Production of D-Phenylglycine fromRacemic (D,L)-Phenylglycine ViaIsoelectrically-Trapped PenicillinG Acylase

Alessandra Bossi, Marina Cretich, Pier Giorgio Righetti

University of Verona, Department of Industrial and AgriculturalBiotechnologies, Strada Le Grazie, Ca’ Vignal, 37100 Verona, Italy;telephone/fax: 045-8098901; e-mail: RIGHETTI@IMIUCCA.CSI.UNIMI.IT

Received 13 August 1997; accepted 30 April 1998

Abstract: Penicillin G acylase (PGA) is exploited for pro-ducing pure D-phenylglycine from a racemate mixture,via an acylation reaction onto a cosubstrate, the estermethyl-4-hydroxyphenyl acetate. The reaction, when car-ried in a batch, is severely hampered by the reverse pro-cess, by which the product, 4-hydroxyphenylacetyl-(L)phenyl glycine, upon consumption of L-phenylglycine,is converted by the enzyme back into free substrate and4-hydroxyphenyl acetic acid via lysis of the amido bond.To prevent this noxious reaction, a multicompartmentelectrolyzer with isoelectric membranes (MIER) is usedas enzyme reactor, operating in an electric field. PGA istrapped between pI 5.5 and pI 10.5 membranes, togetherwith an amphoteric, isoelectric buffer (lysine). As the 4-hydroxyphenylacetyl-(L)phenyl glycine product isformed, it vacates the reaction chamber by electropho-retic transport and is collected close to the anode, in achamber delimited by pI 2.5 and 4.0 membranes. Thesame fate occurs to the free acid 4-hydroxyphenyl aceticacid, formed upon spontaneous (and enzyme-driven) hy-drolysis of the methyl ester in the reaction chamber.These combined processes leave behind, in the enzymereaction chamber, the desired product, pure D-phenylglycine. The advantages of the MIER reactor overbatch operations: the consumption of the L-form in theracemate is driven to completion and the enzyme is keptin a highly stable form, maintaining 100% activity afterone day of operation, during which time the PGA en-zyme, in the batch reactor, has already lost >75% cata-lytic activity. © 1998 John Wiley & Sons, Inc. Biotechnol Bio-eng 60: 454–461, 1998.Keywords: isoelectric membranes; immobilized reactors;penicillin G acylase

INTRODUCTION

The search for efficient methods for the production of op-tically active compounds is a very important goal in thechemical industry. Optically active D-aminoacids arewidely used in the pharmaceutical industry as intermediatesfor the synthesis of semisynthetic antibiotics, pesticides, andvarious compounds (Kim and Kim, 1993; Rosell et al.,1993, 1995; Sheldon et al., 1985). Among them, the demand

for D-para-hydroxy-phenylglycine and D-phenylglycinewill be significantly increased for the synthesis of newdrugs such as aspoxicillin, cefbuperazone, and cefpyramide(Kim and Kim, 1993). Penicillin G acylase (PGA) (E.C.3.5.1.11) catalyzes acylation/deacylation reactions by ac-cepting a broad variety of nucleophiles. This enzyme selec-tively transfers theN-phenylacetyl moiety of benzylpenicil-lin to water yielding 6-amino-penicillanic acid (6-APA)(Fuganti et al., 1986), which is the basic raw material forindustrial production of semisynthetic penicillins. PenicillinG acylase has been used for obtaining optically active com-pounds by means of phenylacetyl derivative hydrolysis ofinteresting compounds.

Zmijewski Jr. et al. (1991) used for the first time theenzyme in the synthetic direction: They opted to acylate thesubstrate, by assuming that the precipitation of water in-soluble products would be the driving force for the acylationand would protect the product from hydrolysis by the en-zyme. Recently, Rosell et al. (1995) proposed an alternativeroute for directly resolving unmodified racemic mixtures byway of synthetic reactions catalyzed by a stabilized deriva-tive of PGA. The PGA immobilization-stabilization con-sisted in multipoint covalent attachment to agarose gels (Al-varo et al., 1990).

Whereas enzyme immobilization has been typically ac-complished by chemical and/or physical attachment to solidsurfaces (Gekas, 1986; Gerhartz, 1990), there is an increas-ing interest in developing membrane bioreactors (Prazeresand Cabral, 1994), in particular those in which the drivingforce for transport of solutes is an electric potential (Lee andHong, 1987; Furusaki et al., 1990). Thus, Ishimura and Suga(1992) proposed a reactor based on the coupling of electro-dialysis and immobilized PGA. The reason for this experi-mental set-up is that PGA is severely product inhibited:Phenylacetic acid (PAA) inhibits competitively the enzy-matic hydrolysis. It is thus, important and useful to maintainthe concentration of PAA at low levels by removing it fromthe reaction mixture continuously.

A few years ago, we introduced multicompartment elec-Correspondence to:Pier Giorgio Righetti

© 1998 John Wiley & Sons, Inc. CCC 0006-3592/98/040454-08

trolyzers (ME) with isoelectric membranes for processinglarge volumes and amounts of proteins to homogeneity(Righetti et al., 1989, 1990). With this technique, the com-pound to be purified is always kept in a liquid vein (thus, itis not lost by adsorption onto surfaces, as customary inchromatographic procedures), and it is trapped into a cham-ber delimited by two membranes having pIs encompassingthe pI value of the protein being purified. Thus, by a con-tinuous titration process, all other impurities, either non-isoelectric or having different pI values, are forced to leavethe chamber, in which the protein of interest will ultimatelybe present as the sole species, characterized by being iso-electric and isoionic as well. Recently, it occurred to us thatthe ME could be used as an enzyme reactor: The enzymecould be trapped between two isoelectric membranes andcharged reaction products could be removed by the electricfield. As the ME is an open system, it offers the opportunityof developing continuous processes. The first example of anenzyme reactor based on the ME (nicknamed with the ac-ronym MIER: multicompartment immobilized enzyme re-actor) was proposed by Chiari et al. (1996), for transformingdehydrocholic acid into 3b-hydroxy-7,12-dioxo-5b-cholanoic acid, via an enzymatic reaction catalysed byb-hydroxysteroid dehydrogenase. In a subsequent work, Nem-bri et al. (1997) applied the MIER technology to the studyof the reaction catalyzed by urease. In a batch reactor, ureaamido hydrolase is strongly product-inhibited and quicklyinactivated by pH changes engendered by the developingammonia. On the contrary, in the MIER apparatus, the samereaction can be sustained for long periods of time whilemaintaining enzyme integrity and ensuring > 90% conver-sion of the substrate into the desired product. In yet anotherapplication, Righetti et al. (1997) demonstrated the possi-bility of continuous enzymatic hydrolysis ofb-casein in theMIER reactor, while simultaneously recovering, in a pureform, three major peptides of industrial and pharmaceuticalinterest. In this last work, an additional selectivity principlewas applied during MIER operation because only trypsincould be blocked into the isoelectric trap, whereas caseinhad a much lower pI value, the latter protein was capturedinside the reaction chamber by adding a sieving membraneto the anodal side of the trap. Whereas casein was effec-tively blocked in the trap, the produced peptides were per-meable to the sieve and could be recovered, by an isoelectricmechanism, in adjacent, anodal chambers.

In the present article, we apply the MIER technology tothe resolution of a racemic mixture of phenylglycine by asynthetic reaction continuously catalyzed by PGA. In thisreaction, PGA performs a stereo-specific addition of L-phenylglycine to the acceptor methyl-4-hydroxyphenyl ac-etate. The product [4-hydroxyphenylacetyl-(L)phenyl gly-cine], being a free acid, is continuously transported by theelectric field towards the anode, thus leaving behind, in theenzyme reaction chamber, the pure optical isomer D-phenylglycine.

MATERIALS AND METHODS

Equipment and Reagents

Penicillin G acylase (PGA) (E.C. 3.5.1.11), fromEsch-erichia coli,22 U/mg, was obtained from Sigma (St. Louis,MO). D,L-phenylglycine, methyl-4-hydroxyphenylacetate,D-lysine and PGA immobilized on Eupergit were fromFluka (Buchs, Switzerland). NaOH, glycerol, borate andortho-phosphoric acid were from Merck (Darmstadt, Ger-many). Acrylamide,N,N8-methylene bisacrylamide (Bis),N,N,N8,N8-tetramethylethylene diamine (TEMED) and per-suphate were from Bio-Rad (Hercules, CA), Immobilinechemicals (acrylamido weak acids and bases) were fromPharmacia Biotech (Uppsala, Sweden). The multicompart-ment electrolyzer was from Talent (Trieste, Italy). The cap-illary electrophoresis apparatus was a Quanta 4000E fromWaters, Bedford (MA). Fused silica capillaries (50mm I.D.,375mm O.D.) were obtained from Polymicro Technologies(Phoenix, AZ). The Microcon 30000 ultrafiltration systemwas from Millipore (Bedford, MA).

Batch Reactions

Batch reactions were performed in 40 mM D-lysine zwit-terionic buffering agent at pH 9.7 with a 50 mM constantconcentration of methyl-4-hydroxyphenylacetate and differ-ent phenylglycine concentrations ranging from 1.5 mM upto 20 mM. An internal standard of acrylamide was added(10 mL/mL of a 10% solution). The system was ther-mostated at 8°C and 1.144 U/mL of PGA was added.Samples of 100mL were withdrawn at different times, thereaction was stopped by centrifugation at 14000g for 6 minwith Microcon 30 microconcentrators and subsequentlyanalyzed by capillary zone electrophoresis (CZE).

Capillary Electrophoresis Analysis

Each sample was injected by hydrostatic pressure for 3 s inan uncoated capillary (50 cm length, 50mm I.D.). Separa-tion conditions: borate-NaOH buffer, pH 10; 15 kV appliedvoltage. Detection was at 214 nm.

Scheme of the Enzyme Reaction

Figure 1 gives the scheme of the reaction adopted in thepresent study. The enzyme, penicillin G acylase (PGA) uti-lizes the ester methyl-4-hydroxyphenylacetate as an accep-tor for the stereo-specific addition of L-phenylglycine froma racemate. The reaction products are: the free acid 4-hydroxyphenylacetyl-(L)phenyl glycine, the optical isomerD-phenylglycine and free methanol. Two noxious side re-actions are immediately apparent: At the high pH of opti-mum of activity, the ester initially present as cosubstrate canbe converted chemically to the free acid. Additionally, thereaction product, if left in the enzyme chamber, can also behydrolyzed, because it contains an amido bond, and thus

BOSSI ET AL.: ISOELECTRICALLY TRAPPED ELECTROENZYMATIC BIOREACTOR 455

gives origin to free L-phenylglycine and to the free 4-hydroxylphenyl acetic acid. This last reaction is also cata-lyzed by the enzyme, because PGA can drive the equilib-rium in both directions. In any case this last event, whenoccurring by either mechanism, would be devastating, be-cause it would maim the ultimate goal of this enzymaticreaction, namely the production of the pure D-phenylglycine isomer, an important precursor in the synthe-sis of a number of drugs (Kim and Kim, 1993).

Membrane Preparation

The isoelectric membranes were made to contain a 8%T,4%C polyacrylamide matrix, admixed with appropriateamounts of charged Immobiline monomers (buffering andcounter-ions) so as to obtain the desired pI values. Fourisoelectric membranes were cast, having pI values of 2.5,4.0, 5.5, and 10.5. Their composition was calculated withthe aid of the computer program of Giaffreda et al. (1993),so as to ensure an average buffering power of 8 milliequivalents L−1 pH−1. The membranes have a diameter of4.7 cm and a thickness of ca. 1 mm. To render them tear-resistant, the polymeric material is cast onto glass fiberfilters, to which polyacrylamide strongly adheres. Afterstandard polymerization conditions (1 h at 50°C), the mem-branes are extensively washed in distilled water, so as toremove all noxious material (catalysts, ungrafted mono-mers), and then assembled into the MIER apparatus.

Multicompartment Immobilized EnzymeReactor (MIER)

Figure 2 shows the experimental set up of the MIER. Theenzyme, which was found to have a pI value of 5.9, istrapped in chamber R (between the pI 5.5 and 10.5 mem-branes) at a concentration of 0.36 U/mL. The pH in thereaction chamber was maintained at the value of 9.7 by

addition of 40 mM D-lysine. Substrate concentration was 50mM methyl-4-hydroxyphenylacetate and 10 mM phenylgly-cine. Chamber 2, filled with 10% glycerol, was kept at avolume of 10 mL, while chamber R and chamber 1 wereboth filled to 30 mL. The anolyte was 50 mL of 10 mMphosphoric acid in 10% glycerol. The catholyte was 50 mLof NaOH 10 mM in 10% glycerol. Anolyte, catholyte, andthe contents of chambers R and 1, (the latter being theproduct collection chamber) were recirculated with a peri-staltic pump with a flow rate of 2 mL/min. The runningconditions were 200 V constant for several days. Sampleswere withdrawn at different times and the reaction wasstopped by centrifugation.

Eupergit-Immobilized Enzyme Reactor

A column containing 1 g of Eupergit-immobilized PGA(126 U) was prepared and equilibrated with 40 mM D-Lysbuffer prior to feeding the substrate. The reaction conditionswere the same as described for the batch reactor. Substratesand cofactors were continuously injected into the columnfor several hours (2–4 h). Different fractions, with a volumeof about 500mL, were collected during the process. Theanalysis of each fraction was performed by CZE as de-scribed above.

RESULTS

Monitoring the Reaction by Capillary ZoneElectrophoresis CZE

To follow the reaction kinetics both in the batch and in theMIER reactors, we had to devise a simple, yet effective

Figure 1. Scheme of the reaction of penicillin G acylase in the presenceof methyl-4-hydroxyphenylacetate and (D,L)-phenylalanine. The optimumof activity is at pH 9.7 and the reaction is carried out at 8°C to minimizehydrolysis of the amido bond in the product [4-hydroxyphenylacetyl-(L)phenyl glycine].

Figure 2. Scheme of the multicompartment immobilized enzyme reactor(MIER). The enzyme is loaded in R (reaction chamber), trapped betweenthe pI 5.5 and 10.5 isoelectric membranes. The product [4-hydroxyphenylacetyl-(L)phenyl glycine] is collected in chamber 1, be-tween the pI 2.5 and 4.0 membranes. Abbreviations: P: peristaltic pump forrecycling; A: anodic reservoir; Pcc: product collection chamber; RC: re-action chamber reservoir; C: cathodic reservoir. The curved arrows indicatestirring in the various vessels.

456 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 4, NOVEMBER 20, 1998

method for monitoring the content of the reaction chamber.Capillary zone electrophoresis was found to be an excellenttool in this respect, because it allowed a fast and reproduc-ible separation of all components of interest, with easyquantitation of each compound by simple integration of thevarious isolated peaks. Figure 3 gives an example of theelectropherograms obtained. In the upper panel, the reactionmixture is monitored just before the onset of enzyme ca-talysis. An internal standard (acrylamide, peak 1) was addedprior to injection, to correct any difference in analyte mi-gration due to unreproducible electroendoosmotic flow andany variation in peak area due to slightly different injection

volumes in sequential runs. Peak no. 2 represents the cosub-strate (methyl-4-hydroxyphenyl acetate); peak no. 3 is theracemate mixture (D,L-phenylglycine) and peak no. 4 rep-resents small amounts of free acid generated by the spon-taneous hydrolysis of the ester cosubstrate. The lower panelis the electropherogram of the reaction mixture after 30 minfrom the onset of enzyme catalysis. One can immediatelyappreciate the marked decrement of the racemate peak (dueto consumption of the L-isomer) and the appearance of anew peak, i.e., the reaction product [4-hydroxyphenylace-tyl-(L)phenyl glycine, a free acid]. A series of CZE runstaken at appropriate times were the base for constructing thereaction kinetics in both batch and MIER reactors.

Batch Reactions

Batch experiments were performed to measure the kineticparameters of the enzyme and compare them to those ob-tained in the MIER experiments. PGA kinetics were studiedby the initial rate method. A broad range of D,L-phenylglycine concentrations was used for initial rate mea-surements: 1.5, 2, 3, 5, 7.4, 10, 14, and 20 mM (whichappears to be the upper solubility level under our experi-mental conditions). Spontaneous hydrolysis was minimizedby working at 8°C and alkaline conditions (pH 9.7) wereadopted for favoring the synthetic reaction vs. the hydroly-sis. Figure 4 shows the reaction kinetics obtained underthese experimental conditions with two substrate levels: 1.5and 14 mM D,L-phenylglycine. The initial, fast decay, cor-responds to the consumption of the L-isomer: A 0.5% levelwould be equivalent to 100% consumption of L-

Figure 3. Capillary electrophoresis separation of the substrate and prod-ucts in the PGA reaction. Upper profile: CZE separation of the reactionmixture at time T4 0. The peaks are (from left to right) 1: internalstandard; 2: methyl-4-hydroxyphenyl acetate; 3: (D,L)-phenylglycine; 4: asmall amount of acid produced by hydrolysis of the methyl ester (peak 2).Lower electropherogram: CZE analysis of the reaction mixture after 30min of enzyme catalysis. The peaks are as above. The extra peak between3 and 4 (labeled as 5) is the synthetic product [4-hydroxyphenylacetyl-(L)phenylglycine].

Figure 4. Reaction kinetics of PGA in a batch reactor. The two curvesrepresent the kinetics in presence of 1.5 (solid circles) and 14 mM (opensquares) substrate, respectively. The reaction is followed by monitoring thedecrement of the (D,L)-phenylglycine peak in CZE. Note that, whereasafter ca. 10 (for the 1.5 mM) or 25 (for 14 mM) min about 45% substrateis consumed, after 120 min this level goes back to the initial value, due toconcomitant hydrolysis of the product [4-hydroxyphenylacetyl-(L)phenylglycine].

BOSSI ET AL.: ISOELECTRICALLY TRAPPED ELECTROENZYMATIC BIOREACTOR 457

phenylglycine and would indicate the yield of the pure D-form in the reaction chamber (at the various substrate con-centrations, the mean-yield of the synthesis was ca. 47%).Unfortunately, as described in the above section ‘‘Schemeof the Enzyme Reaction,’’ upon prolonged incubation in thepresence of the enzyme (possibly also favored by the alka-line pH), the opposite reaction, i.e., breakdown of the prod-uct of synthesis, begins to develop, as indicated by the sec-ond part of the U-shaped kinetic profile. Deacylation is thenfavored by the presence of the excess of ester and, as aconsequence of working in a closed-batch system, the syn-thetic product will be totally cleaved within the time periodof 120 min, fully reverting to the initial racemic mixture ofD,L-phenylglycine. By linearizing the data with theLineawer-Burk equation, PGA was found to have a Vm of1.36 mM/min ? U and a Km of 3.37 mM/U. The syntheticand hydrolytic rate of the enzyme were calculated duringthe early stages of the reaction and the results are summa-rized in Table I.

MIER Reactions

A description of the characteristic behavior of the MIERreactor will be given below. Due to the fact that enzymecatalysis occurs in the presence of an electric field, it ispossible to follow the reaction progress by also monitoringcurrent changes during the reaction time (Fig. 5). At thevery early stages there is a marked current increment whichcorresponds to the production of ionic species in the reac-tion chamber, because the enzyme converts the neutral esterinto an anionic product [4-hydroxyphenylacetyl-(L)phenylglycine]. In the first 40 min, the enzyme works in the syn-thetic direction, then after consumption of the entire amountof L-phenylglycine, it begins the deacylation of the methyl-ester, thus forming an additional free acid (4-hydroxyphenylacetic acid). The sudden current jump at 40 min indicatesthe transport of these additional charged products towardsthe anode (in chamber 2). At 80 min, there is a strongmovement of 4-hydroxyphenylacetic acid from chamber Rand 2 towards chamber 1. These two flexes indicate thepassage of ions from a chamber to the next in the multi-compartment electrolyzer. As the enzyme reaction is com-pleted (due to exhaustion of the L-isomer), the current rap-idly declines to minimal values, indicating the final stages

of the transport process of the two acids towards the anode.The peculiar advantage of the MIER set-up is that the elec-tric driving force subtracts the synthetic product out of thereaction chamber, causing a shift of the kinetic equilibriumthat favors the enzymatic synthesis over the hydrolysis. Thepresence of the electric field is important also because itforces the hydrolytic product (the acid), which inhibits theenzyme, to rapidly leave the reaction chamber, allowinghigher reaction yields.

The overall reaction progress, as monitored by CZE, issummarized in Figure 6. Figure 6A gives the evolution of allspecies present in the enzyme reaction chamber (R): the linewith triangles indicates the consumption of the ester cosub-strate (methyl-4-hydroxyphenylacetate), accompanied bythe concomitant disappearance of L-phenylglycine (solidcircles). The empty circles give the production of the amidoderivative [4-hydroxyphenylacetyl-(L)phenylglycine]: Be-cause this acidic compound is transported out of chamber R,its profiling reaches a maximum and then slowly decays to0. The concomitant hydrolysis of the excess, free ester (in-duced both by the enzyme, after consumption of L-phenylglycine, and by the alkaline milieu) is indicated bythe line with crosses: This production is continuous and, dueto the large amounts of ester initially present, the free acidproduced (4-hydroxyphenyl acetic acid) does not start de-clining (due to electric transport) until after 200 min ofoperation. Figure 6B gives the profiling of the content ofchamber 2, which temporarily hosts the two acids producedin the reaction chamber. Finally, Figure 6C (the receivingchamber for both acids) gives the evolution in time of thecollection of both 4-hydroxyphenylacetyl-(L)phenylglycineand 4-hydroxyphenyl acetic acid. As a result of both en-zyme reaction and simultaneous electrophoretic transport,the pure D-phenylglycine isomer remains as the desiredproduct in the reaction chamber. The rate of synthesis andhydrolysis was calculated within the first stage of the reac-tion and the Vs / Vh was found to be 4.5 (Table II), which

Table I. Kinetic parameters of the PGA reaction in a batch system.a

Phenylglycine(mM)

Vsynthesis

(mM/min?U)Vhydrolysis

(mM/min?U) Vs/Vhb

1.5 280 175 1.63 480 231 2.037.4 580 258 2.31

14 790 559 1.41

aReaction conditions: 50 mM methyl-4-hydroxyphenylacetate pH 9.7,8°C.

bInitial reaction rate.

Figure 5. Monitoring the course of the enzyme reaction by plotting thecurrent profile vs. time in the MIER apparatus.

458 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 4, NOVEMBER 20, 1998

is more than twofold the ratio Vs / Vh obtained in batchreactions (see Table I).

Figure 7 summarizes the yields of the reaction in time. Asthe reaction progresses, it is seen that, due to total consump-tion of the L-phenylglycine isomer, the total concentrationof the D,L-racemate falls to one half. Concomitantly, thelevel of synthetic product [4-hydroxyphenylacetyl-(L)phenylglycine] rises from 0 to equimolar levels with theremaining D-phenylglycine. Because all substrate has been

exhausted, the level of the two species remains constant intime.

Figure 8 shows the other unique advantage of the MIERreactor over the batch procedure. It plots the residual en-zyme activity vs. time of operation for the two systems. It is

Figure 6. Time courses of the various species in the different chambersof the reactor.A: reaction chamber. The triangles indicate consumption ofthe ester; solid circles: consumption of L-phenylglycine; open circles: rateof synthesis of the product of the enzyme reaction [4-hydroxyphenylacetyl-(L)phenylglycine]; crosses: rate of production of 4-hydroxyphenylaceticacid via hydrolysis of the enzyme reaction product.B: same asA, but withmonitoring of the content of chamber 2 (delimited by the pIs 4.0 and 5.5membranes) of the MIER.C: same asA, but with monitoring of thecontent of chamber 1 (delimited by the pIs 2.5 and 4.0 membranes) of theMIER.

Table II. Kinetic parameters of the PGA reaction in MIER.a

Phenylglycine(mM)

Vsynthesis

(mM/min?U)Vhydrolysis

(mM/min?U) Vs/Vhb

10 120 28 4.3

aReaction conditions: 200 V constant, 50 mM methyl-4-hydroxy-phenylacetate pH 9.7, 8°C.

bInitial reaction rate.

Figure 7. Rate of consumption of L-phenylglycine (solid circles) and ofproduction of 4-hydroxyphenylacetyl-(L)phenyl glycine (open circles) inthe MIER apparatus. The reaction yield is 50%, since the enzyme catalysesan enantioselective synthesis. In the reaction chamber R (see Fig. 2) re-mains D-phenylglycine essentially pure. As a comparison, the uppermost(solid diamonds) and lowermost (open diamonds) curves give the corre-sponding profiles of consumption of D,L-phenylalanine and production of4-hydroxyphenylacetyl-(L)-phenylglycine in an Eupergit-immobilizedconventional reactor.

Figure 8. Time courses of inactivation of PGA in the MIER reactor ascompared with a batch system. The enzymatic activity is fully maintainedduring the first 48 h of operation in the MIER (solid line with diamonds),whereas in the batch system > 80% of enzyme activity is lost (broken linewith open circles). After 4 d, there is still 60% of initial catalytic activityin the MIER, vs. complete loss of activity in batch operations.

BOSSI ET AL.: ISOELECTRICALLY TRAPPED ELECTROENZYMATIC BIOREACTOR 459

seen that, whereas after only one day of operation, the batchreactor experiences a 75% loss of enzymatic activity; 100%of enzyme integrity is maintained in the MIER approach.After four days of operation, the enzyme activity falls to 0in a batch reactor, whereas 60% of enzyme activity stillremains in the MIER set-up.

Eupergit-PGA Reactor

This conventional immobilized reactor exhibits some pecu-liar behavior. In terms of enzyme activity, it is maintainedfor long periods of time (e.g., after 12 days of operation, theenzyme still retains 85% of the original activity). Thus thestability of the covalently bound enzyme seems to be betterthan that of the free form. However, the behavior in terms ofconsumption of the L-phenylglycine form and production of4-hydroxyphenylacetyl-(L)phenylglycine is less than satis-factory. As shown in Figure 7, there appears to be no morethan 15–20% consumption of the L-form (uppermostcurve), with a concomitant yield of 15–20% of product(lowermost tracing) (cf. the near 100% consumption in thecase of the MIER, two central curves). Why should it be sois not quite clear; we hypothesize that the amido bond of thereaction product generated in the upper layers of the columnmight be hydrolyzed as the product migrates down the resin.Thus, this type of reactor seems not to be efficient for thepurpose of properly resolving the racemate mixture.

DISCUSSION

Two main advantages of the MIER reactor, over conven-tional batch operations, are immediately apparent: the veryhigh conversion rates (close to 100% consumption of theL-isomer in the reaction chamber, see Fig. 7) and the con-comitant high stability of the enzyme in the reactor (see Fig.8). This speaks in favor of such a reactor in industrial ap-plications, for large-scale production of important reactionintermediates for pharmaceuticals and other valuable indus-trial products. The small size of our present MIER apparatusshould not be a deterrent. At present, we only operate withrelatively small-size membranes (4.7 cm diameter), butthere is no upper limit for a scaled-up version. The majorlimitation, in fact, would be the poor resistance of large-surface membranes to tear or to pulsing originating from thepumping devices. Nevertheless, large-surface membranescan be built in a very simple and effective way by exploitingthe technology available today for large optical telescopesin astronomy. Due to difficulties in cooling down the meltedsilica mirrors, and to the high risk of refractive index gra-dients produced by differential cooling of such a large mass(which often require years of cooling at a very slow rate),such lenses are produced as relatively small diameter opticalobjects and then assembled into a single, large optical de-vice for collecting faint star light. Much in the same fashion,we could easily build a MIER instrument with large mem-brane surface areas (even one square meter) by simply as-sembling smaller membranes into a single array. The pro-

cessing rate will then be proportional to the available mem-brane area; all other experimental conditions being keptconstant (voltage gradient, substrate concentration in thefeed, etc.). The cost of the membranes would not be pro-hibitive too, because we have described novel membranes,made of extremely resistant and highly hydrophilicN-substituted acrylamide monomers, capable of operating inan electric field for long periods of time (Chiari et al., 1994),with minimal variations of the Immobiline pK values (Bossiet al., 1994). Additionally, our isoelectric membranes areknown not to be fouled by adsorption of proteins and en-zymes, thus they can be reused for several MIER operations(Righetti et al., 1992). The Joule heat production in suchlarge instruments should also not be a problem. As in thepresent, laboratory scale version, the distance between ad-jacent membranes in the MIER can be kept to a minimum,thus greatly reducing the amount of liquid present at anytime in the electric field. The largest proportion of feed willthen be found in the reservoirs, which can be adequatelycooled (or heated) by standard thermostats.

One major limitation could be the availability of suitableisoelectric buffers to be trapped in the enzyme chamber, forcontrolling the pH during operation. At present, we are lim-ited by a list reported some time ago for conventional iso-electric focusing in soluble, amphoteric buffers (Righetti,1983). However, with the advent of immobilized pH gradi-ents (Righetti, 1990), an almost endless number of ampho-teric buffers could be made, covering effectively the pHscale even at minute pI increments. Such soluble Immobi-line buffers can only be made polymeric, because in theprocess of polymerizing the weak acids and bases, admixedwith the neutral monomer (acrylamide, but in the absence ofcross-linker) an amphoteric, isoelectric polymer will begenerated. Such a polymer might have substantial viscosity,however, we have described methods, based on chain-transfer agents coupled to high temperatures, for producingshort-chain, low-viscosity acrylamide polymers (Gelfi et al.,1995). Experiments are now in progress to verify this pro-cedure, which would render the MIER technology easilyapplicable to any enzyme reaction at any desired pH value.

An additional drawback of the present method is the factthat, although the desired product (D-phenylglycine) can berecovered essentially pure, it still remains in the reactionchamber, together with the enzyme. Thus, an additionalprocessing step will be required for separating the enzymefrom the product.

Supported by grants from Agenzia Spaziale Italiana (ASI, Roma)and by Progetto Finalizzato Biotecnologie e Biostrumentazione(CNR, Roma, grant No. 97.01199.PF9).

References

Alvaro, G., Fernandez-Lafuente, R., Blanco, R. M., Guisan, J. M. 1990.Immobilization-stabilization of penicillin G acylase fromE. coli. Ap-plied Biochem. Biotechnol.26: 181.

Bossi, A., Righetti, P. G., Chiari, M. 1994. Immobilized pH gradients: New

460 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 4, NOVEMBER 20, 1998

pK values of acrylamido buffers in poly(N-acryloylamino ethoxy etha-nol) matrices. Electrophoresis15: 1112.

Chiari, M., Dell’Orto, N., Mendozza, M., Carrea, G., Righetti, P. G. 1996.Enzyme reactions in a multicompartment electrolyzer with isoelectri-cally trapped enzymes. J. Biochem. Biophys. Methods.31: 93.

Chiari, M., Nesi, M., Roncada, P., Righetti, P. G. 1994. Preparative iso-electric focusing in multicompartment electrolysers: Novel, hydrolyti-cally stable and hydrophilic isoelectric membranes. Electrophoresis15: 953.

Fuganti, C., Grasselli, P., Seneci, P. F., Servi, S. 1986. Immobilized ben-zylpenicillin acylase: Application to the synthesis of optically activeforms of carnitine and propranalol. Tetrahedron. Lett.27: 2061.

Furusaki, S., Nozawa, T., Nomura, S. 1990. Membrane enzyme reactorwith simultaneous separation using electrophoresis. Bioproc. Eng.5:73.

Gekas, V. 1986. Artificial membranes as carriers for the immobilization ofbiocatalysts. Enzyme Microb. Technol.8: 450.

Gelfi, C., Orsi, A., Leoncini, F., Righetti, P. G. 1995. Fluidified polyacryl-amides as molecular sieves in capillary zone electrophoresis of DNAfragments. J. Chromatogr. A689: 97.

Gerhartz, W. 1990. General production methods. In: W. Gerhartz (ed.)Enzymes in Industry: Production and Applications. VCH, Weinheim,pp. 33–74.

Giaffreda, E., Tonani, C.,Righetti, P. G. 1993. A pH gradient simulator forelectrophoretic techniques in a windows environment.J. Chromatogr.630: 313.

Ishimura, F., Suga, K. I. 1992. Hydrolysis of penicillin G by combinationof immobilized penicillin G acylase and electrodialysis. Biotechnol.Bioeng.39: 171.

Kim, D. M., Kim, H. S. 1993. Enzymatic synthesis of D-p-hydroxyphenylglycine from DL-p-hydroxyphenylhydantoin in thepresence of organic solvents. Enzyme Microb. Technol.15: 530.

Lee, C. K., Hong, J. 1987. Enzyme reaction in a membrane cell coupledwith electrophoresis. Ann. N. Y. Acad. Sci.506: 499.

Nembri, F., Bossi, A., Ermakov, S., Righetti, P. G. 1997. Isoelectricallytrapped enzymatic bioreactors in a multimembrane cell coupled to an

electric field: Theoretical modeling and experimental validation withurease. Biotechnol. Bioeng.53: 110.

Prazeres, D. M. F., Cabral, J. M. S. 1994. Enzymatic membrane bioreac-tors and their appplications. Enzyme Microb. Technol.16: 738.

Righetti, P. G. 1983. Isoelectric focusing: Theory, methodology and appli-cations. Elsevier, Amsterdam.

Righetti, P. G., Wenisch, E., Faupel, M.. 1989. Preparative protein purifi-cation in a multi-compartment electrolyzer with Immobiline mem-branes. J. Chromatogr.475: 293.

Righetti, P. G. 1990. Immobilized pH gradients: Theory and methodology.Elsevier, Amsterdam.

Righetti, P. G., Wenisch, E., Jungbauer, A., Katinger, H., Faupel. M. 1990.Preparative purification of human monoclonal antibody isoforms in amulticompartment electrolyzer with Immobiline membranes. J. Chro-matogr.500: 681.

Righetti, P. G., Faupel, M., Wenisch, E. 1992. Preparative electrophoresiswith and without immobilized pH gradients, pp. 159–200. In: A.Chrambach, M. J. Dunn, B. J. Radola (eds.), Advances in Electropho-resis, Vol. 5. VCH, Weinheim, Germany.

Righetti, P. G., Nembri, F., Bossi, A., Mortarino, M. 1997. Continuousenzymatic hydrolysis ofb-casein and isoelectric collection of some ofthe biologically active peptides in an electric field. Biotechnol. Prog.13: 258.

Rosell, C. M., Fernandez-Lafuente, R., Guisan, J. M. 1993. Resolution ofracemic mixtures by synthesis reaction catalysed by immobilized de-rivatives of the enzyme penicillin G acylase. J. Mol. Catal.84: 365.

Rosell, C. M., Fernandez-Lafuente, R., Guisan, J. M. 1995. Resolution ofracemic mixtures through stereospecific kinetically controlled synthe-sis catalyzed by Penicillin G acylase derivatives. Ann. N.Y. Acad. Sci.750: 425.

Sheldon, R. A., Porskamp, P. A., Ten Hoeve, W. 1985. Immobilized en-zymes,pp.59–0. In: J. Tramper, H. C. van der Plas, P. Linko (eds.),Biocatalysis in organic synthesis. Elsevier, Amsterdam.

Zmijewski Jr., M. J., Briggs, B. S., Thompson, A. R., Wright, I. G. 1991.Enantioselective acylation of ab-lactam intermediate in the synthesisof Loracarbef using penicillin G amidase. Tetrahedron Lett.32: 1622.

BOSSI ET AL.: ISOELECTRICALLY TRAPPED ELECTROENZYMATIC BIOREACTOR 461