Kinetic and operational study of a cross-flow reactor with immobilized pectolytic enzymes

Pedro Lozano, Arturo Manjon, Jos6 L. Iborra, Manuel C~inovas and F61ix Romojaro*

Depar tment o f Biochemistry and Molecular Biology, Faculty o f Chemistry, University o f Murcia, Murcia, Spain *C.E.B.A.S . (CSIC), Murcia, Spain

The kinetics and operational behavior of a nylon membrane derivative with immobilized pectolytic enzymes in a cross-flow reactor were analyzed. A high dependence on the recycling flow rate was observed. A design equation of the system has been proposed by taking into account both the shear stress deactivation and the control of the external diffusional limitations. Integration of the resulting design equation allowed us to study the effect of different operational parameters on substrate conver- sion. The catalytic efficiency of the immobilized derivative in a cross-flow reactor showed the highest pectin hydrolysis capability when it was compared with two different configurations of packed-bed reactors.

Keywords: Membrane reactor; cross-flow reactor; immobilized pectinases; shear stress; enzyme deactivation; fruit juice clarification

Introduction

The industrial viability of a biocatalytic process per- formed with immobilized enzymes is based upon im- mobilization efficiency, operational stability, and pro- ductivity. J

The utilization of an immobilized pectolytic enzyme system in fruit juice clarification could be very useful for continuous industrial use. However , the activity yield of these immobilized derivatives is negatively in- fluenced by the colloidal characteristics of the pectic substrate solution. 2 Thus, it is interesting to develop an efficient biochemical reactor in order to reduce the hydrodynamic limitations and to yield a high substrate conversion capability for long operational time.

A cross-flow reactor with immobilized enzymes onto the membrane surface shows several advantages for viscous substrate solution processing, since the ex- ternal diffusional limitations are clearly reduced by the high recycling flow rate. 3 This bioreactor was used by several authors 4,5 for the hydrolysis of whey and vege- table proteins and for the development of a mathemati- cal model concerning diffusion and reaction at only one recycling flow rate.

Address reprint request to Dr. Iborra at the Department of Biochem- istry and Molecular Biology, Faculty of Chemistry, University of Murcia, 30001 Murcia, Spain Received 5 January 1989; revised 7 April 1989

Synthetic membranes containing active covalently bound pectolytic enzymes have been previously devel- oped 6 and applied in a cross-flow reactor configuration for the continuous clarification of apricot juice with excellent results. 7 This paper concerns the kinetics and operational behavior of this immobilized pecto- lytic enzyme derivative in the membrane cross-flow reactor, as a function of the recycling flow rate, and presents a reactor design equation in order to study the effect of the different parameters of the system (recy- cling flow rate, filtrate flow rate, reaction volume, and amount of immobilized enzyme) on the conversion profiles. Additionally, the catalytic efficiency of this bioreactor was tested.

Materials and methods

M a t e r i a l s

The pectolytic enzyme preparation used was a com- mercial product (Pectinol D from R6hm GmbH) re- ported to have five polygalacturonases (EC 3.2.1.15), four pectin lyases (EC 4.2.2.2), one pectinesterase (EC 3.1.I.11), and minor quantities of other hydrolases, such as cellulases, galactomanases, proteases, etc., and having a protein content of 2.4% (w/w). The en- zyme preparation was used without previous purifica- tion except for a cross-flow ultrafiltration step with PTGC membranes from Millipore (10,000 cutoff), in order to remove low molecular weight stabilizers.

© 1990 Butterworth Publishers Enzyme Microb. Technol., 1990, vol. 12, July 499

Papers

Citrus pectin (grade II) f rom Sigma Chemical Co., having a 40-60% esterification degree, was used as substrate.

Nylon 6 membrane (NY I0 Super; open surface = 4.25%; water permeabil i ty = 82 1 • m -2 - s ]; weight = 50 g • m 2) f rom ZBF (Switzerland) was used as sup- port.

All remaining reagents were Merck, analytical grade, and were used without additional purification.

Methods

Immobilization process. The pectolytic enzyme prepa- ration was immobilized on nylon membrane by an O- alkylation activation method developed and previ- ously described in detail by Thompson et al. ~ In brief, the process was as follows: support activation with dimethyl sulfate; then, the activated support was coated with polyethyleneimine (PEI). The n y l o n - P E l derivat ive was react ivated with glutaraldehyde, to which the pectolytic enzyme preparat ion was finally at tached.

Activity measurements. Overall viscosity-reducing ac- tivity of the immobilized Pectinol D was measured in a thermosta ted C a n n o n - F e n s k e viscometer , using a 0.5% (w/v) pectin solution in 0.1 M acetate buffer, pH 4.0, as standard substrate. One unit of activity was taken as the amount of enzyme that hydrolyses 1 mg of substrate per minute, calibrated viscometrical ly, at 40°C.

Operational stability in a cross-flow reactor. A MINI- TAN S ® sys tem (from Millipore), containing between 0.15 and 0.20 g of derivative, was used as the cross- flow biocatalytic module, and a closed vessel was used as the reservoir tank. The reaction volume was 300 ml of standard substrate solution. The operational stabil- ity of immobilized Pectinol D on nylon membranes was tested by recycling the standard substrate solution through the biocatalytic module at 40°C and at differ- ent flow rates. The sys tem was operated continuously at a filtrate flow rate of 0.6 ml rain J and the samples were taken f rom the outlet product.

Operational stability in a packed-bed reactor. A jack- eted column (0.5 cm I.D.) with either 0.75 g of nylon pellets of immobilized derivative (243 U g i of sup- port) or 0.25 g of a nylon membrane immobilized en- zyme derivative fashioned into circles (208.1 U g ] of support) and packed perpendicularly to the column axis (6.5 cm bed height) was used as a packed-bed reactor; the standard substrate solution was pumped through at a flow rate of 0.12 ml min t and the samples were taken f rom the outlet product. 6

The degree of pectin hydrolysis in both operational stability exper iments was measured with a C a n n o n - Fenske viscometer , a method equivalent to the indus- trial process control method.

Implementation. To solve the design equation, the im- plementat ion proposed by Howell et al. 9 of an embed-

500 Enzyme Microb. Technol . , 1990, vol. 12, Ju ly

ded fourth-fifth order R u n g e - K u t t a algorithm due to Butcher was used.J° The implementat ion is such that the integration step is always updated to optimize inte- gration size and to minimize error.

Results and discussion

Reactor operation and kinetic behavior analysis

The reactor setup consisted of four parts: biocatalytic module (B M), reservoir tank (R T), recycling pump (PR), and filtrate pump (Pv) (Figure 1). The reactor was operated with recycle and tengentiai flow pattern and allowing the outlet of product at a constant flow rate controlled by the filtrate pump. The reservoir tank was pressure-t ight, so that the outlet product produced a pressure drop into the system, allowing an inlet of fresh substrate solution into it at the same flow rate as the outlet product. Thus, the sys tem behaves as a con- tinuous reactor , where the recycling pump only con- trols the mass- t ransfer rate. Hence , the external diffu- sional limitations were dependent on the recycling flow rate/ react ion volume ratio.

Since the recycling flow rate was largely higher than the feed flow rate, the reactor was assumed to be a whole system formed of both the reservoir tank and the biocatalytic module. That is similar to the continu- ous feed stirred tank reactor (CFSTR), where the recy- cling flow rate might represent the stirring rate and the residence time was dependent on the filtrate flow rate and reaction volume. In this type of reactor, the sub- strate mass-balance equation can be stated as fol- lows lI-

dS Qr" (So - S) - g . r = v • d--t (I)

where Qf is the filtrate flow rate (ml • min ]); So is the initial substrate concentrat ion (mg • ml J); S is the substrate concentrat ion into the reactor (mg • ml ~) at time t; g is the quantity of immobilized derivat ive (g); v is the reaction volume (reservoir tank plus biocata- lytic module) (ml); and r is the reaction rate (milligram

i~ET

z~ P~

Figure I Experimental setup: (R T) reservoir tank; (B M) biocat- alytic module; (PR) recycling pump; (PF) filtrate pump

of substrate hydrolysed per minute and per gram of derivative).

On the other hand, the behavior of this immobilized derivat ive in a stirred-tank reactor without external diffusional limitations was michaelian, 7 thus:

r = Vmap" S KMap + S (2)

which allows the subst ra te-mass balance equation (1) to be as follows:

gmap • S a s Q f . (So - S ) - g " KMap + S = u" d t (3)

The pectolytic activity of the immobilized deriva- tive, as well as its kinetic behavior , in the cross-flow reactor is controlled mainly by the external mass- t ransfer rate (since the solid support is nonporous) , dependent on the recycling flow rate/react ion volume ratio (Qr/v). Therefore , the effect of this paramete r was studied in order to establish an adequate design equation. Fur thermore , in continuous operat ion, the immobilized derivative showed a decay of the catalytic activity with time; thus it was necessary to introduce a deact ivat ion function within the reaction rate expres- sion for the design equation.

The apparent kinetic parameters , VM,p and KMap, for the immobilized derivative in use were previously determined at different recycling flow rates and con- stant reaction volume. 7 As can be seen in Figure 2, these apparent kinetic parameters could be linearized as a function of the recycling flow rate/react ion vol- ume ratio (Q~/v) as:

KM.,,p = A • e B'Q,/u (4)

Qr/o Vmap -- (5)

m • Q ~ / v + n

where B and n were empirical constants (B = 1.95 min; n = 1.74. 10 3 g 1ME • mg s-~), a n d A and m were the max imum values of the KMa p and of the reciprocal of

10 ¸

%/v8 VmQp

x | O 3

(=) 6

2

i I I i

0.35 0.70 1.12 1.40 n r / V (rain 1 )

5

/+ [n KMa p (,,)

3

2

i

o

Figure 2 Dependence of the apparent kinetic parameters of the immobi l ized pectolytic enzymes derivative wi th the mass-trans- fer control l ing parameter (Qr/v) in the cross-f low reactor

Immobilized pectolytic enzymes: P. Lozano et al.

Vmap, respect ively (A = 160.6 mg • ml- t ; m = 4.805 • 10 -3 min • g IME • mg s i).

Additionally, the immobilized derivat ive on nylon pellets showed a michaelian behavior in a batch reac- tor. 7 Since the support was nonporous and the experi- ments were carried out at high stirring rate, obtaining a linear H a n e s - W o o l f plot, it could be concluded that the external diffusional limitations had been avoided, which allowed the determinat ion of the intrinsic ki- netic parameters as Vmi = 208.1 U " g J IME and KMi = 0.42 mg • ml ~.

As can be checked, l / m = Vmi, then in equation (5) Vm,,p could be expressed as a function of Vm~ and Qr/o, as follows:

gmi" Qr/v (6) Vmap Qr/v + n " Vmi

In this expression, when Qr/v increases to infinity, Vmap approaches Vmi, and when Qr/v tends to zero, Vm, p also tends to zero.

On the other hand, from equation (4), when Qr/o approaches infinity, Kmap tends to zero, instead of ap- proaching the intrinsic value, KMi, as physically ex- pected. Hence KMi, as an addition term, was intro- duced into equation (4) with a small error in the standard experimental working conditions (0.25%). Then, KMa p could be expressed by the following equa- tion:

KM,p = KMi" (1 + C " e B.Q,/v) (7)

where C = A/KMi = 382.4, a dimensionless empirical constant.

As can be shown, the kinetic paramete r equations (6) and (7) include the external diffusional limitation effects into the mass-balance, since both functions contain the controlling mass- t ransfer paramete r Qr/v.

Thus, the reaction rate term of equation (2) could be expressed as follows:

Vmi " Qrlv . S

Q~/v + n . Vmi r = (8)

KMi" (l + C" e BO,/v) + S

This rate equation contained both the michaelian kinetic behavior and the external diffusional limita- tions of the immobilized system, controlled by the re- cycling flow rate and the volume of reaction.

A n a l y s i s o f t he o p e r a t i o n a l b e h a v i o r

In order to study the effect of the recycling flow rate on the operat ional behavior of the immobilized en- zymes derivative in the cross-flow reactor , the opera- tional stability was tested at different Qr values (50, 100, 200, and 400 ml • min ', respectively) and at only one filtrate flow rate (Qr = 0.6 mi - min-º) , using a total volume of substrate solution of 300 ml in the reservoir tank.

Figure 3 shows first-order deact ivat ion profiles for the immobilized derivat ive at those recycling flow rates assayed. As can be seen, an increase of the recy-

Enzyme Microb. Technol. , 1990, vol. 12, Ju ly 501

Papers

2.0

.~ 1.8

~ 1.6

"~ 1.4

~ 1.2

I l I i I I I

0 1 2 3 4 5 6 7

Time (cl)

1.0

Figure 3 Deactivation profiles of the immobil ized derivative in the cross-flow reactor at different recycling f low rates: (0) 50, (A) 100, (11) 200, and (0) 400 ml min

-9.2

-9.3

[nKD -9.4

-9.5

-9.6 I I I I

100 200 300 400 Qr (m[/min)

Figure 4 Dependence of the operational deactivation constant on the recycling f low rate

cling flow rate resulted in a concomitant increase in the max imum level of activity. This fact was a conse- quence of the high dependence of the reaction rate on the substrate recycling flow rate. An increase of the mass- t ransfer rate obviously reduces the external diffusional limitations. However , a decrease in the level of the substrate convers ion at long-time reaction was observed when the recycling flow rate was in- creased. This latter was related to the shear stress deact ivat ion of the immobilized enzymes due to the tangential interactions of the recycling substrate solu- tion flow with the external surface of the support , jz

Additionally, since the nylon net membrane has both external and internal surfaces, the shear stress phenomena at low recycling flow rates occurred at two different levels of magnitude on the immobilized pec- tinase system, which was supported by the two-step

deact ivat ion profiles obtained for recycling flow rates lower than 100 ml min -~. On the contrary, the increase of the recycling flow rate might have reduced the thickness of the l iquid-solid interface for the substrate diffusion, resulting in a high shear stress which was similar in both external and internal surfaces of the nylon net membrane , and, obviously, the deactivat ion profiles were shown in only one step.

In order to establish the deact ivat ion function of the immobilized derivat ive in the cross-flow reactor , only the first-step decay constants were used.

Figure 4 shows the decay constants as a function of the recycling flow rate (ln KD = 8.75 • l 0 -4 • Qr - 9.65; r = 0.995). Since the operational stabilities showed a first-order deactivat ion kinetics, the deact ivat ion func- tion, FD, could be expressed as follows'3:

FD = e-K~ ' t (9)

where KD is the decay constant (min -q and t is t ime (min).

By introducing the KD function from Figure 4 into equation (9), the deactivat ion function (10) could be expressed as:

F D = e K't'eJ'Qr (I0)

where K and J were empirical constants (K = e-965 = 6.44 • 10 -5 min i and J = 8.75 • 10 -4 min • ml i). In this equation, it is noticeable that KD tends to K when Qr approaches zero; then, K could be considered as the operational deactivat ion constant and the term e x p ( J • Q0 as a shear deactivat ion factor.

Design equation

If the convers ion of substrate (X) is defined as:

So - S dX -1 dS X - - - , then -

So dt So dt

This term can be introduced into the substra te-mass balance equation (3), as well as equations (6), (7), and (10), to obtain the design equation proposed for this cross-flow reactor, expressed as follows:

Q f • x

o

Vmi" Qr/v Qr/v + n" Vmi

• g " (1 - - X ) " e -K ' t ' eJ 'Qr

O " ( K M i " (1 + C " e -8Q'/v) + So • (I - X ) )

- d X - d t (I 1)

The integration of this equation allows us to study the effect of the different operat ional conditions on the convers ion profiles. In this way, Figure 5 (A-D) shows the substrate convers ion profiles versus t ime when only one of the four parameters (g, Qr, v, or Qf) was changed while the others were kept constant . As can be seen, by increasing the amount of derivat ive (g), the max imum level of convers ion was also in-

502 Enzyme Microb. Technol., 1990, vol. 12, July

Immobilized pectolytic enzymes: P. Lozano et al.

v

c o tn t _

>

o L.)

100

75

50

25

0 100

75

50

25

0

1.0 400

0.3 ~ 250

0.2 200

150

0.1

0.05

I |

50

50

B |

1

i ", f c D

i m |

0 2 4 0 2 4 6

I00 2

~ ! 125 4

150 6

C , 175 8 200 12

20 300

Time (h)

Figure 5 Simulated convers ion- t ime prof i les at: (A) Different quanti t ies of der ivat ive (g); v = 300 ml; Or = 300 ml rain 1. Of = 2 ml m i n - t (B) Dif ferent recycl ing f low rates (Or, ml min-1); v = 300 ml; Of = 2 ml ra in- l ; g = 0.5 g. (C) Different reaction v o l u m e (v, ml). Qr = 200 ml min-1; g = 0.5 g Qf = 12 ml min 1. (D) Different f i l trate f low rates (Of, ml ra in- l ) ; Qr = 300 ml rain ~; v = 300 ml; g = 0.5 g

creased, whereas the time necessary to obtain this maximum level was reduced (Figure 5A). A similar effect could be observed when the recycling flow rate was increased (Figure 5B), or when v was decreased (Figure 5C), as a consequence of the high dependence of Vm,p on Qr/v. However, if the filtrate flow rate was increased, the maximum conversion consequently de- creased (Figure 5D), because the residence time of the substrate solution in the reactor decreased. In spite of this latter, the time necessary to obtain the maximum conversion did not change, because it is only related to the catalytic capability of the system, which is con- trolled both by the recycling flow rate and the amount of enzyme derivative.

The design equation here proposed is adequate in view of the correlation between theoretical substrate conversion data and their respective experimental val- ues, obtained when the hydrolysis of pectin was fol- lowed every 8 h during 6 days of continuous operation, for all of the recycling flow rates assayed (Figure 6). The degree of correlation for each one of the four sets of values was better than -+6%, showing good agree- ment between the proposed design equation and the experimental behavior of the reactor.

Catalytic efficiency of the cross-flow reactor

The continuous processing of pectic substrate by the immobilized Pectinol D in the cross-flow and packed

bed reactors showed a viscosity drop greater than 50% during the experimentation time, a value largely suffi- cient for the product to be considered as clarified.t4

In order to test the catalytic efficiency of the nylon membrane with the immobilized pectolytic enzymes in the cross-flow reactor, the conversion capability, cal- culated as substrate conversion times the outlet flow per amount of immobilized derivative expressed in grams (X. Qf/g), was determined. This parameter was used to compare the performance of the cross-flow reactor with that of the immobilized derivative on either nylon membrane circles or pellets in a packed- bed reactor.

In the cross-flow reactor, the immobilized Pectinol D derivative showed the best pectic substrate conver- sion capability, as depicted in Figure 7. This is a con- sequence of the higher mass-transfer rate produced by the recycling flow rate that approximates the observed kinetic parameters to their intrinsic values, which were previously worked out in a batch reactor 7 and that are severely affected by the external diffusional limitations. Under such conditions, the catalytic effi- ciency of the immobilized biocatalysts is the highest they can exhibit.

Alternatively, when using the immobilized deriva- tive as nylon pellets in a packed-bed reactor, 6 the oper- ational stability (KD = 2.05 × 10 3 h l) was higher than that of the nylon membrane derivative in the cross- flow reactor, obviously due to the fact that the immo-

Enzyme Microb. Technol., 1990, vol. 12, July 503

Papers

Xtheor

Or= 50 "/ O0

60

t,0

20 0

80 Q r ~

6 ,~, of/ 20 • •

0 20 t,0 60 80 0

(~r=1~ /

. / 20 t,0 60 8'0

Xexp Figure 6 Correlat ion between theoret ical and exper imenta l substrate convers ions for the hydrolysis of pectins by the immo- bil ized pectolyt ic der ivat ive in the cross-f low reactor at di f ferent recycl ing f low rates (ml min-1), v = 300 ml; Qf = 0.6 ml min 1; g = 0.15-10.20 g

3°° X x°,,° f \ -

I ~ ' - - " ~ - - " - - - " - - - " - - - - -31

0 V : , -7 ~ , "~'" 0 1 2 3 4 5 6

Time {d )

Figure 7 Substrate convers ion capabi l i ty of the immobi l ized Pectinol D wi th t ime in: (O) cross-f low reactor, (Q, = 100 ml min 1); ( l l ) packed-bed reactor wi th membrane circles; (A) packed-bed reactor wi th pellets

bilized derivative as pellets was subjected to a lower shear stress. However , the use of the immobilized de- rivative membrane in the packed-bed reactor (placed perpendicularly to the liquid flow) showed an impor- tant deactivation phenomenon during the first 2 days of operation (KDI = 14.1 x 10 3 h-~), whereas in the following days, this deactivation effect decreased (Ko2 = 1.78 × 10 -3 h-l) .

Furthermore, in the packed-bed reactor, the immo- bilized pectolytic derivative on membrane configura-

tion showed a higher catalytic efficiency than that on nylon pellets. This latter could be a consequence of the low pressure drop into the reactor column when the membrane configuration was used. Given the high vis- cosity of the substrate solution, its flow through the biocatalyst bed causes a pressure drop that can en- hance the generation of preferential channeling in the pellet bed and, obviously, reduce the substrate conver- sion capability of the immobilized derivative.J5

Conclusions

Due to the high viscosity of the pectin solutions, the use of pectinases immobilized on a nylon membrane and arranged in a cross-flow system provides an effi- cient way to perform the pectin degradation process. The shear stress enzyme deactivation effect, produced by the tangential flow on the immobilized derivative in this bioreactor, is largely compensated by its high con- version capability. This finding appears to be very promising for continuous juice clarification on an in- dustrial scale basis.

Acknowledgements

Pedro Lozano is a fellow of Instituto de Fomento de la Comunidad Aut6noma de la Regi6n de Murcia, Spain. This work was partially supported by a CICYT grant No. BT87-0026. We thank Millipore Ib6rica S.A. for the gift of MINITAN S ®.

Nomenclature

FD Deactivation function g Quantity of derivative (g) KD Deactivation constant (min -]) gMa p Apparent Michael is -Menten constant (mg •

ml -l) gMi Intrinsic Michael is -Menten constant (rag •

ml -j) Qf Filtrate flow rate (ml • rain -1) Qr Recycling flow rate (ml • min l) r Reaction rate (mg S • min ~) S Pectic substrate concentrat ion at time t (rag •

ml -]) So Pectic substrate concentrat ion at time t = 0

(rag. m1-1) v Reaction volume (ml) Vmap Apparent maximum velocity (mg S • min -] •

g-~ I ME) Vmi Intrinsic maximum velocity (rag S • min 1 . g-]

I ME)

References 1 Buchholz, K. and Klein, J. Methods Enzymol 1987, 135, 3-30 2 Kminkovfi, M. and Kucerfi, J. Enzyme Mierob. Technol. 1983,

5, 204-208

504 Enzyme Microb. Technol. , 1990, vol. 12, Ju ly

3 Chang, H. N. Biotechnol. Adv. 1987, S, 129-145 4 Lopez-Leiva, M. and Gekas, V. Process Biochem. 1986, 21,

27-29 5 Adu-Amankawa, B. and Constantinides, A. Biotechnol.

Bioeng. 1984, 26, 156-166 6 Lozano, P., Manj6n, A., Romojaro, F. and lborra, J. L. Pro-

cess Biochem. 1988, 23, 75-78 7 Lozano, P., Manj6n, A., Romojaro, F., Canovas, M. and

lborra, J. L. Biotechnol. Lett. 1987, 9, 875-880 8 Thompson, R. Q., Mandoke, C. S. and Womack, J. P. Anal.

Lett. 1985, 18, 93-107 9 Howell, J. A., Merion, G. J., Klu, M. K. and Khan, A. R. in

Immobilized pectolytic enzymes: P. Lozano et al.

Waste Treatment and Utilization (Moo-Young, M. and Farqu- har, G. J., eds.) New York, Pergamon Press 1978, pp. 395-420

10 Butcher, J. C. J. Austral. Math. Soc. 1964, 4, 179-185 11 Lilly, M. D. and Dunnill, P. Methods. Enzymol. 1976, 44, 717-

738 12 Charm, S. E. and Wong, B. L. Enzyme Microb. Technol. 1981,

3, 111-118 13 Henley, J. P. and Sadana, A. Enzyme Microb. Technol. 1985,

7, 50-60 14 Neubeck, C. E. in Enzymes in Food Processing (Reed, G.,

ed.~ Academic Press, New York, 1975, pp. 397-442 15 Kim, K. H. and Chang, H. N. Biotechnol. Bioeng. 1986, 28,

452-455

Enzyme Microb. Technol., 1990, vol. 12, July 505