Aust. J. Bioi. Sci., 1984, 37, 257-65

A Sensitive Procedure for Screening Microorganisms for the Presence of Penicillin Amidase

W L. Baker Chemistry Department, Swinburne Institute of Technology, John Street, Hawthorn, Vic. 3122.

Abstract A procedure is described for screening bacteria for the presence of penicillin amidase. Cells, grown in the presence of phenylacetic acid, are incubated with phenoxymethylpenicillin (type I), benzylpenicillin (type II) or ampicillin and the 6-aminopenicillanic acid formed is detected and quantitatively estimated by its strong reaction with fluorescamine at pH 4. There is no requirement for separation of the penicillin substrate from the product but when a-aminobenzylpenicillin derivatives are used as enzyme substrates the amount of 6-aminopenicillanic acid formed must be determined by calculation. The procedure allowed positive and reliable identification of penicillin amidases in six organisms known to produce the enzyme and indicated that some of these enzymes had different properties in reactivity towards a-aminobenzylpenicillin derivatives.

Extra keywords: amoxycillin; epicillin.

Introduction Penicillin amidase (penicillin amidohydrolase, EC 3.5.1.11) is used to produce semisynthetic penicillins but its role in microbial metabolism has not been established. There are conflicting reports concerning its distribution in bacteria (Holt and Stewart 1964; Cole and Sutherland 1966) but these are primarily due to interpretation of data obtained from different methods of determination of enzyme activity. The usual chemical methods for measuring penicillin amidase activity estimate the 6-aminopenicillanic acid (6-APA) formed from the penicillin substrate (Batchelor et al. 1961; Bomstein and Evans 1965), but are neither sensitive nor specific for 6-APA. Vandamme and Voets (1974) considered that a rapid, reliable chemical technique did not exist for unequivocal detection and determination of penicillin amidase activity. The bioassay technique that has been used (Uri and Sztaricskai 1961) requires separation of 6-APA from substrate prior to phenylacetylation and determination of inhibition of microbial growth. This procedure is not convenient in rate studies and separation of 6-APA from ex-aminobenzylpenicillin derivatives, such as ampicillin, is not readily accomplished. The reaction of 6-APA with fluorescamine at pH 7 has been used to follow the kinetics of penicillin amidase (Veronese et al. 1981), but its use is restricted to low concentrations of purified enzyme because amino acids and polypeptides also react strongly at this pH. At pH 4, 6-APA develops a stronger fluorescence intensity with fluorescamine than at pH 7 (Baker 1983a). The reaction shows considerable specificity for 6-APA because polypeptides and amino. acids react minimally at this pH, and it has been possible to detect penicillin amidase activity against benzylpenicillin in whole-cell cultures without any purification. In the present work the reaction of 6-APA with fluorescamine at pH 4 0004-9417/84/040257$02.00 258 W. L. Baker

has been used to detect and estimate penicillin amidase activity of organisms deacylating phenoxymethylpenicillin (type I), benzylpenicillin (type II), and a-aminobenzylpenicillin derivatives.

Materials and Methods Chemicals Sodium benzylpenicillin, potassium phenoxymethylpenicillin and bovine serum albumin were obtained from Commonwealth Serum Laboratories, Parkville, Vic. 6-APA (96% pure) was obtained from Aldrich Chemicals, Milwaukee, Wisconsin, and L-tyrosine obtained from Sigma Chemicals, St Louis, Missouri, U.S.A. Fluorescamine was purchased from Roche Laboratories, Dee Why, N.S.W. Ampicillin and amoxycillin trihydrates came from the sources previously described (Baker 1983a) and epicillin was a gift from Mr R. Hems, of Squibb Laboratories, Noble Park, Melbourne. DL-Alanine, DL-leucine, glycine and N-glycylglycine were obtained from BDH Chemicals Ltd, Poole, United Kingdom.

Instrumental All absorbance readings were made on a Varian-Techtron 635 recording spectrophotometer. Total fluorescence measurements were made on the same instrument using the total fluorescence attachment, low intensity lamp source and an excitation wavelength of 390 nm for fluorescamine complexes.

Analytical Methods Fluorescamine solution was prepared in A.R. acetone at a concentration of 15 mg/IOO ml. Protein concentrations in dilutions of bacterial cell suspensions were measured in the absence of penicillins by the method of Lowry et al. (195 I), bovine serum albumin being used as a standard. Turbidity was controlled by cell blanks treated with distilled water rather than reducing colour reagent. Cell debris was removed by centrifugation after colour had developed and the small amount of colour sometimes found in the precipitate was ignored. The protein estimated in the suspensions under these conditions is referred to as soluble protein.

Source of Organisms Escherichia coli ATCC9637 was obtained from the Beecham Research Laboratories, Brockham Park, Surrey, U.K. Mr V. Davey of the Commonwealth Serum Laboratories provided Escherichia coli ATCCll105 and the Squibb Institute for Medical Research, Princeton, New Jersey, U.S.A. provided the culture of Bacillus megaterium ATCC14945. Acinetobacter calcoaceticus ATCC21288, Nocardia globerula ATCC21022 and Alcaligenes faecalis ATCCI5246 were kindly donated by Dr R. Ohira of The Tokyo Research Laboratories of Kwoya Hakka Kogyo Co. Ltd, Machida, Japan, and correspond to strains originally designated Micrococcus ureae KY3767, N. globerula KY3901 and Pseudomonas aeruginosa KY3951 respectively (Nara et al. 1971).

Maintenance and Growth of Organisms Organisms. maintained at 4°C on nutrient agar slopes, were inoculated into nutrient broth and grown at 37°C for 48 h; 1 ml of this culture was inoculated into 100 ml of sterile growth medium in 300-ml Erlenmeyer flasks which were then shaken at 24°C and 200 rpm in a Gallenkamp orbital incubator for 48-60 h. Cells were harvested by centrifugation at 7710 g for 20 min, washed twice in cold distilled water and resuspended in 10 ml of O· OS M phosphate buffer, pH 7· 8. The culture of B. megaterium was used without concentration or washing because the enzyme from this organism is extracellular (Chiang and Bennett 1967). Two growth media were used for penicillin amidase production: medium A contained phenylacetic acid as inducing agent and its composition was as previously described (Baker 1983b); medium B was of the same composition as that used by Nara et al. (1971). E. coli ATCC9637 cells were ruptured by treatment at 13· 8 X 104 kPa in an Aminco French press (2·5 cm diam piston). Some spheroplasts remained in solution but by using phase-contrast microscopy it was estimated that over 80% of the cells were disrupted. Screening Microorganisms for Penicillin Amidase 259

Detection and Estimation of Penicillin Amidase Activity Bacterial suspension (I ml) obtained after growth as described above was mixed with 4 ml of a solution of penicillin in 0·05 M phosphate buffer, pH 7·8. Tubes were incubated for 16 h in a reciprocating water-bath at 37"C. Solutions were diluted I: 10 to 1: 100 with O· 5 M acetate buffer, pH 4, and I ml treated with 0·5 ml of fluorescamine solution. The fluorescence was examined under ultraviolet light (366 nm) I h later. In quantitative analyses the same ratio of cell suspension to penicillin solution as described above was used. Dilutions with 0·5 M acetate buffer, pH 4, were usually in the range 1 : 50-1 : 500. After addition of fluorescamine solution (0·5 ml/ml dilute suspension) the final volume was made to 3 ml with buffer at pH 4. After fluorescence developed the tubes were centrifuged and total fluorescence readings, taken I h after addition offluorigenic agent, were compared with those ofa 6-APA standard curve. All calculations were corrected for blanks containing whole cells without penicillin. When enzyme activity was very low, quantitative estimations were not made because turbidity at the lower dilutions made readings less reliable. Initial screening experiments were performed using penicillins at final concentrations between 2·5-2·8 mM (I mg/ml of ampicillin, phenoxymethylpenicillin and benzylpenicillin). In quantitative work the final concentrations were-benzylpenicillin 0·56, 2 8, 28 and 93·4 mM; phenoxymethylpenicillin 43 mM; and amoxycillin, ampicillin and epicillin 9·6, 9·9 and 10·8 mM respectively (4 mg/ml). The highest concentrations of all substrates used were based on previous work (Nara et al. 1971; Baker 1983b).

Calculations for a-Aminobenzylpenicillin Substrates Penicillin amidase activity against a-aminobenzylpenicillin derivatives was determined from the reaction a-Aminobenzylpenicillin =' 6-APA + amino side-chain compound (I) In the general case, I /Lmol of substrate would yield x /Lmol of 6-APA. At the dilutions used phenylglycine and other side-chain derivatives did not fluoresce. A control solution was diluted, after incubation, to contain 20 /Lg of a-aminobenzylpenicillin per millilitre (Wa) and the relative fluorescence of this solution (F20 ) was obtained in comparison to the fluorescence given by 10 /Lg 6-APA. The relative fluorescence of the enzyme test solution (FT ) was obtained at the same dilution as the control and represented the contribution of 6-APA (n . x /Lmol) and remaining penicillin [n. (I-x) /Lmol] to the total measured fluorescence. Hence

FT = n. (I-x). FA + n. x. F6 , (2) where FA and F6 were the molar relative fluorescence values of penicillin and 6-APA respectively. Rearrangement of expression (2) and substitution of appropriate quantities and molecular weights leads to

W6 = [215. (20. FT - Wa' F2o )]/(215. 200 - F20 • M) = k. (20. FT - Wa' F20 ), (3) where W6 is the weight of6-APA in micrograms (molecular weight of anion 215) and Mis the molecular weight of the respective penicillin. The value of k was of a similar order for all preparations used.

Results Fluorescence of Reagents at pH 4 At pH 4 concentration quenching effects of the fluorescamine 6-APA complex began above 10 JIg/ml (Fig. I). Ampicillin gave linear, and less intense, fluorescence emission, whilst benzylpenicillin gave a negligible response in the concentration range to 1 mg/m!. After dilution the final concentration of intact benzylpenicillin was in the concentration range 50-300 JIg/ml prior to addition of fluorigenic agent. The intensity of the fluorescence of the reaction of amoxycillin and epicillin with fluorescamine at pH 4 was similar to ampicillin (Table I). The reaction with a higher concentration of glycylglycine (2·23 JImoljO· 5 ml) was also detectable but other amino compounds did not give a reaction at this pH. 6-APA was the only compound tested which gave a more intense reaction at pH 4 than at pH 7 (ratio 1·7: I). 260 W. L. Baker

1·0

~ 0·8 c Q) c..> (/) oQ) 0·6 :l Fig. 1. Relative fluorescence of ampicillin (0), 6-APA (e) and benzylpenicillin (_) with -=Q) 0·4 > fluorescamine at pH 4. 100% relative ~ fluorescence taken from a solution of 6-APA Q) cr: at a concentration of 150 ILg/ml. All readings were taken I h after mixing reagents. ~~4-~~~~~~ o 0·2 0·4 0·6 0·8 1·0 Benzylpenicillin (mg/ml) I o 20 40 60 80 100 Ampicillin or 6-APA (fLg/ml)

Table 1. Substances reacting with f1uorescamine at pH 4 The compounds listed were treated with fluorescamine as described for quantitative estimation of enzyme activity under Methods

Compound Concentration Fluorescence Compound Concentration Fluorescence (I'm 0110 -5 ml) ratio (I'm 0110 -5 mil ratio

6-APA 009 100 Alanine o 56 051 Glycine o 67 0 Tyrosine 028 1-24 Glycine 3 13 063 Ampicillin 005 25-7 Glycylglycine 223 3-70 Amoxycillin 005 226 Leucine 038 030 Epicillin 005 225

Substrate Levels and Incubation Times Penicillin amidase activity was readily measured in whole cells of E. coli ATCC9637 at benzylpenicillin concentrations of 2 -8 mM or greater (Table 2). This substrate level was

Table 2. Activity of E. coli ATCC9637 against benzylpenicillin Incubation mixtures contained per millilitre: 50 I'mol of phosphate buffer, pH 7 -8, the concentration of penicillin shown and either whole cells providing 0 -25 mg soluble protein or broken cells providing 0 -24 mg soluble protein

Enzyme Concentration of 6-APA formed (I'g) per I mg protein in: preparation benzylpenicillin I h 2 5 h 24 h (mM)

Whole 056 0 0 0 cells 2 8 19 210 209 28 490 1130 2950 934 490 1510 3310

Disrupted o 56 80 180 171 cells 28 240 325 342 28 600 1480 3420 93 4 610 1670 3930 considered the minimum required for satisfactory estimation of enzyme activity although it was possible to estimate the amount of 6-APA formed from an 0 -56 mM solution of penicillin by the disrupted cell preparation. This preparation was slightly more active ------Screening Microorganisms for Penicillin Amidase 261

against benzylpenicillin than the whole cell suspension and apart from results obtained with 0·56 mM substrate, the penicillin appeared to be accessible to the intracellular enzyme of the intact organism. No evidence was obtained of substrate inhibition of the enzyme with the concentrations of penicillin used. The rate of formation of 6-APA was initially linear. Over a 4-h period samples, taken hourly, gave respective assay results of o· 6, 1·6, 2·1 and 3·2 mg 6-APA formed per milligram of protein. The rate of formation decreased with time and after 24 h only 5 mg of 6-AP A had been formed per milligram of protein. This effect is probably due to the accumulation of phenylacetic acid which is an established inhibitor of penicillin amidase (Chiang and Bennett 1967; Balasingham et al. 1972).

6

= 5 E '" '" "-OJ .s 4 Fig. 2. Formation of 6-APA with '0 Q) variation in whole cell volumes of E 3 E. coli ATCC9637. Reaction mixture .E was as described for benzylpenicillin in « • Table 4. Samples were taken at 2 '" ~ I I h (0),4 h (e) and 24 h (L'», diluted I co appropriately and treated with fluorescamine.

0·25 0·5 0·75 1·0 1·25 Bacterial protein (mg/ml)

In a solution of 93·4 mM benzylpenicillin the rate of formation of 6-APA was also proportional to cell concentration over an incubation period of 4 h (Fig. 2). Linearity was not maintained over 24 h when the cell protein concentration was above 0·4 mg/m\. This concentration of benzylpenicillin or 43 mM phenoxymethylpenicillin (Nara et al. 1971) was used for screening and quantitative purposes with most organisms. A standard incubation time of 16 h was adopted for convenience and proved adequate and reliable for organisms known to possess penicillin amidases. These conditions allowed estimations to be made directly on a dilution of bacterial incubation mixture and were also satisfactory for detecting low enzyme activity in non-induced bacteria. Use of high concentrations of penicillin substrates also eliminated amino acid interference by adequate dilution prior to making estimations. It also permitted detection of induced penicillin amidases in the possible presence of /3-lactamases.

Detection and Quantitative Estimation of Penicillin Amidase Activity Using 2·8 mM benzylpenicillin it was possible to confirm the presence of a type I penicillin amidase in A. calcoaceticus and N. globerula and a type II enzyme in four other organisms which were grown in medium A (Table 3). The organisms possessing a type II enzyme showed some activity against phenoxymethylpenicillin and three of these organisms showed definite activity against ampicillin. Both strains of E. coli probably have an elevated basal level of enzyme because activity was readily detected when they were grown in the absence of phenylacetic acid as inducing agent (medium B). Detection of enzyme activity in other organisms grown in this medium was only achieved with the higher substrate levels indicated above. N 0"­ N

Table 3. Detection of penicillin amidase activity A and 8 refer to growth media (see Methods); n.d. means that the test was not done. Other symbols: strong fluorescence +++; moderate fluorescence ++; weak fluorescence +; very weak fluorescence ±; no fluorescence -. All readings were made against blanks containing cells but no penicillin. Cell blanks containing I : 100 cetyltrimethylammonium bromide were also used

Organism Induced cells (A), substrate concn constant: Non-induced cells (8), substrate concn varied:

Pen. VK Pen. G Ampicillin Pen. VK Pen. VK Pen. G Pen. G (1 mg/ml, (1 mg/ml, (1 mg/ml, (1 mg/ml, (16.7 mg/ml, (I mg/ml, (33 mg/ml, 2·7 mM) 28 mM) 2·5 mM) 2·7 mM) 43 mM) 2·8 mM) 93·4 mM)

E. coli ATCC9637 ± +++ +++ n.d. +++ n.d. E. coli ATCCI I 105 ± +++ +++ n.d. +++ n.d. B. metaterium ATCCI4945 ± +++ +++ n.d. +++ A. calcoaceticus ATCC21288 +++ +++ A.faecalis ATCCI5246 + +++ ± + ± ++ N. globerula ATCC21022 +++ +++ +

~ r I:tl "';0;- ...,n> Screening Microorganisms for Penicillin Amidase 263

Considerable variation occurred when enzyme activity was measured in the respective organisms (Table 4). The results clearly confirm the low enzyme activity of A. faecalis against phenoxymethylpenicillin. The rates of deacylation by each organism are not strictly comparable because the B. megaterium enzyme is extracellular and there are possible

Table 4. Quantitative estimation of penicillin amidase in bacteria All organisms grown in medium A, incubation time 16 h. Incubation mixtures contained in I ml: 50 I'mol phosphate buffer, pH 7·8, and either 43 I'mol penicillin VK (phenoxymethylpenicillin) or 93·4 I'mol penicillin G (benzylpenicillin) and cells yielding the amount of soluble protein shown

Organism Soluble 6-APA formed (I'g) per milligram protein soluble cell protein: (mg/ml) Penicillin VK Penicillin G (type I) (type II)

E. coli ATCC9637 03 Not estimated 734 E. coli ATCCIII05 0·6 Not estimated 786 B. megaterium ATCCI4945 0·24 Not estimated 167 A. calcoaceticus ATCC21288 o 17 1965 Not estimated A.faecalis ATCC 15246 043 31 255 N globerula ATCC21 022 o 16 2010 Not estimated differences in substrate binding and permeability between Gram-positive and Gram­ negative organisms (Franklin 1974). In addition the kinetic parameters of the enzymes of some organisms are not known. Despite these limitations the analytical procedure was adequate, sensitive and convenient for estimating enzyme activity over a period of 16 h.

Table 5. Penicillin amidase activity against a-aminobenzylpenicillins Cells grown in medium A, incubation time 16 h. Incubation mixture contained in I ml: 50l'mol phosphate buffer, pH 7·8, 4 mg a-aminobenzylpenicillin (9·9, 9·6 and 10·8 mM ampicillin, amoxycillin or epicillin) and cells yielding the amount of soluble protein shown

Organism Soluble 6-APA formed (I'g) per milligram protein protein per 16 h from: (mg/ml) Ampicillin Amoxycillin Epicillin

E. coli ATCC9637 o 7 1803 1856 1247 E. coli ATCCIII05 0·6 2893 3622 1607 B. megaterium ATCCI4945 07 387 Nil 896 A. faecalis ATCC 15246 13 327 Nil Nil a-Aminobenzylpenicillin Substrates Significant differences were observed in the deacylation of a-aminobenzylpenicillins by the penicillin amidases of individual organisms (Table 5). Lack of deacylation of amoxycillin by the enzyme in B. megaterium (Baker 1983b) was confirmed and activity of the A. faecalis enzyme was not detected against amoxycillin or epicillin substrates. The fluorescent properties of some a-aminobenzylpenicillin derivatives altered after incubation for 40 h or longer (probably due to polymerization) but consistent properties were always observed after 16 h incubation.

Discussion Use of the reaction of6-APA with fluorescamine at pH 4 to detect and measure penicillin amidase activity eliminated separation techniques prior to differentiating the product of 264 W. L. Baker

enzyme activity from the substrate. The simple reaction was found to be very convenient for use with whole or even disrupted bacterial cells. Any penicillin which is deacylated to expose the free amino group of the 6-APA nucleus may be used as a substrate for the enzyme. a-Aminobenzylpenicillins gave a significant fluorescence at pH 4 but for quantitative purposes this problem was overcome by a calculation which accounted for the molar fluorescence of substrate and the more intense molar fluorescence of 6-APA. Low levels of penicillin substrates (2·8 mM) were adequate for detection of penicillin amidases when the organisms were grown in the presence of phenylacetic acid. However, it is suggested that the enzyme be induced and exposed to higher levels of substrate (Nara et al. 1971) in all screening experiments. These conditions are recommended because the formation of 6-APA from benzyl- or phenoxymethylpenicillin should be detectable if substrate inhibition of the enzyme occurs (Balasingham et al. 1972) or if iJ-Iactamases are present in the microorganism. The iJ-Iactamases are considered to be more active enzymes than penicillin amidase (Sykes and Matthew 1979) which in turn has been shown to be less active against benzylpenicilloic acid than benzylpenicillin (Cole 1964). The presence of broad-spectrum iJ-Iactamase in organisms, which hydrolyse the iJ-Iactam ring of6-APA, would still mask penicillin amidase activity. The fluorescamine reaction of 6-APA has been used to follow enzyme activity and obtain enzyme parameters (Veronese et al. 1981; Baker 1983a). Lack of serious interference of amino and other compounds at pH 4, as shown in this work, means that it can be used in selection programs to obtain high-yielding and active penicillin-amidase-producing microorganisms for industrial purposes. The sensitivity of the reaction makes it ideal for monitoring enzyme purification procedures. U seof the screening technique to determine the range of organisms possessing penicillin amidases coupled with the quantitative reactions to obtain substrate profiles, kinetic parameters and other properties of individual enzymes will help establish the biochemical relationships which exist between type I, type II and ampicillin amidases (Okachi et al. 1973). It should also help in assessing the role which penicillin amidases have in microbial metabolism and indicate whether an organism may possess more than one type of enzyme or isozymes.

Acknowledgments The author wishes to express appreciation to Mr R. Hems and Squibb Laboratories, Mr V. Davey (C.S.L.), Dr R. Okachi (Kwoya Hakka Kogyo Co.), and the Beecham Research Laboratories for their gifts, and to Mr P. J. Havlicek, of this department, for advice and suggestions in the preparation of the manuscript.

References Baker, W. L. (1983a). Application of the fluorescamine reaction with 6-aminopenicillanic acid to estimation and detection of penicillin acylase activity. Antirnicrob. Agents Chernother. 23, 26-30. Baker, W. L. (1983b). Activity of penicillin acylase producing bacteria against a-aminobenzylpenicillins. Antonie van Leeuwenhoek J. Microbial. Serol. 49, 551-8. Balasingham, K., Warburton, D., Dunnill, P., and Lilly, M. D. (1972). The isolation and kinetics of penicillin amidase from Escherichia coli. Biochirn. Biophys. Acta 276, 250-6. Batchelor, F. R., Chain, E. B., Hardy, T. L., Mansford, K. R. L., and Rolinson, G. N. (1961). 6-Aminopenicillanic acid. III. Isolation and purification. Proc. R. Soc. Land. B 154, 498-508. Bomstein, J., and Evans, W. G. (1965). Automated colorimetric determination of 6-aminopenicillanic acid in fermentation media. Anal. Chern. 37, 576-8. Chiang, C, and Bennett, R. E. (1967). Purification and properties of penicillin amidase from Bacillus rnegateriurn. J. Bacterial. 93, 302-8. Cole, M. (1964). Properties of the pencillin deacylase enzyme of Escherichia coli. Nature (Land.) 203, 519-20. Cole, M., and Sutherland, M. (1966). The role of penicillin acylase in the resistance of Gram-negative bacteria to penicillins. J. Gen. Microbial. 42, 345-56. Screening Microorganisms for Penicillin Amidase 265

Franklin, T. J. (1974). Factors affecting the penetration of antibiotics into bacteria. In 'Industrial Aspects of Biochemistry'. (Ed. B. Spencer.) Vol. II. pp. 549-77. (North-Holland/American Elsevier: Amsterdam, London.) Holt, R. J., and Stewart, G. T. (1964). Production of amidase and iJ-Iactamase by bacteria. J. Gen. Microbiol. 36, 203-\3. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951). Protein measurement with The Folin phenol reagent. J. Bioi. Chern. 193, 265-75. Nara, T., Misawa, M., Okachi, R., and Yamamoto, M. (1971). Enzymatic synthesis of D(-)-a­ aminobenzy1penicillin. Part I. Selection of penicillin acylase-producing bacteria. Agric. Bioi. Chern. 35, 1676-1682. Okachi, R., Kato, F., Miyamura, Y., and Nara, T. (1973). Selection of Pseudomonas melanogenum KY3987 as a new ampicillin-producing bacteria. Agric. Bioi. Chern. 37, \953-57. Sykes, R. B., and Matthew, M. (1979). Detection, assay and immunology of iJ-lactamases. In 'Beta­ Lactamases'. (Eds J. M. T. Hamilton-Miller and J. T. Smith.) pp. 17-49. (Academic Press: London, New York, San Francisco, Toronto, Sydney.) Uri, J., and Sztaricskai, F. (1961). Simple, quick methods for detection of 6-aminopenicillanic acid. Nature (Lond.) 191, 1223. Vandamme, E. J., and Voets, J. P. (1974). Microbial penicillin acylases. Adv. Appl. Microbiol. 17, 311-69. Veronese, F., Franchi, D., Boccu, E., Guerrato, A., and Orsolini, P. (1981). E. coli penicillin acylase: extraction, purification and kinetic characterisation by improved methods. II. Farm. Ed. Sci. 36, 663-70.

Manuscript received 18 October 1983, accepted I June 1984