Báo cáo Y học: Agmatine oxidation by copper amine oxidase Biosynthesis and biochemical characterization of N-amidino-2-hydroxypyrrolidine

Eur. J. Biochem. 269, 884–892 (2002) (cid:211) FEBS 2002

Agmatine oxidation by copper amine oxidase Biosynthesis and biochemical characterization of N-amidino-2-hydroxypyrrolidine

Paolo Ascenzi1,*, Mauro Fasano2,*, Maria Marino1, Giorgio Venturini1 and Rodolfo Federico1 1Department of Biology, University (cid:212)Roma Tre(cid:213), Rome, Italy; 2Department of Structural and Functional Biology, University of Insubria, Varese, Italy

M)

M and (2.2 (cid:139) 0.4) · 10)8

significantly higher than that observed for agmatine and clonidine binding. Furthermore, N-amidino-2-hydroxy- pyrrolidine and agmatine are more e(cid:129)cient than clonidine in displacing [3H]clonidine ((cid:136) 1.0 · 10)8 from specific binding sites in heart rat membranes, values of IC50 being (1.3 (cid:139) 0.4) · 10)9 M, respec- tively (at pH 7.4 and 37.0 (cid:176)C).

Keywords: copper amine oxidase; agmatine; N-amidino-2- hydroxypyrrolidine; enzyme inhibition; type 1 imidazoline receptor binding.

The product of agmatine oxidation catalyzed by Pisum sativum L. copper amine oxidase has been identified by means of one- and two-dimensional 1H-NMR spectroscopy to be N-amidino-2-hydroxypyrrolidine. This compound inhibits competitively rat nitric oxide synthase type I and type II (NOS-I and NOS-II, respectively) and bovine trypsin (trypsin) activity, values of Ki being (1.1 (cid:139) 0.1) · 10)5 M (at pH 7.5 and 37.0 (cid:176)C), (2.1 (cid:139) 0.1) · 10)5 M (at pH 7.5 and 37.0 (cid:176)C), and (8.9 (cid:139) 0.4) · 10)5 M (at pH 6.8 and 21.0 (cid:176)C), the a(cid:129)nity of N-amidino- respectively. Remarkably, 2-hydroxypyrrolidine for NOS-I, NOS-II and trypsin is

with TPQ in the reductive part of the process forming a Schiff base complex (reaction 1). Proton abstraction of the substrate, catalyzed by an invariant Asp residue, leads to the release of product aldehyde and leaves the enzyme in the reduced aminoquinol form (reaction 1) [1–4]. The oxidative part (reaction 2) leads to reoxidation of the aminoquinol cofactor with the release of ammonia and hydrogen peroxide [1–4].

Copper amine oxidase has been identified in bacteria, yeasts, fungi, plants, and animals. This enzyme is a homodimer of 70- to 90-kDa subunits, each containing a single copper ion and a covalently bound cofactor formed by the post- translational modification of the catalytic tyrosyl residue to 2,4,5-trihydroxyphenylalanine quinone (TPQ) [1–4]. Copper amine oxidase catalyzes the oxidative deamination of biogenic amines, including mono, di, and polyamines, neurotransmitters such as catecholamines, histamine and xenobiotic amines, with substrate preferences depending upon the enzyme source [1–5]. The copper amine oxidase catalyzed reactions follow the general scheme:

Eox (cid:135) R-CH2-NH2 ! Ered (cid:135) R-CHO (cid:133)reaction 1(cid:134)

Ered (cid:135) O2 (cid:135) H2O ! Eox (cid:135) NH3 (cid:135) H2O2 (cid:133)reaction 2(cid:134)

where Eox represents the enzyme–quinone, R-CH2-NH2 is the substrate, Ered is the enzyme–aminoquinol, and R-CHO is the product aldehyde. Substrate amines interact directly

Copper amine oxidase catalyzes also the oxidation of agmatine [3–5], which has been recognized to be an impor- tant bioactive molecule, being identified as a novel neuro- transmitter and modulator of cardiovascular functions via binding to type 1 imidazoline (I1-R) and a-adrenergic receptors [6,7]. Interestingly, agmatine inhibits nitric oxide synthase isoforms [8,9] and induces the release of some peptide hormones [7]. To date, the product(s) of the copper amine oxidase catalyzed oxidation of agmatine has not been identified. Moreover, no information is available on the role played by the product(s) of agmatine metabolism on cell function(s). Here, the biosynthesis and the biochemical characterization of N-amidino-2-hydroxypyrrolidine, the product of agmatine oxidation by Pisum sativum L. copper amine oxidase, is reported.

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

Proteins

M EGTA, 1.0 · 10)3

P. sativum copper amine oxidase was purified as previously reported [10]. Rat nitric oxide synthase type I (NOS-I) was prepared from the rat brain homogenate [11]. Rat nitric oxide synthase type II (NOS-II) was prepared from the lung homogenate of rats treated with E. coli lipopolysaccharide (10 mgÆkg)1) [11]. NOS-I and NOS-II containing specimens were homogenized at pH 7.5 (5.0 · 10)2 M Hepes buffer), 5.0 · 10)4 M dithiothreitol, and 0.1 mgÆmL)1 phenylmethanesulfonyl fluoride [11]. Then,

Correspondence to P. Ascenzi, Dipartimento di Biologia, Universita` (cid:212)Roma Tre(cid:213), Viale Guglielmo Marconi 446, I-00146 Rome, Italy. Fax: + 39 06 55176321, Tel.: + 39 06 55176329, E-mail: ascenzi@uniroma3.it Abbreviations: I1-R, type 1 imidazoline receptor; MMFF, Merck Molecular Force Field; NOS-I, rat nitric oxide synthase type I (neu- ronal constitutive isoform); NOS-II, rat nitric oxide synthase type II (inducible isoform); TPQ, 2,4,5-trihydroxyphenylalanine quinone; trypsin, bovine trypsin. Enzymes: bovine catalase (EC 1.11.1.6); bovine trypsin (EC 3.4.21.4); Pisum sativum L. copper amine oxidase (EC 1.4.3.6); rat nitric oxide synthase type I (EC 1.14.13.39); rat nitric oxide synthase type II (EC 1.14.13.39). *Note: These authors contributed equally to this work. (Received 26 July 2001, revised 17 October 2001, accepted 3 December 2001)

N-Amidino-2-hydroxypyrrolidine characterization (Eur. J. Biochem. 269) 885

(cid:211) FEBS 2002

M and 5.0 · 10)3

NOS-I and NOS-II containing homogenates were desalted by chromatography over disposable PD-10 columns packed with Sephadex G-25 medium (Amersham Pharmacia Bio- tech, Uppsala, Sweden). Bovine calmodulin, bovine cata- lase, bovine serum albumin, bovine trypsin (trypsin), and horseradish peroxidase were purchased from Sigma Chemical Co (St Louis, MO, USA). Proteins were of reagent grade and used without further purification.

Chemicals

In the enzyme assay, the P. sativum copper amine oxidase concentration was 5.0 · 10)9 M and the agmatine concen- tration ranged between 5.0 · 10)5 M. The enzyme activity was linear up to 5 min of incubation and results were expressed as lmol productÆs)1Æ(lmol enzyme))1. Under all the experimental conditions, the initial velocity for the P. sativum copper amine oxidase catalyzed oxidation of agmatine was unaffected by the enzyme/substrate incuba- tion time. In fact, the enzyme/substrate equilibration time was very short, being completed within the mixing time ((cid:25) 15 s).

Values of the first-order rate-limiting catalytic constant (kcat) and of the Michaelis constant, as determined in the absence of the inhibitor (K 0 m) for the P. sativum copper amine oxidase catalyzed oxidation of agmatine, were obtained from the dependence of the initial velocity for agmatine oxidation (vi) on the substrate (i.e. agmatine) concentration ([S]), according to Eqn (1) [13]:

(cid:133)1(cid:134)

vi (cid:136) kcat(cid:137)S(cid:138)=(cid:133)K0

m (cid:135) (cid:137)S(cid:138)(cid:134)

Agmatine, aminoantipyrine, N-a-benzoyl-L-arginine p-nitro- anilide, clonidine, 3,5-dichloro-2-hydroxybenzenesulfonic acid, epinephrine, phenylmethanesulfonyl fluoride, and Escherichia coli lipopolysaccharide (serotype 0127:B8) were obtained from Sigma Chemical Co. [3H]L-arginine (specific activity 2.0 TBqÆmmol)1) and [3H]clonidine (specific activity 2.6 TBqÆmmol)1) were purchased from NENTM Life Science Products (Boston, MA, USA). Deuterium oxide (99.8% isotopic enrichment) was obtained from Cortec (Paris, France). All the other chemicals were from Merck AG (Darmstadt, Germany). All products were of analytical or reagent grade and used without further purification.

Values of kcat and K 0 m for the P. sativum copper amine oxidase catalyzed oxidation of agmatine are 1.3 (cid:139) 0.1 s)1 and (3.8 (cid:139) 0.3) · 10)4 M, respectively, at pH 7.0 (1.0 · 10)1 M phosphate buffer) and 25.0 (cid:176)C (Fig. 1). Values of kcat and K 0

Animals

m are independent of the enzyme assay.

Biosynthesis of N-amidino-2-hydroxypyrrolidine

Male Sprague–Dawley rats (from Morini, Italy), 4- to 5-month-old, were housed and acclimatized for 1 week under controlled temperature (20 (cid:139) 1 (cid:176)C), humidity (55 (cid:139) 10%), and light (from 7 a.m. to 7 p.m) conditions. The rats were anaesthetized with ether in a fume hood, and organs removed and rapidly chilled in liquid nitrogen (brain and lung) or in ice-cold medium solution (2.0 · 10)2 M NaHCO3; heart). Animal experiments were performed accor- ding to ethical guidelines for the conduct of animal research.

P.sativumcopper amine oxidase assay

N-Amidino-2-hydroxypyrrolidine was synthesized as fol- lows. Twenty micrograms of P. sativum copper amine oxidase were added to 1.0 mL of a buffered 2.0 · 10)3 M agmatine solution (5.0 · 10)2 M phosphate buffer, pH 7.4). 28 lg of bovine catalase were also added to the reaction solution (1.0 mL) in order to remove H2O2, arising from the P. sativum copper amine oxidase catalyzed oxidation of agmatine. The reaction solution was stirred vigorously at 25.0 (cid:176)C for 20 min, and the product recovered by ultrafil- tration on Amicon PM10 membranes (Amicon, Inc., Beverly, MA, USA).

M

Oxidation of agmatine by P. sativum copper amine oxi- dase was investigated spectrophotometrically by follo- wing the formation of a pink adduct (e515nm (cid:136) 2.6 · )1Æcm)1), as a result of the oxidation of aminoanti- 104 pyrine and 3,5-dichloro-2-hydroxybenzenesulfonic acid cat- alyzed by horseradish peroxidase, at pH 7.0 (1.0 · 10)1 M phosphate buffer) and 25.0 (cid:176)C [5,6,10]. In a typical experi- ment, 20 lL of a buffered P. sativum copper amine oxidase solution (1.0 · 10)1 M phosphate buffer, pH 7.0) were added to a buffered solution (1.0 mL; 1.0 · 10)1 M phosphate buffer, pH 7.0) containing the substrate (i.e. agmatine), aminoantipyrine (1.0 · 10)4 M), 3,5-dichloro- 2-hydroxybenzenesulfonic acid (1.0 · 10)3 M), and horse- radish peroxidase (1.5 · 10)6 M). The initial velocity for the enzymatic oxidation of agmatine was then measured.

P. sativum copper amine oxidase activity was also assayed polarographically with a Clark electrode (Hansa- tech Instruments Ltd, Norfolk, UK) by following the O2 consumption, at pH 7.0 (1.0 · 10)1 M phosphate buffer) and 25.0 (cid:176)C [12]. In a typical experiment, 20 lL of a buffered agmatine solution (1.0 · 10)1 M phosphate buffer, pH 7.0) were added to a buffered solution (1.0 mL; 1.0 · 10)1 containing M phosphate buffer, pH 7.0) P. sativum copper amine oxidase. The initial velocity for the enzymatic oxidation of agmatine was then measured.

Fig. 1. Effect of substrate (i.e. agmatine) concentration on values of vi for the P. sativum copper amine oxidase catalyzed oxidation of agma- tine. The continuous line was calculated according to Eqn (1), with the following values of kcat((cid:136) 1.3 (cid:139) 0.1 s)1) and K 0 m[(cid:136) (3.8 (cid:139) 0.3) · 10)4 M]. Data were obtained at pH 7.0 and 25.0 (cid:176)C, mean (cid:139) SD. For further details, see text.

886 P. Ascenzi et al.

(Eur. J. Biochem. 269)

(cid:211) FEBS 2002

M FMN,

M FAD, 1.0 · 10)5

The total conversion of agmatine to N-amidino-2- hydroxypyrrolidine was detected by 1H-NMR spectroscopy. Moreover, the agmatine/N-amidino-2-hydroxypyrrolidine stoichiometry is 1 : 1 as shown by 1H-NMR spectroscopy. The N-amidino-2-hydroxypyrrolidine concentration was determined from 100% conversion of agmatine to N-amidino-2-hydroxypyrrolidine as demonstrated by 1H-NMR spectroscopy.

Under all the experimental conditions, the formation of free 4-guanidinobutyraldehyde was observed neither by the o-aminobenzaldehyde assay [14] (data not shown) nor 1H-NMR spectroscopy (Figs 2 and 3).

NMR spectroscopy

(5.0 · 10)2 M Hepes buffer) and 37.0 (cid:176)C, in the absence and presence of N-amidino-2-hydroxypyrrolidine. In a typical experiment, a NOS-I or NOS-II aliquot (50 lL) was added to the reaction mixture (100 lL) containing 1.0 · 10)3 M M CaCl2, 1.0 lgÆmL)1 calmodulin, NADPH, 1.2 · 10)3 1.0 · 10)5 [3H]L-arginine (from 12 to 185 kBq) and L-arginine (from 1.0 · 10)6 M to 1.0 · 10)4 M), in the absence and presence of N-ami- dino-2-hydroxypyrrolidine (from 5.0 · 10)6 M and 5.0 · 10)5 M). For the determination of NOS-II activity, CaCl2 and calmodulin were omitted, and 1.0 · 10)3 M EGTA was added to the reaction mixture. NOS-I and NOS-II activity was assayed in the presence of 5.0 · 10)5 M BH4 [16]. In the enzyme assay, the NOS-I or NOS-II concentration was 2.0 · 10)7 M. After 15 min incubation, the reaction was stopped by addition of an ice-cold 2.0 · 10)2 M Hepes buffer solution (700 lL), pH 5.5, containing 2.0 · 10)3 M EDTA. [3H]L-citrulline was separated from [3H]L-arginine by ion exchange chromatography on Dowex 50WX8 (Fluka Chemie AG) [11,16]. The enzyme activity was linear up to 30 min of incubation and results were expressed as pmol productÆmin)1Æ(mg protein))1. Under all the experimental conditions, the initial velocity for NOS-I and NOS-II catalyzed conversion of L-arginine to L-citrulline was unaffected by the enzyme/inhibitor/substrate incuba- tion time. In fact, the enzyme/inhibitor/substrate equilibra- tion time was very short, being completed within the mixing time ((cid:25) 15 s).

P. sativum copper amine oxidase catalyzed oxidation of agmatine was conducted as described above, in deuterated phosphate buffer (pD 7.4; uncorrected pH-meter reading 7.0); residual oxygen was removed with a mild nitrogen stream. A control spectrum was recorded prior to addition of P. sativum copper amine oxidase. 1H-NMR one- and two-dimensional spectra were recorded at 25.0 (cid:176)C on a Bruker AVANCE 600 NMR spectrometer (Bruker Ana- lytik, Rheinstetten, Germany), operating at a magnetic field strength of 14.1 T. The residual water signal was suppressed by a 2-s presaturation before the observation pulse. The duration of the pulse corresponding to a flip angle of 90(cid:176) was 7.4 ls. The spin system of the agmatine oxidation product was assigned by COSY, by setting the flip angle of the second pulse to 35(cid:176). To this purpose, 256 t1 increments were recorded (4096 points each). The resulting matrix was zero-filled to 1024 · 4096 complex points and processed with a 5(cid:176)-shifted squared sinebell in both dimensions [15].

Building of the N-amidino-2-hydroxypyrrolidine structure

Energy minimization of the proposed structure of N-amidino-2-hydroxypyrrolidine was performed on a Silicon Graphics Octane workstation (SGI, Mountain View, CA, USA) by using the program SPARTAN (Wave- function Inc., Irvine, CA, USA).

M, respectively, at pH 7.5 and 37.0 (cid:176)C [17].

Values of the first-order rate-limiting catalytic constant (kcat) and of the Michaelis constant, as determined in the absence and presence of the inhibitor (K 0 m and K app m , respectively), for NOS-I and NOS-II catalyzed conversion of L-arginine to L-citrulline were obtained from the depen- dence of the initial velocity for substrate conversion (vi) on the L-arginine concentration ([S]), according to Eqn (1) [13]. Values of kcat and K 0 m for the NOS-I catalyzed conversion of L-arginine to L-citrulline were 1.4 (cid:139) 102 pmol prod- uctÆmin)1Æ(mg protein))1 and 4.0 · 10)6 M, respectively, at pH 7.5 and 37.0 (cid:176)C [11]. Values of kcat and K 0 m for the NOS-II catalyzed conversion of L-arginine to L-citrulline were 4.7 · 101 pmol productÆmin)1Æ(mg protein))1 and 1.8 · 10)5

NOS-I and NOS-II assay

NO production was also monitored spectrophotometri- cally (between 350 and 460 nm) following the NO-mediated conversion of human oxy-hemoglobin (6.0 · 10)6 M), added to the NOS-I and NOS-II preparations, to met-hemoglobin,

NOS-I and NOS-II activity was assessed by evaluating the conversion of [3H]L-arginine to [3H]L-citrulline at pH 7.5

M

Fig. 2. 1H-NMR spectra of 2.0 · 10)3 agmatine before (A) and after (B) oxidation catalyzed by P. sativum copper amine oxidase, at pD 7.4 and 25.0 (cid:176)C. Acquisition param- eters: 4 scans, flip angle 45(cid:176), relaxation delay 2 s. The residual water signal was suppressed by presaturation. For further details, see text.

N-Amidino-2-hydroxypyrrolidine characterization (Eur. J. Biochem. 269) 887

(cid:211) FEBS 2002

M and 8.0 · 10)5

2-hydroxypyrrolidine. In a typical experiment, 20 lL of a buffered trypsin solution (1.0 · 10)1 M phosphate buffer, pH 6.8) were added to 1.0 mL of a buffered solution (1.0 · 10)1 M phosphate buffer, pH 6.8) containing the substrate (i.e. N-a-benzoyl-L-arginine p-nitroanilide) and the inhibitor (i.e. N-amidino-2-hydroxypyrrolidine). The initial velocity for the enzymatic hydrolysis of N-a-benzoyl- L-arginine p-nitroanilide was then measured. In the enzyme assay, the trypsin concentration was 1.0 · 10)6 M, the N-a-benzoyl-L-arginine p-nitroanilide concentration ranged M and 1.0 · 10)3 between 1.0 · 10)5 M, and the N-ami- dino-2-hydroxypyrrolidine concentration ranged between 2.0 · 10)5 M. The enzyme activity was linear up to 10 min of incubation and results were expressed as lmol productÆs)1Æ(lmol enzyme))1. Under all the exper- imental conditions, the initial velocity for the trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroani- lide was unaffected by the enzyme/inhibitor/substrate incubation time. In fact, the enzyme/inhibitor/substrate equilibration time was very short, being completed within the mixing time ((cid:25) 15 s).

for

the initial velocity for

M, respectively, at pH 6.8 and 21.0 (cid:176)C [20].

Values of the first-order rate-limiting catalytic constant (kcat) and of the Michaelis constant determined in the absence and presence of the inhibitor (K 0 m and K app m , the trypsin catalyzed hydrolysis of respectively) N-a-benzoyl-L-arginine p-nitroanilide were obtained from the dependence of substrate hydrolysis (vi) on the N-a-benzoyl-L-arginine p-nitroani- lide concentration ([S]), according to Eqn (1) [13]. Values of kcat and K 0 m for the trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroanilide were 0.70 s)1 and 3.0 · 10)4

Determination of values of the inhibition dissociation equilibrium constant (Ki) for N-amidino-2-hydroxypyrrolidine binding to NOS-I, NOS-II, and trypsin

m /K 0

Values of the inhibition dissociation equilibrium constant (Ki) for the competitive inhibition of the NOS-I and NOS-II catalyzed conversion of L-arginine to L-citrulline (at pH 7.5 and 37.0 (cid:176)C) and of the trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroanilide (at pH 6.8 and 21.0 (cid:176)C) by N-amidino-2-hydroxypyrrolidine were deter- mined from the linear dependence of the K app m ratio on the inhibitor concentration (i.e. [I]), according to Eqn (2) [13]:

Kapp

(cid:137)I(cid:138) (cid:135) 1

(cid:133)2(cid:134)

m =K0

i

m (cid:136) K (cid:255) 1 As expected for a simple competitive inhibition system [13], values of kcat for the NOS-I and NOS-II catalyzed conversion of L-arginine to L-citrulline and for the trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroani- lide were unaffected by the inhibitor concentration within the standard deviation ((cid:139) 5%).

in the presence of N-amidino-2-hydroxypyrrolidine as the substrate instead of L-arginine, at pH 7.5 (5.0 · 10)2 M Hepes buffer) and 37.0 (cid:176)C [18,19].

Trypsin assay

Model building of the NOS-II: and trypsin: N-amidino-2-hydroxypyrrolidine complexes

Molecular models of the human NOS-II: and bovine trypsin:N-amidino-2-hydroxypyrrolidine complexes were built using the coordinates of the human NOS-II:S-ethyl-

The trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroanilide was investigated spectrophotometrically (at 408 nm), at pH 6.8 (1.0 · 10)1 M phosphate buffer) and 21.0 (cid:176)C [20], in the absence and presence of N-amidino-

Fig. 3. Two-dimensional COSY spectrum of N-amidino-2-hydroxy- pyrrolidine, the cyclic oxidation product of agmatine, at pD 7.4 and 25.0 (cid:176)C (top) and ball-and-stick model of N-amidino-2-hydroxypyrro- lidine (bottom). Acquisition parameters: 4 scans, 16 dummy scans, relaxation delay 2 s. Labels refer to the resonance assignment in Fig. 1B. For further details see text.

888 P. Ascenzi et al.

(Eur. J. Biochem. 269)

(cid:211) FEBS 2002

of the ligand (i.e. N-amidino-2-hydroxypyrrolidine, agma- tine, or clonidine) [29].

R E S U L T S

(i.e.

between

explored

5.0 · 10)5

isothiourea complex (PDB accession no. 4NOS) [21] and the bovine trypsin:benzamidine adduct (PDB accession no. 1CE5) [22] as templates, respectively. The atomic coordi- nates of rat NOS-II are not yet available [23], the homologous human enzyme was used instead. The confor- mations of the N-amidino-2-hydroxypyrrolidine in the enzyme:inhibitor complexes were obtained after 10 ps molecular dynamics. Energy minimization and molecular dynamics were performed on a Silicon Graphics O2 workstation (SGI, Irvine, CA, USA) with HYPERCHEM 4.5 for SGI (Hypercube Inc., Gainesville, FL, USA).

I1-R binding assay

Over the whole substrate (i.e. agmatine) concentration M and range 5.0 · 10)3 M), the P. sativum copper amine oxidase cata- lyzed oxidation of agmatine follows simple Michaelis– Menten kinetics (Fig. 1). According to the literature [30], values of kcat and K 0 m for the P. sativum copper amine oxidase catalyzed oxidation of agmatine are 1.3 (cid:139) 0.1 s)1 and (3.8 (cid:139) 0.3) · 10)4 M, respectively, at pH 7.0 and 25.0 (cid:176)C. Moreover, values of kcat and K 0 m were independent of the enzymatic assay used (spectrophotometric vs. pola- rographic). The stoichiometric analysis of the enzymatic oxidation of agmatine yields a molar ratio of substrate (i.e. agmatine) to O2 and H2O2 of 1 : 1 : 1.

the -CHO signal was not observed,

Figure 2 shows the 1H-NMR spectra of agmatine before (Fig. 2A) and after (Fig. 2B) oxidation catalyzed by P. sativum copper amine oxidase, at pD 7.4 and 25.0 (cid:176)C. The agmatine sample shows some signals at the impurity level, which however do not hamper the observation of the main component. The main features of Fig. 2B with respect to Fig. 2A are: (a) the upset of a downfield-shifted signal at d (cid:136) 5.5 p.p.m., and (b) the splitting of CH2 signals in magnetically unequivalent components. On the basis of the general mechanism (see reactions 1 and 2), one triplet (relative area 1) should occur at about d (cid:136) 9 p.p.m., corresponding to the formyl proton, one triplet at about d (cid:136) 3 p.p.m. (relative area 2), and two multiplets at about d (cid:136) 2 p.p.m. (relative area 2 each). As the formation of the corresponding free aldehyde (i.e. 4- guanidobutyraldehyde) was ruled out. To note that the agmatine/N-amidino-2-hydroxypyrrolidine stoichiometry is 1 : 1 as shown by 1H-NMR spectroscopy.

the resolution of

A possible explanation for

between

ranged

1.5 · 10)4

the magnetic equivalence of CH2 groups would be the formation of an intramolecular Schiff base in its emiac- etalic form, deriving from nucleophilic attack of the guanidinic eN nitrogen to the (transient) aldehydic carbonyl. This implies the formation of a chiral center on the ring, with all CH2 protons consequently becoming diastereotopic and hence magnetically non equivalent (see Scheme 1). As the presence of free 4-guanidobutyralde- hyde was never detected, the formation of the cyclic product N-amidino-2-hydroxypyrrolidine should occur within the enzyme catalytic center (shown within square brackets in Scheme 1).

from 1.0 · 10)9

Cardiac muscle (cleaned of connective tissue and fat) was finely minced and homogenized in ice-cold medium solution 2.0 · 10)2 M NaHCO3, containing 1.0 · 10)4 M phen- ylmethanesulfonyl fluoride, with a wet weight to volume ratio of 1 : 7, using a glass-Teflon homogenizer (10 · 30 s) [24]. The homogenate was centrifuged at 1500 g for 15 min (4.0 (cid:176)C). The supernatant was centrifuged at 45 000 g for 5 min (at 4.0 (cid:176)C). The pellet was washed twice, then re-suspended in 2 mL of ice-cold 5.0 · 10)3 M Hepes buffer, containing 5.0 · 10)4 M EGTA, 5.0 · 10)4 M MgCl2, and 1.0 · 10)4 M ascorbic acid (pH 7.4) [25]. Membrane pre- parations were free of mitochondria and nuclei as confirmed by subcellular enzymatic marker assays (data not shown). Two-hundred and forty micrograms of membrane pro- tein were incubated for 55 min with 1.3 nmol to 40 nmol [3H]clonidine at 37.0 (cid:176)C in a final volume of 0.5 mL of M Hepes buffer, containing 5.0 · 10)4 5.0 · 10)3 M EGTA, 5.0 · 10)4 M MgCl2, and 1.0 · 10)4 M ascorbic acid (pH 7.4). The reaction was stopped by rapid vacuum filtration with a Millipore harvester through Whatman GF/C glass fiber filters (Whatman International Ltd Maidstone, UK) presoaked with 10% polyethyleneglycol in Tris/HCl 2.0 · 10)2 M, containing MgCl2 1.0 · 10)2 M, followed by rapid washing of filters with 10 mL ice-cold 5.0 · 10)3 M M EGTA, 5.0 · 10)4 Hepes buffer, containing 5.0 · 10)4 M MgCl2, and 1.0 · 10)4 M ascorbic acid (pH 7.4). Filters were placed in a 6-mL scintillation fluid and the radio- activity determined by liquid scintillation counting. Epine- phrine (1.0 · 10)5 M), which does not bind to imidazoline sites [26,27], was added to the assay to prevent [3H]clonidine from binding to a-adrenergic receptors. Nonspecific binding was defined as [3H]clonidine-binding (the [3H]clonidine M and concentration 5.0 · 10)4 M). Saturation studies were performed with 1.0 · 10)8 M [3H]clonidine and increasing concentrations of the unlabelled ligand (i.e. N-amidino-2-hydroxypyrroli- M to dine, agmatine, and clonidine; 1.0 · 10)6 M). Protein concentration was measured by the method of Bradford [28], using bovine serum albumin as the standard.

Values of IC50 for [3H]clonidine displacement from I1-R in heart rat membranes by N-amidino-2-hydroxypyrroli- dine, agmatine, and clonidine were determined according to Eqn (3):

(cid:133)3(cid:134)

a (cid:136) 1=f1 (cid:135) (cid:133)(cid:137)L(cid:138)=IC50(cid:134)g

where a is the molar fraction of [3H]clonidine bound to I1-R present in heart rat membranes and [L] is the concentration

Figure 3 (top panel) shows the magnitude COSY spec- trum of the product of agmatine oxidation catalyzed by P. sativum copper amine oxidase. Starting from the emiac- etalic proton A, it is possible to walk over the whole spin system and identify the connectivities on the basis of 3J scalar couplings [15]. As three-bond couplings were not observed, it was assumed that the involved protons form dihedral angles close to 90(cid:176) [31]. In other words, the absence of scalar coupling between A and, say, C identified the axial- equatorial pairs. Figure 3 (bottom panel) shows the ball- and-stick model of N-amidino-2-hydroxypyrrolidine (the product of agmatine oxidation catalyzed by P. sativum copper amine oxidase) after 200 cycles of energy minimi-

N-Amidino-2-hydroxypyrrolidine characterization (Eur. J. Biochem. 269) 889

(cid:211) FEBS 2002

m =K 0

Fig. 4. Effect of N-amidino-2-hydroxypyrrolidine concentration (i.e. [Inhibitor]) on the Kapp m ratio for the competitive inhibition of NOS-I (squares) and NOS-II (triangles) catalyzed conversion of L-arginine to L-citrulline, and of the trypsin (circles) catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroanilide. The continuous lines were cal- culated according to Eqn (2) with values of Ki given in Table 1. Data were obtained between pH 6.8 and 7.5 and between 21.0 (cid:176)C and 37.0 (cid:176)C, mean (cid:139) SD, for further details, see text.

As shown in Fig. 4, N-amidino-2-hydroxypyrrolidine inhibits competitively the NOS-I and NOS-II catalyzed conversion of L-arginine to L-citrulline and the trypsin catalyzed hydrolysis of N-a-benzoyl-L-arginine p-nitroani- lide. Table 1 gives Ki values for N-amidino-2-hydroxy- pyrrolidine (present study), agmatine [8,30], and clonidine [16,30] binding to NOS-I, NOS-II, and trypsin. Remark- ably, the affinity of N-amidino-2-hydroxypyrrolidine for NOS-I, NOS-II, and trypsin is systematically higher than that observed for agmatine and clonidine binding (see Table 1). As reported for agmatine [8] and clonidine [16], N-amidino-2-hydroxypyrrolidine is not a NO precursor. In fact, human oxy-hemoglobin added to NOS-I and NOS-II preparations is not converted to met-hemoglobin in the presence of N-amidino-2-hydroxypyrrolidine as the sub- strate instead of L-arginine (data not shown).

Figure 5 shows the molecular models of the human NOS-II: and bovine trypsin:N-amidino-2-hydroxypyrro-

zation in the MMFF force field [32], with torsion angles constrained according to the results of the COSY spectrum (Fig. 3, top panel).

Scheme 1.

Table 1. Values of Ki for N-amidino-2-hydroxypyrrolidine, agmatine, and clonidine binding to NOS-I, NOS-II, and trypsin.

Ki (M)

Enzyme N-Amidino-2- hydroxypyrrolidine Agmatine Clonidine

a pH 7.5 and 37.0 (cid:176)C. Present study. b pH 7.8 and 37.0 (cid:176)C. From [8]. c pH 7.5 and 37.0 (cid:176)C. From [16]. d pH 6.8 and 21.0 (cid:176)C. Present study. e pH 7.0 and 25.0 (cid:176)C. From [30].

NOS-I NOS-II Trypsin (6.6 (cid:139) 1.1) · 10)4 b (2.2 (cid:139) 0.2) · 10)4 b >10)2 e (5.0 (cid:139) 0.2) · 10)3 c >5 · 10)2 c >10)2 e (1.1 (cid:139) 0.1) · 10)5 a (2.1 (cid:139) 0.1) · 10)5 a (8.9 (cid:139) 0.4) · 10)5 d

890 P. Ascenzi et al.

(Eur. J. Biochem. 269)

(cid:211) FEBS 2002

of N-amidino-2-hydroxypyrrolidine

which is required for L-arginine binding [21,33]. By homo- logy, this residue corresponds to Glu597 and Glu371 in rat NOS-I and NOS-II, respectively [23]. Moreover, as previ- ously reported for the bovine trypsin:benzamidine complex [22,34], N-amidino-2-hydroxypyrrolidine binds to the enzyme primary specificity subsite S1 (bottom panel). Interestingly, the alicyclic group is extended in a semichair conformation, with the positively charged amidino group of N-amidino-2-hydroxypyrrolidine forming a salt bridge with the negatively charged carboxylate of the trypsin Asp189 residue. The latter is required for recognition of the cationic amino acid residue present at the P1 position of substrates and inhibitors of trypsin-like serine proteinases [35,36].

Fig. 6. Competition (circles), agmatine (triangles), and clonidine (squares) with [3H]clonidine for its specific binding sites in rat heart membranes. The filled diamond indi- cates [3H]clonidine saturating specific binding (a (cid:136) 1) in the absence of the ligand (i.e. clonidine, or agmatine or N-amidino-2-hydroxypyrro- lidine). The continuous lines were calculated according to Eqn (3) with the following IC50 values: N-amidino-2-hydroxypyrrolidine and agmatine, IC50 (cid:136) (1.3 (cid:139) 0.4) · 10)9 M, and clonidine, IC50 (cid:136) (2.2 (cid:139) 0.4) · 10)8 M. Data were obtained at pH 7.4 and 37.0 (cid:176)C, mean (cid:139) SD. For further details, see text.

lidine complexes. In human NOS-II (top panel), N-amidino- 2-hydroxypyrrolidine is hosted in the hydrophobic cavity defined by the heme prosthetic group and by the facing hydrophobic residues Ala270 and Val271, as observed for a number of nitrogen heterocycles [21,33] (note that N-ami- dino-2-hydroxypyrrolidine is constrained in a semiboot conformation, with the nitrogen lone pair directed towards the heme iron). The positively charged amidino group of N-amidino-2-hydroxypyrrolidine appears to be stabilized by the negatively charged carboxylate of the Glu296 residue

N-Amidino-2-hydroxypyrrolidine, agmatine, and cloni- dine bind to I1-binding sites (i.e. I1-R). In fact, the I2 sites, which are not considered as receptors and showing a mitochondrial localization possibly corresponding to monoamine oxidase [37,38], are removed from rat heart membrane preparations. Figure 6 shows [3H]clonidine displacement from I1-R present in rat heart membranes by N-amidino-2-hydroxypyrrolidine, agmatine, and cloni- dine. As observed in other target tissues [25], the specific binding of [3H]clonidine to rat heart membranes is saturable (data not shown). Moreover, specific binding amounts to 3650 (cid:139) 294 d.p.m.Æh)1Æ(mg protein))1, at saturating [3H]clonidine concentration ((cid:136) 1.0 · 10)8 M). N-amidino- 2-hydroxypyrrolidine and agmatine are more efficient than clonidine in displacing [3H]clonidine from specific binding sites in heart rat membranes, values of IC50 being

Fig. 5. N-Amidino-2-hydroxypyrrolidine binding mode to human NOS-II (top) and bovine trypsin (bottom). The conformations of N-amidino- 2-hydroxypyrrolidine in the enzyme:inhibitor complexes were obtained after 10 ps molecular dynamics. For further details, see text.

N-Amidino-2-hydroxypyrrolidine characterization (Eur. J. Biochem. 269) 891

(cid:211) FEBS 2002

M, respec-

M and (2.2 (cid:139) 0.4) · 10)8

(1.3 (cid:139) 0.4) · 10)9 tively (at pH 7.4 and 37.0 (cid:176)C) (Fig. 6).

2. Fontecave, M. & Eklund, H. (1995) Copper amine oxidase: a novel use for a tyrosine. Structure 3, 1127–1129.

D I S C U S S I O N

3. Klinman, J.P. (1996) Mechanisms whereby mononuclear copper proteins functionalize organic substrates. Chem. Rev. 96, 2541– 2561.

4. Buffoni, F. & Ignesti, G. (2000) The copper-containing amine oxidases: biochemical aspects and functional role. Mol. Genet. Metab. 71, 559–564.

5. Federico, R., Angelini, R., Ercolini, L., Venturini, G., Mattevi, A. & Ascenzi, P. (1997) Competitive inhibition of swine kidney copper amine oxidase by drugs: amiloride, clonidine, and gabexate mesylate. Biochem. Biophys. Res. Commun. 240, 150–152.

For the first time, N-amidino-2-hydroxypyrrolidine, the product of agmatine oxidation by P. sativum copper amine oxidase, has been identified and characterized from the structural and biochemical viewpoints. Notably, the enzy- matic oxidation of agmatine leads to the cyclic compound N-amidino-2-hydroxypyrrolidine, as the only detectable reaction product (Figs 2 and 3). In fact, the formation of 4-guanidinobutyraldehyde was never observed. Therefore, 4-guanidinobutyraldehyde, the best substrate of the alde- hyde dehydrogenase that occurs in Fabaceae plants and rat hepatocytes with copper amine oxidase [39–42], does not appear to originate from the enzymatic cycling of agmatine to N-amidino-2-hydroxypyrrolidine.

6. Holt, A. & Baker, G.B. (1995) Metabolism of agmatine (clonidine- displacing substance) by diamine oxidase and the possible impli- cations for studies of imidazoline receptors. Prog. Brain. Res. 106, 187–197. 7. Reis, D.J. & Regunathan, S. (2000) Is agmatine a novel neuro- transmitter in brain? Trends Pharmacol. Sci. 21, 187–193.

8. Galea, E., Regunathan, S., Eliopoulos, V., Feinstein, D.L. & Reis, D.J. (1996) Inhibition of mammalian nitric oxide synthases by agmatine, an endogenous polyamine formed by decarboxylation of arginine. Biochem. J. 316, 247–249.

9. Demady, D.R., Jianmongkol, S., Vuletich, J.L., Bender, A.T. & Osawa, Y. (2001) Agmatine enhances the NADPH oxidase activity of neuronal NO synthase and leads to oxidative inacti- vation of the enzyme. Mol. Pharmacol. 59, 24–29.

N-Amidino-2-hydroxypyrrolidine inhibits competitively NOS-I, NOS-II, and trypsin (Fig. 4). This compound binds to the Glu597 and Glu371 carboxylate, present in NOS-I and NOS-II, respectively (Glu296 in human NOS- II; see Fig. 5), which is required for substrate (i.e. L-arginine) recognition [21,33]. Moreover, N-amidino-2- hydroxypyrrolidine binds to the trypsin primary specificity forming a salt bridge with the Asp189 subsite S1 carboxylate (Fig. 5). The latter is required for recognition of the cationic amino acid residue present at the P1 position of substrates and inhibitors of trypsin-like serine proteinases [35,36].

10. McGuirl, M.A., McCahon, C.D., McKeown, K.A. & Dooley, D.M. (1994) Purification and characterization of pea seedling amine oxidase for crystallization studies. Plant Physiol. 106, 1205– 1211.

11. Stuehr, D.J. & Gri(cid:129)th, O.W. (1996) Purification, assay and properties of mammalian nitric oxide synthases. In Methods in Nitric Oxide Research (Feelisch, M. & Stamler, J.S., eds), pp. 177– 186. John Wiley & Sons Ltd, Chichester.

12. Rinaldi, A., Floris, G. & Finazzi-Agro` , A. (1982) Purification and properties of diamine oxidase from Euphorbia latex. Eur. J. Biochem. 127, 417–422.

N-Amidino-2-hydroxypyrrolidine and agmatine displace efficiently [3H]clonidine from I1-R present in heart rat membranes (Fig. 6). Interestingly, different physiological roles (i.e. neuronal neurotransmission and hypotensive protection of cardiovascular system) have been linked to agmatine, which has been reported to be the endogenous the N-amidino- ligand for I-R1 [7] and to represent 2-hydroxypyrrolidine precursor. In this respect, pleiotropic functional role(s) of N-amidino-2-hydroxypyrrolidine may be envisaged, as reported for agmatine [7].

13. Ascenzi, P., Ascenzi, M.G. & Amiconi, G. (1987) Enzyme competitive inhibition: graphical determination of Ki and presen- tation of data in comparative studies. Biochem. Educ. 15, 134– 135.

14. Holmestedt, B., Larsson, L. & Tham, R. (1961) Further studies on spectrophotometric method for the determination of amine oxidase activity. Biochim. Biophys. Acta 48, 182–186. 15. Braun, S., Kalinowski, H.-O. & Berger, S. (1998) 150 and More Basic NMR Experiments. Wiley-VCH, Weinheim.

As a whole, agmatine oxidation by P. sativum copper amine oxidase may represent a new biocatalytic route for the synthesis of N-amidino-2-hydroxypyrrolidine, possibly representing a lead compound for the development of NOS and trypsin-like serine protease inhibitors. Moreover, N-amidino-2-hydroxypyrrolidine may represent a new ligand for I1-R.

16. Venturini, M., Colasanti, M., Persichini, T., Fioravanti, E., Federico, R. & Ascenzi, P. (2000) Selective inhibition of nitric oxide synthase type I by clonidine, an antihypertensive drug. Biochem. Pharmacol. 60, 539–544.

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

17. Venturini, G., Colasanti, M., Fioravanti, E., Bianchini, A. & Ascenzi, P. (1999) Direct effect of temperature on the catalytic activity of nitric oxide synthases types I, II, and III. Nitric Oxide 3, 375–382.

18. Feelisch, M., Kubitzek, D. & Werringloer, J. (1996) The oxyhe- moglobin assay. In Methods in Nitric Oxide Research (Feelish, M. & Stamler, J.S., eds), pp. 455–478. John Wiley & Sons Ltd, Chichester.

Authors wish to thank Prof S. Aime and Dr G. Rea for helpful discussions and Dr L. Leone and Mr A. Merante for technical assistance. This study was partially supported by grants from the National Research Council of Italy (CNR, target oriented project (cid:212)Biotechnology(cid:213), 99.00280.PF49 to P. A., and 99.00360.PF49 to M. F.). Access to the 600 MHz NMR facility has been granted by Bioindustry Park Canavese, Colleretto Giacosa, TO, Italy. 19. Venturini, G., Menegatti, E. & Ascenzi, P. (1997) Competitive inhibition of nitric oxide synthase by p-aminobenzamidine, a serine proteinase inhibitor. Biochem. Biophys. Res. Commun. 232, 88–90.

R E F E R E N C E S

1. McIntire, W.S. & Hartmann, C. (1993) Copper-containing amine oxidases. In Principles and Applications of Quinoproteins (Davison, V.L., ed.), pp. 99–171. Marcel Dekker, New York. 20. Ascenzi, P., Menegatti, E., Guarneri, M., Bortolotti, F. & Anto- nini, E. (1982) Catalytic properties of serine proteases. 2. Com- parison between human urinary kallikrein and human urokinase, bovine b-trypsin, bovine thrombin, and bovine a-chymotrypsin. Biochemistry 21, 2483–2490.

892 P. Ascenzi et al.

(Eur. J. Biochem. 269)

(cid:211) FEBS 2002

31. Friebolin, H. (1993) Basic One- and Two-Dimensional NMR Spectroscopy. VCH, Weinheim.

21. Fischmann, T.O., Hruza, A., Niu, X.D., Fossetta, J.D., Lunn, C.A., Dolphin, E., Prongay, A.J., Reichert, P., Lundell, D.J., Narula, S.K. & Weber, P.C. (1999) Structural characterization of nitric oxide synthase isoforms reveals striking active-site conser- vation. Nat. Struct. Biol. 6, 233–242. 32. Halgren, T.A. (1996) Merck Molecular Force Field. I. Basis, form, scope, parametrization, and performance of MMFF94. J. Com- put. Chem. 17, 490–519.

33. Crane, B.R., Arvai, A.S., Gachhui, R., Wu, C., Ghosh, D.K., Getzoff, E.D., Stuehr, D.J. & Tainer, J.A. (1997) The structure of nitric oxide synthase oxygenase domain and inhibitor complexes. Science 278, 425–431. 22. Ota, N., Stroupe, C., Ferreira da Silva, J.M.S., Shah, S.S., Mares- Guia, M. & Brunger, A.T. (1999) Non-Boltzmann Thermo- dynamic Integration (NBTI) for macromolecular systems: relative free energy of binding of trypsin to benzamidine and benzylamine. Proteins 37, 641–653.

23. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N. & Bourne, P.E. (2000) The Protein Data Bank. Nucleic Acids Res. 28, 235–242. 34. Bode, W. & Schwager, P. (1975) The refined crystal structure of bovine b-trypsin at 1.8 A˚ resolution. II. Crystallographic refine- ment, calcium binding site, benzamidine binding site and active site at pH 7.0. J. Mol. Biol. 98, 693–717. 24. Glossmann, H. & Ferry, D.R. (1985) Assay for calcium channels. Methods Enzymol. 109, 513–550. 35. Bode, W. & Huber, R. (1992) Natural protein proteinase inhibi- tors and their interaction with proteinases. Eur. J. Biochem. 204, 433–451.

36. Bode, W. & Huber, R. (2000) Structural basis of the endopro- teinase–protein inhibitor interaction. Biochim. Biophys. Acta 1477, 241–252.

37. Tesson, F., Limon-Boulez, I., Urban, P., Puype, M., Van- dekerckhove, J., Coupry, I., Pompon, D. & Parini, A. (1995) Localization of I2-imidazoline binding sites on monoamine oxi- dases. J. Biol. Chem. 270, 9856–9861. 25. Molderings, G.J., Donecker, K. & Go¨ thert, M. (1995) Charac- terization of non-adrenergic [3H]clonidine binding sites in rat stomach: high a(cid:129)nity of imidazolines, guanidines and sigma ligands. Naunyn-Schmiedeberg’s Arch. Pharmacol. 351, 561–564. 26. Molderings, G.J., Moura, D., Fink, K., Bo¨ nisch, H. & Go¨ thert, M. (1993) Binding of [3H]clonidine to I1-imidazoline sites in bovine adrenal medullary membranes. Naunyn-Schmiedeberg’s Arch. Pharmacol. 348, 70–76.

38. Ernsberger, P. & Haxhiu, M.A. (1997) The I1-imidazoline-binding site is a functional receptor mediating vasodepression via the ventral medulla. Am. J. Physiol. 273, R1572–R1579. 27. Molderings, G.J., Kundt, L. & Go¨ thert, M. (1994) [3H]Idazoxan binding to bovine adrenal medullary membranes: identification and pharmacological characterization of I2-imidazoline sites. Naunyn-Schmiedeberg’s Arch. Pharmacol. 350, 252–257.

39. Matsuda, H. & Suzuki, Y. (1981) Purification and properties of the diamine oxidase from Vicia faba leaves. Plant Cell Physiol. 22, 737–746. 28. Bradford, M.M. (1976) A rapid, sensitive method for the quanti- fication of mg quantities of protein, utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254.

40. Matsuda, H. & Suzuki, Y. (1984) c-Guanidinobutyraldehyde dehydrogenase of Vicia faba leaves. Plant Physiol. 76, 654–657. 41. Kaneoke, M., Shimizu, E. & Yorifuji, T. (1994) Metabolism of L-arginine, agmatine, and related-compounds in Nocardioides simplex. Biosci. Biotechn Biochem. 58, 244–249. 29. Ascenzi, P., Desideri, A., Amiconi, G., Bertollini, A., Bolognesi, M., Castagnola, M., Coletta, M. & Brunori, M. (1988) Effect of inositol hexakisphosphate on the spectroscopic properties of the nitric oxide derivative of ferrous naturally glycated human hemoglobin HbA1c. J. Inorg. Biochem. 34, 19–24.

42. Cabella, C., Gardini, G., Corpillo, D., Testore, G., Bedino, S., Solinas, S.P., Cravanzola, C., Vargiu, C., Grillo, M.A. & Colombatto, S. (2001) Transport and metabolism of agmatine in rat hepatocyte cultures. Eur. J. Biochem. 268, 940–947. 30. Federico, R., Leone, L., Botta, M., Binda, C., Angelini, R., Venturini, G. & Ascenzi, P. (2001) Inhibition of pig liver and Zea mays L. polyamine oxidase: a comparative study. J. Enzyme Inhib. 16, 147–155.