Introduction

The thermal stability of a series of homologous proteins correlates with the optimal growth temperatures of the source organisms from mesophiles growing at more than 20 °C to thermophiles growing at more than 45 °C (Gromiha et al. 1999; Akanuma et al. 2013; Sambongi et al. 2002). Therefore, the stability of the proteins appears to be related directly to the source’s natural temperature environment at more than 20 °C.

Cytochrome c′, the target protein in this study, is a heme protein found in a variety of Gram-negative bacteria (Ambler et al. 1991). It forms homo dimer, where each monomer has a four α-helix bundle structure containing a covalently bound penta-coordinate heme (Moore 1991). We have so far investigated the thermal stability of three cytochromes c′ from one thermophile, Hydrogenophilus thermoluteolus, which optimally grows at 52 °C (Goto et al. 1977) and two mesophiles, Shewanella amazonensis and Allochromatium vinosum, which optimally grow at 37 and 25 °C, respectively (Venkateswaran et al. 1998; Holm and Vennes 1970). These three cytochromes c′ exhibited different thermal stability, which was in accordance with the optimal growth temperatures of the source bacteria (Inoue et al. 2011; Fujii et al. 2013; Kato et al. 2015). Therefore, cytochromes c′ show a relationship between protein stability and bacterial growth temperature when they originate from mesophiles and thermophiles, as observed in the other proteins (Gromiha et al. 1999; Akanuma et al. 2013; Sambongi et al. 2002), but it remains unknown whether this relationship applies to cytochromes c′ from psychrophiles.

Fortunately, a highly homologous series of cytochromes c′ is found in Shewanella species, which include both psychrophiles and mesophiles. In this study, the stability and structure of four homologous cytochromes c′ from three psychrophilic and one mesophilic Shewanella species were compared. The cytochromes c′ used in this study were SBCP, SVCP, SLCP, and SACP, isolated from Shewanella benthica, Shewanella violacea, Shewanella livingstonensis, and Shewanella amazonensis, which optimally grow at 4, 8, 20, and 37 °C, respectively, under 0.1 MPa conditions (Kato et al. 1995; Kulakova et al. 1999; Venkateswaran et al. 1998). In particular, the amino acid sequence identity between SBCP and SVCP is strikingly high, 92%, and their psychrophilic source bacteria, S. benthica and S. violacea, are also both piezophiles, being isolated from extremely deep-sea environments, i.e., at more than 5000 m depth. In spite of the environmental similarity between S. benthica and S. violacea, SBCP was unexpectedly more stable than SVCP, and thus the both are examples regarding complex multiple adaptation of proteins to low temperature and high hydrostatic pressure.

Materials and methods

Gene sequencing and cloning

Chromosomal DNA of S. benthica DB6705, S. violacea DSS12, and S. amazonensis SB2B was prepared as described previously (Takenaka et al. 2010; Masanari et al. 2014), which were then used to amplify the SBCP, SVCP, and SACP genes to confirm their sequences. The PCR primers used for the gene amplification are listed in Table 1. The SLCP gene sequence was kindly provided by Dr. Kurihara of Kyoto University, Japan.

Table 1 Primers used in this study for gene sequencing and mutagenesis experiments
Full size table

The genes encoding for the mature SBCP, SVCP, SLCP, and SACP proteins fused with the Pseudomonas aeruginosa cytochrome c551 signal gene sequence at the N-terminal region were synthesized using optimized codons for Escherichia coli by a manufacturer (Thermo Fisher Scientific). The fused genes were inserted into the pKK223-3 vector at the EcoRI and SalI restriction sites under the control of a tac promoter.

Protein expression and purification

The plasmids carrying the genes for SBCP, SVCP, SLCP, and SACP were transformed into E. coli JCB387 containing the pEC86 plasmid carrying the genes for cytochrome c maturation (ccmABCDEFGH). E. coli JCB387 cells expressing cytochromes c′ were initially pre-cultured in liquid LB medium containing 100 µg/mL ampicillin and 30 µg/mL chloramphenicol. These LB pre-cultures (2 mL) were inoculated into 20 mL LB medium containing the same antibiotics, followed by shaking at 37 °C for 3 h. After this pre-cultivation, 20 mL LB medium was inoculated into 2 L minimal medium (Sambongi and Ferguson 1994) containing the same antibiotic, 0.8% (v/v) glycerol, and 50 μg/mL 5′-aminolevulinic acid in a 5 L Sakaguchi flask, and then the cells were grown at 37 °C with shaking for 21 h. The periplasmic protein fractions of the resulting cells were obtained by the cold osmotic shock method as described previously (Sambongi et al. 1996).

The periplasmic protein fractions containing the SBCP and SVCP proteins were dialyzed against 10 mM Tris-HCl buffer (pH 8.0) at 4 °C containing 100 mM NaCl to prevent protein aggregation. These fractions were then loaded onto a DEAE column (GE Healthcare, Tokyo) that had been equilibrated with the same buffer. The fractions containing cytochrome c′ proteins were eluted with the same buffer containing 300 mM NaCl. The cytochrome c′ fractions were dialyzed against the same buffer containing 100 mM NaCl, and then loaded onto a HiTrap Q column (GE Healthcare, Tokyo) that had been equilibrated with the same buffer. The cytochrome c′ fractions were eluted with the same buffer using the NaCl concentration gradient method (100–500 mM), and then dialyzed against 25 mM sodium acetate buffer (pH 5.0) at 4 °C containing 100 mM NaCl. The fractions were further loaded onto a HiTrap SP column (GE Healthcare, Tokyo) that had been equilibrated with the same buffer. The cytochrome c′ fractions were eluted with the same buffer using the NaCl concentration gradient method (100–300 mM), followed by loading onto a Superdex™ 75 column (GE Healthcare, Tokyo) equilibrated and eluted with 10 mM Tris-HCl buffer (pH 7.0) containing 150 mM NaCl at 4 °C.

The SLCP and SACP proteins were purified by the similar method as for SBCP and SVCP without the addition of 100 mM NaCl to the dialyzing buffer before the column chromatography procedure. The protein concentrations were determined spectrophotometrically using an extinction coefficient of 205 nm due to the peptide bond (Scopes 1974). The purity of the cytochrome c′ proteins was checked by sodium dodecyl sulfate polyacrylamide gel electrophoresis.

Spectral analysis of the recombinant proteins

The recombinant SBCP, SVCP, SLCP, and SACP proteins were air-oxidized or reduced with a grain of sodium dithionite. Visible absorption spectra (350–700 nm) of the oxidized and reduced SBVP, SVCP, SLCP, and SACP protein solutions (6 μM) in 100 mM potassium phosphate buffer (pH 7.0) at 25 °C were obtained with a JASCO V-730-Bio spectrometer at 25 °C (Jasco, Tokyo).

Sedimentation velocity analysis of the SBCP, SVCP, SLCP, and SACP protein solutions was performed with a Beckman Optima XL-I analytical ultracentrifuge (Beckman Coulter, California) equipped with a 4-hole An60 Ti rotor at 20 °C using 12-mm double-sector charcoal-filled Epon centerpieces with quartz windows to determine their subunit compositions. Each protein solution (5 μM) was dissolved in 10 mM Tris-HCl buffer (pH 7.5) containing 100 mM NaCl. Data were collected using absorbance optics at 390−400 nm with a radial increment of 0.003 cm and at a rotor speed of 42,000 rpm. The distribution of sedimentation coefficients (S) was analyzed using the continuous c(s) distribution model in the SEDFIT version 14.1 program (Schuck 2000), as previously reported (Fujii et al. 2017). The positions of the meniscus and cell bottom were floated during the fitting. The range of sedimentation coefficients for fitting was 0–15S with a resolution of 300.

Circular dichroism (CD) spectra (190–260 nm) of the oxidized SBCP, SVCP, SLCP, and SACP protein solutions (20 μM) in 100 mM potassium phosphate buffer (pH 7.0) were obtained with a JASCO J-820 CD spectrometer at 25 °C (Jasco, Tokyo).

Crystallization

The purified SBCP and SVCP protein solutions were dialyzed against milliQ water containing 100 mM NaCl, and then concentrated to ~ 15 mg/mL. The initial crystallized conditions were screened by the hanging-drop vapor diffusion method at 4 °C using Crystal Screens I and II (Hampton Research, California). The solution drops for crystallization were prepared by mixing 2 μL of protein solution and 1 μL of reservoir solution, followed by equilibration against 400 μL of a reservoir solution. The SBCP protein was crystalized using a reservoir solution consisting of 100 mM N-cyclohexyl-3-aminopropanesulfonic acid/sodium hydroxide buffer (pH 10.5) and 30% (v/v) polyethylene glycol 400 within 5 days, and the SVCP protein was also crystalized using a reservoir solution consisting of 100 mM N-cyclohexyl-3-aminopropanesulfonic acid/sodium hydroxide buffer (pH 9.4) and 26% (v/v) polyethylene glycol 400 within 7 days.

X-ray diffraction

The X-ray diffraction experiments on the SBCP and SVCP crystals were carried out on a Rigaku Micromax-007 rotating anode X-ray generator with a Cu target and an R-AXIS IV++ imaging plate detector (Rigaku, Tokyo). The SBCP and SVCP crystals were flash-cooled at − 173 °C in a stream of cold nitrogen, using the following cryoprotectant solution: 100 mM N-cyclohexyl-3-aminopropanesulfonic acid/sodium hydroxide buffer (pH 10.5) and 30% (v/v) polyethylene glycol 400. A total of 210 images was collected for each of the SBCP and SVCP crystals, respectively, at camera lengths of 120 mm for SBCP and 150 mm for SVCP and an oscillation angle of 1.0°. The SBCP and SVCP crystals diffracted to maximal resolutions of 1.71 and 2.14 Å, respectively (Table 2). The diffraction data sets were processed using the CrystalClear program (Rigaku, Japan). The SBCP and SVCP crystals belonged to space groups C21 (a = 72.67 Å, b = 52.47 Å, c = 50.93 Å) and P21 (a = 44.75 Å, b = 69.71 Å, c = 45.60 Å), respectively. The Matthews coefficient values were 2.45 Å3/Da and 2.27 Å3/Da, respectively, which correspond to solvent contents of 49.9 and 42.7%, respectively (Matthews 1968).

Table 2 Data collection, phasing and refinement statistics for the SBCP and SVCP crystals
Full size table

Molecular replacement and structure refinement

The structures of SBCP and SVCP were determined by the molecular replacement method using the Phaser program included in the PHENIX software suite (McCoy et al. 2007), and the crystal structure of Shewanella frigidimarina cytochrome c′ (PDB ID: 4ULV, Manole et al. 2015) was applied as a search model, which showed 68 and 70% amino acid sequence identity with SBCP and SVCP, respectively. The structural refinement, manual model building, and addition of water molecules were carried out using the phenix.refine and Coot programs (Adams et al. 2010; Emsley and Cowtan 2004). Finally, the refinement converged to an Rwork of 0.166 and an Rfree of 0.198 for SBCP, and an Rwork of 0.215 and an Rfree of 0.272 for SVCP. The geometry of the final SBCP and SVCP structure models was checked using the MolProbity program based on Ramachandran plot statistics (Table 2) (Chen et al. 2010). The final coordinates of SBCP and SVCP have been deposited in the Protein Data Bank (PDB ID: 6A3K and 6A3L).

Three-dimensional model structures of SLCP and SACP

As the amino acid sequence of SLCP from psychrophilic S. livingstonensis was 100% identical to that of similarly psychrophilic S. frigidimarina cytochrome c′ that had been determined through X-ray analysis (PDB ID: 4ULV, Manole et al. 2015), the three-dimensional structure of S. frigidimarina cytochrome c′ was adopted as a model structure of SLCP in this study. The three-dimensional model structure of SACP could be successfully simulated using the crystal structure of S. frigidimarina cytochrome c′ (PDB ID: 4ULV) as a template with the SWISS-MODEL workplace (Biasini et al. 2014). These SLCP and SACP three-dimensional model structures were used for comparison with X-ray crystal structures of SBCP and SVCP experimentally obtained in this study.

Thermal denaturation experiment

Thermal denaturation of the SBCP, SVCP, SLCP, and SACP proteins was measured as circular dichroism (CD) spectra in a pressure-proof cell compartment (Jasco, Tokyo) attached to a CD spectrometer. The protein solutions (20 μM) were dialyzed against 100 mM potassium phosphate buffer (pH 7.0) at 25 °C. The temperature-dependent CD ellipticity changes at 222 nm were monitored in a cuvette of 1-mm path length from 25 to 100 °C at intervals of 0.5 °C and a heating rate of 1.0 °C/min under 0.9 MPa pressure conditions.

The raw data were subjected to nonlinear least-squares fitting as described previously (Uchiyama et al. 2004). The data points were corrected for the slopes of the baselines for the native and denatured forms, and were normalized to calculate the fraction of protein denatured. The fraction denatured was plotted as a function of temperature, and the resulting thermal denaturation curves were used to determine the temperature at the midpoint of the transition (Tm).

Mutagenesis

Plasmids for expression of SBCP variants were prepared with a PrimeSTAR Mutagenesis Basal kit (TaKaRa Bio, Shiga), which had been applied to the wild-type gene inserted in the pKK223-3 vector. Ala-43, Pro-62, and Lys-87 in SBCP were separately substituted with Thr, Gln, and Val, respectively, which were the corresponding residues in SVCP. The primers used for the mutagenesis experiments are listed in Table 1. The correct introduction of mutations was confirmed by DNA sequencing. All the resulting plasmids were used for the expression of SBCP variants in E. coli JCB387 as described for the wild-type protein.

Results and discussion

Quality of the recombinant proteins

To confirm the quality of the recombinant proteins used in this study, the following analyses were performed. The sequence of SBCP gene from S. benthica DB6705 was determined using PCR primers constructed from the genome sequence of S. benthica KT99 that had been deposited in NCBI (Lauro et al. 2013). From the present sequencing results, we realized that three amino acid residues differed in the mature SBCP sequence from the two strains; the Glu-47, Ala-88, and Arg-105 residues of the KT99 strain were replaced with Gln, Thr, and Lys in the DB6705 strain, respectively. Here we adopted the strain DB6705 sequence (Accession number: LC406697), and the experiments in this study were carried out using the SBCP protein from that strain. The SVCP and SACP gene sequences obtained in this study, respectively, were confirmed to be 100% identical to the genome sequences that had been deposited.

The N-terminal sequences of the recombinant SBCP, SVCP, SLCP, and SACP proteins were determined to be Ser-Asn-Phe-Lys-Glu-, Gly-Asn-Phe-Lys-Glu-, Asn-Asn-Phe-Glu-Glu-, and Asn-Asn-Phe-Glu-Ser-, respectively, which appeared to be authentic N-terminal sequences. Therefore, the signal peptide of the P. aeruginosa cytochrome c551 was successfully cleaved from these mature proteins in the E. coli cells.

Visible absorption spectra of the oxidized SBCP, SVCP, SLCP, and SACP proteins showed typical peaks at 339–400 and 635–637 nm, and those of the reduced forms showed peaks at 427 and 545–549 nm (Fig. 1a). These results indicated that these proteins have penta-coordinate heme, as reported for other authentic mesophilic and thermophilic cytochromes c′ (Inoue et al. 2011; Kato et al. 2015).

Fig. 1
figure 1

Characteristics of SBCP, SVCP, SLCP, and SACP with spectral techniques. a Visible absorption spectra of the four proteins. Red and black solid lines represent the spectra of oxidized and reduced proteins, respectively. Relevant wavelengths are indicated by arrows. b Distribution of the sedimentation coefficients of the four proteins on analytical ultracentrifugation analysis. The data are fitted using continuous c(s) distribution method with the SEDFIT program. c CD spectra of the four proteins. Relevant wavelengths are indicated by arrows

Full size image

The subunit compositions of the four cytochromes c′ were determined through an analytical ultracentrifugation experiment. The S values of SBCP, SVCP, SLCP, and SACP were 2.7S with a population of more than 91%, the calculated molecular mass values being 31.1, 31.3, 30.5, and 29.5 kDa, respectively (Fig. 1b). All theoretical molecular masses of these cytochromes c′ were 14 kDa, calculated as a single subunit, indicating that these 2.7S peaks indicated dimeric structures. However, parts of the populations showed monomeric structures reflected by smaller S values (~ 0.9–1.1) than those of the dimeric structures, indicating that these four proteins each existed predominantly as a dimer with a small quantity of monomer.

The CD spectra of the SBCP, SVCP, SLCP, and SACP proteins showed specific negative peaks at 208 and 222 nm (Fig. 1c), indicating that they contain an α-helical structure. The results obtained on the present spectral analyses indicated that the heme environment, subunit composition, and secondary structure were quite similar in these four recombinant proteins, which were used in the following experiments.

Overall structures of SBCP and SVCP

As to the crystal structures of cytochromes c′ from psychrophilic Shewanella species, that of shallow-sea S. frigidimarina cytochrome c′ (PDB ID: 4ULV, Manole et al. 2015) is only available, and those of deep-sea Shewanella species remain unknown. Here, the X-ray crystal structures of SBCP and SVCP from deep-sea Shewanella species were first determined at 1.71 and 2.14 Å resolutions, respectively (Fig. 2a), by the molecular replacement method (McCoy et al. 2007) using the crystal structure of S. frigidimarina cytochrome c′, which exhibited 68 and 70% amino acid sequence identity with SBCP and SVCP, respectively. The two main chain structures superimposed well, the root-mean-square-deviation value being 0.42 Å. They consisted of homo dimeric subunits where each monomer had a four α-helix bundle structure, as observed in the other cytochrome c′ proteins. Around the heme in SBCP and SVCP, the side chains of His-123 residues were coordinated to the heme iron at the proximal site and those of Phe-17 were located at the vacant sixth distal site parallel with the heme (Fig. 2b). All these structural results for SBCP and SVCP are consistent with those of spectral analyses described above.

Fig. 2
figure 2

Structure comparison between SBCP (green) and SVCP (violet). a Overall structures. The main chain and heme structures are shown as ribbon and stick models, respectively. The N- and C-termini of each subunit are also indicated. b Heme environment

Full size image

Stability prediction through the structure comparison

The X-ray crystal structures of SBCP and SVCP obtained in this study were compared with the three-dimensional model structures of SLCP and SACP (see “Materials and methods”) to predict their stability. Previously we found that the stability of a dimeric cytochrome c′ from a thermophile, as compared with that of a homologous mesophilic counterpart, is due to strengthened interactions around the heme and at the subunit–subunit interface, both of which are in molecular interior regions (Fujii et al. 2017; Yamane-Koshizawa et al. 2018). We here examined whether interactions in molecular interior regions differed in the SBCP, SVCP, SLCP, and SACP proteins from psychrophiles to a mesophile to predict their stability differences.

Around the heme, Asp-70 in SACP formed a hydrogen bond with the propionate group of the heme, whereas Glu-70 in SBCP, SVCP, and SLCP did not (Fig. 3a). The Leu-89 residues in SBCP and SVCP hydrophobically interacted with the heme, and the corresponding Phe-89 residues in SLCP and SACP also interacted hydrophobically with the heme (Fig. 3a). The Phe-89 residues in SLCP and SACP appeared to fill the space around the heme more favorably than those in SBCP and SVCP, which may strengthen the interaction in SLCP and SACP. Furthermore, at the subunit–subunit interface, only one hydrogen bond between the Ser-15 residues in the two subunits was formed in SBCP and SVCP, while two between the Tyr-22 and Glu-58 residues, and three between the Ser-15 residues and the Tyr-22 and Glu-58 residues were formed in SLCP and SACP, respectively (Fig. 3b). These results indicated that the interactions in the molecular interior regions of SBCP and SVCP were similarly weakened as compared with those of SLCP and SACP, consistent with our previous findings for the cytochrome c′ proteins from mesophiles and a thermophile where thermophilic one had strengthened molecular interior as compared with mesophilic ones (Fujii et al. 2017).

Fig. 3
figure 3

Structural comparison among SBCP (green), SVCP (violet), SLCP (cyan), and SACP (yellow). a Interactions around the heme. Hydrophobic and other relevant residues are shown as sphere and stick models, respectively. b Interactions at the subunit–subunit interface. In both panels, the hydrogen bonds formed are indicated by red dotted lines

Full size image

Validation of stability prediction through thermal stability analysis

To validate the stability prediction through structural comparison among SBCP, SVCP, SLCP, and SACP described above, thermal denaturation experiments were carried out by means of CD spectral measurement. The normalized data for CD spectral changes at 222 nm during the thermal denaturation were plotted as a function of temperature (Fig. 4a). The values for the transition temperatures (Tm) during thermal denaturation of the SBCP, SVCP, SLCP, and SACP proteins were 55.5 ± 0.0, 48.3 ± 0.4, 51.3 ± 0.7, and 63.7 ± 0.0 °C, respectively, the mean and standard deviation values being obtained for three independent measurements. Among the four proteins, the Tm values of SVCP, SLCP, and SACP revealed a clear correlation with the optimal growth temperatures of the source bacteria (Fig. 4b), as predicted from their structures in the molecular interior regions (Fig. 3). Unexpectedly, however, the Tm value of SBCP was significantly higher than that of SVCP (Fig. 4b), although the interactions in the molecular interior regions of SBCP were similar to those in SVCP. There must be some other clues in their structures for this stability difference.

Fig. 4
figure 4

Thermal stability of four aShewanella cytochromes c′. Thermal denaturation, measured as CD spectra. The fractions of representative raw data are plotted at temperature interval of 2.5 °C for SVCP (violet circles), SLCP (cyan circles), SBCP (green circles), and SACP (yellow circles). The fitting curves are also shown. b Relationship between optimal growth temperatures of Shewanella species at 0.1 MPa and the Tm values of cytochromes c′. The symbol colors used are same as those in panel a. The dashed line shows the fitted linear line obtained using the data for SACP, SLCP, and SVCP

Full size image

Structural insights into the stability difference between SBCP and SVCP

Although SBCP and SVCP significantly differed in the stability, the two proteins exhibited 92% sequence identity, only ten out of 129 residues differed. Ser-1, Ala-43, Gln-47, Pro-62, Gln-66, Glu-68, Glu-78, Lys-87, Thr-88, and Gln-114 in SBCP were substituted by Gly-1, Thr-43, Glu-47, Gln-62, Leu-66, Asp-68, Asp-78, Val-87, Ser-88, and Lys-114 in SVCP, respectively. Among these differences, the SBCP Ala-43 residue was located in an α-helix region, in the interior of molecule (Fig. 5), and its helix conformation tendency (Chou and Fasman 1978) is higher than that of the corresponding Thr-43 in SVCP. The Pro-62 residue of SBCP was located in a loop on the molecular surface, which may stabilize its structure by reducing the entropy because its pyrrolidine ring adopts fewer conformations than the corresponding Gln-62 residue in SVCP, as discussed for many other proteins previously (Matthews et al. 1987; Sriprapundh et al. 2000; Goihberg et al. 2007). In addition, the SBCP Lys-87 residue formed a salt bridge with Asp-91 on the molecular surface, whereas the corresponding hydrophobic Val-87 residue in SVCP showed no interaction. Salt bridge formation on the molecular surface is reported to result in enhanced thermal stability of the proteins (Perl et al. 2000; Kuribayashi et al. 2017), whereas the hydrophobic residue on the molecular surface is reported to result in decreased thermal stability of the protein (Lee et al. 2005). Some of these specific structural differences may contribute to the higher thermal stability of SBCP than that of SVCP.

Fig. 5
figure 5

Structure comparison between SBCP (green) and SVCP (violet). The relevant amino acid residues are indicated

Full size image

Thermal stability analysis of the SBCP variants

To verify the structural insights described above, thermal denaturation experiments were carried out for the SBCP variants, which had been mutated so as to have the corresponding SVCP residues, by means of CD spectral measurement. The normalized data for CD spectral changes at 222 nm during thermal denaturation were plotted as a function of temperature (Fig. 6). The Tm values during thermal denaturation of the SBCP Ala-43 to Thr (A43T), Pro-62 to Gln (P62Q), and Lys-87 to Val (K87V) variants were 55.6 ± 0.2, 53.4 ± 0.2, and 51.9 ± 0.4 °C, respectively, the mean and standard deviation values being obtained for three independent measurements.

Fig. 6
figure 6

Thermal stability profiles of the SBCP variants. The fractions of representative raw data are plotted at a temperature interval of 2.5 °C for the A43T (closed triangles), P62Q (open circles), and K87V (closed circles) variants. The fitting curves are also shown, and red dashed line is the fitting curve of thermal denaturation for the SBCP wild-type protein, which has been already shown in Fig. 4a in this paper

Full size image

Among these variants, the Tm value of the A43T variant was essentially the same as that of the wild-type, and those of P62Q and K87V variants significantly decreased as compared with that of the wild-type. These results indicated that the molecular interior Ala-43 did not contribute to the SBCP stability as compared with SVCP, but the molecular surface Pro-62 and Lys-87 residues were responsible for the higher stability of SBCP, consistent with the structural prediction above.

Plausibly, the thermal stability difference between SBCP and SVCP can be related to the growth pressure of their source bacteria, psychrophilic S. benthica and S. violacea, growing optimally at 50 and 30 MPa, respectively. It could be possible that SBCP was more stable than SVCP because S. benthica is more piezophilic than S. violacea. However, there is presently no clear explanation about the relationship between the protein thermal stability and piezophilicity of source organism, presumably representing complex multiple adaptation to low temperature and high hydrostatic pressure. More information, together with those from other proteins (Hamajima et al. 2016, Fujii et al. 2018), should be collected for a better explanation of this relationship.

Conclusion

This study was aimed at further elucidation of the stability regulation in cytochromes c′ previously observed in thermophilic and mesophilic ones, extending to psychrophilic ones using Shewanella cytochromes c′. As expected, together with the previous findings (Fujii et al. 2017), the stability of cytochromes c′ from the psychrophile, mesophile, and thermophile is systematically regulated in molecular interior regions. An unanticipated observation in the present study was the stability difference between the SBCP and SVCP proteins, which was attributed to the differences in the amino acid residues located on the molecular surface. As the psychrophilic sources of SBCP and SVCP are also piezophilic, the adaptation of these proteins to multiple extreme environments of low temperature and high hydrostatic pressure appears to be complex, which may be required for proper protein function in the source organisms.