. 64 (3), 322–332. Pivarnik, L.F., Senecal, A.G., Rand, A.G., 1995. Hydrolytic and transgalactosylic activities of commercial -galactosidase (lac- tase) in food processing. In: Kinsella, J.E., Taylor, D.L. (Eds.), Advances in Food and Nutrition: Research, vol. 3. Academic, NY, pp. 1–101. Reimerdes, E.H., 1985. Entwicklung von Enzymreaktoren zur Lact asebehandlung von Milch- und Folgeprodukten. Abschlußbericht BM f. Forschung und Technologie, Reference number 038491, Follow up report PTB 0382628, Bonn, 3–63. Richmond, M.L., Gray, J.I., Stine, C.M., 1981. -Galactosidase:

review of recent research, related to technological applica- tion, nutritional concerns and immobilization. J. Dairy Sci. 64,


Ruttloff, H., 1994. Lactase. Industrielle Enzyme, second ed. Behrs Verlag, pp. 766–777. Splechtna, B., Petzelbauer, I., Kuhn, B., Kulbe, K.D., Nidetzky, B., 2002. Hydrolysis of lactose by beta-glycosidase CelB from hyperthermophilic archaeon Pyrococcus furiosus: comparison of hollow-fiber membrane and packed-bed immobilized enzyme reactors for continuous processing of ultrahigh temperature- treated skim milk. Appl. Biochem. Biotechnol. 98–100, 473–488. Zadow, J.G., 1992. In: Zadow, J.G. (Ed.), Lactose Hydrolysis in Whey and Lactose Processing. Elsevier Science Publishers Ltd., Amsterdam, pp. 2–21.