Inactivation of Aeromonas hydrophila metallo-β-lactamase by cephamycins and moxalactam

Inactivation of Aeromonas hydrophila metallo-b-lactamase bycephamycins and moxalactam

Astrid Zervosen1,*, Maria Hernandez Valladares1,3*, Bart Devreese2, Christelle Prosperi-Meys1,Hans-Werner Adolph3, Paola Sandra Mercuri1, Marc Vanhove1, Gianfranco Amicosante4,Jozef van Beeumen2, Jean-Marie FreÁ re1 and Moreno Galleni1

1Centre for Protein Engineering, University of LieÁge, Belgium; 2Laboratory of Protein Biochemistry and Protein Engineering,

University of Gent, Belgium; 3Fachrichtung 8.8 Biochemie, UniversitaÈt des Saarlandes, SaarbruÈcken, Germany;4Dipartimento di Scienze e Technologie Biomediche, UniversitaÁ del'Aquila, Italy

Incubation of moxalactam and cefoxitin with the Aero-

monas hydrophila metallo-b-lactamase CphA leads to

enzyme-catalyzed hydrolysis of both compounds and to

irreversible inactivation of the enzyme by the reaction

products. As shown by electrospray mass spectrometry, the

inactivation of CphA by cefoxitin and moxalactam is

accompanied by the formation of stable adducts with mass

increases of 445 and 111 Da, respectively. The single thiol

group of the inactivated enzyme is no longer titrable, and

dithiothreitol treatment of the complexes partially restores

the catalytic activity. The mechanism of inactivation by

moxalactam was studied in detail. Hydrolysis of moxa-

lactam is followed by elimination of the 3 0 leaving group

(5-mercapto-1-methyltetrazole), which forms a disulfide

bond with the cysteine residue of CphA located in the

active site. Interestingly, this reaction is catalyzed by

cacodylate.

Keywords: cefoxitin; disulfide bond; inactivation; metallo-

b-lactamase; moxalactam.

The synthesis of potent broad-spectrum and selectiveinhibitors of b-lactamases has always been a contestbetween chemists and bacteria. To date, all the effectiveand clinically used inhibitors, such as clavulanic acid, havebeen designed to act against active-site serine b-lactamases.They are substrate analogs of the enzyme and form aninactive acyl-enzyme [1]. Unfortunately, metallo-b-lacta-mases rapidly hydrolyze most classes of b-lactams, includ-ing the broad-spectrum carbapenems [2], and are notinhibited by classic b-lactamase inhibitors [3]. Theseenzymes are metalloproteins that require a bivalent tran-sition metal ion for their activity, usually Zn21 [3]. Theyhave been found in various clinical pathogens such asBacteroides fragilis, Aeromonas hydrophila, Klebsiellapneumoniae, Stenotrophomonas maltophilia, Chryseobac-terium meningosepticum, Pseudomonas aeruginosa andSerratia marcescens [4±6].

The first synthetic inhibitors of these enzymes were thea-amidotrifluoromethyl alcohols and ketones [7]. Severalmercaptoacetic acid thiol ester derivatives also inhibitmetallo-b-lactamases from S. maltophilia L1, B. fragilisCfiA, A. hydrophila CphA and IMP-1 [8±10]. For theB. cereus enzyme, a mechanism-based formation of

mercaptoacetic acid was suggested, with the subsequentformation of a disulfide bond with the active-site cysteineresidue [9]. Compounds such as thiols [11,12], amino acid-derived hydroxamates [13], various carbapenem derivatives[14], and biphenyltetrazoles [15] have been reported to begood inhibitors.

The crystal structures of the CcrA enzyme in complexwith a biphenyltetrazole [16] and IMP-1 in complex with amercaptocarboxylate [17] have been studied. In both cases,the structures showed that the three critical interactions thatprovide selective inhibition against metallo-b-lactamasesare binding in a hydrophobic pocket, metal ion interactions,and triggering conformational changes in a loop thatincrease in interaction between the enzyme and the inhibitor.

Interestingly, most of these compounds were found tobe poor inhibitors or irreversible inactivators of theA. hydrophila AE036 metallo-b-lactamase.

Although cefoxitin, moxalactam and ceftriaxone aresubstrates for most known metallo-b-lactamases, theyirreversibly inactivate the CphA enzyme [18±21]. In thepresent work, the interactions between moxalactam orcefoxitin and the metallo-b-lactamase of A. hydrophila arestudied in detail.

M A T E R I A L S A N D M E T H O D S

Chemicals

Moxalactam was a gift from Eli Lilly and Co. (Indianapolis,IN, USA). Cefoxitin, cephalothin and ceftriaxone werepurchased from Sigma (St Louis, MO, USA). Imipenemwas from Merck, Sharp and Dohme Research Laboratories(West Point, PA, USA). 3 0-Decarbamoylcefoxitin and3 0-de(carbamoyloxy)cefoxitin were gifts from R. F. Pratt

Eur. J. Biochem. 268, 3840±3850 (2001) q FEBS 2001

Correspondence to M. Galleni, Centre for Protein Engineering (CIP),

Institut de Chimie (B6), University of LieÁge, Sart-Tilman, B-4000

LieÁge, Belgium. Fax: 1 3243 663364, Tel.: 1 3243 663549,

E-mail: [email protected]

Abbreviations: DTNB, 5,5 0-dithiobis-(2-nitrobenzoate); MALDI-TOF,

matrix-assisted laser desorption ionization time of flight; ESMS,

electrospray mass spectrometry.

*Note: these two authors contributed equally to this work.

(Received 7 March 2001, revised 10 May 2001, accepted

17 May 2001)

(Wesleyan University, Middletown, USA) [22]. Centa was agift from J. Fastrez (UCL, Louvain la Neuve, Belgium).The structures of the antibiotics are presented in Fig. 1.Acetonitrile was purchased from Acros Organics (Springfield,NJ, USA). Dithiothreitol, 5,5 0-dithio-bis-(2-nitrobenzoate)(DTNB) and d-cysteine were supplied by Sigma. 5-Mercapto-1-methyltetrazole was from Avocado Research Chemicals(Heysham, Lancs., UK). All buffer materials were reagentgrade.

Production of metallo-b-lactamase

The A. hydrophila AE036 metallo-b-lactamase was pro-duced as previously described [23,24]. The S. maltophiliazinc b-lactamase ULA511 was produced as describedby P. S. Mercuri (personal communication). The TEMb-lactamase was purified as previously described [25].

HPLC analysis

The antibiotics and their hydrolysis products were analysedby RP-HPLC on a Lichrospher 100 column (RP18; 5 mm;250 mm � 4 mm) from Merck (Darmstadt, Germany).Samples (volume 50 mL) were eluted by a gradient methodat room temperature using a flow rate of 1 mL´min21. Thegradient was composed of 25 mm potassium phosphate,pH 6.8 (solution A) and acetonitrile (HPLC-grade, solution

B): gradient, 0±2 min: 2% B, 17 min; 30% B, 19±21 min;80% B, 22±30 min; 2% B. A photodiode array detector wasused for UV detection.

Determination of kinetic parameters

Hydrolysis of imipenem by CphA metallo-b-lactamase wasfollowed by monitoring the variation in A300 of 200 mmimipenem in 50 mm sodium cacodylate, pH 6.5. All theexperiments were performed at 30 8C. Cefoxitin andmoxalactam were also tested as substrates for the mono-Zn21 enzyme (6.3 mm final concentration). Km and kcat

were determined under initial-rate conditions, with the helpof the Hanes' linearization of the Henri±Michaelisequation. Kinetic data were fitted using non-linear regres-sion methods on a PC computer with the programkaleidagraph 3.09. Hydrolysis of these compounds wasmonitored at 260 nm with a Uvikon 860 spectrophotometerconnected to a microcomputer via an RS232 interface.

Inactivation of the A. hydrophila metallo-b-lactamase

Inactivation of CphA with cephamycins, oxacephamycinsand cephem. The A. hydrophila AE036 (3.6 mm) enzymewas incubated in buffer (50 mm Hepes/NaOH, pH 7.2,50 mm sodium cacodylate, pH 6.5, or 50 mm potassiumphosphate, pH 6.5) containing 1 mm antibiotic. Inactivationwas followed by withdrawing samples after differentincubation times. Residual activity was determined asdescribed above. The pseudo-first-order inactivation rateconstant ki was calculated for a fixed concentration ofinhibitor.

Inactivation of CphA by hydrolyzed cephamycins andoxacephamycins. Cefoxitin, its substrate analogs3 0-decarbamoylcefoxitin and 3 0-de(carbamoyloxy)cefoxi-tin, and moxalactam are good substrates of the metallo-b-lactamase from S. maltophilia ULA511 [18,20]. Theantibiotics (2 mm) were hydrolyzed by this enzyme in50 mm Hepes/NaOH, pH 7.2, or in 50 mm sodiumcacodylate, pH 6.5, at 30 8C. After complete hydrolysis,monitored at 260 nm, the product was separated from theenzyme by ultrafiltration on a Millipore-Ultrafree-MCFilter Unit (10000 NMWL; Millipore, Bedford, MA,USA). The filtrate was analyzed by HPLC as describedabove and stored at 4 8C. Hydrolyzed antibiotics at aconcentration of 1 mm were used to inactivate 3.6 mmCphA enzyme at 4 8C. Inactivation was followed bymeasuring the residual activity as described above.

Inactivation of CphA with moxalactam and hydrolyzedmoxalactam. CphA (70 mm) was incubated in 50 mmsodium cacodylate buffer, pH 6.5, in the presence of140 mm moxalactam at 4 8C. Moxalactam (2 mm) washydrolyzed with the S. maltophilia ULA511 enzyme asdescribed above. Hydrolyzed moxalactam at a concentra-tion of 140 mm was used to inactivate 70 mm A. hydrophilaenzyme in 50 mm Hepes/NaOH, pH 7.2, at 4 8C. Thereaction of CphA with hydrolyzed moxalactam wasperformed in Hepes to prevent any reaction betweencacodylate and hydrolyzed moxalactam (Fig. 2). Thereaction was followed by HPLC and by measuring theresidual activity as described above.

Fig. 1. Structures of the b-lactam compounds used in this study. 1,

Moxalactam; 2, centa; 3, ceftriaxone; 4, cephalothin; 5, imipenem; 6,

cefoxitin; 7, 3 0-de(carbamoyloxy)cefoxitin; 8, 3 0-decarbamoylcefoxitin.

q FEBS 2001 Inactivation of the CphA metallo-b-lactamase (Eur. J. Biochem. 268) 3841

Inactivation of CphA with 5-mercapto-1-methyltetrazole.CphA (3.6 mm) was incubated with 1 mm 5-mercapto-1-methyltetrazole in 50 mm sodium cacodylate, pH 6.5, inthe absence or presence of 1 mm moxalactam at 4 8C.Inactivation was followed by determining enzyme activityin samples withdrawn after different incubation times asdescribed above.

MS of the enzymes

Cefoxitin and moxalactam both at a concentration of 1 mmwere used to inactivate the A. hydrophila enzyme (40 mm)as described above. Before MS analysis, excess antibioticsand salts were removed by RP-HPLC. About 0.2 nmol ofthe active and inactive enzyme were injected onto a ZorbaxSB-C3 column (DuPont, Wilmington, DE, USA). TheHPLC system consisted of a 140A solvent-delivery systemand a 1000S diode array detector (Applied Biosystems,Foster City, CA, USA). A linear gradient, 5±80% aceto-nitrile in water containing 0.05% trifluoroacetic acid, wasapplied over 30 min at a flow rate of 0.2 mL´min21. Theprotein fractions were dried using a Speed Vac concen-trator. For the electrospray MS (ESMS) experiment, about100 pmol of the native and the inactive forms weredissolved in 20 mL 0.05% formic acid/50% acetonitrile inwater. The samples were injected into the source of themass spectrometer using a Harvard 11 syringe pump

(Harvard Instruments, South Natick, MA, USA) at a flowrate of 6 mL´min21. The mass spectrometer was a VGBio-Q instrument, upgraded with a Platform source(covering 600±1500 amu. Scans were accumulated during135 s and processed using the masslynx software deliveredwith the instrument). Calibration was performed usinghorse heart myoglobin. Tryptic digestion of the desaltedprotein, protein±cefoxitin and protein±moxalactam com-plexes (200 pmol) was carried out at 37 8C for 2 h in100 mL 50 mm Tris/HCl, pH 7.5, containing 0.1 mg trypsin(Worthington). The capillary was held at 2.7 kV, and thecone voltage was set at 40 V. Fifteen-scan matrix-assistedlaser desorption ionization time of flight (MALDI-TOF)analyses of the digestion mixtures were performed on a VGTofSpec SE mass spectrometer (Micromass, Wythenshawe,UK) using time-lag focusing. A 1-mL volume of thedigested solution was mixed with 1 mL 50 mm a-cyano-hydroxycinnamic acid in 0.1% trifluoroacetic acid/50%acetonitrile in ethanol. These mixtures were applied to themetal target and allowed to dry in the air. The masses of thepeptides obtained after tryptic digestion were measured inthe reflectron mode using 25 kV in the source and 28.5 kVon the reflectron. External calibration was performed usingsubstance P and ACTH clip (Sigma). After the masses ofthe peptides had been determined and the peptides iden-tified by comparing expected and obtained masses, internalcalibration was performed using two peptides in thespectrum.

Determination of the thiol groups

The thiol group content of native and inactivated CphA(12 mm) was determined by addition of 1 mm DTNB underdenaturing conditions (0.2% SDS) in 50 mm phosphatebuffer, pH 7.0, and monitoring the increase in A412. Theenzyme was purified by HPLC before the DTNB reaction.

Reactivation of the inactivated enzymes

A. hydrophila b-lactamase (3.6 mm) was inactivated in50 mm sodium cacodylate buffer, pH 6.5, at 4 8C in thepresence of 1 mm hydrolyzed cefoxitin or moxalactam(sample volume � 150 mL). Residual enzyme activity wasthen determined as described above. After 4 h of incuba-tion, 100 mL of the solution was mixed with 900 mLreactivation buffer (50 mm Hepes/NaOH, pH 7.2, contain-ing 5 mm dithiothreitol). The solution was concentrated byultrafiltration using Centricon-10 (Millipore) up to 400 mLat 8 8C. To avoid nonspecific interactions between theenzyme and the membrane, the membrane was treated for30 min with 1 mL 0.1 mg´mL21 BSA solution beforefiltration. After filtration, the volume was adjusted to 1 mLwith reactivation buffer, and the sample was stored at 4 8Covernight. In order to measure residual enzyme activityafter the reactivation with dithiothreitol, the dithiothreitolconcentration in the enzyme preparation was decreased byseveral ultrafiltration steps. Dithiothreitol (Ki � 1 mm)reversibly inhibits CphA activity (M. HernandezValladares, unpublished results). The solution was concen-trated to 100 mL and, before the next filtration step, 900 mL50 mm sodium cacodylate buffer, pH 6.5, was added. Theprocess was repeated three times and, after the last

Fig. 2. HPLC profile of the hydrolysis products obtained by

incubation of 1 mm moxalactam with ULA511 metallo-b-lactamase

in 50 mm Hepes/NaOH, pH 7.2 (A) and 50 mm sodium cacodylate,

pH 6.5 (B). The reaction was performed at 30 8C. The peaks in (A)

correspond to: 2.33 min, moxalactamhydr.; 3.04 min, Hepes; 4.58 min,

5-mercapto-1-methyltetrazole. Those in (B) correspond to: 2.26 min,

moxalactamhydr.; 4.31 min, 5-mercapto-1-methyltetrazole.

3842 A. Zervosen et al. (Eur. J. Biochem. 268) q FEBS 2001

filtration, the volume of the retentate and the residualenzyme activity were determined as described above.

Reaction of cefoxitin with cysteine

HPLC analysis of cefoxitin hydrolysis. Cefoxitin (1 mm)was hydrolyzed with 0.1 mg´mL21 ULA511 and3.6 mg´mL21 TEM at 30 8C in 50 mm Hepes/NaOH,pH 7.2. The hydrolysis of 1 mm cefoxitin by the ULA511enzyme was also carried out in the presence of 1 mmcysteine. The products were separated from the enzymes byultrafiltration and analyzed by HPLC.

MS of the hydrolysis products of cefoxitin. Cefoxitin (1 mm)was hydrolyzed enzymatically in the presence and absenceof 1 mm d-cysteine as described above. After ultrafiltration,the products were separated by capillary HPLC on aPepmap C18 column (LC-Packings, Amsterdam, the Nether-lands) at a flow rate of 4 mL´min21. A linear gradient wasused as described above in the HPLC analysis section. Inthese experiments, 25 mm ammonium hydrogen carbonate,pH 7.8, was used as buffer A. The main products with anelution time of 7.3 min (cefoxitinhydr.) and 9.0 min (adductof cefoxitinhydr. and cysteine) were directly analyzed on aQ-TOF mass spectrometer (Micromass, Manchester, UK).

NMR analysis of cefoxitin hydrolysis. Cefoxitin (10 mm)was hydrolyzed with 0.07 mg´mL21 ULA511 in 50 mmpotassium phosphate, pH 6.5, in 90% D2O in the presenceand absence of 10 mm d-cysteine at room temperature.NMR analyses were carried out with a Bruker DRX 400Avance spectrometer. Proton-NMR spectra with waterpresaturation were obtained with a spectral width of4 kHz for 16 000 frequency and time domain data points.

R E S U LT S

Kinetic parameters

Cefoxitin and moxalactam were poor substrates of CphA.The steady-state kinetic parameters, measured at pH 6.5,were: cefoxitin, Km � 615 mm, kcat. � 0.02 s21, kcat /Km �33 m21´s21; moxalactam, kcat /Km � 5.6 m21´s21. At highconcentrations of the different compounds (1 mm at

pH 6.5), time-dependent inactivation was observed(Table 1). The same was observed when CphA wasincubated with enzymatically hydrolyzed cefoxitin andmoxalactam (Table 1). In both cases, the pseudo-first-orderinactivation constant ki was higher when the enzyme wasincubated with the hydrolyzed antibiotic. The inactivationof CphA by hydrolyzed cefoxitin was significantly faster ifa fresh solution was used. Prolonged storage of the solution(24 h at 4 8C) decreased the inactivating potential(ki � 5 � 1023 min21 vs. ki � 3.5 � 1022 min21 with afresh solution). HPLC analysis of the fresh and stored(24 h) hydrolyzed cefoxitin solution showed degradation ofthe molecules (data not shown).

In the presence of antibiotics with good 3 0 leavinggroups, high rates of inactivation (1.3 � 1022 min21 forcefoxitin) were found. However, even after prolongedincubation of CphA with 3 0-decarbamoylcefoxitin and3 0-de(carbamoyloxy)cefoxitin, the enzyme remained fullyactive. The 3 0-OH group of 3 0-decarbamoylcefoxitin is nota good leaving group. For the TEM-b-lactamase it has beenshown that this group will not be eliminated at the activesite of the enzyme [22]. Inactivation could be detected inthe presence of the hydrolyzed form of 3 0-de(carbamoyl)-cefoxitin (ki � 1.1 � 1022 min21), whereas no inactivationwas observed with 3 0-de(carbamoyloxy)cefoxitin. Theseresults also emphasized that hydrolysis of cephamycins by

Table 1. Rates of inactivation of b-lactamase from A. hydrophila

with 1 mmm inactivator. Buffer A, 50 mm sodium cacodylate, pH 6.5;

buffer B, 50 mm Hepes/NaOH, pH 7.2. SD values were ^ 20%. ND,

Not determined.

ki (min21)

Inactivator Buffer A Buffer B

None , 3 �� 1025 , 3 �� 1025

Cefoxitin 1.3 �� 1022 2.0 �� 1023

Hydrolyzed cefoxitina 3.5 �� 1022 ND

Moxalactam 5.7 �� 1023 6.5 �� 1024

Hydrolyzed moxalactama 1.7 �� 1022 ND

a Prior incubation with CphA; cefoxitin and moxalactam were

completely hydrolyzed by the ULA511 metallo-b-lactamase.

Fig. 3. Inactivation of 70 mmm b-lactamase from A. hydrophila with

140 mmm moxalactam in 50 mmm sodium cacodylate, pH 6.5 at 4 8C.

The HPLC profiles were recorded at the beginning (A) (residual

activity � 100%) and after 44 h (B) of reaction (residual activity �9%). The peaks in (A) correspond to: 4.36 min, 5-mercapto-1-

methyltetrazole; 11.55 and12.05 min, moxalactam diastereoisomeres

A and B. Those in (B) correspond to: 4.36 min, 5-mercapto-1-

methyltetrazole; 11.6 and 12.08 min, moxalactam diastereoisomeres A

and B.


CphA is dependent on the presence of a good 3 0 leavinggroup and is essential for the inactivation event. A similarobservation was made for cephem compounds containing agood 3 0 leaving group, such as centa, cephalothin andceftriaxone. Prolonged incubation of CphA with thesecompounds yielded a residual activity lower than 30% ofthat of the untreated enzyme.

Interestingly, the inactivation kinetics for the mono-zincand di-zinc forms of CphA were different for moxalactam.The mono-Zn21 enzyme [24] was fully inactivated by1 mm moxalactam after 14 h (ki � 5.7 � 1023 min21,t1/2 � 121 min) whereas, in the presence of 200 mmZn21, inactivation of the di-Zn21 enzyme with 1 mmmoxalactam was significantly slower (ki � 3.0 � 1023 min21,t1/2 � 231 min).

The rates of inactivation of CphA by cephamycins andoxacephamycin were significantly lower in 50 mm Hepes/NaOH, pH 7.2, than in 50 mm sodium cacodylate, pH 6.5(Table 1). To investigate the effect of buffers and pH on theinactivation reaction, 3.6 mm CphA was inactivated with1 mm moxalactam in different buffers at 4 8C. The enzymewas stable in all buffers for about 24 h (data not shown).Residual activity was determined after 18 h. Inactivationwas significantly faster in 50 mm sodium cacodylate,pH 6.5 (residual activity 0.5%) than in 50 mm Hepes/NaOH, pH 7.2 (residual activity 27.2%) and 50 mmpotassium phosphate, pH 6.5 (residual activity 17.5%).The increased rate of inactivation in cacodylate at pH 6.5,compared with phosphate buffer at the same pH, demon-strates the influence of cacodylate on the inactivation reaction.

The HPLC profiles of the hydrolysis products of moxa-lactam with the ULA511 metallo-b-lactamase in Hepes andin cacodylate were different (Fig. 2), indicating thatcacodylate reacts with the reaction products.

HPLC analysis of the reaction between CphA andmoxalactam

The two diastereoisomers of moxalactam were readilyseparated by HPLC (Fig. 3). They were characterized byretention times of 11.55 and 12.05 min. At pH 7.2, nospontaneous hydrolysis of moxalactam was observed, evenafter prolonged incubation (44 h). In the presence of CphA,stereospecific hydrolysis of moxalactam was observed.Analysis of the HPLC profile indicates that the diastereo-isomer with a retention time of 12.05 min is hydrolyzedmore rapidly than that with a retention time of 11.55 min.In addition, a peak at 4.36 min corresponding to thedisulfide of 5-mercapto-1-methyltetrazole appeared. Theproduct is probably from oxidation of the 5-mercapto-1-methyltetrazole produced by the hydrolysis of moxalactam.The HPLC profile of a solution of moxalactam hydrolyzedby the ULA511 enzyme shows that the peak at 2.92 mincorresponds to the retention time of Hepes. After incuba-tion of the solution with CphA, a peak at 2.18 min wassignificantly reduced, indicating that the species inactivatesthe metallo-b-lactamase.

In the presence of synthetic 5-mercapto-1-methyltetra-zole (1 mm), slow inactivation of the enzyme wasobserved. A residual activity of 83% was measuredafter a 24-h incubation at 4 8C. The inactivation ofCphA by moxalactam was not affected when 1 mmsynthetic 5-mercapto-1-methyltetrazole was added to theinactivation mixture (ki � 5.1 � 1023 min21). The lowinactivating potential of the molecule can be explainedby its rapid oxidation in aqueous solution. No free thiolcould be detected by the DTNB assay, indicatingformation of a disulfide bond between two moleculesof 5-mercapto-1-methyltetrazole.

Fig. 4. ESMS of the native A. hydrophila metallo-b-lactamase (A), the fully inactivated enzyme±cefoxitin adduct (B), and the fully

inactivated enzyme±moxalactam adduct (C).


ESMS

The maximum entropy-converted spectra of the protein andprotein±antibiotic complexes are shown in Fig. 4. Thespectrum of the native mono-Zn21 enzyme showed a singlepeak at 25 189 ^ 1 Da which was in perfect agreementwith the mass determined from the nucleotide sequence(25 189 Da). The spectrum of the inactive protein±cefoxitin complex also showed a single peak, at25 634 ^ 2 Da, which corresponds to the mass of a simpleadduct in which the b-lactam ring of cefoxitin has beenhydrolyzed (theoretical value 25 631 Da). The protein±cefoxitin complex was not stable on storage at 220 8C.After 2 weeks, the ESMS spectrum showed a new peak at25 573 ^ 2 Da, probably the result of the departure of theleaving group at the 3 0-position.

The spectrum of the inactive protein±moxalactam com-plex exhibited two peaks, at 25 300 ^ 2 and 25 319 ^ 2 Da.The former corresponds to the sum of the masses of theprotein and the 3 0 leaving group adduct. The latter may beassigned to a hydrated form of the protein±leaving group

adduct. The mass spectrum of moxalactam alone showed asimilar pattern. The protein±moxalactam complex wasstable on storage at 220 8C.

In the presence of Hepes, inactivation of the enzyme wasslower than in cacodylate. The moxalactam±CphA andcefoxitin±CphA complexes inactivated in Hepes were alsoinvestigated by ESMS. The same masses were observed asfor the cacodylate buffered protein, although additionalpeaks with higher masses were also present. The resultsindicate that the same inactivation products are mainlyformed and that the slow rate of inactivation is probablybecause cacodylate catalyses the inactivation reaction.

Estimation of free thiol groups in the native enzyme andin the inactive complexes

The reaction of the native protein with Ellman's reagent(DTNB) in the presence of SDS (0.2% final concentration)revealed the presence of a single thiol group, a DA412 of0.152 being observed with 12 mm enzyme. This result is inagreement with the primary structure of the A. hydrophila

Fig. 5. MALDI-TOF analyses of a trypsin digest of the A. hydrophila enzyme before (A) and after (B) treatment with 1 mmm moxalactam. The

cysteine-containing peptide (theoretical mass 3347 Da) was found in the free enzyme (3349 Da). In the adduct, the peptide has undergone a 113-Da

shift (3462 Da), indicating the addition of the 3 0 leaving group of moxalactam.


enzyme. However, the thiol group content of the inactiveand chromatographically pure adducts formed with cefoxi-tin and moxalactam, under denaturing conditions, was lessthan 0.05 per enzyme molecule. Thus, the cysteine residueseemed to be directly involved in the formation of theprotein±antibiotic adducts.

CphA was partially reactivated by incubating thedifferent inactivated complexes in the presence of 4.5 mmdithiothreitol. After overnight incubation at 4 8C, theactivities recovered with CphA±cefoxitin and CphA±moxalactam complexes were 28% and 37% of that of thenative enzyme, respectively. With the native enzyme, 81%of activity was recovered under the same conditions.

Direct identification of the residues

The ESMS analysis and estimation of free thiol groups ofthe protein±cefoxitin and protein±moxalactam adductssuggested the formation of covalent bonds with the solecysteine residue at the active site. Digestions with trypsinand CNBr were carried out to confirm this.

The Ala149±Lys169 peptide (3349 Da), which containsthe sole cysteine residue (at position 166) of CphA, wasobserved in MALDI-TOF analyses after a 2-h incubation inthe presence of trypsin at 37 8C (Fig. 5). The correspondingpeptide of the enzyme after the reaction with moxalactamincreased in mass by 112.5 Da (3461.5 Da), confirmingthat the 116-Da fragment from the moxalactam moleculewas indeed bound to the cysteine.

However, after reaction with cefoxitin, the same peptidecould not be detected in the digestion mixture, probablybecause of increased hydrophobicity. In fact, the peptidewas only recovered from the digestion mixture when theplastic container was extensively washed with an acidicsolution (formic acid). The addition of 0.1% n-octyl-b-d-glucopyranoside to the reaction mixture did not improve therecovery of the peptide.

Reaction of cefoxitin with D-cysteine

To learn more about the mechanism of inactivation ofcefoxitin, the reaction between d-cysteine and the productsof the enzymatic hydrolysis of cefoxitin was investigated byHPLC, HPLC/MS and NMR.

The hydrolysis of cefoxitin by TEM was followed bymonitoring the elimination of the 3 0 leaving group [22]. Inthe HPLC profiles of the product of hydrolysis of cefoxitinwith ULA511 and with TEM (data not shown), a main peakwith an elution time of 10 min was detected. The massspectrum of the products of cefoxitin hydrolysis withULA511 shows the presence of a molecule peak at m/z385.2 (Fig. 6B) corresponding to cefoxitin after the hydro-lysis and elimination reactions. Formation of the b-lactam-ring-opened 7a-methoxy-exo-methylenethiazine could beobserved by NMR. These data are in agreement with theNMR spectra of the cefoxitin hydrolyzed by TEM aspreviously described [22]. The typical splitting of the signald� 3.98, corresponding to the thiazine group, was observed

Fig. 6. Mass spectra of the main products of the hydrolysis of 1 mmm cefoxitin in the presence (A) and absence (B) of 1 mmm cysteine with

ULA511 metallo-b-lactamase in 50 mmm Hepes/NaOH, pH 7.2, at 30 8C.


indicating hydrolysis of the molecule. The hydrolysis wasfollowed by monitoring the elimination of the 3 0 leavinggroup, which is characterized by the appearance of thesignals d� 5.65 and d� 5.69 of the 3-exomethylene groupand the increase in the signal d� 4.66 of the 3-CH2 groupof cefoxitin. The latter signal could not be observed in thepresence of the enzyme because of the large HDO peak.

The reaction of the hydrolyzed cefoxitin with d-cysteinecould be readily detected by HPLC. The main product waseluted after 12.3 min. The mass spectrum of this productcontained four main peaks at m/z 506.3, m/z 430.2, m/z385.2 and m/z 309.1 Da (Fig. 6A). In addition to theproducts of mass 385.2 and 309.1 Da, which have alreadybeen detected in the spectrum of hydrolyzed cefoxitin, twonew peaks were observed. The mass difference of 121.1 Dabetween the peaks m/z 506.3 and m/z 385.2 and m/z 430.2and m/z 309.1 indicated that the reaction of the product ofcefoxitin hydrolysis with cysteine (121.1 Da) had takenplace. In a control experiment, cefoxitin was infused atseveral source energies. A linear relationship between theloss of 61 Da (3 0 leaving group) and the cone voltagewas observed, indicating that cefoxitin and probably theproducts of the hydrolysis are MS-labile.

Hydrolysis of 10 mm cefoxitin with ULA511 in thepresence of 10 mm d-cysteine was investigated by NMRover about 1 h. In a control experiment, 10 mm cefoxitinwas hydrolyzed. After complete hydrolysis and eliminationof the 3 0 leaving group, 10 mm d-cysteine was added. Inboth experiments, the same NMR data were obtained,indicating that the same reaction products were formed:400 MHz 1H-NMR(D2O) d 3.22 (m), 3.53 (s, 3, OCH3),3.74 (s, ThCH2, degradation product), 3.96 (AB, q, 2,J � 17 Hz, ThCH2), 7.08 (m, 2, Th 3 05 0), 7.40 (m, 1, Th 4 0).

In the presence of d-cysteine, no signals at d 5.65 and5.69 were observed. After addition of d-cysteine in thecontrol experiment, these signals immediately disappeared,indicating the addition of d-cysteine to the 3-exomethylenegroup of hydrolyzed cefoxitin.

D I S C U S S I O N

In this report, we show that there are two types ofinteraction between moxalactam or cefoxitin and theA. hydrophila metallo-b-lactamase CphA. First, the enzymecatalyzes the hydrolysis of both compounds. Secondly, theCphA b-lactamase is irreversibly inactivated by cefoxitin andmoxalactam. Both mechanisms of inactivation are discussed.

Cefoxitin and moxalactam are both poor substrates ofA. hydrophila metallo-b-lactamase [kcat /Km (cefoxitin) �33 m´s21, kcat /Km (moxalactam) � 5.6 m´s21]. Rates ofinactivation by the hydrolyzed antibiotics are significantlyhigher, indicating that hydrolysis of the b-lactams is therate-limiting step. The rates of inactivation of the hydro-lyzed antibiotics also indicate higher reactivity of hydro-lyzed cefoxitin than hydrolyzed moxalactam. In both cases,the inactivation rate depends on the affinity of the enzymefor the hydrolyzed antibiotic, the rate of formation of thereactive inactivator, and formation of the covalent bondbetween the inactivator and the enzyme.

The experiments performed with cefoxitin and itsanalogs indicate that the presence of a good 3 0 leavinggroup is also essential for the inactivating potential ofcephamycins.

The rates of inactivation by antibiotics such as centa (2),moxalactam (1) and cefoxitin (6) were increased incacodylate buffer. For hydrolysis of imipenem, pH in therange 6.5±7.2 is not significant [kcat /Km (pH 6.5) �7.6 � 106 m´s21, kcat /Km (pH 7.2) � 6.9 � 106 m´s21].The observation that, at pH 6.5, the inactivation inpotassium phosphate buffer was significantly slower thanin cacodylate disproves the hypothesis that pH affects theinactivation rate and indicates the active participation ofcacodylate in the inactivation mechanism.

Under acidic conditions, cacodylate can act as anoxidizing agent. The reaction of cacodylate with organicthiols is accompanied by the disappearance of the acidfunction of cacodylate and the SH groups [26]. The reactionbetween an aqueous solution of dimethlyarsinic acid andcysteine [(CH3)2As(O)OH 1 3Cys ! (CH3)2As-Cys 1cystine 1 H2O] has previously been described [27].Furthermore, interactions between cacodylate and otherenzymes have previously been observed [28]. For example,the inhibition by 2-mercaptoethanol of phenylalanyl-tRNAsynthetase of Neurospora crassa with valine tRNA ofEscherichia coli was much more severe in the presence ofcacodylate buffer than Tris [26,29]. This is probably due tothe formation of a reactive product of cacodylate andmercaptoethanol.

Figure 2 shows that a reaction between moxalactam andcacodylate is possible. Presumably the product of thisreaction is a better inactivator than hydrolyzed moxalactamalone.

The ESMS analysis of the complexes showed massincreases of 445 and 111 Da after inactivation by cefoxitinand moxalactam, respectively, indicating the formation ofcovalent adducts with hydrolyzed cefoxitin and with the 3 0leaving group 5-mercapto-1-methyltetrazole obtained frommoxalactam. Estimation of the free thiol group content ofthe denatured inactive complexes suggests a reaction withthe sole cysteine present in active site of the enzyme [24].The importance of this residue in the catalytic activity ofCphA has been demonstrated by modifying the residuewith iodoacetonitrile and iodoacetate. The activity of themodified enzyme against imipenem is greatly reduced(M. Hernandez Valladares, unpublished results). TheMALDI-TOF experiments confirmed that, after trypticdigestion, the 3 0 leaving group of moxalactam is indeedbound to the peptide containing the cysteine residue. Incontrast, the cysteine-containing peptide from the cefoxitinadduct could not be found, probably because of higherhydrophobicity.

With moxalactam, the only possible reaction is that ofthe 3 0 leaving group with the Cys thiol group, resultingin the formation of a disulfide bond (Fig. 7). The slowinactivation of the enzyme in the presence of the synthetic5-mercapto-1-methyltetrazole (9) and the unchanged rate ofinactivation by hydrolyzed moxalactam in the presence ofthe synthetic thiol indicates that the elimination and theformation of the disulfide bond are probably consecutivereactions in the active site of the enzyme. Nevertheless, theinactivation reaction cannot be described by an equimolarreaction between moxalactam and the enzyme.

With cefoxitin, three possible adducts could initially beproposed. The first would involve formation of a similardisulfide bond with the hydrolyzed cefoxitin molecule(Fig. 8, reaction II), a completely unexpected reaction,


which is in agreement with the mass of 445 Da observedduring on MS. The second and the third adducts, inagreement with a more classical explanation, would involvethe replacement of the 3 0 leaving group with cysteine(Fig. 8, reaction III) or the acylation of the cysteine thiolgroup by the b-lactam carbonyl, as observed with the serinehydroxy group of active-site serine penicillin-recognizingenzymes (Fig. 8, reaction I). These reactions would resultin a mass increase corresponding to that of hydrolyzedcefoxitin without the 3 0 leaving group or, in the case of thethioester, to the mass of intact cefoxitin, and not thehydrolyzed b-lactam molecule as seen in the mass spectraof the adduct. But, as found with moxalactam, the presenceof a hydrated adduct remains possible.

An indication that a disulfide bond is formed in theCphA±cefoxitin complex comes from the partial reactiva-tion of the inactive enzyme with dithiothreitol. As theformation of a disulfide bond with the hydrolyzed cefoxitinmolecule was quite unexpected, the reaction betweenhydrolyzed cefoxitin and d-cysteine was investigated. TheHPLC and HPLC/MS results indicate a reaction betweenhydrolyzed cefoxitin and d-cysteine. The mass of thisproduct (506.2 Da; Fig. 6A) excludes the formation of athioester with a theoretical mass of 488.5 Da. MS analysisof cefoxitin with different cone voltages indicated that the3 0 leaving group is MS-labile. Thus, it is not possible todistinguish between the other two proposed reactions ofhydrolyzed cefoxitin with cysteine (Fig. 8 reactions II andIII). The disappearance of the NMR signals d 5.65 and 5.69after the addition of d-cysteine to a solution of hydrolyzedcefoxitin indicates the occurrence of a reaction between

Fig. 7. Postulated mechanism of the inactivation of the A. hydro-

phila metallo-b-lactamase by moxalactam (1). Compound 9 is

5-mercapto-1-methyltetrazole.

Fig. 8. Postulated mechanisms of the inactivation of the A. hydrophila metallo-b-lactamase by cefoxitin.


d-cysteine and the 3-exomethylene group of hydrolyzedcefoxitin. Thus, in solution, the hydrolysis of cefoxitin isfollowed by the elimination of the 3 0 leaving group and there-addition of d-cysteine.

In conclusion, moxalactam and cefoxitin are veryspecific inactivators of the A. hydrophila enzyme. Enzym-atic hydrolysis of the b-lactams is a prerequisite for thesubsequent inactivation reaction. For moxalactam, wehave shown that the inactivation mechanism involves theformation of a mixed disulfide between the sole cysteine inthe active site and the 3 0 leaving group of the hydrolyzedmoxalactam.

In the case of cefoxitin, formation of a thioester can beexcluded. Two possible mechanisms remain. First, asshown by the reaction of hydrolyzed cefoxitin and cysteine,the free thiol of CphA reacts with the exo-methylene group,which appears after elimination of the 3 0 leaving group.Thesecond alternative, based on MS data and dithiothreitoltreatment of the inactivated complex, is formation of adisulfide bond between the cysteine residue of the enzymeand the dihydrothiazine sulfur. Although, this reaction isunexpected, we could not exclude it.

Finally, this work suggests that synthesis of cepha-mycins, cephems or oxacephamycins containing a good 3 0leaving group could be a new approach to the design ofinactivators of the metallo-b-lactamases.

A C K N O W L E D G E M E N T S

This work was supported by a grant from the European Union

(ERB3512-ICI15-CT98-0914) and as part of the training and mobility

of researchers program and by the Belgian Program PoÃles d'Attraction

Interuniveritaire initiated by the Belgian state, the prime minister's

office, and the Services FeÂdeÂraux des affaires Economiques, Tech-

niques and Culturelles (PAI P4/03). B. D. is a postdoctoral fellow of

the FWO-Vlaanderen. J. V. B. is indebted to the Fund for Scientific

Research-Flanders for the project grant G.422.98.

R E F E R E N C E S

1. Pratt, R.F. (1992) b-Lactamase inhibition. In The Chemistry of

b-Lactams (Page, M.I., ed.), pp 229±272. Blackie Academic &

Professional, London, UK.

2. Felici, A., Amicosante, G., Oratore, A., Strom, R., Ledent, P.,

Joris, B., Fanuel, L. & FreÁre, J.-M. (1993) An overview of the

kinetic parameters of class B b-lactamases. Biochem. J. 291,

151±155.

3. Matagne, A., Dubus, A., Galleni, M. & FreÁre, J.M. (1999) The b-

lactamase cycle: a tale of selective pressure and bacterial ingenity.

Nat. Prod. Rep. 16, 1±19.

4. Bush, K. & Jacoby, G. (1995) A functional classification scheme

for b-lactamases and its correlation with molecular structure.

Antimicrob. Agents Chemother. 39, 1211±1233.

5. Rasmussen, B.A. & Bush, K. (1997) Carbapenem-hydrolyzing

b-lactamases. Antimicrob. Agents Chemother. 41, 223±232.

6. Rossolini, G.M., Franceschini, N., Riccio, M.L., Mercuri, P.S.,

Perilli, M.G., Galleni, M., FreÁre, J.M. & Amicosante, G. (1998)

Characterization and sequence of the Chryseobacterium (Flavo-

bacterium) meningosepticum carbapenemase: a new molecular

class B b-lactamase showing a broad substrate profile. Biochem. J.

332, 145±152.

7. Walter, M.W., Felici, A., Galleni, M., Paul Soto, R., Adlington,

R.M., Baldwin, J.E., FreÁre, J.M., Gololobov, M. & Schofield, C.J.

(1996) Trifluoromethyl alcohol and ketone inhibitors of metallo-b-

lactamases. Bioorg. Med. Chem. Lett. 6, 2455±2458.

8. Hammond, G.G., Huber, J.L., Greenlee, M.L., Laub, J.B., Young,

K., Silver, L.L., Balkovec, J.M., Pryor, K.D., Wu, J.K., Leiting, B.,

Pompliano, D.L. & Toney, J.H. (1999) Inhibition of IMP-1

metallo-b-lactamase and sensitization of IMP-1-producing

bacteria by thioester derivates. FEMS Microbiol. Lett. 459,

289±296.

9. Payne, D.L., Bateson, J.H., Gasson, B.C., Proctor, D., Khushi, T.,

Farmer, T.H., Tolson, D.A., Bell, D., Skett, P.W., Marshall, A.C.,

Reid, R., Ghosez, L., Combret, Y. & Marchand-Brynaert, J. (1997)

Inhibition of metallo-b-lactamases by a series of mercaptoacetic

acid thiol ester derivatives. Antimicrob. Agents Chemother. 41,

135±140..

10. Payne, D.L., Bateson, J.H., Gasson, B.C., Khushi, T., Proctor, D.,

Pearson, S.C. & Reid, R. (1997) Inhibition of metallo-b-

lactamases by a series of thiol ester derivatives of mercapto-

phenylacetic acid. FEMS Microb. Lett. 157, 171±175.

11. Bounaga, S., Laws, A.P., Galleni, M. & Page, M.I. (1998) The

mechanism of catalysis and the inhibition of the Bacillus cereus

zinc-dependent b-lactamase. Biochem. J. 331, 703±711.

12. Goto, M., Takahashi, T., Yamashita, F., Koreeda, A., Mori, H.,

Ohta, M. & Arakawa, Y. (1997) Inhibition of the metallo-b-

lactamase produced from Serratia marcescens by thiol com-

pounds. Biol. Pharm. Bull. 20, 1136±1140.

13. Walter, M.W., Hernandez Valladares, M., Adlington, R.M.,

Amicosante, G., Baldwin, J.E., FreÁre, J.M., Galleni, M., Rossolini,

G.M. & Schofield, C.J. (1999) Hydroxamate inhibitors of

Aeromonas hydrophila AE036 metallo-b-lactamase. Bioorganic

Chem. 27, 35±40.

14. Nagano, R., Adachi, Y., Imamura, H., Yamada, K., Hashizume, T.

& Morishima, H. (1999) Carbapenem derivates as potential

inhibitors of various b-lactamases, including class B metallo-b-

lactamases. Antimicrob. Agents Chemother. 43, 2497±2503.

15. Toney, J.H., Cleary, K.A., Hammond, G.G., Yuan, X., May, W.J.,

Hutchins, S.M., Ashton, W.T. & Vanderwall, D.E. (1999)

Structure-activity relationships of biphenyl tetrazoles as

metallo-b-lactamase inhibitors. Bioorg. Med. Chem. Lett. 9,

2741±2746.

16. Toney, J.H., Fitzgerald, P.M., Grover-Sharma, N., Olson, S.H.,

May, W.J., Sundelof, J.G., Vanderwall, D.E., Cleary, K.A., Grant,

S.K., Wu, J.K., Kozarich, J.W., Pompliano, D.L. & Hammond,

G.G. (1998) Antibiotic sensitization using biphenyl tetrazoles as

potent inhibitors of Bacteroides fragilis metallo-beta-lactamase.

Chem Biol. 5, 185±196.

17. Concha, N.O., Janson, C.A., Rowling, P., Pearson, S., Cheever,

C.A., Clarke, B.P., Lewis, C., Galleni, M., FreÁre, J.M., Payne, D.J.,

Bateson, J.H. & Abdel-Meguid, S.S. (2000) Crystal structure of

the IMP-1 metallo beta-lactamase from Pseudomonas aeruginosa

and its complex with a mercaptocarboxylate inhibitor: binding

determinants of a potent, broad-spectrum inhibitor. Biochemistry

39, 4288±4298.

18. Felici, A. & Amicosante, G. (1995) Kinetic analysis of extension

of substrate specificity with Xanthomonas maltophilia, Aeromonas

hydrophila, and Bacillus cereus metallo-b-lactamases. Antimicrob.

Agents Chemother. 39, 192±199.

19. Felici, A., Perilli, M., Segatore, B., Franceschini, N., Setacci, D.,

Oratore, A., Stefani, S., Galleni, M. & Amicosante, G. (1995)

Interactions of biapenem with active-site serine and metallo-b-

lactamases. Antimicrob. Agents Chemother. 39, 1300±1305.

20. Felici, A., Perilli, M., Franceschini, N., Rossolini, G.M., Galleni,

M., FreÁre, J.M., Oratore, A. & Amicosante, G. (1997) Sensitivity

of Aeromonas hydrophila carbapenemase to D3-cephems: Com-

parative study with other metallo-b-lactamases. Antimicrob.

Agents Chemother. 41, 866±868.

21. Quiroga, M.I., Franceschini, N., Rossolini, G.M., Gutkind, G.,

Bonfiglio, G., Franchino, L. & Amicosante, G. (2000) Interaction

of cefotetan and the metallo-b-lactamases produced in Aeromonas

spp. and in vitro activity. Chemotherapy 46, 177±185.


22. Faraci, W.S. & Pratt, R.F. (1986) Mechanism of inhibition of

RTEM-2 b-lactamase by cephamycins: relative importance of the

7a-methoxy group and the 3 0 leaving group. Biochemistry 25,

2934±2941.

23. Hernandez Valladares, M., Galleni, M., FreÁre, J.-M., Felici, A.,

Perilli, M., Franceschini, N., Rossolini, G.M., Oratore, A. &

Amicosante, G. (1996) Overproduction and purification of the

Aeromonas hydrophila CphA metallo-b-lactamase expressed in

Escherichia coli. Microb. Drug Resist. 2, 253±256.

24. Hernandes Valladares, M., Felici, A., Weber, G., Adolph, H.W.,

Zeppezauer, M., Rossolini, G.M., Amicosante, G., FreÁre, J.M. &

Galleni, M. (1997) Zn(II) dependence of the Aeromonas

hydrophila AE036 metallo-b-lactamase activity and stability.

Biochemistry 36, 11534±11541.

25. Dubus, A., Wilkin, J.M., Raquet, X., Normark, S. & FreÁre, J.M.

(1994) Catalytic mechanism of active-site serine b-lactamases:

role of the conserved hydroxy group of the Lys-Thr(Ser)-Gly triad.

Biochem. J. 301, 485±494.

26. Jacobson, K.B., Murphy, J.B. & Sarma, B.D. (1972) Reaction of

cacodylic acid with organic thiols. FEBS Lett. 22, 80±82.

27. Banks, C.H., Daniel, J.R. & Zingaro, R.A. (1979) Biomolecules

bearing the S- or SeAsMe2 function: amino acid and steroid

derivatives. J. Med. Chem. 22, 572±575.

28. Tsao, D.H.H. & Maki, A.H. (1991) Optically detected magnetic

resonance study of the interaction of an arsenic (III) derivative of

cacodylic acid with EcoRI methyl transferase. Biochemistry 30,

4565±4572.

29. Ritter, P.O., Kull, F.J. & Jacobson, K.B. (1970) Aminoacylation of

Escherichia coli valine transfer ribonucleic acid by Neurospora

crassa phenylalanyl transfer ribonucleic acid synthetase in Tris

(hydroxymethyl) aminomethane hydrochloric acid and potassium

cacodylate buffers. J. Biol. Chem. 245, 2114±2120.


Inactivation of Aeromonas hydrophila metallo-β-lactamase by cephamycins and moxalactam

Documents