Ceramic membrane microfilter as an immobil ized enzyme reactor

T. J. Harrington, J. L. Gainer and D. J. Kirwan

C e n t e r f o r B i o P r o c e s s D e v e l o p m e n t , D e p a r t m e n t o f C h e m i c a l Eng ineer ing , Un ivers i t y o f Virginia, Char lo t tesv i l l e , VA

This study investigated the use o f a ceramic microfilter as an immobilized enzyme reactor. In this type o f reactor, the substrate solution permeates the ceramic membrane and reacts with an enzyme that has been immobilized within its porous interior. The objective o f this study was to examine the effect o f permeation rate on the observed kinetic" parameters for the immobilized enzyme in order to assess possible mass transfer influences or shear effects. Kinetic parameters were found to be independent o f f low rate for immobilized penicillinase and lactate dehydrogenase. Therefore, neither mass transfer nor shear effects were observed fi)r enzymes immobilized within the ceramic membrane. Both the residence time and the conversion in the microfilter reactor could be controlled simply by regulating the transmem- brane pressure drop. This study suggests that a ceramic microfilter reactor can be a desirable alternative to a packed bed o f porous particles, especially when an immobilized enzyme has high activity and a low Michaelis constant,

Keywords: Immobilized enzyme: membrane reactor: kinetics; mass transfer

Introduction

This work has been per formed to character ize the use of a C E R A F L O W (Norton Company) tubular ceramic microfilter as an immobil ized enzyme reactor. In this type of reactor , the substrate solution flows through the membrane and reacts with an enzyme that has been immobil ized within its porous interior. The ceramic membrane is const ructed from sintered alpha alumina particles that are approximate ly 0.5 p~m in the filtration skin layer and 6/xm in the outer support layer. ~ Simple engineering calculations indicate that the surface area per unit volume in the microfilter is very large [a - 6 (1 - e ) / d p ~ 103-104 cm-1]. The membrane wall can be envis ioned as a packed-bed reactor of fluid-impervious microscopic particles. Because there is fluid flow con- tact to this large surface area, there should be excellent mass t ransfer within the membrane without severe pressure drops across the membrane wall, which is only 0.1 cm thick.

Since the pores in the membrane are quite small,

Address reprint requests to Dr. Kirwan at the Center for BioProcess Development, Department of Chemical Engineering, University of Virginia, Charlottesville, VA 22903 The present address ofT. J. Harrington is Abbott Laboratories, 1401 Sheridan Road, North Chicago, IL 60064 Received 9 January 1992; revised 17 May 1992

fluid flow through the membrane results in significant shear forces which may adverse ly influence enzyme activity. These fluid shear s tresses are es t imated to be on the order of 50 to 100 Pa. Previous work using tubular nylon reactors suggests that immobil ized en- zyme activity was unaffected by fluid shear at shear rates of up to 10,350 s -j and at shear forces of up to 73 Pa for immobil ized penicillinase and lactate dehydroge- nase.= In this work, we also seek to confirm this expec- tation.

M i c r o f i l t e r m e m b r a n e r e a c t o r

A ceramic membrane microfilter immobil ized enzyme reactor was used by Nakaj ima et al. 3-5 to immobil ize invertase. These authors conducted their investiga- tions with the microfilter in the cross-f low configuration and used high substrate concentrat ions (0.3 and 1.79 M). Nakaj ima et al. modeled reaction in the membrane by considering an imaginary pore with react ion oc- curring at the pore wall, and with axial convect ion, radial diffusion, and axial dispersion occurr ing within the p o r e ? In their latest paper , 5 the authors concluded that both radial diffusion and axial dispersion in the imaginary pore could be ignored due to the magnitude of their characteris t ic t imes relative to the residence t ime in the pore and the react ion rate t ime scale. Naka- j ima et al. used a Michae l i s -Menten model to describe the reaction. 3-5 In effect, the authors concluded that

© 1992 B u t t e r w o r t h - H e i n e r n a n n Enzyme Microb. Technol., 1992, vol. 14, October 813

Papers

reaction in the membrane could be modeled as a packed-bed reactor without external mass transfer ef- fects, s but did not demonstrate this by showing that kinetic parameters were independent of the membrane permeation rate.

Packed-bed reactor

For the ideal case of an immobifized-enzyme, packed- bed reactor, operating at steady state and under plug flow conditions, the integrated rate expression reduces tO6-11:

Vm~xZ= S o X - K M~ppln(1 - X ) (1)

The Jumped parameter, K~.po, includes the effects of the external film diffusional resistance:

KM app ~ K~ + Vm~,x/kLa (2)

Equations (1) and (2) have been used to model the observed kinetics for many immobilized enzyme reac- tions.~°-~4.

In equation (2), it is noted that the value of K M app will approach the value of KM when kLa is very large. A conservative estimate of the magnitude of the transport coefficient can be obtained by considering mass trans- fer to a single particle by diffusion of the solute through a stagnant liquid film adjacent to the well-mixed bulk fluidSS:

Sh = kLdp- = D 2 (3)

Because of the small particle diameter, this estimate gives a very large value for the overall mass transfer coefficient (kka ~ I03-104 cm rain i). Therefore, the measured value of KM ,vp in the membrane is expected to be the intrinsic K M for the immobilized enzyme. Indeed, this high mass transfer rate due to small parti- cles provides significant advantage to the membrane configuration, as opposed to a porous support where substrate must be supplied by molecular diffusion. This study then seeks to confirm this advantage along with a demonstration that there are no deleterious shear effects due to flow through the small pores.

Materials and methods

Materials

CERAFLOW Ceramic Microfilter tubes were donated by the Norton Company (Worcester, MA). The CERAFLOW microfilters are constructed of alpha alu- mina and have a nominal filter rating of 0.45/zm. These tubes have an internal diameter of 0.3 cm and an exter- nal diameter of 0,5 cm. Penicillinase (Difco Labora- tories, Bacto Penase Concentrate, 20,000 LU m1-1, code 0346) was purchased from Fisher Scientific (Springfield, N J). All other chemicals were purchased from Sigma Chemical Company (St. Louis, MO): 3-aminopropyltriethoxysilane (A3648), glutaraldehyde (G5882), lactate dehydrogenase (LAD solution 340-10), Penicillin G (Pen K), pyruvate (P2256), and NADH (N8129).

S T

V R

V T, S T

Figure1 Ceramicmicrofilter reactor in dead-end configuration under batch-recycle operation

Enzyme immobilization me thod

The enzyme was immobilized to the ceramic microfil- ters by activating the ceramic with an aminosilane and using glutara[dehyde to couple the enzyme to the acti- vated ceramic.L~ All attachment procedures were con- ducted in a recirculation system with the ceramic mi- crofilter operated in the dead-end configuration so that all fluid permeated the pores (see Figure 1 ). The mi- crofilter was first flushed with a 10% nitric acid solu- tion, tbllowed by distilled water, and finally with iso- propyl alcohol. The microfilter was then dried overnight in an oven at 100°C. A 10% aqueous solution of 3-aminopropyltriethoxysilane was adjusted to pH 5 with glacial acetic acid. This solution was heated to 70°C and then pumped through the microfilter for 3 h. The microfilter was rinsed with distilled water and then cured overnight in an oven at 100°C. The silanized microfilter was rinsed with isopropyl alcohol and then with distilled water. A solution of 2.5% glutaraldehyde in 0.1 M PO 4 buffer (pH 7.0) was pumped through the microfilter for 1 h at room temperature. Last, the pur- chased enzyme solution was pumped through the mi- crofilter overnight at 4°C.

Experimental condit ions/experimental apparatus/assay methods

The experimental conditions and experimental appara- tus that were used in this work are reported elsewhere. 2

814 Enzyme Microb. Technol., 1992, vol. 14, October

Penicillin concentra t ion was measured with a modified hydroxylamine assay.~7 N A D H concentrat ion was de- termined by measuring the absorbance at 340 nm. ~8 N A D H concentrat ions were linear up to concentra- tions of 0.3 mM NADH with 10 mM pyruvate in 0.1 M Tris buffer (pH 7.3). Higher NADH concentrat ions were measured by diluting them with 10 mM pyruvate in 0.1 M Tris buffer (pH 7.3).

All experiments were conducted with the microfilter operated in the dead-end filtration mode because this permitted maximal permeation rates. In this paper, im- mobilized enzyme kinetic parameters are discussed in terms of flow rate through the microfilter. The mem- brane permeation rate is equal to the flow rate divided by the membrane cross-sectional area. The liquid vol- ume of the ceramic microfilter reactor (VR) was calcu- lated from the dimensions of the membrane and the volume void fraction (e), which was taken as 0.4. ~

Measurement of reaction kinetics

Most kinetic studies were performed with a batch-recy- cle reactor configuration. A discussion of the material balance for the batch-recycle reactor, and the determi- nation of both zero-order and first-order reaction rate constants appears in our previous paper. 2 In this work, differential reactor operation in the batch-recycle reac- tor was used to obtain both kinetic parameters when Michael is -Menten reaction kinetics were observed. The ceramic microfilter immobil ized-enzyme reactor was also operated as a single-pass reactor for some experiments.

Di f f e ren t i a l r eac tor opera t i on

In the case of differential reactor operation within the batch-recycle reactor, the conversion per pass ap- proaches zero, so that the concentrat ion in the tank is approximately equal to the concentrat ion in the reac- tor. Therefore , the observed rate is assumed to occur at the concentrat ion measured in the reservoir.~9 For the case of differential conversion in the batch-recycle reactor with Michael is -Menten kinetics, the material balance reduces to:

V T dS T VmaxS T ~RR'-Z0-- = rate°b~ = - ~ (4)

Operation in the differential regime with high flow rates and/or high substrate concentrat ions is suitable for determining Vm~x and K M in the absence of external film diffusional effects. This differential method was used as long as the single pass conversion was less than 5%. Equation (4) can be linearized to extract the kinetic parameters , Vm,x and KM:

Sx/rateob~ = KM/Vm.~x + STIVmax (5)

This linearized form of the Michael is-Menten rate ex- pression was used because it tends to spred the x-axis and to give more equal weighting to each data point. 2°

Ceramic membrane microfilter: T. J. Harrington et al.

Single-pass operation

Single-pass reactor operat ion was also used to deter- mine the first-order rate constant within the membrane reactor. Because the single pass convers ion ranged between 14 and 75%, an integral rate method was used to determine the first-order reaction rate constant. Op- eration in the first-order regime was ensured when the reactor inlet substrate concentrat ion was less than K M for the immobilized enzyme.

The first-order rate constant was determined by measuring steady-state inlet and outlet reactor concen- trations as a function of the flow rate through the mem- brane reactor. The single-pass convers ion was then calculated directly from reactor inlet and outlet concen- trations:

X = (S o - Sv:)/S o (6)

Because the residence time in the reactor can also be calculated directly (~- = Q/V~), the expression for con- version as a function of residence time for a first-order reaction can be used to calculate the observed first- order rate constant, k:

X = 1 - exp [ - k r] (7)

Results and discussion

Penic i l l inase

Determination of maximal reaction velocity (Vmax). The batch-recycle reactor system was used to determine V,~x for penicillinase immobilized within the ceramic microfiiter. Figure 2 is a plot of substrate concentrat ion in the reservoir versus time for different flow rates through the immobilized penicillinase reactor. Note that the initial concentrat ion is slightly different in each run because this simply represents the time when the first measurement was taken. The plots are linear over the entire concentrat ion range and have the same slope independent of flow rate; only two lines are shown for clarity. Since Vm,x is proportional to the slope of these

15 {- I I 7 - - t - o

0 5.8 ml/min.

(~ . 1£ - - , , , ~ A 21.6 rnl/min. - - cj L..

0 r - ~ 5 - . ~ r r -

"K .~-

a_ 0 r j I E ~

0 10 20 3O 40 5O

Time, min.

Figure 2 Zero -o rder p lo t fo r immob i l i zed penicillinase reactor under batch-recyc le operation at different f l o w rates (/. = 34.5 cm)

Enzyme Microb. Technol. , 1992, vol. 14, October 815

Papers 2.01

E

-~- 1.5 o >

o9 1.0 '

r r

~" 0.5 £3 < Z 0 . 0 - -

0

F ~ T 1 3 I

E3 27.7 ml/min. - - . ' , - -

L__ [ I I _1_ L J 10 20 30 40 50 60 70

Time, min.

Figure3 NADH concen t ra t i on in the reservo i r ve rsus t ime fo r i m m o b i l i z e d lactate d e h y d r o g e n a s e reactor unde r batch- recyc le ope ra t i on at d i f f e ren t f l o w rates (L = 40 cm)

lines (Vm~,~ = - slope [Vv/VR]), it is independent of the flow rate through the microfilter for these runs. Hence, V ...... is not influenced by shear, in agreement with our previous results for penicillinase immobilized on the inner wall of nylon tubes, 2 The value of Vm, ~ from Fi~,,ure 2 is 12.5 mM rain 1.

Determination of first-order rate constant (Vmax/KM). In Figure 2 it is noted that the reaction within the immobil ized penicillinase reactor follows zero-order kinetic behavior down to very low concentrat ions (~0.5 raM). Because it was not possible to accurately measure substrate concentrat ions much lower than 0.5 raM, it was not possible to measure the first-order rate constant (V,n,x/K M) with this assay method. The K M for penicillinase immobil ized on aminoethyl cellulose has been reported as 3.3 mg21; the K M for penicillinase in solution has been reported as 0.05 raM. 2"-

Lacta te dehydrogenase

Determination of maximal reaction velocity (Vmax), The batch-recycle reactor sys tem was used to determine V.~.~ for lactate dehydrogenase immobilized within the ceramic microfilter. With relatively high flow rates and substrate concentra t ions , the convers ion per pass was approximate ly 3% and Vm~ ~ was determined with a differential rate method. This was done by taking mea- surements of the concentra t ion in the reservoir versus t ime and fitting the data to a second-order polynomial , as seen in Figure 3. The observed rate was determined by differentiating the polynomial expression with re- spect to time. Since the convers ion per pass was less than 5%, the observed rate was taken to occur at the measured concentra t ion in the reservoir . A linearized form of the Michae l i s -Menten rate expression was used to obtain V . . . . "

Figure 4 shows the plot of ST/rateob~ versus S T that was used to obtain V~×. In Figure 4 the maximal reac- tion velocity is equal to the reciprocal of the slope, and the first-order rate constant (Vm,~/K M) is equal to the

reciprocal of the intercept. It is noted that the slope and the intercept are the same for both of the flow rates. This implies that the measured values are not influenced by either shear or t ransport effects within the membrane wall.

The values of the kinetic pa ramete rs f rom Figure 4 are as tbllows: Vm, x = 3.2 mM min -I and K M = 1.1 mM (for NADH) . Since the value of VmJkLa in equation (2) is of the order of 10 3 raM, the measured KM is the intrinsic value for the immobil ized enzyme. K M values of 1.1 mM (for N A D H ) were obtained for other immobi- lized lactate dehydrogenase preparat ions . Wykes et al. 2~ have reported the K M for lactate dehydrogenase immobilized on DEAE-cel lu lose as 0.84 and 1.33 mM (for NADH) , and the K M for lactate dehydrogenase in solution as 0.028 mM (for NADH) .

Determination of first-order rate constant (Vmax/KM). In addition to the differential rate method discussed above, two additional methods were used to measure the first-order rate constant (Vm,JK M) for immobilized lactate dehydrogenase . Each method was per formed using a different enzyme- tube preparat ion.

The batch-recycle reactor sys tem was used to mew sure the first-order rate constant at low substrate con- centrations. Figure 5 is a plot of log (S T) versus time for different flow rates in the batch-recycle reactor s y s t e m The first-order rate constant was calculated for each flow rate by obtaining the convers ion f rom the slope of each p lo t ] Table 1 presents the calculated values of the first-order rate constant , k, f rom the data in Figure 5. The calculated first-order rate constant was found to be independent of flow rate for this data set and is therefore taken as equal to V ..... /K M . The devia- tions at the higher flow rates result f rom inaccuracies in the determinat ion of the reactor residence time.

The first-order rate constant was also determined by running the ceramic microfilter as a single-pass reactor. Since the single-pass convers ion ranged between 14 and 75%, an integral rate method was used to determine the first-order reaction rate constant f rom the measured convers ion at various flow rates. From equat ion (7), it

"T. 1.0 e--

E

. 1 3

~ o

re

I £3 < z

0.5

-- ~ 27.7 mFmin. - - O 55.0 ml/min.

l _ I L . 0.~3.0n 0.5 1.0 1.5 2.0

[NADH] in Reservoir , mM

Figure 4 Plot of ST/rat%b s ve rsus S T f r om Figure 3

816 Enzyme Microb. Technol . , 1992, vol. 14, Oc tober

0.40,[ I [ I

0 . 3 ~

~ 0.2

~ 0 . 1 0

0.40 L ~ L ~ " , - 0 5 10 15 20

Time, min.

Figure 5 First-order plot for immobi l i zed lactate dehydrogenase reactor under batch-recycle operat ion at di f ferent f low rates (L = 40 cm)

1.5

1.0 X i

c-

' 0.5

0 . 0 0.0

- - i

° i I

i t ~ I l ~

I

0.1 0.2 0.3 0.4 0.5

Residence Time, min.

Figure 6 Reactor convers ion as a funct ion of residence t ime for immobi l i zed lactate dehydrogenase

is seen that a plot of - In (1 - X) versus r will have a slope equal to the first-order rate constant (V~,~/K M) and an intercept at the origin. Figure 6 is such a plot for single-pass operation. The value of the first-order rate constant from Figure 6 is 3.6 rain ~.

Comparison of methods for determining the first-order rate constant (Vm~x/K~a). Table 2 presents a comparison of the values of the first-order rate constant that were obtained by all three methods using the same enzyme- tube preparation: the batch-recycle reactor under dif- ferential operation with high substrate concentrations, the batch-recycle reactor with low substrate concentra- tions, and single-pass operation with appreciable conversion levels. All three methods show good agreement. This is significant because the batch-recy- cle reactor under differential operation is the only one of these methods that allows determination of both the maximal reaction velocity (Vma 0 and the Michaelis constant (KM).

Note that the values for Vm~x/K M shown in Table 1 differ from those in Table 2. This is because different

Ceramic membrane microfi/ter: T. J. Harrington et al.

enzyme-tube preparations (with different enzyme load- ings) were used. The tube that was used to collect the data in Table 1 was prepared using twice the volume of enzyme attachment solution as the preparation in Table 2. In using double the volume of enzyme attach- ment solution, it appears that a higher enzyme loading resulted.

Conclusions

The observed values of Vm~ ~ for immobilized penicil- linase and lactate dehydrogenase, and K M for lactate dehydrogenase, were not affected by membrane per- meation rate. This supports our previous finding with tubular reactors that these immobilized enzymes are not influenced by shear at the levels testedfl Further, the lack of flow rate dependence of K~a supports our hypothesis that external film diffusional resistances to mass transfer are negligible for immobilized enzyme reactions within the ceramic microfilter reactor.

The membrane reactor configuration has a large cross-sectional area and a shallow bed depth. High membrane permeation rates can be achieved at low transmembrane pressure drops. The high surface area to volume of this reactor allows for high immobilized enzyme activity, which allows high reactor conver- sions at relatively small residence times. These prop- erties make the membrane reactor configuration an attractive alternative to a packed bed of porous par- ticles.

Table 1 First-order rate constant for immobi l i zed lactate dehy- drogenase in batch-recycle reactor

Q (cm 3 min -1) X ~- (rain) k (min -1)

5.1 0.90 0.39 6.0 12.8 0.60 0.16 6.0 27.5 0.36 0.07 6.2 52,3 0.23 0.04 6.8

Data f rom Figure 5

Table 2 Compar ison of methods for de termin ing f i rst-order rate constant for immobi l i zed lactate dehydrogenase

Vmax/KM Vma× KM Method (rain -1) (raM rain 1) (raM)

Batch-recycle 3.4 3.7 1.1 di f ferent ial (data not shown)

Batch-recycle 3.6 * * constant X, (data not shown)

Single-pass 3.6 * * (Figure 6)

* Method does not permi t de terminat ion

Enzyme Microb. Technol., 1992, vol. 14, October 817

Papers

N o m e n c l a t u r e (.I

dp D k kL

KM KM app L Q rateobs SE So ST V

m a x

VR VT X

surface area to volume ratio, cm- particle diameter, cm diffusion coefficient, cm 2 rain- observed first-order rate constant, rain i external film mass transfer coefficient, cm min ~ intrinsic Michaelis constant, mM apparent Michaelis constant, mM reactor length, cm flow rate through reactor, cm 3 min- observed rate, mM min concentration leaving reactor, mM concentration entering reactor, mM concentration in the reservoir (tank), mM maximal reaction velocity, mM min 1 reactor volume, cm 3 reservoir (tank) volume, cm 3 conversion in reactor, (Cr - C~)/CT

Greek symbols s v o l u m e t r i c v o i d fract ion in reactor O r e s i d e n c e t ime in the reservo ir , VT/Q,

min r r e s i d e n c e t ime in reac tor = VR/Q, min

References

1 Norton Company CERAFLOW Asymmetric Ceramic Micro- filter literature

2 Harrington, T. J., Gainer, J. L. and Kirwan, D. J. Enzyme Microb. Technol. 1991, 13, 610-616

3 Nakajima, M., Jimbo, N., Nishizawa, K., Nabetani, H. and Watanabe, A. Proc. Biochem. 1988, 23, 32-35

4 Nakajima, M.. Watanabe, A., Nabetani, H., Horitaka, H. and Nakao, S. Agric. Bio. Chem. 1988, 52, 357-365

5 Nakajima, M.. Watanabe, A., Jimbo, N., Nishizawa, K., and Nakao, S. Biotechnol. Bioeng. 1989, 33, 856-861

6 Patwardhan, V. S. and Karanth, N. G, Biotechnol. Bioenq. 1982, 24, 763-780

7 Bunting, P. S, and Laidler, K. J. Biotechnol. Bioen~,,. 1974.16, 119-134

8 Daka, N. J. and Laidler, K. J, Can. J. Biochem. 1978, 56, 774-779

9 Kobayashi, T. and Laidler, K. J, Biotec/mol. Bioen~,. 1974, 16, 99-118

10 Lilly. M. D.. Hornby, W. E. and Crook, E. M. Biochem. J. 1966, 100, 718

I I Daka, J. N., Laidler, K. J., Sipehia, R. and Chang, T. M. S. Biote('hnol. Bioen~. [988, 32, 213-219

12 Sharp, A. K., Kay. G. and Lilly, M. D. Biotechnol. Bioeng. 1969, 11,363-380

13 Tosa. T., Mori, T. and Chibata, 1. J. Ferment, Technol. 1971. 49, 522-528

14 Kobayashi. T. and Moo-Young, M. Biotechnol. Bioet ,~. 1973, 15, 47-67

15 Bird, R. B., Stewart, W. E, and Lightfoot, E. N. Transport Phenomem~. John Wiley & Sons, New York, 1960, pp. 642-648

16 Weetall, H. H. lmmobili:,ed Enzymes, Antigen~s, Antibodies, and Peptides. Vol. 1. Marcel Dekker, New York, 1975, pp. 7-13

17 Hamihon-Millcr. J. M. T. and Smith. J. T. Beta-Lactamase,~. Academic Press, New York, 1979, pp. 37-39

18 Sigma Chemical Company, Procedure No. 340-UV 19 Smith. J. M. Chemical Engineering Kinetics, 3rd ed. McGraw-

Hill, New York, 1981, pp. 199-202 20 Wilkinson. G. N. Biochem. J. 1961, 80, 324-332 21 Klemes, Y. and Citri, N. Biotechnol. Bioeng. 1979, 21,897-905 22 Pollock, M. R. in The Enzyme,~, 2nd ed. Academic Press, 1963,

Vo/. 4. pp. 269-278 23 Wykes, J. R.. Dunnill. P. and Lilly, M. D. Bioteclmol. Bioen~,,.

1975. 17, 51-68

818 Enzyme Microb. Technol., 1992, vol. 14, October