Antarctic marine ciliates under stress: superoxide dismutases from the psychrophilic Euplotes focardii are cold-active yet heat tolerant enzymes

Oxidative stress is a particularly severe threat to Antarctic marine polar organisms because they are exposed to high dissolved oxygen and to intense UV radiation. This paper reports the features of three superoxide dismutases from the Antarctic psychrophilic ciliate Euplotes focardii that faces two environmental challenges, oxidative stress and low temperature. Two out of these are Cu,Zn superoxide dismutases (named Ef-SOD1a and Ef-SOD1b) and one belongs to the Mn-containing group (Ef-SOD2). Ef-SOD1s and Ef-SOD2 differ in their evolutionary history, expression and overall structural features. Ef-SOD1 genes are expressed at different levels, with Ef-SOD1b mRNA 20-fold higher at the ciliate optimal temperature of growth (4 °C). All Ef-SOD enzymes are active at 4 °C, consistent with the definition of cold-adapted enzymes. At the same time, they display temperatures of melting in the range 50–70 °C and retain residual activity after incubation at 65–75 °C. Supported by data of molecular dynamics simulation, we conclude that the E. focardii SODs combine cold activity, local molecular flexibility and thermo tolerance.

Polar marine organisms face a number of environmental challenges, in particular the adverse effects of cold on key biological processes and high oxidative stress 1,2 . Low temperature greatly impairs enzyme activity and membrane fluidity, slows down secretory processes and affects the stability and activity of macromolecular machines (replication and transcription complexes, mitotic spindles, ribosomes) which depend on weak, noncovalent molecular interactions [3][4][5] . To survive, "psychrophilic" organisms rely on adaptive changes, that include the up-regulation of genes encoding proteins involved in metabolite transport, the synthesis of cryoprotectors (mannitol, polyamines), increased membrane fluidity and the production of enzymes endowed with high activity at low temperature 4 . Moreover, cold-adapted marine organisms from Antarctica are exposed to both high concentration of oxygen and to UV radiation (the latter due to the ozone depletion) that boost the production of reactive oxygen species (ROS). ROS are involved in several physiological processes, such as cellular signaling pathways, resistance to microbial pathogens, apoptosis 6 . Nevertheless, at high concentration ROS threaten essential cellular macromolecules, notably proteins, lipids, and nucleic acids. As a consequence, antioxidant defences are of key relevance to aquatic polar living beings, as ROS control is required to balance their concentration and avoid cellular damage 7 . One of the key players in this process are the ubiquitous metalloenzymes superoxide dismutases (SOD, EC 1. 15 . SODs are grouped into three protein families, based on the metal cofactor they contain and on the protein fold 10 . Copper,zinc SODs (Cu,Zn SODs) are found in the cytoplasm of eukaryotes, in the chloroplasts of some plants and in the periplasmic space of bacteria 11,12 . This group of SODs is often referred to as SOD1. SOD1s are homodimers, with each subunit organized in eight antiparallel β strands and three external loops 13 . SOD3s, the Cu,Zn enzymes present in the extracellular fluids of eukaryotes, are similar to the previous ones but have tetrameric quaternary structure 14,15 . Iron-and manganese-containing SODs (FeSOD and MnSOD) are believed to be the more primitive forms of SODs 16,17 . FeSODs are found in prokaryotes and chloroplasts, while MnSODs are present both in prokaryotes and in the mitochondrial matrix of eukaryotes. Several fungi possess both cytosolic and mitochondrial MnSODs 18 . Cambialistic SODs, whose activity depends either on iron or on manganese incorporated in the same protein moiety, have been identified in a few classes of bacteria and eukaryotes [19][20][21] . MnSOD and FeSOD can be either homodimers or homotetramers and share high similarity in sequence and structure 22,23,24 , strongly suggesting a common evolutionary origin. In contrast, Cu,Zn and Mn/Fe SODs never shared a common ancestry, as shown by their distinctive amino acid sequences 25 and completely different tertiary structures 13 . The discovery of a nickel containing enzyme in Streptomyces and cyanobacteria established a novel SOD group with a unique Ni-hook structural motif 26 .
This study focuses on one MnSOD and two Cu,Zn SODs, identified in the strict psychropilic protist Euplotes focardii, a hypotrichous ciliated protozoan isolated from the coastal seawaters of Terra Nova Bay in Antarctica 27 . Under laboratory condition, E. focardii has an optimal temperature of growth of about 4-5 °C and is not viable above 12 °C 28 . We show that the two SOD families likely followed different evolutionary history. However, they share similar temperature-dependent regulation of expression and combine cold activity with thermo-tolerance, a feature that could have played a role in the successful colonization of the Antarctic marine habitat by this ciliate.

Results
Evolution of E. focardii superoxide dismutases. Putative Cu,Zn SOD coding sequences identified by BLAST search in the E. focardii transcriptome were reported in a previous work and named Ef-SOD1a (accession number KF740481) and Ef-SOD1b (accession number KF740482) 29 . Ef-SOD2 (MnSOD, accession number MG575644) is described in this work.
The molecular evolution of E. focardii SODs (Ef-SODs) was studied in the frame of known SOD sequences from other species of ciliates. The data set included also the Cu,Zn SOD from the amoeba Dictyostelium (Dp-SOD1) as a non-ciliate member of the protista group and sequences (both Cu,Zn and Fe/Mn SODs) from Antarctic bacteria to have a reference describing cold-active SODs (Fig. 1). In the phylogenetic tree, SODs assemble in two separate clusters. The two clusters correspond to the different families, SOD1 and SOD2, defined based on the metal cofactors they contain (either Cu,Zn or Fe/Mn). Overall, the length of the branches is consistent with the low degree of sequence conservation highlighted also in the multiple sequence alignments (Figs S1 and S2). Within the Cu,Zn SOD (SOD1) family, Ef-SOD1a and Ef-SOD1b isoforms cluster in two different clades. The clade that includes Ef-SOD1a contains SOD1 sequences from Euplotes (Ec-SOD1a), Stylonychia (Sl-SOD1) and Oxytricha (Ot-SOD1a and Ot-SOD1b) all of them belonging to the same class of ciliates (Spirotrichea). By contrast, the clade containing Ef-SOD1b is more heterogeneous, since it groups SOD sequences from both Spirotrichea and Oligohymenophorea (Paramecium, Tetrahymena and Ichthyophthirius) and that from Dictyostelium. Cu,Zn SODs from Antarctic bacteria (Flavobacterium hibernum, Nesterenkonia sp.AN1 and  Table S1. Cyan and red diamond symbols indicate ciliates that belong to Spirotrichea and Oligohymenophorea classes. The tree was obtained by using Neighbour-Joining (NJ); the numbers on the branches represent bootstrap values for 1000 replicates. Planococcus antarcticus) are completely separated from the protistan group. Their position in the tree excludes a close phylogenetic relationship with the homologous sequences from the Antarctic ciliate E. focardii. By contrast, the bacterial and the protistan SODs of the Fe/Mn family (SOD2) cluster together in a single clade. In conclusion, the topology of the tree suggests that SOD1 evolutionary history would have witnessed multiple events of gene duplication followed by sequence diversification even inside homogeneous groups of ciliate and bacteria, whereas SOD2 sequences are more conserved.
Expression of E. focardii superoxide dismutases. In order to investigate how gene expression is regulated by temperature, we analyzed by qRT-PCR the transcription levels of SODs coding genes in E. focardii cells both at physiological temperature (4 °C) and under temperature stress that is after exposure either at 0 °C, a mild temperature drop, or at 12 °C, the highest temperature permissive for E. focardii growing. At the constant temperature of 4 °C, the three SOD genes are expressed at different levels, with the Ef-SOD1b mRNA nearly 20-fold higher and the Ef-SOD2 mRNA ca. 2,5-fold higher than that of Ef-SOD1a ( Fig. 2A). After 1-hour shift at 0 °C both SOD1s were up-regulated (white bars in Fig. 2B), whereas at 12 °C a down-regulation was detected (red bars). Ef-SOD2 mRNA levels decreased both at 0 °C and at 12 °C (Fig. 2B). Regulation effects of all Ef-SODs fainted with time, hinting to cell adaptation to the new temperature condition (see results at 2 hours of incubation).  amino acid signal peptide for translocation to mitochondria. Based on these results, in the synthetic genes the corresponding nucleotide stretches were omitted (Fig. S3).

Biochemical and conformational features of recombinant superoxide dismutases.
Truncated Ef-SOD1a and Ef-SOD2 were produced in E. coli as soluble proteins and in the following are referred to as Ef-SOD2 Δ and Ef-SOD1a Δ . Consistent with the lack of any predicted sequence for transport, the full-length Ef-SOD1b sequence was obtained at high yield. Recombinant proteins were produced by 16 hours cultures at 25 °C in Zym5052 medium in the presence or in the absence of Cu and Zn (see materials and methods) for Ef-SOD1a Δ and Ef-SOD1b, with or without Mn in the case of Ef-SOD2 Δ . Added metals did not produce any effect neither on cell viability nor on the amount of recombinant proteins (data not shown). Protein yields as determined after IMAC purification were 5.3 mg/L Ef-SOD2 Δ , 3.1 mg/L Ef-SOD1a Δ , 39 mg/L Ef-SOD1b.
We assayed for activity at 4 °C and 27 °C recombinant proteins produced both in the presence and in the absence of metal cofactors in the growth medium during the heterologous expression in E. coli (Fig. 3). Given assay temperatures were selected since the first one (4 °C) is the optimal temperature of growth of E. focardii cells under laboratory conditions and the second one approaches the optimal temperature of most cold-active enzymes, included a phospholipase and an alpha-amylase previously characterized from this Antarctic ciliate 32,33 .
Overall, the activity of Ef-SOD2 Δ was poorly affected both by temperature and by added metals. The combined effects of temperature and metals on SOD1s are complex and are therefore described separately. We observed that the specific activities of Ef-SOD1a Δ and Ef-SOD1b increased by ~6-7 folds in the presence of metals available to cells during production. As for temperature dependence, Ef-SOD1a is more active (~25%) at the highest temperature, while Ef-SOD1b shows similar activity at 4 °C and 27 °C. Overall, the data presented demonstrate cold activity for all Ef-SODs.
In order to gain a deeper insight into the temperature dependence and stability of the three SODs, we measured the residual activity of enzymes pre-incubated 20 minutes at temperatures in the range 5 °C to 90 °C ( Fig. 4A). We observed that both Ef-SOD1s were kinetically stable in the range 5 °C-55/60 °C, with midpoints at ~60 °C for Ef-SOD1a Δ and of 55 °C for Ef-SOD1b, whereas the activity of Ef-SOD2 Δ kept constant from 5 °C to 70 °C. Pre-incubation above 70 °C induced a sharp decrease in Ef-SOD2 activity with a midpoint of ~75 °C. This set of experiments showed that Ef-SOD1 enzymes are less stable to temperature than Ef-SOD2. We showed were incubated for 20 minutes at indicated temperatures. After incubation, residual activities were assayed at appropriate dilution at room temperature. Measured activity was normalized taking the initial activity as 100%. (B) Far UV CD spectra of SODs at 4 °C. (C) Thermal stability of Ef-SODs. Ellipticity values were recorded at 215 nm (Ef-SOD1a Δ ), 210 nm (Ef-SOD1b), 208 nm (Ef-SOD2 Δ ) during heating from 5 °C to 90 °C. Initial CD signal was taken as 100% for normalization. Recombinant proteins were produced in the presence of metal cofactors. previously ( Fig. 3) that SOD1s specific activity is affected by metals. Nevertheless, we did not observe any difference in the temperature stability of proteins produced with and without metals in the medium (data not shown). Accordingly, the stability profiles are fully superimposable and in Fig. 4A we show results obtained with proteins produced in the presence of metals only. These data suggest that metals are relevant for specific activity but not for protein stability. One can hypothesize that Cu 2+ intracellular concentration is insufficient to saturate the overexpressed Ef-SOD1s, as reported for other SOD1s 34,35 .
With this information in our hands, it was of interest to check the impact of temperature on the proteins structure by circular dichroism (CD) analysis. Spectra obtained at 4 °C and 27 °C were identical. As shown in Fig. 4B, the CD spectra of Ef-SOD1s were similar with a peak at ~210 nm. This profile is typical of proteins with mainly-β structure and it is reported for other SOD1s 36,37 . The two minima detected at ~208 and ~222 nm reveal high content of α-helix structures in Ef-SOD2 Δ . These data are consistent with the 3D structural model described in a later section (Fig. 5). Thermal stability was investigated by CD analysis at fixed wavelength and temperature raising from 5 °C to 90 °C. The secondary structure of Ef-SOD2 Δ was unaltered in the range 5 °C-60 °C. Above 60 °C, the CD signal was rapidly lost with a midpoint temperature (Tm) of 72.8 °C ± 2.1. Above 40 °C a dramatic loss of CD signal was detected in Ef-SOD1a Δ and Ef-SOD1b samples, with a Tm of 58.4 °C ± 2.3 and 51.3 °C ± 1.5, respectively (Fig. 4C).
Structural models of E. focardii superoxide dismutases. In order to correlate functional analyses with structural information we modeled the 3D structure of Ef-SODs with SwissModel 38 using the following templates: C. elegans Cu,Zn SOD1 (PDB: 3KBF) for Ef-SOD1a (sequence identity: 45%), chimeric SOD1 from Mus musculus and Homo sapiens (PDB: 3LTV) for Ef-SOD1b (sequence identity: 55%) and human mitochondrial SOD (PDB: 1VAR) for Ef-SOD2 (sequence identity: 44%). Templates were chosen based on high sequence identity with the target proteins and the high resolution of available 3D structures. Moreover, the oligomerization state is in agreement with that experimentally determine (see later). The quality of the structural models was estimated using the QMEAN function (Fig. S4).
Structural models of Ef-SOD1s show the typical structure of Cu,Zn SODs ( Fig. 5A and B) with a β-barrel composed by eight β-sheets containing two Greek key domains and two large loops 13 68 bridge Cu 2+ to a Zn 2+ ions in Ef-SOD1a and Ef-SOD1b, respectively. Sequence analysis reveals that the SOD1 E-loop is less conserved in ciliates than in higher eukaryotes, where it usually contains the electrostatic triad and a conserved arginine residue 42 (Fig. S3A). Among ciliates, the "canonical" electrostatic triad is observed only in Ot-SOD1a and Ot-SOD1b from Oxytricha trifallax and Sl-SOD from Stylonychia lemnae (Fig. S1), while both Ef-SOD1a and Ef-SOD1b contain a non-canonical triad (Fig. S3A). In the 3D model of Ef-SOD2, two long α-helices at the N-terminus form a helical hairpin structure, while the C-terminus contains a three-stranded β-sheet flanked by four α-helices on both sides (Fig. 5C). The Mn 2+ ion is coordinated by three His (His 46 , His 94 , His 188 ) and one Asp residue (Asp 184 ) conserved among MnSODs ( Fig. S2 and S3B). In homologous enzymes the helical hairpin structure plays a key role in the formation and stabilization of the tetrameric structure 43,44 . Indeed, in human MnSOD, the substitution of Ile 58 (boxed in Fig. S3B) with Thr destabilizes the tetramer and promotes a dimeric structure by perturbing the methyl group interaction network and hence impairing the oligomerization interface 44 . We showed by SECS-MALS analysis (Table S2) that Ef-SOD2 is tetrameric, although it contains a Thr residue at an equivalent position (Thr 78 , boxed in Fig. S3A). This might indicate that the interaction network involved in oligomerization of the human MnSOD is not conserved in Ef-SOD2.
Recombinant Ef-SODs, were analyzed by SEC-MALS to determine their molecular weight and oligomerization state (Table S2). The molecular mass of Ef-SOD2 Δ is close to the theoretical mass of the tetramer, whereas the molecular mass of Ef-SOD1b is close to that expected for the dimer. Unfortunately, the high propensity of Ef-SOD1a to form aggregates interferes with MALS measurements. The comparison of SEC chromatographic profiles obtained for Ef-SOD1a and Ef-SOD1b, sharing similar molecular mass, suggests that Ef-SOD1a is a dimer because elutes at the same time of Ef-SOD1b (data not show).

Molecular dynamic simulation.
Most cold active proteins are endowed with higher either local or global flexibility than the mesophilic and thermophilic counterparts to cope with the reduction of dynamics and activity at low temperature [45][46][47] . Therefore, we set up to investigate Ef-SODs by Molecular Dynamic (MD) simulations, in which the flexibility of monomeric and oligomeric conformations (dimers for Ef-SOD1s and tetramers for Ef-SOD2) was compared at 4 °C and 27 °C (Fig. 6), i.e. the same temperatures used for the enzymatic assays.
Flexibility was estimated based on the average root-mean-square fluctuation (RMSF) values recorded for proteins both in the monomeric and the oligomeric form of the enzymes. Due to the limitations inherent to the use of 3D models, in the following we report only relevant differences (>1.5 Å) observed in simulations performed at 4 °C and 27 °C.
In SOD1 enzymes, RMSF differences detected at the two temperatures are small. In particular the E-loop of Ef-SOD1a is more flexible at 4 °C than at 27 °C, what might be of importance for catalytic activity in the cold (Fig. 6A) The RMSF profile of Ef-SOD1b shows no relevant differences in flexibility between the monomeric and the dimeric forms at both considered temperatures (Fig. 6B). Indeed, only loop I of chain A and loop V in both chains (Fig. 6B) of the native dimeric conformation are slightly more flexible at 27 °C. However, such loop regions are localised on the opposite sites of the structure with respect to regions involved in subunits interaction 40 , and are not assumed to be involved in protein stability.
In the native tetrameric form of Ef-SOD2 but not in the monomer, the α7 helix in chain A, loop V in chain C and residues spanning 66-71 belonging to α1 helix of chain D exhibit higher flexibility at 4 °C (Fig. 6C). The α1 helix is involved in protein oligomerization and may favour the formation of Ef-SOD2 tetramers at low temperatures. Worth of note, in some cases we observed differential effects of temperature among the subunits building the oligomer. These observations may surmise a non-equivalence of the subunits in the quaternary structure or non-symmetric long range effects of oligomerization, suggesting cooperativity of SOD subunits 48 .
To analyse more in depth the effects of temperature on ions-coordinating loops, the flexibility of His and Asp residues involved in ions coordination (i.e. His 80 , His 82 His 105, His 114 , Asp 117 ; His 154 in Ef-SOD1a and by His 51 , His 53 His 76 , Asp 88 His 85 and His 125 in Ef-SOD1b and His 46 , His 94 , His 188 , Asp 184 in Ef-SOD2) was evaluated using the root mean square deviation (RMSD) profile (Fig. S5A). Overall, Cu 2+ coordinating His and Asp residues of Ef-SOD1a and Ef-SOD1b were slightly more flexible in the dimeric conformation, whereas Mn 2+ coordinating residues of the Ef-SOD2 were more stable in the tetrameric form. The stabilizing effect of oligomerization, reported in several proteins is therefore relevant for the tetrameric SOD2 enzyme but not for the two dimeric enzymes (Fig. S5A).
A key role in the catalytic activity of Cu,Zn SOD is played by Arg residue which follows the electrostatic triad 49 . This residue is conserved in both Ef-SOD1s and is found in position 177 and 148 in Ef-SOD1a and Ef-SOD1b, respectively (Fig. S3A). MD simulations show that temperature and oligomerization state do not change the flexibility of these residues (Fig. S5B), suggesting that catalytic Arg flexibility is nor relevant for the activity of Cu,Zn SODs in the cold.

Discussion
E. focardii is an obligate psychrophilic stenothermal organism restricted to live within a very narrow temperature range. Its optimal growing temperature is 4 °C-5 °C and cell death occurs over 12 °C-15 °C 28 . Besides strict temperature requirements, a major issue for E. focardii is oxidative stress since it experiences high dissolved oxygen typical of Antarctic marine environments 2 . For these reasons, it allowed us studying the combined effects of oxidative stress and cold on key cell enzymes, superoxide dismutases. Transcriptomic analyses of other Antarctic marine organisms such as notothenioid fishes and the krill Euphasia superba show that they overexpress catalases, glutathione peroxidase and SODs under physiological conditions, to avoid cell damages induced by reactive oxygen species (ROS) 50 . Differently from SOD2, the SOD1 family was believed to be absent in protists 10 until whole genomes sequencing revealed SOD1 encoding genes in a number of different ciliates 29,[51][52][53] . Recently, SODs from the mesophilic ciliate Tetrahymena thermophila were described 29 showing their importance in the detoxification pathway in continued, elevated presence of metals in the environment. The evolutionary history of SODs is controversial. Fink and Scandalios suggested that bacterial and eukaryotic SOD1 sequences evolved from a common ancestor 54 . Lee and co-authors proposed that SOD1 evolved rapidly in relatively recent times, differently from SOD2, which appear to have evolved at a relatively constant rate over the entire history of eukaryotes 55 . The large differences between eukaryotic and bacterial SOD1s, highlighted also in the phylogenetic tree reported in this paper, do not surmise any obvious common ancestry inside the SOD1 family. Therefore, it was proposed that some eukaryotic Cu,Zn SODs would derive from horizontal gene transfer from the endosymbionts that gave rise to mitochondria 10 . The discovery of the SOD1 family in several ciliates and other protists, including the amoeba Dictyostelium, opened new scenarios. The phylogenetic tree reported in this work suggests that bacterial and eukaryotic SOD1s do not derive from a common ancestor. Furthermore, most of the SOD1 isoforms of individual ciliate species do not cluster together but rather belong to separated clades, including the SOD1 isoforms from Paramecium and Tetrahymena. This result suggests that the evolution of SOD1s may have witnessed several events of gene duplication and diversification, occurred very early in ciliate SOD1 genes, rather than horizontal gene transfer. By contrast, bacterial and eukaryotic SOD2s appear to derive from a common ancestor. Whatever was the evolutionary history, the study of todays E. focardii SODs pinpoint some peculiarities in these enzymes that reflect a complex interplay of environmental pressures. Evolution of key cell enzymes is restricted by survival constrains. Nevertheless, gene duplication enlarges the kit of available catalysts and favors diversification. Moreover, SODs under study belong to two different families. Even though we do not have any direct evidence about the sub-cellular localization of the SOD isoenzymes within the ciliate cells, we assume from signal peptide characterized by in silico analysis that Ef-SOD2 is localized in mitochondria, Ef-SOD1a could be either extracellular or membrane bound and Ef-SOD1b is cytosolic (given the absence of the sequence of signal peptides).
We report regulation of gene expression by temperature changes permissive for E. focardii viability (0 °C and 12 °C). Both Ef-SOD1a and Ef-SOD1b expression is induced by a mild temperature drop, consistent with similar results from other Antarctic organisms. Deschampsia antarctica, a plant adapted to the cold climate, overexpress antioxidant enzymes, including peroxidase, SOD and glutathione reductase that cope with damages by ROS 56,57 . Noteworthy, when this plant was acclimated at 4 °C the expression level of Cu,Zn SODs was higher compared to non-acclimated plants incubated at 13 °C 58 . Similar behaviour was observed for the MnSOD transcription after a mild cold exposure of the Antarctic yeast Glaciozyma antarctica 59 . In the bivalve Yoldia eightsi, SOD activity is higher at low temperatures 60 . Higher expression of SOD1s during the drop of temperature may be explained by the need to cope with the increased concentration of dissolved oxygen in cool water. Incubation at increased temperature causes a decrease in the mRNA levels of both enzymes that is partly recovered with time. Regulation of Ef-SOD2 expression seem to be different. Changes of temperature (both down and up) induce a partial reduction of expression in the short time that is fully recovered within two hours. This would be consistent with the mitochondrial localization of this enzyme suggested by sequence analysis. Mitochondria in fact are exposed to high ROS concentration under any condition.
We studied the temperature dependence of activity and the structural stability of SODs in a broad temperature range (4 °C to 90 °C). It is useful to recall that in most cases the optimal temperature of cold-active enzymes is higher than optimal temperature of organism growth and that an enzyme is classified as "cold active" if it retains activity at low temperature, independently on its optimal temperature 61,62 . All three SODs are active at low temperature. At the same time, they retain high activity upon 20 minutes incubation up to 55/60 °C. This feature is unusual in cold-active enzymes that are often heat sensitive and undergo inactivation and unfolding even at mild temperature. Nevertheless, thermo-tolerance or even thermostability of cold-adapted enzymes was reported previously. For example, Yang and colleagues reported 2.07 hours half-life at 70 °C for a cold-active patatin-like phospholipase from E.focardii 33 . Similarly, cold active SODs from Antarctic bacteria such as Pseudoalteromonas haloplankitis 63,64 , Exiguobacterium sp 65 and Aliivibrio salmonicida 66 possess high thermal stability associated to catalitic activity in the cold. It has been previously proposed that structural flexibility and rigidity may co-exist in the same molecule in psychrophilic enzymes, because only domains involved in the conformational changes during catalysis need to be flexible 62,67 . This is consistent with the observation that enzyme activity can be impaired by temperature before structural damage occurs 61,62 . However, in our results the Tm describing loss of activity and protein unfolding are very close, and the small differences observed can be accounted for by the different experimental methods employed.
Therefore, we propose that cold activity of Ef-SODs is supported by the presence of restricted flexible regions sufficient for low temperature catalysis but not enough extended or mobile to unfold independently of the overall protein structure. Molecular dynamics simulation of the three Ef-SODs revealed only minor differences in the temperature-dependent flexibility. This result is not in contrast with our experimental results; on the contrary, it confirms that Ef-SODs are indeed flexible at both low and mild temperatures, and that slightly higher local flexibility at 4 °C is sufficient for low temperature catalysis yet maintaining heat tolerance. In conclusion, Ef-SODs combine cold activity with thermostability, a characteristic that may be due to the evolutionary origin of this ciliate before Antarctica broke up from Pangaea. In Cenozoic, due to Pangaea fragmentation the complete oceanic circulation around Antarctica and the decreasing atmospheric carbon dioxide concentrations caused a rapid cooling of Antarctica and allowed glaciers to form. Unicellular organisms adapted at higher temperatures, once trapped in the cold Southern Ocean gradually adapted their enzyme structures to the new condition. The combined features of cold activity with overall structural robustness of Ef-SODs seem to be functional to confer to key enzymes the ability to be functional under changing environmental conditions and could have been of advantage for the success of this ciliate in the colonization of the Antarctic marine habitat. Escherichia coli strain DH5 α ™ (Invitrogen, Waltham, USA) was the host for plasmid DNA amplification, while strain BL21 (DE3) (EMD Millipore, Billerica, USA) was used for recombinant protein expression. Random hexamer primers for cDNA synthesis were from Thermo Fisher Scientific (Thermo Fisher Scientific, Waltham, MA USA) and oligonucleotides from Metabion (Metabion International AG, Steinkirchen, Germany). Q5 ® High-Fidelity DNA Polymerase was purchased from New England Biolabs (New England Biolabs, Ipswich, MA). Materials for SOD activity assay (cytochrome C, xanthine and xanthine oxidase), TRIzol reagent were from Sigma-Aldrich (Saint Louis, Missouri, USA). 2X SYBR Green Mix was purchased from Carlo Erba (Milan, Italy).
RNA Extraction and cDNA synthesis. Cell cultures of E. focardii constantly grown at 4 °C were incubated at 0 °C, 4 °C and 12 °C, for 1 and 2 hours. Total RNA was extracted from ~20000 cells (representing a mix of both TN1 and TN15 strains) using TRIzol reagent (Sigma, Milan, Italy) according to manufacturer's instructions. Isolated RNA was resuspended in diethylpyrocarbonate treated water (Sigma) and subjected to DNase I (from Carlo Erba, Milan Italy) treatment to remove genomic DNA. Total RNA concentration was measured at 260 nm with UV spectrophotometer (UV 1600PC). RNA integrity was verified by denaturing electrophoresis of a 2 µg sample on 1% agarose gel. DNase I-treated RNA samples were used as the template to amplify the E. focardii SSUrRNA gene to verify the absence of genomic DNA contamination. First strand cDNA was synthesized at 42 °C for 1 h from 2 µg of total RNA using random hexamer primers (5 ng/µl final concentration) and Moloney murine leukemia virus reverse transcriptase (Lucigen, Middleton, WI, USA).
Quantitative real time PCR. The relative expression patterns of Ef-SODs coding genes were measured in cDNA samples from control (4 °C) and stressed (0 °C and 12 °C) cells by comparative-threshold qPCR using the SYBR green DNA-binding method 68 . Quantitative real time PCR analysis was carried out in a final volume of 25 µl containing 12.5 µl of 2X SYBR Green Mix (Carlo Erba, Milan Italy), 10 µM of forward and reverse primers (Table S3), 1 µl of cDNA template (100 ng/µl) and 9.5 µl of nuclease-free water. The E. focardii SSUrDNA (GenBank ID: EF094961) gene was used for normalization. Amplification reactions were performed in triplicate in a Multicolor qPCR MX3000P thermocycler (Stratagene, Milan, Italy) as follows: 95 °C for 2 min; 40 cycles (95 °C for 15 s, 55 °C for 1 min) and a final cycle (95 °C for 1 min, 55 °C for 30 s, 95 °C for 30 s).
The relative ratio of each Ef-SOD gene was determined according to 69 , by estimating the ∆∆CT, it correspond to where ∆CT is the difference between the mean Ct of the target gene and the mean Ct of housekeeping gene. Ct is the PCR cycle number at which fluorescent signal is above the threshold that is set to exclude the fluorescent signal background. ∆Ct is the Ct deviation of the control minus the sample of the target (SOD) or reference (housekeeping) gene. The unknown and control ∆Cts in the ∆∆CT formula stand for stress and physiological condition respectively.

Recombinant proteins production and purification. Recombinant proteins produced by cells grown
in Zym-5052 medium 72 were extracted as described 73 . To evaluate the effect of metal cofactors on activity, copper (CuSO 4 ) and zinc (ZnCl 2 ) were added at 250 µM final concentration to cultures producing Ef-SOD1s, whereas manganese (MnSO 4 ) were added at the final concentration of 250 µM to Ef-SOD2 producing cultures.
Proteins were purified at 4 °C by metal ion affinity chromatography on a nickel-nitrilotriacetic acid agarose resin (Jena bioscience, Germany). Samples containing protein at the highest concentration were pooled and buffer exchanged twice by gel filtration on PD10 columns (GE healthcare) against 200 mM sodium phosphate buffer, pH 7.0. Protein concentration was determined by the Bradford protein assay (Bio-Rad, California, USA), using bovine serum albumin as a standard. Activity assays. Enzyme activity was assayed according to McCord and Fridovich 8 using a Jasco V-530 UV/ VIS spectrophotometer (JASCO International Co. Ltd., Hachioji, Tokyo, Japan). The activity assay was performed at 4 and 27 °C with Ef-SODs produced either in the absence or in the presence of the respective metal cofactor in the culture medium. Experiments were in triplicate. One unit of SOD was defined as the amount of protein required to inhibit the reduction of cytochrome c by the superoxide radical by approximately 50% 8 .
To study the effect of temperature on Ef-SODs activity, proteins were incubated 20 minutes at temperatures in the range 5 °C-90 °C and then assayed for activity at room temperature. Experiments were in triplicate. CD spectroscopy. CD spectra of 2 µM Ef-SODs were measured with a J-815 spectropolarimeter (Jasco Corp., Easton, MD, USA), using 0.1 cm path-length cuvette. Measurements were performed in the range 190-260 nm, with 0.2-nm data pitch and 20-nm/min scanning speed. All spectra were corrected for buffer contribution, averaged from two independent acquisitions, and smoothed by using a third-order least square polynomial fit.
Thermal denaturation spectra were obtained measuring the CD signal in correspondence of the following minimum peaks: 205 nm for Ef-SOD1a Δ , 215 nm for Ef-SOD1b, 210 nm and 208 nm for Ef-SOD2 Δ fixed wavelength when progressively heating the sample from 5 °C to 90 °C. Measurements were performed in triplicate with a data pitch of 2 °C and a temperature slope of 0.5 °C/min. Protein modelling and simulations. 3D structures were modeled through the SwissModel server 38 .
In order to compare the effect of temperature and oligomerization status on structure flexibility, molecular dynamic (MD) simulations were performed. 3D Models complexed with Cu, Zn (Ef-SOD1s) or Mn (Ef-SOD2) were subjected to 100 ns MD simulations at either 227.15 K (4 °C) or 300.15 K (27 °C) under constant temperature and pressure (1 atm). Analysis was performed on both monomers and dimers (Ef-SOD1s) or tetramers (Ef-SOD2). MD simulations were performed with the GPU implementation of the pmemd code 75 from AMBER14 76 employing the ff99SSBildn forcefield 77 .
Root mean square fluctuation (RMSF), the average residues flexibility during the trajectory and the root mean square deviation (RMSD) of particular residues, like ions binders, were evaluated in order to analyze temperature and oligomerization status effect on protein structure.