Enzymatic oxidation of NADP + to its 4-oxo derivative is a side-reaction displayed only by the adrenodoxin reductase type of ferredoxin-NADP + reductases Matteo de Rosa 1 , Andrea Pennati 1 , Vittorio Pandini 1 , Enrico Monzani 2 , Giuliana Zanetti 1 and Alessandro Aliverti 1 1 Dipartimento di Scienze Biomolecolari e Biotecnologie, Universita ` degli Studi di Milano, Italy 2 Dipartimento di Chimica Generale, Universita ` degli Studi di Pavia, Italy Ferredoxin-NADP + reductases (FNRs, EC 1.18.1.2) can be classified into two phylogenetically distinct subgroups: the mitochondrial-type and the plastid-type (or plant-type) enzymes [1]. The prototype of the former enzymes is the mammalian adrenodoxin reduc- tase (AdR), whereas the latter subgroup is best exemplified by the photosynthetic FNR. Although both FNR types catalyze the same physiologic reaction, i.e. Keywords adrenodoxin reductase; 3-carboxamide-4- pyridone adenine dinucleotide phosphate; flavoprotein; Mycobacterium tuberculosis; NADP derivative Correspondence A. Aliverti, Dipartimento di Scienze Biomolecolari e Biotecnologie, via Celoria 26, 20133 Milano, Italy Fax: +39 02 50314895 Tel: +39 02 50314897 E-mail: alessandro.aliverti@unimi.it Website: http://www.sbb.unimi.it/index.htm (Received 19 March 2007, revised 7 May 2007, accepted 11 June 2007) doi:10.1111/j.1742-4658.2007.05934.x We have previously shown that Mycobacterium tuberculosis FprA, an NADPH-ferredoxin reductase homologous to mammalian adrenodoxin reductase, promotes the oxidation of NADP + to its 4-oxo derivative 3-car- boxamide-4-pyridone adenine dinucleotide phosphate [Bossi RT, Aliverti A, Raimondi D, Fischer F, Zanetti G, Ferrari D, Tahallah N, Maier CS, Heck AJ, Rizzi M et al. (2002) Biochemistry 41, 8807–8818]. Here, we pro- vide a detailed study of this unusual enzyme reaction, showing that it occurs at a very slow rate (0.14 h )1 ), requires the participation of the enzyme-bound FAD, and is regiospecific in affecting only the C4 of the NADP nicotinamide ring. By protein engineering, we excluded the involve- ment in catalysis of residues Glu214 and His57, previously suggested to be implicated on the basis of their localization in the three-dimensional struc- ture of the enzyme. Our results substantiate a catalytic mechanism for 3-carboxamide-4-pyridone adenine dinucleotide phosphate formation in which the initial and rate-determining step is the nucleophilic attack of the nicotinamide moiety by an active site water molecule. Whereas plant-type ferredoxin reductases were unable to oxidize NADP + , the mammalian adrenodoxin reductase also catalyzed this unusual reaction. Thus, the 3-carboxamide-4-pyridone adenine dinucleotide phosphate formation reac- tion seems to be a peculiar feature of the mitochondrial type of ferredoxin reductases, possibly reflecting conserved properties of their active sites. Furthermore, we showed that 3-carboxamide-4-pyridone adenine dinucleo- tide phosphate is good ligand and a competitive inhibitor of various dehy- drogenases, making this nucleotide analog a useful tool for the characterization of the cosubstrate-binding site of NADPH-dependent enzymes. Abbreviations AdR, adrenodoxin reductase; Amplex Red, 10-acetyl-3,7-dihydroxyphenoxazine; amu, atomic mass units; CT1, charge transfer complex between FprA and NADPH; FNR, ferredoxin-NADP + reductase; INT, iodo-nitro-tetrazolium chloride; NADPO, 3-carboxamide-4-pyridone adenine dinucleotide phosphate; NMNO, 3-carboxamide-4-pyridone mononucleotide; 2P-AMP, 2¢-phospho-AMP. 3998 FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS the transfer of a couple of electrons from NADPH to two successive ferredoxin molecules, they mark- edly differ in their structure and functional properties [2,3]. In the recent past, we have obtained the crystal structure of FprA, a Mycobacterium tuberculosis homolog of AdR (41% identity spread over the entire polypeptide length) at a very high (1.05 A ˚ ) resolution [4]. This accomplishment, together with the fact that the structures of AdR and FprA are very similar, made the bacterial protein an ideal representative of mitochondrial-type FNRs for structure–function rela- tionship studies. The atomic resolution of the FprA structure allowed us to discover that this enzyme, in addition to the well-known NADPH-dependent ferre- doxin reduction, catalyzed an unprecedented reaction, i.e. the oxidation of NADP + to yield its 4-oxo deriv- ative, which was named 3-carboxamide-4-pyridone adenine dinucleotide phosphate (NADPO). The first evidence that such a reaction was occurring was the observation that crystals of FprA grown in the pres- ence of NADP + contained NADPO bound to the act- ive site instead of NADP + [4]. Comparison of the FprA–NADPO and FprA–NADPH structures clearly showed that in the latter complex an additional ordered water molecule (water 1) was sitting in the act- ive site in a position very close to that occupied by the carbonyl oxygen atom of NADPO in the former complex [4]. This observation prompted us to propose the hypothetical mechanism for the FprA-catalyzed NADPO formation reaction depicted in Fig. 1. NADP + oxidation is initiated by water 1 addition to the electron-poor C4 atom of the nicotinamide ring of NADP + . His57 and Glu214 have been previously pro- posed to increase the nucleophilicity of the water mole- cule by favoring its deprotonation [4]. Both residues, highly conserved in AdR-like enzymes, have been recently changed to nonionizable ones by site-directed mutagenesis [5]. Characterization of the resulting FprA variants allowed us to conclude that His57 but not Glu214 played a significant role in the physiologic, NADPH-dependent activity of FprA. However, crys- tals of FprA-H57Q grown in the presence of NADP + again displayed NADPO as the bound nucleotide by X-ray diffractometry [5], showing that the H57Q muta- tion did not abolish the NADP + oxidation activity of FprA. The previous studies summarized above only gave a qualitative evidence of the ability of FprA to catalyze the production of NADPO. In the present article, we provide the first quantitative description of this reac- tion, along with a detailed analysis of the spectroscopic properties of the product of NADP + oxidation. Results and Discussion NADPO isolation, quantitation, and spectral characterization In order to study the kinetics of NADP + oxidation to NADPO catalyzed by FprA, an NADPO assay method was required. After testing different analytical chromatographic procedures, we found the ion exchange method described by Orr & Blanchard to be particularly reliable and robust for our purposes [6]. Inclusion of AMP as an internal standard increased precision and accuracy, and replacement of the ori- ginal mobile phase with a volatile ammonium formate buffer allowed the recovery of purified nucleotide as salt-free preparations after vacuum drying. Figure 2 shows the typical analysis of aliquots sampled at increasing incubation times from a reaction mixture where FprA was incubated with NADP + in air. As Fig. 1. Hypothetical mechanism of the reductive half-reaction of the catalytic cycle of FprA in the oxidation of NADP + to yield NADPO. The reaction scheme was based on the crystal structures of the FprA–NADPO and FprA–NADPH complexes [4]. The depicted water molecule is referred to in the text as water 1. B1 and B2 rep- resent hypothetical groups acting as base catalysts. B1 has been proposed to be the imidazyl of His57 [4]. M. de Rosa et al. Oxidation of NADP + to its 4-oxo derivative FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 3999 can be seen, incubation over several hours resulted in a gradual decrease of the NADP + peak while a single new peak appeared and progressively increased in intensity. By tandem MS, the latter peak was found to unambiguously contain an NADP derivative bearing an oxygen atom bound to the nicotinamide ring. Previ- ously, the same NADP + modification was observed by MS analysis of the whole reaction mixture [4]. How- ever, even though only 4-oxo-NADP was observed by X-ray crystallography in complex with FprA, in princi- ple it cannot be excluded that FprA could also intro- duce an oxygen atom at other positions of the nicotinamide ring of NADP + , leading to other nucleo- tide derivatives with a lower affinity for the enzyme active site. Thus, it was of interest to ascertain the identity of the isolated compound by a detailed spect- roscopic characterization. Values of 17 100 and 15 600 m )1 Æcm )1 were found for the extinction coeffi- cient of the NADP derivative at 260 and 254 nm, respectively, as calculated by determining the concen- tration of the nucleotide on the basis of the phosphate released by alkaline phosphatase treatment. The absorbance spectrum of the nucleotide is shown in Fig. 3A in comparison to those of NADP + and NADPH. A peculiar feature of the modified NADP + is a relatively high absorption in the 280–310 nm region. To study the effect of pH on its spectral prop- erties, the adenylate portion of the modified nucleotide was removed by enzymatically splitting the pyrophos- phate link in order to get rid of its large absorbance contribution. As shown in Fig. 3B, the spectrum of chromatographically purified NMN derivative under- goes a large spectral transition when the pH is decreased from 7.7 to 1.0. The two spectra are very similar to those described for N -methyl-4-pyridone- 5-carboxamide, and completely different from those of N-methyl-2-pyridone-5-carboxamide, under similar pH conditions [7]. Furthermore, the isolated NADP deriv- ative was found to lack any fluorescence when excited with light in the UV region. This observation is rele- vant for excluding the presence of species modified in position 6 of the nicotinamide ring, because, unlike the compounds with the oxo group in positions 2 and 4, N-methyl-6-pyridone-5-carboxamide has been shown to emit blue light by fluorescence [8]. We exclude the formation of the species modified in position 3 of the nicotinamide ring because of the poor reactivity of this position towards nucleophilic attack. Indeed, the posit- ive charge of the pyridinium moiety favors attack by nucleophiles at positions 2 and 4 under mild condi- tions [9]. Furthermore, and more importantly, the modification at position 3 can be excluded because the resulting N-substituted 3-oxo-nicotinamide moiety would not be neutral, resulting in a dinucleotide with spectral and chromatographic properties expected to be markedly different from those of 4-oxo-NADP. These data allow us to conclude that FprA does not Fig. 2. Anion exchange chromatographic pattern of NADP + oxida- tion reaction mixtures. NADP + at 500 lM was reacted with O 2 at its air equilibrium concentration (c. 250 l M) in the presence of 100 l M FprA at 25 °C. Forty-microliter aliquots were analyzed by high-performance chromatography as described in Experimental procedures. Reaction times were 0 h (dotted trace), 3 h (dashed trace), 6 h (dot-dashed trace), and 24 h (continuous trace). Fig. 3. Extinction coefficients and absorption spectra of NADPO and NMNO. (A) Dinucleotides in 20 m M Tris ⁄ HCl (pH 7.7). (B) NMNO in 20 m M Tris ⁄ HCl (pH 7.7), and after bringing the pH to 1 by the addition of HCl. Oxidation of NADP + to its 4-oxo derivative M. de Rosa et al. 4000 FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS produce detectable amounts of NADP derivatives bearing a carbonylic oxygen at sites other than posi- tion 4 of the nicotinamide ring, indicating that the NADP + oxidation reaction is highly regiospecific. Kinetics of NADPO formation as catalyzed by FprA and AdR Figure 4A shows the time courses of NADP + oxida- tion to NADPO catalyzed by FprA or AdR in the presence of air oxygen. Clearly, both enzymes are able to catalyze this reaction. Studies at various NADP + concentrations were performed, showing that the sub- strate concentration of 500 lm was fully saturating. The initial rate of NADPO formation promoted by AdR was slightly lower (75%) than that of FprA under the same conditions. A peculiar feature of the NADP + oxidation kinetics is the progressive decrease in the reaction rate. When FprA was the catalyst, this behavior was particularly marked: NADPO formation sharply decreased after the first enzyme turnover. We excluded the possibility that this was a consequence of enzyme denaturation by assaying the enzyme during incubation. A reasonable explanation is that NADPO was acting as a competitive inhibitor of the enzyme with respect to NADP + (see below). In the case of FprA, the reaction was studied at different enzyme concentrations. As shown in Fig. 4B, the initial rate of the reaction was proportional to FprA concentration, showing that the NADP + oxidation, although very slow, was strictly enzyme-dependent. The NADP + oxidation reaction catalyzed by FprA was much slower (0.14 h )1 ) than that of its NADPH-dependent, physiologic reaction (336 min )1 when Mycobacterium smegmatis FdxA was used as electron acceptor [10]). Plant-type FNRs do not catalyze NADPO formation To verify whether NADP + oxidation to yield NADPO was a common feature of the members of the FNR class of enzymes, we assayed Toxoplasma gondii FNR [11], Plasmodium falciparum FNR [12] and spinach (Spinacia oleracea) leaf FNR [13] (which all are plant- type FNRs) for their ability to catalyze this reaction, and found them to be completely inactive. Although FNRs from other sources should be assayed before drawing general conclusions, it is suggested that NADP + oxidation to NADPO most probably repre- sents a unique feature of AdR-type FNRs and reflects very specific organization and reactivity of their active sites. Role of O 2 in enzyme-catalyzed NADP + oxidation When the O 2 -dependent NADPO production reaction catalyzed by FprA was allowed to proceed in the pres- ence of excess peroxidase and its fluorogenic substrate 10-acetyl-3,7-dihydroxyphenoxazine (Amplex Red), a progressive increase in fluorescence emission at 585 nm was observed (not shown), indicating accumulation of the fluorescent product resorufin. No fluorescence built up in the absence of peroxidase, allowing us to attrib- ute resorufin formation entirely to the 1 : 1 reaction between H 2 O 2 and Amplex Red catalyzed by the per- oxidase. This conclusion was confirmed by the effect of the presence of catalase in the reaction mixture, which completely abolished resorufin formation. On the basis of the quantum yield experimentally deter- mined for resorufin, a rate of 0.08 mol H 2 O 2 (mol FprA) )1 was calculated, a value comparable to the rate of NADPO formation. In addition to H 2 O 2 ,a small amount of superoxide was apparently produced in the reaction, as judged from the slight increase in the rate of fluorescence appearance after superoxide dismutase addition. However, it could not be excluded that the peroxide ⁄ superoxide ratio was substantially lower than that found, as the low rate of the reaction would allow enough time for spontaneous disproportio- nation of O 2 – to H 2 O 2 and O 2 . In any case, the produc- tion of H 2 O 2 and ⁄ or O 2 – strongly supports the FAD prosthetic group as the direct oxidant of the nicotina- mide ring, as these species are the usual products of FADH 2 reoxidation by O 2 in most flavoproteins. To verify whether the direct reaction between NADP + and O 2 was not required in the enzymatic production of NADPO, the latter reaction was studied in the presence of electron acceptors different from O 2 . NADP + was incubated with FprA under anaerobic Fig. 4. (A) Time-courses of NADPO production catalyzed by FprA and AdR in the presence of O 2 . The reaction was carried out at 25 °C in the presence of 100 l M FprA (filled circles) or bovine AdR (open circles) and 500 l M NADP + . A control experiment was per- formed, omitting the enzyme (squares). (B) Initial rate of NADPO production calculated over the first 6 h of the reaction plotted as a function of the enzyme concentration. M. de Rosa et al. Oxidation of NADP + to its 4-oxo derivative FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 4001 conditions in the presence or in the absence of K 3 Fe(CN) 6 , an artificial electron acceptor of the enzyme. As shown in Fig. 5A, as NADP + was added to the reaction mixture, a progressive bleaching of the absorbance contributed by ferricyanide was observed, indicating its reduction to ferrocyanide. The spectrum of enzyme-bound FAD remained unaltered, until all the ferricyanide was consumed. Starting from this point, the FAD spectrum undergoes a progressive per- turbation (not shown), similar to that observed by incubating FprA with NADP + in the absence of ferri- cyanide (Fig. 5B). The ability of FprA to carry on NADP + oxidation using either O 2 or ferricyanide sup- ports our previous proposal, made on the basis of structural data [4], that the mechanism of NADPO formation can be split in two half-reactions: the first leads to NADPO formation coupled to FAD reduc- tion; and the second consists of the reoxidation of FADH 2 by O 2 or other oxidants (Fig. 1). With the aim of trapping possible intermediates produced in the first half-reaction, the reaction between NADP + and FprA was studied in the absence of any oxidant, thus preventing the second, oxidative, half-reaction of the catalytic cycle. During incubation, the A 340 of the mix- ture progressively increased, and the visible absorbance spectrum of the enzyme-bound FAD underwent a pro- gressive partial bleaching, leading to a final stable spectrum strongly reminiscent of that of the charge transfer complex between NADPH and oxidized FprA [10]. The time course of the FAD spectral change approximately followed a single exponential decay equation (inset of Fig. 5B) with a first-order rate con- stant of 0.095 ± 0.009 h )1 , a value similar to the rate of NADPO formation measured under aerobic condi- tions. The enzyme present in the endpoint reaction mixture was denatured and precipitated before admit- ting air into the anaerobic cuvette, and the enzyme-free solution was analyzed by anion exchange chromato- graphy as previously described. The sample was found to contain both NADPO and NADPH. Thus, in the absence of oxidants, FprA catalyzes the disproportio- nation of NADP + according to the following equa- tion: 2NADP þ þ OH À ! NADPO þ NADPH þ H þ ð1Þ As FprA has a single binding site for NADPH [4,5], the reaction must proceed through a ping-pong mech- anism, with the transferred electron couple being tran- siently stored on the enzyme prosthetic group. Thus, experimental evidence points to the involvement of FAD in the mechanism of NADPO formation. The observed spectral changes can be interpreted as result- ing from the following reaction mechanism, where E(FAD) and E(FADH – ) represent the oxidized and fully reduced form of FprA, respectively: E(FAD) þ NADP þ ! E(FAD)ÀNADP þ ð2Þ E(FAD)ÀNADP þ þ OH À ! E(FADH À ÞÀNADPO þ H þ ð3Þ NADP þ þ E(FADH À ÞÀNADPO ! NADPO þ E(FAD)ÀNADPH (CT1) ð4Þ In the reaction shown in Eqn (4), the FprA– NADPH complex represents a charge transfer species (CT1) [10]. The reactions shown in Eqn (2) and Eqn (4) are expected to be much faster than that shown in Eqn (3), as dinucleotide binding and release to and from FprA have never been found to be limiting steps in reactions of the enzyme with NADPH or NADH, which occurred at rates much higher than that of NADPO formation [5]. Thus, the only process access- ible to experimental observation was the conversion of the E(FAD)–NADP + complex to CT1, precluding any further dissection of the reaction mechanism by time- Fig. 5. Spectral changes resulting from the incubation of FprA with NADP + in the presence and absence of potassium ferricyanide under anaerobic conditions. (A) FprA at c.20l M was incubated with 200 l M NADP + and 140 lM K 3 Fe(CN) 6 at 25 °Cin20mM Hepes ⁄ NaOH (pH 7.0), containing 100 mM NaCl and 10% glycerol, within a gas-tight cuvette made anaerobic by several vacuum–N 2 flushing cycles. Reaction times were 0 min to 330 min. (B) Same conditions as in (A), with the omission of K 3 Fe(CN) 6 . Reaction times were 0 h to 19 h. Inset: absorbance at 580 nm as a function of incubation time. The curve represents the best fit according to a single exponential decay equation (k ¼ 0.095 h )1 ). Oxidation of NADP + to its 4-oxo derivative M. de Rosa et al. 4002 FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS resolved spectral analysis. The fact that CT1 formation follows single-phase kinetics without the formation of any observable transient intermediate suggests that water molecule addition to the nicotinamide moiety (Fig. 1) might be the limiting step of the whole reaction. Investigating the role of Glu214 and His57 of FprA in the catalysis of NADPO production As it was not possible to obtain detailed information on the catalytic mechanism of the NADP + oxidation reaction by kinetic studies, we attempted to gain insights into the role of specific residues of FprA by site-directed mutagenesis. On the basis of structural data, Glu214 and His57 of FprA have been proposed to be involved in promoting nucleophilic attack by a water molecule on the C4 of the nicotinamide moiety of NADP + [4]. The production and purification of the enzyme variants, where these residues were replaced with Ala or Gln, have been described else- where [5]. The physiologic NADPH-dependent activity of the mutant FprA forms was characterized in detail. Unlike Glu214, which had essentially no effect on this activity, His57 turned out to decrease by 4–5-fold the hydride transfer rate from NADPH or NADH to the enzyme-bound FAD [5]. Here, we have assayed FprA- E214A and FprA-H57Q for their ability to catalyze NADP + oxidation to NADPO. As shown in Fig. 6, both mutant forms supported the production of NADPO, with time-courses similar to that of the wild-type enzyme. It is noteworthy that both single mutations slightly increased the initial rate of NADPO synthesis. At 500 lm NADP + , the reaction rates were 0.21 h )1 and 0.19 h )1 for FprA-E214A and FprA-H57Q, respectively. If the rate-determining step of NADPO production is water addition, the proposed activating role of Glu214 and His57 should be dismissed. The crystal structure of the complex between FprA-H57Q and NADPO has been obtained at 1.8 A ˚ resolution [5]. Slight alterations in the posi- tioning of the nicotinamide ring in the active site have been observed by comparing the mutant with the wild-type enzyme. The described 0.7 A ˚ shift of the nicotinamide moiety could represent the structural basis of the observed small increase in NADP + oxidation rate observed with this mutant. NADPO as an inhibitor of FNRs To the best of our knowledge, this is the first time that enzymatic oxidation of NADP + has been reported and the spectral properties of the resulting NADP derivative have been described. It was thus of interest to provide a first characterization of NADPO as a possible probe for studying the NADPH-binding site of dehydrogenases. By inhibition studies under steady- state conditions, we have found that NADPO acts as a competitive inhibitor with respect to NADPH on var- ious FNRs (both photosynthetic and nonphotosynthet- ic), with K i values ranging between 1 and 30 lm (Table 1). FprA is the enzyme most strongly inhibited by this nucleotide. To better characterize the interac- tion of NADPO with FprA, enzyme–ligand binding was studied by difference spectrophotometry using var- ious nucleotides: NADPO, NADP + , thio-NADP + and 2¢-phospho-AMP (2P-AMP). The K d determined for the complex between the enzyme and NADPO was slightly but significantly lower than that measured for NADP + (Table 2). As 2P-AMP, which lacks the NMN portion of NADP + but still strongly binds to FprA, yielded very weak perturbations of the absorp- tion spectrum of the enzyme, it is clear that the intense difference spectra of the FprA–dinucleotide complexes are due to the alteration of the isoalloxazine microen- vironment induced by the binding of the NMN group. The small structural differences in the nicotinamide rings of NADP + , NADPO and thio-NADP + resulted in small but significant differences in the corresponding difference spectra (Table 2). In plant FNRs, we found a correlation between the intensity of the difference spectrum induced by binding and the degree of Fig. 6. Time-courses of NADPO production catalyzed by wild-type and mutant forms of FprA. The reaction was carried out at 25 °Cin the presence of 100 l M wild-type FprA (filled circles), FprA-E214A (open circles) or FprA-H57Q (squares) and 500 l M NADP + in air- equilibrated buffer. M. de Rosa et al. Oxidation of NADP + to its 4-oxo derivative FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 4003 occupancy of the nicotinamide ring of the bound ligand in the active site [14]. In this view, it is signifi- cant that the perturbations induced by NADPO bind- ing to FprA were more intense than those induced by the other dinucleotides tested. This suggests that the 4-pyridone-5-carboximide ring of NADPO is partic- ularly well fitted to stack over the isoalloxazine ring, resulting in a higher occupancy than with dinucleotides carrying other pyridine derivatives. These observations support the hypothesis that accumulation of NADPO would substantially inhibit its own production by FprA, due to competition with NADP + for binding to the enzyme active site. Conclusions In this article, we provide clear evidence that M. tuberculosis FprA and bovine AdR, but not plant- like FNRs, catalyze the FAD-dependent oxidation of NADP + to NADPO. This enzyme activity, possibly shared by all AdR-like enzymes, is highly regiospecif- ic, in that it targets only the 4-position of the pyrid- ine ring of the substrate. The very low reaction rate tends to exclude a physiologic role for NADPO, although at least one example exists of an NADP derivative, i.e. nicotinic acid adenine dinucleotide phosphate, with documented signaling functions [15]. Rather, we feel that the NADP + oxidation activity of AdR-like enzymes reflects a specific and conserved reactivity of their active sites, where a water molecule exerts considerable strain and possibly a polarizing effect on the C4 atom of the nicotinamide moiety of the bound substrate. The strict interaction between a zinc-bound water molecule and the nicotinamide ring has been suggested to have a role in activating NADH for efficient hydride transfer in the catalytic cycle of horse liver alcohol dehydrogenase [16]. It can be speculated that water 1 plays a similar role in the active site of AdR-type FNRs, i.e. it favors the hydride transfer between NADPH and FAD in the physiologic reaction catalyzed by these enzymes. When NADP + is the enzyme ligand, the electrophilicity of the nicotinamide C4 promotes water 1 addition to the nicotinamide and subsequent FAD-dependent oxi- dation to give NADPO. It is interesting to note that no ordered water molecules are present in the prox- imity of the nicotinamide ring in the crystal structure of the complexes between plant FNR and NADP + or NADPH [14]. This observation could provide a rationale for the lack of NADP + oxidation activity in plant-type FNRs. Further work will be required to fully elucidate the actual role of active site water mole- cules in modulating the reactivity of NADP + and NADPH bound to FprA. Experimental procedures Enzymes and chemicals Calf intestine alkaline phosphatase was obtained from GE Healthcare (Milano, Italy). Crotalus durissus phosphodiest- erase, beef liver catalase, superoxide dismutase from bovine Table 1. Inhibition constants of NADPO for different FNRs. 2,6-Dichloroindophenol reductase activity was measured using either 62 nM FprA or 2.5 n M T. gondii FNR in 0.2 M Tris ⁄ HCl (pH 8.2) at 25 °C at a fixed concentration of 66 lM 2,6-dichloroindophenol; the NADPH concentra- tion was varied between 0.1 and 10 l M in the case of FprA, and between 2 and 27 lM in the case of T. gondii FNR. The concentration of NADPO was varied between 0 and 20 l M. INT reductase activity was measured using 2.5 nM spinach leaf FNR in 0.2 M Tris ⁄ HCl (pH 9.0), 70 m M NaCl and 0.1% Triton X-100 at 25 °C at a fixed concentration of 100 lM INT; the NADPH concentration was varied between 2 and 27 l M; the NADPO concentration was varied between 0 and 20 lM. All assay mixtures included an NADPH-regenerating system comprising glucose 6-phosphate and glucose-6-phosphate dehydrogenase. Enzyme Diaphorase reaction K NADPH m (lM) a k cat (s )1 ) a K NADPO i (lM) FprA NADPH fi 2,6-dichloroindophenol 0.2 ± 0.01 4.1 ± 0.05 1.2 ± 0.2 Spinach leaf FNR NADPH fi INT 5.0 ± 0.4 54 ± 1 25 ± 5 T. gondii FNR NADPH fi 2,6-dichloroindophenol 3.5 ± 0.2 64 ± 1.5 30 ± 5 a Values reported in the table should be considered as ‘apparent’ kinetic parameters, as they were determined at a nonsaturating fixed con- centration of the electron acceptor. Table 2. Affinities of various nucleotides for FprA and extent of the spectral perturbations induced by their binding to the enzyme. Ligand K d (lM) De a (mM )1 Æcm )1 ) 2P-AMP 3.6 ± 2 0.22 (489 nm) NADP + 6.3 ± 0.7 1.2 (499 nm) Thio-NADP + 2.0 ± 0.2 0.83 (496 nm) NADPO 3.0 ± 1 1.8 (494 nm) a Difference extinction coefficients at the wavelength indicated in parentheses were calculated by subtracting the absorbance of free FprA from that extrapolated at infinite ligand concentration. Oxidation of NADP + to its 4-oxo derivative M. de Rosa et al. 4004 FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS erythrocytes and yeast glucose-6-phosphate dehydrogenase were all from Roche Diagnostics (Monza, Italy). Horserad- ish peroxidase was bought from Invitrogen (San Giuliano Milanese, Milano, Italy). Recombinant M. tuberculosis wild-type FprA was produced and isolated in two different molecular forms: without extra residues [10], and with an N-terminal poly-His extension [5]. The two enzyme forms were found to be indistinguishable in their functional prop- erties. The site-directed mutants FprA-H57Q and FprA- E214A were obtained in poly-histidinylated form only [5]. Recombinant S. oleracea leaf FNR, T. gondii FNR and P. falciparum FNR were purified as described elsewhere [11,17,18]. Purified recombinant bovine AdR was a gener- ous gift of R Bernhardt (Universita ¨ t des Saarlandes, Saar- bru ¨ cken, Germany). NADP + , NADPH, NAD + and AMP were purchased from Sigma-Aldrich (Milano, Italy). Amplex Red was from Invitrogen. All other chemicals were of the highest possible grade. Chromatographic separation of nucleotides Variable volumes of enzyme reaction mixtures (see below) were treated with equal volumes of acetonitrile to denatur- ate and precipitate the protein. After centrifugation at 12 000 g for 10 min, the supernatants were dried, resuspend- ed in 50 mm ammonium formate, and chromatographed by a modification of the high-performance ion exchange proce- dure described in Orr & Blanchard [6]. Using an A ¨ KTA FPLC apparatus (GE Healthcare), samples were loaded on a MonoQ HR 5 ⁄ 5 column (1 mL; GE Healthcare), equili- brated in the above volatile buffer. Nucleotides were separ- ated at room temperature using a 50–600 mm ammonium formate gradient in 25 column volumes at a flow rate of 1mLÆmin )1 . The eluate was monitored continuously by measuring its absorbance at 254 nm. Fractions containing NADPO or 3-carboxamide-4-pyridone mononucleotide (NMNO) were dried under vacuum and stored at ) 20 °C. MS MS and MS ⁄ MS data were obtained using an LCQ ADV MAX ion trap mass spectrometer equipped with an ESI ion source and controlled by xcalibur software v.1.3 (Thermo-Finnigan, San Jose, CA, USA). ESI experi- ments were carried out in positive ion mode under the following constant instrumental conditions: source voltage 5.0 kV, capillary voltage 10 V, capillary temperature 250 °C, and tube lens voltage 55 V. MS ⁄ MS spectra obtained by collision-induced dissociation were performed with an isolation width of 2 Th (m ⁄ z), and the activation amplitude was around 35% of the ejection RF amplitude of the instrument, which corresponds to 1.58 V. ESI-MS of NADPO yielded ions at m ⁄ z 760.18 [MH] + and 782.06, amu [MNa] + ESI-MS ⁄ MS of the former ion yielded fragment ions at m ⁄ z 741.74 [MH–H 2 O] + , 624.71 [MH–adenine] + , 603.66 [MH)4-oxo-nicotinamide– H 2 O] + , 489.71 [MH)4-oxo-nicotinamide-ribose–H 2 O] + , and 329.86 [MH)4-oxo-nicotinamide-ribose-diphosphate– H 2 O] + , amu. Determination of the absorption and fluorescence properties of NADPO and NMNO All absorption and fluorescence emission spectra were recor- ded on a UV–visible 8453 diode array spectrophotometer (Agilent, Cernusco sul Naviglio, Milano, Italy) and a Cary Eclipse spectrofluorimeter (Varian, Leini, Torino, Italy), respectively. In order to determine its extinction coefficient, NADPO was quantified on the basis of the amount of the phosphate released by phosphatase treatment. NADPO at 10–20 nmol was incubated for 1 h with 0.25 units of alka- line phosphatase at 25 °C in 0.5 m Tris ⁄ HCl (pH 9.0), con- taining 10 mm MgCl 2 , to hydrolyze the 2¢-phosphate group. Free phosphate content was determined by the method of Chen et al. [19]. Known amounts of NADP + were used as controls, verifying the accuracy of the procedure. To isolate NMNO, c. 10 nmol of NADPO was treated with 2 lgof phosphodiesterase for 20 min at 25 °Cin20mm Tris ⁄ HCl (pH 7.7). NMNO was then purified chromatographically as described above. The absorbance spectrum of NMNO was recorded both in 20 mm Tris ⁄ HCl (pH 7.7) and after adjust- ing the pH to c. 1 by the addition of HCl. Monitoring of the time-course of NADPO formation catalyzed by various enzymes The enzymatic conversion of NADP + to NADPO was studied in both aerobic and anaerobic conditions. The aero- bic reactions were carried out at 25 °C by mixing 50–150 lm enzyme with variable concentrations of NADP + , ranging from 150 lm to 10 mm,in20mm Hepes ⁄ NaOH (pH 7.0), containing 100 mm NaCl and 10% glycerol. At different incubation times, 40 lL aliquots were withdrawn, AMP was added as internal standard, and sam- ples were analyzed by ion exchange chromatography as des- cribed above. The amount of NADPO was determined on the basis of peak integration data provided by unicorn 5 software (GE Healthcare), and the experimentally estimated e 254 value of 15.6 mm )1 Æcm )1 for NADPO. To monitor the reaction in the absence of molecular oxygen as the oxidant, c.20lm FprA in the same buffer as above, either in the absence or in the presence of 140 lm K 3 Fe(CN) 6 , was placed in an anaerobic cuvette, containing a concentrated NADP + solution in the side arm to yield a final concentra- tion of 200 lm. After anaerobiosis was established by suc- cessive cycles of N 2 -flushing and evacuation, reactants were mixed. Spectral changes were recorded over a period of sev- eral hours at 20 °C using a UV–visible 8453 diode array spectrophotometer (Agilent). M. de Rosa et al. Oxidation of NADP + to its 4-oxo derivative FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 4005 Identification of the reactive oxygen species produced in the reaction between NADP + and O 2 catalyzed by FprA Air-equilibrated mixtures of 10 lm FprA and 0.5 mm NADP + were incubated as described in the previous para- graph in the presence of 0.1 unitÆmL )1 horseradish peroxi- dase and 100 lm Amplex Red. Peroxidase-catalyzed Amplex Red conversion to resorufin was monitored by measuring the fluorescence emission at 585 nm of the solu- tion upon excitation at 571 nm. When superoxide dismu- tase and catalase were present, the concentrations were 0.5 lgÆmL )1 and 1 lgÆmL )1 , respectively. Enzyme activity assays Assays of the NADPH-dependent catalytic activities of FprA, spinach leaf (S. oleracea) FNR and T. gondii FNR were performed under steady-state conditions with different artificial electron acceptors [iodo-nitro-tetrazolium chloride (INT) or 2,6-dichloroindophenol] by continuously monitor- ing the reactions using either an Agilent 8453 diode array or a Varian Cary 100 double-beam spectrophotometer. Reaction conditions have been described elsewhere [10]. To evaluate the inhibitory effect of NADPO, the concentration of NADPO was varied between 0 and 20 lm, whereas that of NADPH was independently varied between 0.1 and 10 lm and between 2 and 27 lm, in the case of FprA and FNRs, respectively. K i values of NADPO were determined by fitting the experimental data points to the theoretical equation for the competitive inhibition mechanism, using the nonlinear fitting feature of grafit 5 (Erithacus Soft- ware Ltd, Horley, Surrey, UK). Ligand-binding studies Spectrophotometric titrations of FprA (c.15lm) with either NADPO, NADP + , 2P-AMP or thio-NADP + were performed at 15 °Cin20mm Hepes ⁄ NaOH (pH 7.0), con- taining 50 mm NaCl, using a Cary 100 double-beam spec- trophotometer (Varian). The spectra were recorded before and after successive additions of equal amounts of the nuc- leotide to the sample and reference cells. Difference spectra were computed by subtracting the initial spectrum, correc- ted for dilution, from those recorded after each ligand addi- tion. K d values were computed by fitting the data points to the theoretical equation for 1 : 1 binding [20], using the nonlinear fitting feature of grafit 5 (Erithacus Software Ltd). Acknowledgements We are grateful to Rita Bernhardt for providing a sample of purified recombinant bovine AdR. We also thank Federico Fischer for his contribution to the ini- tial part of this research. This work was supported by grants from Ministero dell’Universita ` e della Ricerca of Italy (PRIN 2005) and Fondazione Cariplo, Milano, Italy. References 1 Carrillo N & Ceccarelli EA (2003) Open questions in ferredoxin-NADP + reductase catalytic mechanism. Eur J Biochem 270, 1900–1915. 2 Ziegler GA & Schulz GE (2000) Crystal structures of adrenodoxin reductase in complex with NADP + and NADPH suggesting a mechanism for the electron trans- fer of an enzyme family. Biochemistry 39, 10986–10995. 3 Karplus PA & Faber HR (2004) Structural aspects of plant ferredoxin:NADP + oxidoreductases. Photosynth Res 81, 303–315. 4 Bossi RT, Aliverti A, Raimondi D, Fischer F, Zanetti G, Ferrari D, Tahallah N, Maier CS, Heck AJ, Rizzi M et al. (2002) A covalent modification of NADP + revealed by the atomic resolution structure of FprA, a Mycobacterium tuberculosis oxidoreductase. Biochemis- try 41, 8807–8818. 5 Pennati A, Razeto A, de Rosa M, Pandini V, Vanoni MA, Mattevi A, Coda A, Aliverti A & Zanetti G (2006) Role of the His57–Glu214 ionic couple located in the active site of Mycobacterium tuberculosis FprA. Bio- chemistry 45, 8712–8720. 6 Orr GA & Blanchard JS (1984) High-performance ion- exchange separation of oxidized and reduced nicotina- mide adenine dinucleotides. Anal Biochem 142, 232– 234. 7 Wong P, Bachki A, Banerjee K & Leyland-Jones B (2002) Identification of N1-methyl-2-pyridone-5-carbox- amide and N1-methyl-4-pyridone-5-carboxamide as components in urine extracts of individuals consuming coffee. J Pharm Biomed Anal 30, 773–780. 8 Chang MLW & Johnson BC (1959) N-Methyl-4-pyri- done-5-carboxamide, a new major normal metabolite of nicotinic acid in rat urine. J Biol Chem 234, 1817–1821. 9 Scriven EFV (1984) Pyridines and their benzo deriva- tives: (ii) reactivity at ring atoms. In Comprehensive Heterocycle Chemistry, Vol. 2 (Katritzky AR & Rees CW, eds), pp. 167–168. Pergamon Press, Oxford, UK. 10 Fischer F, Raimondi D, Aliverti A & Zanetti G (2002) Mycobacterium tuberculosis FprA, a novel bacterial NADPH-ferredoxin reductase. Eur J Biochem 269, 3005–3013. 11 Pandini V, Caprini G, Thomsen N, Aliverti A, Seeber F & Zanetti G (2002) Ferredoxin-NADP + reductase and ferredoxin of the protozoan parasite Toxoplasma gondii interact productively in vitro and in vivo. J Biol Chem 277, 48463–48471. Oxidation of NADP + to its 4-oxo derivative M. de Rosa et al. 4006 FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 12 Milani M, Balconi E, Aliverti A, Mastrangelo E, Seeber F, Bolognesi M & Zanetti G (2007) Ferredoxin- NADP + reductase from Plasmodium falciparum under- goes NADP + -dependent dimerization and inactivation: functional and crystallographic analysis. J Mol Biol 367, 501–513. 13 Aliverti A, Jansen T, Zanetti G, Ronchi S, Herrmann RG & Curti B (1990) Expression in Escherichia coli of ferredoxin:NADP + reductase from spinach. Bacterial synthesis of the holoflavoprotein and of an active enzyme form lacking the first 28 amino acid residues of the sequence. Eur J Biochem 191, 551–555. 14 Deng Z, Aliverti A, Zanetti G, Arakaki AK, Ottado J, Orellano EG, Calcaterra NB, Ceccarelli EA, Carrillo N & Karplus PA (1999) A productive NADP + binding mode of ferredoxin-NADP + reductase revealed by pro- tein engineering and crystallographic studies. Nat Struct Biol 6, 847–853. 15 Soares S, Thompson M, White T, Isbell A, Yamasaki M, Prakash Y, Lund FE, Galione A & Chini EN (2006) NAADP as a second messenger: neither CD38 nor base-exchange reaction are necessary for in vivo genera- tion of NAADP in myometrial cells. Am J Physiol Cell Physiol 292, C227–C239. 16 Meijers R, Morris RJ, Adolph HW, Merlii A, Lamzin VS & Cedergren-Zeppezauer ES (2001) On the enzymat- ic activation of NADH. J Biol Chem 276, 9316–9321. 17 Aliverti A, Deng Z, Ravasi D, Piubelli L, Karplus PA & Zanetti G (1998) Probing the function of the invari- ant glutamyl residue 312 in spinach ferredoxin-NADP + reductase. J Biol Chem 273, 34008–34015. 18 Ro ¨ hrich RC, Englert N, Troschke K, Reichenberg A, Hintz M, Seeber F, Balconi E, Aliverti A, Zanetti G, Ko ¨ hler U et al. (2005) Reconstitution of an apicoplast- localised electron transfer pathway involved in the isoprenoid biosynthesis of Plasmodium falciparum. FEBS Lett 579, 6433–6438. 19 Chen PS, Toribara TY & Warber H (1956) Microdeter- mination of phosphorus. Anal Chem 28, 1756–1758. 20 Wang ZX, Kumar NR & Srivastava DK (1992) A novel spectroscopic titration method for determining the dis- sociation constant and stoichiometry of protein–ligand complex. Anal Biochem 206, 376–381. M. de Rosa et al. Oxidation of NADP + to its 4-oxo derivative FEBS Journal 274 (2007) 3998–4007 ª 2007 The Authors Journal compilation ª 2007 FEBS 4007 . Enzymatic oxidation of NADP + to its 4-oxo derivative is a side-reaction displayed only by the adrenodoxin reductase type of ferredoxin-NADP + reductases Matteo. FprA, an NADPH-ferredoxin reductase homologous to mammalian adrenodoxin reductase, promotes the oxidation of NADP + to its 4-oxo derivative 3-car- boxamide-4-pyridone