Characterization of the Bradyrhizobium japonicum CycY protein, a membrane-anchored periplasmic thioredoxin that may play a role as a reductant in the biogenesis of c-type cytochromes.

A new member of membrane-anchored periplasmic thioredoxin-like proteins was identified in Bradyrhizobium japonicum It is the product of cycY, the last gene in a cluster of cytochrome c biogenesis genes. Mutational analysis revealed that cycY is essential for the biosynthesis of all c-type cytochromes in this bacterium. The CycY protein was shown to be exported to the periplasm by its N-terminal signal sequence-like domain. Results from Western blot analyses of membrane and soluble fractions indicated that the CycY protein remains bound to the membrane. A soluble version of the protein devoid of its N-terminal membrane anchor (CycY*) was expressed in Escherichia coli and purified to homogeneity from the periplasmic fraction. The protein showed redox reactivity and properties similar to other thioredoxins such as fluorescence quenching in the oxidized form. Its equilibrium constant with glutathione was determined to be 168 mM, from which a standard redox potential of −0.217 V was calculated, suggesting that CycY might act as a reductant in the otherwise oxidative environment of the periplasm. This is in agreement with our hypothesis that CycY is required, directly or indirectly, for the reduction of the heme-binding site cysteines in the CXXCH motif of c-type apocytochromes before heme attachment occurs.

A key step in the post-translational maturation of c-type cytochromes is the covalent ligation of the heme cofactor to the reduced apocytochrome (1). In bacteria, this reaction occurs in the periplasm in a rather oxidative environment. Thus, a reductive step in the cytochrome c maturation pathway may be postulated that would ensure that the cysteines in the hemebinding site motif (CXXCH) of the apocytochrome are in the reduced dithiol form before heme is attached.
Bacterial cytochrome c biosynthesis is known to depend on the products of at least nine different genes. These encode an ABC transporter, a putative cytochrome-c-heme lyase complex, and a thioredoxin-like protein (1)(2)(3)(4)(5)(6)(7). The gene for the latter has been identified in Rhodobacter capsulatus (helX) (8) and Rhizobium leguminosarum (cycY) (9) because of the presence of two conserved redox-active cysteines in the derived amino acid sequence that are typically present in thioredoxins. Mutations in these genes were found to affect the maturation of c-type cytochromes. In contrast to conventional cytoplasmic thioredoxins, the predicted HelX and CycY polypeptides carry an N-terminal extension resembling a typical signal sequence. A topological analysis with the R. capsulatus HelX protein, using alkaline phosphatase (PhoA) fusions, suggested its periplasmic location. It was postulated that HelX might function as a periplasmic reductant of the heme-binding site in apocytochromes (10).
In the soybean root nodule symbiont Bradyrhizobium japonicum, an open reading frame (orf194) with a high degree of similarity to helX and cycY has been identified downstream of the cytochrome c biogenesis gene cycX (11). A transposon (Tn5) insertion detaching the last eight codons of the gene led to a partial loss of cytochrome c biosynthesis, suggesting that the gene was essential for this process. Surprisingly, however, a kanamycin resistance cassette insertion (mutant 98) disrupting the gene in the presumptive fourth codon had no effect on cytochrome c biogenesis. It was concluded that the gene started downstream of the site of the cassette insertion in mutant 98. The next possible start codon, a TTG at position 3216 in the published nucleotide sequence, was therefore predicted to represent the start of the gene, which was thereupon called orf132 (11). However, the orf132 product lacked the hydrophobic Nterminal domain present in HelX and CycY. Unfortunately, a true null mutant of this gene was not obtained.
Intrigued by the high degree of similarity of B. japonicum orf132 to genes for other bacterial thioredoxin-like proteins, we further investigated its function. Here we present an extensive molecular analysis of this gene, now called cycY, which shows unequivocally that it codes for a membrane-anchored periplasmic thioredoxin and is essential for cytochrome c maturation. We overexpressed and purified a periplasmic soluble variant of the protein (CycY*) from Escherichia coli and determined its redox potential. This allowed us to make several important predictions with respect to the role of this peculiar thioredoxin in cytochrome c maturation.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-B. japonicum 110spc4 (12) is called the wild type throughout this work. E. coli DH5␣ (13) was used as a host for DNA cloning. E. coli S17-1 (14) was used as donor strain in bacterial conjugation experiments. E. coli BL21(DE3) (15) was used for overexpression of CycY* from the phage T7 promoter. B. japonicum strains were grown aerobically at 28°C in PSY medium (12) or anaerobically in YEM medium supplemented with 10 mM KNO 3 (16). Antibiotics were added at the following final concentrations: kanamycin, 100 g/ml; spectinomycin, 100 g/ml; tetracycline, 60 g/ml; and chloramphenicol, 10 g/ml. E. coli cells were grown at 30 or 37°C in LB medium to which antibiotics were added at the following concentra-tions: ampicillin, 100 -200 g/ml; tetracycline, 10 g/ml; and carbenicillin, 400 g/ml.
Recombinant DNA Work-Molecular cloning procedures were carried out according to Sambrook et al. (17). DNA sequence analysis was performed using the chain termination method and the equipment for automated DNA sequencing (Sequencer Model 370A and fluorescent dye terminators, Applied Biosystems, Inc., Foster City, CA).
Plasmid and Mutant Constructions-The lacZ gene of pNM480X (18) was fused in frame to the sixth codon of cycY at a SalI site, resulting in plasmid pRJ2801. The lacZ-containing plasmid pNM482 (19) was modified first by inserting an XhoI linker (5Ј-CCGCTCGAGCGG-3Ј; Promega, Wallisellen, Switzerland) into the unique StuI site, and subsequently, the lacZ gene was fused in frame to the sixtieth codon of cycY at an XhoI site, which gave plasmid pRJ2802. The phoA gene lacking the coding region for its own signal sequence was fused to the sixtieth codon of cycY by inserting the 2.9-kb 1 PstI fragment of pCH40 (20), yielding pRJ2806. All fusions contained the entire cycVWZXYЈ DNA and were confirmed to be in frame by sequencing of the fusion site. They were subsequently transferred into the broad host range plasmid pRK290 (21), giving plasmids pRJ2803 (cycY 6 Ј-ЈlacZ), pRJ2804 (cycY 60 Ј-ЈlacZ), and pRJ2757 (cycY 60 Ј-ЈphoA). These plasmids were conjugated into the B. japonicum wild type. The cycAЈ-ЈphoA fusion was created by digesting the 3Ј-end of cycA with Bal-31 nuclease and by inserting a PstI linker at codon 127 plus the 2.9-kb PstI fragment of pCH39 (20). This resulted in pRJ3280, whose 3.7-kb XhoI fragment was subcloned into the suicide plasmid pSUP202X (52) and cointegrated into the B. japonicum wild-type chromosome after conjugation.
The cycY Ϫ mutant was constructed by first inserting the 1.3-kb SmaI fragment containing the kanamycin cassette of pUC4-KIXX into the blunt-end cycY internal RsaI site and subsequently by subcloning the disrupted gene into pSUP202 (14). After conjugation, the mutant was obtained by double crossover events, yielding strain Bj2746.
To generate a CycY antigen for antibody production, a plasmid encoding a soluble version of the protein with a C-terminal His tag was constructed. For this purpose, a DNA fragment was amplified by polymerase chain reaction that contained an NcoI site at codon 38 of cycY and six additional histidine codons at the end of the gene. The fragment was amplified using the following primers: primer 1, 5Ј-GCATGCCAT-GGATCCTTCGCGGATTCCCTCCG-3Ј; and primer 2, 5Ј-GCAGATG-GAGAAGGCGCTGAAGCACCACCACCACCACCACTAGGATCCCG-3Ј. The polymerase chain reaction product was cloned as an NcoI-BamHI fragment into pUCBM21 (Boehringer, Mannheim, Germany) and sequenced. From this plasmid (pRJ2760), the NcoI-BamHI fragment was cloned into pET22b (Novagen, Madison, WI), resulting in pRJ2762. On the protein level, this led to the N-terminal fusion of the signal sequence from the Erwinia carotovora PelB protein with the soluble part of CycY (residues 38 -194) carrying a C-terminal His tag. This protein variant was called CycY His *. A similar CycY His version with the original B. japonicum-derived N-terminal hydrophobic CycY segment instead of the PelB signal sequence is encoded by plasmid pRJ2764. This plasmid was constructed by cloning the 0.44-kb XhoI-EcoRI fragment from pRJ2760 (cycY) into pRK290X (23), yielding pRJ2763. Subsequently, the adjacent 3.1-kb wild-type XhoI fragment containing the DNA upstream of the cycY internal XhoI site was cloned into pRJ2763, resulting in pRJ2764. This plasmid was conjugated into mutant Bj2746 and tested for complementation.
CycY*, the soluble form of CycY lacking the C-terminal His tag, is encoded by plasmid pRJ2766, in which the XhoI fragment of pRJ2762 (containing codons 61-194 plus the extra histidine codons; see above) was replaced by the corresponding 471-base pair wild-type XhoI fragment. The correct orientation of the insert and the intact fusion site were confirmed by sequencing.
Expression and Purification of CycY*-E. coli BL21(DE3) cells were transformed with the expression plasmid pRJ2766 and grown in 10 liters of LB medium containing ampicillin (200 g/ml) at 30°C for 36 h. (Induction of gene expression by isopropyl-␤-D-thiogalactopyranoside was not required for efficient expression of CycY* under these conditions.) The cells were harvested by centrifugation at 5000 ϫ g and washed in cold buffer A (10 mM MOPS/NaOH, 150 mM NaCl, and 5 mM EDTA, pH 7.0). After centrifugation at 12,000 ϫ g for 10 min, the cell pellet was suspended in cold extraction buffer (buffer A with 1 mg/ml polymyxin B sulfate (Sigma), 2 ml/g of cells, wet weight). The suspen-sion was gently shaken on ice for 1 h and centrifuged at 39,000 ϫ g for 30 min at 4°C. To increase the yield of periplasmic proteins, the extraction procedure was performed twice. The supernatants were combined, dialyzed extensively against buffer B (10 mM MOPS/NaOH, pH 7.0), and applied to a DE52 cellulose column (40 ml; Whatman, Maidstone, United Kingdom) equilibrated with buffer B. The eluate containing CycY* was directly loaded onto a CM52 cellulose column (40 ml; Whatman) equilibrated with buffer B. The column was washed with buffer B, and CycY* was eluted by a linear gradient (900 ml) from 0 to 500 mM NaCl in buffer B. Fractions containing CycY* (corresponding to 50 -100 mM NaCl) were pooled, and ammonium sulfate was added to a final concentration of 1.5 M. The solution was applied to a phenyl-Sepharose column (37 ml; Pharmacia, Uppsala) equilibrated with 20 mM Tris-HCl, pH 8.0, containing 1.5 M ammonium sulfate. An ammonium sulfate gradient (400 ml, 1.5 to 0 M) in the same buffer led to elution of pure CycY* between 0.8 and 0 M ammonium sulfate. The CycY*-containing fractions were dialyzed extensively against distilled water.
The CycY* concentration was determined by its absorbance at 280 nm. The molar extinction coefficients of unfolded CycY* (⑀ 280 nm ϭ 22,310 M Ϫ1 cm Ϫ1 ) and native oxidized CycY* (⑀ 280 nm ϭ 23,594 M Ϫ1 cm Ϫ1 ) were calculated as described by Gill and von Hippel (24). Analysis of free thiol groups in the CycY* preparation was carried out according to Ellman (25). The relative molecular weight of the native protein was determined on a calibrated Superdex 200HR gel filtration column (Pharmacia), revealing that the native protein is a monomer. The protein was stored at Ϫ20°C for several weeks without detectable degradation.
Biochemical Characterization-The redox reactivity of CycY* was analyzed by the different motility of the reduced and oxidized forms of the protein during SDS-PAGE. To reduce CycY*, 2 volumes of SDS loading dye containing 1.4 M 2-mercaptoethanol were added, and the sample was incubated for 15 min at room temperature. In parallel, an equivalent sample was incubated in SDS loading dye without 2-mercaptoethanol. The samples were boiled for 5 min and separated by SDS-PAGE.
Redox-dependent fluorescence of CycY* was measured according to Loferer et al. (26). To determine the equilibrium constant of the CycY*/ glutathione redox system, oxidized CycY* was incubated at 30°C in 100 mM sodium phosphate, pH 7.0, and 1 mM EDTA containing 10 M GSSG and different concentrations of GSH (125 M to 125 mM). The redox mixtures had to be incubated for 3 days under argon to reach equilibrium. Samples were removed immediately before fluorometric measurements were performed. To minimize interference from GSSG absorption, an excitation wavelength of 295 nm was used.
The relative amount of reduced CycY* at equilibrium (R) was calculated according to Equation 1, where F is the measured fluorescence intensity at 335 nm, and F red and F ox are the fluorescence intensities of completely reduced or oxidized CycY*, respectively. The GSSG concentration present in the reaction mixtures after equilibrium was quantified enzymatically using NADPH and yeast glutathione reductase (EC 1.6.4.2; Boehringer) in order to account for air oxidation of reduced glutathione during the incubation period and the GSSG contaminations present in the GSH stock solution. The assay was performed as described by Loferer et al. (26), except that the determination was done separately for each sample.
To determine the equilibrium constant between CycY* and glutathione by HPLC analysis, identical redox mixtures were made as described above, except that they contained 11.7 M CycY*. After an incubation time of 3 days, the equilibrium concentration of GSSG was determined, and thiol-disulfide exchange reactions were quenched by the addition of formic acid (pH Յ2). Oxidized CycY*, reduced CycY*, and CycY*/glutathione mixed disulfides were separated on a reversed-phase Vydac TM 218TP54 column with a gradient from 40 to 40.5% acetonitrile in 0.1% (v/v) trifluoroacetic acid during 55 min (flow rate: 0.5 ml/min). Proteins were detected by their absorbance at 275 nm, and the peaks were quantified by integration.
The overall equilibrium constant between CycY* and glutathione (K eq ) was calculated according to Equations 2 and 3 and verified by determining the individual equilibrium constants K 1 and K 2 independently according to Equations 4 and 5, where CycY* SH SH and CycY* S S are the reduced and oxidized forms of the protein, respectively, and CycC* SSG SH is the single mixed disulfide species.
The rate of the reaction between CycY* and DTT was determined by recording the change in fluorescence intensity at 335 nm at an excitation wavelength of 295 nm. Ten l of different DTT stock solutions (1 mM to 1 M) were added to 1 ml of oxidized CycY* (1 M) in degassed 100 mM potassium phosphate, pH 7.0, and 2 mM EDTA, and the change in fluorescence intensity was recorded for 15 min. The apparent secondorder rate constant of CycY* reduction was determined using the measured pseudo first-order rate constants.
Production of CycY His * Antigen and Antiserum-E. coli BL21(DE3) cells were transformed with the expression plasmid pRJ2762 and grown in 2 liters of LB medium at 37°C to an absorbance (550 nm) of 0.5. After induction by isopropyl-␤-D-thiogalactopyranoside (final concentration: 0.2 mM), the cells were grown at 37°C for 2 h. The cells were harvested by centrifugation and washed in cold buffer C (10 mM Tris-HCl, pH 7.9, 150 mM NaCl, and 5 mM EDTA). The preparation of the periplasmic extract was done as described above, except that buffer D (buffer C with 1 mg/ml polymyxin B sulfate, 2 ml/g of cells, wet weight) was used. The purification by nickel affinity chromatography (His-Bind metal chelation resin, Novagen) was performed according to the instructions of the manufacturer. The purified fraction was analyzed by SDS-PAGE, and the protein was stained with 0.01% Coomassie Brilliant Blue R-250. The gel piece containing the pure protein was excised and used to immunize a New Zealand White rabbit. The antiserum was purified following the method of Smith and Fisher (27).
Cell Fractionation-Membrane and soluble fractions of B. japonicum cells were prepared as described by Loferer et al. (28), except that the extraction buffer contained 1 mM MgCl 2 . Protein concentrations were measured using the Bio-Rad assay with bovine ␥-globulin as standard. Heme stains were performed as described previously (29).
Western Blotting-Proteins were separated by SDS-PAGE and electroblotted onto Hybond-C nitrocellulose (Amersham International, Buckinghamshire, United Kingdom). Cross-reacting polypeptides were detected by the biochemiluminescence method according to the instructions of the manufacturer (Boehringer).
Enzymatic Assays-␤-Galactosidase activity of strains harboring cycYЈ-ЈlacZ fusions was measured from 100-l samples of two independent cultures as described by Miller (30). Alkaline phosphatase activity and N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine oxidase activity on bacterial colonies were measured according to Loferer et al. (28). Symbiotic nitrogen fixation activity was determined in plant infection tests with soybean (Glycine max L. Merr) (31), and nitrogen fixation was determined by the acetylene reduction assay (32).

Characterization of the B. japonicum cycY Gene-
The open reading frame downstream of the B. japonicum cycX gene was investigated with respect to its translation start site (Fig. 1). Two principal possibilities were considered. (i) orf194 might start with an ATG codon overlapping with the cycX stop codon. The derived gene product would have the capacity to be exported by its N-terminal signal sequence-like domain. (ii) The start codon might be the rare TTG, as suggested previously by Ramseier et al. (11), thus leading to an orf132-encoded soluble thioredoxin-like protein that did not carry an export sequence. To investigate the true start of the gene downstream of cycX, E. coli lacZ and phoA genes were fused in frame proximal and distal to the coding sequence of the putative transmembrane helix at the N terminus of the orf194 gene product, i.e. at codons 6 and 60, respectively (Fig. 1). The lacZ fusion at codon 6 produced significant amounts of ␤-galactosidase activity, showing that translation of the gene downstream of cycX in fact starts at the ATG codon that overlaps with the end of cycX. Hence, orf132 must be abandoned in favor of an extended orf194. By analogy with the situation found in R. leguminosarum (9), we named the gene cycY. The LacZ fusion at amino acid 60 after the hydrophobic segment gave significantly lower ␤-galactosidase activity (Table I). This result is in agreement with the hypothesis that the hydrophobic segment translocates the C-terminal portion of CycY through the membrane, and translocated ␤-galactosidase is well known to be inactive. To further substantiate this finding, alkaline phosphatase was also fused at position 60 of CycY. Significant levels of PhoA activity were found in whole cell extracts, as expected for a periplasmically located fusion protein (Table I). Thus, the hydrophobic segment underlined in Fig. 1 most likely functions as a transmembrane domain and orients the hydrophilic C-terminal portion of the attached protein moiety toward the periplasm.
A cycY Ϫ null mutant was constructed by introducing a kanamycin resistance cassette into the RsaI site that maps precisely to the coding sequence between the two conserved cysteines (Fig. 1). The mutant was tested for its ability to synthesize c-type cytochromes. Clearly, physiological reactions depending on the presence of c-type cytochromes such as N,N,NЈ,NЈ-tetramethyl-p-phenylenediamine oxidase activity, anaerobic growth with nitrate as the terminal electron acceptor, and symbiotic nitrogen fixation were completely defective in the mutant (data not shown). The presence of the different c-type cytochromes was also tested by staining of proteins for covalently bound heme in membrane and soluble fractions separated by SDS-PAGE ( Fig. 2A). None of the c-type cytochromes present in the wild type were found in the cycY Ϫ mutant. Also, the apoproteins of the cytochromes c 550 (Fig. 2B), c 1 , CycM, FixO, and FixP (data not shown) were not detectable in the mutant by Western blot analysis. Thus, cycY is essential for maturation of all B. japonicum c-type cytochromes.
Subcellular Localization of CycY-The alkaline phosphatase activity derived from the cycY 60 Ј-ЈphoA fusion suggested that the C-terminal bulk of the CycY protein is located on the periplasmic side of the membrane. In fact, CycY contains a possible cleavable signal peptide (Fig. 1). Alternatively, the 19-amino acid hydrophobic stretch might allow this segment to span the membrane, which might anchor the otherwise periplasmic protein to the membrane. We tested whether the cycY 60 Ј-ЈphoA fusion produced membrane-bound or periplasmic enzymatic activity. Table I shows that there was an approximately even distribution of PhoA activity in membrane and soluble fractions, suggesting that at least part of the CycY-PhoA hybrid remained attached to the membranes. For comparison, we constructed a phoA fusion to cycA, the gene encoding the soluble B. japonicum cytochrome c 550 . This protein is known to be exported into the periplasm by its N-terminal cleavable signal peptide (33). In this case, most of the PhoA activity was found in the soluble fraction, indicating that efficient cleavage of the signal peptide and release of the CycA-PhoA hybrid into the periplasm had taken place (Table I). We also attempted to localize the CycY protein using a more direct immunological approach. A CycY variant carrying a C-terminal His 6 tag was expressed in E. coli and secreted into the periplasm by an N-terminally fused PelB leader sequence (see "Experimental Procedures" and Fig. 1). The purified protein was used as an antigen to produce polyclonal CycY-specific antiserum for detection of CycY in B. japonicum membrane and soluble fractions (Fig. 3). Clearly, wild-type membranes (ϩ lanes) contained CycY protein that approximately corresponded in size to CycY His , a His-tagged version of an otherwise wild-type CycY (H lanes) that was used for complementation of the cycY Ϫ mutant. The CycY polypeptide was absent in the cycY Ϫ mutant (Ϫ lanes). Neither strain exhibited detectable amounts of CycY cross-reacting material in the soluble fraction (Fig. 3). We concluded that CycY is a membrane-anchored periplasmic thioredoxin, similar to B. japonicum TlpA (26) or E. coli DipZ (7).
Overexpression and Purification of Soluble CycY*-DNA encoding a soluble form of CycY without the N-terminal hydrophobic domain (codons 39 -194; CycY*) was cloned downstream of the DNA for the PelB signal sequence in the expression vector pET22b. The N-terminal PelB signal sequence caused an efficient secretion of the protein into the periplasm, whereby the signal peptide was cleaved off quantitatively. Soluble CycY* was purified to homogeneity from the periplasm of E. coli BL21(DE3)/pRJ2766 cells. Purification was achieved by sequential chromatography on DE52-and CM52-cellulose and on phenyl-Sepharose, yielding 290 mg of homogeneous CycY* from a 10-liter culture (Fig. 4A, lane 4). N-terminal amino acid sequence analysis revealed the sequence MDPSRIP, confirming that cleavage of the PelB leader peptide had occurred at the expected site (cf. Fig. 1). Correct cleavage of the signal sequence was also shown by electrospray mass spectrometry of purified CycY* (calculated mass: 17005.4 Da; experimentally determined mass: 17005.4 Da). Analysis of free sulfhydryl groups, using Ellman's reagent (5,5Ј-dithiobis(2-nitrobenzoic acid)), showed that 99.7% of the protein was obtained in the oxidized state.
Characterization of Redox Properties-CycY* contains two cysteine residues separated by only two amino acids in the proposed active site (CVPC) (Fig. 1). We investigated the protein by SDS-PAGE under reducing and nonreducing conditions. Fig. 4B shows that the oxidized form of CycY* migrated slightly faster on the gel than the reduced form. This confirmed that the two cysteine residues are capable of forming a disulfide bond.
It is known that many proteins of the thiol-disulfide oxidoreductase family such as thioredoxin and DsbA of E. coli (34,35) or TlpA of B. japonicum (26) show an increase in fluorescence intensity upon reduction of their active-site cystines. Therefore, the fluorescence spectra of oxidized and reduced CycY* were compared. When oxidized CycY* was excited at 295 nm (selective excitation of tryptophan residues), the addition of 1 mM DTT led to a 2.3-fold increase in the fluorescence intensity and to a shift of the emission maximum from 327 to 332 nm (Fig. 5A). We assume that, like in thioredoxin, the fluorescence of at least one of the two homologous tryptophans adjacent to the cysteines is quenched by the disulfide bond (Fig. 1). In the presence of 7 M guanidinium chloride, almost identical spectra for unfolded oxidized and unfolded reduced CycY* were obtained with emission maxima of 354 nm (Fig. 5A), which are typical for denatured proteins. The minor difference in fluorescence intensity between unfolded oxidized and unfolded reduced CycY* may reflect the fact that the quenched tryptophan residues are located in the immediate vicinity of the active-site  disulfide in the primary sequence of the protein.
Using the fluorescence properties of CycY*, the CycY*/glutathione redox equilibrium constant K eq was determined, which is given by Equations 6 and 7.
Oxidized CycY* was incubated in the presence of 10 M GSSG and increasing concentrations of GSH. The fraction of reduced CycY* at equilibrium was measured by the intrinsic CycY* fluorescence assuming no significant equilibrium concentrations of CycY*/glutathione mixed disulfides (Fig. 5B). After fitting the data by nonlinear regression according to Equation 8, an equilibrium constant for the CycY*/glutathione system of 0.168 Ϯ 0.008 M was determined (correlation coefficient: 0.993). A standard redox potential of Ϫ0.217 V for the active-site cysteines of CycY* at 30°C and pH 7.0 (EЈ 0C ) was calculated from the Nernst equation using a value of Ϫ0.240 V (36) for the glutathione standard redox potential (EЈ 0G ) (Equation 9).
The redox potential of CycY* is closer to that of cytoplasmic thioredoxins (Ϫ0.23 to Ϫ0.27 V) (37)(38)(39) than to that of eukaryotic protein disulfide-isomerase (Ϫ0.11 V) (40) and DsbA (Ϫ0.124 V) (35), indicating that CycY* is a rather reducing enzyme. (For direct comparison, a redox potential of Ϫ0.240 V was used for the glutathione redox potential, and the published value for DsbA was recalculated with that value.) The redox mixtures had to be incubated for at least 3 days to reach equilibrium, which is much longer than in the case of DsbA (35) or TlpA (Ͻ20 h) (26).
To measure the CycY*/glutathione equilibrium by an independent method, it was quenched by acid. Oxidized CycY*, reduced CycY*, and CycY*/glutathione mixed disulfides were separated by reversed-phase HPLC. Quantification of the peak areas allowed determination of the equilibrium constants K 1 and K 2 , whose product corresponds to K eq (Equation 2). Under any redox conditions, the fraction of the CycY*/glutathione mixed disulfide was below 16%. The values for K 1 , K 2 , and K eq are 3.8 and 4.7 ϫ 10 Ϫ2 M and 0.182 M, respectively, the latter of which is in good agreement with the value of K eq obtained by fluorescence spectrometry.

DISCUSSION
The results of this work show that B. japonicum contains a cytochrome c biogenesis gene that codes for a periplasmically oriented membrane-anchored protein with a thioredoxin-like function. The sequence of cycY has been published before (11); however, at that time, it was not clear where the translation of the open reading frame started and whether or not the gene was involved in maturation of c-type cytochromes. The results presented here have clarified both points. First, the lacZ and phoA fusion analyses unambiguously showed that the hydrophobic segment at the N terminus, encoded by what was previously called orf194 (11), is translated and thus part of the protein. However, the identity of orf194 as the bona fide cycY gene was previously questioned due to the inconspicuous wildtype phenotype of B. japonicum mutant 98, which carries a kanamycin resistance cassette inserted between codons 4 and 5. How can this be explained? We sequenced the border fragments of the pUC4-KIXX-derived kanamycin cassette and found that by inserting the XhoI fragment of the cassette into the SalI site overlapping codons 4 and 5 of cycY, an in-frame fusion of cycY to a sequence encoding a newly created Nterminal tripeptide (MRI) is obtained (Fig. 1). Interestingly, the ATG start codon of this tripeptide is preceded by the sequence AAG at a distance of nine nucleotides, which might serve as a ribosome-binding site (data not shown). Thus, assuming there is promoter activity within the kanamycin cassette, the mutant might express an almost normal CycY protein in which only the first four amino acids (MSEQ) are replaced by MRI (Fig. 1). Hence, strain 98 might produce an artifactual CycY protein that is normally exported and functional, leading to a wild-type phenotype. Some mutations in B. japonicum cytochrome c biogenesis genes lead to small colonies on plates, probably because (micro)aerobic respiration is affected when c-type cytochromes are not synthesized (43). We now succeeded in constructing a true cycY Ϫ mutant by screening for small colonies after conjugation. The slow-growing exconjugants proved to be the correct mutants. Therefore, although cycY is important, it is not essential for (micro)aerobic growth of B. japonicum due to the presence of an alternative quinol oxidase (44). However, anaerobic growth with nitrate, which depends on certain c-type cytochromes such as cytochrome c 550 (45), was affected. The absence of all c-type cytochromes in the cycY Ϫ mutant indicated that a general step of the maturation pathway was affected by the mutation.
The subcellular localization of the CycY protein was investigated by two different approaches. The result with the cycYЈ-ЈphoA fusion concurred very well with the direct immunological localization of the protein to the membrane fraction. This is a new aspect in view of some indirect evidence for the periplasmic location of the R. capsulatus CycY homologue, HelX. helXЈ-ЈphoA fusions were expressed in E. coli and gave high alkaline phosphatase activities in the periplasmic fraction, suggesting that the hybrid protein was cleaved after translocation (8). Indeed, several putative signal sequence cleavage sites were identified in the HelX presequence. Nevertheless, there is no direct evidence for the precise subcellular location of HelX in R. capsulatus. In the N-terminal leader sequence of the B. japonicum CycY protein, the sequence GSG (positions 36 -38; see Fig.  1) following a hydrophobic stretch of 19 amino acids might serve as a signal peptidase cleavage site. However, our results support the idea that CycY is anchored in the cytoplasmic membrane and faces the periplasmic space, which is reminiscent of another membrane-bound thioredoxin-like protein identified in this organism, TlpA (28). The latter is necessary for the biosynthesis of the aa 3 -type cytochrome oxidase.
What might be the biological function of the CycY protein? Its resemblance to thioredoxins suggests that it is a periplasmic protein thiol-disulfide oxidoreductase. Knowing that CycY is required for one of the steps in the cytochrome c maturation pathway, it is tempting to speculate that its function is to keep the cysteine residues in the heme-binding motif of apocytochromes c reduced before heme is attached. This hypothesis predicts that CycY is a reductant, with a possible intramolecular disulfide in apocytochrome c being the target. CycY*, the soluble version of CycY expressed in E. coli, was tested for its redox reactivity, and its redox potential was determined. The only two cysteines of the polypeptide are located in the WCVPC motif ( Fig. 1) that presumably forms the active site and is thus reminiscent to that of E. coli thioredoxin (WCGPC), supporting the view that CycY is a thioredoxin-like, redox-active enzyme.
The spectroscopic characterization of CycY* also revealed strong similarities to thioredoxin. The fluorescence of CycY increased by a factor of 2.3 upon reduction of its disulfide, compared with a ϳ3-fold increase in fluorescence observed for the reduction of thioredoxin (34). These fluorescence properties of CycY* were used to determine its intrinsic redox potential. The obtained value of Ϫ0.217 V is similar to that of E. coli glutaredoxin, a protein that complements thioredoxin deficiency in ribonucleotide synthesis, and is closer to that of the reductant thioredoxin (Ϫ0.23 to Ϫ0.27 V) than to that of the oxidants DsbA and protein disulfide-isomerase (Ϫ0.124 and Ϫ0.11 V, respectively). This indicates that CycY* may indeed act as a reductant in the otherwise oxidizing environment of the periplasm. The determination of the equilibrium constant between CycY* and glutathione by fluorescence measurements was based on the assumption that the population of CycY*/ glutathione mixed disulfides at equilibrium was negligible. This was proven by HPLC analysis of acid-quenched equilibrium mixtures. Since the fraction of the mixed disulfides was Ͻ16% even under oxidizing conditions, the fluorescence analysis yields a reliable value for K eq . This result was confirmed independently by evaluating the HPLC elution profiles under different redox conditions, which also allowed the determination of the "microscopic" equilibrium constants K 1 and K 2 (see Equations 4 and 5).
In conclusion, there is now genetic and biochemical evidence for CycY being a membrane-bound thioredoxin involved in a redox reaction required for cytochrome c maturation in the periplasm. However, the natural substrate of CycY, which could be apocytochrome c, heme, or yet another periplasmic molecule that is required for cytochrome c maturation, has not yet been identified. Furthermore, we do not know how CycY, once oxidized in the course of a reaction cycle, becomes recycled as a reductant in the oxidative environment of the periplasm. In this context, it is interesting to note that another cytochrome c biogenesis protein of B. japonicum, the cycL gene product, also contains a periplasmically oriented CXXC motif, but otherwise does not resemble thioredoxins (3). This protein might also catalyze disulfide reduction in apocytochromes c or CycY. In E. coli, several periplasmic protein thiol-disulfide oxidoreductases have been shown to be required for cytochrome c maturation, among which are DsbA, DipZ, and CcmG (5,7,46,47). CcmG shares a high degree of sequence similarity with HelX and CycY. Apart from the extra N-terminal hydrophobic stretch in this class of thioredoxins, an additional sequence of 27 highly conserved amino acids in the C-terminal third of these polypeptides is remarkable, which might be essential for substrate recognition (isoleucine 149 to proline 175).
A relevant observation in this context may be that, in contrast to other thiol-disulfide oxidoreductases including TlpA (26), CycY* was not capable of mediating the reduction of insulin disulfides by DTT, one of the standard assays for thioldisulfide oxidoreductase activity (41). Measurement of the apparent second-order rate constant of the reduction of CycY* by DTT (3.5 M Ϫ1 s Ϫ1 at pH 7.0) showed that this reaction is ϳ3 orders of magnitude slower compared with thioredoxin (ϳ10 3 M Ϫ1 s Ϫ1 ) (41) and ϳ6 orders of magnitude slower compared with DsbA (ϳ10 6 M Ϫ1 s Ϫ1 ) (Ref. 42 and data not shown). It follows that generation of reduced CycY* by DTT as catalytic reductant of insulin disulfides becomes rate-limiting under the conditions used, which explains why CycY* does not act as a catalyst in the insulin reduction assay (48). The slow disulfide exchange reactions of CycY* were also reflected by the fact that 3 days of incubation were necessary to reach equilibrium with glutathione redox buffers. The slow reactions between CycY* and small organic thiol compounds suggest that the enzyme may specifically act on a single substrate molecule in vivo. It is known from studies on the polypeptide specificity of DsbA and protein disulfide-isomerase (42, 49 -51) that a specific substrate-binding site next to the active-site disulfide increases the rate constants of the disulfide exchange between the enzyme and its substrates. To investigate the participation of CycY in redox reactions in a more general way, two further possibilities of CycY reactivity were explored: (i) the interaction between oxidized DsbA from E. coli and reduced CycY* was investigated, revealing that DsbA was not capable of oxidizing CycY* in vitro; (ii) the alkaline phosphatase activity of the cycAЈ-ЈphoA fusion was compared in the B. japonicum wild type and in the cycY Ϫ mutant. The activity was identical in both strains, indicating that, in contrast to DsbA (22), CycY is not involved in disulfide bond formation of alkaline phosphatase (data not shown). Our main goal for future experiments is now the identification of the target molecule of CycY.