M. Scarpa, Dipartimento di Fisica, Via Sommarive 14, 38050 Povo-Trento, Italy Fax: ++39 0461881696 Tel: ++39 0461882029 E-mail: email@example.com
The structures of copper amine oxidases from various sources show good similarity, suggesting similar catalytic mechanisms for all members of this enzyme family. However, the optimal substrates for each member differ, depending on the source of the enzyme and its location. The structural factors underlying substrate selectivity still remain to be discovered. With this in view, we examined the kinetic behaviour of pea seedling amine oxidase with cadaverine and hexylamine, the first bearing two, and the second only one, positively charged amino group. The dependence of Km and catalytic constant (kc) values on pH, ionic strength and temperature indicates that binding of the monoamine is driven by hydrophobic interactions. Instead, binding of the diamine is strongly facilitated by electrostatic factors, controlled by polar side-chains and two titratable residues present in the active site. The position of the docked substrate is also essential for the participation of titratable amino acid residues in the following catalytic steps. A new mechanistic model explaining the substrate-dependent kinetics of the reaction is discussed.
Copper amine oxidases (CuAOs; EC 18.104.22.168) are widespread in nature, being present in both prokaryotic and eukaryotic organisms. They are homodimers, each subunit containing a copper and a redox cofactor, 2,4,5-trihydroxyphenylalanine quinone (TPQ) . CuAOs catalyse the oxidative deamination of primary amines to the corresponding aldehydes, according to the overall reaction:
Catalysis occurs by a ping-pong mechanism, in which the amine is converted to the product aldehyde while reducing the enzyme cofactor (reductive half-reaction); this is followed by reoxidation of the cofactor by oxygen, which completes the catalytic cycle (oxidative half-reaction) .
To date, several amine oxidase crystal structures have been solved [3–10]. The structures for Escherichia coli (ECAO) , Pisum sativum (PSAO) , Arthrobacter globiformis (AGAO) , Hansenula polimorpha (HPAO) , Pichia pastoris , bovine serum amine oxidase (BSAO)  and human semicarbazide sensitive amine oxidase [9,10] reveal the similarity of the overall fold of these enzymes from various sources and point to the importance of the channel involved in amine substrate binding. The domain including the catalytic region (called D4) exhibits a rather high sequence similarity. All these features may implicate a similar catalytic mechanism for all members of the CuAO family. The pathway for the reductive half-reaction has been extensively studied, particularly for the ECAO [11,12], the HPAO  and the BSAO [14,15]. The fundamental reaction steps appear to be similar for enzymes from various sources [2,15,16]. In particular, C-H cleavage of the intermediate Schiff base generated upon amine substrate binding to TPQ appears to be a crucial step in the mechanism. Dependence of the kinetic isotope effect on pH , site-specific mutations at the active centre [17–19] and the crystal structure of ECAO in complex with a covalently bound inhibitor , indicate that a fully conserved aspartate residue (Asp300 for PSAO) serves to abstract the proton from the Schiff base. This residue also plays a role in ensuring the correct orientation of the cofactor during catalytic turnover . In the course of nucleophilic attack, the TPQ ring must be oriented with O5 pointing towards the general base , called the ‘productive’ conformation. A detailed theoretical study of the reductive half-reaction of PSAO suggests the possible role of Lys296, located near TPQ, as a proton donor . However, the conversion of amine to aldehyde groups involves several proton transfer steps and more than one proton donor or acceptor residue is involved in catalysis. In spite of their structural similarities, the substrate specificities of CuAOs vary among enzymes from different sources. In fact, the best substrates for different CuAOs have different structure and charge distribution, indicating that substrate-specific interactions govern substrate binding. The molecular nature of the substrate entry channel controls substrate binding and subsequent catalysis for two HPAOs . Electrostatic or hydrophobic forces have been suggested to drive polyamines (spermine and spermidine) and long-chain diamines, respectively, into the BSAO active site . In contrast to the reductive half-cycle, the oxidative half-cycle is a matter of debate, and a reaction pathway has been proposed for plant enzymes, which differs in some steps from that for mammalian or bacterial enzymes. The oxidative half-cycle of plant enzymes is not rate-limiting [25,26] and a semiquinone state may be involved in the catalytic cycle [26,27]. Conversely, in the catalytic cycle of BSAO and HPAO, the one-electron reduction of dioxygen is partially rate-limiting and involves electron transfer from reduced TPQ to produce superoxide anion, a reaction intermediate .
In this work, we examined the structural factors underlying substrate specificity and catalytic rates in copper amine oxidase from pea seedlings. We compared the kinetic behaviour of PSAO with two substrates – the diamine cadaverine (CAD) and the monoamine hexylamine (HEX) (structures shown in Fig. 1) – that are very different regarding affinities (Km) and catalytic constants (kc). The results from integration of kinetic studies with docking studies, and computation of the pKa values of the titratable residues of the active site, suggest that the formation of the enzyme–substrate complex is precisely regulated by specific interactions. In particular, the binding of HEX to the active site is controlled by hydrophobic contacts, whereas the approach of CAD is facilitated by electrostatic factors, primarily dependent on residues Glu359 and Glu412. The difference in the binding mode of the two substrates may modulate the participation of the residues surrounding TPQ in crucial catalytic steps. In this regard, we confirmed the role of Asp300 as a general base in the rate-limiting step of the reaction and indicated Lys296 as playing a substrate-dependent role in the prototropic shift accompanying cleavage of the Cα-H bond. With the predicted pKa values, a new mechanistic model is proposed which can satisfactorily explain substrate-dependent variations in the kinetic data.
A steady-state approach was followed to obtain the kinetic parameters of PSAO with CAD and HEX as substrates. In particular, the dependence of kinetic constants on pH, ionic strength and temperature was studied to elucidate the electrostatic factors that affect substrate specificity. Steady-state kinetic experiments were performed in air-equilibrated solutions at 27 °C. In these conditions, the rate cannot be affected by the co-substrate O2 because the saturation level for O2 has been reached. In fact, the concentration of O2 was about 0.25 mm  and the Km(O2) values were much lower: the Km(O2) was calculated as 17 ± 5 μm when CAD was used as the saturating amine substate and the Km(O2) was calculated to be lower than 2 μm when HEX was used as the saturating amine, at pH 7.2 (Fig. S1). These low Km(O2) values match those obtained with putrescine and benzylamine as substrates, respectively . In addition, according to the rate constant values for the individual steps reported by Padiglia et al.  for lentil seedling CuAO and to the kinetic isotope effect reported for PSAO , the oxidative half-cycle is not rate-limiting in air-saturated solutions and the reductive half-cycle is monitored.
Effect of pH
The dependence of kc, log(1/Km) and log(kc/Km) values on pH are shown in Fig. 1. In the pH range explored, the kc values of CAD are always higher than those of HEX, but become similar at a pH of > 9.5 (Fig 1A). For both substrates, kc profiles appear bell-shaped with the peaks centred at pH values of ∼ 7.2 and 9.3, respectively. Regarding the dependence of 1/Km on pH, a bell-shaped curve with maximum values around pH 8.2 was found in the case of CAD, whereas the HEX Km value was independent of pH, within experimental error (Fig. 1B). The plots of log (kc/Km) versus pH are bell-shaped profiles, with the maximum values centred at about pH 9 and 8 for CAD and HEX, respectively (Fig. 1C).
Effect of ionic strength
The effect of the ionic environment on the kinetics of the catalyzed oxidation of CAD and HEX was measured by varying the ionic strength (I) in the range 20–220 mm, at pH 7.20. Assuming that, for the enzyme–substrate system under investigation, 1/Km is the equilibrium dissociation constant (this point will be discussed later), the electrostatic effects in PSAO catalysis were studied by varying the ionic strength, and the data were analysed according to the Debey-Huckel theory applied to both 1/Km and kc . The plots of log(kc) and log(1/Km) versus (I)1/2 were straight lines, which were fitted to the following equation:
where k0 is a kinetic constant or the equilibrium dissociation constant at I = 0, and zA and zB are the overall electrostatic charges of the interacting ionic species (the substrate and the active site).
Constant C is ∼ 0.5 in water at 300 K . Values for the (2*C*za*zb) term have been derived for both substrates and are listed in Table 1. As this term was found to be close to zero for HEX, the effect of pH on (2*C*za*zb) was tested for CAD only (see Table 1). The data for HEX suggested that both binding (1/Km, see below for a detailed description) and chemical (kc) steps of the catalysis are not controlled by ionic interactions. Conversely, in the case of CAD, a slope [2*C*za*zb = −3] from log(1/Km) data was obtained. The overall CAD charge (za) sensed by the environment is reported to be za∼ 1.3 at pH 7.20  and this value is expected to be independent of pH up to about pH 9, where the amino groups can be titrated. We may thus argue that the positively charged substrate senses an overall charge by about −2 when it binds into the active site (before the chemical step). At pH 6, the total negative charge of the active site is reduced (perhaps one negatively charged group is protonated) and the slope of the plot of log(1/Km) versus (I)1/2 decreases from about −3 to about −2. The 2*C*za*zb of ∼ −1.6 found for the CAD substrate from log(kc) data in the range of pH explored (pH 6.0–9.2) reflects the fact that the chemical steps of the reaction are affected by electrostatic interactions between the negative charges of the enzyme and the positive charge of the substrate. The amino group in the substrate tail, which is positively charged in the pH range explored, may facilitate the correct positioning of the tail and anchor the substrate at the beginning of the catalytic cycle.
Table 1. Effect of ionic strength on kc and 1/Km at various pH values. Experimental values of kc and 1/Km versus (I)1/2 were fitted to Eqn (1) and values of the linear coefficient (2*C*za*zb) are reported. ND, not determined owing to the low Km value.
From log(1/Km) data
From log(kc) data
−2.9 ± 0.2
−1.5 ± 0.3
−1.8 ± 0.4
−1.7 ± 0.1
−1.7 ± 0.1
0.2 ± 0.8
−0.1 ± 0.2
Effect of temperature
The dependence of kc and 1/Km on temperature, measured in the range of 290–320 K at pH 7.20, 150 mm ionic strength, with CAD and HEX as substrates, indicates that these kinetic parameters increase with an increased temperature, with the exception of the 1/Km value of CAD, which is independent of the temperature. According to the steady-state approach of Briggs and Haldane, kc is included in Km (Km = (k−1 + kc)/k1): hence, the independence of the Km of HEX from pH and that of the Km of CAD from temperature, and the strong dependence of kc values of both substrates on pH and temperature, suggests that kc << k−1, which leads to Km ≅ (k−1/k1), that is, to the enzyme–substrate dissociation constant. The deuterium kinetic isotope effects investigated by Mukherjee et al.  are consistent with this hypothesis. These authors observed a strong kinetic isotope effect on both kc and kc/Km with putrescine and benzylamine as substrates of PSAO (conversely, if Km contains kc, the kinetic isotope effect on kc/Km should vanish). According to this hypothesis, from the dependence on temperature (T) of Km, we calculated the ΔH and ΔS accompanying the binding of substrate to enzyme according to the van’t Hoff equation. ΔH* and ΔS*, the enthalpy and entropy of activation accompanying the formation of the activated complex, were calculated from the dependence of kc on T according to the transition state theory, and the resulting values are listed in Table 2. ΔH* increases with decreasing pH, whereas -TΔS* decreases with decreasing pH, as shown by the ΔH* and -TΔS* values plotted as a function of the pH (Fig. 2). Accordingly, the energy cost of the heterolytic cleavage of the Cα-H bond is greater at higher H+ concentrations, although entropy changes become less unfavourable. Interestingly, the ΔH* values of CAD and HEX show better agreement in the high-pH range, where the neutral forms of these compounds predominate.
Table 2. Thermodynamic parameters of the reductive half-reaction of CAD and HEX by PSAO. Experiments were performed at pH 7.2 and 150 mm ionic strength by varying the temperature in the range 290–320 K. Values of the activation enthalpy (ΔH*) and the activation entropy (ΔS*) were calculated by fitting data of kc at various T to Eqn (12). Values of enthalpy (ΔH) and entropy (ΔS) change were obtained by fitting data of 1/Km at various temperatures to Eqn (13).
ΔH* (from kc)
ΔH (from 1/Km)
ΔS* (from kc)
ΔS (from 1/Km)
9.1 ± 0.2
0.1 ± 0.1
10.9 ± 0.5
2.7 ± 0.4
Modelling of substrate–PSAO interactions
Docking CAD into the active site revealed that the head amino group is located at the bottom of the narrow channel and always forms hydrogen bonds with O5 of TPQ and the carboxylic group of Asp300. There are three stable conformations for CAD (Fig 3A–C). In the first stable conformation showing the lowest energy (−13.4 kcal·mol−1) (Fig. 3A), the charged tail amino group is in contact with polar or negatively charged residues (Glu412 and Asn386, located at the bottom of the channel near TPQ). In the second stable conformation (−13.0 kcal·mol−1), the amino group is located in a polar pocket composed of Gln108, Ser138 and Ser139 (Fig. 3B). A third stable conformation (−11.8 kcal·mol−1) (Fig. 3C), albeit energetically less favoured, shows the CAD tail close to Ser138 and Tyr168, which is hydrogen-bonded to Glu359 of the other subunit; the hydroxyl group of Tyr168 can form a hydrogen bond with both Glu359 and the substrate. In all three conformations, the charged side-chain of Lys296 is stabilized by forming a salt bridge with Glu412, and the dihedral angle χ2 of Phe298 is about −85°, whereas it is about −30° in the original crystal structure.
The docking simulation of HEX finds two stable conformations for this substrate with similar binding energy. In one conformation (−10.8 kcal·mol−1; Fig. 3D), the head amino group of HEX is located between TPQ and Asp300, like CAD, and the dihedral angle χ2 of Phe298 is about −85°. The other conformation (−10.9 kcal·mol−1; Fig. 3E) shows the amino group far from Asp300, close to the other side of the TPQ ring, forming a salt bridge with Glu412 and the O4 of TPQ, and a hydrogen bond with Asn386. Unlike the first conformation, the dihedral angle χ2 of Phe298 is about −30°. In both conformations the uncharged side-chain of Lys296 is hydrogen bonded with the O4 of TPQ.
To determine the charged-state of residues involved in substrate binding or in catalytic steps, we computed the pKa of the titratable residues in the presence of the substrate. For the general base candidate Asp300, a pKa value of 8.7 was obtained for the free enzyme, decreasing to 6.6 with CAD or HEX bound at the active site. As noted previously for the free enzyme , the large pKa shift of this residue in PSAO is caused by the highly hydrophobic microenvironment at the enzyme active site. The calculated pKa of 6.6 for Asp300 in the active site with substrate bound match the suggestions for BSAO in an early work by Klinman et al. . Similar results for the pKa value of the catalytic aspartate with the substrate or inhibitor in the active site have also been obtained for ECAO [17,18] and AGAO . In addition, the carboxylic group of Asp300 forms a hydrogen bond with a TPQ carbonyl in the crystal structure of PSAO, suggesting the presence of the protonated form at pH 4.8, the pH of crystallization. The calculated pKa of Lys296 (pKa= 8.3) is reduced by more than two pH units compared with its value in water. The pKa values obtained for Glu359 and Glu412 (7.3 and 5.2, respectively), suggest that these residues change their protonation states in the pH range explored (i.e. by electrostatic interactions, they may interfere with the binding of charged substrates). Lastly, a pKa of ∼ 11 was found for Tyr286. It is difficult to assess the error range of pKa values because they have not been experimentally determined in the enzyme. Hence, we estimate the error range, based on the uncertainty of the method, as 0.5 pKa units [34,35].
pKa calculations were also performed with HEX bound at the active site and the values obtained were very similar to those with CAD and those reported above.
The above results for PSAO indicate that the binding of CAD, a substrate which bears one positive charge on the head and one on the tail, occurs with maximum efficiency (highest kc/Km and lowest Km values) at a pH of about 8. According to ionic strength dependence, binding appears to be driven by the electrostatic interactions occurring between CAD and polar or negatively charged residues located close to the active site. Based on the modelled structure with CAD bound, Glu359 and Glu412 favour stable conformations of the enzyme–substrate complex when negatively charged. The independence of Km of the only head charge-bearing HEX on ionic strength indicates the lack of charge–charge interactions of this substrate. These observations, together with the negligible variations of Km on pH, and the positive values of ΔH and ΔS, all suggest that binding of the HEX substrate is primarily driven by hydrophobic interactions . The high and positive values of ΔS of both CAD and HEX (+22 and +24 cal mol−1·K−1), calculated from the temperature dependence of the Km, suggest that substrate binding is accompanied by the release of water molecules. Concerning the chemical steps, the bell-shaped profile of the kc of CAD and HEX versus pH (Fig. 1B) indicates the involvement of at least two acid–base couples, (B1-H+/B1) and (B2-H+/B2), in the rate-determining step, like the two-protonation state model of Tipton and Dixon . Fig. 1B shows that the pKa values of B1 and B2 are substrate dependent; alternatively, and more probably, different residues behave as B1 and B2, depending on the structure of the interacting substrate.
According to the above results and the fundamental steps of the reaction described in the literature, shown in Fig. 4A, we propose a kinetic model (for details see Doc. S1), in which the only charged forms of CAD are considered as reactive species, because the charged amine groups favour interactions with the active site. Hence, we included [S]R = [SH+] + [SH22+] for CAD. Conversely, ([S]R = [S] + [SH+]) was considered for HEX because hydrophobic interactions with the active site prevail. However, we assumed that, in both substrates, the attacking amino group was neutral at the beginning of the catalytic cycle so that the nucleophilic attack on TPQ could take place . During the substrate entry (not explicitly shown in the simplified scheme of Fig. 4A) the CAD tail is addressed towards stable enzyme–complex conformations by two negatively charged residues (Glu359 and Glu412) and by polar residues (Asn386, Ser138 and Tyr168), as shown in Fig. 3. The two charged residues are titrated in the pH range explored and facilitate interaction between enzyme and substrate. The kinetic rate constant, k1, of the recognition step, which leads to the formation of the enzyme–substrate complex (before the chemical events), may be written as:
where the energy of the electrostatic interaction (δ1 and δ2) of the substrate with two titratable residues, D1 and D2 (probably Glu359 and Glu412), with ionization constants of KD1 and KD2, is explicitly reported. The terms KD1/(KD1 + [H+]) and KD2/(KD2 + [H+]) are weighting factors taking into account the molar fraction of D1 and D2 in the deprotonated state. In the case of HEX the ionization contributions to k1 vanish. includes all the other energy terms contributing to the kinetic constant (i.e. the contribution of the electrostatic interaction between substrate and polar residues and of the hydrophobic interaction).
The fundamental points of our approach describing the catalytic events are as follows.
1 The role played by proton-exchanging residues on the recognition step and on the chemical reaction is explicitly introduced both in Km by Eqn (2), showing residues D1 and D2, and in kc by assuming the presence of B1 and B2 residues.
2 Small differences in substrate structure produce a different enzyme complex, so that the enzyme residues involved in catalysis, in both recognition and reaction steps, are substrate dependent (see also Fig. 4A showing the fundamental steps of the reaction pathway). The position of the TPQ intermediate (the ketimine I±) is consequently modified.
This hypothesis was confirmed by docking computations, which indicated that deprotonated Lys296 points towards TPQ only in the stable conformations of the HEX–enzyme complex. Conversely, the charge–charge interaction facilitating the binding and positioning of CAD also helps to accelerate the chemical events leading to an increase in kc. The importance of this interaction is also supported by the similar behaviour of CAD and HEX above pH 9.5, when both substrates are present in their neutral form.
3 The heterolytic cleavage of the Cα-H bond of the amine is assumed to control the kc of the reductive half-step, as supported in the literature [2,23]. The concerted prototropic shift converting the Schiff base from the ketimine form (I±) to the aldimine form () is assisted by two acid–base couples: (B1-H+/B1) and (B2-H+/ B2), which interact simultaneously with the Schiff base (see Fig. 4B). From our data it appears that the identity of these residues is substrate dependent, which may account for the differences in the pH dependence of the kc values (Fig. 1A).
On the basis of the three points described above, and assuming that the deprotonation of the head amino group is not rate-limiting, the following equations were derived for CAD (the detailed kinetic model is reported in Doc. S1):
and for HEX, respectively:
where [S]0 and [E]0 are the total concentrations of the substrate and enzyme, respectively.
KB1 and KB2 are the ionization constants of the two general bases which control kc, and α and β are empirical constants representing the partial activity at extremal pH ; k1 is given by Eqn (2).
According to Eqns (3,4), the experimental data of Fig. 1 were fitted to the following equations (solid lines of Fig. 1):
The resulting δ1, δ2 and pKa values are listed in Table 3.
Table 3. Ionization constants and energy contributions from the pH profile of PSAO kinetic parameters. Column 3: pKa values and free-energy contributions (δ1 and δ2) were obtained by fitting experimental data of kc and Km or pseudo-first-order kc/Km constants as a function of pH according to the equations described in the Discussion. HEX: kc (Eqn 8) and kc/Km (Eqn 9) fitting were obtained leaving all unknown parameters (i.e. KB1 and KB2) floating. CAD: kc (Eqn 5) fitting was obtained leaving all unknown parameters floating; in the fitting of Km (Eqn 6) and kc/Km (Eqn 7), KD1, KD2, δ1 and δ2 terms were left to float but pKB1 = 6.66 and pKB2 = 8.30 were maintained fixed, as calculated from the kc data; pKs = 10 was also maintained fixed. Column 4: pKa values from the dependence on pH of Km (Eqn 10) and kc/Km (Eqn 11) using CAD as a substrate, according to Dixon’s model .
pKa according to the proposed model
pKa according to Dixon’s model
pKB1 = 6.66 ± 0.15
pKB2 = 8.30 ± 0.11
pKD1 = 5.37 ± 0.32
pKD1 = 5.59± 0.31
pKD2 = 6.90 ± 0.29
pKD2 = 7.23 ± 0.28
δ1 = −3.97 ± 0.79 kcal·mol−1
δ2 = −1.21 ± 0.46 kcal·mol−1
pKS = 9.95 ± 0.12
pKS = 10.04 ± 0.13
pKD1 = 5.22 ± 0.62
pKD2 = 6.16 ± 0.61
pKD2 = 7.28 ± 0.26
pKD2 = 7.14 ± 0.28
δ1 = −2.5 ± 1.50 kcal·mol−1
δ2 = −1.20 ± 0.28 kcal·mol−1
pKS = 10.0 (fixed)
pKS = 8.95 ± 0.34
pKB1 = 8.41 ± 0.17
pKB2 = 10.36 ± 0.23
pKB1 = 8.30 ± 0.18
pKB2 = 10.00 ± 0.35
Equation (8) is equivalent to the Tipton and Dixon equation for kc (according to their ‘Simplified reaction scheme’ ), where the α and β factors  may be included to obtain Eqn (5).
The Dixon’s models , which are usually utilized to predict pKa values from kinetic data, were used for comparison. In the case of CAD, the fit of Km and kc/Km were performed with a three pKa model, fitting a bell-shaped curve with an increase with two pKa values and a decrease with one pKa value.
A good match was found between the two sets of data, that is pKa values according to Dixon and to the model we are proposing. However, the models of Dixon do not estimate the contributions to the Gibbs energy of the recognition step due to D1 and D2 (δ1 and δ2).
The equation for (kc/Km)HEX (a two-pKa model), according to the approach of Dixon, is formally equivalent to Eqn (9).
In addition, from Table 3 it appears that the pKa values obtained by the experimental data are in good agreement also with the computed pKa values reported in the Modelling of substrate-PSAO interactions section. In particular, D1 could be Glu412 (computed pKa = 5.2) and D2 could be identified with Glu359 (computed pKa = 7.2).
The pKa values calculated from kc with the CAD substrate are also in accordance with those obtained by Pec et al.  for the similar, but more rigid, 1,4-diamino-2-butene substrate (pKa values of 6.9 and 8.1 were obtained from the fit of the kc data).
The structure of the catalytic site and the calculated and experimentally obtained pKa values identify Asp300 (pKa = 6.6) and Lys296 (pKa = 8.3) as catalytically important residues, with pKa values falling into the pH range delimiting the kc bells. Based on the kc versus pH profiles and on the docking studies, which show a Lys296 orientation that is substrate dependent (Lys296 forms a salt bridge with TPQ and with Glu412 when HEX or CAD, respectively, are in the active site), we proposed the role for Lys296 as a proton donor in the case of CAD and as a proton acceptor in the case of HEX. Asp300 is the proton acceptor candidate in the case of CAD. The position of Tyr286 indicates this residue as a possible candidate for donating a proton (Fig. 4B). Its role in proton transfer has already been suggested by Hevel et al.  on HPAO.
The results from Pietrangeli et al.  with two aliphatic amines (putrescine and spermidine) and four aromatic amines have been interpreted in terms of hydrophobic interactions prevailing over polar interactions in PSAO. Our results partially match those of these authors, in that the substrate contains a hydrophobic tail. However the tail amino group of CAD not only affects Km but increases, in orders of magnitude, kc at the optimum pH value. Consequently, the electrostatic-driven docking of CAD appears to be crucial for the substrate preference of PSAO. Conversely, if the electrostatic contribution is lacking, increased flexibility of the substrate Schiff base would be expected. A similar effect (although of hydrophobic rather than of electrostatic nature) was reported by Taki et al.  studying the stereo-selectivity of a bacterial amine oxidase.
In conclusion, in a combination of kinetic, structural and computational procedures, this study shows that the substrate-specific interactions underlying the selectivity of PSAO not only affect the binding mode of the amine in the active site, but also the identity of the residues recruited in the catalytic steps. In particular, the new role of Lys206 is proposed in the catalytic cycle. Because this Lys is a conserved residue in plant CuAOs and has been proposed to play a role in the formation of TPQsq upon oxidative deamination of its side-chain , future study of site-directed mutagenesis will be necessary to confirm our findings and to have a better understanding of the structural factors controlling substrate preferences and catalysis of CuAOs, enzymes with many still unknown physiological functions.
Enzyme purification and activity testing
All reagents were from Fluka (Milan, Italy). PSAO was purified from Pisum sativum seedlings according to Vianello et al. , reaching a final specific activity of 1.6 μkat·mg−1.
Initial-rate measurements were carried out by monitoring H2O2 production using a peroxidise–cytochrome c-coupled assay . Kinetic runs were performed at 27 °C, in various experimental conditions, particularly at variable amine substrate concentrations, pH values (range 5.20–10.20) and ionic strength (20–220 mm), equilibrated with air. Steady-state kinetic parameters (kc and Km) were calculated from nonlinear fitting of the reaction rate plots to the Michaelis–Menten equation using sigmaplot 2004, Version 9.01 (Systat Software Inc., Richmond, CA, USA). Michaelis–Menten behaviour was observed independently of substrate, pH and ionic strength.
Experiments were performed in solutions containing 25 mm buffer and 125 mm NaCl at various pH values. The buffers used were: sodium acetate (pH 5.2–5.6), Mes (pH 5.6–6.4), Mops (pH 6.61–7.03), Hepes (pH 8.00–8.65), sodium borate (pH 8.71–9.71) and sodium carbonate (pH 9.71–10.20). Kinetic measurements performed in these buffers at overlapping pH values gave identical results within the experimental error, excluding specific salt effects.
Experiments were performed at pH 7.20, in solution containing 25 mm Hepes at various ionic strengths (10–200 mm NaCl was added).
The heat of activation (ΔH*) and entropy (ΔS*) were obtained by measuring the effect of temperature on kc, according to the law:
where ΔH* is the heat of activation, ΔS* is the entropy of activation, kB is the Boltzmann constant, h is the Plank constant, R is the gas constant and κ is the transmission coefficient. As κ is usually close to unity  this equation simplifies into:
The changes in enthalpy (ΔH) and entropy (ΔS) of the binding process were obtained by measuring the effect of temperature on Km, according to the equation:
assuming that 1/Km values are the association constants of the enzyme–substrate complex and ΔH and ΔS are the thermodynamic parameters of enzyme–substrate complex formation.
The constants kc and Km at various temperatures were calculated from Michaelis–Menten plots obtained in the range 290–320 K.
We studied the binding modes of CAD and HEX by means of docking simulation in the PSAO active site. The crystal structure of free PSAO with the Protein Data Bank code 1KSI  was used as a starting model for all calculations. In this structure the TPQ ring adopts a nonproductive conformation (i.e. O2 of TPQ points towards Asp300 and O5 points towards the copper ion cofactor) . Hence, to generate an appropriate model for the reaction, the TPQ ring was rotated by 180°.
As the two subunits in PSAO operate simultaneously, but not cooperatively , substrate docking was simulated only in subunit A. AutoDockTools version 1.5.2 (the Scripps Research Institute, La Jolla, CA, USA) was used to add polar hydrogens to the PSAO crystal structure and to assign Gasteiger charges to the atoms, with the exception of TPQ, the charges of which were calculated using the petra web server [ http://www2.chemie.uni-erlangen.de/software/petra]. autodock 4 software was used to perform docking simulations, employing the Lamarckian genetic algorithm . Default settings were used for docking parameters. Other details are available in Doc. S2.
As previously described , the active site of PSAO is extremely hydrophobic, and therefore in order to account properly for the pKa shift of titratable residues at the catalytic centre, a microenvironment-dependent method had to be applied. Hence, the pKa values of titratable active-site residues in the presence of various substrates were calculated using the screened Coulomb potential method, with microenvironment-dependent dielectric screening functions [34,35]. (Other details can be found in Doc. S2.)
This work was partly funded by Istituto Nazionale Biostrutture Biosistemi (Rome, Italy) and by Hungarian Research Fund (OTKA) K72579, M.F for Bolyai János fellowship.