(Received for publication, August 9, 1996, and in revised form, January 27, 1997)
From the Universität Regensburg, Lehrstuhl
für Zellbiologie und Pflanzenphysiologie, 93040 Regensburg,
Germany, § Institute of Horticulture, The Volcani Center,
Bet-Dagan 50250, Israel, and ¶ The Hebrew University of Jerusalem,
Division of Microbial and Molecular Ecology, Life Sciences,
Jerusalem 91904, Israel
A sulfide-quinone oxidoreductase (SQR, EC
1.8.5..) has been purified to homogeneity from chromatophores of the
non-sulfur purple bacterium Rhodobacter capsulatus DSM 155. It is composed of a single polypeptide with an apparent molecular mass
of about 55 kDa, exhibiting absorption and fluorescence spectra typical for a flavoprotein and similar to the SQR from the cyanobacterium Oscillatoria limnetica.
From N-terminal and tryptic peptide sequences of the pure protein a
genomic DNA clone was obtained by polymerase chain reaction amplification. Its sequence contains an open reading frame of 1275 base
pairs (EMBL nucleotide sequence data base, accession no. X97478[GenBank])
encoding the SQR of R. capsulatus. The deduced polypeptide
consists of 425 amino acid residues with a molecular mass of 47 kDa and
a net charge of +9. The high similarity (72%)/identity (48%) between
the N termini of the cyanobacterial and the bacterial enzyme was
confirmed and extended. Both enzymes exhibit the FAD/NAD(P) binding
-fold (Wierenga, R. K., Terpstra, P., and Hol, W. G. S. (1986)
J. Mol. Biol. 187, 101-107). The complete sequence of the
SQR from R. capsulatus shows further similarity to
flavoproteins, in particular glutathione reductase and lipoamide
dehydrogenase. The cloned sqr was expressed in
Escherichia coli in a functional form.
Anoxygenic photosynthesis with sulfide serving as an electron donor is a property unique to prokaryotes. In cyanobacteria this property has been demonstrated and most extensively studied in Oscillatoria limnetica, which is capable of both oxygenic and sulfide-oxidizing, anoxygenic photosynthesis (1). Anoxygenic photosynthesis in this organism is induced by sulfide, and the inducible factor was characterized as a sulfide-quinone reductase (SQR)1 by Arieli et al. (2). It was isolated recently as a single 57-kDa polypeptide, and shown to be a flavoprotein (3).
Constitutive sulfide photooxidation by anoxygenic photosynthetic bacteria has long been known and extensively reviewed (4, 5). The sulfide oxidizing entity in Chlorobiaceae and Chromatiaceae has been considered to be a flavocytochrome c (6). However, not all sulfide photooxidizing species contain this component, and alternatives have been suggested (4). In fact, SQR activity (7) and the involvement of the cytochrome bc complex (8) has recently been described for Chlorobium.
Rhodospirillaceae were comparatively little studied with respect to sulfide oxidation and even grouped as Athiorhodaceae. However, their ability to grow with sulfide as an electron donor has been noticed (9). Based on inhibition by quinone analogs, the pathway of sulfide oxidation in Rhodobacter sulfidophilus was proposed to involve the quinone pool (4, 10).
Rhodobacter capsulatus tolerates sulfide up to 2 mM and utilizes it as an electron donor (9). Recently we have demonstrated SQR activity in chromatophores of R. capsulatus and shown that sulfide oxidation is connected to the reduction of the cytochrome bc1 complex via the ubiquinone pool (11). Furthermore we have purified the SQR from R. capsulatus (12, 13), and the N-terminal amino acid sequences of this enzyme and that of O. limnetica were found highly homologous (3, 13). Remarkably, all the expected fingerprint residues of the FAD binding domain (14, 15) are conserved in the SQR of R. capsulatus and O. limnetica (3, 13). We have therefore proposed that SQR is a universal flavoenzyme that plays a major role in sulfide-dependent, anoxygenic photosynthesis.
Whereas the genetic system of O. limnetica has not yet been studied, that of R. capsulatus is well established. Thus cloning of the sqr gene from R. capsulatus will provide an experimental system to apply molecular genetics to the study of SQR.
In the present work we describe the purification to homogeneity and the characterization of SQR from the purple photosynthetic bacterium R. capsulatus DSM 155. DNA primers based on sequences of the N terminus and tryptic peptides of the pure SQR allowed the cloning of the sqr gene. This novel gene was functionally expressed in Escherichia coli.
R. capsulatus DSM 155 cultures were grown photoheterotrophically in 20-liter screw cap bottles containing RCV medium (16), supplemented with 150 µM Na2S at 25 °C under 10,000 lux.
Purification of the SQR ProteinChromatophores were prepared and their bacteriochlorophyll a content was measured as described previously (17). Frozen chromatophores resuspended (0.8 mg of bacteriochlorophyll/ml) in a buffer containing 50 mM glycylglycine, pH 7.0, and 20 mM EDTA were homogenized, and Thesit (Boehringer Mannheim, Germany) was added to a final concentration of 10 mM. After stirring on ice for 30 min in the dark, the mixture was centrifuged for 2 h at 200,000 × g. 40 ml of the supernatant were applied on a DEAE-cellulose column (2 × 20 cm), equilibrated with a buffer containing 10 mM bis-Tris-HCl, pH 6.5, 2 mM EDTA, and 0.4 mM Thesit (BITE). After washing, elution was performed with a linear gradient from 0 to 0.4 M NaCl in BITE. The activity of the fractions was determined. and the most active fractions were pooled and concentrated with Centricon-100 microconcentrators (Amicon, Beverly, MA). 2 ml of the concentrate were loaded on a 23-ml density gradient from 10 to 40% sucrose in BITE and centrifuged for 14 h at 120,000 × g (Beckmann Ti-60). After fractionation, the most active fractions were again combined, concentrated, and applied on a fast protein liquid chromatography-Mono-Q column (HR5/5, Pharmacia, Uppsala, Sweden) equilibrated with BITE. SQR was eluted by 40 ml of a linear gradient from 0 to 0.4 M NaCl in BITE. The activity of the fractions was determined, and after SDS-PAGE the pure fractions were combined.
Spectroscopic AssaysSQR activity was determined as sulfide-dependent decyl-UQ (Sigma) reduction in a spectrophotometer (Kontron Uvikon 860), recorded at 275 nm under N2 atmosphere using a millimolar extinction coefficient of 15 (18). The reaction mixture routinely contained 10 mM bis-Tris-HCl, pH 6.5, 20 µM decyl-UQ. The reaction was started by addition of Na2S to 40 µM. Fluorescence spectra were measured in a Spex Fluoromax spectrofluorimeter (SPEX Industries Inc., Edison, NJ) at room temperature under air. Sulfide concentration was measured by the methylene blue method (19), and protein was determined with BCA protein assay from Pierce.
SDS-PAGE and Western BlottingSDS-polyacrylamide gel electrophoresis was carried out according to Laemmli (20) and stained with silver (ICN Rapid-Ag-Stain, ICN, Irving, CA) or Coomassie. Western blotting was carried out as described in Towbin et al. (21). Rabbit antibodies against purified SQR were obtained from Eurogentec (Seraing, Belgium). Antibodies were purified by binding to blotted SQR and removed at acidic pH (22). For detection the BM-Chemiluminescence kit was used (Boehringer Mannheim, Germany).
Tryptic Digestion and Peptide SequencingThe Coomassie-stained 55-kDa band was cut out of the SDS-polyacrylamide gel, ground, and washed three times with 50% methanol, 10% acetic acid. After incubation in 90% acetic acid, the slices were lyophilized. Digestion was performed overnight at 37 °C in 300 µl of 0.2 M NH4HCO3 containing 20 µg of tosylphenylalanyl chloromethyl ketone-treated trypsin (Merck, Darmstadt, Germany). The gel slices were spun down and washed three times with the same buffer. The supernatants were pooled and lyophilized. The peptides were dissolved in 0.1% trifluoroacetic acid and separated by reversed phase high pressure liquid chromatography chromatography (ERC system, Alteglofsheim, Germany, column: Octadecyl-Si 60 10 µm, 46 × 250 mm, Serva, Heidelberg, Germany) at 53 °C with a gradient from 0 to 70% acetonitrile in trifluoroacetic acid. Peptide sequencing was performed with an Applied Biosystems gas phase sequenator by R. Deutzmann, Universität Regensburg, Germany.
Polymerase Chain Reaction, DNA Purification, and SequencingOligonucleotides were obtained from MWG-BIOTECH (Ebersberg, Germany). Polymerase chain reaction was performed with the apparatus and kit from Perkin-Elmer Co. Plasmid DNA was purified using standard procedures (23). Genomic DNA was isolated as described previously (24). For constructing a genomic library of R. capsulatus DNA, PstI fragments between 2.5 and 5.5 kb were cloned into the PstI site of pUC19 (25). Screening was performed using the DIG luminescence detection kit from Boehringer (Mannheim, Germany). For DNA sequencing the T7SequencingTM kit (Pharmacia, Uppsala, Sweden) was used. Sequence analysis were performed with UWGCG version 7.3 (26).
Expression in E. coliFor expression of the SQR in E. coli the NdeI/BamHI fragment was excised
from the pUC19 derivative denoted pUSQR, and inserted into the
NdeI/BamHI site of pT7-7 (27), yielding pTSQR.
E. coli BL21(DE3) (28) was transformed with pTSQR, and the
expression was induced at an A600 of 0.8-1 with
40 µM isopropyl-1-thio--D-galctopyranoside for 4 h. For SDS-PAGE analysis 0.5-ml cultures were centrifuged, resuspended in 100 µl of 2% SDS gel loading buffer, and after boiling for 5 min 15 µl of the samples were loaded onto the gel. To
test the activity, 50 ml of induced cells were centrifuged, washed with
50 mM sodium phosphate buffer, pH 8, and resuspended in 1 ml of the same buffer. After sonication (8 cycles of 15 s sonication (40 W Sonifier B-12, Branson Sonic Unix Co., Danbury, CT)
followed by 30 s of cooling on ice) and centrifugation for 10 min
at 10,000 × g, the supernatant was centrifuged 30 min
at 200,000 × g. The SQR activity was assayed in the
low speed supernatant containing membranes and soluble proteins
(100%), high speed supernatant, and the resuspended membranes.
To isolate the SQR from chromatophores, various solubilization methods were compared by determining the SQR activity in the extract after ultracentrifugation (100% = activity of the untreated chromatophores). Attempts to solubilize the SQR with 25 mM dodecylmaltoside yielded only 10% (12). Solubilization with 0.5% cholate yielded about 45% of the SQR activity, but it was unstable. NaBr treatment extracted around 36% (12). Sonication (Sonifier B-12, Branson) of the chromatophores (40 W, 15 s, eight times with intervals of 30 s at 4 °C) resulted in 30% loss of the total activity, and after the ultracentrifugation step less than 5% of the residual activity remained in the supernatant. Sonication followed by NaBr treatment extracted around 45% of the total activity. Although the activity in this extract was stable, dialysis or further purification with ammonium-sulfate precipitation, hydrophobic chromatography (phenyl-Sepharose), or gel filtration chromatography inactivated most of the enzyme. The most effective procedure was solubilization with 10 mM Thesit at a bacteriochlorophyll a concentration of 0.8 mg/ml. Thesit is a nonionic detergent of the polyoxyethylene type similar to Triton X-100, but without aromatic residues, which would have interfered with the test for SQR activity. Around 45% of the total activity of the chromatophores was found in the supernatant (Table I) and remained stable for at least 2 days at 4 °C. Further purification as described under "Materials and Methods" was very similar to that of the SQR of O. limnetica (3) and resulted in a total purification factor of only 35 (Table I). The partial loss of activity during the purification procedure is most probably due to the removal of FAD, the prosthetic group of SQR.
|
The size of the purified single polypeptide was found to be 55 kDa (in SDS-PAGE), very similar to the cyanobacterial SQR (57 kDa) (3). However, in contrast to the O. limnetica SQR, which is tightly membrane bound, the partial solubilization by NaBr suggests that the bacterial SQR is loosely bound to the membrane. At high concentration NaBr acts as a chaotropic agent removing peripheral protein complexes bound by hydrophobic interactions, similar to the F1 part of the F0F1 ATP-synthase (29).
Nearly all SQR activity was retained in the concentrate after filtration with Centricon-100. Filtration of the NaBr extract (which does not contain detergent) gave the same results. This may reflect aggregation of the SQR protein. A homooligomeric organization similar to that of the glutathione reductase (30) is suggestive, since as described below SQR is homologous to this enzyme.
The apparent Km of the purified SQR for sulfide, as
well as for decyl-UQ was about 2 µM (Fig.
1). The Km for sulfide was similar to
the values obtained for the reduction of exogeneous decyl-UQ (5 µM) or cytochromes (3 µM) in chromatophores of R. capsulatus (11) as well as for the sulfide dependent
growth of this bacterium (9). The Km values of
R. capsulatus SQR for both substrates are lower than those
of the cyanobacterial enzyme, which were 8 and 31 µM for sulfide and plastoquinone-1, respectively (3).
The optimum pH for the electron transfer from sulfide to decyl-UQ was
around pH 6.3. This pH optimum could merely reflect a preference of
H2S over HS as a substrate for the enzyme.
With a pK around 7 the H2S:HS
ratio at pH 6.3 is about 5:1. The activity drops very drastically below pH 6.
Similar to the cyanobacterial SQR the bacterial enzyme is a
flavoprotein. Fig. 2 shows the fluorescence excitation
and emission spectra of the purified SQR. The excitation maxima of the
oxidized enzyme in the visible light are at 375 and 458 nm, the maximum of the emission is at 518 nm, typical to many flavoproteins and flavins
(31). The additional excitation peak around 280 nm might reflect energy
transfer between the flavin and aromatic amino acids in its vicinity.
The reduction of the enzyme by sulfide (160 µM) has led
to almost complete loss of the fluorescence (Fig. 2). Thus, the SQR of
R. capsulatus is a flavoprotein which fluoresces in the
oxidized state. The identification of the SQR as a flavoprotein was
further supported by the primary structure, which harbors the FAD
binding domain (see below).
Cloning and Sequencing of the sqr Gene
The N terminus and
additional four partial peptides of the purified SQR were sequenced.
Polymerase chain reaction-amplified probes obtained with
oligonucleotides deduced from the peptide sequences were used to clone
a 3.6-kb PstI fragment from a genomic DNA library of
R. capsulatus in which 1791 bases were sequenced (Fig.
3).
The amino acid sequence deduced from the open reading frame from bp 394 to 1668 contains the sequence of the N-terminal as well as the four
internal peptides of the SQR (Fig. 3). The GC content of the clone
(63%) was found to be lower than that characterizing the genomic DNA
of R. capsulatus (65.5-66.8%) (32), although the codon
usage is similar (33). From bp 384 to 388 a putative ribosome
binding site (GGAGG) was found, matching well with the consensus
sequence determined from 43 potential translation initiation sequences
of R. capsulatus genes (34). The sequence from bp 319 to 324 (TTGACA) is identical to the 35 region of the E. coli-like
70 promoter consensus sequence (34). Downstream there
are two hexanucleotides with a score of 4 compared with an E. coli
10 sequence, one (CATATT) 11 and the other (TATATG) 22 nucleotides distant from the
35 region. However, these distances
differ from that given for E. coli-like
70
promoter regions (15-19 nucleotides). Neither additional ribosome binding site nor potential promoter sequences were found further upstream. A sequence encoding a transcript of a hairpin-like structure, containing the stop codon (bp 1669) occurs at the end of the
sqr gene without a subsequent poly(T) region (Fig. 3).
The SQR polypeptide deduced from the sequence consists of 425 amino
acids with a molecular mass of 47 kDa. The theoretical isoelectric
point is 9.7 with a net charge of +9. This might cause the shift up to
55 kDa in the apparent molecular weight observed in SDS-PAGE. Structure
prediction analysis gives 36% -helices and 23%
-sheets.
Hydropathy analysis by Husar, Heidelberg Unix Sequence Analysis
Resources 2.1 with the program "PepPlot" using "Kyte and
Doolittle's Hydropathy" and the "Chou and Fasmann Alpha and Beta
Propensities" as well as "Peptidestucture" Hydrophilicity according to "Kyte-Doolittle and Hopp-Woods method," did not
indicate any membrane spanning or anchoring
-helix.
The first amino acid of the mature SQR, which was identified by the peptide sequencing, is alanine encoded by GCT (bp 397-399). Thus, either the preceding methionine or an N-terminal leader peptide of 16 amino acid residues including the next methionine is cleaved off (Fig. 3). Prokaryotic cleavable leader sequences generally have one or more basic residues near the N terminus followed by a core region of about 7-13 hydrophobic residues preceding to 5-7 residues of higher polarity, with an alanine at the site of cleavage from the mature protein, and another alanine upstream the cleavage site (35). Indeed, there are two arginine residues at the N terminus of the putative leader peptide, but the other features are missing (Fig. 3). Although the real transcription and translation start site has not been experimentally determined, the absence of a putative ribosome binding site upstream bp 348 and the lack of a characteristic leader sequence suggest that the ATG bp 394-396 is the first translated codon.
The N terminus of the SQR from R. capsulatus is highly
homologous to that of O. limnetica (Fig.
4A) (3, 13). This is confirmed now and
extended up to position 30, as shown in Fig. 4A (48%
identity (14/29) and 72% similarity (21/29)). Both enzymes contain all
expected fingerprint residues of the ADP-binding site characterizing
many NAD(P)/FAD containing proteins (14, 15). According to the rule
(14) fingerprints with a score of 10 with one deviating uncharged amino
acid most likely fold as ADP-binding -unit. The R. capsulatus SQR shows a score of 10 out of 11 within the N-terminal
36 amino acids. The deviating amino acid residue is Thr34.
Secondary structure analysis for both SQR enzymes predicts a
-fold consistent with the one proposed for ADP binding, with the exception that the loop is positioned between the first
-sheet and the
-helix. Probably the two adjacent glycine residues in the
bacterial enzyme and the proline in the cyanobacterial enzyme function
as helix breakers. Unlike the NAD(P)-binding domains, the FAD binding
domains of many flavoenzymes, including the SQR, appear to be located
close to the N terminus (14, 15).
Two additional segments of the bacterial enzyme are suggested to be involved in FAD binding (Fig. 4, B and C), since they contain motives involved in the binding of the FAD in two enzymes of known three-dimensional structure, glutathione reductase (36, 37) and lipoamide dehydrogenase (38). Remarkably, within the first region (Fig. 4B) the overall homology is around 64%, although there are only three positions conserved: a glycine, an alanine and a hydroxyl residue, serine or threonine. As shown in the three-dimensional structure, the carbonyl O atom of the conserved alanine (Ala155) in glutathione reductase is in contact with the phosphate group bound to the ribityl chain of FAD via solvent molecules (36). Higher homology was found to other pyridine nucleotide oxidoreductases (Fig. 4B).
Within the second motive (Fig. 4C) in glutathione reductase
Gly330 is conserved because of a turn in the
C-chain to accommodate the PO4 groups of FAD
(36). The conserved Asp331 at the end of a
-sheet forms
hydrogen bonds with the O-3 group of the ribityl chain of the flavin
moiety in FAD and with both phosphate groups. The carbonyl O of
Ile329 and the amide N of residue 332 interact via the same
solvent molecule with the phosphate group near the ribityl chain. The hydrophobic residues at position 326, 327, and 329 are probably conserved to form a hydrophobic environment (36). In the SQR from
R. capsulatus, with the exception of the highly conserved aspartate (Asp331 in glutathione reductase), all other
consensus features of this region are conserved, suggesting the
existence of a hydrophobic
-sheet in a similar distance to the
-fold as in pyridine nucleotide-disulfide reductases.
Taken together, these sequence similarities suggest that the flavin of
the SQR is a FAD which is bound to the protein moiety as in pyridine
nucleotide-disulfide oxidoreductases. However, in the SQR protein there
is no acidic residue at the end of this putative -sheet. In line
with the fact that pyridine nucleotides are not substrates for SQR, the
additional
-fold involved in binding NAD(P) in pyridine
nucleotide oxidoreductases is missing in SQR.
Except for the three regions involved in FAD binding, no other homologies were detected for SQR in the EMBL Data base. Amino acid residues involved in quinone binding, for example in cytochrome bc complexes (41) or in quinone-type photosynthetic reaction centers (42), are not clustered. Therefore, it is impossible to identify residues involved in the quinone binding site of SQR by comparing primary structures.
Expression in E. coliThe SQR was expressed in an active form in E. coli BL21(DE3). After induction, a dominant band around 55 kDa (in SDS-PAGE), which reacted with purified antibodies against the R. capsulatus SQR, appeared only in extracts obtained from cells containing the sqr gene (data not shown). Furthermore these extracts were active in reducing decyl-UQ with sulfide. The activity was lost upon heating of the preparation. Surprisingly a substantial portion of the total activity (30-50%) was found in the soluble fraction (specific activity of 4 µmol of decyl-UQ reduced/mg of protein × min).
This heterologous expression indicates that the FAD and, if present, additional prosthetic groups are incorporated into the SQR in the correct way by E. coli. The relatively high proportion of soluble SQR obtained from E. coli supports the conclusion that in R. capsulatus SQR is a peripheral protein. It is tempting to suggest that it is bound to the membrane of R. capsulatus via a membrane associated factor that is absent in E. coli. This putative factor, however, is not required for the reduction of decyl-UQ by sulfide.
The SQR of R. capsulatus functionally expressed in E. coli now allows for the isolation of larger quantities of the enzyme for more detailed structural and functional studies. In particular the question of another cofactor/prosthetic group is being pursued.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X97478[GenBank].