Introduction

Dihydrofolate reductase (DHFR) is an essential enzyme for the biosynthesis of purine nucleotides and certain amino acids and has, therefore, been studied as an indicator protein for bacterial adaptation to environments with extreme temperatures (Kim et al. 2005; Loveridge et al. 2009), pH (Redecke et al. 2007), salt concentrations (Wright et al. 2002), and pressures (Xu et al. 2003; Murakami et al. 2010). Although DHFRs isolated from these extremophiles show characteristic structures and properties, no definite correlation has been found between the observed molecular characteristics and the environmental conditions due to limited comparative studies.

It is known that DHFR from Escherichia coli (ecDHFR) living in ambient atmospheric conditions is highly flexible, and its structure and function are sensitively influenced by pressure (Kitahara et al. 2000; Ohmae et al. 2008). We previously found that the activity of ecDHFR decreases with increasing pressure, whereas DHFR from Shewanella violacea strain DSS12 (svDHFR), which was isolated from the Ryukyu Trench at a depth of 5,110 m, has an optimal activity at approximately 100 MPa (Ohmae et al. 2004). This finding suggests that DHFRs from deep-sea bacteria have characteristic pressure susceptibility and structural dynamics that are unique from those of organisms living at ambient atmospheric pressure.

Despite the apparent association between pressure and the function of DHFR from piezophilic S. violacea, other DHFRs from deep-sea isolates do not necessarily show such a relationship. For example, DHFRs from the genera Moritella and Photobacterium were not activated by pressure under the identical experimental conditions and no common amino acid sequences among DHFRs from high-pressure environments were identified (Hay et al. 2009; Murakami et al. 2010). However, as these results were obtained with bacteria of different genera, a direct comparison is unreasonable to clarify complicated pressure-adaptation mechanisms because their respective proteins may have undergone different molecular evolution. Therefore, comparative studies on the structure and function of DHFRs within an individual genus consisting of piezophilic and piezotolerant bacteria are necessary to give new insight into the pressure-adaptation mechanisms and molecular evolution of proteins. A target protein for such comparative studies has been so far limited to a single-stranded DNA-binding protein from marine Shewanella species: pressure tolerance of tetramer–monomer equilibrium of this protein correlated with the living pressure of these species (Chilukuri and Bartlett 1997; Chilukuri et al. 2002).

In this study, we therefore cloned and purified novel DHFRs from four piezophilic, one piezotolerant, and three piezosensitive Shewanella species inhabiting various extreme environments, as listed in Table 1, to survey the pressure-adaptation mechanisms of Shewanella species at the protein level. We followed up our previous work with svDHFR, which was pressure activated, to examine if DHFRs from Shewanella species too have the same pressure-tolerance behavior as has been seen in a single-stranded DNA-binding protein (Chilukuri and Bartlett 1997; Chilukuri et al. 2002). To our knowledge, this is the second example for the comparative study of proteins from Shewanella species inhabiting different pressure environments.

Table 1 Description of Shewanella species and DHFR names used in this study
Full size table

Materials and methods

Cloning and construction of Shewanella DHFR overexpression plasmids

Construction of the svDHFR overexpression plasmid was described previously (Ohmae et al. 2004). Genomic DNAs of seven bacteria (S. oneidensis, S. frigidimarina, S. putrefaciens, S. gelidimarina, S. benthica strain ATCC43992, S. benthica strain DB6705, and S. benthica strain DB21MT-2) were kindly provided by Dr. Nogi of JAMSTEC and were amplified prior to use with a GenomiPhi DNA Amplification kit (GE Healthcare UK Ltd., Buckinghamshire, UK). Plasmids overexpressing the DHFRs from S. oneidensis (soDHFR), S. frigidimarina (sfDHFR), and S. putrefaciens (spDHFR) were constructed by a PCR method using KOD-Plus-DNA polymerase (Toyobo, Osaka, Japan) and primer sets derived from the genome sequences of the three species that were previously registered in the DDBJ/GenBank/EMBL database under accession numbers and locus tags SO_3647 of AE014299 (S. oneidensis), Sfri_3084 of CP000447 (S. frigidimarina), and Sputcn32_0984 of CP000681 (S. putrefaciens). The amplified DNA fragments were ligated to pUC118 vectors digested with SmaI, and used to transform competent E. coli HB101 cells (Takara Bio, Otsu, Japan). The transformants carrying the overexpression plasmids were selected on LB plates containing ampicillin (100 μg/ml) and trimethoprim (20 μg/ml) at 37°C. The DNA sequences of the purified plasmids were confirmed using an ABI Prism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) with a DYEnamic ET Terminator Cycle Sequencing kit (GE Healthcare UK Ltd., Buckinghamshire, UK).

The overexpression plasmids for DHFRs from S. benthica strain ATCC43992 (sb43992DHFR), S. benthica strain DB6705 (sb6705DHFR), S. benthica strain DB21MT-2 (sb21DHFR), and S. gelidimarina strain ACAM456 (sgDHFR) were amplified by a PCR method with a mixed primer set based on sequences of the DHFRs from S. oneidensis, S. frigidimarina, S. putrefaciens, and six deep-sea bacteria (S. violacea, Moritella profunda, M. yayanosii, M. japonica, Photobacterium profundum, and Colwellia psychrerythraea with corresponding accession numbers and a locus tag of AB519243, AJ487535, AB505967, AB505968, AB505966, and CPS_4505 of CP000083, respectively). The amplified DNA fragments were purified using a Wizard SV Gel and PCR Clean-up System (Promega, Madison, WI, USA) and sequenced as described above. The up- and downstream regions were amplified using a DNA Walking Speed-up kit (Seegene, Seoul, Korea) with three sets of primers synthesized for each DHFR gene and were then sequenced. Finally, the entire DHFR gene of each species was amplified with a set of up- and downstream primers, KOD-Plus-DNA polymerase, and 0.1 ng of original genomic DNA. The DNA sequences of the amplified fragments were determined and registered in the DDBJ/GenBank/EMBL database under accession nos. AB538271 (sb43992DHFR), AB538272 (sb6705DHFR), AB538270 (sb21DHFR), and AB538273 (sgDHFR). The overexpression plasmids of these four DHFRs were then constructed using the primer set of svDHFR for sb43992DHFR, sb6705DHFR, and sb21DHFR, which have the identical amino acid sequences at the N- and C-terminal regions, and with the newly synthesized primers for sgDHFR. The DNA sequences of the resulting plasmids were confirmed as described above. The DNA sequences of the primers used in this study are listed in supplementary Table S1.

Protein purification

All DHFR proteins were over-expressed and purified from E. coli HB101 cells as described previously (Murakami et al. 2010). Briefly, cells were grown for 48 h at 25°C, harvested by centrifugation, washed with 20 mM Tris-hydrochloride (pH 8.0) containing 14 mM 2-mercaptoethanol and 0.1 mM EDTA (Buffer A), and broken by sonication. After centrifugation, 2% (wt/vol) streptomycin sulfate was added with stirring to the supernatant to precipitate nucleic acids. After centrifugation, a saturated ammonium sulfate solution was added with stirring to the supernatant to achieve a final concentration of 45%. The resulting precipitates were removed by centrifugation, and the supernatant was then loaded on a methotrexate-agarose affinity column (3 × 10 cm) (Sigma-Aldrich, St. Louis, MO, USA) equilibrated with Buffer A. The column was washed with Buffer A containing 0.5 M NaCl and the DHFR protein was eluted with Buffer A containing 3 mM folate. DHFR-containing fractions were dialyzed against 20 mM Tris-hydrochloride (pH 9.0) containing 14 mM 2-mercaptoethanol and 0.1 mM EDTA (Buffer B) for 3 days and concentrated with a DE-52 anion-exchange column (1 × 5 cm) (Whatman, Kent, UK). The samples were then dialyzed against Buffer B containing 3 M guanidine hydrochloride to completely remove the folate and against Buffer B containing 6 M urea to completely remove the guanidine hydrochloride. Proteins were then refolded by dilution with Buffer B to a concentration of approximately 10 μM. Refolded DHFRs were dialyzed against 20 mM Tris-hydrochloride (pH 8.0) containing 0.1 mM dithiothreitol and 0.1 mM EDTA for 3 days and concentrated with an Amicon Ultra filter (Millipore, Billerica, MA, USA). The aggregates were then removed by centrifugation at 18,000×g. Whole cell lysates and purified proteins (15 μg) were separated on a 15% SDS-PAGE gel.

The concentrations of ecDHFR and svDHFR were determined from the absorbance at 280 nm using molar extinction coefficients of 31,100 M−1cm−1 (Fierke et al. 1987) and 22,900 M−1cm−1 (Murakami et al. 2010), respectively. The concentrations of sb6705DHFR, sb21DHFR, sfDHFR, soDHFR, and spDHFR were determined using molar extinction coefficients of 27,300, 34,600, 22,800, 23,000, and 24,400 M−1cm−1 at 280 nm, respectively, which were estimated from the fluorescence intensities of the fully unfolded DHFRs in 6 M guanidine hydrochloride solutions.

Circular dichroism spectrometry

Far-ultraviolet circular dichroism (CD) spectra of the DHFRs were measured using a Jasco J-720W spectropolarimeter. The temperature was maintained at 15°C using a Peltier-controlled thermobath (Jasco PTC-348W). The solvent used was 20 mM Tris-hydrochloride (pH 8.0) containing 0.1 mM EDTA and 0.1 mM dithiothreitol, and the protein concentration was 10 μM.

Fluorescence spectrometry

Fluorescence spectra of DHFRs were measured at 15°C using a Jasco FP-750 spectrofluorometer with an excitation wavelength of 290 nm and an emission wavelength ranging from 300 to 450 nm. The solvent used for the measurements was identical to that used for the CD measurements. The protein concentration was 0.3–0.8 μM and the intensities of the spectra were normalized to 1 μM protein.

Equilibrium unfolding

Equilibrium unfolding of the DHFRs in the presence of urea (ultrapure product from ICN Biomedicals, Irvine, CA, USA) was monitored by fluorescence measurements at 15°C with a Jasco FP-750 spectrofluorometer. The excitation and emission wavelengths, solvent, and protein concentration were identical to those used in the above spectrometric measurements. All samples were fully equilibrated at each denaturant concentration before the fluorescence spectra were measured. The center of spectral mass (CSM) of each spectrum was calculated from the following equation (Vidugiris and Royer 1998):

$$ {\text{CSM}} = {{\sum F_{\text{i}} \nu_{\text{i}} } \mathord{\left/ {\vphantom {{\sum F_{\text{i}} \nu_{\text{i}} } {\sum F_{\text{i}} }}} \right. \kern-\nulldelimiterspace} {\sum F_{\text{i}} }} $$
(1)

where F i and ν i are the fluorescence intensity and wave number at wavelength i, respectively. The obtained CSM values were plotted against the urea concentration and fitted directly to the two-state unfolding model, native (N) ⇔ unfolded (U), by non-linear least-squares regression analysis as follows:

$$ {\text{CSM}} = {{\left\{ {{\text{CSM}}_{\text{N}} + {\text{CSM}}_{\text{U}} \exp \left( { - \Updelta G_{\text{u}} /RT} \right)} \right\}} \mathord{\left/ {\vphantom {{\left\{ {{\text{CSM}}_{\text{N}} + {\text{CSM}}_{\text{U}} \exp \left( { - \Updelta G_{\text{u}} /RT} \right)} \right\}} {\left\{ {1 + \exp \left( { - \Updelta G_{\text{u}} /RT} \right)} \right\}}}} \right. \kern-\nulldelimiterspace} {\left\{ {1 + \exp \left( { - \Updelta G_{\text{u}} /RT} \right)} \right\}}} $$
(2)

where ΔG u is the change in Gibbs free energy for unfolding, R is the gas constant, T is the absolute temperature, and CSMN and CSMU are the CSM values of the native and unfolded forms, respectively. CSMN and CSMU at a given urea concentration were estimated by assuming the same linear dependence of CSM values in the transition region as in the pure native (pre-transition region) and unfolded (post-transition region) states. The free energy change of unfolding, ΔG u, was assumed to be linearly dependent on the urea concentration (Pace 1985):

$$ \Updelta G_{\text{u}} = \Updelta G^{^\circ}_{{\text{u}}} - m\;[{\text{urea}}] $$
(3)

where ΔG °u is the free energy change of unfolding in the absence of a denaturant, and the slope, m, is a parameter reflecting the cooperativity of the transition. The urea concentration at the midpoint of the transition (ΔG °u = 0) was defined as C m.

Steady-state kinetics at atmospheric pressure

Steady-state kinetics of the enzymatic reaction was measured at 25°C at atmospheric pressure with a Jasco V-560 spectrophotometer. The solvent used was 20 mM Tris-hydrochloride (pH 7.0) containing 0.1 mM EDTA and 0.1 mM dithiothreitol. The concentrations of DHF (Sigma-Aldrich, St. Louis, MO, USA) and NADPH (Oriental Yeast, Tokyo, Japan) were determined spectrophotometrically using molar extinction coefficients of 28,000 M−1cm−1 at 282 nm for DHF and 6,200 M−1cm−1 at 339 nm for NADPH (Dawson et al. 1969). The enzymatic reactions were measured under various concentrations of NADPH or DHF while maintaining either ligand at a saturating concentration of 50 μM. The concentration of the DHFRs was adjusted to 1–2 nM by a methotrexate titration method (Williams et al. 1979). The enzymatic reactions were initiated by mixing the DHF solution with the enzyme-NADPH solution. Prior to mixing, both solutions were preincubated for 10 min at 25°C to equilibrate the temperature and eliminate the hysteresis effect (Penner and Frieden 1985). The initial velocity (v) of the enzymatic reaction was calculated from the time course of absorbance at 340 nm using a differential molar extinction coefficient of 11,800 M−1cm−1 (Stone and Morrison 1982). The Michaelis (K m) and catalytic (k cat) constants were determined with the program Origin (OriginLab, Northampton, MA, USA) by direct fitting of the initial velocity of the enzymatic reaction to the Michaelis–Menten equation using non-linear least-squares regression analysis.

Enzyme assay under high pressures

Pressure dependence of the enzyme activity was examined using a Jasco V-560 spectrophotometer equipped with a high-pressure absorbance cell unit (Teramecs PCI-400) and a hand pump (Teramecs PCI-500) as described previously (Ohmae et al. 2008; Murakami et al. 2010). Temperature was maintained at 25°C with a circulating thermobath (NESLAB RTE-5). The buffer used was 20 mM Tris-hydrochloride (pH 7.0) containing 0.1 mM EDTA, 0.1 mM dithiothreitol, 250 μM NADPH, and 250 μM DHF. The reaction mixture (1.5 ml) containing DHF and the enzyme-NADPH solution was loaded into the high-pressure cell, and the absorbance at 370 nm was then measured for 1–5 min at pressures ranging from 0 to 250 MPa. The initial velocity of the enzymatic reaction was calculated from the time course of the absorbance using the differential molar extinction coefficient at each pressure (e.g., 3,180 and 3,710 M−1cm−1 at 0.1 and 250 MPa, respectively) (Ohmae et al. 2008). The activation free energy (ΔG*) and the activation volume (ΔV*) of the enzymatic reactions were calculated using the following equation, which is applicable at saturating substrate concentrations,

$$ \Updelta V^{ * } = \partial \Updelta G^{ * } /\partial P = \partial \left( { - RT\ln k_{\text{cat}} } \right)/\partial P = \partial \left( { - RT\ln v} \right)/\partial P $$
(4)

where R is the gas constant, T is the temperature, P is the pressure, and v is the initial velocity of the enzymatic reaction (Ohmae et al. 2008).

Results and discussion

Amino acid sequences of Shewanella DHFRs

To gain insight into the pressure-adaptation mechanisms and molecular evolution of DHFR proteins, we first compared the amino acid sequences of sb43992DHFR, sb6705DHFR, sb21DHFR, sfDHFR, sgDHFR, soDHFR, and spDHFR, which were determined from the DNA sequences, to those of ecDHFR and svDHFR (Fig. 1). From the alignments, the sequence identities of sb43992DHFR, sb6705DHFR, sb21DHFR, sfDHFR, sgDHFR, soDHFR, spDHFR, and ecDHFR to svDHFR were 89, 90, 89, 74, 76, 76, 74, and 55%, respectively. Overall, the N-terminal regions, particularly in the first 60 residues, showed high sequence conservation, whereas the C-terminal regions showed wider sequence variations, as previously found for DHFRs from other deep-sea species (Murakami et al. 2010). The Gly95-Gly96 residues, which are known to be essential for the proper folding of ecDHFR, were also completely conserved in all of the Shewanella DHFRs. Moreover, nearly all residues in the substrate and cofactor binding sites (with the exception of Val6 of sb43992DHFR, Asn17 of sb21DHFR, Gln19 of svDHFR and sb21DHFR, Lys76 of sfDHFR, Cys77 of sgDHFR, Met94 of spDHFR, and a residue 102) were conserved in all Shewanella DHFRs (Fig. 1). The Met20 loop comprising residues 10–24, which is a well-characterized and essential functional region, was also fully conserved in all of the Shewanella DHFRs, with the exception of only Gln19 of svDHFR, and Asn17 and Gln19 of sb21DHFR.

Fig. 1
figure 1

Amino acid sequences of DHFRs from E. coli and eight Shewanella species. Multiple alignments were conducted with the CLUSTALW program (Thompson et al. 1994) on a DNA Data Bank of Japan server (http://www.ddbj.nig.ac.jp/). The numbering of residues is based on the ecDHFR sequence. The symbols “>” before sequences indicate the DHFRs used for protein purification. The symbols “*”, “:” and “.” below the alignment indicate fully, strongly, and weakly conserved residues, respectively. Amino acid residues different from those of svDHFR are indicated by gray boxes. Non-conserved residues in the substrate- and cofactor-binding sites are enclosed by open boxes. Fully conserved cis-glycine bond (Gly95-Gly96) is indicated by arrows above the sequences. Sequence lengths, levels of identity to svDHFR, and estimated isoelectric points (pI) are also indicated at the end of each sequence

Full size image

It is noteworthy that the amino acid sequences of the Shewanella DHFRs from the deep-sea isolates (svDHFR, sb43992DHFR, sb6705DHFR, and sb21DHFR) are similar to those from the ambient-atmospheric isolates (sfDHFR, sgDHFR, soDHFR, and spDHFR), but largely different from that of ecDHFR. This result is consistent to those for the single-stranded DNA-binding protein (Chilukuri and Bartlett 1997) and the phylogenetic studies from 16S rDNA sequences (Venkateswaran et al. 1999; Kato and Nogi 2001), indicating that members within the genus Shewanella migrated to the deep sea after it separated from the genus Escherichia. From the DHFR alignments, it was also identified that only three residues were different between all four deep-sea isolates and the ambient atmospheric Shewanella species, including E. coli. The residues Ala53, Leu58, and Ser105 of the deep-sea DHFRs were substituted to proline, histidine or lysine, and cysteine or proline, respectively, in the ambient atmospheric bacteria.

Interestingly, for the translation of sgDHFR, S. gelidimarina appears to have adopted a unique initiation codon, TTG, since the nearest stop codon is located at the 36th upstream DNA base before the estimated N-terminus. However, we could not purify and determine the N-terminal residue of sgDHFR directly from this bacterium. Therefore, the structural and enzymatic properties of sgDHFR were not examined in the present study. In addition, sb43992DHFR was also omitted from the comparative studies of structure and function because this protein was predicted to have very similar properties to sb6705DHFR as they differed by only one amino acid residue (Ala6 or Val6). Thus, we compared the structure, stability, and activity of six Shewanella DHFRs, with the exception of sgDHFR and sb43992DHFR.

Following overexpression and purification of the six purified Shewanella DHFRs, the proteins were examined on a SDS-PAGE gel (Fig. 2). Although single bands corresponding to the DHFRs were predominantly observed, these proteins had additional 14 residues (MTMITNSSSVPGTS) at the N-terminal region that originated from the lacZ′ gene on the vector as judged from the mass spectrometry. As the expression of the enzymes significantly decreased after elimination of these residues, we used the DHFRs with these additional residues in all experiments.

Fig. 2
figure 2

SDS-PAGE of E. coli and Shewanella DHFRs. Lanes 1–8 whole protein extracts from E. coli transformants overexpressing svDHFR, sb43992DHFR, sb21DHFR, sb6705DHFR, sfDHFR, sgDHFR, soDHFR, and spDHFR. Lanes 9–14 affinity-column purified svDHFR, sb21DHFR, sb6705DHFR, sfDHFR, soDHFR, and spDHFR. Lane 15 molecular weight markers. The acrylamide concentration of the gel was 15%

Full size image

Structure of Shewanella DHFRs

To examine the structure of the Shewanella DHFRs, they were subjected to far-ultraviolet CD spectrometry at 15°C and pH 8.0 at atmospheric pressure (Fig. 3a). Compared with ecDHFR, the CD spectra of the Shewanella DHFRs were largely blue-shifted. The peak intensities of sfDHFR and soDHFR were nearly identical to those of ecDHFR; however, svDHFR and sb6705DHFR clearly showed increased peak intensities whereas those of spDHFR and sb21DHFR were slightly decreased. These modified CD spectra do not necessarily indicate that the DHFR secondary structures differ from that of ecDHFR because even single amino acid substitutions in ecDHFR can induce large CD spectral changes (Ohmae et al. 1996, 2005). It has also been observed that DHFRs from other deep-sea bacteria, such as P. profundum, M. yayanosii, and M. japonica, show large changes in CD spectra (Murakami et al. 2010). Rather, these large CD spectral changes may be ascribable to the highly flexible structure of DHFR, which could perturb the CD excitation through movements of the secondary structure and/or altered contributions of aromatic side chains typically induced by exciton coupling of Trp47 and Trp74 in ecDHFR (Kuwajima et al. 1991, Ohmae et al. 2001). As shown in the inset of Fig. 3a, no significant difference in the CD spectra of spDHFR was induced by the addition of either cofactor (NADPH) or substrate (DHF), or by the addition of both cofactor and a substrate analog (folate). Similar results were also observed for the five other Shewanella DHFRs (data not shown). These results confirm that all Shewanella apo-DHFRs are in a folded (native) conformation even though they possess an additional 14 residues at the N-terminal region.

Fig. 3
figure 3

Far-ultraviolet CD (a) and fluorescence (b) spectra of E. coli and Shewanella DHFRs at pH 8.0 and 15°C at atmospheric pressure. The solvent used was 20 mM Tris-hydrochloride containing 0.1 mM EDTA and 0.1 mM dithiothreitol. Open circles ecDHFR, filled circles svDHFR, filled triangles sb21DHFR, open triangles sb6705DHFR, filled squares sfDHFR, open squares soDHFR, and cross symbol spDHFR. Inset of panel a far-ultraviolet CD spectra of spDHFR with and without ligands. Open circle apo-spDHFR, filled circle spDHFR with 250 μM NADPH, filled triangle spDHFR with 250 μM dihydrofolate, and open triangle spDHFR with 250 μM NADPH and 250 μM folate

Full size image

In addition to the CD spectra, the fluorescence spectra of the Shewanella DHFRs and ecDHFR were determined at 15°C and pH 8.0 at atmospheric pressure (Fig. 3b). The observed fluorescence spectra were mainly attributable to the content and environment of the tryptophan residues in the proteins. All Shewanella DHFRs, with only three tryptophan residues, showed reduced peak intensities compared with ecDHFR, which has five tryptophan residues. Small blue shifts of the peak wavelengths from 345 nm to 340–343 nm, with the exception of soDHFR, suggest that their tryptophan residues exist in more hydrophobic environments than those of ecDHFR. The spectrum of soDHFR showed the identical peak wavelength to that of ecDHFR, indicating that the tryptophan residues of soDHFR locate in more hydrophilic environments than those of the other Shewanella DHFRs. The observation that spDHFR showed a more significantly decreased peak intensity than the other Shewanella DHFRs even though the location of all three tryptophan residues is completely conserved (Fig. 1), suggests that the quantum yield of tryptophan residues and the surrounding polarity may be considerably different in spDHFR. Taken together, the spectral data of the Shewanella DHFRs presented here indicate that the enzymes have not undergone global structural changes, but rather, have retained localized structural alterations, as evidenced by changes in their CD and fluorescence spectra compared with those of ecDHFR. These observations are confirmed by the fact that these DHFRs retained enzymatic activities comparable to that of ecDHFR, as shown below.

Recently, we determined the crystal structure of the DHFR from the deep-sea bacterium M. profunda (PDB ID: 2zza), which has a 55% sequence identity with ecDHFR (Xu et al. 2003), in a ternary complex with NADP+ and folate. The crystal structures of the apo form and DHFR–NADPH–MTX ternary complex of the identical protein have also been reported (Hay et al. 2009). These crystal structures are nearly identical to those of ecDHFR, which is consistent with our present spectroscopic observations that have revealed the Shewanella DHFRs have only marginal structural differences compared with ecDHFR.

Structural stability of Shewanella DHFRs

The structural stability of the Shewanella DHFRs was evaluated based on the CSM of the fluorescence spectra as a function of urea concentration at 15°C and pH 8.0 (Fig. 4). Although ecDHFR showed a cooperative transition as monitored by CD spectroscopy (Gekko et al. 1994; Ohmae et al. 1996), the transition curves of all Shewanella DHFRs were clearly shifted to lower urea concentrations with reduced cooperativity, suggesting that their structures were destabilized. The CSM values in the absence of urea varied considerably among the DHFRs from the different bacterial species (Fig. 4); however, they were nearly identical in the post-transition region (>4 M urea) for all DHFRs. These results indicate that the tryptophan residues of the DHFRs examined here, which are exposed to different environments in the native state, are fully exposed to the solvent in the unfolded state.

Fig. 4
figure 4

Urea concentration dependence of the center of fluorescence spectral mass (CSM) of E. coli and Shewanella DHFRs at pH 8.0 and 15°C. The solvent used was 20 mM Tris-hydrochloride containing 0.1 mM EDTA, and 0.1 mM dithiothreitol. Open circle ecDHFR, filled circle svDHFR, filled triangle sb21DHFR, open triangle sb6705DHFR, filled square sfDHFR, open square soDHFR, and cross symbol spDHFR. The solid lines represent the theoretical fit to a two-state transition model with the parameter values shown in Table 2. The figure inset shows the dependence of the apparent free energy change of unfolding (ΔG u) on the urea concentration

Full size image

The transition curve of svDHFR obtained from the CSM values closely resembled the curve from the molar ellipticity data (data not shown), suggesting that the secondary and tertiary structures are disrupted by urea at the same time and the unfolding of Shewanella DHFRs follows the two-state unfolding model as found for ecDHFR (Touchette et al. 1986). As shown in Fig. 4, the transition curves of all Shewanella DHFRs showed a good theoretical fit to the two-state unfolding model as determined using Eqs. 2 and 3. The obtained thermodynamic parameters are listed in Table 2. The values of ΔG u (26.9  ± 3.7 kJ/mol), m (9.4 ± 1.2 kJ mol−1 M−1), and C m (2.9 ± 0.5 M) for ecDHFR were consistent with our previous findings (25.4 ± 0.8 kJ/mol, 8.2 ± 0.3 kJ mol−1 M−1, and 3.11 M, respectively) from the molar ellipticity changes under nearly identical conditions (Gekko et al. 1994). The ΔG u values of Shewanella DHFRs ranged from 6.7 ± 1.0 kJ/mol (soDHFR) to 8.7 ± 1.2 kJ/mol (sb21DHFR), which were clearly smaller than that of ecDHFR. Reduced structural stability (3.2 kcal/mol) was also reported for another deep-sea DHFR from M. profunda, (Xu et al. 2003). However, the obtained ΔG u values were not correlated with the environmental pressure from which the Shewanella species were isolated. Similar trends were also observed for the values of m and C m. Taken together, these results suggest that the stability of the Shewanella DHFRs at atmospheric pressure does not largely vary between the strains and has not significantly changed during the evolutional adaptation of bacteria to the high-pressure environment of the deep sea.

Table 2 Thermodynamic parameters for urea denaturation of ecDHFR and six Shewanella DHFRs at 15°C and pH 8.0
Full size table

Steady-state kinetics

To characterize the enzymatic function of the Shewanella DHFRs, their steady-state enzyme kinetics were examined at pH 7.0 and 25°C under atmospheric pressure. The kinetic parameters, K m and k cat, obtained for ecDHFR and the six Shewanella DHFRs are summarized in Table 3. The K m values of soDHFR were 1.0 ± 0.1 and 1.4 ± 0.2 μM for DHF and NADPH, respectively, which were slightly smaller than the corresponding values (1.1 ± 0.1 and 1.9 ± 0.0 μM) of ecDHFR, suggesting that soDHFR has slightly increased affinities for the substrate and cofactor. However, the K m values of the other Shewanella DHFRs ranged from 1.4 (sfDHFR) to 2.4 μM (sb6705DHFR) for DHF and from 2.6 (sb21DHFR) to 4.9 μM (spDHFR) for NADPH, indicating that they have a little weaker affinities for the substrate and cofactor than ecDHFR. In contrast, the k cat values of the Shewanella DHFRs ranged from approximately 45 s−1 (soDHFR) to 100 s−1 (spDHFR), which were 2.3- to 5.2-fold larger than that of ecDHFR and resulted in 2.0- to 3.9-fold increases in k cat /K m. These results indicate that under these experimental conditions, the enzyme activities of the Shewanella DHFRs are higher than that of ecDHFR predominantly due to the enhanced turnover rates that serve to overcome the slightly decreased substrate and cofactor affinities. Three other DHFRs from the deep-sea bacteria, M. japonica, M. yayanosii, and P. profundum, also exhibited larger K m and k cat values than ecDHFR (Murakami et al. 2010). Therefore, the increase of the K m and k cat values compared to ecDHFR seems common in many DHFRs. Further, the fact that no significant differences in the kinetic parameters were observed between Shewanella species suggests that in addition to structural stability, the enzyme function of these DHFRs at atmospheric pressure has not largely changed during the evolutional adaptation of bacteria to the high-pressure environment of the deep sea.

Table 3 Steady-state kinetic parameters of ecDHFR and six Shewanella DHFRs at 25°C and pH 7.0 under atmospheric pressure
Full size table

Pressure effects on enzyme activity

The effects of pressure on enzymatic activity were measured to investigate the potential high-pressure adaptation of DHFR function. Figure 5a shows the relative activity of ecDHFR and six Shewanella DHFRs as a function of the hydrostatic pressure over a range of 0–250 MPa at pH 7.0 and 25°C. It is clear that the enzyme activities of sfDHFR and spDHFR obtained from the ambient atmospheric species monotonously decreased with increasing pressure to 10–20% at 250 MPa. However, the DHFR from the ambient atmospheric species S. oneidensis showed pressure tolerance to approximately 50 MPa and retained 80% of its activity at 100 MPa, indicating that soDHFR also could function sufficiently in the deep sea. Although the biological meanings of this character are not clear, it is noteworthy that DHFR protein can obtain the pressure-tolerant activity independent of high-pressure environment. On the other hand, the DHFRs from two deep-sea Shewanella species, sb21DHFR and svDHFR, had clearly enhanced activities at moderate pressures of 25–150 MPa, suggesting that these DHFRs are adapted to the deep-sea environment. In contrast, the DHFR from the other deep-sea isolate, sb6705DHFR, showed only limited pressure tolerance up to 25 MPa and was monotonously inactivated at higher pressure at levels similar to sfDHFR and spDHFR, although these enzymes are still functional within the cells at their living pressure. Thus, DHFRs from ambient atmospheric species are not necessarily incompatible with deep-sea environments, and interestingly, DHFRs from deep-sea species are not necessarily tolerant to high pressure. These results also support the idea that DHFRs from deep-sea Shewanella species did not largely change functional characters during the evolutional adaptation to high-pressure environments after migrating into the deep sea.

Fig. 5
figure 5

Pressure dependence of relative enzyme activities (a) and activation free energies (b) of E. coli and Shewanella DHFRs at pH 7.0 and 25°C. The solvent used was 20 mM Tris-hydrochloride (pH 7.0) containing 0.1 mM EDTA, 0.1 mM dithiothreitol, 250 μM NADPH, and 250 μM dihydrofolate. Open circle ecDHFR data from Murakami et al. 2010, filled circle svDHFR, filled triangle sb21DHFR, open triangle sb6705DHFR, filled square sfDHFR, open square soDHFR, and cross symbol spDHFR. Lines in b were drawn using least-squares linear regression. The analyses were performed in duplicate and the values represent the mean ± error

Full size image

The activation free energy (ΔG*) calculated using Eq. 4 and the data from Fig. 5a were plotted against pressure (Fig. 5b). All Shewanella DHFRs showed a bimodal linear relationship between ΔG* and pressure, whereas a linear relationship was observed for ecDHFR. The activation volumes, ΔV*, of the enzyme reactions calculated from the slopes are listed in Table 4. Although the quantitative interpretation of these ΔV* values is limited because the enzyme kinetics and structures of these DHFRs are unknown, these volumetric data clearly demonstrate a pressure-dependent function of the Shewanella DHFRs. All DHFRs show positive ΔV* values in the pressure regions examined, except the low-pressure regions for svDHFR and sb21DHFR. The positive ΔV* values indicate that the activated state has a larger partial volume than the reactant in the rate-limiting process of the enzyme reaction, leading to pressure-induced deactivation. Conversely, the negative ΔV* values are indicative of pressure-induced activation. The DHFRs with a small positive ΔV* value may also be pressure tolerant.

Table 4 Activation volumes for enzymatic reactions of ecDHFR and six Shewanella DHFRs at 25°C and pH 7.0
Full size table

Interestingly, the calculated ΔV* values for Shewanella DHFRs are considerably different in the low- and high-pressure regions. As the ΔV* values are larger in the high-pressure regions, the rate-limiting step of the enzyme reaction likely shifts to a more pressure-sensitive step at pressures above a critical pressure between 25 and 125 MPa. Such modification of the enzyme reaction would be expected in the two deep-sea DHFRs, svDHFR and sb21DHFR, which have small positive ΔV* values (5.6 and 6.5 ml/mol, respectively), because S. benthica strain DB21MT-2 has pressure tolerance to the deepest sea on the earth. It is likely that these DHFRs did not adapt to the deep-sea environment, but were able to retain function under high pressure by previously accumulated mutations. In contrast, the very large positive ΔV* values of 30.5, 23.3, and 29.0 ml/mol observed for sfDHFR, spDHFR, and sb6705DHFR, respectively, suggest that more substantial conformational changes occur in these DHFRs under high pressure, which can also be inferred from their negligibly small enzyme activities.

Molecular adaptation of DHFR to high-pressure environments

As revealed in this study, there were no significant differences in the structure, stability, and function of DHFRs between Shewanella species living in deep-sea and ambient atmospheric-pressure environments. DHFRs from Shewanella species living at atmospheric pressures had more similar amino acid sequences to those from deep-sea Shewanella species than to ecDHFR (Fig. 1). The structural stability and steady-state kinetics of the Shewanella DHFRs were also similar to each other, but markedly differed from those of ecDHFR (Tables 2, 3). The finding that the DHFR from S. oneidensis, which typically inhabits ambient atmospheric-pressure environments, retained enzyme activity under moderate pressures (100 MPa) indicates that this DHFR also has the potential to adapt to deep-sea environments. In contrast, the DHFR from the deep-sea isolate S. benthica strain DB6705 was strongly inactivated by pressure as shown in Fig. 5a, and DHFRs from two other deep-sea bacteria, M. japonica (isolated from the Japan Trench at a depth of 6,356 m) and P. profundum (isolated from the Sulu Sea at a depth of 2,551 m), were monotonously inactivated by pressure similar to ecDHFR (Murakami et al. 2010). Although the pressure dependence of the enzyme activity may be influenced by the differences between in vitro and in vivo conditions such as temperature, pH, concentrations of salt, substrate, and cofactor, and the additional N-terminal residues presented in the DHFRs used, these results suggest that DHFRs from deep-sea Shewanella species need not undergo major structural and functional changes for adaptation within the chemical and physical constraints of their normal cellular milieu, and members whose enzymes could retain functional activity under high pressure migrated into the deep sea. This speculation is also supported by the fact that all five genera include piezophilic members, namely, Shewanella, Moritella, Photobacterium, Psychromonas, and Colwellia, also include nonpiezophilic species living at atmospheric pressure (DeLong et al. 1997; Nogi et al. 2002). Further comparative studies on other enzymes will provide useful information concerning the mechanism of molecular evolution and high-pressure adaptation of deep-sea bacteria.