INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Flavocytochrome c-sulfide dehydrogenase (FCSD) was first identified by Bartsch and Kamen (6) in the purple phototrophic bacterium Chromatium vinosum, recently renamed Allochromatium vinosum (21). It was subsequently found in the green phototrophic bacterium Chlorobium limicola f. sp. thiosulfatophilum (7, 30) and in six other species of purple and green bacteria, Chromatium gracile (4, 5) and Chromatium purpuratum (24), both now assigned to the genus Marichromatium (21); Chromatium tepidum (20); Thiocapsa roseopersicina (54); Chlorobium limicola f. sp. thiosulfatophilum (43); and Chlorobium phaeobacteroides (16). A membrane-bound form of FCSD was discovered in the nonphototrophic aerobic bacterium Thiobacillus sp. W5 (50). A gene homologous to the flavoprotein subunit of FCSD was found to be associated with the genes for thiosulfate oxidation in Paracoccus denitrificans, but the adjacent cytochrome gene was more divergent than expected (52). The genome sequence of Aquifex aeolicus contains two FCSD flavoprotein genes associated with one for a Rieske iron-sulfur protein, and a thiosulfate utilization operon is located elsewhere in the genome (14). Thus, there appears to be a correlation between the presence of FCSD and sulfur metabolism.

The Chromatium and Chlorobium flavocytochromes c were found to have sulfide dehydrogenase activity in vitro (19, 25) and presumably are involved in sulfur metabolism in vivo. A common characteristic of the bacteria in which this enzyme is found is the ability to utilize reduced sulfur compounds as electron donors for carbon dioxide fixation. Thus, all species of green and purple sulfur bacteria utilize elemental sulfur and hydrogen sulfide, approximately half the species use thiosulfate, and a few use sulfite or tetrathionate (10). However, thiosulfate and sulfite are not oxidized by FCSD. Once the genes for C. vinosum FCSD were cloned and sequenced, the periplasmic location of the enzyme was established by the presence of signal sequences (15).

The discovery of a homologous enzyme, sulfide-quinone reductase (SQR), in some photosynthetic and nonphotosynthetic sulfur bacteria (2, 39, 40, 41) plus the isolation of the Rhodobacter capsulatus SQR gene brought a new perspective on the involvement of FCSD and SQR in sulfide oxidation. During the oxidation of sulfide, sulfur globules are deposited either in the periplasmic space, as in Chromatium species (34), or extracellularly (in Chlorobium and Ectothiorhodospira species) (18). Thus, sulfide oxidation generally occurs on the periplasmic side of the cytoplasmic membrane in purple and green sulfur bacteria, where FCSD is localized. The location of the R. capsulatus SQR was not clear until recently because the protein appeared to have no N-terminal signal peptide for translocation into the periplasmic space (41). It has now been documented from gene fusion experiments that R. capsulatus SQR functions on the periplasmic side of the cytoplasmic membrane, using an unknown mechanism for translocation (38). Genes for FCSD have not been found in R. capsulatus, nor have SQR genes been conclusively demonstrated in either Chromatium or Chlorobium. On the other hand, disruption of the FCSD genes in C. vinosum had no effect on sulfide oxidation (35). However, there are two FCSD genes in the C. tepidum genome (http://www.tigr.org), suggesting that if there are two sets of FCSD genes in C. vinosum, the effects of a knockout mutation would be negated. Nevertheless, which of the two enzymes, FCSD or SQR, is the sulfide-oxidizing enzyme in Chromatium and Chlorobium remains to be determined.

Ectothiorhodospira species make up a small family of mostly marine and halophilic purple phototrophic bacteria, all of which utilize sulfide for growth. Unlike Chromatium, but similar to Chlorobium, elemental sulfur is deposited outside the cells in the growth medium. The soluble electron transfer proteins of Ectothiorhodospira are generally similar to those of Chromatium. Thus, they are dominated by HiPIP (high-potential iron-sulfur protein) and cytochrome c' (26, 27, 31). FCSD has not been previously reported to occur in this family of bacteria.

The amino acid sequences of the cytochrome subunits of Chromatium and Chlorobium FCSDs have been determined (47, 48), as has that of the Chromatium flavoprotein subunit (49). The three-dimensional structure of Chromatium FCSD has also been determined (12). We now report the nucleotide sequence of an Ectothiorhodospira vacuolata FCSD gene and its flanking regions, which shows that FCSD is more widespread than previously thought and that the genetic context is not the same as that in Chromatium. The membrane localization of the E. vacuolata protein is documented by immunoblotting.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Strains and media. E. vacuolata 1 strain 2111, obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) was grown on DSMZ medium 1448, which is a modification of the American Type Culture Collection medium 1410 (30 g of NaCl/liter instead of 140 g/liter). The culture was grown by anaerobic photosynthesis in light provided by a 40-W tungsten lamp at 30°C. Escherichia coli strains were grown on Luria-Bertani medium (36) supplemented with 100 µg of carbenicillin. Strain XL-1 blue was used as a recipient to detect -complementation for pUC18 derivatives on Luria-Bertani plates supplemented with 80 µM IPTG (isopropyl--D-thiogalactoside) and 32 µg of X-Gal (5-bromo-4-chloro-3-indolyl--D-galactoside) per ml.

Isolation and purification of cytochrome c-552. Cells were harvested by centrifugation and broken in a Ribi cell fractionator (an automated French press). The membranes were sedimented by ultracentrifugation for 3 h in a Spinco Ti45 rotor. They were resuspended in water at 4°C, and an equal volume of n-butanol at 20°C was added with stirring (46). After centrifugation, the top layer containing butanol and colored pigments was removed. The underlying water layer contained the protein of interest. Cytochrome c-552 chromatographed with HiPIP on DEAE-cellulose and eluted after cytochrome c₄. It appeared to exist as both monomer and dimer when chromatographed on Sephadex G-50. The combined Sephadex fractions were chromatographed on DEAE-cellulose developed with 20 mM Tris-HCl containing 2.5 mM NaCl, which resolved the cytochrome c-552 from a HiPIP isozyme.

Amino acid sequence determination of cytochrome c-552. The covalently bound heme was removed from the native protein by treatment with HgCl₂ in 8 M urea-0.1 M HCl at 37°C for 16 h (1). After separation of the apoprotein from heme and salts by gel filtration (Sephadex G-25, 5% HCOOH), the N-terminal sequence was determined by using a freeze-dried aliquot of 400 pmol of the apoprotein. To identify the cysteine residues of the heme binding site, a 1.3-nmol aliquot of apoprotein was treated with 3-bromopropylamine. The alkylated cysteine residues could then be detected by sequence analysis (22). A second aliquot of 9.5 nmol of the apoprotein was digested for 3 h with Staphylococcus aureus protease at pH 4, using an enzyme-to-substrate ratio of 1:40. A third aliquot of 9 nmol was subjected to partial acid hydrolysis in 2% formic acid for 2 h at 106°C. The peptides were separated by reversed-phase high-performance liquid chromatography (SMART system; Pharmacia, Uppsala, Sweden) on a PEPSII column using a gradient of 0.07% trifluoroacetic acid in water (solvent A) and 0.05% trifluoroacetic acid in acetonitrile (solvent B). Sequence determination was performed on a 477A or 476A pulsed liquid sequenator, with on-line analysis of the phenylthiohydantoin-amino acids on a 120A analyzer (PE Biosystems, Foster City, Calif.). Sequencing reagents were from the same firm. The masses of the holoprotein, apoprotein, and peptides were determined by using either plasma desorption, electrospray ionization, or matrix-assisted laser desorption mass spectrometry on a Biopolymer (Uppsala, Sweden) analyzer (BIO-ION), a BIO-Q triple-quadruple instrument (Micromass, Altrincham, United Kingdom), or a TOF-SPEC SE time-of-flight analyzer (Micromass, Whytenshaw, United Kingdom), respectively.

DNA techniques. E. vacuolata genomic DNA was isolated by the cetyltrimethylammonium bromide method (3). On the basis of the cytochrome c-552 protein sequence, a set of degenerate primers was designed. The N-terminal primer, EVM8, had the sequence 5' ATGGCHACHACHTGYTAYG 3', and the C-terminal primer, EVT61, had the sequence 5' AGCTTGATYTCYTCRTCHGTRTA 3'. The probe was amplified by PCR using Taq polymerase (Amersham Pharmacia Biotech, Uppsala, Sweden) under the following conditions: 95°C, 2 min; 5 precycles (94°C, 30 s; 50°C, 60 s; 72°C, 30 s); 30 cycles (94°C, 30 s; 52°C, 30 s; 72°C, 60 s); 72°C, 10 min (followed by 4°C). The amplified fragment of 177 bp was cloned in the pGEM-T vector (Promega) and labeled via PCR with dioxigenin-dUTP (Roche Molecular Biochemicals). Via Southern hybridization, a BamHI fragment was identified. A BamHI pUC18 library was then constructed and analyzed (36). Detection and identification of transformants were done with the nonradioactive digoxigenin-DNA detection system (Roche Molecular Biochemicals). Double-stranded plasmid DNA was sequenced using dye terminator cycle sequencing (PE Biosystems). The sequencing was started from both ends with the universal primers M13F and M13R (England Biolabs, Inc., Beverly, Mass.) and was continued with the specific primers 552/3 (5' CGCCACCGTCATGGATC 3') and 552/4 (5' GCTGCCGGCACTGTGTC 3'), created on the basis of the c-552 DNA sequences. New primers were synthesized at approximately 450-nucleotide intervals based on the results of the previous sequencing.

Membrane protein purification and Western blotting. Ten liters of E. vacuolata culture was harvested in the exponential phase. Membranes were purified from French press-lysed cells via ultracentrifugation (160,000 × g 3 h) combining the EDTA-lysozyme method (32) and fractionation with 20% (NH₄)₂SO₄. The membranes were solubilized with 1% Triton X-100-10 mM Tris-HCl and 10 mM EDTA, pH 8 (3-h incubation; 4°C), and insoluble material was removed by ultracentrifugation. The supernatant was desalted and adsorbed on a DEAE-Sepharose column. The purification protocol was performed basically as described in reference 50 with the exception of using the above-mentioned Triton-Tris-EDTA buffer. Proteins were eluted with a linear gradient of 0 to 0.5 M KCl in the same buffer. The pooled fractions were desalted, concentrated, and further purified by loading them on a Q-Sepharose Fast Flow column in the above-mentioned buffer and gradient. The purified protein was identified on sodium dodecyl sulfate (SDS) gels by silver and heme staining. Western blotting was performed following the manufacturer's instructions for the enhanced chemiluminescence membrane (Amersham Pharmacia Biotech, Roosendaal, The Netherlands) and using an antibody against C. vinosum FCSD.

Protein electrophoresis. SDS-polyacrylamide gel electrophoresis (PAGE) was performed on vertical 10% polyacrylamide gels (28) stained for protein with Coomassie blue or silver. The gels used for heme staining were incubated for 10 min in a 10% trichloroacetic acid solution; the staining itself was performed as described in reference 45. PAGE as well as protein-blotting experiments were performed using the Mini Protean equipment (Bio-Rad, Veenendaal, The Netherlands).

RNA extraction and Northern analysis. Total RNA from E. vacuolata was extracted with LiCl (3) and electrophoresed in 2% agarose formaldehyde gels. All RNA samples were treated with RNase-free DNase I (Sigma) for 10 min at 37°C. Afterwards, 14 µl of denaturation mix (9.2 µl of formamide, 3 µl of formaldehyde, and 1.8 µl of MOPS buffer (40 mM morpholinopropanesulfonic acid, 10 mM sodium acetate, 2 mM EDTA, pH 7.2) was added to 15 µg of RNA. Denaturation was carried out at 65°C for 5 min, followed by cooling on ice. Before the mixture was loaded onto the gels, 2 µl of loading buffer was added. Digoxigenin-labeled pSPTNeo RNA (Roche Molecular Biochemicals) was treated as described above and used as a size marker on the gels. The RNA was transferred by capillary blotting in SSC buffer (3 M NaCl, 300 mM sodium citrate, pH 7), fixed to positively charged nylon membranes (Roche Molecular Biochemicals), and fixed by UV irradiation. Prehybridization, hybridization (42°C for 16 h), and chemiluminescent detection were carried out essentially as prescribed by Boehringer (The DIG System User's Guide for Filter Hybridization, Boehringer Mannheim, Mannheim, Germany, 1993). In order to study the regulation of FCSD by sulfide, it was necessary to use 0.05 mg of cysteine/ml in the growth medium to maintain anaerobiosis.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Isolation of cytochrome c-552. The soluble electron transfer proteins of E. vacuolata strain 1 (DSM 2111) are very similar to those of Ectothiorhodospira shaposhnikovii (26, 27). There are at least two HiPIP isozymes and cytochromes c', c₄, and b₅ (unpublished work). Following buffer solubilization of these proteins, we extracted the membrane fraction with butanol to identify peripheral electron transfer proteins. Additional HiPIP and cytochrome c₄ were released, along with a small cytochrome c-552 that cochromatographed with a third HiPIP isozyme from which it was difficult to separate it. (We have also found four soluble HiPIP isozymes as well as cytochrome c₄ in the related species Ectothiorhodospira mobilis [unpublished]). In addition, an abundant, high-molecular-weight cytochrome c-553 was easily separated by gel filtration. This protein has not been further characterized but is likely to be the tetraheme reaction center cytochrome, based upon its general occurrence in purple bacteria and its membrane localization. Cytochrome c-552 was purified as described in Materials and Methods.

Amino acid sequence analysis of the solubilized cytochrome c-552. The complete amino acid sequence of the small cytochrome c-552 was determined as shown in Fig. 1. Cytochrome c-552 contains 77 residues and a single heme binding site near the N terminus. The sequence is clearly that of a class I cytochrome and is most closely related to the 86-residue cytochrome subunit of C. limicola FCSD (49% identity with no internal insertions or deletions) (48) (Fig. 2). It is 26% identical to the first half, including one gap (a gap is defined as an insertion or deletion), and 19% identical (with four gaps) to the second half of the 174-residue diheme cytochrome subunit of C. vinosum FCSD (47). These results strongly suggested the presence of an associated flavoprotein subunit in E. vacuolata.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Amino acid sequence of the cytochrome subunit of flavocytochrome c from E. vacuolata. The N-terminal sequences of the apoprotein and the modified apoprotein are indicated by N-apo and N-mod, respectively. Peptides obtained after cleavage with S. aureus protease or after partial acid hydrolysis are named Ec and AH, respectively. The arrows indicate residues chemically identified during Edman degradation. An asterisk means that the sequence analysis was deliberately stopped. Mass spectrometry was used to determine the molecular weight of each of the peptides (results not shown).

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Alignment of the amino acid sequences of the FCSD E. vacuolata (1) and C. limicola f. sp. thiosulfatophilum (2) monoheme cytochrome subunits (48), with the diheme cytochrome subunit of C. vinosum (47), first half (3) and second half (4). The amino acids in boldface represent the heme binding site.

Gene sequence of the FCSD locus. Based upon the amino acid sequence of cytochrome c-552, we obtained a subgenomic clone of 5.5 kb. The translated cytochrome gene contains a 28-residue signal sequence and has a normal Ala-Thr-Ala recognition sequence for cleavage by the signal peptidase (51). The derived mature protein sequence was identical to that of the cytochrome sequenced by Edman degradation. Sixteen bases downstream of the cytochrome gene is a 1,290-base open reading frame which encodes a 430-residue protein homologous to the flavoprotein subunit of Chromatium FCSD. There is a 33-residue signal peptide which does not have an obvious cleavage site. It thus appears that the leader is not cleaved or is cleaved in front of Cys 25, to which a diacyl glycerol may be attached, as is the case in lipoproteins (33). The flavoprotein subunit is 50% identical to that of Chromatium, and there are only six small gaps in the protein sequence alignment (Fig. 3). The FCSD of Thiobacillus sp strain W5 is membrane bound and has a small monoheme cytochrome subunit, suggesting close similarity to the E. vacuolata protein (50).

View larger version (59K): [in this window] [in a new window]   FIG. 3.   Alignment of the complete amino acid sequences of the FCSD flavoprotein subunits from E. vacuolata (1), C. vinosum (49) (2), P. denitrificans (52) (partial) (3), and A. aeolicus (14) (starts with position 40) (4). The boxed amino acids represent more than 75% conserved positions.

FCSD is a membrane protein. Since no FCSD was found in the soluble extracts and only the cytochrome subunit was solubilized with butanol, we attempted to confirm that FCSD was localized in the membrane fraction by extraction with the detergent Triton X-100. The purification protocol yielded a small amount of partially purified FCSD. SDS-PAGE of the enzyme showed two abundant protein bands with molecular masses of 46 and 10 kDa. Heme staining revealed that the 10-kDa band was the cytochrome c-552 subunit (Fig. 4). The spectra of the dithionite-reduced enzyme showed the characteristic maxima for cytochrome c, 417 ( band), 524 ( band), and 552 ( band) nm, similar to those of the butanol-solubilized cytochrome. The oxidized spectrum did not display shoulders at 450 and 480 nm or bleaching with dithionite, which is typical of FCSD flavin. This may be due to modification of the native enzyme during the treatment with Triton X-100. Immunoblotting of the solubilized membrane fraction and of the soluble protein fraction clearly indicated that the flavoprotein subunit reacted with an antibody against the C. vinosum flavoprotein. It was present in the membrane fraction but not in the soluble fraction (Fig. 4B).

View larger version (48K): [in this window] [in a new window]   FIG. 4.   (A) SDS-PAGE profiles of FCSD from E. vacuolata. Lane 1, heme-stained cytochrome c-552; lane 2, Coomassie blue-stained flavoprotein and cytochrome c-552 subunits solubilized with Triton X-100 and partially purified as described in the text. (B) Localization of E. vacuolata FCSD in the membrane protein fraction. The Western blot was performed using C. vinosum flavoprotein subunit antibodies. Lanes: 1, E. vacuolata soluble protein fraction; 2, E. vacuolata membrane protein fraction; 3, C. vinosum FCSD (1 µg of native protein).

Flanking genes. Two possible open reading frames were found adjacent to the E. vacuolata FCSD genes. Upstream, in the same orientation, there is the 3' end of a gene for a 132-residue-long protein which by BLAST search appears to be homologous to the htrB gene of E. coli. The HtrB gene codes for lauroyl acyltransferase involved in the biosynthesis of the outer membrane lipid A (13). It is separated from the FCSD cytochrome subunit gene by 558 bases. The HtrB gene product is also a heat shock protein required for cell viability at high temperature in E. coli and is present in Haemophilus influenzae (29) as well. Separated by 236 bases downstream of the flavoprotein subunit gene, and in the opposite orientation, there is a large open reading frame (orf4) encoding a protein of at least 567 amino acid residues. It is unclear where it begins, but the initiator codon seems to be a TTG triplet starting at base 4313, 9 bases downstream of a possible ribosome binding site (GGAG). A BLAST search shows that it has a large number of homologs. The strongest similarity is to four proteins derived from Synechocystis (SLR359, SLL267, SLR1305, and SLR2077) (23), with more than 40% identity. The functional roles of these proteins are unknown, but the orf4 product is also related to a lesser extent to diguanylate cyclases and phosphodiesterases. These proteins show the greatest conservation in the C-terminal 400 residues, which contain characteristic GGDEF and EAL motifs (44). It is remarkable that a gene similar to orf4 is located downstream of R. capsulatus SQR in the same opposing orientation as in E. vacuolata. This context is quite different from that of the FCSD gene in Chromatium, where it was found that the flavoprotein gene is separated from the cytochrome gene by only 15 nucleotides (15, 35). Upstream, a tetraheme cytochrome and a homolog of ankyrin were found. The tetraheme cytochrome is part of a multigene locus in Thiosphaera pantotropha, H. influenzae, and Alcaligenes eutrophus (8, 17, 42) which contains several electron transfer proteins involved in nitrate reduction. Ankyrin serves to bind proteins together and/or to bind them to the membrane (9). Downstream of the Chromatium FCSD gene, there are no genes located in the 446 bases that were sequenced. Thus, the FCSD gene is probably not part of a multigene operon in Chromatium and E. vacuolata because of the lack of conservation of flanking genes and the small sizes of the transcripts (see below).

Influence of sulfur compounds on the growth of E. vacuolata. FCSD transcription and expression in E. vacuolata was studied by examining growth in different media. The best growth was obtained under photoheterotrophic conditions in DSMZ medium combining acetate as a carbon source and sulfide as an electron donor. These results are in agreement with the report of Zakharchuk and Ivanovskii (53), who described increased assimilation of ¹⁴C-acetate in E. shaposhnikovii cells in the presence of thiosulfate, sulfide, and bicarbonate. To study the influence of carbon sources on photosynthetic activity, cells were grown on minimal medium supplemented with acetate, succinate, or malate, all of which supported growth. Removal of sulfide, however, reduced growth compared to that in a photomixotrophic medium, as did replacing acetate with sodium malate or succinate. RNA was extracted from these cells and analyzed by Northern blotting with probes specific for both the heme and the flavoprotein subunits of FCSD. The size and the hybridization pattern were the same with both probes. The transcript of about 3 kb corresponds to an operon that contains the cytochrome and flavoprotein genes but no others. The Chromatium FCSD transcript is also just large enough to include the two subunit genes (15). The intensity of the RNA transcripts increased in cells grown on sulfide or thiosulfate plus CO₂ or on sulfide plus acetate. In view of the fact that some sulfide can be produced from thiosulfate (10, 11), we conclude that sulfide induces FCSD expression (Fig. 5). In the absence of sulfide or thiosulfate, FCSD expression is significantly reduced, confirming the role of sulfide for FCSD induction under anoxygenic photosynthesis.

View larger version (54K): [in this window] [in a new window]   FIG. 5.   Northern blot of RNA isolated from E. vacuolata. Total RNA was isolated from a 100-ml culture growing exponentially in minimal Pfennig medium supplemented with the following compounds: malate (lane 1), sodium thiosulfate (lane 2), sodium sulfide (lane 3), sodium sulfide and acetate (lane 4), acetate (lane 5), and succinate (lane 6). Fifty micrograms of RNA was loaded per lane. Northern hybridization was performed with a digoxigenin-labeled probe amplified with the fcsd gene template. Note the elevated levels of FCSD transcript in the cultures containing sulfide.

Functional importance of conserved residues in FCSD proteins. Several residues in the flavoprotein were identified from the three-dimensional structure as having possible functional importance (12). The FAD is covalently bound to Cys 42 (Chromatium numbering), and this residue is conserved in E. vacuolata. Glu 167 is conserved in nearly all the proteins of the glutathione reductase family of enzymes (37), including the Chromatium and E. vacuolata FCSDs, and is located near the N5 position of the FAD. In FCSD, Glu 167 modulates the reactivity of the FAD with sulfite when ionized at pH 6 and presumably has a role to play in the oxidation of sulfide. The disulfide Cys 161-Cys 337 is adjacent to the FAD; it is conserved and also modulates the reactivity of the FAD with sulfite when it is cleaved by sulfite above pH 8.5. Incidentally, this disulfide is also present in SQR (38), suggesting that it may be essential for the oxidation of sulfide. Trp 128, Tyr 306, and Trp 391 are all near the flavin, and any one of them could act as the charge transfer acceptor for the flavin-sulfite adduct. Trp 128 and Trp 391 are conserved, but Tyr 306 is replaced by His in E. vacuolata. This should affect the redox potential of the FAD in a pH-dependent manner because of its location at the positive N-terminal end of the helix. Thus, FAD may have a higher redox potential at low pH when the His is protonated, provided that the charge on the helix does not prevent His protonation within the physiological range of pH. A higher potential should make the FAD more reactive with sulfite and other nucleophiles.