Biotechnology Advances 24 (2006) 27 41 www.elsevier.

com/locate/biotechadv

Research review paper

Kinetics of h-lactam antibiotics synthesis by penicillin G acylase (PGA) from the viewpoint of the industrial enzymatic reactor optimization
Roberto C. Giordano *, Marcelo P.A. Ribeiro, Raquel L.C. Giordano
DEQ/UFSCar, Via Washington Luiz, km 235, C.P. 676, 13564-210, Sao Carlos, SP, Brazil Received 11 March 2005; accepted 15 May 2005 Available online 28 June 2005

Abstract Competition with well-established, fine-tuned chemical processes is a major challenge for the industrial implementation of the enzymatic synthesis of h-lactam antibiotics. Enzyme-based routes are acknowledged as an environmental-friendly approach, avoiding organochloride solvents and working at room temperatures. Among different alternatives, the kinetically controlled synthesis, using immobilized penicillin G acylase (PGA) in aqueous environment, with the simultaneous crystallization of the product, is the most promising one. However, PGA may act either as a transferase or as a hydrolase, catalyzing two undesired side reactions: the hydrolysis of the acyl side-chain precursor (an ester or amide, a parallel reaction) and the hydrolysis of the antibiotic itself (a consecutive reaction). This review focuses specially on aspects of the reactions kinetics that may affect the performance of the enzymatic reactor. D 2005 Elsevier Inc. All rights reserved.
Keywords: Industrial synthesis of h-lactams; Penicillin G acylase; Enzyme kinetics; Enzymatic reactors

Contents 1. 2. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The process: thermodynamically and kinetically controlled approaches . . . . . . . . . . . . . . . . . 2.1. Thermodynamically controlled synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Kinetically controlled synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The biocatalyst: enzyme immobilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The enzyme: penicillin acylase (hydrolase, amidase). . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. h-Lactam acylases/synthases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. E. coli PGA characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalytic mechanism of E. coli PGA and the effects of process variables (pH, temperature, substrate concentrations) on the reaction rates: key points for the integrated reactor optimization. . . . . . . . . 5.1. Literature consensuses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 28 28 29 30 30 30 32 32 33

3. 4.

5.

. . . . . . . . . . . . . .

* Corresponding author. Tel.: +55 16 3351 8708; fax: +55 16 3351 8266. E-mail address: roberto@power.ufscar.br (R.C. Giordano). 0734-9750/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.biotechadv.2005.05.003

28

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

Acylation steps and the binding site of the h-lactam nucleus to E. coli PGA: X-ray crystallography insights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2. pH and temperature effects on the synthesis . . . . . . . . . . . . . . . . . . . . . . 5.1.3. Inhibitory and activation effects of reactants and products on the reaction .rates . . . 5.2. Open questions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1. Understanding the deacylation step during synthesis: a decisive point to define optimal concentrations and molar ratio of substrates . . . . . . . . . . . . . . . . . . . . . . 5.2.2. Inhibitory effect of the acyl donor derivative . . . . . . . . . . . . . . . . . . . . . . 5.2.3. Do classical mechanistic models represent a convenient approach for this system? . . 6. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1.1.

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

33 33 34 34 34 37 37 38 38

1. Introduction h-Lactam antibiotics are among the most used pharmaceuticals. Semi-synthetic penicillins and cephalosporins (amoxicillin, ampicillin, cephalexin, cefadroxil, cefazolin, and several others) correspond to 65% of the ever rising worldwide production of antibiotics (Elander, 2003), exceeding 45 000 tons in 2000 (Bruggink and Roy, 2001). Industrial manufacturing of these drugs began during the 1960s and 1970s (Beecham patented ampicillin in 1961 and amoxicillin in 1972; Lilly, cephalexin in 1970; BMS, cefadroxil in 1977). These conventional processes demand low temperatures (less than 30 8C), organochloride solvents and protection/deprotection of side groups, and generate great amounts of non-recyclable waste (Ospina et al., 1996; Youshko et al., 2002). Cole (1969) envisioned the possibility of replacing the chemical by the enzymatic synthesis, preferably working in aqueous media and at physiological conditions. Ever since, interest on bcleanerQ, enzymatic syntheses that complies with growing environmental regulations has increased steadily. This benvironmental-friendlyQ approach can be included in the bgreen chemistryQ field. Nevertheless the enzymatic synthesis still has a major drawback: competing with payback times of finetuned conventional processes. In fact, to the best of our knowledge, there is only one plant making use of the enzymatic route nowadays (Wegman et al., 2001). The success in establishing the enzymatic-based production of these antibiotics in new plants will rely on the process optimizationand especially of the bioreactor operation. Industrial semi-synthetic antibiotics result from the condensation of a modified side-chain (the acyl donor) to a h-lactam nucleus. To produce these nuclei in industry, enzymatic hydrolyses of microbial penicillins or cephalosporins are employed: penicillin (usually, penicillin G), providing 6-aminopenicillanic acid (6APA) or cephalosporin C (giving 7-aminocephalos-

poranic acid, 7-ACA). 7-Aminodeacetoxycephalosporanic acid (7-ADCA) is obtained after the chemical ring expansion of 6-APA (Bruggink and Roy, 2001); alternatively, the hydrolysis of cephalosporin G (Shewale et al., 1990) or of microbial adipyl-7-ADCA (Schroen et al., 2002a) may be used. Penicillin G acylase (PGA) (EC 3.5.1.1.11), immobilized on insoluble supports, catalyzes these processes in industry. Most of the time, the strain ATCC 11105 of Escherichia coli (and its mutants) is used. Bacillus megaterium (ATCC 14945) is also employed (Rajendhran and Gunasekaran, 2004). The enzymatic production of semi-synthetic h-lactam antibiotics, on its turn, has been extensively studied both in academia and industry (Bruggink, 2001) but there are still some key controversial points in the literature with respect to the reaction mechanism, which carry consequences for the economical feasibility of the process. In this review, we focus for the most part on the bmainstreamQ configuration for the enzymatic synthesis reactor, i.e. the kinetically controlled route using integrated fed-batch reactorsand especially on aspects of the reactions kinetics that may be important to optimize the reactor. 2. The process: thermodynamically and kinetically controlled approaches 2.1. Thermodynamically controlled synthesis The enzymatic synthesis of h-lactam antibiotics may depart directly from the acyl donor, or from an ester or amide derivative (Kasche et al., 1987). In the first case, one faces a classical thermodynamically controlled problem, having to shift the reaction equilibrium towards the products (see Fig. 1A). This thermodynamically controlled synthesis would certainly be an elegant solution. Still, attempts in this direction, using PGA from different microorganisms, in the presence of dif-

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

29

A
O R1 OH H2N S N O

Synthesis
O CH3 CH3 COOH NH N O

R1

PGA antibiotic Hydrolysis

O R1 NH

CH3 CH3 COOH

H OH

water

side chain

6-APA

O R1 R2

H2N

Synthesis
S N CH3 CH3 COOH

+
O

S N

CH3 CH3 COOH

R2 H

Acyl-derivative
H2O

6-APA PGA

antibiotic
H2O

leaving group

Hydrolysis 1
O R1 OH

Hydrolysis 2
O H2N S N O

R2 H

R1

OH

CH3 CH3 COOH

side chain

leaving group

side chain

6-APA

Fig. 1. (A) Thermodynamically controlled synthesis of semi-synthetic penicillins, catalyzed by penicillin G acylase (PGA). (B) Kinetically controlled synthesis. PGA acts as transferase (for synthesis) and as hydrolase, promoting two undesired side reactions (hydrolysis 1, of the sidechain acyl donor derivative, and hydrolysis 2, of the antibiotic). For instance, in the synthesis of ampicillin R1-COOH = d-()-phenylglycine (PG); for amoxicillin, R1-COOH = p-OH-d-()-PG, etc. The acyl donor derivative may be an ester (e.g., , R2 = OCH3) or amide (R2 = NH2). Semisynthetic cephalosporins will have cepham nuclei, 7-ADCA or 7-ACA, in place of the penam nucleus, 6-APA.

ferent co-solvents, have only been successful for a few h-lactams. When using E. coli PGA for the thermodynamically controlled synthesis, the carboxyl group of the sidechain must be neutral, ready for the nucleophilic attack of the enzyme, while, at the same time, the amino group of the h-lactam nucleus is also neutral, available for nucleophilic interactions (see the discussion of the reaction mechanism in the next sections). However, in the enzyme active range (pH 68), the number of substrate molecules with the proper charge is negligible. A different enzyme could overcome this kinetic hindrance (for example, Xanthomonas citri PGA, Svedas et al., 1980; Blinkovsky and Markaryan, 1993). All the same, thermodynamics disfavors the formation of the amide bond in aqueous medium (as Blinkovsky and Markaryan, 1993 showed for ampicillin and cephalexin). Schroen et al. (1999) tried the thermodynamic-con trolled synthesis of cephalexin using PGA from E. coli and X. citri, but only very low concentrations of antibiotic were produced (less than 0.1 mM at pH 5.5, 530 8C). On the other hand, the low solubility of phenylglycine in water impairs displacement of the equilibrium towards the synthesis (Ospina et al., 1996). Of course, different co-solvents might be used to shift the equilibrium (Fernandez-Lafuente et al., 1991, 1995, 1996a,b; Diender et al., 1998). In particular,

Fernandez-Lafuente et al., 1996a, used PGA from Kluyvera citrophila and established profiles of pH, temperature and co-solvent (diethylformamide) during batch syntheses, thus reaching 20 g/L of cefalotine. The loss of enzyme activity in the presence of organic co-solvents is another drawback of the thermodynamically controlled process. Abian et al. (2001, 2002) described how a multipoint covalent attachment of the enzyme to the immobilization matrix, and the existence of a hydrophilic micro-environment surrounding the molecule may stabilize the enzyme. The same group tested this concept on different condensation reactions (Abian et al., 2004a). Even so, in spite of the improvements in the design of thermodynamically controlled synthesis, the mainstream alternative is still the kinetic-controlled route. 2.2. Kinetically controlled synthesis In the bkinetically controlledQ process, an activated precursor of the acyl moiety (e.g., an ester or amide of the d-acyl side-chain of the antibiotic) reacts with the h-lactam nucleus, releasing the antibiotic in addition to alcohol or ammonium. Fig. 1B illustrates this process. Changing the acyl donor, one obtains different semisynthetic penicillins. Cephalosporins result when 6APA is replaced by 7-ACA or 7-ADCA.

30

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

The kinetically controlled process falls into a classic optimization problem, when seeking the intermediate product in a set of series-parallel reactions. If the process is interrupted at the point of maximum conversion towards the desired product, yields well above the thermodynamic equilibrium can be achieved. The nucleophilic attack to the ester/amide bond is direct, since the problem of the charged carboxyl has been overcome. But there is a drawback in this approach: PGA acts either as a transferase or as a hydrolase, and water molecules compete with h-lactam nuclei for the nucleophilic attack to the acyl-enzyme intermediate. Therefore, the economical feasibility of this route, to a large degree, will be determined by the optimization of the reactors operational conditions. The stoichiometry of the reactions shown in Fig. 1B provides two degrees of freedom for the system, so the yield (with respect to the acyl derivative or to the nucleus; one of them must be chosen as the key component) and the selectivity (S/H = synthesis/hydrolysis ratio) are independent. Productivity, on the other side, is related to the reactions kinetics. Hence, these three variables should be considered when defining an objective function for the enzymatic synthesis reactor (Ferreira et al., 2004). This is even more important because opposite kinetic effects are present: for example, high concentrations of 6-APA may enhance the selectivity but at the expense of productivity. Several options have been put forth for the kinetically controlled process: product crystallization in integrated reactors (Schroen et al., 2002b; Youshko et al., 2000, 2001), pH gradients during batch or fed-batch operation (Youshko et al., 2002), modulation of the enzyme performance after covalent multipoint immobilization (Guisan, 1988), increasing the enzyme affinity with respect to acyl donors with Ca-substitutes via site-directed mutagenesis (Alkema et al., 2004), enhancement of the enantioselectivity of PGAs from different microorganisms via immobilization (Rocchietti et al., 2002), membrane non-isothermal reactors (Schroen et al., 2001a, Travas cio et al., 2002a,b), use of two-phase systems (Hernandez-Justiz et al., 1998; Wei et al., 2002), co-solvents (Kim and Lee, 1996), co-solvents and low temperatures (Illanes et al., 2003, 2004) or even frozen media (van Langen et al., 1999) are some of the alternatives. Whatever strategy is chosen, a rational approach to the reactor optimization problem demands a reliable kinetic model. The mechanism of the three reactions (synthesis of the antibiotic, hydrolysis of the side-chain precursor and hydrolysis of the antibiotic) has been thoroughly studied and there are several consensual aspects in the literature, as it will be shortly discussed

in the next sections. Yet, some important questions relating to the kinetic mechanism of E. coli PGA remain disputable, as will be shown. 3. The biocatalyst: enzyme immobilization The cost of the enzyme poses the need of immobilization for industrial use. A number of commercial catalysts in the form of insoluble particles are available. Guisan (1988), in a seminal paper, described the multipoint covalent immobilization and stabilization of PGA on glyoxil-agarose. Different non-conventional strategies are reported in the literature. Phadtare et al. (2002) proposed a PGAfatty lipid biocomposite film. Ozturk et al., 2002, linking PGA to carboxymethylcellulose. Kazan et al. (1996) and Cao et al. (2001) used enzyme crosslinking. Wilson et al. (2004a,b), co-aggregated PGA and polyionic polymers, aiming to stabilize the catalyst in organic media. With the same purpose, Wilson et al. (2004a,b) entrapped crosslinked PGA in polyvinil alcohol polymeric matrix. An interesting approach, coupling enzyme and genetic engineering, is reported by Abian et al. (2004b). Site-directed mutations of PGA are promoted to enhance the number of lysine residues, capable of being covalently linked to glyoxyl-agarose, thus increasing the catalyst stability. Immobilization on insoluble matrix beads or particles is still the conventional solution. With this conception, intra-pore diffusion resistances might become important, and intra-particle gradients of concentration and pH may affect the performance of the kinetically controlled process. Moreover, the simultaneous crystallization of product inside the pores has to be considered. Kasche and Galunsky (1994) presented a correlation to predict the appearance of crystals in the catalyst pores, using different reactions (hyrolyses and syntheses of peptides) as case studies. Different conceptions of the enzymatic reactor, facilitating the separation of the biocatalyst from the products, have been patented (e.g., Kaasgaard and Karlsen, 1992; Clausen and Dekkers, 2000; Giordano et al., 2004). 4. The enzyme: penicillin acylase (hydrolase, amidase) 4.1. b-Lactam acylases/synthases Penicillin acylases are important enzymes in the pharmaceutical industry. Although their major industrial application is the production of 6-amino penicil-

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

31

lanic acid (6-APA), catalyzing the hydrolysis of penicillins (Parmar et al., 2000), they can be used in several other biotechnological applications, such as peptide synthesis and racemic resolution (Arroyo et al., 2003; Schoemaker et al., 1997; Shewale et al., 1990). They are classified in three sub-groups,

according to their preferential substrate: penicillin V acylase, penicillin G acylase (PGA) and ampicillin amidase (Valle et al., 1991). Following the classification of the SCOP data base, Murzin et al., 1995, penicillin acylases belong to the N-terminal hydrolase superfamily.

Asn 241

NH

O H H3C

O N

A
O O O O N C NH S O H O O H N H H H CH3 CH3

Ala 69
H N

O O N O H2N CH3 CH3 O O O H N H H

O O N

Pen G

O NH O H H + N H O H H S

6-APA
S CH3 CH3 H

Ser 1

Ser 1

Ser 1

Asn 241

NH

O H N H3C

Ala 69
N

O O OH O H O O H + N H H H O H O HH N H H O O H O H OH

PAA

H N H H

Ser 1

Ser 1

Ser 1

NH2 O

Gln 23

B
O O H H N O H H HO O O N NH O H H + N H H NH O O H S CH3 CH3 H H O H O H N H NH +H H O H N H H3C

Asn 241

NH

Ala 69

Ser 1

Phe 146
O

NH NH

Arg 145

Fig. 2. (A) Hydrolysis of penicillin G (Pen G, or benzylpenicillin) yielding 6-APA and phenyl acetic acid (PAA). Mechanism suggested by Duggleby et al. (1995). The role of the water molecule acting as a virtual base was questioned by McVey et al. (2001). (B) Scheme of intermediary step of the hydrolysis of Pen G. The figure depicts the interaction of Arg a145 with the carboxyl group of the h-lactam moiety as reported by Alkema et al. (2000). Bold style bounds indicate the enzyme backbone.

32

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

However, penicillin acylases are not the only enzymes that can be used for the synthesis of hlactam antibiotics. The structure and activity of cephalosporin acylases have been recently reported (for example, Kim et al., 2002; Yoon et al., 2004); aamino acid ester hydrolases of X. citri (Barends et al., 2003) and of Acetobacter turbidans (PoldermanTijmes et al., 2002a,b) are h-lactam synthases, as well. Fernandez-Lafuente et al., 2001a,b) reported promising results for the synthesis of ampicillin using a-amino acid ester hydrolase from A. turbidans, which is also enantioselective with respect to the d-acyl donor. Still, the enzyme mostly used in industrial processes is penicillin G acylase (PGA) from E. coli, EC 3.5.1.11. In vivo, it is involved in the catabolism of phenyl-acetyl compounds (Galan et al., 2004). Several other bacteria, yeasts and fungi express penicillins acylases (Shewale et al., 1990; Rajendhran and Gunasekaran, 2004; Sio and Quax, 2004). Gram-negative bacteria such as E. coli, K. citrophila, Providencia rettgeri, Alcaligenes faecalis, accumulate PGA in the peripalsmatic space, while Gram-positive bacteria (B. megaterium, Arthrobacter viscosus) secrete it (Rajendhran and Gunasekaran, 2004). Other h-lactam acylases have been described recently, with different substrate specificities (Sio and Quax, 2004). 4.2. E. coli PGA characterization PGA from E. coli is a heterodimer (Bock et al., 1983). Sub-unit a has 209 amino acids and sub-unit h, 557 (Bruns et al., 1985). The active site is at the bottom of a conic depression formed by residues of the two sub-units, which are tightly intertwined. E. coli has the mature PGA (86 kDa) located into the perisplasmatic space, and processed from a single precursor, which has a 26-amino acids signal transport sequence and a 54-amino acids spacer (Duggleby et al., 1995; Brannigan et al., 2000; McVey et al., 2001). This cleavage is autocatalytic, exposing the active site of the mature enzyme (Kasche et al., 1999). This enzyme promotes a nucleophilic attack to the carbonyl carbon of amide or ester bonds. In this aspect, it is similar to serine proteases, but instead of the catalytic triad, characteristic of these proteases, PGA has a single amino acid as a catalytic center: Ser h1, with the nucleophilicity of the Og enhanced by its aamino group (Duggleby et al., 1995). Therefore, this enzyme can be classified as an N-terminal nucleophile hydrolase (Brannigan et al., 1995).

The industrial importance of PGA has stimulated research concerning its structure and catalytic action. Two main approaches are employed: X-ray crystallographic structural analyses (usually in the presence of ligands) and kinetic studies. In either case, mutants and the wild enzyme are studied. There are many consensual aspects in the literature regarding the catalytic mechanism of PGA, mostly concerning the initial elementary steps, i.e. acylation of the Ser h1. Still, deacylation in the presence of the h-lactam nucleus remains elusive. During hydrolysis of penicillins, the deacylation agent is water and its high concentrations permit the lumping of this variable into the kinetic constants of the rate equations. Yet, in order to optimize the synthesis of antibiotics with competing nucleophiles, elucidating the second part of the catalytic mechanism (i.e., deacylation) remains an essential task. The following subsections briefly describe the state-of-the-art reaction mechanism of PGA. Fig. 2 illustrates the steps of the hydrolysis of penicillin G by PGA, as reported by different authors. A number of papers show the crystal structure of the enzyme complexed with different side-chains ligands (Duggleby et al., 1995; Done et al., 1998; Alkema et al., 2000, 2002, 2004; McVey et al., 2001). However, there are no data available for enzyme-(h-lactam) complexes, because the solubility of h-lactam nuclei and of PGA is similar, making their co-crystallization unfeasible, as well as the soaking of protein crystals with these nuclei (Alkema et al., 2000). To circumvent this problem, it is possible to soak crystals of an inactive mutant (obtained through site-directed mutagenesis) with penicillin (for instance), or of the wild enzyme (or some mutant) with a modified, inert substrate. This approach may provide some insights concerning the reaction mechanism, but these are always indirect clues. They must be complemented with kinetic assays, which will add essential information to the analysis. 5. Catalytic mechanism of E. coli PGA and the effects of process variables (pH, temperature, substrate concentrations) on the reaction rates: key points for the integrated reactor optimization The following discussion on the catalytic mechanism of PGA is far from exhaustive. Instead, it focuses on the most important aspects for the optimization of a discontinuous, heterogeneous enzymatic integrated reactor, where particles with immobilized enzyme catalyze the synthesis/hydrolyses reactions (see Fig. 1B), while the product crystallizes.

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

33

5.1. Literature consensuses 5.1.1. Acylation steps and the binding site of the blactam nucleus to E. coli PGA: X-ray crystallography insights Duggleby et al. (1995) presented the first crystal structure of E. coli PGA (at 1.9 A resolution). They also reported complexes (2.5 A) with phenyl acetic acid (PAA), and phenylmethylsulphonyl fluoride (PMSF), which links covalently to Ser h1, emulating the tetrahedral hemiacetal intermediate. The authors proposed that during the hydrolysis of penicillin an acyl-enzyme intermediate is formed, after the nucleophilic attack of Og(Fig. 2A). The resulting oxianion can be stabilized by Ny of Asn h241, by the backbone NH of Ala h69 and of Gln h23 (Fig. 2B). Alkema et al. (2000) proposed that the backbone oxygen of Gln h23 would hydrogen-bond, via a water molecule, with the carbonyl oxygen of PAA. Done et al. (1998) reported conformational changes after the binding of phenylacetamide derivatives to PGA from E. coli. PAA analogues were lumped into two different subsets: for subset 1, the conformation of the native protein structure was preserved (this subset includes PAA). Subset 2 ligands induced a conformational change: Arg a145 and Phe a146, which have an important role in the interaction with the h-lactam nucleus, shift towards the solvent, in an open configuration. This two residues lie at the end of a 16-amino acids a-helix (a131a146), which is practically not affected by their large movement. Alkema et al. (2000) promoted site-directed mutageneses of E. coli PGA. Crystals soaked with penicillin G (Pen G) had their structure analyzed. Pen G induced Arg a145 and Phe a146 to assume an open configuration, similar to the one observed by Done et al. (1998), for subset 2 ligands. Arg a145 protrudes 8 A towards the solvent, but still interacts with the carboxyl group of the nucleus (see Fig. 2B) via two bridging water molecules, confirming the observations of Kasche et al. (1984) and Svedas et al. (1996), that a positively charged residue might exist in the nucleus binding site. Alkema et al. (2002) mutated Phe a146, Phe h24 and Phe h57. The substitutes were Tyr, Trp, Ala and Leu (all hydrophobic). The authors observed that Phe h57, located at the bottom of the hydrophobic cleft, was important to preserve the structure of the site. A series of kinetic assays was run, at 30 8C and pH 7.0. Hydrolysis of NIPAB and synthesis of ampicillin (using phenylglycine amide or methyl ester) were carried out. The effect of the mutations on the synthesis selectivity, S/H

(ratio between the velocities of synthesis and hydrolysis) was smaller than on the hydrolysis of NIPAB. It should be noticed, however, that at pH 7.0 effects on S/H are damped. A three-fold increase of S/H was obtained for the mutant hF24A, using the ester as side-chain precursor, without loss of activity (acylation of hF24A was slower for the amide substrate, but not for the ester). McVey et al. (2001) presented crystallographic data (1.3 A), using two approaches: the wild enzyme was soaked with penicillin G sulfoxide (an almost not hydrolysable substrate) and the inactive mutant hN241A was soaked with penicillin G. The flexibility of Phe a146, Arg a145 and Phe h71, interacting with the thyazolidine ring, was pointed out as essential for the binding of the nucleus. Michaelis affinity constants (K M) of wild PGA for PAA are 102103 lower than for Ca-derivatives of PAA such as phenylglycine (PG; Margolin et al., 1980; Svedas et al., 1996). Alkema et al. (2004) observed that the mutants aF146Y, hF24A and aF146Y/hF24A all display greater affinity for Ca-modified acyl donors (for instance, PG) than the wild enzyme. At the same time, these mutants have lower affinity for phenyl acetic acid (PAA). The rates of reaction with the mutants increasedalthough the consequences for selectivity in the course of an industrial process should still be confirmed. Nonetheless, reducing the strong inhibition of the enzyme by PAA is a favorable aspect, since traces of this compound will probably come along with 6-APA from the upstream hydrolysis of penicillin G. In short, the binding mechanism of the h-lactam nucleus has important consequences for the reactor operation, as it will be stressed in the next sessions. Arg a145, Phe a146 and Phe h71 form this binding site, which, in the absence of substrate, is not in the correct spatial conformation. 5.1.2. pH and temperature effects on the synthesis The formation of the acyl-enzyme complex demands that the terminal amine of Ser h1 be deprotonated (see Fig. 2A). The pK of free a-amine group ranges between 6.8 and 7.9: this is the reason for using pH above 8 to hydrolyze penicillin G. In spite of this, the optimum pH for the synthesis of antibiotics departing from ester acyl derivatives is between 6.0 and 6.5. That is because selectivity (S/H) rises at lower pHs, thus compensating the decrease in productivity (due to the lower activity of the enzyme). On the other hand, Schroen et al. (2001a,b) recommended pH 8.0 when phenylglycine amide was used in the synthesis of cephalexin with a commercial catalyst (AssemblaseR, DSM). However,

34

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

the solubility of h-lactams increases with pH, making the crystallization of the product more difficult. Ospina et al. (1996) observed that the selectivity for ampicillin, at 25 8C, is maximum at pH 6.0. Goncalves et al. (2000), studying the synthesis of amoxicillin from d-p-hydroxyphenylglycine methyl ester (PHPGME) and 6-APA, with PGA immobilized on agarose, confirmed that both synthesis and hydrolysis rates decreased when the pH was decreased from 7.5 to 6.5 at 25 8C, but the rate of hydrolysis diminished faster than that of synthesis, thus improving the selectivity. Youshko et al. (2002) affirm that selectivity depends on pH and on the concentration of 6-APA during the synthesis of ampicillin at 25 8C using soluble enzyme. Accordingly, they propose a profile of pH to optimize a fed-batch run, starting at pH 7.0 and decreasing to 6.3 throughout the run. When PGA is immobilized, diffusional resistances in the pores of the biocatalyst may give rise to profiles of concentrations and pH (Schroen et al., 2002c). Ex perimental pH profiles inside the catalyst are measured by Spie et al. (1999), using fluorescence and a fiber of gel, for the hydrolysis of penicillin G by PGA. For an external pH of 8.0, pH differences between 1.0 and 2.5 are observed. Certainly, more work should be done with respect to optimization regarding pH, especially for discontinuous reactors in the presence of immobilized PGA and with simultaneous crystallization of products. In this case, taking into account all the complex phenomena involved (selectivity versus productivity, solidliquid equilibrium of products or even reactants, diffusion delays in the catalyst pores, crystallization kinetics), an optimal trajectory for the pH could be sought. To accomplish this task, however, the pH-dependency of the kinetic constants should be considered or, alternatively, the equilibrium constants of dissociation of the charged groups of substrates and products must be known (Diender et al., 2000 followed this approach, using a simplified kinetic model for amoxicillin synthesis; see also van der Wielen et al., 1997, for the hydrolysis of penicillin). A difficulty here is that this approach, to be consistent, should also take into account the equilibrium of all the catalytically active residues of the enzymeundoubtedly a complicated undertaking. Schroen et al. (2001b) used a simplified kinetic model and estimated the temperature dependence of its constants for the synthesis of cephalexin. Ferreira et al. (2004) observed that low temperatures (4 8C) maximize the selectivity (S/H) in the synthesis of ampicillin (with PGA immobilized on agarose), at the

expense of the productivity. Goncalves (2001) noticed the same behavior for amoxicillin. Once again, an optimization problem presents itself. van Langen et al. (1999) synthesized cephalexin in frozen medium ( 20 8C), with the reaction occurring in a liquid micro-interphase. The increase of selectivity was expressive (initial S/H rises from 1.3 at 20 8C to 25.0 at 20 8C). Generally speaking, in a trade-off between productivity and selectivity, and taking into account energy costs, mild temperatures (approx. 2025 8C) and pH between 6.0 and 6.5 are the usual choices for the kinetically controlled synthesis. 5.1.3. Inhibitory and activation effects of reactants and products on the reaction rates One of the most used derivatives of the side-chains (acyl donors) is their methyl esters. As a consequence, methanol is released to the medium (either because of the hydrolysis of the ester or the synthesis of the antibiotic). Fernandez-Lafuente et al. (1998) studied the effect of methanol on the selectivity for two different sidechains (phenylglycine methyl ester, PGME and mandelic acid) and three nucleus (6-APA, 7-ADCA and 7ACA). The behavior of E. coli PGA is not monotonic: for PGME, methanol increases S/H, but increases the hydrolysis of the ester as well. With mandelic acid, methanol increases S/H for both the antibiotic and the ester. Goncalves et al. (2003), studying the synthesis of amoxicillin, observed that methanol inhibits the undesired hydrolyses. Travascio et al. (2002b) noticed the same effect for the synthesis of cephalexin. Ferreira et al. (2000) (for ampicillin) and Goncalves et al. (2000) (for amoxicillin) observed that the antibiotic is an inhibitor of the hydrolysis of the ester, and vice-versa. Sheldon et al. (2001) reported an unexpected behavior: the presence of d-phenylglycine increased S/H twice, at the beginning of a batch synthesis of cephalexin. L-PG did not present any effect. 5.2. Open questions 5.2.1. Understanding the deacylation step during synthesis: a decisive point to define optimal concentrations and molar ratio of substrates The acylation mechanism of PGA seems to be well established. Yet, the deacylation step in the presence of h-lactam nuclei remains disputable. PGA is very specific for phenylacetyl groups, but admits several side groups in the hydrolysis reactions. As already pointed

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

35

out, direct X-ray analysis of protein crystals saturated with h-lactam nuclei is a difficult task and therefore a proposed mechanism for deacylation will have to rely on kinetic assays. One point not contested in the literature is that the hlactam nucleus must bind to the enzyme before its nucleophilic attack on the acyl-enzyme complex. Kasche et al. (1984) proposed a test to verify whether the nucleophile is bound to the enzyme before deacylation. Fig. 3 helps to understand the underlying idea of this test: if an unbound nucleus started the nucleophilic attack, the ratio (k H/k S)app would be invariant with respect to the concentrations of substrates and products. The gray box in Fig. 3 stands for any of the mechanisms where the acyl-enzyme complex is involved. From Fig. 3, the rates of synthesis and hydrolysis (m AN and m AOH) can be quantified in terms of bapparentQ kinetic constants, namely: mAN kS app CNH CEA V mAOH kH app CH2 O CEA 1 2

Here, C NH, C EA and C H2O are the concentrations of the h-lactam nucleus, of the acyl-enzyme complex and of water, respectively. (k S)app and (kVH)app are pseudo-kinetic, apparent, constants of the simplified reaction bschemeQ (see Fig. 3). Lumping water concentrations in a new constant, (k H)app = C H2O(k H app, the ratio of V) apparent deacylation constants becomes:   kH mAOH CNH 3 kS app mAN Of course, if the nucleophiles (water or the nucleus) did not bind to the enzyme, the constants would

BH EH + AB NH EA (kS)app EH + AN
Fig. 3. A black-box representation of the kinetically controlled synthesis of h-lactam antibiotics. EH = enzyme. EA= acyl-enzyme complex. NH = h-lactam nucleus. AB = ester or amide side-chain precursor. AOH = hydrolysis product. BH = alcohol or ammonia. AN = antibiotic. (k S)app and (k H)app are the apparent kinetic constants for synthesis and for hydrolysis, respectively. (k H)app lumps the acyl donor derivative (ester and amine) and the antibiotic rates of hydrolysis.

(kH)app H2 O

EH + AOH

not be bapparentQ, and their ratio would be invariant with respect to the concentrations of substrates and products. Kasche et al. (1984), Kasche (1986) and Goncalves et al. (2002a), among others, plotting experimental data of CNH mmAOH against C NH, showed that there is a linear AN dependency of (k H/k S)app with respect to the h-lactam nucleus (6-APA) concentration, C NH. Hence, 6-APA must bind to the enzyme before the deacylation step. Another important aspect of the reaction kinetics is whether the rate-determining step is acylation or deacylation. Alkema et al. (2003) used stopped-flow experiments for hydrolysis of penicillin and ampicillin analogues. None of the analogues presented an initial burst, an indication that the acyl-enzyme formation is the rate-controlling step of the hydrolysis. The phenylacetyl-enzyme hydrolysis rate constant was of the order of 1000 s 1 while the phenylglycine-enzyme one was 75 s 1. A proton inventory for these reactions after kinetic solvent isotope effect (KIE) experiments indicated that one proton was transferred while in the transition state, confirming the acylation mechanism described beforehand. The influence of saturating PGA with 6-APA, preceding the synthesis of amoxicillin, was also tested by Goncalves (2001). No statistically meaningful differ ence was detected between S/H ratios at pH 7.0, but at pH 6.0 the selectivity increased when 6-APA was previously linked to PGA. Hence, at lower pHs, where less neutral Ser h1 amines are available, the importance of the nuclei binding to PGA becomes noticeable. The most common approach in the literature regarding the deacylation step during the antibiotic synthesis admits that all acyl-enzyme intermediates are formed before the adsorption of 6-APA to the enzyme. This hypothesis simplifies considerably the rate equations and was assumed by Youshko and Svedas (2000), Youshko et al. (2003), Schroen et al. (2001b) and Alkema et al. (2003). This is based on an analogy with the serine-proteases mechanism (especially that of a-chymotrypsin), with competing nucleophiles (for instance, during the synthesis of peptides in aqueous medium). Fig. 4A shows the corresponding kinetic scheme. A different simplified kinetic model was proposed by Goncalves et al. (2000, 2003): the acyl-enzyme complex might be formed either after or before the adsorption of 6APA, but only in the first case would the antibiotic be synthesized. The rate of formation of the acylenzyme complex would not be affected by the presence of adsorbed 6-APA.

36

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

A
EH + AB

KAB

EH-AB

k2

EA + NH K NH

k3

EH + AOH

BH

k5

NH

EAN

k4 k-4

KAN EH-AN EH + AN

EH + AB + NH kN k-N

k1 k-1

EH-AB + NH

k2 BH kN

EA + NH k-N

k3

EH + AOH

H2O

EH : NH H2O

EA : NH k4 EH + AOH : NH

k5 k-5

EH + AN

B
EH + AB + NH kN k-N k5 k-5 k1 k-1 EH-AB + NH kN k-N k2 EA + NH kN k-N k7 k-7 k3 EH + AOH k4 k-4 H2O BH EH-AOH

EH + AB : NH

EH-AB : NH

k6

EA : NH k8 EH + AOH : NH

EH + AN

H2O

Fig. 4. (A) Simplified mechanism of the kinetically controlled enzymatic synthesis of h-lactam antibiotics. (B) More complex mechanistic scheme for the same reactions. EH = free enzyme; AB = acyl-donor derivative; NH = 6-APA; BH = leaving group; AOH = side-chain; AN = antibiotic; EA= acyl-enzyme complex; EAd d NH = acyl-enzymenucleus complex. Non-covalent bonds are depicted by dots.

In either case, the rate equations of these simplified models would provide the same functional expression for the selectivity (synthesis/hydrolysis ratio, S/H = m AN/m AOH) in the absence of antibiotic:     S mAN P1 CNH 4 H CAN 0 mAOH CAN 0 P2 P3 CNH

where P 1, P 2 and P 3 are lumped kinetic constants and C AN is the concentration of antibiotic. Therefore, for these models, the ratio of initial rates of synthesis and of hydrolysis would be invariant with respect to the concentration of the acyl derivative (ester or amide). More complex rate equations appear if the nucleophilic attack to the acyl-enzyme complex

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741

37

comes from a h-lactam nucleus that was already linked to the enzyme before the acylation step (see Fig. 4B). For such a bcompleteQ model, the

selectivity (calculated using initial rates, i.e., without any antibiotic in the medium) is given by:

S=H CAN 0

2 CNH P1 P2 CAB P3 CNH P4 CAB CNH P5 CNH 3 2 2 P6 P7 CAB P8 CNH P9 CNH P10 CNH P11 CAB CNH P12 CNH CAB

where P i (i = 1,2,. . .,12) are lumped kinetic constants. A test for these models, hence, would be plotting S/H as a function of C AB for constant C NH. This is a controversial point. Ribeiro et al. (2005), studying the synthesis of ampicillin from phenylglycine methyl ester (PGME) covering high concentrations of substrates at pH 6.5, obtained S/H values that vary with C AB, according to Eq. (5). Moreover, high concentrations of 6-APA strongly inhibited the hydrolysis of the ester and of the antibiotic. On the other hand, Youshko and Svedas (2000), working with the same system, observed that S/H followed Eq. (4). Goncalves et al. (2002a), working with amoxicillin, synthesized from 6-APA and PHPGME at pH 6.5, showed experimental evidence that 6-APA (NH in Fig. 4) may either inhibit (at high concentrations) or activate (at low concentrations) the formation of the acyl-enzyme complex (EA). An implication of this behavior is that S/H can vary with C AB. Travascio et al. (2002b), synthesizing cephalexin from PGME and 7-amino-3-deacetoxycephalosporanic acid (7-ADCA) at pH 6.5, with PGA immobilized on a membrane, concluded that maximum yields are obtained for a 4:1 molar excess of the ester (although the authors did not extend their assays to the less-than-1:1 region). This is actually a critical point for the optimization of the enzymatic reactor. For instance, to design an optimal fed-batch run one must define the start up and the feeding trajectory, working either with excess of nucleus, excess of acyl donor or close to the stoichiometric ratio. The kinetic model will provide the basis for this decision. Once more, it should be stressed that the linkage of the nucleus to the enzyme before or after the acylation step is the key point in this discussion. As already noticed, no X-ray crystallographic data exist with the nucleus as ligand, however, from the available information one may translate the dilemma into the following question: in the simplified model, the acyl-enzyme formation would be necessary to induce the conformational change involving Phe a146 and Arg a145, opening the site for the binding of the h-lactam nucleus. The bcompleteQ model assumes that the nucleus might also

bind on its site (probably, there would be weak interactions with Phe a146, Arg a145 and Phe h71) without hindering the acyl-enzyme formation and the release of an alcohol or ammonia molecule after enzyme acylation by the side-chain derivative. 5.2.2. Inhibitory effect of the acyl donor derivative Literature disagrees regarding a possible inhibitory effect of the acyl moiety (product of the hydrolysis of the antibiotic and of the ester/amide derivative) on the rate of hydrolysis. Goncalves et al. (2003) affirmed that d-p-hydroxyphenilglycine (PHPG) inhibited the hydrolysis of PHPGME and of ampicillin, while Alkema et al. (2004) observed that phenylglycine (PG) did not inhibit the hydrolysis of NIPGB (d-2-nitro-5-[(phnylglycil)amino] benzoic acid). These authors suggested that the positive charge of the Ca-amine prevented the linking of PG to the non-polar site. Fernandez-Lafuente et al. (1996b), studying the effect of pH on the synthesis of ampicillin, also suggested that the low affinity of the enzyme with PG derivatives was due to the protonated amine. At variance with this, Ospina et al. (1996), changing pH between 5.5 and 8.5, observed that, although the activity of the enzyme for the hydrolysis of PGME changed with pH, the affinity constant (K M) was invariant. This is an indication that the charged Caamine did not disturb the binding of PG (PGME has pK a c 7). 5.2.3. Do classical mechanistic models represent a convenient approach for this system? Goncalves et al. (2002b) showed that if the complete mechanism in Fig. 4B is followed, the problem will probably become intractable using the classical BriggsHaldane approach. Some of the numerous parameters of the rate equations will have a very high statistical correlation, and their estimation from kinetic experiments becomes unfeasible. If integrated reactors (i.e., with simultaneous precipitation of the products and even with insoluble substrates) were to be used, a mechanistic model of the system would have to consider the kinetics of crystallization. Hence, the complexity of the mathematical model would become even greater.

38

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741 Abian O, Wilson L, Fernandez-Lorente G, Palomo JM, Fuentes M, Fernandez-Lafuente R, et al. Preparation of hydrophilic microenvironments (polymeric salts) surrounding enzyme molecules New enzyme derivatives to be used in any reaction medium. J Mol Catal B Enzym 2002;19:295 503. Abian O, Mateo C, Fernandez-Lorente G, Guisan JM, FernandezLafuente R. Thermodinamically controlled synthesis of amide bonds catalyzed by highly organic solvent-resistant penicillin acylase derivatives. Biotechnol Prog 2004;20:117 21. Abian O, Grazu V, Hermoso J, Gonzalez R, Garcia JL, FernandezLafuente R, et al. Stabilization of penicillin G acylase freom Escherichia coli: site-directed mutagenesis of the protein surface to increase multipoint covalent attachment. Appl Environ Microbiol 2004;70:1249 51. Alkema WBL, Hensgens CMH, Kroezinga EHK, de Vries E, Floris R, van der Laan J-M, et al. Characterization of the h-lactam binding site of penicillin acylase of Escherichia coli by structural and sitedirected mutagenesis studies. Protein Eng 2000;13:857 63. Alkema WBL, Dijkhuis A-J, Vries E, Janssen DB. The role of hydrophobic active-site residues in substrate specificity and acyl transfer activity of penicillin acylase. Eur J Biochem 2002; 269:2093 100. Alkema WBL, de Vries E, Floris R, Janssen DB. Kinetics of enzyme acylation and deacylation in the penicillin acylase-catalyzed synthesis of h-lactam antibiotics. Eur J Biochem 2003; 270:3675 83. Alkema WBL, Hensgens CMH, Snijder HJ, Keizer E, Dijkstra BW, Janssen DB. Structural and kinetic studies on ligand binding in wild-type and active-site mutants of penicillin acylase. Protein Eng Des Sel 2004;17:473 80. Arroyo M, de la Mata I, Acebal C, Castillon MP. Biotechnological applications of penicillin acylases: state-of-the-art. Appl Microbiol Biotechnol 2003;60:507 14. Barends TRM, Polderman-Tijmes JJ, Jekel PA, Hensgens CMH, de Vries EJ, Janssen DB, et al. The sequence and crystal structure of the a-amino acid ester hydrolase from Xanthomonas citri define a nes family of h-lactam antibiotic acylases. J Biol Chem 2003; 278:23076 84. Blinkovsky AM, Markaryan AN. Synthesis of h-lactam antibiotics containing a-aminophenylacetyl group in the acyl moiety catalyzed by d()-phenylglycyl-h-lactamide amidohydrolase. Enzyme Microbial Technol 1993;15:965 73. Bock A, Wirth R, Schmid G, Shumacher G, Lang G, Buckel P. The penicillin acylase from Escherichia coli ATCC 11105 consists of two dissimilar subunits. FEMS Microbiol Lett 1983;20:135 9. Brannigan JA, Dodson GG, Duggleby HJ, Moody PCE, Smith JL, Tomchick DR, et al. A protein catalytic framework with an Nterminal nucleophile is capable of self activation. Nature 1995; 378:416 9. Brannigan JA, Dodson GG, Done SH, Hewitt L, McVey CE, Wilson KS. Structural studies of penicillin acylase. Appl Biochem Biotechnol 2000;88:313 9. Bruggink A, editor. Synthesis of h-lactam antibiotics. Dordrecht7 Kluwer; 2001. Bruggink A, Roy PD. Industrial synthesis of semi-synthetic antibiotics. In: Bruggink A, editor. Synthesis of h-lactam antibiotics. Dordrecht7 Kluwer; 2001. p. 12 54. Bruns W, Hoppe J, Tsai H, Bruning HJ, Maywald F, Collins JE, et al. Structure of the penicillin acylase gene from Escherichia coli: a periplasmic enzyme that undergoes multiple proteolytic processing. J Mol Appl Genet 1985;3:36 44.

An alternative approach to this problem is the use of neural networks as empirical models, providing the values of the reaction rates (Goncalves et al., 2002b). The solidliquid mass transfer in integrated reactors could also be lumped in the neural-network model, if enough experimental data are available. The same could be said concerning pH effects. 6. Concluding remarks Process optimization is essential to make the enzymatic synthesis of h-lactam antibiotics economically feasible. This, however, is an exceedingly complex process. The mainstream configuration of the industrial reactor, a fed-batch system with immobilized E. coli PGA, in aqueous medium, with product crystallization, is appropriate for the application of model-based dynamic optimization, optimal control, among other techniques of reactor engineering. The state of the art, as mentioned earlier, indicates that we are still far from a complete, mechanistic model that could take into account all the phenomena involved: pH, temperature and concentration effects on the reaction rates, liquidsolid mass transfer, intraparticle diffusion, and so forth. Much progress has already been achieved in understanding the catalytic mechanism and the configuration of the active site, helping on the accounting of the effect of each reaction component on the reaction rates. The knowledge of these interactions is ultimately the keystone to define operational conditions for the industrial reactor. For example, the definition of an optimal molar ratio between h-lactam nucleus and acyl derivative, along the course of the reaction, depends on the comprehension of the deacylation mechanism, in the presence of the nucleus, which is not well established yet. In conclusion, for the biochemical engineer two approaches are possible: using simplified models, which consider only some selected aspects that are supposedly more relevant for a particular situation, or, alternatively, turning to empirical or hybrid models such as neural networks, which may be very effective, but are strongly dependant on the quality of the data basis used for model fitting and validation, and do not support extrapolation. References
Abian O, Mateo C, Fernandez-Lorente G, Palomo JM, FernandezLafuente R, Guisan JM. Stabilization of immobilized enzymes against water-soluble organic cosolvents and generation of hyperhydrophilic micro-environments surrounding enzyme molecules. Biocatal Biotransform 2001;19:489 583.

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741 Cao L, van Langen LM, van Rantwijk F, Sheldon RA. Cross-linked aggregates of penicillin acylase: robust catalysts for the synthesis of h-lactam antibiotics. J Mol Catal B Enzym 2001;11:665 70. Clausen K, Dekkers RM. Process for preparation of h-lactams at constantly high concentration of reactants USPTO 6048708; 2000. Cole M. Hydrolysis of penicillins and related compounds by the cellbound penicillin acylase of Escherichia coli. Biochem J 1969;115:733 9. Diender MB, Straathof AJJ, Van der Wielen LAMS, Ras C, Heijnen JJ. Feasibility of the thermodynamic controlled synthesis of amoxicillin. J Mol Catal A Chem 1998;5:249 53. Diender MB, Straathof AJJ, Does T, Zomerdijk M, Heijnen JJ. Course of pH during the formation of amoxicilin by a suspension-to-suspension reaction. Enzyme Microb Technol 2000; 27:576 82. Done SH, Brannigan JA, Moody PC, Hubbard RE. Ligand-induced conformational change in penicillin acylase. J Mol Biol 1998;284:463 75. Duggleby HJ, Tolley SP, Hill CP, Dodson EJ, Dodson GD, Moody PCE. Penicillin acylase has a single-amino-acid catalytic centre. Nature 1995;373:264 8. Elander RP. Industrial production of h-lactam antibiotics. Appl Microbiol Biotechnol 2003;61:385 92. Fernandez-Lafuente R, Rossel CM, Guisan JM. Enzyme reaction engineering: synthesis of antibiotics catalyzed by stabilized penicillin G acylase in the presence of organic cosolvents. Enzyme Microb Technol 1991;13:898 905. Fernandez-Lafuente R, Rossel CM, Guisan JM. The use of stabilized penicillin acylase derivatives improves the design of kinetically controlled synthesis. J Mol Catal A Chem 1995;101:91 7. Fernandez-Lafuente RF, Rossel CM, Guisan JM. Dynamic reaction design of enzymes biotransformations in organic media: equilibrium-controlled synthesis of antibiotics by penicillin G acylase. Biotechnol Appl Biochem 1996;24:139 43. Fernandez-Lafuente R, Rossel CM, Piatkowska B, Guisan JM. Syn thesis of antibiotics (cephaloglycin) catalyzed by penicillin G acylase: evaluation and optimization of different synthetic approaches. Enzyme Microb Technol 1996;19:9 14. Fernandez-Lafuente RF, Rossel CM, Guisan JM. The presence of methanol exerts a strong and complex modulation of the synthesis of different antibiotics by immobilized penicillin G acylase. Enz Microbiol Technol 1998;23:305 10. Fernandez-Lafuente R, Hernandez-Justiz O, Mateo C, Terrini M, Fernandez-Lorente G, Moreno MA, et al. Biotransformations catalyzed by multimeric enzymes: stabilization of tetrameric ampicillin acylase permits the optimization of ampicillin synthesis under dissociation conditions. Biomacromolecules 2001a; 2:95 104. Fernandez-Lafuente R, Hernandez-Justiz O, Mateo C, Terrini M, Alonso J, Garcia-Lopez JL, et al. Stabilization of a tetrameric enzyme (a-amino acid ester hydrolase from Acetobacter turbidans) enables a very improved performance of ampicillin synthesis. J Mol Catal B Enzym 2001b;11:633 8. Ferreira ALO, Goncalves LRB, Giordano RC, Giordano RLC. A simplified kinetic model for the side reactions occurring during the enzymatic synthesis of ampicillin. Braz J Chem Eng 2000;17:835 9. Ferreira ALO, Giordano RLC, Giordano RC. Improving the selectivity and productivity of enzymatic synthesis of ampicillin with immobilized penicillin G acylase. Braz J Chem Eng 2004; 21:519 29.

39

Galan B, Garcia JL, Prieto MA. The paaX repressor, a link between penicillin G acylase and the phenyl-acetyl coenzyme A catabolism of Escherichia coli. W J Bacteriol 2004;186:2215 20. Giordano RC, Giordano RLC, Ferreira ALO. Process for protection of insoluble enzymatic biocatalysts, biocatalyst obtained thereof and bioreactor with the immobilized biocatalyst WO 04/050822A1; 2004. Goncalves LRB, Fernandez-Lafuente R, Guisan JM, Giordano RLC. A kinetic study of the synthesis of amoxicillin using penicillin G acylase immobilized on agarose. Appl Biochem Biotechnol 2000;8486:931 45. Goncalves, L.R.B. Kinetic study of the enzymatic synthesis of amox icillin catalyzed by penicillin G acylase immobilized on agarose. PhD thesis, UFSCar, Brazil; 2001 [in Portuguese]. Goncalves LRB, Fernandez-Lafuente R, Guisan JM, Giordano RLC. The role of 6-aminopenicilanic acid on the kinetics of the enzymatic synthesis of amoxicillin catalyzed by penicillin G acylase immobilized on glyoxyl-agarose. Enzyme Microb Technol 2002a;31:464 71. Goncalves LRB, Sousa R Jr, Fernandez-Lafuente R, Guisan JM, Giordano RLC, Giordano RC. Enzymatic synthesis of amoxicillin: avoiding limitations of the mechanistic approach for reaction kinetics. Biotechnol Bioeng 2002b;80:622 31. Goncalves LRB, Fernandez-Lafuente R, Guisan JM, Giordano RC, Giordano RLC. Inhibitory effects in the side reactions occurring during the enzymatic synthesis of amoxicillin: p-hydroxyphenylglycine methyl ester and amoxicillin hydrolysis. Biotechnol Appl Biochem 2003;38:77 85. Guisan JM. Aldehyde-agarose gels as activated supports for immobilizationstabilization of enzymes. Enzyme Microb Technol 1988;10:375 82. Hernandez-Justiz O, Fernandez-Lafuente R, Terreni M, Guisan JM. Use of aqueous two-phase systems for in situ extraction of water soluble antibiotics during their synthesis by enzymes immobilized on porous supports. Biotechnol Bioeng 1998;59:73 9. Illanes A, Cabrera Z, Wilson L, Aguirre C. Synthesis of cephalexin in ethylene glycol with glyoxil-agarose immobilized penicillin acylase: temperature and pH optimization. Proc Biochem 2003; 39:111 7. Illanes A, Anjari MS, Altamirano C, Aguirre C. Optimization of cephalexin synthesis with immobilized penicillin acylase in ethylene glycol medium at low temperatures. J Mol Catal B Enzym 2004;30:94 103. Kaasgaard SG, Karlsen L. Process for separation of two solid components, WO 92/12782; 1992. Kasche V. Mechanism and yields in enzyme catalyzed equilibrium and kinetically controlled synthesis of h-lactam antibiotics peptides and other condensation products. Enzyme Microb Technol 1986;8:4 16. Kasche V, Galunsky B. Enzyme catalyzed biotransformations in aqueous two-phase systems with precipitated substrate and/or product. Biotechnol Bioeng 1994;45:261 7. Kasche V, Haufler U, Zollner R. Kinetic studies on the mechanism of the penicillin acylase-catalyzed synthesis of ampicillin and benzylpenicillin. Hoppe-Seylers Physiol Chem 1984;365:1435 43. Kasche V, 0 Haufler U, Riechmann K. Equilibrium and kinetically controlled synthesis with enzymes: semisynthesis of penicillins and peptides. Methods Enzymol 1987;136:280 93. Kasche V, Lummer K, Nurk A, Piotrashke E, Rieks A, Atreva A, et al. Intramolecular autoproteolysis initiates the maturation of penicillin amidase from Escherichia coli. Biochim Biophys Acta 1999;1433:76 86.

40

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741 Schroen CGPH, Nierstrasz VA, Moody HM, Hoogschagen MJ, Kroon OJ, Bosma R, et al. Modeling of the enzymatic kinetic synthesis of cephalexininfluence of substrate concentration and temperature. Biotechnol Bioeng 2001;73:171 8. Schroen CGPH, Nierstrasz VA, Bosma R, Dijskstra ZJ, Vandesandt EJAX, Beeftink HH, et al. Process design for adipyl-7-ADCA hydrolysis. Biotechnol Prog 2002a;18:745 51. Schroen CGPH, Nierstrasz VA, Bosma R, Kroon PJ, Tjeerdsma PS, Devroom E, et al. Integrated reactor concepts for the enzymatic kinetic synthesis of cephalexin. Biotechnol Bioeng 2002b; 80:144 55. Schroen CGPH, Fretz CB, Debruin VH, Berendsen W, Moody HM, Roos EC, et al. Modelling of the enzymatic kinetically controlled synthesis of cephalexininfluence of diffusion limitation. Biotechnol Bioeng 2002c;80:331 40. Shewale JG, Deshpande BS, Sudhakaran VK, Ambedkar SS. Penicillin acylases: applications and potentials. Proc Biochem Int 1990;25:97 103. Sheldon RA, van Rantwijk F, van Langen LM, Wegman MA, Cao L, Janssen MHA. Biocatalysts and biocatalysis in the synthesis of hlactam antibiotics. In: Bruggink A, editor. Synthesis of h-lactam antibiotics. Dordrecht7 Kluwer; 2001. p. 130. Sio CF, Quax WJ. Improved h-lactam acylases and their use as industrial biocatalysts. Curr Opin Biotechnol 2004;15: 349 355. Spie A, Schlothauer RC, Hinrichs J, Scheidat BE, Kasche V. pH gradients in immobilized amidases and their influence on rates and yields of h-lactam hydrolysis. Biotechnol Bioeng 1999; 62:267 77. Svedas VK, Margolin AL, Borisov IL, Berezin IV. Kinetics of enzymatic synthesis of benzylpenicillin. Enzyme Microb Technol 1980; 2:313 7. Svedas VK, Savchenko MV, Belster AI, Guranda DF. Enantioselective penicillin acylase-catalyzed reactions: factors governing substrate and sterospecificity of the enzyme. Ann N Y Acad Sci 1996;799:659 69. Rocchietti S, Urrutia ASV, Pregnolato M, Tagliani A, Guisan JM, Fernandez-Lafuente R, et al. Influence of the enzyme derivative preparation and substrate structure on the enantioselectivity of penicillin G acylase. Enzyme Microb Technol 2002; 31:89 93. Travascio P, Zito E, De Maio A, Schroen CGPH, Durante D, De Luca P, et al. Advantages of using non-isothermal bioreactors for the enzymatic synthesis of antibiotics: the penicillin G acylase as enzyme model. Biotechnol Bioeng 2002a;79:334 46. Travascio P, Zito E, Portaccio M, Diano N, Grano V, Di Martino S, et al. Enzyme reaction engineering: effect of methanol on the synthesis of antibiotics catalyzed by immobilized penicillin G acylase under isothermal and non-isothermal conditions. Biotechnol Prog 2002b;18:975 85. Valle F, Balbas P, Merino E, Bolivar F. The role of penicillin amidases in nature and in industry. TIBS 1991;16:36 40. van der Wielen LAM, van Bue lMJ, Straathof AJJ, Luyben KChAM. Modelling the enzymatic deacylation of penicillin G: equilibrium and kinetic considerations. Biocatal Biotransform 1997;15:121 46. van Langen LM, de Vroom E, van Rantwijk F, Sheldon R. Enzymatic synthesis of h-lactam antibiotics using penicillin-G acylase in frozen media. FEBS Lett 1999;456:89 92. Wegman MA, Janssen MHA, van Rantwijk F, Sheldon RA. Towards biocatalytic synthesis of h-lactam antibiotics. Adv Synth Catal 2001;343:559 76.

Kazan D, Ertan H, Erarslan A. Stabilization of penicillin G acylase against pH by chemical cross-linking. Proc Biochem 1996;31:135 40. Kim GM, Lee SB. Effect of organic solvents on penicillin acylasecatalyzed reactions: interaction of organic solvents with enzymes. J Mol Catal B Enzym 1996;1:181 90. Kim Y, Kim S, Earnest TN, Hol WGJ. Precursor structure of cephalosporin acylase Insights into autoproteolytic activation in a new N-terminal hydrolase family. J Biol Chem 2002; 277:2823 9. Margolin AL, Svedas VKS, Berezin IV. Substrate specificity of penicillin amidase from Escherichia coli. Biochim Biophys Acta 1980;616:283 9. McVey CE, Walsh MA, Dodson GG, Wilson KS, Brannigan JA. Crystal structures of penicillin acylase enzymesubstrate complexes: structural insights into the catalytic mechanism. J Mol Biol 2001;313:139-5. Murzin AG, Brenner SE, Hubbard T, Chothia C. A structural classification of proteins database for the investigation of sequences and structures. J Mol Biol 1995;247:536 40. Ospina S, Barzana E, Ramirez OT, Lopez-Munguia A. Effect of pH in the synthesis of ampicillin by penicillin acylase. Enzyme Microb Technol 1996;19:462 9. Ozturk DC, Kazan D, Erarslan A. Stabilization and functional properties of Escherichia coli penicillin G acylase by covalent conjugation of anionic polysaccharide carboxymethylcellulose. World J Microb Biotechnol 2002;18:881 8. Parmar A, Kumar H, Marwaha SS, Kennedy JF. Advances in enzymatic transformation of penicillins to 6-aminopenicillanic acid (6APA). Biotechnol Adv 2000;18:289 301. Phadtare S, Parekh P, Gole A, Patil M, Pundle A, Prabhune A, et al. Penicillin G acylase-fatty lipid biocomposite films show excellent catalytic activity and long term stability/reusability. Biotechnol Prog 2002;18:483 8. Polderman-Tijmes JJ, Jeckel PA, de Vries EJ, van Merode AE, Floris R, van der Laan JM, et al. Cloning, sequence analysis, and expression in Eschirichia coli of the gene encoding an a-amino acid ester hydrolase from Acetobacter turbidans. Appl Eviron Microbiol 2002;68:211 8. Polderman-Tijmes JJ, Jeckel PA, Jeronimus-Stratingh CM, Bruins AP, van der Laan JM, Sonke T, et al. Identification of the catalytic residues of a-amino acid ester hydrolase from Acetobacter turbidans by labeling and site-directed mutagenesis. J Biol Chem 2002;277:29474 82. Rajendhran J, Gunasekaran P. Recent biotechnology interventions for developing improved penicillin G acylases. J Biosci Bioeng 2004;97:1 13. Ribeiro MPA, Ferreira ALO, Giordano RLC, Giordano RC. Selectivity of the enzymatic synthesis of ampicillin by E coli PGA in the presence of high concentrations of substrates. J Mol Catal B Enzym 2005;33:81 6. Schoemaker HE, Boesten WHJ, Broxterman QB, Roos EC, Kaptein B, van den Tweel WJJ, et al. Application of enzymes in industrial organic synthesis. Chimia 1997;51:308 10. Schroen CGPH, Nierstrasz VA, Kroon PJ, Bosma R, Janssen AEM, Beeftink HHE, et al. Thermodynamically controlled synthesis of h-lactam antibiotics Equilibrium concentrations and side-chain properties. Enzyme Microb Technol 1999;24:498 506. Schroen CGPH, Eldin MSM, Janssen AEM, Mita GDE, Tramper P. Cephalexin synthesis by immobilised penicillin G acylase under non-isothermal conditions: reduction of diffusion limitation. J Mol Catal B Enzym 2001;15:163 72.

R.C. Giordano et al. / Biotechnology Advances 24 (2006) 2741 Wei D-Z, Zhu J-H, Cao X-J. Enzymatic synthesis of cephalexin in aqueous two-phase systems. Biochem Eng J 2002;11:95 9. Wilson L, Illanes A, Abian O, Pessela BCC, Fernandez-Lafuente R, Guisan JM. Co-aggregation of penicillin G acylase and polyionic polymers: an easy methodology to prepare enzyme biocatalysts stable in organic media. Biomacromol 2004a;5:852 7. Wilson L, Illanes A, Pessela BCC, Abian O, Fernandez-Lafuente R, Guisan JM. Encapsulation of crosslinked penicillin G acylase aggregates in lentikats: evaluation of a novel biocatalyst in organic media. Biotechnol Bioeng 2004b;86:558 62. Yoon J, Oh B, Kim K, Park J, Han D, Kim KK, et al. A bound water molecule is crucial in initiating autocatalytic precursor activation in an N-terminal hydrolase. J Biol Chem 2004;279:341 7. Youshko MI, Svedas VK. Kinetics of ampicillin synthesis catalyzed by penicillin G acylase form E coli in homogeneous and heterogeneous systems Quantitative characterization of nucleophile reactivity and mathematical modeling of the process. Biochem (Moscow) 2000;65:367 75.

41

Youshko MI, Langen LM, Vroom E, Moody HM, Rantwijk F, Sheldon RA, et al. Penicillin acylase-catalyzed synthesis of ampicillin in baqueous solutionprecipitateQ systems High substrate concentration and supersaturation effect. J Mol Catal B Enzym 2000; 10:509 15. Youshko MI, Langen LM, Vroom E, Rantwijk F, Sheldon RA, Svedas VK. Highly efficient synthesis of ampicillin in baqueous solution precipitateQ systems: repetitive addition of substrates in a semicontinuous process. Biotechnol Bioeng 2001;73:426 30. Youshko MI, Langen LM, Vroom E, Rantwijk F, Sheldon RA, Svedas VK. Penicillin acylase catalyzed ampicillin synthesis using a pH gradient: a new approach to optimization. Biotechnol Bioeng 2002;78:589 93. Youshko MI, Bukhanov AL, Svedas VK. Study of nucleophile binding in the penicillin acylase active center Kinetic analysis. Biochem (Moscow) 2003;68:334 8.