Introduction

Superoxide dismutases (SODs, EC 1.15.1.1) are a group of ubiquitous metalloenzymes that catalyze the dismutation of toxic superoxide anions to molecular oxygen and hydrogen peroxide (McCord and Fridovich 1969; Fridovich 1995). In concert with catalase, SOD eliminates reactive oxygen species from cells. SODs have been reported to possess a potentially important role in host invasion and development of pathogenic fungi (Belinky et al. 2002; Giles et al. 2005) and Microsporidia (El-Taweel et al. 2007), but the exact mechanism has not been clearly demonstrated. SODs are classified into three main groups according to the type of metal cofactor: manganese SODs (MnSODs) are found in prokaryotes and in the mitochondrial matrices of eukaryotes; iron SODs (FeSODs) are found in prokaryotes, protists and plants; and copper/zinc SODs (Cu/ZnSODs) are found in bacteria and the cytosol and extracellular compartments of eukaryotes (Bordo et al. 1994; Brouwer et al. 2003). FeSODs and MnSODs resemble each other in their sequence and structure (Parker et al. 1987), suggesting a common ancestry (Hassan 1989), but they are unrelated to Cu/ZnSODs (Tainer et al. 1982; Smith and Doolittle 1992). Cambialistic SODs are those that function efficiently with both manganese and iron (Martin et al. 1986) and are regarded as a transitional form between FeSODs and MnSODs.

Similarity among MnSODs suggests that these enzymes have a common ancestry (Fitch and Ayala 1994), but their distribution among species is variable. Studies have shown that most organisms contain only a single MnSOD gene (Bagnoli et al. 1998; Kliebenstein et al. 1998; Babitha et al. 2002; Fang et al. 2002; Regelsberger et al. 2002), and only a small number of organisms, such as Zea mays (Zhu and Scandalios 1993), Nicotiana tabacum (Van Camp et al. 1997), Callinectes sapidus (Brouwer et al. 1997), Caenorhabditis elegans (Hunter et al. 1997), and Prunus persica (Bagnoli et al. 2002), have been found to have multiple MnSODs. To date, however, there has been no report of a tandem duplication of the MnSOD gene.

Each MnSOD monomer adopts an α/β fold, and combines to form an oligomeric structure in solution (Stallings et al. 1984; Jackson and Cooper 1998). MnSODs in eukaryotes are generally tetrameric (Natvig et al. 1996; Palma et al. 1998; Alscher et al. 2002) while in bacteria, MnSODs are typically dimeric (Natvig et al. 1996; Alscher et al. 2002), although tetramers (Wagner et al. 1993) have been reported. In prokaryotes, MnSODs are usually associated with the cytosol (Lynch and Kuramitsu 2000), though they have also been found in the thylakoid and plasma membranes of cyanobacteria (Regelsberger et al. 2002). In eukaryotes, MnSODs are usually synthesized in the cytosol and imported post-translationally into the mitochondrial matrix (Weisiger and Fridovich 1973; Zhu and Scandalios 1993; Natvig et al. 1996; Alscher et al. 2002). Some MnSODs, such as those from Callinectes sapidus (Brouwer et al. 1997), Ganoderma microsporum (Pan et al. 1997), Penicillium chrysogenum (Diez et al. 1998), and Candida albicans (Lamarre et al. 2001) are targeted to the cytosol.

Microsporidia are a group of obligate intracellular eukaryotic parasites that infect a wide variety of species, including humans (Desportes et al. 1985; Canning 1993; Snowden 2004). Members of this phylum possess many unusual characteristics, including a prokaryotic type rRNA gene arrangement with the 5.8S rRNA homologue attached to the large subunit rRNA (Vossbrinck and Woese 1986) and a highly divergent small subunit rRNA (ssrRNA) gene. Comparative analysis of the ssrRNA led to the view that Microsporidia may represent an early branch of the eukaryotes (Vossbrinck et al. 1987). However, some recent studies have suggested that Microsporidia emerge within the fungi (Thomarat et al. 2004; Gill and Fast 2006; James et al. 2006), and that long-branch attraction (LBA) (Felsenstein 1978) explains the original placement of Microsporidia at the base of the eukaryotes (Germot and Philippe 1999). Another possible “primitive” feature of the Microsporidia is the lack of mitochondria. It has been shown, however, that Microsporidia retain a mitosome, which is a double-membrane-bound structure that is thought to be a remnant of the mitochondria (Williams et al. 2002). The very small genome of Microsporidia (the smallest known genome among eukaryotes) could also be considered a primitive trait but, based on an analysis of the completed genome of another Microsporidia, Encephalitozoon cuniculi, it has been suggested that the small size of the microsporidial genome is a recently acquired feature (Keeling et al. 2005).

Genomic analysis of E. cuniculi has shown that this species uses a unique MnSOD to deal with oxidative stress (Katinka et al. 2001). Our analysis shows that the MnSOD from the microsporidian Nosema bombycis, a parasite of the silk moth Bombyx mori, is closely related to that of E. cuniculi. In addition, the MnSOD shows a tandem duplication event in N. bombycis. Here we predict the structure and the subcellular localization of the two MnSODs from N. bombycis and perform a comparative phylogenetic analysis on a wide array of MnSODs.

Materials and Methods

Obtaining SOD Sequence Data

Nearly two hundred thousand random shotgun reads have resulted in a 7.8-fold genomic database as part of the N. bombycis genomic sequencing project in our laboratory. N. bombycis isolate CQ1, purified from infected silkworms in Chongqing, China, is contained in the China Veterinary Culture Collection Center (CVCC number 102059) (Xu et al. 2006). All initial reads were assembled using the RePS program (Wang et al. 2002). The N. bombycis manganese superoxide dismutase (NbMnSOD) genes were located by the GLIMMER 3.0 gene sequence prediction program (Delcher et al. 2007). Two MnSOD genes were identified as a tandem repeat and designated here as NbMnSOD1 and NbMnSOD2.

Using the two NbMnSODs in a BLAST (Altschul et al. 1990) search, 53 homologous amino acid sequences were retrieved from the NCBI database. In addition, 18 fungal sequences were obtained from NCBI based on previously published information (Frealle et al. 2006). Accession numbers of all amino acid sequences are indicated in the Supplementary Table S1. The nucleotide sequences determined in this study for the N. bombycis SODs have been deposited in Genbank under accession numbers FJ377710 for NbMnSOD1 and FJ377711 for NbMnSOD2.

RT-PCR

Total RNA was extracted from purified spores of N. bombycis using the Trizol reagent (Invitrogen, USA), according to the protocol provided by the manufacturer, and treated using RNase-free DNase I. Single-strand cDNA (sscDNA) was synthesized using M-MLV Reverse Transcriptase following the recommended protocols (Promega, USA). The RNA was subsequently digested using RNase H (MBI Fermentas).

The specific primers were designed by Oligo6 software (Molecular Biology Insights, Inc., Cascade, CO, USA). Primer sequences are as follows:

  • NbMnSOD1F: 5′-GGACTTATGAAAGAAATGGA-3′ and

  • NbMnSOD1R: 5′-AATTACTAACGTATTCAGGC-3′, with a 268-bp predicted product;

  • NbMnSOD2F: 5′-ATTTGGCAGTTGATCTTAGG-3′ and

  • NbMnSOD2R: 5′-AACTTCTTTCCTTCAACTAC-3′, with a 337-bp predicted product.

Gene amplification was carried out using both sscDNA and genomic DNA (as a control) templates.

Prediction of Subcellular Targeting

All superoxide dismutase proteins were analyzed for the presence of an amino-terminal extension of 20–40 residues targeting the protein for transport either within or out of the cell. Targeting sequence prediction programs: PSORT II (Nakai and Horton 1999), TargetP (Emanuelsson et al. 2000), and MitoProt II (Claros and Vincens 1996) were used.

Sequence Alignment

To define the structures and perform phylogenetic analysis, the amino acid sequences of 72 SODs were aligned using Muscle software (Edgar 2004). The alignment was then adjusted using the SOD structures determined from X-ray diffraction data for 3 of the MnSODs and 8 of the FeSODs (Wintjens et al. 2004). These structures are the following: MnSODs from Homo sapiens mitochondria (Protein Data Bank accession number 1n0j; X-ray resolution 2.20 Å; MnSOD; tetramer (Borgstahl et al. 1992)), Aspergillus fumigatus (1kkc; 2.00 Å; MnSOD; tetramer (Fluckiger et al. 2002)), Escherichia coli (1d5n and 1isa; 1.55 and 1.80 Å; MnSOD and FeSOD, respectively; dimer (Lah et al. 1995; Borgstahl et al. 2000)), Porphyromonas gingivalis (1qnn; 1.80 Å; FeSOD; dimer (Sugio et al. 2003)), Sulfolobus acidocaldarius (1b06; 2.20 Å; FeSOD; tetramer (Knapp et al. 1999)), Sulfolobus solfataricus (1sss; 2.30 Å; FeSOD; tetramer (Ursby et al. 1999)), Aquifex pyrophilus (1coj; 1.90 Å; FeSOD; tetramer (Lim et al. 1997)), Propionium freudenreichii subsp. shermanii (1bsm; 1.35 Å; FeSOD; tetramer (Schmidt 1999)), Mycobacterium tuberculosis/vaccae (1ids; 2.00 Å; FeSOD; tetramer (Cooper et al. 1995)), and Pseudomonas ovalis/putida (1dt0; 2.10 Å; FeSOD; dimer (Bond et al. 2000)).

Hydrophobic Cluster Analysis

Sequences of 12 SOD proteins (6 tetramers, 4 dimers, and the 2 N. bombycis SODs) were compared using Hydrophobic Cluster Analysis (HCA) and an HCA plot was drawn with the drawhca program (Woodcock et al. 1992). HCA was developed to show similarities among protein structures when amino acid sequence similarities are low by comparing the placement of hydrophobic regions in the protein. For proteins which have a greater than 50% sequence homology to one with a known X-ray structure, direct structural inferences can be made. At the same time, it has been shown that three-dimensional structural similarities based on X-ray structural analysis can be seen among proteins which have as little as 18% amino acid similarity. HCA can show similarities among proteins for which there is no X-ray data. HCA compares hydrophobic regions between two proteins while marking regions with prolines and glycines as representing possible loops and positions cystines to allow disulphide bonds. In addition to representing similarities visually, HCA can be used to obtain a numerical homology score between two proteins. This is done by comparing the number of hydrophobic residues that are in correspondence between two sequences as follows:

$$ {\text{HCA homology score (\%)}} = {\frac{{ 2 {\text{CR }} \times { 100}}}{{{\text{RC1 }} + {\text{ RC2}}}}} $$

where RC1 and RC2 are the number of hydrophobic residues in each protein and CR is the number of hydrophobic residues which are in correspondence between the two proteins (Gaboriaud et al. 1987). HCA shows an 80% homology in the hydrophobic regions between human hemoglobin α-chain and lupine leghemoglobin, although the two proteins have less than 15% sequence identity.

Analysis of Selection

We were interested in knowing whether the two NbMnSODs have evolved under different selective constraints. Ka/Ks analysis is based on the fact that synonymous (Ks) or silent mutations in the coding regions (exons) of genes are much more frequent than non-synonymous (Ka) mutations which result in an amino acid change in the protein. The analysis involves comparing the same genes from two organisms. Ka values smaller than Ks values (Ka/Ks < 1) imply that a gene is under purifying selection and may be functional and under constraint not to change at the amino acid level. Ka values closer to Ks values (Ka/Ks = 1) implies that the gene in question is either not a functional gene or that it is not constrained and is undergoing more rapid change at the structural level. Ka values greater than Ks values (Ka/Ks > 1), indicate that positive selection may be acting on the gene and that a new functionality may have evolved. Nucleotide sequences were aligned to correspond to the amino acid alignment described above and shown in Fig. 3. Ka/Ks ratios were calculated by the 2p method of (Kimura 1980) using the K-Estimator software (Comeron 1999). Because the A. locustae MnSOD sequence was incomplete, nucleotide sequences from the C-domain of the N. bombycis, E. cuniculi, and A. locustae MnSODs were compared. We also analyzed the selection of each individual codon for three complete MnSOD sequences (both NbMnSODs and EcMnSOD) using both the modified branch-site model A and site models M1a, M2a, M7, and M8 of the CODEML program from the PAML package (Yang 1997). Posterior probabilities were calculated using the Bates empirical Bates (BEB) method (Yang 1997). In order to investigate whether the rate of evolutionary change was greater for one N. bombycis MnSOD gene than for the other, relative-rate tests were carried out using the RRTree program (Robinson-Rechavi and Huchon 2000), with the MnSOD sequence of E. coli (GenBank accession number P00448) as an outgroup.

Phylogenetic Analysis

The phylogenetic analysis included all 72 SOD amino acid sequences from the alignment described above and given in Supplemental Table S2. The most suitable substitution model, WAG with an alpha of 0.80, was determined by ProtTest software (Abascal et al. 2005). Phylogenetic analysis was done using all sites of SOD sequences using the maximum-likelihood Phyml software (Guindon and Gascuel 2003). Bootstrap values were obtained using 500 replicates. In order to check for LBA artifacts, the method of long-branch abstractions (Siddall and Whiting 1999) was used.

Results

Tandem Nosema bombycis MnSODs and Their Genomic Context

A search of the N. bombycis genomic dataset revealed a tandem array of two MnSOD genes. Each gene has a length of 678 bp, coding for 225 amino acid residues, with an intergenic region of 788 bp separating the two genes. Figure 1 shows a 9,227 bp region at one end of the superscaffold, which contains NbMnSOD1 and NbMnSOD2 and four genes of unknown function. There is a possible transposition-mediated insertion region, directly downstream of the NbMnSODs relative to the arrangement of the MnSOD gene found on chromosome XI of E. cuniculi (Fig. 1).

Fig. 1
figure 1

Genomic context of the tandem duplication of NbMnSOD with a weak synteny between the N. bombycis superscaffold7 and that of E. cuniculi chromosome XI. The orientation and position of NbMnSOD1, NbMnSOD2, and EcMnSOD are shown in gray (purple). The pol polyprotein and possible transposition-mediated insertion region are illustrated in gray (red). The four coding sequences in light gray (green) are functionally unknown genes and have no homology to any other genes, based on a BLASTX search. Arrows represent gene positions and transcriptional orientation in the genome. The scale bar is in kilobases (Color figure online)

Full size image

A complete structure containing the N-domains and C-domains for both proteins is predicted based on our analysis using the SMART (http://smart.embl-heidelberg.de/) and PDB (http://www.rcsb.org/pdb/) databases. We have also found two independent candidate promoter regions upstream from each gene through SIB-EPD (http://www.epd.isb-sib.ch/), PROSCAN (http://www-bimas.cit.nih.gov/molbio/proscan/), and PLACE (http://www.dna.affrc.go.jp/PLACE/signalscan.html) analyses. The amino acid identity between the two NbMnSODs is 55.66%. BLASTP searches of the genomes of E. cuniculi and A. locustae revealed that both organisms have only one copy of the MnSOD gene and their identities were 45.54 and 45.69%, respectively, to NbMnSOD1 and 36.20 and 37.61% to NbMnSOD2.

Transcription of MnSODs and Selection Analysis

Primers specific for NbMnSOD1 and NbMnSOD2, respectively, were used to amplify N. bombycis cDNA and genomic DNA isolated from N. bombycis spores. Figure 2 shows a single band of the expected size (268 bp for NbMnSOD1; 337 bp for NbMnSOD2) for both the cDNA and genomic DNA for each of the NbMnSODs.

Fig. 2
figure 2

Transcriptional activity of NbMnSOD1 and NbMnSOD2 in vivo. Lane C amplification of N. bombycis β-tubulin using sscDNA as the template as an internal control (739 bp). Lanes 1, 2 amplification of NbMnSOD2 using genomic and sscDNA as template, respectively; Lanes 3, 4 amplification of NbMnSOD1 using genomic and sscDNA as template, respectively; Lane M DL2000 marker

Full size image

Table 1a shows that the Ka/Ks ratios calculated for the C-domain of each of the NbMnSODs in a comparison with the C-domain of two other Microsporidia (E. cuniculi and A. locustae) are less than 1. This is indicative of purifying selection to eliminate deleterious mutations to keep the protein from changing (Hurst 2002). This agrees with the evidence from the cDNA amplification experiments (Fig. 2) indicating that the two NbMnSODs are transcribed and function in vivo. The bigger Ks values and smaller Ka values for NbMnSOD1 than for NbMnSOD2 (Table 1a) may indicate that NbMnSOD2 is less constrained and evolving more rapidly than NbMnSOD1.

Table 1 Non-synonymous (Ka) and synonymous (Ks) substitution rates (a) and the P value of relative-rate tests (b) for two NbMnSODs
Full size table

The relative-rate tests of microsporidial MnSODs are shown in Table 1b. Although relative rates could not be computed for the Ks values, the relative-rate tests for the Ka and amino acid values of NbMnSOD2 compared to the other microsporidial MnSODs are significant (P < 0.05), suggesting that NbMnSOD2 evolves faster than NbMnSOD1. Positive selection tests and posterior probabilities were calculated for the two NbMnSODs and EcMnSOD using the PAML package (Table 2). Branch-site model A reveals several positively selected codons within the foreground branch of NbMnSOD2 (Table 2).

Table 2 Positive selection tests for three microsporidian MnSODs (NbMnSOD1, NbMnSOD2, and EcMnSOD), calculated using the CODEML program from the PAML package
Full size table

Alignment of Sequences

The metal specificity and oligomeric state of 72 SOD enzymes was predicted based the known X-ray structures of 11 SODs included in the alignment. Fifty-five MnSODs (28 tetramers, 25 dimers, and 2 archaebacterial SODs of unknown oligomeric state), 1 dimeric cambialistic SOD, and 16 FeSODs (2 tetramers and 14 dimers) were analyzed (Supplementary Table S1).

Figure 3 shows an amino acid alignment of 12 of the above sequences, including three MnSODs for which the structure has been determined based on X-ray diffraction studies (H. sapiens (tetramer; mitochondria), A. fumigatus (tetramer), and E. coli (dimer)). Figure 3 also includes three microsporidial sequences (the two MnSODs from N. bombycis and the E. cuniculi MnSOD) and pairs of MnSODs from Phytophthora nicotianae (PnMnSOD1 (dimer) and PnMnSOD2 (tetramer)) and Nematostella vectensis (NvMnSOD1 (dimer) and NvMnSOD2 (tetramer)). Two protistan FeSODs from Tetrahymena thermophila (TtFeSOD) and Tetrahymena pyriformis (TpFeSOD) (Barra et al. 1990; Takao et al. 1991; Wintjens et al. 2004), which show a sister group relationship to the MnSODs (Fig. 5), are also included in Fig. 3.

Fig. 3
figure 3

Structural alignment of 12 SOD proteins. Amino acid sequences from NbMnSOD1, NbMnSOD2, EcMnSOD, PnMnSOD1, PnMnSOD2, NvMnSOD1, NvMnSOD2, TtFeSOD, and TpFeSOD were aligned with 3 MnSODs for which the X-ray structure has been determined (Homo sapiens mitochondria, Aspergillus fumigatus, and Escherichia coli). Residues conserved in 100% of the SOD sequences are in dark gray. Residues common to MnSODs (Met24, Gly79, Gly80, Phe87, Gln160, and Asp161) are in yellow, while Tyr87, found in the two FeSODs, is in pink; residues common to dimers (Asp20, Arg75, Arg135, and Ser156) are in green. Residues Thr23, Asn76, Phe136, and Pro163 in red are systematically encountered in dimers and never in tetramers, whereas Phe76 and Gln136 (in dark blue) are conserved among tetramers. NbMnSOD2 residues Tyr76 and Gln136 (light blue) show a covariant deviation compared to other dimeric SODs, while Asn160 (also in light blue) shows a deviation from all other SODs presented. The forward arrows at position 0 represent the omitted amino-terminal extensions found in 6 of the SODs shown here. These extensions were identified by our analysis as mitochondrial target sequences. An interchain aromatic–polar interaction formed by the amino acid residues at sites 76 and 136 is indicated by the black vertical arrows. The conserved metal binding site is boxed. Residue numbers are indicated below the sequence (Color figure online)

Full size image

The 12 SODs shown in Fig. 3 all have the conserved metal binding motif DxWEHAYYx (box) as well as 4 additional constant amino acid positions (Gly79, Gly80, Gln160 (with the exception of NbMnSOD2, which has Asn160), and Asp161) common to MnSODs and the Tetrahymena FeSODs (Parker and Blake 1988). Two additional conserved amino acids (Met24, Phe87) are specific to the manganese-containing SODs (Parker and Blake 1988; Jackson and Cooper 1998; Wintjens et al. 2004).

The oligomeric structure for three of the MnSODs shown in the Fig. 3 alignment has been determined by X-ray diffraction studies (A. fumigatus and H. sapiens, tetrameric; E. coli, dimeric:). NbMnSOD1, PnMnSOD1, NvMnSOD1, and EcMnSOD have several amino acid residues found in dimeric SODs, while PnMnSOD2, NvMnSOD2, TtFeSOD, and TpFeSOD have several amino acid residues found in the tetrameric SODs. In MnSOD dimers, the Asn76 residue and the aromatic Phe136 residue are linked to form an interchain aromatic–polar interaction across the dimer interface situated not far from the entrance of the main substrate channel. This interaction plays an important role in the structure and function of MnSODs (Wintjens et al. 2004). The NbMnSOD2 protein shows an interesting co-variation at these key positions: Asn76 is replaced by the aromatic amino acid Tyr76, while the aromatic Phe136 is replaced by Gln136, allowing for a Tyr76–Gln136 interaction similar to the Phe76–Gln136 aromatic–polar interaction seen in tetramers.

Hydrophobic Cluster Analysis

Figure 4 is an HCA alignment of the 12 SODs shown in Fig. 3. The tetrameric SODs are on the left and the dimeric SODs are on the right. The presence of helix 2 between residue numbers 50 and 60 (highlighted) in the microsporidial forms implies that they are dimeric (Lah et al. 1995); in general, this helix is absent from tetrameric MnSODs (Borgstahl et al. 1992; Cooper et al. 1995).

Fig. 4
figure 4

This is an HCA alignment of the 12 SODs shown in Fig. 3, including Homo sapiens mitochondria, Aspergillus fumigatus, and Escherichia coli which have known X-ray structures. Tetrameric forms are on left, while dimeric forms are on right. Amino acids are symbolized by their single letter code with the following exceptions: threonine (open square), serine (square with dot), glycine (black diamond), and proline (black star). The HCA plot is created by placing the sequence on a cylinder of 3.6 amino acids per turn. After 5 turns, residue i and i + 18 have the same position on the cylinder. The cylinder is then cut lengthwise and unrolled (see Fig. 4). Since this would make some adjacent residues widely separated, the cylinder is duplicated to allow visualization of adjacent residues (Gaboriaud et al. 1987). Groups of encircled amino acids represent clusters of hydrophobic residues (W, Y, M, F, I, L, V are considered as hydrophobic amino acids). Seven helices and three strands are shown. Dimer-specific helix 2 is shaded. Amino acid residue numbers are indicated above each sequence

Full size image

Our calculations of HCA homology scores are presented in Table 3. Table 3a illustrates a greater similarity among MnSODs with similar quaternary structure. Tetramers show significantly higher HCA values when compared with other tetramers (P value of T test = 0.0084, <0.01) and dimers show higher values when compared with other dimers (P value of T test = 0.0327, <0.05).

Table 3 The HCA homology scores and BLASTP identity scores (in parentheses) of the 12 SODs shown in Fig. 4
Full size table

The HCA homology scores for NbMnSOD1 compared to the other SODs shown in Table 3b are significantly higher (P value of T test = 0.0033, <0.01) than those for NbMnSOD2 compared to the other SODs of Table 3b. This agrees with the phylogenetic and Ka/Ks analyses, indicating that NbMnSOD2 has changed more rapidly than NbMnSOD1 since the time of duplication and may have developed a slightly different function.

Subcellular Localization

Subcellular localization of 72 SODs was predicted using PSORT II (Nakai and Horton 1999), TargetP (Nakai and Horton 1999) and MitoProt II (Claros and Vincens 1996). Target sequences were identified for 24 of the MnSODs (12 from fungi, 7 from metazoa, 2 from Plantae, and 3 from cyanobacteria) and 6 of the protistan FeSODs (Supplementary Table S1). The signals for PnMnSOD2 and TpFeSOD could not be determined in our analysis due to incomplete sequence information but previous studies have shown that these proteins are located in the mitochondria (Barra et al. 1990; Blackman et al. 2005).

The microsporidial MnSODs NbMnSOD1, NbMnSOD2, and EcMnSOD show no terminal extensions. The fungi typically possess two MnSODs, one targeted for the mitochondria and one without a target sequence. Similarly, both the oomycete Phytophthora nicotianae and the cnidarian Nematostella vectensis have two MnSODs, one targeted for the mitochondria and one lacking a target sequence. Interestingly, two of the cyanobacterial MnSODs with target sequences showed mitochondrial-type target sequences and one showed a chloroplast-type target sequence.

Phylogenetic Analysis

Figure 5 shows the most likely tree inferred from the 72 SOD amino acid sequences. The analysis indicates that NbMnSOD1 and NbMnSOD2 sequences are more closely related to each other than to any other sequence (bootstrap values of 100%). The monophyly of Microsporidia is strongly supported by the bootstrap values (85% of the replicates).

Fig. 5
figure 5

A phylogenetic tree of 72 SOD proteins. Accession numbers of the SOD sequences retrieved from NCBI are shown in the Supplementary Table S1. Numbers near the branching nodes indicate bootstrap values given as percentages. Nodes with values of <50% are not shown. Names in bold are MnSODs from the Microsporidia N. bombycis and E. cuniculi. Black circles indicate the protistan FeSODs, diamonds indicate the two MnSODs of P. nicotianae, and squares indicate the two MnSODs of N. vectensis. Predicted cytosolic, mitochondrial, and chloroplast SODs are shown in red, blue, and green letters, respectively. Predicted dimeric or tetrameric structure of MnSODs is indicated by red or blue boxes, respectively. 1 PnMnSOD2 and TpFeSOD are located in the mitochondria (Barra et al. 1990; Blackman et al. 2005). 2 TtFeSOD and TpFeSOD are two iron-containing SOD enzymes, but they are similar to MnSODs in amino acid sequence. 3 Quaternary not determined for the two Archaea MnSODs. 4 This organism is a green algae which belong to the Kingdom Plantae (Color figure online)

Full size image

Interestingly, the Microsporidia do not cluster with fungi but resolve as a sister taxon to the alpha- and delta/epsilon-proteobacteria albeit with low bootstrap values (43%). Our analysis indicates that LBA artifact is not a factor and therefore would not explain the divergent nature of the microsporidial MnSODs.

The MnSODs from the Archaea form a separate group and thus the MnSOD phylogeny separates these taxa (with some exceptions) into the three domains of life (Woese et al. 1990).

Figure 5 shows a fundamental dichotomy between the FeSODs and the MnSODs. In addition, we can see a clear distinction between the bacterial MnSODs, which are dimeric, and those from eukaryotes, which are usually tetrameric. Phytophthora nicotianae and N. vectensis each have two MnSODs. PnMnSOD1 and NvMnSOD1 branch within the bacteria while PnMnSOD2 and NvMnSOD2 cluster together with Plantae and metazoa. Thus, these two organisms have both the “prokaryotic” dimeric and “eukaryotic” tetrameric forms of MnSOD.

Figure 5 also shows two distinct clades of fungal MnSODs, one targeted to the cytosol and the other to the mitochondria. It appears that for these SODs, phylogenetic position correlates reasonably well with protein structure, subcellular localization, and metal cofactor.

Discussion

A tandem duplication of the manganese superoxide dismutase gene in N. bombycis has been identified and seems to have occurred at some time since N. bombycis diverged from the common ancestor with E. cuniculi (Fig. 1). Both NbMnSOD1 and NbMnSOD2 transcripts are present in the spores, indicating that both isozymes are expressed. Ka/Ks calculations and relative-rate tests obtained from a comparison of the nucleotide sequences of the two NbMnSODs with the nucleotide sequences of the A. locustae and E. cuniculi MnSODs indicate that NbMnSOD2 is under less constraint from evolutionary change and evolves faster than NbMnSOD1 (Table 1). Hydrophobic cluster analysis (Fig. 4) shows that both NbMnSOD1 and NbMnSOD2 have the alpha helix (between positions 50 and 60) typically observed for dimeric MnSODs. Pairwise comparison of HCA homology scores (Table 3), however, shows larger differences between NbMnSOD2 and the other SODs than between NbMnSOD1 and the other SODs, providing further evidence that NbMnSOD2 is under less evolutionary constraint than NbMnSOD1.

An amino acid alignment of 6 tetrameric and 6 dimeric SODs (Fig. 3) shows several significant changes in the primary structure of NbMnSOD2 compared to NbMnSOD1 and the other MnSODs. NbMnSOD2 has an asparagine at position 160, unlike the other MnSODs shown in Fig. 3, which all have a glutamine at this position. Particularly notable amino acids changes occur at residues 76 and 136, which form an interchain aromatic–polar interaction in both dimers (Asn76 and Phe136) and tetramers (Phe76 and Gln136). Unlike the other MnSODs in Fig. 3, NbMnSOD2 has a tyrosine at position 76, and a glutamine at position 136. The codons for residue 76 in EcMnSOD, NbMnSOD1, and NbMnSOD2 are AAC, AAT, and TAC, respectively. We presume that AAC is the ancestral codon (present in most dimeric MnSODs), and that a synonymous mutation to AAT occurred in NbMnSOD1, while a nonsynonymous mutation to TAC occurred in NbMnSOD2. The change from Gln136 to Phe136 in NbMnSOD2 would allow the important interchain aromatic-polar interaction between residues 76 and 136 to be conserved, and as Table 2 shows, Gln136 in NbMnSOD2 is under strong positive selection, with a posterior probability >0.95.

Based on these analyses, a more detailed investigation of NbMnSOD2 to determine its physical and enzymatic properties is warranted. This would allow us to understand whether the NbMnSOD gene duplication represents an added dose effect to enhance antioxidant efficiency in N. bombycis, or if the duplication of this gene has resulted in the acquisition of a new gene function (Ohno 1970). In plants multiple forms of SODs are present which vary based on their location within cells, tissues and organelles. Changes in plant SOD isozymes occur during development and in response to environmental factors, suggesting different roles for these SOD isozymes (Scandalios 1993). In animals different SODs are expressed at different levels depending on tissue type (Marklund 1984), indicating a need for SODs with different functions. The two NbMnSOD isozymes may be expressed at different times in spore and meront development, in different insect host tissues, or at different temperatures.

Phylogenetic analysis reveals that MnSODs group according to subcellular localization, quaternary structure, and taxonomic relationship. There are clear groupings separating those proteins that are targeted for the cytosol (Fig. 5, red lettering) from those that are targeted for the mitochondria (Fig. 5, blue lettering). The ascomycote fungi presented in Fig. 5 have tetrameric MnSODs targeted for the cytosol and the mitochondria. Both types of MnSOD (cytosolic and mitochondrial) have been identified in several of the fungi shown, suggesting that gene duplication occurred before the radiation of these species. We also see a sharp dichotomy between the dimeric (primarily bacterial) MnSODs (Fig. 5, red boxes) and the tetrameric (eukaryotic) MnSODs (Fig. 5, blue boxes). In addition, it is clear from Fig. 5 that many of the eukaryotic MnSODs are targeted for the mitochondria (blue lettering), while most of the bacterial and a number of the fungal MnSODs presented here are located in the cytosol (red lettering). Several exceptions are seen in the cyanobacteria, two of which (Thermosynechococcus elongatus and Gloeobacter violaceus) have mitochondrial target sequences and one of which (Nostoc punctiforme) has a chloroplast target sequence. The oomycete P. nicotianae and the cnidarian N. vectensis each have both a tetrameric mitochondrial and a dimeric “prokaryotic” MnSOD (PnMnSOD1, 2 and NvMnSOD1, 2, respectively). The final noteworthy exception is the Microsporidia, which group weakly with the bacteria and have dimeric MnSODs.

Our phylogenetic analysis (Fig. 5) correlates well with what we know about the relationships among these organisms. We see a clear separation of the 14 protistan FeSODs which we use here as the outgroup for the remaining, primarily manganese SODs. Figure 5 shows a clear division of the 3 domains of life (Woese et al. 1990). Our analysis groups the metazoa together arranging the vertebrates in the expected order. The fungi for which the SOD gene has been sequenced all belong to the Phylum Ascomycota, and show relationships which vary only in minor respects to those previously published for a six-gene phylogeny (Thomarat et al. 2004; Gill and Fast 2006; James et al. 2006). The bacteria group together into previously established clades, and phylogenetic relationships among the cyanobacteria presented here are in agreement with those determined using ssrDNA (Tomitani et al. 2006) and whole genomes (Zhaxybayeva et al. 2006). We believe that the phylogeny presented here, based on MnSOD analysis, shows a remarkable amount of resolution among a wide range of taxa.

There are several possible explanations regarding the placement of the Microsporidia in Fig. 5. One possibility is that the Microsporidia obtained MnSOD through horizontal gene transfer (HGT) from a proteobacterium (perhaps from the common ancestor of the alpha and delta/epsilon group). As long as the Microsporidia are considered to be derived fungi (Fast et al. 2003; Keeling et al. 2005), HGT to the Microsporidia will remain an explanation for the numerous genes which are very different from those of the fungi. Another possibility is that the microsporidial MnSOD is a divergent eukaryotic gene. The bootstrap value of Fig. 5 linking this gene to the proteobacteria is not high (43%). In none of the analyses, with any of the alignments obtained, did the MnSODs from the Microsporidia group with those of the fungi, or show a sister relationship with the fungi presented.