Purified PfFeSOD was resolved by electrophoresis for 3 hours at 100 V in a 10% native polyacrylamide gel. The gel was soaked in riboflavin and NADPH to generate superoxide radicals, and stained by Nitro Blue Tetrazolium (NBT) dye. NBT reacts with superoxide to form a blue precipitate when fixed in a gel. Any regions on the gel that exhibit SOD activity will therefore be white against a blue background. A clear white band coincident with the PfFeSOD protein indicated that the protein was active. Analytical ultracentrifugation (AUC) sedimentation equilibrium experiments estimated the molecular weight of the protein to be 47,400 ± 700 Da. Since the calculated molecular weight of the monomer is 23,799 Da, this indicates that PfFeSOD is a dimer, consistent with previously characterised FeSODs, which are either dimers or tetramers.

The crystal structure of PfFeSOD has been solved and refined against data extending to 2.5 Å spacing. The refined model consists of two polypeptide chains, A and B, which are related by a non-crystallographic two-fold axis of symmetry, together with two iron ions and 140 solvent water molecules. The electron density maps are contiguous from residue 1 to residue 197 of both chains. The C-terminal PfFeSOD residue (Lys¹⁹⁸), together with the eight residues of the C-terminal tag are not observed in the electron density maps. Details of the refinement statistics are given in Table 1. There is a solitary outlier in the Ramachandran plot at residue Lys⁴³ in the B chain, which occurs in a partially ordered surface loop. The Ser¹⁵-Pro¹⁶ peptide bond in both chains is in the cis conformation.

The overall structure is similar to those of previously solved iron superoxide dismutases, consisting of an amino terminal α-helical domain linked to a carboxy terminal α/β domain. The first eleven residues have an extended structure and are involved in crystal packing. The next 70 residues form two long anti-parallel α-helices (α1 and α3) connected by a meandering loop containing a third helix, α2. In both subunits, helix α1 has a noticeable kink at the conserved Lys²⁹ residue, a feature that has been noted before 13. This domain is connected via an extended loop to the α/β domain (Figure 2A), which comprises a three-stranded anti-parallel β-pleated sheet (strand order β3-β1-β2) surrounded by four α-helices. Superposition of equivalent Cα atoms of chain A and chain B of PfFeSOD gives a positional root mean squared deviation (rmsΔ) of 0.3 Å. The PfFeSOD structure superposes onto the molecular replacement model, E. coli FeSOD, with an rmsΔ of 0.55 Å over all equivalent Cα atoms.

The tertiary structure is very similar to that of a number of other Fe- or Mn-dependent SOD enzymes. Of particular interest is comparison with the tetrameric MnSOD found in human mitochondria. Subtle differences in the structures may allow the design of inhibitors to block the activity of FeSOD selectively without affecting the MnSOD. The arrangement of the active site residues is well conserved between the Fe and MnSODs; the only significant difference in main chain conformation arises in the loop connecting α1 and α3, which is remote from the active site and the dimer interface.

As is typical of SOD proteins, only a relatively small accessible surface area (9.7 %) becomes buried in the dimer interface. The main contributions to the dimer interface arise from helix α1, the loop that connects α5 to β1 and the loop connecting β3 to α6. 41% of the atoms forming the dimer interface are polar, while 59% are non-polar. There are five bridging water molecules within the dimer interface. There is an interesting pair of adjacent charge-charge interactions between the carboxylates of the Asp¹⁴⁰ residues and the side chain imidazoles of His¹³⁹. Perhaps associated with this, the dihedral angles of residue Asp¹⁴⁰ appear in a generously allowed location on the Ramachandran plot. These residues lie on what would be a classical β-turn were it not for the fact that residue 141 is an alanine rather than a glycine. All of the dimer stabilising interactions in the E. coli enzyme are conserved in PfFeSOD. Such high conservation of the dimer interface reflects the importance of dimerisation to the activity of dismutases.

The metal ion in both subunits of the enzyme is five-coordinate. Three bonds arise from coordination with histidine Nε2 atoms of residues His²⁶ (mean Fe-Nε2 distance 2.2 Å), His⁷³ (2.1 Å) and His¹⁶¹ (2.0 Å). A further bond is between the metal and the Oδ2 atom of Asp¹⁵⁷ (mean Fe-Oδ2 distance 1.9 Å). The fifth coordination position is an oxygen atom from either a water molecule or a hydroxide species (mean Fe-O distance 2.1 Å). The coordination geometry is trigonal bipyramidal, with His⁷³, His¹⁶¹ and Asp¹⁵⁷ providing the equatorial ligands and with His²⁶ and the hydroxide or water oxygen atom forming the axial ligands.

The area around the active site is fairly hydrophobic with two tryptophan residues (122 and 159) 5.4 Å and 6.1 Å from the metal, respectively. Trp¹²² forms a hydrogen bond to Gln⁶⁹, which in turn forms a hydrogen bond with the hydroxyl group of Tyr³⁴. Tyr³⁴ is also hydrogen bonded via a water molecule to His³⁰, which is bonded to Tyr¹³⁴ from the other subunit. This classical hydrogen-bonding network in superoxide dismutases may allow communication between the two metal sites 13. In the B chain however, we note that the Gln⁶⁹ Nδ to Tyr³⁴ -OH distance is longer (3.6 Å), perhaps due to coordinate error at this resolution. The spatial arrangement of the residues in the active site is otherwise identical to those of other Fe-dependent superoxide dismutases whose structures are known, consistent with the notion that structure and mechanism are highly conserved in this enzyme class.

A second line of communication between the active sites is afforded by the intermolecular ionic/hydrogen bonding interaction between the Fe-coordinating His¹⁶¹ and the carboxylate of Glu¹⁶⁰. The latter is situated in a loop that links two Fe-coordinating residues (Asp¹⁵⁷ and His¹⁶¹) in the opposing subunit. The relevance, if any, of structural communication between the active sites to the kinetics and mechanism of these enzymes is not established.

The sequence of the major FeSOD is highly conserved in different Plasmodia species (see Additional file 2) and is encoded by a gene on chromosome 8 (PlasmoDB code PF08_0071 14) consisting of a single exon. Another gene product (PfFeSOD2) with significant sequence homology is located on chromosome 6. This PfFeSOD2 gene is interrupted with multiple introns. The original gene model (MAL6P1.194 14) was revised by a bioinformatic analysis to maximise primary sequence homology to PfFeSOD, leading to a gene predicted to consist of nine exons, separated by eight introns with consensus splice donor/acceptor sites (see Additional file 3). Evidence supporting the model was found by Plasmodium interspecies conservation of the intron-exon structure, which is a good predictor for small exons 15 and subsequently confirmed by cDNA cloning 16. A striking feature of the PfFeSOD2 gene is that it encodes an N-terminal sequence typical of proteins that are targeted to sub-cellular organelles. The first two exons coincide precisely with elements of the bipartite signal (see Additional file 4) namely a signal sequence and a transit peptide. A GFP-fusion with the signal is directed to the mitochondrion, despite predictions from the sequence that it would be delivered to the apicoplast 16. However, the gene structure not only hints at its evolution by exon shuffling, but provides the opportunity for alternatively spliced forms 17.

A homology model (see Additional file 1) of the mitochondrially targeted PfFeSOD2 was derived using the cytosolic PfFeSOD structure solved here. The cytosolic PfFeSOD is the closest in sequence of any dismutase to the mitochondrial PfFeSOD and hence provides the most suitable coordinate set from which to construct a homology model. The model generated has an rmsΔ on all Cα atoms of 0.36 Å for the 388 matched residues. Unsurprisingly the secondary structure is predicted to be identical to that of the cytosolic FeSOD of Plasmodium. It appears that the active site and dimer interface are also well conserved. This is consistent across the Fe/MnSOD enzyme family, which is a highly conserved enzyme class. It remains to be seen if the two superoxide dismutases from Plasmodium are indeed structurally and biochemically equivalent, or whether the mitochondrially targeted SOD has a different turnover/stability that is required for function within a specific environment in the cell.

The cytosolic FeSOD coding sequence was amplified from P. falciparum (3D7) genomic DNA using Vent™ polymerase (New England Biolabs) and the oligonucleotide primers:

5'-CATGCCATGGTTATTACATTGCCCAAATTAAAG-3'

5'-CCGCTCGAGCTTTTGCATAGCTTTTTTTAAGTT-3'.

The 597 base pair PCR product was digested with NcoI and XhoI and ligated to similarly cut pET28a (Novagen). This places the coding sequence for PfFeSOD upstream of that for a non-cleavable, C-terminal His₆ tag. The expected expression product therefore comprises residues 1–198 of PfFeSOD with a C-terminal LEHHHHHH tag. The ligation products were introduced into competent E. coli NovaBlue™ cells (Novagen). Kanamycin-resistant transformants harbouring pET28-PfFeSOD were identified and isolated. Plasmid DNA from these cells was purified and subsequently introduced into the expression strain E. coli BL21-CodonPlus (Stratagene). An overnight culture of E. coli BL21/pET28-PfFeSOD was used to inoculate 1 litre of Luria-Bertani medium supplemented with 30μg ml^-1 kanamycin and cells were grown with shaking at 37°C to an OD₆₀₀ of 0.6 at which point expression of recombinant protein was induced by the addition of isopropyl-β, D-thiogalactopyranoside to a final concentration of 1 mM. After a further 4 hours incubation at 30°C, cells were harvested by centrifugation.

The harvested cell pellet was resuspended in 50 mM Tris-HCl buffer pH 7.5 containing 300 mM NaCl (Buffer A). The cells were lysed by sonication and clarified by centrifugation. The supernatant was mixed with 2 ml Ni-NTA agarose resin (Qiagen) at 4°C for 1 hour with gentle shaking. The resin was then washed with Buffer A containing 20 mM imidazole and protein was eluted with Buffer A containing 100 mM imidazole. Fractions were examined by Coomassie staining following SDS-polyacrylamide gel electrophoresis and those containing proteins of approximately 23 kDa corresponding to FeSOD were pooled and further fractionated by size exclusion chromatography on a Superdex 75 (Pharmacia) column run in 50 mM Tris-HCl pH 7.5 containing 100 mM NaCl. Typically 10 – 15 mg of protein was obtained per litre of cell culture.

Sedimentation equilibrium experiments were conducted on a Beckman Optima XL/I analytical ultracentrifuge, in an AN-50Ti rotor. The buffer density (300 mM NaCl, 50 mM Tris-HCl, pH 7.5) and the partial specific volume of the protein were estimated to be 1.012 g.ml^-1 and 0.7302 respectively, using the program SEDNTERP 24. Protein homogeneity was examined on an 8.75 % native polyacrylamide gel. Enzyme activity was confirmed using a gel-based Nitro blue tetrazolium assay 25.

Crystallisation experiments were carried out by the hanging-drop vapour-diffusion method. Drops contained a 1:1 volume ratio of PfSOD at 10 mg ml^-1 in 50 mM Tris-HCl pH 7.5, 100 mM NaCl and a range of precipitants. Rod shaped crystals appeared within a week at 18°C from reservoir solutions that contained 0.1 M Tris-HCl pH 7.5 and 30–40% (w/v) polyethylene glycol (PEG) 600.

Crystals were mounted in a fine rayon loop and transferred into a solution containing 0.1 M Tris-HCl (pH 7.5) and 40 % PEG 600 which acts as a cryoprotectant. They were then flash-cooled in liquid nitrogen. Preliminary in-house X-ray analysis of these crystals showed that they belong to the space group P2₁2₁2₁ with unit cell dimensions a = 55.5 Å, b = 78.4 Å, c = 89.3 Å. A complete X-ray diffraction data set was collected at 100 K, using synchrotron radiation at a wavelength of 0.976 Å, on beamline ID-29 at the ESRF, Grenoble, using an ADSC Quantum4 CCD detector. The data were processed using the programs DENZO and SCALEPACK 26.

Calculations were performed using the CCP4 suite of programs 27. Molecular replacement was performed with AMoRe 28 and refinement conducted with REFMAC 29. The structure was determined by the molecular replacement method using the Escherichia coli FeSOD coordinate set (PDB entry code 1ISC) as a search model. EcSOD and PfSOD share 51% sequence identity. Molecular replacement calculations produced two clear solutions related by a non-crystallographic 2-fold symmetry axis, and giving an R-factor of 40.4 % after rigid body refinement. Model building was performed using the X-AUTOFIT routines 30 implemented in the molecular graphics programme QUANTA (Accelrys, San Diego). During refinement, tight non-crystallographic symmetry restraints were imposed on the main chain atoms and medium restraints on the side chains of the two molecules in the dimer.

A homology model of the mitochondrial FeSOD (PfFeSOD2) was generated on the basis of 52% sequence identity over 188 residues to the known structure of the cytosolic PfFeSOD. All structural visualisation/manipulation was performed using QUANTA. The sequences of PfFeSOD2 and the PfFeSOD template structure were aligned using CLUSTALX 31, and manually corrected for certain regions, based on interrogation of the known structure.

Homology modelling was performed using MODELLER6 32. Ten initial models were generated, and the model with the lowest objective function chosen as the representative PfFeSOD2 structure. After manual checking of backbone and side chain positions, energy minimization was performed using CHARMM 3334 to relax steric clashes within the model. Stereochemical evaluation was then performed using PROCHECK 35. The model possesses good stereochemical quality, with 88.4 % of residues in the most favoured regions of the Ramachandran map, and 10.4 % in additional allowed regions. There are four residues in generously allowed or unfavourable regions, which accurately reflect those in the template structure.