The ferredoxin-NADP+ reductase/ferredoxin electron transfer system of Plasmodium falciparum


A. Aliverti, Dipartimento di Scienze Biomolecolari e Biotecnologie, Università degli Studi di Milano, via Celoria 26, 20133 Milano, Italy
Fax: +39 02 50314895
Tel: +39 02 50314897


In the apicoplast of apicomplexan parasites, plastidic-type ferredoxin and ferredoxin-NADP+ reductase (FNR) form a short electron transport chain that provides reducing power for the synthesis of isoprenoid precursors. These proteins are attractive targets for the development of novel drugs against diseases such as malaria, toxoplasmosis, and coccidiosis. We have obtained ferredoxin and FNR of both Toxoplasma gondii and Plasmodium falciparum in recombinant form, and recently we solved the crystal structure of the P. falciparum reductase. Here we report on the functional properties of the latter enzyme, which differ markedly from those of homologous FNRs. In the physiological reaction, P. falciparum FNR displays a kcat five-fold lower than those usually determined for plastidic-type FNRs. By rapid kinetics, we found that hydride transfer between NADPH and protein-bound FAD is slower in the P. falciparum enzyme. The redox properties of the enzyme were determined, and showed that the FAD semiquinone species is highly destabilized. We propose that these two features, i.e. slow hydride transfer and unstable FAD semiquinone, are responsible for the poor catalytic efficiency of the P. falciparum enzyme. Another unprecedented feature of the malarial parasite FNR is its ability to yield, under oxidizing conditions, an inactive dimeric form stabilized by an intermolecular disulfide bond. Here we show that the monomer–dimer interconversion can be controlled by oxidizing and reducing agents that are possibly present within the apicoplast, such as H2O2, glutathione, and lipoate. This finding suggests that modulation of the quaternary structure of P. falciparum FNR might represent a regulatory mechanism, although this needs to be verified in vivo.

Structured digital abstract


charge transfer complex between NADPH and enzyme-bound FAD


charge transfer complex between NADP+ and fully reduced enzyme-bound FAD


1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride




ferredoxin-NADP+ reductase




isopropyl thio-β-d-galactoside


2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyltetrazolium chloride


Plasmodium falciparum ferredoxin


Plasmodium falciparum ferredoxin-NADP+ reductase


The protozoan parasite Plasmodium falciparum causes more than 1 000 000 deaths every year [1]. The development of resistance in the parasite to the presently used drugs means that new targets for novel antimalarial drugs need to be developed. Plasmodia, like all of the Apicomplexa except for Cryptosporidium spp., contain a unique organelle, the apicoplast, that is phylogenetically and functionally related to plant nonphotosynthetic plastids [2,3]. The apicoplast has been shown to be a major determinant of virulence [4], strongly suggesting that proteins located therein could provide new and promising targets for rational drug design, because it is assumed to contain a number of unique metabolic pathways not found in the mammalian host. Recently, it has been reported that Toxoplasma gondii and P. falciparum possess a nuclear-encoded but apicoplast-localized redox system comprising a plastidic-type ferredoxin-NADP+ reductase (FNR; EC and a [2Fe–2S] ferredoxin (Fd) [5,6]. All other apicomplexan parasites examined so far also possess DNA sequences for apicoplast-targeted Fd and FNR, implying that this redox system is of primary importance for the parasite. Although the physiological role of the FNR/Fd couple in the apicomplexan parasites is still largely unknown, it is conceivable that it could provide reducing power for various reductive biosyntheses. Indeed, we have recently demonstrated that, in vitro, the FNR/Fd system of P. falciparum can supply electrons to LytB, an enzyme of the apicoplast isoprenoid biosynthetic pathway [7], which is a known target of antiplasmodial compounds [8].

In the past, we produced both the FNR and Fd of T. gondii in recombinant form, and characterized this electron transport system in detail [9–12]. Unfortunately, we did not succeed in obtaining crystals of these proteins that were suitable for X-ray analysis. With the aim of obtaining new insights into the structure–function relationships of the apicomplexan FNR/Fd system, we then accomplished the heterologous expression in Escherichia coli of the P. falciparum FNR (PfFNR) gene. Crystals of the recombinant reductase were obtained, and its three-dimensional structure was solved in free form and complexed with the inhibitor 2′-P-AMP at 2.4 and 2.7 Å resolution, respectively [13]. An unexpected feature, peculiar to PfFNR, was the finding that it crystallizes as a dimer stabilized by an intermolecular disulfide bridge linking the Cys99 side chains of each protomer. We demonstrated that PfFNR dimerization in solution occurs under oxidizing conditions and is highly favored by 2′-P-AMP as well as NADP+ binding. The covalent dimer is essentially inactive, and disulfide reduction restores full activity [13]. More recently, the three-dimensional structure of P. falciparum Fd (PfFd) and the identification of its surface regions involved in the interaction with the reductase have been reported [14,15]. Thus, whereas the structural features of PfFNR and PfFd are now known in great detail, the functional properties of the protein couple have been poorly analyzed. In order to fill this gap, we report here on the distinctive functional properties of the P. falciparum FNR/Fd redox system in comparison with other known FNR/Fd couples. The new information that we provide here is expected to increase our knowledge of the structure–function relationships of PfFNR, which could be helpful for the rational design of PfFNR/PfFd inhibitors, possibly endowed with antimalarial activity.


PfFNR gene cloning and heterologous expression

Using plasmid pET–PfFNR, the putative mature PfFNR, starting at Lys56 of the deduced sequence (the first amino acid of the mature form was chosen on the basis of sequence alignments with FNRs where the boundary between the signal N-terminal sequence and mature polypeptide is known), was overproduced in E. coli as a fusion protein with an N-terminal His-tag extension. A factor Xa recognition site, engineered between the poly-His and the PfFNR sequence, allowed tag removal during protein purification. Notwithstanding the high A/T content of its coding sequence, a substantial production of PfFNR was obtained in E. coli using as host the Rosetta(DE3) strain, which is specifically engineered to enhance the expression of eukaryotic proteins containing codons rarely used in E. coli. Optimal production of the enzyme was obtained by growing transformed cells at 20 °C for 24 h after isopropyl thio-β-d-galactoside (IPTG) induction. The purification procedure included a step on a phenyl–Sepharose column after the metal chelate affinity chromatography. Following the factor Xa proteolysis step, a second affinity column step was performed, which allowed the removal of both small amounts of undigested enzyme and contaminants, exploiting weak binding affinity of the untagged PfFNR for the Ni2+–Sepharose resin. About 1.5 mg·g−1 cells of homogeneous enzyme were obtained, with a 26% overall yield. Recombinant PfFd was difficult to produce in substantial amounts with the previously described expression plasmid [5], so the DNA fragment encoding the putative mature form of PfFd, starting at Leu97, was transferred to plasmid pET28b without the addition of any affinity tag. A three-fold improvement of holo-PfFd yield was obtained by cotransforming the E. coli cells with a plasmid carrying the isc gene cluster to favor iron–sulfur biogenesis. About 2.5 mg of purified PfFd were obtained from 1 g of cells with a 26% yield. PfFd showed the typical spectrum of a 2Fe Fd, with peaks at 277, 424 and 460 nm, and a value of 0.59 for the A424 nm/A280 nm ratio.

Steady-state kinetic analyses of PfFNR

The crystal structure of the 2′-P-AMP–PfFNR complex suggested that the His286 side chain needs to be charged to effectively bind the substrate [13]. In order to verify this hypothesis, we determined at pH 7.0 the kinetic parameters of the diaphorase reactions of PfFNR, previously measured at pH 8.2 [13]. Table 1 shows that there is, indeed, a decrease in the Km values for NADPH with both ferricyanide and 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyltetrazolium chloride (INT) as electron acceptors, by approximately four-fold and seven-fold, respectively, whereas the kcat values were only moderately lowered (by ∼ 30%). Thus, PfFNR shows a three-fold to five-fold higher catalytic efficiency as a diaphorase at pH 7.0 than at pH 8.2 [13], which is at variance with the plastidic-type FNRs.

Table 1.   Kinetic parameters for the diaphorase reactions of PfFNR. Activity assays were performed at 25 °C. The kinetic parameters at pH 8.2 were taken from [13].
pHReactionkcat (e eq·s−1)KmNADPHm)kcat/KmNADPH (e eq·s−1·μm−1)
7.0NADPH → Fe(CN)63−170 ± 810 ± 317 ± 5
NADPH → INT130 ± 1150 ± 72.6 ± 0.4
8.2NADPH → Fe(CN)63−250 ± 836 ± 67 ± 1
NADPH → INT180 ± 11350 ± 300.5 ± 0.05

The physiological reaction of PfFNR was initially measured with recombinant T. gondii Fd, which was available in our laboratory. The kcat value was only 20% of that obtained with the homologous couple of T. gondii, and the Km value for T. gondii Fd was two-fold higher. This poor catalytic activity prompted us to produce PfFd in recombinant form. The kinetic parameters of the physiological reaction of PfFNR with PfFd were determined at both pH 8.2 and pH 7.0 (Table 2). At pH 8.2, the Km of PfFNR for the homologous protein substrate was indeed in the same range as those of other plastidic-type FNR/Fd couples. The kcat values with both apicomplexan Fds were similar, and about five-fold lower than those of plastidic-type FNRs [9,16,17]. As expected, at pH 7.0, the Km for NADPH was markedly diminished (11-fold lower than that at pH 8.2), whereas the kcat was halved. The three-fold decrease in the Km value for PfFd on lowering of the pH from 8.2 to 7.0 would suggest stronger binding also of the protein substrate at neutral pH. Thus, also in the physiological reaction, PfFNR shows a higher catalytic efficiency at pH 7 than at pH 8.2.

Table 2.   Kinetic parameters for the cytochrome c reductase (PfFd-dependent) reaction of PfFNR. Activity assays were performed at 25 °C.
pHkcat (e eq·s−1)KmNADPHm)kcat/KmNADPH (e eq·s−1·μm−1)KmPfFdm)kcat/KmPfFd (e eq·s−1·μm−1)
7.013 ± 0.55.5 ± 0.52.4 ± 0.20.8 ± 0.116 ± 2
8.223 ± 161 ± 70.37 ± 0.052.4 ± 0.39.2 ± 1

PfFNR covalent dimer

The covalent dimer of PfFNR, which was shown to maintain only 5% of the kcat value in the NADPH–ferricyanide reaction [13], was found to be fully inactive in catalysing the physiological NADPH–cytochrome c reductase reaction (PfFd-dependent). We further investigated the conditions that promote Cys99 disulfide formation to give the homodimer, and tested the ability of physiological reductants to reverse the dimerization and re-establish the catalytic activity of PfFNR. For easy analysis of the monomer–dimer interconversion under different conditions, the procedure reported by Leichert and Jacob [18] was followed. Briefly, after the desired treatment was performed, N-ethylmaleimide and SDS were added to the samples in order to block all protein-free sulfhydryls. SDS/PAGE was then performed under nonreducing conditions, in order to preserve the disulfide bonds possibly present in the protein. Incubation of PfFNR with different concentrations of hydrogen peroxide in the presence or absence of NADP+ showed that this reactive oxygen species is far more efficient than O2 [13] in promoting formation of the enzyme dimer, which accumulated in significant amounts only when the ligand was present in the incubation mixture (Fig. 1A,B). The time course of the decrease of the enzyme diaphorase activity matched the formation of the dimer (Fig. 1C). Oxidized lipoic acid was also able to induce enzyme dimerization, although at a lower rate: 40% dimer was formed in 3 h in the presence of 100 μm lipoic acid and 1 mm NADP+. The PfFNR dimer purified by gel filtration was converted back to the monomer by incubation with dithiothreitol [13], glutathione (GSH), or dihydrolipoate. The time course of production of monomer by incubation of 7.5 μm PfFNR dimer with 50 μm dihydrolipoate was determined both by nonreducing SDS/PAGE (Fig. 2A) and by measuring the diaphorase activity (Fig. 2B). The half-life for conversion was 7 min. In similar experiments, 100 μm GSH required more time for full conversion (half-life of 11 min).

Figure 1.

 Effect of NADP+ on the hydrogen peroxide-promoted dimerization of PfFNR. PfFNR (∼ 15 μm) was incubated at 20 °C in the presence of 0.5 mm H2O2 in 50 mm Tris/HCl (pH 7.4) containing 10% glycerol. (A, B) The progress of protein dimerization either in the absence (A) or in the presence (B) of 1 mm NADP+, as analyzed by nonreducing SDS/PAGE on 10% polyacrylamide gels. Incubation times are indicated above the respective gel lanes. The molecular masses of the protein standards (M) are: 205, 116, 97.4, 66, 34, and 29 kDa, from top to bottom. (C) Time course of the NADPH-ferricyanide reductase activity of PfFNR incubated either in the absence (open circles) or in the presence (filled circles) of 1 mm NADP+. Curves represent the equations of single exponential decays with nonzero endpoint values fitted to the experimental data points.

Figure 2.

 Effect of dihydrolipoate on the dimeric form of PfFNR. Gel filtration-purified dimeric PfFNR (∼ 15 μm FAD) was incubated at 20 °C in the presence of 50 μm dihydrolipoate in 50 mm Tris/HCl (pH 7.4) containing 10% glycerol. (A) Analysis of the reaction mixture by nonreducing SDS/PAGE on a 10% polyacrylamide gel. Incubation times are indicated above the respective gel lanes. The molecular masses of the protein standards (M) are: 205, 116, 97.4, 66, 34 and 29 kDa, from top to bottom. (B) Time course of the NADPH-ferricyanide reductase activity of PfFNR. The curve represent a single exponential decay equation fitted to the experimental data points.

With the aim of studying the role of Cys99 of PfFNR, we produced the mutant C99A. Furthermore, we reasoned that by eliminating the residue responsible for enzyme dimerization, we could crystallize PfFNR as a monomer and thus obtain the three-dimensional structure of the catalytically active enzyme form. The purified PfFNR-C99A showed the same spectral and kinetic properties as the wild-type enzyme (not shown). The titration of the sulfhydryl groups of wild-type and mutant reductases in the presence and absence of 6 m guanidinium chloride confirmed that the mutant had only five cysteines that were titrable under denaturing conditions and had lost the only –SH group that was titrable under native conditions in the wild-type PfFNR. As expected, no dimer formation was ever seen, even after incubation with NADP+ and diamide, a strong oxidant of sulfhydryl groups. Unfortunately, notwithstanding several trials, the mutant proteins did not yield crystals.

PfFNR-bound FAD reduction in anaerobiosis

The enzyme solution was made anaerobic and, after recording of the spectrum of the oxidized species, a 1.2 molar excess of NADPH was added from the side arm of the anaerobiosis cuvette. Immediately, the yellow solution was bleached and a long-wavelength band appeared, extending over 800 nm (Fig. 3A). This absorbance band can be tentatively assigned to a charge transfer complex between fully reduced enzyme-bound FAD and NADP+ (CT2), on the basis of previously described reduced species of other FNRs [19–21]. No FAD semiquinone was observed, even during the slow reoxidation of the enzyme by oxygen that occurred after the cuvette had been opened to the air.

Figure 3.

 Redox properties of PfFNR-bound FAD. Anaerobic solutions of ∼ 15 μm PfFNR in 50 mm Hepes/NaOH (pH 7.0) containing 10% glycerol were either reacted with a slight excess of NADPH or photoreduced either in the absence or in the presence of a slight excess of NADP+. All reactions were carried out in anaerobic cuvettes thermostated at 15 °C. (A) The spectrum of the PfFNR solution was recorded before (thin line) and after (bold line) the addition of NADPH in a 1.2 : 1 molar ratio to the enzyme from the cuvette side arm. (B) The spectrum of the PfFNR solution, containing 13 mm EDTA and 1.3 μm 5-deaza-riboflavin, was recorded before (thin line) and after (bold lines with different styles) successive light irradiation periods. (C) Spectra were recorded under the same conditions described for (B), with the exception that NADP+ was included in the reaction mixture in a 1.3 : 1 molar ratio with the enzyme. The arrows indicate the direction of the spectral changes observed as irradiation time increased. In the presence of NADP+, the intensity of the smooth band centered at about 800 nm was maximal when the absorbance of the bands at 380 and 450 nm was almost completely bleached (bold line). Further irradiation led to the complete disappearance of absorption above 550 nm.

Stepwise anaerobic photoreductions of the FAD prosthetic group of PfFNR were performed in the presence and absence of NADP+. Again, at variance with other FNRs, unliganded PfFNR was reduced to the hydroquinone level without any transient accumulation of FAD semiquinone (Fig. 3B). In the presence of an approximately equimolar amount of NADP+, during reduction the long-wavelength band extending over 800 nm, described above, was transiently seen. Again, no FAD semiquinone species were observed during the reaction (Fig. 3C). From the absorption spectra recorded in the latter type of experiment, the concentrations of the oxidized and reduced species of the two redox couples (PfFNRox/PfFNRred and NADP+/NADPH) were measured after each irradiation step. Nernst plots yielded Em values for the two-electron reduction of the FAD bound to PfFNR of −280 ± 2 mV at pH 7 and of −298 ± 4 mV at pH 8.

Rapid reaction study on the reductive

Steady-state kinetic analysis of the ferricyanide reductase activity of PfFNR suggested that the reductive half-reaction of the PfFNR catalytic cycle was probably responsible for the low value of kcat [13]. To test this hypothesis and to identify the catalytic step(s) actually involved, the process of reduction of PfFNR-bound FAD by NADPH was studied by stopped-flow diode array spectrophotometry. Experimental conditions were chosen in order to determine values that were directly comparable to data already published for plant FNRs. As a consequence, the solution medium of rapid reaction studies differed from that used in steady-state assays in composition, pH, and ionic strength. PfFNR (∼ 38 μm before mixing) was reacted at 25 °C with NADPH at concentrations ranging from 50 μm to 2 mm (before mixing). Time-resolved spectra were collected between 300 and 700 nm. The spectra of PfFNR reacting with 25 μm NADPH (after mixing) are shown in Fig. 4A. Within the instrument dead-time, the formation of a detectable amount of charge transfer complex between NADPH and enzyme-bound FAD (CT1) occurred, as judged from the high A550 nm of the spectra recorded immediately after mixing, and from the value of A450 nm, which was lower than expected for oxidized PfFNR. The observable part of the reaction consisted of a further bleaching of the flavin main absorbance band (at 450 nm) with a concomitant decrease in the absorbance band around 550 nm. Another spectral change occurring during this phase was a slight increase in the absorbance at wavelengths above 650 nm, indicating the accumulation of a small amount of CT2 at the end of the reaction (Fig. 4A). The analysis of a series of shots at different NADPH concentrations showed that the amount of CT1 observed after the dead-time increased with the concentration of NADPH, whereas the CT2 present at the reaction endpoint followed the opposite trend, being undetectable when the enzyme was reacted with reductant concentrations higher than 100 μm (Fig. 4B). We concluded that the observable part of the reaction occurring under NADPH excess corresponded to the conversion via hydride transfer of CT1 to the complex between reduced PfFNR and NADPH, with accumulation of the CT2 intermediate only when the concentration of NADPH was low, according to the following scheme:

Figure 4.

 Reductive half-reaction of PfFNR as studied by stopped-flow diode array spectrophotometry. PfFNR was reacted anaerobically with NADPH at 25 °C in 50 mm Hepes (pH 7.0). The enzyme concentration was ∼ 19 μm, and the NADPH concentration ranged from 25 μm to 1 mm (after mixing). (A) Reaction of PfFNR with 25 μm NADPH (after mixing). Spectra recorded 1 ms (red), 5 ms (green), 9 ms (brown), 13 ms (magenta), 21 ms (yellow) and 51 ms (blue) after mixing are shown. The spectrum of the oxidized enzyme mixed 1 : 1 with the above buffer is reported for comparison (black trace). (B) Reaction of PfFNR with 500 μm NADPH (after mixing). The spectra shown were recorded at the same reaction times as the traces shown in (A). The insets of (A) and (B) show enlargements of the portions above 520 nm of some of the spectra shown in the respective panels. (C) Time course of the same reaction shown in (B). The curve represents the equation of a single exponential decay fitted to the experimental data points, yielding a kapp of 122 ± 3 s−1. The inset shows the plot of the kapp value as a function of the NADPH concentration (after mixing).

According to this reaction mechanism, high NADPH concentrations would displace NADP+ from its complex with PfFNRred (CT2), by favoring the formation of the PfFNRred–NADPH complex. Absorbance traces at different wavelengths were monophasic and fitted well to single exponential decay equations, yielding the same apparent rate constant (kobs) for any NADPH concentration. Global fitting over the entire wavelength range confirmed that all of the observed spectral changes were due to a single process. A typical result of the analysis of an absorbance trace is shown in Fig. 4C. A plot of kobs as a function of [NADPH] gave an upper limiting value for this constant of 125 s−1 (inset of Fig. 4C). As shown in the same figure, kobs was found to be essentially independent of NADPH concentration when it is higher than 50 μm, suggesting a Kd for the PfFNR–NADPH complex in the lower micromolar range.

Interaction between PfFNR and PfFd

We have investigated the protein–protein interactions through different approaches. Carbodiimide-promoted crosslinking has been extensively used in the past [22–24]. Indeed, it allows the isolation of covalent complexes of the two proteins under study that can be further analyzed. Both T. gondii Fd and PfFd were crosslinked to PfFNR with 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC), yielding a single band of approximately 49.5 kDa in SDS/PAGE. This molecular mass value is consistent with a 1 : 1 stoichiometry of the proteins in the covalent complex. Apparently, the presence of NADP+ did not influence the crosslinking reaction course. In contrast to the spinach FNR–Fd cross-linked complex [22], the P. falciparum covalent complex does not gain the capacity to transfer electrons from NADPH to cytochrome c. These results could be explained by measuring the diaphorase activity of the reductase during the crosslinking reaction. Whereas PfFNR alone maintained more than 90% of its diaphorase activity after carbodiimide treatment, inactivation ensued when it was incubated with the crosslinker in the presence of PfFd. The inactivation was not due to protein–protein crosslinking, because full inactivation was reached with only 30–40% covalent complex produced. It is possible that EDC can modify a reductase essential residue(s) only when PfFNR forms a complex with PfFd.

It is well known that ligands of plant-type FNRs induce perturbations in their visible absorption spectra. The difference absorption spectra obtained by titrating PfFNR with PfFd allowed determination of the Kd of the protein complex. A positive difference spectrum was observed, as in the case of spinach FNR, with the same Δε of 2.8 mm−1·cm−1 at 460 nm [25]; however, the peak in the 300–400 nm wavelength range was less intense and defined. Several titrations were performed at pH 7, with the ionic strength of the medium being varied to allow extrapolation of the complex Kd value at 0 ionic strength (Fig. 5). This value falls in the nanomolar range. At a more physiological values of ionic strength of 50–100 mm, the Kd values range between 0.1 and 0.5 μm, which indicates high affinity between the two oxidized proteins.

Figure 5.

 Effect of the ionic strength on the interaction between PfFNR and PfFd. PfFNR titrations with PfFd were carried out at 16 °C in 50 mm Hepes/NaOH (pH 7.0) containing different concentrations of NaCl. Complex formation was quantified spectrophotometrically on the basis of the perturbations of the absorption spectra of prosthetic groups induced by PfFd binding to PfFNR. Upper panel: titration curves obtained in the presence of 50 mm (filled circles), 100 mm (open circles), 150 mm (filled squares), 200 mm (open squares) and 300 mm (filled triangles) NaCl. Curves represent the theoretical equations for 1 : 1 binding fitted to each experimental dataset by optimizing the Kd values. Lower panel: plot of the calculated Kd values of the PfFNR–PfFd complex as a function of the ionic strength of the medium.

We found that PfFd prevents the formation of the covalent dimer of PfFNR. This was shown by monitoring the inactivation of the diaphorase activity of the enzyme incubated in the presence of both NADP+ and PfFd for more than 1 day. The amount of dimer formed was negligible in comparison to that formed in the absence of PfFd (data not shown).


The catalytic properties of PfFNR will be discussed here using the well-characterized FNRs from both plants and T. gondii as reference, in order to highlight the peculiar properties of the malarial parasite enzyme. The catalytic efficiencies of PfFNR, in both the diaphorase and Fd–reductase reactions, were found to be significantly higher at pH 7 than at pH 8.2, the latter being the pH optimum for plant and T. gondii FNRs. The improvement of this kinetic parameter at lower pH is due to a substantial decrease in the Km values for both NADPH and Fd. Better binding of NADPH at pH 7 can be ascribed to the protonation of His286, which was shown to make contact with the 5′-phosphate of the 2′-P-AMP moiety of the nucleotide substrate (Fig. 6A) [13]. This His is a peculiar, Plasmodium-specific replacement for the aliphatic residue (usually Leu) present at this position in plastidic-type FNRs [26,27], including the T. gondii enzyme [5]. It is important to recall that the NADP-binding site of PfFNR differs substantially from that of typical plastidic-type FNRs. Indeed, PfFNR lacks the two highly conserved positive side chains that bind the 2′-phosphate moiety of NADP (Fig. 6A) [6,13]. PfFNR is also distinguished from other plastidic-type FNRs by the kcat values for both the diaphorase and the Fd–reductase reactions. Such differences cannot arise from a wrong choice of the first residue of the mature PfFNR polypepdide, because we have previously shown that the diaphorase activities of both spinach and T. gondii enzymes were not affected by the length of their N-terminus [9,28]. Furthermore, a longer PfFNR (starting at Leu38 of the precursor) was found to be even less catalytically active [14]. In all FNRs, the rate of the ferricyanide reductase reaction was shown to be limited by hydride transfer from NADPH to FAD. For this reason, we investigated this process in PfFNR by rapid kinetics. As expected, the kobs at 25 °C for hydride transfer from NADPH to FAD (250e eq·s−1) was comparable to the kcat of PfFNR in the NADPH–ferricyanide reaction. The impairment of the hydride transfer rate in PfFNR could tentatively be ascribed to the substantial alteration of the canonical NADP(H)-binding site of the malarial parasite FNR. It is possible that the peculiar interaction of the PfFNR with the adenylate moiety of NADP(H) could either slow down the entry of the nicotinamide ring into the active site or alter its positioning with respect to the isoalloxazine ring of FAD. A similar hypothesis has recently been proposed to explain the very low hydride transfer rate of the bacterial-type FNR of Rhodobacter capsulatus [21].

Figure 6.

 Crystal structure of the PfFNR dimer and details of the interaction between the enzyme and 2′-P-AMP. (A) Ribbon model of an enzyme subunit of the 2′-P-AMP-bound dimeric form of PfFNR (Protein Data Bank accession no. 2OK7, chain A) [13]. FAD, 2′-P-AMP, Cys99, which is involved in the covalent stabilization of the dimer, and the residues most directly involved in the binding of the ligand are shown as wire frames. The αF helix, which undergoes a large conformational transition upon ligand binding, is indicated. The FAD-binding and the NADP-binding domains of the protein are shown in yellow and blue, respectively. (B) Ribbon model of the dimeric form of PfFNR (Protein Data Bank accession no. 2OK7, chains A and B) [13]. The two subunits are shown in different colors. The two bound molecules of the substrate analog 2′-P-AMP and the disulfide bridge linking the Cys99 residues of the two subunits are indicated. The two FAD prosthetic groups can be seen as wire frame models in the central portion of the dimeric assembly.

No simple explanation seems to hold in the case of the low rate of steady-state PfFd reduction, where the kcat of PfFNR is five-fold lower than those of plant and T. gondii FNRs. In PfFNR, the bound FAD possesses an Em suited for electron flow towards the [2Fe–2S] cluster of PfFd [14]. Km and Kd values for PfFd are similar to those reported for other plastidic-type FNR/Fd couples [10,29,30]. On the other hand, FAD semiquinone is highly destabilized in PfFNR. It is thus possible that one-electron transfer to Fd is impaired because of the high energy required to generate the flavin semiquinone. Further studies are required to clarify this point.

We have further analyzed the specific property of PfFNR, unparalleled in other FNRs, to undergo dimerization in the presence of NADP+ or its analogs, with concomitant inactivation. We have already proposed a mechanism for this process, which provide a rationale both for the lack of catalytic activity of the dimeric protein and for the role of NADP+ or 2′-P-AMP in favoring the formation of the intersubunit disulfide bridge [6,13]. Briefly, binding of NADP+ or its analogs to monomeric PfFNR promotes an induced-fit transition consisting of the partial unwinding of the αF helix (Fig. 6A) and other minor conformational changes. In the ligand-bound form of PfFNR, the formation of a homodimer, in which the Cys99 residues from the two interacting subunits are placed close together, is highly favored. In the presence of O2, such thiols are converted to a disulfide that locks the dimer (Fig. 6B). We have previously reported that the disulfide-linked dimeric PfFNR is inactive as a diaphorase, because it adopts a conformation in which the active site is poorly accessible to the solvent (Fig. 6B) [13]. Here, we showed that dimerization fully abolished the Fd-dependent physiological activity of PfFNR. Besides oxygen, other physiological oxidants have been tested for their ability to promote dimerization. Both hydrogen peroxide and oxidized lipoic acid were able to yield high amounts of PfFNR dimer in relatively short times, especially in the presence of NADP+. Conversely, dihydrolipoic acid and, although less efficiently, GSH could convert the inactive dimer back to the active monomer. Whether this dimerization/inactivation is operating in the Plasmodium apicoplast remains to be checked. However, the fact that oxidants and reductants that are possibly present in the organelle actually promote in vitro monomer–dimer interconversion suggests that this process could be relevant in vivo.

We have produced the C99A variant of PfFNR and shown that it lacks the single titrable Cys observed in the wild-type enzyme under native conditions, thus identifying Cys99 as this residue. Furthermore, PfFNR-C99A possesses the same functional properties as wild-type PfFNR, with exception of the ability to undergo dimerization, which is completely abolished in the mutant. PfFNR-C99A is expected to represent a useful model of the wild-type enzyme in studies where the possible complications due to dimerization/inactivation of PfFNR need to be avoided. In particular, it will be a valuable tool in the screening of chemical libraries to search for PfFNR inhibitors, allowing discrimination between compounds that target its active site and compounds that cause its dimerization.

Experimental procedures

NADP+ and NADPH were purchased from Sigma. Horse heart cytochrome c (Sigma-Aldrich, Milano, Italy) was further purified by cation exchange chromatography on SP Sepharose (GE Healthcare, Milano Italy). Restriction endonucleases were obtained from GE Healthcare, and factor Xa from Pierce Biotechnology, Inc. (Rockford, IL, USA). All other chemicals were of the highest grade.

Plasmid construction

The cloning of the coding sequence of PfFNR (gene PFF1115w, located on chromosome 6) and the construction of the expression plasmid pET-PfFNR have already been reported [7]. In brief, the DNA encoding the putative mature PfFNR (starting from Lys56 of the protein precursor) was originally cloned from the genomic DNA of P. falciparum and inserted in the expression vector pET28b (Novagen, Merck KGaA, Darmstadt, Germany) in order to be expressed in E. coli as a cleavable fusion protein with an N-terminal poly-His-tag. Mutation of Cys99 to Ala was achieved by PCR using the QuikChange II site-directed mutagenesis kit (Stratagene, Agilent Technologies, Cernusco sul Naviglio, Milano, Italy) and the oligonucleotides 5′-CATATTAAAAAACAACGAGCTGCCAGATTATATT CTATATCC-3′ (sense primer) and 5′-GGATATAGAATATAATCTGGCAGCTCGTTGTTTTTTAATATG-3′ (antisense primer), which harbor the desired codon change (underlined bases). The DNA sequence encoding mature PfFd was amplified by PCR from pTUK–PfFd [5], using the oligonucleotides 5′-CCATGCCATGGCTTTATTTTATAATATAACATTAAGAAC-3′ (sense primer) and 5′-CCGGAATTCTTAATTCATTACATGTCGTG-3′ (antisense primer), which introduced NcoI and EcoRI restriction sites (underlined bases) at the 5′-end and the 3′-end of the coding sequence, respectively. The fragment was cloned between the same sites of pET28b (Novagen), yielding the plasmid pET–PfFd, which allows the synthesis of mature PfFd with no extra residues. The insert of all expression plasmids was entirely sequenced to exclude the presence of artefacts.

Overexpression and protein purification

For overproduction of PfFNR and PfFNR-C99A, Rosetta(DE3) (Novagen) E. coli cells harboring the plasmids pET–PfFNR and pET–PfFNR-C99A, respectively, were grown at 20 °C in 2× YT medium supplemented with kanamycin and chloramphenicol (30 mg·L−1 each) and induced with 0.1 mm IPTG for 24 h. The E. coli cell paste was resuspended in buffer A (50 mm NaH2PO4, pH 8, 500 mm NaCl, 5 mm imidazole, 1 mm phenylmethanesulfonyl fluoride, 1 mmβ-mercaptoethanol) at a ratio of 3 mL of solvent per gram of cell (fresh weight). After cell disruption by sonication, the clarified cell-free extract was loaded on an Ni2+–Sepharose high performance column (GE Healthcare) pre-equilibrated in buffer A. Extensive washing was performed with buffer A containing 20 mm imidazole and 10% glycerol. The enzyme was eluted with a 20–500 mm imidazole gradient under the conditions recommended by the resin manufacturer. The best fractions were pooled together, 5 mm EDTA was added, and ammonium sulfate was added to 15% saturation, for chromatography on a phenyl–Sepharose high performance column (GE Healthcare). Ammonium sulfate in a decreasing gradient of saturation from 15% to 0% in buffer B (50 mm Tris/HCl, pH 7.4, containing 10% glycerol and 1 mm dithiothreitol) was used to elute the enzyme. The best fractions were pooled and precipitated at 50% ammonium sulfate saturation. The enzyme was solubilized in buffer B and desalted on a HiTrap column (GE Healthcare). After incubation for 48 h at 16 °C with 1 : 500 (w/w) factor Xa to cleave off the His-tag, the mixture was loaded again on the Ni2+–Sepharose column. Owing to the moderate intrinsic affinity of PfFNR for the nickel ion, the enzyme was retained by the column even in the absence of the His-tag, and was specifically eluted with 20 mm imidazole. The FNR-containing fractions were concentrated by ultrafiltration, desalted by gel filtration on a PD10 column (GE Healthcare), and stored at −80 °C in 50 mm Tris/HCl (pH 7.4) containing 10% glycerol and 1 mm dithiothreitol. For Fd overproduction, E. coli strain Rosetta(DE3) was cotransformed with the pET–PfFd plasmid and the pRKISC plasmid. The latter plasmid (kindly provided by Y. Takahashi, Department of Biological Sciences, Graduate School of Science, Osaka University, Japan) harbors the E. coli isc gene cluster [31], which we found to improve the yield of holo-PfFd. The transformed cells were grown at 37 °C in Terrific Broth medium supplemented with 30 mg·L−1 kanamycin, 30 mg·L−1 chloramphenicol, and 10 mg·L−1 tetracycline. Induction was achieved with 0.1 mm IPTG for 3 h. PfFd was purified as previously described for recombinant spinach leaf FdI, omitting the final Sephadex G-75 gel filtration [16]. Purified PfFd was stored at −80 °C in 150 mm Tris/HCl (pH 7.4), under nitrogen.

Spectral analyses and steady-state kinetics

Absorption spectra were recorded either on an 8453 diode array (Agilent) or a Cary 100 double-beam (Varian, Leinì, Torino, Italy) spectrophotometer. The extinction coefficient of the protein-bound flavin of PfFNR was determined spectrophotometrically by quantitating the FAD released from the apoprotein following SDS treatment [32]. Steady-state kinetic parameters were determined for the K3Fe(CN)6 reductase, NADPH-INT reductase and the cytochrome c reductase (Fd-dependent) activities as previously described [9,33]. The concentrations of both the electron donor and the electron acceptor were independently varied within the following ranges: NADPH, 20–500 μm; K3Fe(CN)6, 0.04–1 mm; INT, 20–300 μm; and PfFd, 0.5–20 μm. Initial rate data were fitted to the equation for a ping-pong Bi-Bi mechanism by nonlinear regression using the software grafit 5.0 (Erithacus Software Ltd, Horley, UK).

Analysis of the interconversion between
monomeric and dimeric PfFNR forms

To study PfFNR dimer formation, 15 μm monomeric enzyme was diluted in 50 mm Tris/HCl (pH 7.4) containing 10% glycerol and incubated at 20 °C. The dimerization reaction was started by adding either 0.5 mm H2O2 or diamide as oxidant. When NADP+ was present, the concentration was 1 mm. To study restoration of the monomeric form by disulfide reductants, 7.5 μm gel filtration-purified PfFNR (see below) was incubated with either 50 μm dihydrolipoate or 100 μm GSH under the same conditions described above. Aliquots of the reaction mixtures were withdrawn and analyzed by both enzyme assay and SDS/PAGE. Before electrophoresis, 5 mmN-ethylmaleimide and 4% SDS were added to the samples. After 30 min of incubation at room temperature, samples were loaded on 12% polyacrylamide gels, omitting the addition of β-mercaptoethanol. In order to obtain dimeric PfFNR in purified form for further characterization, the enzyme was incubated for 36 h in the presence of NADP+ under the conditions described above, without the addition of oxidants. The PfFNR dimer formed through Cys99 oxidation by atmospheric O2 was isolated by gel filtration on a Superdex 75 10/30 column (GE Healthcare) in 20 mm Tris/HCl (pH 7.4) containing 100 mm NaCl.

Active site titrations and FAD photoreductions

Titrations of the PfFNR with PfFd were performed spectrophotometrically at 15 °C using a Cary 100 (Varian) double-beam spectrophotometer. The enzyme was gel filtered in 20 mm Hepes/NaOH (pH 7.0) containing 10% glycerol, and diluted in the same buffer to a final concentration of approximately 15 μm. Mixtures included also NaCl to vary the ionic strength of the medium. Difference spectra were computed by subtracting from each spectrum that obtained in the absence of ligand, corrected to account for dilution. Kd values were obtained by fitting datasets by nonlinear regression to the theoretical equation for 1 : 1 binding [34]. Stepwise reduction of the PfFNR was performed by the light–EDTA system [35] in anaerobic cuvettes at 15 °C. The enzyme was diluted to 15 μm in 50 mm Hepes/NaOH (pH 7.0) containing 10% glycerol, 13 mm EDTA, and 1.3 μm 5-carba-5-deazariboflavin. The solution was made anaerobic by successive cycles of equilibration with O2-free nitrogen and evacuation. When present, NADP+ was in a 1.2 molar ratio with respect to the enzyme. The latter type of photoreduction experiment allowed the estimation of the redox potential of PfFNR-bound FAD according to the procedure previously described [36], using the NADP+/NADPH couple as the redox indicator.

Rapid kinetics

Rapid reaction studies of the enzyme reductive half-reaction were performed under anaerobic conditions using a Hi-Tech Scientific SF-6I DX2 stopped-flow spectrophotometer, equipped with a diode array detector (300–700 nm). Enzyme (∼ 38 μm, before mixing) was reacted with various concentrations of NADPH at 25 °C in 50 mm Hepes/NaOH (pH 7.0). Absorbance traces at individual wavelengths were analyzed using kinetasyst 3 software (Hi-Tech Scientific, Bradford-upon-Avon, UK). Global analysis of multiwavelength kinetic datasets was performed with specfit/32 version 3.0 (Spectrum Software Associates, Chapel Hill, NC, USA).

Protein crosslinking

PfFNR (10 μm) and either 50 μm PfFd or T. gondii Fd were incubated at 25 °C with 5 mm EDC in 25 mm phosphate/NaOH (pH 7.0) in the presence and absence of 2 mm NADP+. At time intervals, aliquots were withdrawn from the reaction mixture, 100 mm ammonium acetate was added to quench the crosslinker, and analysis was performed by enzyme assay and SDS/PAGE on 12% polyacrylamide gels.


We thank M. A. Vanoni for assistance in performing rapid kinetics experiments and S. Baroni for technical help. This work was supported by a grant from Ministero dell’Istruzione, dell’Università e della Ricerca of Italy (PRIN 2004).