Functional properties of the protein disulfide oxidoreductase from the archaeon Pyrococcus furiosus

A member of a novel protein family related to protein disulfide-isomerase

Authors


S. Bartolucci, Dipartimento di Chimica Biologica, Università degli Studi di Napoli Federico II, via Mezzocannone 16, 80134 Napoli, Italy. Fax: +39 81 2534614, Tel.: +39 81 2534732, E-mail: bartoluc@unina.it

Abstract

Protein disulfide oxidoreductases are ubiquitous redox enzymes that catalyse dithiol–disulfide exchange reactions with a CXXC sequence motif at their active site. A disulfide oxidoreductase, a highly thermostable protein, was isolated from Pyrococcus furiosus (PfPDO), which is characterized by two redox sites (CXXC) and an unusual molecular mass. Its 3D structure at high resolution suggests that it may be related to the multidomain protein disulfide-isomerase (PDI), which is currently known only in eukaryotes. This work focuses on the functional characterization of PfPDO as well as its relation to the eukaryotic PDIs. Assays of oxidative, reductive, and isomerase activities of PfPDO were performed, which revealed that the archaeal protein not only has oxidative and reductive activity, but also isomerase activity. On the basis of structural data, two single mutants (C35S and C146S) and a double mutant (C35S/C146S) of PfPDO were constructed and analyzed to elucidate the specific roles of the two redox sites. The results indicate that the CPYC site in the C-terminal half of the protein is fundamental to reductive/oxidative activity, whereas isomerase activity requires both active sites. In comparison with PDI, the ATPase activity was tested for PfPDO, which was found to be cation-dependent with a basic pH optimum and an optimum temperature of 90 °C. These results and an investigation on genomic sequence databases indicate that PfPDO may be an ancestor of the eukaryotic PDI and belongs to a novel protein disulfide oxidoreductase family.

Abbreviations
PfPDO

protein disulfide oxidoreductase from the archeon Pyrococcus furiosus

DMS

dimethyl suberimidate

MgATP

5 mm MgCl2, 2 mm ATP

PDI

protein disulfide-isomerase

Protein disulfide oxidoreductases are ubiquitous redox enzymes that catalyse dithiol–disulfide exchange reactions. These enzymes share a CXXC sequence motif at their active sites. The two cysteines can undergo reversible oxidation–reduction by shuttling between a dithiol and a disulfide form in the catalytic process. Protein disulfide oxidoreductases comprise the families of thioredoxin, glutaredoxin, protein disulfide-isomerase (PDI), and DsbA (disulfide-bond forming) and their homologs. Whereas thioredoxin and glutaredoxin mainly catalyse the reduction of disulfides, PDI and DsbA catalyse the formation or rearrangement of disulfide bridges in the protein-folding process.

Protein disulfide oxidoreductases have been well studied in bacteria and eukarya, although to date only a few archaeal members of this protein family have been isolated, and therefore very little is known about protein disulfide oxidoreductases in archaea.

A small redox protein with a molecular mass of 12 kDa was purified from the archaeon Methanobacterium thermoautotrophicum by McFarlan et al. [1]. This protein can catalyse the reduction of insulin disulfides and function as a hydrogen donor for Escherichia coli ribonucleotide reductase. The presence of the active-site motif CPYC, which is conserved in all glutaredoxins, suggested that it acts as a glutaredoxin-like protein. Surprisingly, however, the reduced enzyme does not react with either thioredoxin reductase or glutathione differently from other thioredoxins and glutaredoxins [2]. In the hyperthermophilic archaeon Methanococcus jannaschii[3], a thioredoxin homologue was identified (Mj0307) [4] that has the sequence CPHC, which had never before been observed in either thioredoxins or glutaredoxins. It exhibits biochemical activities similar to thioredoxin, although its structure is more similar to glutaredoxin. The observation that a single thioredoxin system is present in M. jannaschii and Mb. thermoautotrophicum suggested that a single thioredoxin-like protein with a glutaredoxin-like structure is enough to maintain redox homeostasis in the archaeal methanogen [5].

Guagliardi et al. [6] purified a protein disulfide oxidoreductase from the hyperthermophilic archaeon Sulfolobus solfataricus. Given its ability to catalyse the reduction of insulin disulfides in the presence of dithiothreitol, the protein was named thioredoxin. The monomeric form of the enzyme has an unusual molecular mass of about 26 kDa, compared with that observed in thioredoxin and glutaredoxin (12 kDa).

A homologous protein disulfide oxidoreductase was purified from the hyperthermophilic archaeon Pyrococcus furiosus (PfPDO) [7]. PfPDO showed close similarity to the S. solfataricus protein in molecular mass (25 648 Da) and dithiothreitol-dependent insulin reduction activity. In addition, both proteins displayed thiol transferase activity by catalysing the reduction of disulfide bonds in l-cysteine [7,8]. The PfPDO primary structure does not show any overall sequence similarity to known protein disulfide oxidoreductases. Interestingly, it has two potential active sites with the conserved CXXC sequence motif. A CPYC sequence is located at the C-terminal half of PfPDO, which is the conserved active sequence of the glutaredoxin family, usually located at the N-terminus. In addition, a CQYC sequence, which has never been observed in any other protein disulfide oxidoreductase, is present at theN-terminal half of the protein. The PfPDO crystal structure provides some intriguing challenges to the understanding of the enzyme's function [9–11]. The protein consists of two homologous units with low sequence identity (18%). Each unit contains a thioredoxin fold, and the accessibilities of the two CXXC active sites are rather different. The presence of two homologous units in the same protein resembles the structure of PDI; in fact, the PDI molecule possesses two thioredoxin-like domains with two active sites. Interestingly, whereas thioredoxins and glutaredoxins were identified in both prokaryotes and eukaryotes, DsbA was only found in prokaryotes. PDIs, with multiple thioredoxin/glutaredoxin domains within a single polypeptide are known in eukaryotes, and it is likely that the first step in their molecular evolution was the duplication of an ancestral thioredoxin/glutaredoxin domain [12]. The unusual structural features of PfPDO suggest that this enzyme probably represents a new member of the protein disulfide oxidoreductase superfamily and a new form of isomerase compared with PDI and DsbA. Functional studies of PfPDO are essential to support this finding, but have not yet been conducted. Therefore, this work focuses on the functional characterization of the PfPDO protein in an attempt to elucidate its relation with the eukaryotic multidomain PDI. Functional data revealed that the archaeal protein not only has oxidative and reductive activity, but also isomerase activity. This is the first example of an archaeal protein characterized with disulfide isomerase activity.

To investigate the specific roles of each PfPDO redox site, two single mutants (C35S and C146S) were constructed, in which the N-terminal active-site cysteine residue (Cys35 or Cys146) was replaced by serine, and a double mutant (C35S/C146S). All mutants were expressed, purified, and their activities compared with that of the wild-type protein. To compare the PfPDO with PDI for ATP binding and hydrolysis, the archaeal protein was also tested for its ATPase activity.

Experimental Procedures

Materials

Bovine insulin, glutathione disulfide (GSSG), glutathione (GSH), bovine liver PDI, horse liver alcohol dehydrogenase, bovine pancreas scrambled RNase and all the other reagents used were from Sigma. Molecular-mass standards for SDS/PAGE were obtained from Pharmacia or Bio-Rad. E. coli strain JM101 was purchased from Boehringer. Expression vector pET22(b+), E. coli strain BL21(DE3), and CJ236 E. coli strain were from AMS Biotechnology (Abingdon, UK). Radioactive materials were obtained from New England Nuclear/Life Science (Boston, MA, USA). 8-Azido-[32P]ATP[αP] was obtained from ICN. Deoxynucleotides and restriction and modification enzymes were from Boehringer. All materials used for gene amplification were supplied by Stratagene Cloning Systems. All synthetic oligonucleotides and the peptide designed by Ruddock et al. [13] were from PRIMM (Milan, Italy). Bacterial cultures, plasmid purifications, and transformations were performed as described by Sambrook et al. [14].

Construction of E. coli PfPDO and mutants PfPDO(C35S), PfPDO(C146S) and the double mutant (C35S/C146S)

Isolation of chromosomal DNA from P. furiosus was performed as described by Barker [15]. From the PfPDO amino-acid sequence from residues 1–7, the following oligonucleotides were designed and used as primers in the PCR gene amplification procedure, using the chromosomal DNA (200 ng) as template: forward primer, 5′-GGAATTcatatgGGATTGATTAGTGACGCTG-3′, contained a 5′-NdeI site (indicated in lowercase); reverse primer housed the PfPDO stop codon 3′ of a unique BamHI (indicated in lowercase) 5′-GGAATTcatatgGGATTAGTGACGCTG-3′. The amplification was performed as described by Saiki [16] for 35 cycles at 45 °C annealing temperature, on a Perkin–Elmer Cetus Cycler Temp using Pfx polymerase (Stratagene). The amplified DNA fragment (PfPDO), opportunely digested, was inserted into the pET22(b+) plasmid. The recombinant clone, designated pET-PfPDO wild-type, represented the expression vector.

The mutations Cys35Ser (C35S) and Cys146Ser (C146S) were introduced into the PfPDO DNA by the method of Kunkel [17]. The amplified genes, opportunely digested, were ligated to the cloning pET22(b+) plasmid. Insertion of the correct mutations was confirmed by DNA sequencing using Sanger's dideoxy method, with a Sequenase Sequencing Kit from Amersham [18].

Expression and purification of recombinant PfPDO mutants

Competent E. coli BL21(DE3) cells were transformed with pET-PfPDO wild-type, C35S, C146S, and C35S/C146S, and grown at 37 °C to different densities in 500 mL terrific-broth medium; isopropyl thio-β-d-galactoside was added to 1 mm final concentration, varying the induction time from 2 to 24 h. E. coli BL21DE3 cells transformed with pET22(b+) represented a negative control. Optimized overexpression of all the proteins was obtained by exposing the cells to 1 mm isopropyl thio-β-d-galactoside at a cell density of A600 = 2.5 for 18 h. Cell pellets from 500 mL cultures were resuspended in 5 mL 10 mm Tris/HCl, pH 8.4, and crude extracts were prepared by disrupting the cells with 20 min pulses at 20 Hz (Sonicator Ultrasonic liquid processor; Heat System Ultrasonics Inc., Farmingdale, NY, USA) and ultracentrifugation at 160 000 g for 30 min. Recombinant wild-type protein and its mutants were purified in a similar way. The crude extracts were subjected to heat treatment at 80 °C and then centrifuged at 5000 g at 4 °C for 15 min, removing almost 70% of the mesophilic host proteins. The crude extracts were applied to a 2.6 cm × 60 cm column (HiLoad Superdex 75; Pharmacia) connected to an FPLC system (Pharmacia) and eluted with 10 mm Tris/HCl (pH 8.4)/0.2 m NaCl at a flow rate of 2 mL·min−1. The active fractions were pooled, concentrated, and extensively dialysed against 10 mm Tris/HCl, pH 8.4. They were then loaded on an anion-exchange Mono Q column in 10 mm Tris/HCl, pH 8.4, connected to an FPLC system (Pharmacia), and eluted with a linear gradient (0/0.3 m NaCl) in 30 min at a flow rate of 0.5 mL·min−1. A single peak was observed on RP-HPLC and a single protein band on SDS/PAGE.

Analytical methods for protein characterization

Protein concentration was determined using BSA as the standard [19]. The molar absorption coefficient, obtained by the method used by the Schepertz laboratory (http://paris.chem.yale.edu), was 19 724 m−1·cm−1.

Protein homogeneity was assessed by SDS/PAGE [12.5% (w/v) gels] using the silver staining procedure of Rabilloud et al. [20]. In addition, proteins were analysed by nondenaturing electrophoresis [12.5% (w/v) polyacrylamide slab gel].

The molecular mass of the proteins was estimated using electrospray mass spectra recorded on a Bio-Q triple quadrupole instrument (Micromass). Samples were dissolved in 1% (v/v) acetic acid/50% (v/v) acetonitrile and injected into the ion source at a flow rate of 10 mL·min−1 using a Phoenix syringe pump. Spectra were collected and elaborated using masslynx software provided by the manufacturer. Calibration of the mass spectrometer was performed with horse heart myoglobin (16 951.5 Da).

UV-CD spectra in 10 mm sodium phosphate, pH 7.0, using a 1-mm path-length cell at 185–260 nm at 25 °C, were recorded on a Jasco J-710 spectropolarimeter equipped with a Peltier thermostatic cell holder (Jasco, model PTC-343) for all the proteins.

Counting integral numbers of residues by chemical modification

The procedure of Hollecker & Creighton [21] was used to detect the different exposure of the cysteine residues. All the proteins (PfPDO and mutants at a final concentration of 200 mm) were incubated in a final volume of 1 mL for 30 min at 37 °C in 10 mm Tris/HCl (pH 8.0)/10 mm EDTA (pH 7.0) in native, reduced (10 mm dithiothreitol), and reduced and denatured (10 mm dithiothreitol and 8 m urea) conditions. Successively in a final volume of 10 μL, five different solutions containing 0.25 m iodoacetate (in 0.25 m Tris/HCl, pH 8.0, and 0.25 m KOH) and 0.25 m iodoacetamide (in 0.25 m Tris/HCl, pH 8.0) were prepared in the following ratios: 0 : 1 (250 mm); (250 mm) 1 : 0; 1 : 1 (each 125 mm); (187.5 mm) 3 : 1 (62.5 mm); (225 mm) 9 : 1 (25 mm). At the end of the incubation, 40 μL of the mixture was added to each of the five solutions; these were then left to react on ice for 5 min. The reaction mixtures were analysed by nondenaturing electrophoresis [12.5% (w/v) polyacrylamide slab gel]. The ‘ladder’ or control is represented by a mixture of 10 mL taken from each of the five reaction mixtures. The method consists of adding various iodoacetamide and iodoacetate ratios to portions of the protein to generate a complete spectrum of protein molecules with 0, 1, or 4 acidic carboxymethyl groups, where 4 is the integral number of cysteine residues. Protein in which all thiol groups were blocked with iodoacetate, if well exposed, migrated more slowly than that blocked with iodoacetamide, because of the acidic carboxymethyl groups.

Cross-linking with dimethyl suberimidate (DMS)

Following the procedure of Davies & Stark [22], 10 μg PfPDO was incubated for 2 h at room temperature with different quantities of DMS (1 : 1, 1 : 2.5, 1 : 5, 1 : 10) to determine the best protein to DMS ratio. Molecular mass and yield were checked by SDS/PAGE [12.5% (w/v) polyacrylamide gel].

Assay of enzyme activities

Insulin reductase activity.  Reductase activity was assayed by Holmgren's turbidimetric method [23] with a few modifications. The catalytic reduction of insulin disulfide bonds was measured at 30 °C. Protein was added in 1 mL 100 mm sodium phosphate buffer, pH 7.0, containing 2 mm EDTA and 1 mg bovine insulin. A control cuvette contained only buffer and insulin. The reaction was started by the addition of 2 mm dithiothreitol to both cuvettes. Increasing turbidity from precipitation of the insulin B chain was recorded at 650 nm. The stock solution of insulin (10 mg·mL−1) was prepared according to the Holmgren protocol.

Oxidation activity.  The disulfide bond-forming activity of the proteins was monitored using the synthetic decapeptide NRCSQGSCWN containing two cysteine residues at position 3 and 8 designed by Ruddock et al. [13]. The peptide contains a fluorescent group (tryptophan) on one side of one cysteine residue and a protonated group (arginine) on the other side of the second cysteine residue, and the two cysteine residues are separated by a flexible linker region. The linker is long enough to permit the formation of an unstrained disulfide bond, and the peptide is small and water soluble. Oxidation of this dithiol peptide to the disulfide state is accompanied by a change in tryptophan fluorescence emission intensity. In fact, on oxidation, the fluorescent group and the protonated group are brought close together, and quenching on the fluorophore occurs where arginine is the charged quencher. Fluorescence quenching was used as the basis for monitoring the disulfide bond-forming activity of PfPDO.

Spectrofluorimetric analysis.  The assay was performed in McIlvaine buffer (0.2 m disodium hydrogen phosphate/0.1 m citric acid, pH 7.0) with 2 mm GSH, 0.5 mm GSSG and 5 μmPfPDO. The reaction mixture was placed in a fluorescence cuvette with a final assay volume of 1 mL. After mixing, the cuvette was placed in a thermostatically controlled Perkin–Elmer LS50B spectrofluorimeter for 1 min to allow thermal equilibration of the solution to 50 °C. Next, 5 μm substrate peptide was added, mixed, and the change in fluorescence intensity (excitation 295 nm, emission 350 nm, slits 10/10 nm) was monitored over an appropriate time (15 min). As a control, the same experiment was carried out in the absence of any protein; no decrease in fluorescence intensity was observed [13].

HPLC analysis.  Alternatively the oxidation activity was measured by HPLC analysis (Varian). The reduced and oxidized forms of the peptide have different retention times and are eluted separately on reverse-phase chromatography [L. Birolo and A. Tosco (1999) personal communication]. The peptide was eluted in a single peak and stored at −20 °C in the elution buffer (30% acetonitrile in 0.1% trifluoroacetic acid; v/v/v) at a concentration of 1.05 mm. The peptide concentration was determined spectrophotometically using an absorption coefficient of 5600 m−1·cm−1 at 278 nm. The oxidized state of the peptide was generated by incubating the peptide at a concentration of 50 μm in 0.2 m Tris/HCl, pH 8.4, at 20 °C for 15 h. The reduced state was generated by incubating the peptide in McIlvaine buffer (0.2 m disodium hydrogen phosphate/0.1 m citric acid, pH 7.0) at a final concentration of 50 μm and 1 mm dithiothreitol in a final volume of 50 μL.

The assay mixture contained 5 μm reduced peptide, 100 mm GSH (stock solution 60.1 mg·mL−1), 25 mm GSSG (stock solution 30.7 mg·mL−1) and the protein PfPDO (final concentration 10, 50, 100, 150 or 200 μm). The mixture was incubated at different temperatures (50 °C, 60 °C or 70 °C) for different times in the presence of different concentrations of the protein. After incubation, the mixture was loaded on the HPLC reversed-phase Vydac C18 column equilibrated in buffer A [0.1% (v/v) trifluoroacetic acid in water]. Chromatography was carried out with a linear gradient 0–100% buffer B (95% acetonitrile, 0.07% trifluoroacetic acid; v/v/v) in buffer A at a flow rate of 1 mL·min−1 for 35 min.

Re-activation of scrambled RNase.  The isomerase activity was assayed by Lambert's method. Re-activation of scrambled RNase was monitored after incubation of PfPDO in 50 mm sodium phosphate, pH 7.5, in a total volume of 0.9 mL, with 10 μL dithiothreitol (1 mm stock solution, final concentration 10 μm) for 2 min at 30 °C [24]. A 0.1 mL portion of ‘scrambled’ RNase (Sigma; 0.5 mg·mL−1 in 10 mm acetic acid, final concentration 4 μm) was added, and at different times after this addition 10 μL samples were withdrawn and assayed for RNase activity. Each sample was added to an assay mixture of 1 mL 0.5 mg·mL−1 RNA in 50 mm Tris/HCl, pH 7.5. RNase activity on yeast RNA was assayed by the method outlined by Kunitz [25] with some modifications, and under conditions in which the decrease in A300 was linear for at least 3–4 min. Yeast RNA was dissolved in water, and the pH was kept neutral by performing the assay in 50 mm Tris/HCl, pH 7.5. The positive control was re-activation of scrambled RNase catalysed by PDI (bovine liver; Sigma). Nonenzymatic reactivation of scrambled RNase was corrected for by using the same mixture without the addition of any of the proteins.

Detection of ATP binding by CD

CD measurements were performed in a Jasco J-720 spectropolarimeter in 20 mm Tris/HCl (pH 7.5)/5 mm MgCl2 at 25 °C. Each sample was scanned five times, noise reduction was applied, and baseline buffer spectra were subtracted from sample spectra before molar ellipticities were calculated. To obtain spectra in the near-UV region (250–320 nm), the cell path length was 1 cm and the protein concentration 1 mg·mL−1. The CD spectra were evaluated at 260 nm.

Cross-linking of PfPDO with 8-Azido-[32P]ATP[αP]

To analyze the ability of PfPDO to cross-link to 8-azido-ATP, 3 mg protein was incubated in the presence of 2 mCi 8-azido[32P]ATP[αP] for 30 min in 50 mm Tris/HCl, pH 8.0 or 10 mm Gly/NaOH, pH 10.0, containing 2 mm EDTA, 1 mm dithiothreitol and 5 mm MgCl2 at 60° or 70 °C. To induce cross-linking, samples were exposed for 10 min to UV irradiation and then resolved by SDS/PAGE in 12% polyacrilamide gel and visualized by radioautography on a Fuji medical X-ray film. The same procedure was used for incubation of PfPDO at pH 10.0 at 70 °C in the presence of an increasing concentration of unlabeled ATP (50 μm and 1 mm) [26].

Fluorescence measurements

Samples of PfPDO (100 µg) were incubated for 10 min at 70 °C, 80 °C, or 90 °C in the presence of MgATP, and then loaded on a Superdex 75 HiLoad column (Amersham Pharmacia Biotech; 1 × 30 cm; eluent 10 mm Tris/HCl, pH 7.5, 0.2 m NaCl; flow rate 0.3 mL·min−1) to remove the nucleotide excess. The protein samples recovered from the columns and a sample of native PfPDO were analyzed for fluorescence at 3 µm final protein concentration (excitation wavelength 280 nm; emission recorded between 310 and 410 nm) using a Perkin–Elmer LS50B spectrofluorimeter at 25 °C [27].

Assay of ATPase activity

A colorimetric assay was routinely used to measure ATPase activity following the method of Lanzetta et al. [28]. To 100 μL of sample (water and 10 mg protein) was added 800 μL of green malachite/ammonium molybdate in 1 m HCl, followed by mixing. After 1 min, 100 μL 34% citrate was added and mixed. This solution was read immediately at 660 nm.

In an alternative assay, the ATPase activity of PfPDO was assayed in mixtures containing 2 mm ATP, 15 µCi [32P]ATP[αP], 5 mm MgCl2 and 10 µg pure protein in 50 mm Tris/HCl, pH 7.5 (150 µL final volume). After a 5 min incubation at 70 °C, a 25 µL aliquot was withdrawn and added to 0.5 mL of a suspension containing 50 mm HCl, 5 mm H3PO4 and 7% activated charcoal. The mixture was then centrifuged at 4000 g for 20 min. The radioactivity of the supernatant was determined in a 100 µL aliquot using a liquid-scintillation counter (Beckman). In rate calculations, spontaneous ATP hydrolysis in the absence of PfPDO was corrected for [29].

Results

Production of wild-type and mutant PfPDO

To determine the redox state and the accessibility of the cysteine residues of the PfPDO redox sites, electrophoretic analysis was performed, as described by Hollecker et al. [21], on the protein treated under different conditions (native, reduced, and reduced and denatured) (Table 1). By comparing the results obtained from the different gels, it was possible to confirm the crystallographic data that the most reactive cysteine was Cys146, as this was observed at the lowest ratio of iodoacetate to iodoacetamide. This was followed by Cys149, Cys35, and Cys38, which was the last to react and the least accessible residue [21].

Table 1. Exposed cysteine residues by the Hollecker method [21]. The proteins in native, reduced (10 mm dithiothreitol), and reduced and denatured (10 mm dithiothreitol and 8 m urea) conditions were treated with different amounts of iodoacetate/iodoacetamide [1 : 1 (each 250 mm), 1 : 3, 1 : 9 ratios of neutral to acidic reagents] and separated by SDS/PAGE. The appearance of a band in the different conditions used (native, reduced, reduced and denatured) on the different proteins (PfPDO and mutants) is evidence of the exposure of that cysteine residue. nd, Not determined.
 NativeReducedReduced/denatured
Iodoacetate/iodoacetamide1 : 13 : 19 : 11 : 13 : 19 : 11 : 13 : 19 : 1
PfPDO wild-typeC146C146C146/149C146C146/149C146/149/35
C35S mutantC146C146C146/149C146C146/149C146/149/38
C146S mutantC149C149C149C149/35C149/35
C35S/C146S mutantndndndC149C149C149/38C149C149C149/38

To investigate the role of the putative redox sites of PfPDO, three mutants were constructed (C35S, C146S, and C35S/C146S) by mutagenizing the most exposed cysteines of each of the redox sites: specifically, Cys35 at the N-terminal site and Cys146 at the C-terminal site were replaced by serine [30].

PfPDO and mutants were expressed in E. coli BL21(DE3). Overexpression of all the proteins was obtained by exposing the cells to 1 mm isopropyl thio-β-d-galactoside at a cell density of A = 2.5. To optimize the production of the recombinant proteins, transformed cells were exposed to the inducer for 2–24 h; maximum expression was obtained after 18 h of induction.

The crude extract of E. coli was subjected to one thermal precipitation step at 80 °C for 20 min to remove almost 70% of the mesophilic host proteins. During the purification procedure, the proteins were assayed after reduction of protein disulfides on insulin as substrate; when the interchain disulfide bridges are reduced between chains A and B of the insulin, the turbidity of the solution increases because of precipitation of the free B chain [23]. After gel-filtration chromatography and anion-exchange chromatography, a single peak was observed on RP-HPLC, and a single band on SDS/PAGE. The protein yield from 1 L of culture was ≈ 40 mg for all the recombinant proteins.

The molecular mass of the proteins was analysed by electrospray mass spectroscopy. The measured mass of PfPDO was 25 648 ± 0.5 Da. The measured mass of C35S and C146S was 25 628 ± 0.5 Da, and that of C35S/C146S was 25 613 ± 0.4 Da. Thus, the difference in mass was in perfect agreement with the mutations introduced.

To see if the mutations introduced had an effect on the structure of the protein, far-UV CD spectra were recorded for all the proteins. The spectra were very similar, showing that all the proteins are completely folded and indicating that the mutations did not result in any obvious change in overall structure.

Characterization of the activities of wild-type and mutant PfPDO

PfPDO reduces insulin disulfide in the presence of dithiothreitol at 30 °C. The analysis was performed in the presence of increasing concentrations of the pure proteins, as well as in their absence (the spontaneous precipitation reaction), because dithiothreitol is the reducing agent that recycles the oxidized protein (Fig. 1). The activity was assayed at 1.2 μm for all the proteins. Both the wild-type PfPDO and the mutant C35S were active in the insulin reductase assay [23], whereas the activity of the mutant C146S and the double mutant was similar to the control. This shows that the active site in the C-terminal half (CPYC) is responsible for the reductase activity.

Figure 1.

Assay of reductase activity by measuring the reduction of bovine insulin disulfides. The dithiothreitol-dependent reduction of bovine insulin disulfides was carried out as described in Experimental procedures in the absence [control (–––)] or presence of 1.2 µmPfPDO wild-type (····), PfPDO (C35S) (- - - -), or PfPDO (C146S) and PfPDO (C35S)/(C146S) (-·-·).

In the presence of 5 μmPfPDO, oxidation of the dithiol peptide designed by Ruddock was observed at neutral pH by the spectrofluorimetric assay (Fig. 2A). Separation of the oxidized and reduced forms of the peptide by HPLC allowed quantification of the oxidative activity as a ratio between the areas of oxidized/reduced peptide. The assays were performed at a concentration of 100 μm protein, at different times and different temperatures [50 °C, 60 °C, and 70 °C (data not shown)]. The best conditions were 50 °C for 3 h. A linear relation between activity and concentration was detected for all the proteins (Fig. 2B). Wild-type PfPDO and C35S were able to oxidize the peptide with maximum activity at a concentration of 150 and 200 μm, respectively. C146S had residual oxidative activity, but the double mutant was completely inactive, demonstrating the predominant role of the redox site at the C-terminus in the oxidative activity.

Figure 2.

Assay of oxidative activity by measuring the formation of the disulfide bridge in the peptide NRCSQGSCWN.

(A) Spectrophotometric method. The disulfide bond-forming activity of PfPDO was monitored using the synthetic decapeptide NRCSQGSCWN. Oxidation of this dithiol peptide to the disulfide state is accompanied by a change in tryptophan fluorescence emission intensity. (a) Control, the same assay performed without the protein; (b) PfPDO; (c) PfPDO (C35S); (d) PfPDO (C146S). (B) HPLC. The disulfide bond-forming activity of PfPDO is monitored using the synthetic decapeptide NRCSQGSCWN and the oxidation of this dithiol peptide to the disulfide state is accompanied by a change in time of retention on a Vydac C18. Oxidative activity is expressed as a ratio between the peak of oxidized and reduced peptide. The assay was performed at 50 °C, with an incubation time of 210 min, at increasing concentration of PfPDO wild-type (◆), PfPDO (C35S) (▪); PfPDO (C146S) (▴); PfPDO (C35S)/(C146S) and control (•).

The action of PfPDO in catalysing interchange of intramolecular disulfides in scrambled RNase results in restoration of the native disulfide pairing and the concomitant return of RNase activity. Thus, the isomerase activity of PfPDO was assayed by a time-course incubation during which aliquots were removed and RNase activity with RNA was measured. Re-activation of scrambled RNase was performed with all the proteins. Only the wild-type protein was able to refold the scrambled RNase (Fig. 3), indicating that isomerase activity requires the participation of both N-terminal and C-terminal active sites. Refolding of the scrambled RNase in the presence of PDI was used as a positive control, and the absence of the recovery of the RNase activity in the presence of the thioredoxin from Alicyclobacillus acidocaldarius was used as a negative control. The refolding of the scrambled RNase in the presence of PfPDO seems to be less efficient when using PDI. However, the temperature of the assay, which is limited by the stability of the protein substrate RNase, is very far from the optimal growth temperature of the hyperthermopilic micro-organism.

Figure 3.

Assay of isomerase activity of PfPDO by measuring re-activation of scrambled RNase. The recovery of RNase activity as a function of time is presented after preincubation with PDI (▴); PfPDO wild-type (◆); PfPDO (C35S) and PfPDO (C146S) and PfPDO (C35S)/(C146S) (▪); control (•). RNase activity with RNA was measured.

Characterization of wild-type PfPDO

A detailed study of PfPDO structure highlighted certain putative ATP-binding sites (the presence of P-loops, a common motif in ATP-binding proteins), the primary structure of which consists of a glycine-rich sequence followed by a conserved lysine and a serine or a theonine [31]. In particular, PfPDO has the sequences, GKDFG(88–94), GLPAG(97–101), GKGKILG(167–173), which resemble, with some deviations, the glycine-rich motif, GXXGXG, of the ATPase domains of the eukaryotic chaperone hsp90 [32], the type II DNA topoisomerases, and MutL DNA mismatch-repair proteins [33]. The hypothetical nucleotide-binding sites are presumably located in loops between β3 and β4, α4 and β4, and α6 and β6. To study the role played by the putative binding of ATP in the conformation of PfPDO, a spectrofluorimetric analysis was performed. The presence of Trp184 enabled us to perform intrinsic fluorescence experiments. The tryptophan emission spectrum of native PfPDO displayed a maximum around λ = 345 (data not shown). A PfPDO sample incubated in the presence of hydrolysable ATP (MgATP) gave a similar spectrum to that of the native protein. Far-UV CD spectra in the presence and absence of ATP (data not shown) gave the same results as the spectrofluorimetric analysis, i.e. no change in the conformation of the protein in the presence of the nucleotide. These experiments indicate that the binding and/or hydrolysis of ATP do not have any effect on the conformation of PfPDO, possibly because of the localization of the amino-acid residues involved in ATP binding in exposed regions. Near-UV CD spectroscopy was performed, which provides information on the environment of aromatic residues in folded proteins. The aromatic CD spectra of PfPDO in the absence and presence of ATP (up to 324 mm) are shown in Fig. 4. The ellipticity of the protein was positive between 255 and 300 nm. A signal around 279 nm can be assigned to tyrosine residues, and the major intensity at 268 nm and 261.5 nm can be attributed to the numerous phenylalanine residues (12 of them). After ATP was added, the signal attributed to tryptophan and tyrosine residues does not seem to have been affected, whereas the signal attributed to phenylalanine residues changed considerably.

Figure 4.

Measurement of Kd for ATP by CD. Near-UV CD spectra recorded in 20 mm Tris/HCl (pH 7.5)/5 mm MgCl2 and in the presence of increasing concentrations of ATP (0–324 mm). The inset shows normalized CD variation at 260 nm vs. increasing [ATP] concentration. CD values at 260 nm were normalized and elaborated using the programme Microsoft Excel 2000. The curve for the determination of the Kd for ATP was obtained using the program kaleida graph 3.0.

Interestingly, close to the P-loop domain, there is a phenylalanine residue at position 91. In addition, our spectra indicate that other aromatic residues are in close proximity to the ATP-binding domain. The CD data indicate ATP binding with co-operativity and a Kd of 230 μm.

The ATP binding to PfPDO was confirmed by cross-linking to 8-azido-ATP after UV irradiation (Fig 5A,B). The data show ATP binding for PfPDO. Alcohol dehydrogenase (horse liver; Sigma) was used as a negative control because it is known not to bind ATP, even though it contains a putative nucleotide-binding site. The alcohol dehydrogenase did not show any affinity for the ATP analog, suggesting that the binding to PfPDO was specific under the conditions used. It has been reported that some non-ATP-binding proteins (for example, BSA) bind8-azido-ATP in a nonspecific way. However, in these cases, the bound analog could not be displaced by the unlabeled nucleotide [34]. In this work, photoaffinity labeling of PfPDO with 8-azido-[32P]ATP[αP] was decreased by the presence of unlabeled ATP, indicating that ATP and the analog 8-azido-ATP recognize the same binding site.

Figure 5.

ATP-binding capacity of PfPDO. Cross-linking of PfPDO with 8-azido-[32P]ATP[αP]: 3 μg PfPDO was incubated with 2 mCi 8-azido-[32P]ATP[αP] for 30 min at pH 8.0 and pH 10.0 at 60° and 70 °C. To induce cross-linking, samples were exposed for 10 min to UV irradiation and then resolved by SDS/PAGE in 12% polyacrylamide gel and visualized by radioautography. (A) Lanes 1 and 2, pH 8.0 at 60 °C and 70 °C; lanes 3 and 4, pH 10.0 at 60 °C and 70 °C. (B) The same procedure was used by incubating PfPDO at pH 10.0 at 70 °C in the presence of increasing concentrations of unlabeled ATP. Lane 1, 0 mm unlabeled ATP; lane 2, 50 mm unlabeled ATP; lane 3, 1 mm unlabeled ATP.

The ATPase activity of PfPDO was demonstrated. The hydrolysis of ATP was linear for up to 30 min at every temperature examined with the colorimetric and radioactive assays used (see Experimental Procedures). The hydrolysis of ATP by PfPDO required the presence of bivalent metal ions, Mg2+ giving the highest rate (Fig. 6A) compared with the activity observed in the absence of ions. When assayed in the pH range 4.0–10.0, PfPDO catalyzed hydrolysis of ATP with a maximum around basic values (Fig. 6B). Assays performed in the temperature range 30− 90 °C showed that, at 90 °C, PfPDO is still fully able to hydrolyse ATP (Fig. 6C). The rate of spontaneous ATP hydrolysis was followed in the same range of temperature and pH. Freshly purified PfPDO hydrolysed ATP with a Vmax of 127.5 nmol Pi released·min−1·mg−1 (Mg2+, pH 10.0, 90 °C).

Figure 6.

ATPase activity of PfPDO. The assays were performed under standard conditions (see Experimental Procedures) except for the ions at 5 mm. (A) The activity assayed under standard conditions (Mg2+ at 90 °C, pH 7.5) was 67.3 nmol Pi released·min−1·mg−1, which was taken as 100%. Activity was assayed at different pH values [50 mm sodium acetate for pH 4.0–5.5 (▴); 50 mm sodium phosphate for pH 6.0–7.0 (▪); 50 mm Tris/HCl for pH 7.5–8.4 (◊); 50 mm glycine/NaOH for pH 9.0–10.5 (•)] (B) and temperatures (c). The activity assayed under standard conditions was 127.5 nmol Pi released·min−1·mg−1 (Mg2+, pH 10.0, 90 °C), which was taken as 100%. Data are means from at least three independent experiments.

The ability of PfPDO to bind and hydrolyse ATP is another property that links this protein with the multifunctional PDI, as this feature has been observed in the eukaryotic protein [35].

In addition, as PfPDO exists as a dimer in the crystal form and PDI is a dimer in its 3D structure, we analysed the dimerization of PfPDO by gel filtration and in the presence of the cross-linking agent DMS. In all the conditions tested, the presence of the dimer was never observed. It was observed only in the presence of the cross-linking reagent DMS. In particular, a ratio of PfPDO to DMS of 1 : 2.5 proved to be optimal (Fig. 7).

Figure 7.

Cross-linking of PfPDO with DMS. After 2 h of incubation at room temperature in the presence of the cross-linking agent DMS, the samples were loaded on an SDS/12.5% polyacrylamide gel. Lane 1, PfPDPfPD/DMS in a ratio 1 : 2.5; lane 2, control, PfPDO with no DMS; lane 3, markers of molecular mass.

Discussion

Insufficient information is available on protein disulfide oxidoreductases from archaea to define their physiological function(s) with any certainty. Disulfide bonds are now known to occur in many thermophilic and intracellular archaeal proteins, and this observation highlights the importance of the glutaredoxin/thioredoxin system in these micro-organisms.

Hyperthermophiles are generally capable of growing under extreme conditions such as low pH, high pressure, and high salt concentration. Most of these organisms are anaerobes, have extraordinarily heat-stable proteins, and use ingenious strategies for stabilizing nucleic acids and other macromolecules in vivo[36].

Recently, from the resolution of the whole genome sequences of various hyperthermophilic archaea, it is clear that these hyperthermophiles have proteins endowed with thioredoxin/glutaredoxin motifs, suggesting the ubiquity of this system in nature.

The protein from P. furiosus described here may provide an important contribution to our understanding of the function of these proteins in hyperthermophilic archaea and bacteria. In fact, PfPDO is able to catalyse the oxidation of dithiols, as well as the reduction and rearrangement of disulfides. In the presence of glutathione, up to 70 °C, PfPDO catalyses the formation of a disulfide bond between the two cysteines of the peptide, an activity similar to that observed for DsbA at 25 °C [13]. At 30 °C, PfPDO is able to catalyse the reduction of insulin disulfides in the presence of dithiothreitol. Disulfide rearrangement was also observed at a similar temperature using RNase with scrambled disulfides as substrate.

Using the two single mutants (C35S and C146S) and the double mutant (C35S/C146S), we have demonstrated that the C-terminal site (CPYC), which is common to all the glutaredoxins, determines the reductive activity. This result is in agreement with crystallographic data, which suggest a reductive nature for the C-unit. The lower capacity of the N-unit to reduce disulfide bridges may be due to intrinsic factors, such as a higher redox potential and major conformational tension of the disulfide, but it may also depend on external factors such as steric impediments caused by a closed conformation of the active site in the N-unit. As regards the oxidative activity, the two units also display differences in their functional properties, with the site at the C-termus always predominant, the mutant with a nonmutagenized site at the N-terminus showing very low activity at 50 °C. Higher temperatures, closer to the physiological temperature at which the micro-organism P. furiosus lives, may be necessary to obtain more kinetic energy and allow an open conformation at the site. Alternatively, a different substrate may be required because of the polar nature of the amino acids close to the active site. On the other hand, both sites are necessary for the disulfide isomerase activity. In fact, only wild-type PfPDO was able to refold scrambled RNase. This is in agreement with a functional model of PDI in which the domains function synergistically [37,38]. The emerging model of PDI comprises four structural domains, a, b, b′ and a′, plus a linker region between b′ and a′ and a C-terminal acidic extension. In this model of PDI function, individual domains with specialized roles contribute to different activities to enable the catalysis of complex isomerizations in substantially folded protein substrates. Mutations at the first cysteine of the active site in either the N-terminal or C-terminal thioredoxin domain inhibits the capacity of PDI to catalyse thiol–disulfide exchange reactions in vitro, reducing enzymatic activity to negligible levels. In fact, the redox/isomerase activities of PDI, as in thioredoxin, are due to the reactivity of the N-terminal Cys residue in two thioredoxin-like boxes (Cys-Gly-His-Cys) within the a and a′ domains of the protein [39]. Although the two domains do not possess equivalent catalytic activities or substrate-binding affinities, they can function independently from each other.

PfPDO resembles eukaryotic PDI, as it has two thioredoxin-like motifs. In PDI, the thioredoxin-like regions are separated from each other in the primary structure, whereas in PfPDO they are connected directly. In this work, only the first cysteine of each redox site was mutated to investigate the effect on the function of the protein, demonstrating that the active site at the C-terminus is basic for oxidative and reductive activities and that the two units do not seem to be functionally independent, considering that only the wild-type enzyme is able to refold scrambled RNase. Unlike PDI, which is a homodimer of two 57 kDa subunits, PfPDO seems to be a monomer, dimerization only occurring in the presence of the cross-linking agent DMS.

The ability of PfPDO to bind and hydrolyse ATP supports its relationship to PDI [40]. In fact, an ATP-binding site and ATPase activity related to its chaperone role have been reported in PDI [41]. Whereas PDI binds ATP with a Kd of 9.66 μm, PfPDO binds ATP with a Kd of ≈ 230 μm. PfPDO is a hyperthermostable protein, and the studies of its functional and catalytic properties are limited by the temperature at which its activities are studied. Such temperatures are usually far below the physiological temperature (70–103 °C) at which P. furiosus lives. The ATPase activity does not seem to be linked to the isomerase or redox activities, as in the presence of ATP no differences in the activities are observed. This is in full agreement with a report that the site of phosphorylation, and thus probably the ATPase active site, lies somewhere within the central domain of the PDI [42], and that this site is far away from the redox active sites in the sequence. Furthermore, the measurements of the rates of PDI-catalysed refolding of scrambled RNase A, in the absence or presence of ATP, show that ATP has little or no effect on this activity.

Interestingly, comparison of the genomes of archaea and bacteria showed the existence of a group of redox proteins with a similar molecular mass to PfPDO. Clearly, all these proteins also contain two active sites, although they were often initially assigned as hypothetical thioredoxins and glutaredoxins [43–52]. The presence of the redox site, CQYC, at the N-terminus of protein disulfide oxidoreductase in P. furiosus, P. abyssi, and P. horikoshii, and also in the more distant S. solfataricus, further confirm the importance of this site for protein function (Fig. 8). It is worth noting that amino-acid residues that are probably involved in putative ATP binding, such as Gly88, Gly97, Pro99, Gly167 and Gly170, are well conserved, indicating their importance. The genomes of the hyperthemophilic bacteria Aquifex aeolicus, Thermotoga maritima and Thermoanaerobacter tengcongensis do not encode a protein related to bacterial DsbA and no DsbA-like protein in Archaea were found, suggesting that PfPDO-like proteins represent a new family characteristic of extremophiles (like DsbA in bacteria and PDI in eukarya). It should be noted that we found PfPDO-like proteins only in thermophilic bacteria, i.e. Aquifex aeolicus, Thermotoga maritima and Thermoanaerobacter tengcongensis. A preferential horizontal gene transfer has been noticed between archaea and hyperthermophilic bacteria, such as Aquifex and Thermotoga; in fact their proteins show greater similarity to archaeal than to bacterial homologs [53]. The reality of horizontal gene flow from archaea to thermophilic bacteria becomes even more tangible on examination of the proteins encoded in the genome of Thermoanaerobacter tengcongensis which contains more ‘archaeal’ genes than appear in other bacteria.

Figure 8.

Comparison of the amino-acid sequences of different protein disulfide oxidoreductases. The sequences were from the following sources: Pf, P. furiosus; Ph, P. horikoshii; Pa, P. abissi; Ss, S. solfataricus; St, S. tokodaii; Ap, Aeropyrum pernix; Ta, Thermoplasma acidophilum; Tv, Thermoplasma volcanium; Fa, Ferroplasma acidarmanus; Tm, Thermotoga maritima; Aa, Aquifex aeolicus; Tt, Thermoanaerobacter tengcongensis. The residues identical with the sequence of PfPDO in at least 90% of the sequences are indicated in bold. The underlined residues indicate the active sites.

The exclusive presence of PfPDO-like proteins in extremophiles may suggest that they have a special role in the adaptation to extreme conditions. The P. horikoshii genome also contains a glutaredoxin-homolog gene (88% identity with the glutaredoxin from P. furiosus) [54]. This protein is the first glutaredoxin-homolog protein that directly mediates electron transfer from a thioredoxin reductase-like flavoprotein to protein disulfide in archaea. The redox-active sequence motifs CPYC and CQYC suggest that P. horikoshii redox protein (PhRP) belongs to the same family as PfPDO. PhRP has insulin-reducing activity. Site-directed mutagenesis studies revealed that the active site of the redox protein corresponds to a CPYC sequence located in the middle of the sequence, as in PfPDO. As regards PhRP activities, the disulfide formation and its rearrangement were not detected when reduced or scrambled RNases were used as substrates at 25 °C. However, the possibility that CQYC may play some role and that PhRP has PDI-like activity in vivo at the optimum growth temperature of P. horikoshii cannot be excluded.

The various functions of PfPDO make it an interesting model system for clarifying the long-standing debate on the content of cysteine residues and disulfide in thermophilic proteins. Disulfide bonds have only rarely been found in intracellular proteins. The pattern is consistent with a chemically reducing environment inside the cells and with a PDI role in the endoplasmic reticulum. However, recent experiments and new calculations based on genomic data of archaea provide striking contradictions to this pattern. Recent results indicate that the intracellular proteins of certain hyperthermophilic archaea, especially some crenarchaea such as Pyrobaculum aerophilum and Aeropyrum pernix, are rich in disulfide bonds [55]. This finding points to the role of disulfide bonds in stabilizing many thermostable proteins and suggests new chemical environments inside these microbes.

Acknowledgements

We thank Dr Raffaele Cannio and Dr Enrico Bucci for stimulating discussions. This work was supported by grants from MIUR (PRIN 2002).

Ancillary