DESALINATION

ELSEVIER Desalination 149 (2002) 279-285

www.elsevier.com/locate/desal
Enzymatic hollow fiber membrane bioreactor
for penicilin hydrolysis
I.G. Wenten*, I.N. Widiasa
Department of Chemical Engineering, Institut Teknologi Bandung, Jalan Ganesha 10 Bandung, Indonesia, 40132
email: igw@che.itb.ac.id
Received 4 March 2002; accepted 6 April 2002
Abstract
Continuous enzymatic reaction has been proven as an efficient technique for several industrial applications. In
this study, a type of hollow fiber membrane bioreactor where penicillin acylase entrapped within membrane pores
was applied to continuously hydrolyze Penicillin G The influences of various operating conditions on immobilization
and enzymatic reaction processes were assessed. Amathematical model of the reactor behaviour at steady state con-
dition was also developed. The immobilization results show that penicillin acylase was entrapped more than 90%
(100,000 u.a m-?). Due to the much smaller size of 6-APA compared to the membrane pore, the solute diffuses freely
through the membrane. However, the immobilized enzyme membrane retained around 35% of the solute. In addition,
K,,, of immobilized penicillin acylase (8.04 mM) was slightly higher than that of free penicillin acylase (7.75 mM).
The theoretical results indicated that convective transport was the main mechanism of mass transport even in the
case where flux was very low. Low flux rate is important to avoid gel formation or enzyme release from membrane
pores and to maximize the degree of conversion.
Keywords: Hollow fiber membrane bioreactor; Penicillin hydrolysis; Penicillin acylase
1. Introduction
In most enzymatic reactions, batch or con-
tinuous, the recycle of enzyme in a reactor system
is important to reduce production cost. Enzyme
membrane reactor, in which an enzyme reactor
coupled to ultrafiltration or dialysis membrane
with a suitable molecular weight cut-off, is capable
to keep enzyme and other larger components, while
low-molecular-weight-molecules, e.g. products
and/or inhibitor, are allowed to pass freely through
the membrane. This reactor has been used in wide
*Corresponding author.
range of applications, i.e. saccharification of
Presented at the International Congress on Membranes and Membrane Processes (ICOM), Toulouse, France,
Julv 7-12, 2002.
0011-9 164/02/$- See front matter 0 2002 Elsevier Science B.V. All rights reserved
PII:SOOl l-9164(02)00789-0
280 1.G Wenten. I.N. Widiasa /Desalination 149 (2002) 279-285
cassava flour starch [l], protein hydrolysis [2,3],
regeneration of ATP [4], production of aspartic
acid [5], processing of natural substrates [6], etc.
The main problem appears to relate to concen-
tration polarization phenomena, which leads to
severe fouling of membrane and extremely low
production rate, as measured by the mass flux of
product. Furthermore, circulation of biocatalyst
through the system under relatively high shear
stress tends to reduce enzyme activity [7].
Recently, hollow fiber membrane bioreactor
(HFMB) is considered as an attractive alternative
for enzymatic reactions. HFMB has a large
surface area-volume ratio with a variety of module
designs and operation modes. With such charac-
teristics, HFMB gives a high volumetric pro-
ductivity (kg product per cubic meter reactor per
hour). Many investigators have recognized the
potential applications of the reactor. Rony [8] first
suggested immobilizing enzymes within the lumen
of hollow fiber membranes. The reactor was then
investigated by Chang et al [9], Chang and Chung
[lo], and Shao et al. [II].
A different configuration of HFMB was
developed by Waterland et al. [ 121 using asymmetric
hollow fiber membranes. In this reactor, an
enzyme solution contained within the annular
open-cell porous support structure of the fiber is
separated from a substrate flowing through the
fiber lumen by an ultrathin dense membrane
impermeable to enzyme but permeable to substrate
and product. Both of those models rely on mass
diffusion. Their mass transfer limitations make
them impracticable for industrial applications.
This limitation led Breslau [ 131 and Shi et al. [7]
to develop a flow-controlled HFMB operating
under backflush mode. Due to such mode of
operation, the bio-catalyst tends to build up upper
the membrane surface.
This paper concerns with a type of hollow fiber
membrane bioreactor (HFMB) where the enzyme
was entrapped inside membrane pores. It is
expected that the proposed configuration enable
to eliminate mass transfer limitation. By adjusting
flux rate, which can affect the residence time,
maximum conversion can be obtained to reduce
downstream processing cost and unconverted
substrate. The investigation was focused on
enzyme loading, permeation characteristic of 6-
APA, and HFMB performance for hydrolysis of
penicillin G A theoretical mode1 has been developed
to have better understanding of the reactor
behavior at steady state condition.
2. Theoretical model
In the present study, we try to identify a set of
process variables for hollow fiber membrane bio-
reactor based on mode1 proposed by Fabiani et
al. [ 141. The model considers a single fiber (Fig. 1).
In order to simplify the mathematical formulation,
rectangular coordinates are used. Concentrations
in axial direction are assumed to be uniform and
the variation of surface area along the fiber radius
is neglected.
It should be pointed out that the proposed
model allows us to make predictions about the
reactor behavior when the diffusion contribution
to the mass transport is not negligible.
In rectangular coordinate, the equation
describing the complete mass transport pheno-
mena with chemical reaction across the hollow
fiber wall at steady state is:
Da2s as
--z-R=o 3X2
(1)
in the range 0 < r I 6 and with the following
boundary conditions:
s
Fig. 1. A single hollow fiber. R, = 1.5 mm; R, = 1.9 mm;
6=0.2mm;L= 1OOmm.
1.G Wenten, I.N. Widiasa /Desalination 149 (2002) 279-285
281
s = so (x = 0) (2)
-D as/ax = 0 (x = 6) (3)
The second boundary condition means that the
bulk concentration profiles is flat, therefore, there
are no diffusive fluxes on the external wall of the
membrane. Suppose F = -D X/&Y, Eq. (1) can be
rewritten as follows:
with following boundary conditions:
s = s,, (x = 0)
(5)
F=O(x=@
(6)
An iterative shooting method with implicit
finite difference for nonlinear ordinary differential
Eq. (4) is utilized to obtain substrate concentration
atx= 6:
F = 4-1 -J&L ). Ax
,
l--U.hD
(7)
S, = S,_, -Ax. F I D
(8)
The following equations have been used for
chemical reaction rate:
~(S)22_
K,,, + s
(no inhibition)
3. Materials and methods
3. I. Chemicals
Penicillin acylase (penicillin amidase, EC
3.5.1.11) utilized in this study was purchased from
Sigma Chemical Co. It was dissolved in phosphate
buffer solution 0.1 M at pH 7.5. The enzyme is
one of the commercially available enzymes with
high activity and ideal for hydrolysis of penicillin
G into 6-amino-penicillanic acid (6-APA) and
phenyl-acetic acid (PAA) at pH and temperature
ranges from 7.5 to 9 and from 35 to 52C respec-
tively. In order to obtain consistent experimental
results, penicillin G analytical grade was used as
substrate. All other chemicals were purchased
from commercial sources and all were analytical
grade. Distilled water was used exclusively in all
experiments.
3.2. Experimental apparatus
Fig. 2 shows the schematic experimental
apparatus. The HFMB consisted of a double glass
tube and hollow fiber membrane with 0.2 mm pore
size, 0.2 mm thickness, 1.5 mm inside diameter,
and 100 mm effective length. The membrane was
supplied by X-Flow. The volume of the stirred
tank (V,) is 200 ml. The total volume of the
internal circuit is 250 ml.
Penicillin
Membrane unit
Fig. 2. Schematic experimental apparatus of hollow fiber
membrane bioreactor. A: mixer; FL: flow meter; P: pump;
PI: pressure indicator; S: valve; Vl: working tank.
3.3. Enzyme immobilization
First of all, the stirred tank was filled with
solution of penicillin acylase in phosphate buffer.
Solution concentrations were varied from 140 to
1550 u.a.11. The solution was circulated through
the hollow fiber membranes under dead-end
282 1.G Wenten, I.N. Widiasa /Desalination 149 (2002) 279-285
operation mode from lumen side to shell side
during 30 min. Then, distilled water was circulated
through the external and internal circuits until the
rinsing water was free from the enzyme. Penicillin
acylase quantities in permeate and rinsing water
were determined using the assay described below.
3.4. Continuous hydrolysis of penicillin G
After immobilization step, the continuous
experiments were carried out at temperature of
35C by flowing the substrate solution under
dead-end operating mode from lumen side to shell
side. In this manner, substrate will be contacted
with the immobilized penicillin acylase and
converted into 6-APA during a certain time in the
reactor. Data from HFMB were collected under a
variety of operating conditions. The velocities of
feed were varied from 1.2x1 O-2 to 245x1 @2cm/min.
The concentrations of substrate were varied from
1.87x10- to 7.48~10~~ M. After that, the hollow
fiber module was cleaned by sodium hydroxide
solution 1% at temperature of 60C and was then
filled with 0.15% sodium azide to prevent microbial
growth. Prior to each experiment run, experimental
apparatus was rinsed by pumping at least 2 1 of
distilled water.
3.5. Analytical methods
Enzyme activity of penicillin acylase was
assayed by measuring the amount of 6-APA
liberated from penicillin G by means of the p-di-
methylaminobenzaldehyde method [ 151. The
procedure is based upon formation of a 2,4-penta-
nedione derivative of 6-aminopenicillanic acid
followed by a second reaction with p-dimethyl-
amino benzaldehyde, resulting in a red product.
A spectrophotometer (Shimadzu, UV-120-02) was
used to determine the concentration of 6-APA at
wave length of 538 nm. One unit is defined as
the amount of enzyme that liberates 1 mmole of
6-APA from Penicillin G per minute at 34C [ 161.
Moreover, conversion of penicillin G was indirectly
monitored using the same method.
4. Results and discussion
4.1. Immobilization of penicillin acylase
When using reactor with enzymes within the
pores of asymmetric membrane, the molecular
weight cut off and enzyme size must be com-
patible so that no enzyme pass across the membrane.
In this work, we investigated the capability of a
kind of microfiltration membrane for immobilizing
penicillin acylase as a model system. Fig. 3 shows
the results of immobilization of the enzyme using
the hydrophilic hollow fiber membranes (see the
section on experimental apparatus). As can be seen
in this figure, penicillin acylase was entrapped
more than 90% (100,000 u.a/m2). In addition, the
specific unit activity decrease when initial concen-
tration of the enzyme more than 1500 u.a./liter.
This denotes that at a high concentration, molecules
of enzyme may compete to reach the membrane
pores. On the other hand, the enzyme can freely
pass through the membrane pores at low concen-
tration. Thus, the hollow fiber MF membrane can
best serve as an immobilization support under
proper enzyme concentration.
4.2. Permeation characteristic of 6APA
At the beginning of the study, we investigated
the permeation rate of 6-APAacross the membrane.
100
0
0 500 1000 1500 2000
Initial solution concentration (u.a/lt)
Fig. 3. The effect of initial enzyme concentration on enzyme
loading, +: % immobilized membrane; Cl: specific unit
activity.
1.G Wenten, I.N. Widiasa /Desalination 149 (2002) 279-285 283
80
Z
5 6o
z 40
Ci
20
0
0 20 40 60 80
Time (minutes)
Fig. 4. Permeate flux and rejection of 6-APA, W, 0: through
enzyme free-membrane; +, 0: through enzyme immobi-
lized membrane (black markers are permeate flux).
Results of a set of such experiments taken at TMP
of 8 psi are shown in Fig. 4. It can be seen that
the complete permeation of 6-APA was achieved
over a test period of 60 min when no immobilized
enzyme in the membrane pores. Due to the much
smaller size of 6-APA compared to the membrane
pore, the solute diffuses freely through the
membrane. However, the immobilized enzyme
membrane retained around 35% of the solute. This
phenomenon may be caused by the build-up of
enzyme layer followed by partially pore blockage
and the formation of enzyme-6-APA complex.
After the enzyme was saturated by 6-APA,
permeation of the solute was only affected by
fouling characteristics. Although the permeate
flux is relatively low, it should be emphasized that
suitable module design and operating conditions
may enhance the stable flux. This has been
confirmed by the results obtained using different
module and operating conditions.
4.3. Performance of HFMB
The basic experiments were conducted to
obtain the effect of flux rate on the degree of
conversion for various initial substrate concen-
trations. The solution of penicillin G was pumped
100
g
80
x
B
60
C
.o
$ 40
;
0 20
0
0 50 100
Flux rate (UM*/h)
150
Fig. 5. Performance of enzymatic hollow fiber membrane
bioreactor for penicillin G concentration: +( 1.87~16 M),
0 (3.74x10- M) and l .(7.48xlW M).
through HFMB under dead-end operation mode
from lumen side to shell side. In this manner,
penicillin G would be brought into contact with the
immobilized enzyme during a certain time. The
experimental results of the HFMB performance
are shown in Fig. 5. It is found that the conversion
degree of substrate is strongly influenced by flux
rate. Higher conversion could be achieved by
reducing filtration rate. Low flux rate is also
important to avoid gel formation of immobilized
enzyme on the membrane or enzyme release from
the membrane pores.
The nonlinear regression of experimental data
for penicillin G concentrations of 1.87~10-~,
3.74~10~~ and 7.48~10-~ M under a wide range
of flux rates (1.2~10-~ to 24.5x10-*cm/min) was
conducted without taking into account the
possibility of substrate or product inhibitions. The
numerical simulation resulted in values of the
kinetic parameters and diffusivity coefficient as
summarized in Table 1. As shown in this table,
the substrate diffusivity through the membrane
pore was approximately two orders of magnitude
lower than the solution diffusivity. The decrease
of substrate diffusivity depends on several para-
meters such as enzyme loading and membrane
284 I.G. Wenten, I.N. Widiasa /Desalination 149 (2002) 279-285
Table I
Table 2
Comparison of theoretical and experimental results
Summary of the kinetics parameters and diffusivity
coefficient
Parameter Initial substrate concentration, &, M
1.87~10-~ 3.74x1o-3 7.48~10-~
u 2.48~10-~ 1.2ox1o-2 3.13x1o-2
7.66~10-~ 2.71~10-~ 7.08~10-~
1 1.5x1o-2 10.8~10-~ 18.8x1o-2
20.8~10~~ 23.3~10-~ 24.5~10-~
V
ma*
1.48~10-~ 1.21x1o-2 1.36~10-~
G
8.04~10-~ 8.04~10-~ 8.04~10-~
D 3.21~10-~ 3.21~10-~ 3.21~10-~
U Conversion, %
Experiment Convective- Convective
diffusive
1.2ox1o-2 90.0 90.1 90.4
2.71~10-~ 56.7 58.3 58.6
10.8~10-~ 17.6 17.9 17.9
23.3~10-~ 9.6 8.6 8.6
pore structure. The increase of enzyme loading
decreases substrate diffusivity. Therefore, in the
case of high enzyme loading, dead-end operation
mode is more applicable than cross flow operation
mode. In addition, the consistent values of kinetic
parameters (K,, and V,,,) indicate that they were
obtained at saturated substrate concentration. The
apparent Michaelis constant (K,,) of immobilized
penicillin acylase (8.04 n&l) is slightly higher
than that of free penicillin acylase (7.75 mM,
shown in Fig. 6). This indicates that the internal
diffusion resistance, namely the steric hindrance
of the membrane matrix, is negligible. Con-
sequently, the immobilized enzyme is more
accessible for the substrate.
By using kinetic and hydrodynamic parameter
values in Table 1, were then predicted by
convective-diffusion model and by convective
model. As can be seen in Table 2, The results of
both models are very close to each other. This
denotes that convective transport is the main
mechanism of mass transport in our experimental
conditions.
5. Conclusions
3.5
3
z 2.5
- 2
.E 1.5
I
:
1
0.5
0
0 0.2 0.4 0.6 0.8
l/S (mM_)
Fig. 6. Lineweaver-Burk plot of free penicillin acylase at
temperature of 35C.
A type of hollow fiber membrane bioreactor
(HFMB) where the enzyme was entrapped inside
membrane pores was investigated. The immobili-
zation results show that penicillin acylase was
entrapped more than 90% (100,000 u.a/m2). Due
to the much smaller size of 6-APA compared to
the membrane pore, the solute diffuses freely
through the membrane. However, the solute was
retained around 35% in case where enzyme was
immobilized. In addition, Kn, of immobilized
penicillin acylase (8.04 mM) is slightly higher
than that of free penicillin acylase (7.75 mM).
Finally, the theoretical results obtained by con-
vective-diffusion model and by convective model
are very close to each other indicating that con-
vective transport is the main mechanism of mass
transport in these experimental conditions. Higher
degree of conversion can be achieved by reducing
1.~3 Wenten, I.N. Widiasa /Desalination 149 (2002) 279-285 285
filtration rate. Low flux rate is also important to
avoid gel formation of immobilized enzyme on
the membrane or enzyme release from the
membrane pores.
Acknowledgement
Financial support is from the Minister of State
for Research and Technology Republic of Indonesia
under Grant No. 256/SP/RUT/BPPT/IV/1996.
The support of X-Flow in the form of membranes
is gratefully acknowledged. The authors are
indebted to Budiraharjo, Soelistiono, Sylvia and
Prima for technical assistance.
Symbols
D
F
57,
L
R
RI
R2
s
S
U
v,,,
x
6
- Diffusion coefficient, cm2/mnt
- Flux, mmol/cm*/mnt
- Michaelis constant, mmol/ml
- Fiber length, mm
- Chemical reaction rate, mmol/ml/mnt
- Inner radius of fiber, mm
- Inner radius of fiber, mm
- Penicillin G concentration, mmol/ml
-Penicillin G concentration in input,
mmol/ml
- Linear velocity inpores, cm/mm
- Asymtotic value of R, mmol/ml/mnt
- Rectangular coordinate, mm
- Fiber thickness, mm
References
111
121
R.L. Ulibarri and GM. Hall, Saccharification of
cassava flour starch in a hollow-fiber membrane
reactor, Enzyme and Microbial Technol., 21 (1997)
398-404.
W.D. Deeslie and M. Cheryan, A CSTR-hollow fiber
system for continuous hydrolysis of proteins. per-
[31
[41
[51
[61
[71
PI
[91
UOI
u11
[I21
[I31
1141
1151
[IhI
formance and kinetics, Biotechnol. Bioeng., 23 (198 1)
2257-2271.
W.D. Deeslie and M. Cheryan, A CSTR hollow fiber
system for continuous hydrolysis of proteins. factors
affecting long term stability of the reaction, Biotech
and Bioeng., 24 (1982) 69-82.
W. Berke, H.J. Schuz, C. Wandrey, M. Morr, G Denda
and M.-R. Kula, Continuous regeneration of ATP in
enzyme membrane reactor for enzymatic syntheses,
Biotechnol. Bioeng., 32 (1988) 130-139.
C.K. Lee and J. Hong, Membrane reactor coupled with
electrophoresis for enzymatic production of aspartic
acid, Biotechnol. Bioeng., 32 (1988) 647-654.
T.E. Williams, G Catapano, E. Klein and R.A. Ward,
Reduction of product inhibition by use of an enzyme
membrane reactor in the processing of natural
substrates, AIChE Symposium Series, 85 (1989) l-
9.
G Shi, X. Yu and Q. Yuan, Hollow-fiber membrane
reactor under backflush mode for penicillin hydrolysis,
Int. Chem. Eng., 33 (1993) 691-697.
P.R. Rony, Multiphase catalysis II, hollow fiber
catalysis, Biotechnol Bioeng., 13 (1971) 43147.
H.N. Chang, Y.S. Kyung and B.H. Chung, Glucose
oxidation in a dual hollow fiber bioreactor with silicon
tube oxidator, Biotechnol Bioeng., 31 (1987) 552-
557.
B.H. Chung and H.N. Chang, Aerobic fungal cell
immobilization in a dual hollow fiber bioreactor,
Biotechnol Bioeng., 32 (1988) 205-212.
X. Shao, Y. Feng, S. Hu and R. Govind, Pectin
degradation in a spiral membrane reactor, AIChE
Symposium Series, 85 (1989) 85-92.
L.R. Waterland, AS. Michaels and C.R. Robertson,
A theoretical model for enzymatic catalysis using
asymmetric hollow fiber membranes, AIChE J., 20
(1974) 50-59.
B.R. Breslau, Catalytic Process Utilizing Hollow Fiber
Membranes, U.S. Patent 4,226,026 (198 1)
C. Fabiani, G Giubileo, M. Pizzichini and V. dan
Violante, Steady state modelling of a hollow fiber
enzymatic reactor, Biotech and Bioeng., 29 (1987)
458-461.
J.M. Komfeld,Anal. Biochem., 86 (1978) 118-126.
K.-I. Suga, T. Sorai, S. Shioya and E Ishimura, J.
Chem. Eng. Japan, 26 (1993) 709-714.