Deacylation of benzylpenicillin by immobilized penicillin acylase in a continuous four-stage stirred-tank reactor

Deacylation of Benzylpenicillin by Immobilized Penicillin Acylase in a Continuous Four-Stage

Stirred-Tank Reactor

S. W. CARLEYSMITH* and M. D. LILLY, Depurtment of Chemical and Biochernicul Engineering, University College

London, Torrington Place, London, W C l E 7JE England

Summary Immobilized penicillin acylase has been used for the deacylation of benzylpenicillin

at 37°C in a continuous reactor consisting of four I liter stirred tanks connected in series. There was good agreement between the predicted and actual conversions obtained in each tank under steady-state conditions. The operational stability of the immobilized enzyme in the tanks depended on the pH and the rate of addition and concentration of alkali needed to neutralize the acid produced during the reaction. At pH 7 with the addition of 2M NaOH, the half-life for enzyme stability was greater than 400 hr in all tanks. This was over half the value for the immobilized enzyme when stored at 37°C and pH 7.

INTRODUCTION

Penicillin acylase (EC 3.5.1.1 1) from Escherichiu coli hydrolyzes benzylpenicillin to yield 6-aminopenicillanic acid (6-APA), which is used for the synthesis of other penicillins. The commercial pro- duction of 6-aminopenicillanic acid using E.coli and later immobilized penicillin acylase2 in batch operated stirred-tank reactors has been described. The kinetic behaviors of small batch3 and continuous-flow stirred-tank3f4 reactors have been reported. For all other commercial processes at present being operated with immobilized cells or enzymes, packed beds are used thereby exploiting the advantages of continuous operation. Unless operated under recycle conditions2 with very high flow rates through the bed or with high buffer concentrations, column systems are unsuitabie for the deacylation process because of the formation of an acidic product which is dissociated at the reaction pH. One solution is to carry out

* Present address: Beecham Pharmaceuticals ( U . K . Division). Worthing. Eng- land.

Biotechnology and Bioengineering, Vol. XXI, Pp. 1057- 1073 (1979) @ 1979 John Wiley & Sons, Inc. 0006-3592/79/002 I - I057$0 I .OO

1058 CARLEYSMITH AND LILLY

repeated batch reactions in a stirred tank. Continuous-flow operation with a single stirred tank is much less favorable than batch operation because the reaction is subject to severe product inhibition.6-8 In this paper we describe the operation of a continuous- flow system consisting of four stirred tanks in series, which although slightly less favorable kinetically than the batch system, does have the advantage of continuous operation. Since little information has been published on the operational stabilities of immobilized penicillin acylase preparations, we have also obtained data on the stability of our preparation during operation of the continuous-flow reactor.

THEORY

Calculation of the Fractional Conversion in a Series of Well- Stirred Tank Reactors

Consider the passage of a reactant solution through a series of n well-stirred tank reactors containing fixed weights of immobilized enzyme catalyst. The fraction of the reactant converted in each tank and the overall conversion can be found by a graphical procedure similar to that described by L e ~ e n s p i e l , ~

where F is the flow rate of the substrate solution (m3/sec); So is the substrate concentration entering the first reactor (moUliter); S, is the substrate concentration leaving the nth reactor (mol/liter); V , is the reference activity of immobilized enzyme (usually specific activity when S + 00) (katikg); V , is the activity of the immobilized enzyme in conditions of nth reactor (kat/kg); v, is the dimensionless activity of immobilized enzyme in the nth reactor = V,/V,; W , is the weight of the immobilized enzyme in nth reactor (kg); X,-, is the fractional conversion of substrate entering the nth reactor (= (So - S,-l)/S,,). The substrate concentration depletion in nth reactor is Snp1 - S, and equals the total reaction rate in that reactor divided by the substrate flow rate, i.e., Sn-, - S, = W,V,/I03F.

Therefore

V , / S , - S,-, = 103F/W, (1)

IMMOBILIZED PENICILLIN ACYLASE REACTOR 1059

Substituting dimensionless variables vn and X n , eq. ( I ) becomes

Vn/Xn - Xn-1 = lPFS, /Wn V m (2 1 This equation allows the conversion in each reactor of the series to be found by a graphical procedure using a dimensionless reaction velocity-conversion profile. An experimental batch progress curve ( X vs. time) for the deacylation of benzylpenicillin (0.1M) at 37°C in the presence of 0.1M phosphate buffer, pH 8, by penicillin acylase immobilized to XAD-7 particles was used to construct the velocity- conversion profile shown in Figure 1. When a line is drawn from the origin with a slope v , / ( X , - X , ) it cuts the curve at a point (A) , which represents the conditions in the first tank. A perpendicular line from A cuts the X-axis at B, which represents the condition of the feed to the second tank. A line is drawn from B with a slope v,/ ( X , - X I ) and cuts the curve at C, which represents the conditions in the second tank. The graphical procedure is continued until X n is found for the nth tank (n = 4 in Fig. 1 ) .

At the start of this present study, the graphical method was used in reverse to find the amount of activity in each of the tanks required to achieve particular fractional conversions. The total enzyme activity required was expressed as a multiple of the amount of activity

' ' O h 0 8

0 0.2 0.4 0.6 0.8 10 CONVERSION ( X )

Fig. I . Construction for the prediction of the fractional conversion in the multistage reactor. Each datum point is the ratio of the initial rate I,,, to the velocity I, ,

(both rates corrected for nonspecific hydrolysis) at conversion X . Ratios were determined from a batch progress curve (X vs. time) for the deacylation of benzylpenicillin (0. I M ) in the presence of phosphate buffer (0. IM) catalyzed by immobilized penicillin acylase at pH 8 and 37°C.


required to achieve the same fractional conversion in a batch stirred- tank reactor with no down-time (Fig. 2). The velocity-conversion profile was calculated from the kinetic equation of Warburton et

using values for the kinetic constants determined previously.* The value for the product inhibition constant for 6-aminopenicillanic acid was 21mM although a higher value was shown later to be more accurate. In practice the down-time (the time from the end of one batch reaction to the start of the next) for a batch reactor is a significant proportion of the total time. For example a batch reactor with a down-time of ?4 of the total time will require an additional 50% of enzyme for a given output per unit time. In such a case a multistage continuous-flow system can be more efficient in the use of enzyme activity. From the data in Figure 2 and other consider- ations, such as construction costs, we decided to build a continuous reactor consisting of four stirred tanks in series.

MATERIAL AND METHODS

Penicillin acylase and benzylpenicillin were supplied by Beecham Pharmaceuticals ( U . K . Division). All other chemicals were of lab- oratory grade. Immobilized penicillin acylase was prepared by cov-

t 4 W I w Ir 3 a W

m 3 W

f N z W

1 2 3 4 5 6 7 8 NUMBER OF TANKS")

Fig. 2. Effect of the number of stages ( N ) of a multistage continuous stirred reactor on the multiple excess penicillin acylase requirement compared with a batch reactor with no down-time (unit enzyme requirement). for different degrees of conversion X of benzylpenicillin solution (0 .168M). ([?) X = 0.99; (A) X = 0.95; (0) X = 0.90.


alent binding of the enzyme to Amberlite XAD-7 beads (Rohm and Haas (UK) Ltd., Croydon, England).8 Both soluble and immobilized penicillin acylase were assayed in a pH-stat as described previ- ou sly.

The fractional conversions of benzylpenicillin in the continuous- flow reactors were determined by enzymatic measurement of the unconverted benzylpenicillin. A known volume of sample from the reactor was transferred to the pH-stat reaction vessel, made up to about 25 ml with water and allowed to equilibrate to pH 8 and 37°C. A small volume of extensively dialyzed enzyme solution, previously equilibrated to the same conditions, was introduced into the reactor. Sufficient enzyme activity was added that the reaction was complete in 10 min. The benzylpenicillin concentration in the sample was calculated from the alkali uptake corrected for background hydrolysis measured before and after enzyme action. A control sample of the benzylpenicillin feed to the first reactor was assayed similarly. From these two assays the fractional conversion of benzylpenicillin in the reactor sample was calculated. The assay is especially effec- tive at high conversions because it directly determines the amount of unconverted material, a quantity which would be determined less exactly as a difference value if a colorimetric assay for the product, 6-aminopenicillanic acid, were used.

CONSTRUCTION AND OPERATION OF THE CONTINUOUS- FLOW REACTOR

The reactor was designed for continuous upward flow of reactant through four stirred tanks in series, each fitted with seals and filters to retain the immobilized enzyme and each maintained at a constant pH by addition of alkali (Fig. 3). The construction of one tank is shown in Figure 4. Each tank, 1060 & 2.5 ml volume, was fitted with a Perspex two-bladed impeller, each blade (16 x 6 mm) slightly inclined to promote axial circulation, mounted on a single overdri- ven stainless-steel shaft. Temperature control was by rapid recir- culation of water from a heated reservoir through a heating coil in each reactor. The temperature in each tank was measured with a thermocouple switched to an electronic thermometer. The maximum temperature difference between reactors was 0.5"C and the long-term temperature stability was better than 2 0.1 "C. Each reactor was fitted with an EIL or Pye combined pH electrode connected to a pH controller (model 9904, Electronic Instruments Ltd, Chertsey, Surrey, England). Alkali solution (normally 5N NaOH)


Fig. 3. Multistage reactor. Four tanks each contained 0.1 kg immobilized penicillin acylase. Perspex impellers and stainless-steel heating coils can be seen inside the tanks.

was pumped by a peristaltic pump (Minipump model 111, Schuco International London Ltd.) through a stainless-steel syringe needle into the impeller discharge region in the reactor.

The benzylpenicillin feed solution was stored in a stainless-steel reservoir (20 liter) fitted with a cooling coil to maintain the reservoir contents at 0°C. The feed solution was pumped by a DCL metering pump via a coil in the heated reservoir to raise the temperature to 37°C through a Propyltex 25 mm filter into the bottom reactor. Fittings for three of these filters had been placed in the plates between each reactor. For the large immobilized enzyme particles used (420-850 pm) the pressure drop across the filters was negligible


and only one filter was used therefore between reactors. The filter material was Propyltex polypropylene monofilament fabric (Swiss Silk Bolting Co. Ltd, Switzerland) with a thread diameter of 150 pm, mesh opening of 210 pm, 27.8 mesh/cm and a 34% free surface. The mean liquid flow across the filter free surface at operational feed rates was 0.7 to 2.7 mm/sec.

Fig. 4. Sectional elevation of cylindrical stirred tank reactor. Four-stage reactor comprised four tanks mounted vertically and clamped together by tie-bars between plates X . Dimensions in mm. B, flow of reactant solution; D, drain port; F. filter assembly; H, heating coil; 1. impeller (2 blades); P, pH probe; S, sample port; T, thermocouple; X, bottom plate.

1064 CARLEYSMITH AND LlLLY

TABLE I Effect of Protein Concentration on

Stability of Penicillin Acylase”

Half-life of Protein enzyme conc. activity

(giliter) (hr)

0.056 60 2.0 270 2 . 8 320

XADPAb 600

a Conditions: pH 7.35. 37°C. phos-

I, Immobilized preparation (XAD7 phate buffer.

penicillin acylase).

RESULTS

Storugc S;ubilitirs of Soluble und Immobilized Penicillin Acylase

As with many enzymes, the stability of penicillin acylase when stored in solution depended markedly on the protein concentration (Table I), and on the pH of the solution (Fig. 5 ) . Maximum stability was observed at about pH 6.5. Stability is poorer at pH 8 , the optimum for activity.

During deacylation of benzylpenicillin by penicillin acylase, alkali must be added to maintain a constant pH. Thus the pH stability of

3 L 5 6 7 0 9 10 PH

Fig. S. Fraction of initial activity ( ~ t i r , ~ ) measured after the storage of soluble penicillin acylase (0.4 mgiml) at 37°C for I hr (open symbols) and 24 hr (closed symbols) with 10 mM buffer concentration. (0.0) Citrate; (&A) phosphate; (0.m) bicarbonate.


TIME (hours)

Fig. 6 . Fraction (i. ,/v,) of the initial activity ( v , ) of immobilized penicillin acylase remaining after time t of the following treatments: (v) storage in deionized water at 10°C: (a) stirring at pH 5 and 37°C in 0. I M phosphate buffer; repeated batch partial conversions of 50 ml benzylpenicillin: (c) 0. IM. pH 8.0; (A) O.SM, pH 8 .35 , at 37°C in the presence of phosphate buffer (0. I M ) and using the addition of 5N alkali for pH control. Reactor conditions were chosen as not those for the most stable oper- ations of the enzyme so that a large decay of activity was measurable in a short period of time.

the immobilized penicillin acylase is an important factor in reactor design and operation. Loss of activity by immobilized penicillin acylase when stored as a suspension was first order with respect to time (Fig. 6), and so stabilities were expressed as half-lives. The influence of pH on the activity half-life (t l ,J is shown in Figure 7. The profile had a symmetrical shape about an optimum of pH 6.6

4 5 6 7 8 9 1 0 PH

Fig. 7. Effect of pH on the half-life of the activity of immobilized penicillin acylase during storage at 37°C in 0.5 M phosphate buffer. Continuous line is a theoretical plot calculated from eq. (3) with K , = 2.0 X K , = 2 .5 X c = 1.224 and t;;" = 943 hr.


and could be fitted by a simple equation,

tf ly c 1 + H / K , + K 2 / H tl lZ =

where t$;x = 943 hr, c = 1.224, K , = 2.0 x 10-6M, K , = 2.5 x IO-*M, and H is the hydrogen ion concentration ( M ) .

The dependence of the stability of immobilized penicillin acylase on temperature was not large compared with many enzymes. The half-lives at 37, 2 I , and 10°C (under otherwise similar conditions and neutral pH) were 85, 43 I , and about 655 hr, respectively. From these data, the activation energy for the denaturation reaction was calculated to be 14-20 kcal/mol.

Performance of the Four-Stuge Stirrc.d-Tunk Reuctor

In the Theory section of this paper we showed how the predicted conversions in each of the four stages of the reactor could be calculated using a velocity-conversion curve obtained from batch reaction data. The actual behavior was determined by measuring the degree of conversion in each stirred tank during the initial period of operation of the reactor. An example using a benzylpenicillin feed concentration of 0.1M is shown in Figure 8. The immobilized

0 4 l I I I 0 1 2 3

TIME (hours)

Fig. 8. Degree of conversion of benzylpenicillin (0. I M ) in each of the four tanks: ((0) I ; (0) 2; (A) 3; (0) 4) of the multistage continuous reactor at time t after starting the reactor. Immobilized penicillin acylase: 0. I kghank, specific activity (in these conditions but in the absence of product) about 0.85 mkatikg. Reactor conditions: benzylpenicillin feed pumping rate: 56.4 ? 0.6 ml/min; phosphate buffer present (O . IM) , pH 8; 37°C; impeller speed 700 rpm; pH control by addition of 5N NaOH solution.


TABLE I1 Comparison of Predicted and Observed Conversions in the Four-Tank Continuous

Reactor

Benzylpenicillin concentration (M) of feed

0. I 0.5 Tank No. predicted observed predicted observed

I 0.52 0.54 0.2 1 0.22 2 0.80 0.8 I 0.36 0.37 3 0.92 0.91 0.48 0.47 4 0.97 0.96 0.58 0.57

enzyme activities that were used for calculation of the predicted conversions were those measured immediately after stopping the four-stage reactor. The predicted and observed results were in good agreement (Table 11) indicating that each stage was behaving as a well-mixed reactor.

Operutional Stability of Immobilized Penicillin Acylase

Thermal and pH denaturation under storage conditions have been shown to be first order with respect to time. When the immobilized enzyme was used repeatedly for batch reactions in a small reactor, the loss of activity was first order with respect to time although a small rapid loss of activity was observed occasionally at the start of use of a new immobilized enzyme preparation (Fig. 5) . Opera- tional stabilities for the immobilized enzyme in the four-stage continuous-flow reactor were expressed as half-lives calculated from activity assays of samples taken before and after a period of continuous operation. Some values for stabilities are listed in Table 111. Indirect estimates of the half-lives calculated from the decrease in conversion in each reactor were inaccurate except for the first tank because of the cumulative effects through the tanks of the decay in activities. Table 111 shows that in all four stages the stabilities of the immobilized enzyme were much less than that measured under the corresponding storage conditions (320 hr). This large decrease in stability under operational conditions was greatest in the first tank where the rate of alkali addition was highest. Thus denaturation by added alkali seemed to be the main cause and this problem was investigated further.

First the effects of stirrer speed and buffer concentration were examined in a single-stage reactor. For these experiments, substrate was omitted from the reactor. Instead, hydrochloric acid was

1068 CARLEYSMITH A N D LlLLY

TABLE 111 Operational Stability of Immobilized Penicillin Acylase in the Four-Tank

Continuous Reactor

Relative Approximate degree Enzyme half-life mean rate of

Tank No. of conversion ( h r ) alkali addition

0.1.44 Bc,nzylpc,nicillin I 0.4 2 0.65 3 0.8 4 0.9

0.5M Brnzylpc~nicillin I 0.22 2 0.36 3 0.47 4 0.57

23 32 63

> 130

29 26 28 37

1 0.6 0.4 0.25

I .06 0.8 0.65 0.5

pumped into the impeller discharge at a rate that was an approximate molar equivalent of the acid that would have been generated in the first tank during deacylase reaction using 0.1 kg immobilized enzyme. A solution of phosphate buffer, pH 8 , was pumped through the reactor and the pH controlled by addition of 5M NaOH. The losses of enzyme activity after 5 hr at two stirrer speeds and two buffer concentrations are shown in Table IV.

The operational stability of the immobilized enzyme in the presence of 0.1M phosphate, initially at pH 8, but without alkali addition was then examined in the four-stage reactor. After continuous operation for 3 hr without pH control, a feed concentration of O.1M and a stirrer speed of 700 rpm, the pH values in the four tanks were 6.35, 6.45, 6.62, and 7.00 and the conversions were 0.53, 0.72, 0.77,

TABLE IV Loss of Activity of Immobilized Penicillin Acylase by Inactivation with Alkali"

Stirrer speed Phosphate buffer concentration (M) (rpm) 0. I 0.5

700 I400

2 6% < I %

5% -

a Operating conditions: 5 hr in a I liter stirred-tank reactor fed continuously with 3M HCI at 1.73 mlimin and controlled at pH 8 by addition of 5M NaOH. Losses are corrected for thermal decay.

IMMOBlLIZED PENlClLLIN ACYLASE REACTOR 1069

and 0.83. Under these conditions the half-lives of the enzyme in all tanks were greater than the limit of measurement (300 hr).

In a final set of experiments four changes were made to the original operating conditions in an attempt to increase the operational stability of the immobilized enzyme: 1) the pH in the reactor was reduced to 7.5 or to 7.0; 2) the stirrer speed was increased to 1400 rpm; 3) the alkali for pH control was diluted to 2M, and 4) the pH and temperature control in each tank was not started until all four tanks were filled, rather than switched on as individual tanks filled as had been done previously. After an initial stabilization period, the conversion became constant in each tank (Fig. 9). The immobilized enzyme activity in each tank was determined before and after the continuous runs and the half-life calculated. At both pH values the enzyme stability was much improved compared to operation at pH 8 (Table V). Unlike the former method of operation at pH 8 when denaturation was greatest in the first tank, there was no significant difference in the stabilities of the enzyme in each of the four tanks.

10

0.8

- X

Z -

06 (L W > Z 0 U

OLI

2 3 4 17 i a 19 20

TIME (hours)

Fig. 9. Degree of conversion X of benzylpenicillin (O . IM) in each of the four tanks ((0) I ; (v) 2: (A) 3; (c) 4) of the multistage continuous reactor at time t after starting operation. Immobilized penicillin acylase: 0. I kgitank. specific activity (in these conditions but in the absence of product) about 0.87 mkatikg. Reaction conditions; benzylpenicillin feed pumping rate: 30.4 ? 0.5 mlimin; phosphate buffer present at 0. IM; pH 7.5: 37°C: impeller speed 1400 rpm; pH control by addition of 2N NaOH solution.


TABLE V Operational Stability of Immobilized Penicillin Acylase at Three pH Values

Stirrer Alkali Half-life ( t l iz) of Operational speed conc. enzyme activity

PH (rpm) ( M ) (hr)

8 700 5 23-130 7.5 I400 2 290 7.0 I400 2 > 400

DISCUSSION

In this paper we have demonstrated the operation of a four-stage stirred-tank reactor for the deacylation of benzylpenicillin by immobilized enzyme. The good agreement between the measured conversion values in each tank (Table 11) and those predicted from batch show that each tank was behaving as a well-mixed reactor. With the immobilized enzyme particles used (420-850 pm diam) there was no problem in retaining the particles in each reactor especially as little attrition occurred after several days operation.

As indicated in the Theoretical section a four-stage reactor may be more efficient in terms of utilization of enzyme activity than a batch reactor when down-time is taken into account. Nevertheless in the continuous reactor it is impossible with a reasonable number of tanks to achieve complete conversion of benzylpenicillin although conversions up to 0.98 were observed. Savings in labor and running costs may compensate partly for the small loss of unconverted substrate.

In our experiments the continuous reactor was operated with about 1% wt/vol catalyst content. In an industrial reactor higher catalyst contents could be used but the choice would depend on several factors. These include the need to maintain good mixing, the minimization of damage by attrition, and the desired reduction of the residence time of the labile benzylpenicillin and 6-APA in the reactor. An increased reaction rate means more rapid acid produc- tion and therefore the need to add alkali faster. Our experiments show that alkali addition is a major factor in the operational stability of the immobilized penicillin acylase.

One of the most important aspects of any immobilized enzyme system is the operational stability of the enzyme. Immobilization of penicillin acylase increased its stability when stored at 37°C (Table


1). The influence of temperature on the stability of immobilized penicillin acylase was less than that observed for many other en- zymes1° and suggests a less complex mechanism than protein denaturation for the loss of activity of penicillin acylase. In practice, the optimum temperature for the deacylation process is determined not by the stability of the enzyme but by the stability of benzylpenicillin and to a lesser extent 6-APA.

Immobilized penicillin acylase is most stable at pH 6.6 (Fig. 7) but i t is necessary to operate the reaction at a higher pH to achieve maximum conversion since the reaction is reversible. In a batch reaction the pH profile within the particles will change with time. In the absence of added buffer the intraparticle pH fell toward a minimum at about pH 4.5 . s At this pH the predicted half-life for activity from eq. (3) was 70 hr. The minimum pH that can be attained inside a particle rises as the conversion proceeds and the reaction rate falls. The products formed not only cause a slowing of the reaction rate but also act as buffers. Only a small rise in the intraparticle pH is necessary for a large increase in stability (Fig. 7). A rise of 1.25 pH units above pH 4.5 increases the predicted half-life tenfold. When high buffer concentrations were used in the reactor there was not a large pH depression in the particles and the stability of the immobilized enzyme could be predicted from eq. (3) using the external pH value.

In a continuous multitank reactor the pH depression in the particles will be most severe in the first stage, where the reactant conversion is lowest and the rate of reaction is highest. However, in the four-tank reactor that we used the conversion in the first tank was normally greater than 0.5. Under these conditions the minimum pH that can be attained in the particles is greater than 5.5 for a benzylpenicillin feed concentration of 0 . M . Above pH 5.5 the intrinsic stability of the enzyme is good. Furthermore, the self- buffering effect of the products will tend to suppress the pH depression. The enzyme was therefore not expected to be destabilized by the internal pH depression in the operation of any of the tanks of the continuous reactor.

The operational stability of the immobilized enzyme in the four- stage reactor without added buffer was much less than expected on the basis of the intrinsic storage stability. In the presence of a large concentration of phosphate buffer ( 0 . l M ) the stability was higher but still less than expected (Table 111). Under operational conditions the main reason for loss of activity was inactivation by alkali added to maintain a constant pH. This was confirmed in several ways.


First there was a correlation between the rate of addition of alkali and the loss of activity of immobilized enzyme (Table 111). In the absence of pH control and therefore no alkali addition the immobilized enzyme stability was good and approached the intrinsic stability. The good stability in this experiment showed that stirring at 700 rpm had no deleterious effect on the enzyme stability. In the experiments (Table IV) where acid was added to simulate the enzymatic reaction the loss of enzyme activity was decreased both by increased stirrer speed and buffer concentration. Since the results at high stirrer speed were satisfactory, this parameter was not investigated further. The agitation speed will be limited by the rate of attrition caused by the agitator. For a given tip speed the mixing time in the reactor will increase with reactor size. Thus one advantage of a multistage reactor will be the smaller volumes in each reactor.

In the final experiments (Table V) it was possible to achieve very good operational enzyme stabilities in each of the four stages of the reactor. The extent of the possible changes to the reactor conditions is limited. The use of a lower alkali concentration causes a greater dilution of the product. For total conversion of O.1M benzylpenicillin addition of 5M and 2M NaOH for pH control dilute the product by 2 and 5%, respectively. In each reactor the alkali was pumped into the discharge (jet stream) of the impeller, which has been shown experimentally to be the best position for rapid mixing. l1

The use of pH 7.0 or 7.5 instead of 8.0 means a reduction of the initial reaction rate in the presence of phosphate buffer of 16 and 8%, respectively. Another effect of lower pH is a reduction in the maximum conversion that can be reached. From the equilibrium constant for the conversion the equilibrium values for conversion at pH 7 were calculated to be 0.996,0.97, and 0.87 for corresponding initial benzylpenicillin concentrations of 0.01, 0. I , and 0.5M. In Figure 9 a conversion of 0.98 was obtained in the fourth stage when operating at pH 7.5. Further improvements could be made by con- trolling the tanks at different pH values.

All of these modifications to the operating conditions are con- cerned with reducing the loss of activity by exposure to alkaline conditions. It is possible to prepare particles with the enzyme immobilized not near the surface as in the present case but at a defined depth. Such preparations may be more stable toward denaturation by alkali.I2

The authors wish to thank Beecham Pharmaceuticals and the Science Research Council for their support of this work.


References

I . T. R. Carrington, Proc Roy. Soc. Lond. B , 179, 321 (1971). 2. E. Lagerlof, L. Nathorst-Westfelt, B. Ekstrom, and B. Sjoberg, in Methods

3. D. Warburton, P. Dunnill, and M. D. Lilly,Biotechnol. Bioeng., 15, 13 (1973). 4. D. Y. Ryu, C. F. Bruno, B. K. Lee, and K. Venkatasubramanian, in Fer-

mentation Technology Today, G. Terui, Ed. (Society of Fermentation Technology, Tokyo, 1972), p. 307.

5 . M. D. Lilly, in Biotechnological Applications of Proteins and Enzymes, Z. Bohak and N. Sharon, Eds. (Academic, New York, 1977), p. 127.

6 . K. Balasingham, D. Warburton, P. Dunnill, and M. D. Lilly, Biochim. Bio- phys. Acta, 276, 250 (1973).

7. C. Kutzbach and E. Rauenbusch, Hoppe-Seyler's Z . Physiol. Chem., 354, 45 ( 1974).

8. S. Carleysmith, P. Dunnill, and M. D. Lilly, Biotechnol. Bioeng., submitted for publication.

9. 0. Levenspiel, Chemical Reaction Engineering, 2nd ed. (Wiley International, New York, 1972), p. 141.

10. K. J . Laidler and P. S. Bunting, The Chemical Kinetics of Enzyme Action, 2nd ed. (Claredon, Oxford, 1973).

11. E. Paul and R. E. TreybaLAIChE. J . , 17, 718 (1971). 12. S . Carleysmith, M. Eames, and M. D. Lilly,Biotechnol. Bioengng., submitted

in Enzymology, K. Mosbach, Ed. (Academic, New York, 1976), Vol. 44, p. 759.

for publication.

Accepted for Publication September 27, 1978

Deacylation of benzylpenicillin by immobilized penicillin acylase in a continuous four-stage stirred-tank reactor

Documents