Superoxide dismutases (SODs, E.C. 1.15.1.1) are the superfamily of metalloenzymes that dismutases the highly toxic and reactive superoxide radical (O₂ ^-, by-product of aerobic metabolism) through a cyclic oxidation-reduction ('ping-pong') mechanism. As described by McCord and Fridovich 1, it is the first line of defense to alleviate oxidative stress virtually in all living organisms that survive in oxic environment.

The evolutionary trajectory has favored SOD as a ubiquitous enzyme in multiple forms within a single organism or cell, indicating a fail-safe redundancy that emphasizes the importance of this family of enzymes against reactive oxygen species (ROS). Based on metal cofactors, four known (canonical) isoforms viz., iron (Fe), manganese (Mn), copper/zinc (Cu/Zn) and nickel (Ni) SODs have been identified. In general, SODs have a strict metal binding specificity for enzymatic activities with the exception of a class of enzymes which show enzymatic activity regardless of whether Fe or Mn is bound at the active site; these are known as cambialistic forms 2345.

Cyanoprokaryotes are oxygen evolving photosynthetic organisms occupying a crucial position between pro- and eukaryotes. They are considered to be primeval having evolved about 3.2 billion years ago 6. In addition, they succeeded in linking photosynthetic electron flow from water as the photoreductant through an oxygen-evolving complex at the high-potential side of the newly elaborated photosystem II, which is thought to have originated from a uniform primordial photosystem by gene duplication 7. The resultant tandem operation of two photosystems is now known as oxygenic or plant-type photosynthesis 8. This marked the turning point in the evolution of earth, opening up the era of an aerobic, oxygen-containing biosphere and SOD is found to play a critical role in mitigating the toxic effect of superoxide ion. The first implication on the protective role of cyanobacterial SOD in photo-oxidative damage was shown in Anacystis nidulans 9. Subsequently, several studies on protective role of SODs of cyanobacteria in response to various physiological processes/stresses like photosynthesis 10, desiccation 1112, chilling 13, nitrogen starvation 14 and with azo dyes (unpublished) have been reported.

Metal preferences in Fe and MnSODs have been well documented in both pro- and eukaryotic forms 151617. However, no information is available on distinguishing the canonical isoforms of cyanobacteria. Hence, the present study focuses on structure and sequence pattern of subsets of cyanobacterial SODs to explore the possibility of solving the ambiguity.

For the survival of cyanobacteria with oxygenic photosynthesis, the selection pressure led to the evolution of SODs as the first antioxidant arsenal against nascent oxygen species. Studies on cyanobacterial SODs would serve as a window into the past and present evolutionary events of these primitive phototrophs.

On comparison, the canonical isoforms of SOD, Fe and MnSOD's are structurally distinct from Cu/Zn and NiSOD. Both Fe and MnSOD are typically homodimers or tetramers (Fig 1A,C) sharing identical metal chelating residues at the active site with a high degree of sequence and structural homology except for slight differences in amino acid residues. For instance, the amino acid range in cyanobacterial FeSOD is 199–229 residues with a molecular weight of 21–25 KDa, whereas in MnSOD, it is 200–316 amino acids with a molecular weight of 22–34 KDa.

Both SODs revealed a common topology with all α N-terminal (Pfam:PF00081) and a α/β C terminal domains (Pfam:PF02777) (Fig 1B,D). The sequence pattern for Fe and MnSODs of eukaryotes and other non-cyanobacterial prokaryotes is D-X-[WF]-E-H-[STA]-[FY]-[FY] 18; whereas, the analysis of the sequence conservation in cyanobacteria (based on available data) showed a specific motif DVWEHAYY [D282-Y289, based on Fig 2]. This motif extends between the second α-helix and the first β-sheet of the C-terminal domain in both the SOD's. The highly conserved residues aspartate D282 and histidine H286, a constituent of the motif are the metal binding ligands. In addition, glutamic acid E285 and tyrosine Y289 form a dimer surface spanning the interface and bridging the active sites between the opposite halves of each subunit, see Figure 2 (For full image, please see Additional file 1).

Structural analysis of available cyanobacterial Fe and MnSODs, confirms that both share a similar active site (i.e., metal ion) being coordinated in the respective isoform by three histidine and an aspartate residue with a ligating solvent molecule (water or OH), a five coordinated trigonal bipyramidal geometry. In Thermosynechococcus elongatus (PDB code 1my6); the Fe ion is coordinated by the carboxylate oxygen (Oδ2) of D161 with the amino group (Nε2) of H79, 27, 165 along with the oxygen atom of the water molecule. The hydrogen bonding distance between Oδ2 (D161) and Nε2 (H27 and H79) is 2.79Å and 3.27Å respectively (Table 1). In case of Anabaena sp (PDB code: 1gv3), the Mn is coordinated by Nε2 of H117, 204, 62 and Oδ2 of D200. The hydrogen bonding between Oδ2 (D200) and Nε2 (H62 and H117) is 2.19Å and 3.33Å respectively. These hydrogen bonds are involved in stabilizing the orientation of the ligand residues in MnSOD 8. The observed contact surface area (31–35 Ų) between the side chain aspartate oxygen atom (Oδ2) and histidine (Nε2) implies that the metal coordination ligands in the exposed region may perhaps tune the redox potential (Fig 3, 4).

The motif and metal binding sites of Fe and Mn isoforms appear to exhibit similar function. However, the sequence alignment and structural analysis reveal their possible discrimination by three traits to specifically differentiate Fe and Mn isoforms (Table 1 Additional file 1).

First, is the change in conserved amino acid signature F184X₃A188Q189_.......T280_......F/Y303 in Fe being replaced by R184X₃G188G189_.......G280_......W303 in MnSOD (see Figures 2 and 5).

The second notable feature is related to the metal bound solvent molecule that serves as a hydrogen bond to the non-coordinated oxygen of the carbonyl group of the aspartate ligand accepting a hydrogen bond from an outer sphere residue 19. In MnSOD, it is glutamine Q262 (Fig 2) arising from the end of the β₂-strand and H ₉ in the C-terminal domain, while in FeSOD, it is tryptophan W243 arising from the middle of the sequence (within the β₁) in the C-terminal domain. In the case of cambialistic Fe/MnSOD metalloform reported in archaea (Pyrobaculum aerophilum) 19, the outer-sphere H-bonding residue is histidine. This residue plays a major role in altering the solvent interaction with the active site metal ion in cambialistic Fe/Mn SOD isoform 19. The sequence analysis of cyanobacterial SODs showed the absence of this histidine residue which probably suggests the absence of cambialistic forms in cyanobacteria. Vance and Miller 20 reported that the most highly conserved residues glutamine Q262 in Mn and Q189 of FeSOD forms the outer sphere hydrogen-bond network exerts a large influence on redox midpoint potential tuning for catalytic activity of SOD's.

The third difference is the presence of two lysine residues, K201 and 255 in FeSOD but not in MnSOD (Fig 2 and 5). These residues seem to be unique and function specific to cyanobacteria among prokaryotes 21. K201 lines a small pit at the surface of the T. elongatus and of higher plants FeSOD, formed by the loop P202-G203-G204 connecting N and C terminal domains. Likewise, K255 is restricted only to cyanobacteria, indicating its importance in the photosynthetic context 21.

Cyanobacterial MnSOD is the only SOD to be membrane anchored by transmembrane helix 22. The factor that determines localization of MnSOD is found to span the N terminal which is a hydrophobic transmembrane helix (Fig 1D, 6). The cyanobacterial representatives such as (Synechococcus sp. WH5701 (EAQ76095), Synechococcus sp. RS9917 (EAQ68777), Trichodesmium erythraeum IMS101 (EAO27349), Anabaena variabilis ATCC29413 (ABA21068) and Nostoc sp. PCC7120 (BAB77594)) clearly corroborate this (Fig 6).

Cyanobacterial Cu/ZnSOD isoform bears no resemblance to Fe or Mn or Ni isoform in relation to its primary and tertiary structure. The theoretical molecular weight ranges between 16–23 KDa with an amino acid length of 174–233 residues. Further, study on amino acid composition illustrates that it is rich in Gly (11–16%) forming eight β-sheets (Fig 7A) accredited to be involved in conformation 23 and stability in repeated freeze/thaw cycles and prolonged refrigeration 9. These isoforms in general have a copper containing domain (Pfam:PF00080) with two different signatures. The first is G-F-H-[ILV]-H-x-[NGT]-[GPDA]-[SQK]-C where the conserved histidine is involved in copper binding, and the second being G-[GA]-G-G-[AEG]-R-[FIL]-[AG]-C-G where C is involved in disulfide bonding (Fig 8). G. violaceus SOD (NP_925116, NP_924927) annotated as 'similar to SOD' contains only copper binding domain and both the signatures are absent. Further confirmation requires additional structural data. Each monomer is comprised of a binuclear metal centre with one Cu and one Zn atom. The noticeable β parallel fold of cyanobacterial Cu/Zn isoform mimics the structure of Salmonella typhimurium Cu/ZnSOD 24 (Fig 7B). The catalytic coordination sphere of Cu²⁺ ion is by Nδ1 of H103, Nε2 of H105, H147 and H215 and Zn²⁺ by Nδ1 of three H147, 157, 171 and Oδ1 of one D174 (Fig 8). Besides this, structural comparison designates the two specific hydrogen bonds between the Zn²⁺ coordinating residues D174-Oδ1... H157-Nδ1 (3.25 Å) and D174-Oδ1... H171-Nε1 (3.18 Å) to ligand stability.

The fourth canonical form NiSOD is a hexamer (Fig 9A) found only in cyanobacteria 25 and Streptomyces 2627 with amino acids ranging from 140–163 and molecular weight between 15–18 KDa. Analysis of available sequences and complete genome sequences revealed that, unicellular Prochlorococcus forms possess only NiSOD, whereas, multicellular filamentous heterocystous and heterotrichous forms lacks this isoform (Table 2). The key for the ubiquity of NiSOD in Prochlorococcus may be due to the primitive photosynthetic machinery and its smallest genome size (between 1669–2434 Kb) by gene rearrangement or loss to maximize the energy economy 28. The sequence conservation, motif with eleven-residues (HCDGPCVYDPA) in N-terminal region of Ni-hook, along with a nickel containing SOD domain (Pfam:PF09055) forms an unique pattern to identify cyanobacterial NiSOD. Cyanobacterial NiSODs seem to have an assembly of four alpha helices bundle with a short connecting alpha helix, as that of Streptomyces sp. (Fig 9B). The catalytic Ni ion of cyanobacteria is very much analogous to the reported square planar active center with thiolate (C2, based on 1t6u), backbone nitrogen (H1 and C6) ligands and of square pyramidal Ni (II) with an added axial His₁ side chain of Streptomyces sp. 29.

The analysis is based on 64 cyanobacterial SODs available to date in public databases. Among them 2 are described as Fe/Mn, 4 as Cu/Zn and Mn precursor, 16 as putative NiSOD, 11 annotated as Fe, Mn and Cu/Zn isoforms, 29 as possible/putative SOD and 2 as hypothetical proteins.

Thus the present study resolves the incompletely annotated SODs among cyanobacteria (Table 2). Further, 64 cyanobacterial SOD sequences are clearly categorized into 17 NiSOD, 7 Cu/ZnSOD, 24 FeSOD and 14 MnSOD genes, 2 non assignable as they require further structural data. The strict metal specificity, precise sequence and structure among the metalloforms led to discriminate Mn and FeSOD (Table 1). The highly homologous Fe and MnSODs shares a metal binding motif DVWEHAYY without any variation, compared to D-X-[WF]-E-H-[STA]-[FY]-[FY] found in other pro – and eukaryotes.

The whole genome sequences analyses of cyanobacteria reveal that the primitive unicellular Prochlorococcus with simple photosynthetic apparatus possesses only NiSOD. The more evolved middle order forms of cyanobacteria posses a combination of Fe and Ni or Fe and Mn SODs. The most evolved filamentous, heterotrichous and heterocystous forms predominantly have only Fe and Mn metalloforms. However, CuZn also occurs rarely (Table 2).

The non-redundant database of protein sequences (National center for Biotechnology Information, NIH, Bethesda) were retrieved using the PHI-BLAST 30 search tool using BLOSOM 62 matrix with gap penalities (Existence – 11 and Extension – 1) with a threshold value of 0.005 and optimal limit for cyanobacteria. The query sequence used were Synechococcus sp. JA-3-3Ab with Expasy-PROSITE pattern D-x-[WF]-E-H-[STA]-[FY]2 for Fe/MnSOD; Synechococcus sp. RSS9916 with signature 1 [GA]-[IMFAT]-H-[LIVF]-H-{S}-x-[GP]-[SDG]-x-[STAGDE] and signature 2 (G-[GNHD]-[SGA]-[GR]-x-R-x-[SGAWRV]-C-X(2)-[IV]) for Cu/ZnSOD. In addition, the individual sequences of all the SOD metalloforms were also manually retrieved from public databases (NCBI, KEGG). Identical sequences from the same organism were removed manually. Intoto, 64 sequences representing 24 complete genomes and individual submissions obtained are listed in Table 2 together with the accession numbers and the organisms. Identification of domains associated with SOD proteins were realized using NCBI Conserved Domain Search and Pfam servers

The secondary structure consensus was carried out using nnPREDICT 31 and JPRED 32 for each protein to refine the multiple sequence alignment. Multiple alignments for cyanobacterial Fe and MnSODs; and Cu/ZnSOD sequences were generated using the Clustal W (neighbor-joining) of BioEdit V.7.0.5 33 program. Default parameter for both the alignments was gap initial penalty- 8 and gap extension penalty of 2. The alignment was fixed under the PAM40 series protein-weight matrices in both the cases. The sequence alignments were displayed graphically using BIOEDIT package 28 with a threshold of 95% consensus residue shading.

Representative crystal structures of available cyanobacterial FeSOD (1my6-Thermosynechococcus elongates BP-1) and MnSOD (1gv3-Anabaena sp. PCC7120) with exception for NiSOD (1t6u-Streptomyces coelicolor) and Cu/ZnSOD (1eqw-Salmonella typhimurium) were retrieved from PDB. The 3D structures were analyzed using SWISS-PDB viewer 34 and graphical representations were done with WebLab viewer lite (V.4.2)

BP and JP contributed equally in carrying out the sequence analysis studies and participated in the sequence alignment. RTD carried out further confirmation of the results and helped BP in visualization of the structures. TS helped in carrying out the structural comparison. LU and DP participated equally in the study, its design and coordination. GS helped in fine tuning of the manuscript. All authors read and approved the final manuscript written by BP.