Heterogeneous Rieske Proteins in the Cytochromeb 6 f Complex ofSynechocystis PCC6803?*

The completely sequenced genome of the cyanobacterium Synechocystis PCC6803 contains three open reading frames, petC1, petC2, and petC3,encoding putative Rieske iron-sulfur proteins. After heterologous overexpression, all three gene products have been characterized and shown to be Rieske proteins as typified by sequence analysis and EPR spectroscopy. Two of the overproduced proteins contained already incorporated iron-sulfur clusters, whereas the third one formed unstable aggregates, in which the FeS cluster had to be reconstituted after refolding of the denatured protein. Although EPR spectroscopy showed typical FeS signals for all Rieske proteins, an unusual low midpoint potential was revealed for PetC3 by EPR redox titration. Detailed characterization of Synechocystismembranes indicated that all three Rieske proteins are expressed under physiological conditions. Both for PetC1 and PetC3 the association with the thylakoid membrane was shown, and both could be identified, although in different amounts, in the isolated cytochromeb 6 f complex. The considerably lower redox potential determined for PetC3 indicates heterogeneous cytochromeb 6 f complexes inSynechocystis and suggests still to be established alternative electron transport routes.

The functional core of cytochrome bc complexes consists of three subunits, which are a b-type cytochrome, a c-type cytochrome, and the Rieske iron-sulfur protein (1). In contrast to the cytochrome bc 1 complex of mitochondria and bacteria with three subunits, cytochrome b, cytochrome c 1 , and the Rieske protein, the homologous cytochrome b 6 f complex of chloroplasts and cyanobacteria contains four core subunits named cytochrome b 6 , subunit IV, cytochrome f, and the Rieske protein (2).
The Rieske protein consists of an N-terminal transmembrane ␣-helix, which anchors the protein in the membrane (3,4), and a large soluble domain on the luminal side of the membrane. The three-dimensional structure of this luminal part has been solved both for the Rieske protein of a cytochrome b 6 f complex (5) and of a cytochrome bc 1 complex (6) to a resolution of 1.8 and 1.5 Å, respectively. The redox active [2Fe-2S] cluster, which is bound by the luminal domain, is ligated by two cysteine and two histidine residues with one iron being ligated by the two cysteines and the other by the two histidines. It is likely that this unusual coordination causes the high midpoint potential of about ϩ100 to ϩ400 mV. Besides these coordinating amino acids there are two additional cysteine residues, which form a disulfide bridge close to the ironsulfur cluster. In the amino acid sequence the coordinating amino acids as well as the additional cysteines are organized in two boxes with highly conserved sequences in all cytochrome bc complexes. It was shown that substitutions of single amino acids within or around these boxes affect the midpoint potential of the Rieske proteins (7) and that this midpoint potential can be predicted to some extent from the amino acid sequence. Proteins with the sequence CPCHGSXY of the C-terminal ironsulfur cluster-binding site (Box II) oxidize high potential quinones like ubi-and plastoquinone, and these proteins have a midpoint potential between ϩ250 and ϩ375 mV. In contrast, in organisms utilizing menaquinone, one or both of the lateral residues Ser and Tyr are replaced by nonhydroxylated amino acids resulting in a lowered midpoint potential of the Rieske protein. The midpoint potentials of these menaquinone-oxidizing Rieske proteins are in the range of ϩ100 to ϩ180 mV. Whereas Rieske cluster of cytochrome b 6 f complexes have been found to have midpoint potentials in the range of about ϩ270 to ϩ360 mV, in cytochrome bc 1 complexes also proteins with a lower midpoint potentials can be found.
Although heterologous overexpression of proteins is a routine method to produce large amounts of a protein, the overexpression of membrane proteins in Escherichia coli often leads to cell death due to the toxicity of the foreign protein (8). There are, however, reports on the successful overexpression of Rieske proteins. The heterologous overexpression of the Rieske protein from the cyanobacterium Nostoc PCC7906 resulted in copious quantities of the protein located in inclusion bodies (9). Also, the previously described overexpression of the Synechocystis Rieske protein PetC1 yielded predominantly a non-native form clustered in inclusion bodies (10). However, both studies agree that a small fraction of the overexpressed protein can be found in E. coli membranes, although without incorporated iron-sulfur cluster. A successful overexpression of a functional Rieske protein with incorporated iron-sulfur cluster was reported for two proteins from hyperthermophilic archaea (11,12), which was not achieved up to now with "mesophilic" Rieske proteins.
The completely sequenced genome of the cyanobacterium Synechocystis PCC6803 contains three open reading frames named sll1316 (petC1), slr1185 (petC2), and sll1182 (petC3) encoding for three putative Rieske iron-sulfur proteins. The characterization of the overexpressed proteins presented in this paper reveals that all three petC genes encode functional Rieske proteins. Although two proteins, PetC1 and PetC2, are rather similar, the encoded protein PetC3 displays different structural and functional properties. In this report the functional expression of the three petC genes in the cyanobacterium Synechocystis PCC6803 is shown, and at least two PetC proteins were found to be associated with the cytochrome b 6 f complex of this organism.

MATERIALS AND METHODS
Growth Conditions-Synechocystis PCC6803 wild type and mutant strains 1 were grown at 30°C in BG11 medium according to Ref. 13. Constant illumination was provided by white fluorescent lamps with a photon flux density of 40 mol photons m Ϫ2 s Ϫ1 .
E. coli strains were grown in broth or agar LB, supplemented by 100 g of ampicillin/ml or 30 g of chloramphenicol/ml where indicated. Strain DH5␣ was used for plasmid propagation and BL21(DE3) for heterologous overexpression of the Rieske proteins. DNA Techniques-Molecular cloning was carried out using standard techniques as described previously (14). Enzymes used for PCR and cloning were obtained from MBI Fermentas. PCR was carried out using a Bio-Rad thermocycler. DNA sequencing was performed by MWG Biotech (Ebersberg, Germany). For PCR, genomic Synechocystis DNA was prepared according to Ref. 15.
Overproduction of the Rieske Proteins-Heterologous overexpression of the three petC genes of Synechocystis was done by cloning them separately into the expression plasmid pRSET6a (16). The three petC genes were amplified by PCR using the primers NTPetC1 (gggaattccatatgacccagatttctggctcccc) and CTPetC1 (gagtcggatcccttaagcccaccagggatcttcgtcg) for the amplification of petC1, NTPetC2 (gggaattccatatggataacacacaggcgatcg) and CTPetC2 (gtacaagatctttaaacccaccagggtttctctcc) for the amplification of petC2, and NTPetC3 (gggaattccatatggtcaaacgacgcaagctaatttc) and CTPetC3 (gcgatccttaggctttgaccagaatgttgttgc) to amplify petC3. For cloning of the PCR-amplified full-length Rieske genes, a NdeI side was introduced at the 5Ј end of the gene including the translation start codon ATG, and a BamHI or BglII side was added at the 3Ј terminus of petC1/petC2 and petC3, respectively. The PCRamplified genes were cleaved with NdeI and BamHI/BglII and cloned into the appropriately restricted expression plasmid pRSET6a. After transformation of the expression plasmids into E. coli BL21(DE3), the expression cultures were grown to an A 600 ϭ 0.6 at 37°C in LB media containing 100 g of ampicillin/ml. Expression of the foreign genes was induced by addition of 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside. Cells were harvested after 4 h of incubation, resuspended in buffer PBS (150 mM NaCl, 2.5 mM KCl, 13 mM NaHPO 4 , 1.8 mM KH 2 PO 4 , pH 7.2), and broken by sonication. Soluble E. coli proteins and membranes were separated from insoluble components ("low speed pellet") by centrifugation at 10,000 ϫ g. E. coli membranes were isolated by centrifugation of the supernatant at 100,000 ϫ g ("high speed pellet").
Spectroscopic Methods-EPR spectra of the isolated cytochrome b 6 f complex and membranes were recorded on a Bruker EPR200 spectrometer; the sample temperature was kept at 15 K with a helium cryostat. Other parameters for the measurements include 6.3 milliwatts of microwave power, 10 G modulation amplitude, and 9.44-gigahertz microwave frequency. E. coli membranes with the incorporated iron-sulfur proteins were adjusted to pH 5.5 and a protein concentration of 40 mg/ml. EPR-monitored redox titrations were done as outlined previously (18).
Protein Analysis-Membranes of the cyanobacterium Synechocystis PCC 6803 were isolated as described previously (17), and the cytochrome b 6 f complex was extracted from the thylakoids by incubation with 1% (w/v) n-dodecyl-␤-D-maltoside for 30 min at room temperature. After centrifugation at 40,000 ϫ g, the cytochrome b 6 f complex was purified by two high performance liquid chromatography steps according to Wenk et al. 2 Characterization of the complex by SDS-PAGE and immunoblotting revealed the presence of cytochrome f (29 kDa), cytochrome b 6 (24 kDa), the Rieske protein (19 kDa), subunit IV (17 kDa), and several small subunits in the range of 3-8 kDa. Further characterization of the complex by gel filtration chromatography showed that the isolated complex was monomeric.
Components of Synechocystis PCC 6803 membranes or the purified cytochrome b 6 f complex were separated by 15% SDS-PAGE gels (19) and stained with Coomassie Blue or blotted on a polyvinylidene difluoride membrane. Rieske proteins were detected by immunoblotting using polyclonal antibodies raised against unique peptide fragments deduced from the three different petC genes (Fig. 1). The rabbit primary antibodies and the goat anti-rabbit IgG alkaline phosphatase conjugate (Sigma) were used at a dilution of 1:2000 and 1:30,000, respectively. To visualize the cross-reacting protein bands, the membranes were incubated with CSPD® (Roche Molecular Biochemicals).
Miscellaneous Techniques-Sequences of the Rieske proteins involved in respiratory or photosynthetic electron transport were retrieved from GenBank TM via PubMed (www.ncbi.nlm.nih.gov/PubMed/).
The program ClustalX (version 1.64b) (20) was used to calculate the alignments and the phylogenetic trees.
All gaps were reset before the beginning of the individual alignments of the two protein groups. For pairwise alignment parameters, the identity matrix was used. All other parameters remained as pre-set by the program. For multiple alignment parameters, the threshold for delaying divergent sequences was lowered to 10% in order to inactivate this option. The BLOSUM series was used. The "gap opening penalty" was increased to 15. The "gap separation distance" was set to 12. All other parameters remained as pre-set by the program. Positions containing gaps were excluded from the calculation of the phylogenetic trees.

RESULTS
Computational Analysis-Genome analysis of the completely sequenced cyanobacterium Synechocystis PCC6803 (21) revealed gene multiplications for at least 22 genes (22). Besides the already sequenced petC gene (23), two additional open reading frames were found, which encode for possible Riesketype proteins. The three DNA sequences of the genes are deposited in the genome data base Cyano Base (www.kazusa. or.jp/cyano/cyano.html) under accession numbers sll1316 (petC1), slr1185 (petC2), and sll1182 (petC3).
The deduced amino acid sequence of the PetC1 protein shows the highest homology to other known cytochrome b 6 f complex Rieske protein sequences. Also, its organization in an operon with the cytochrome f encoding gene petA, which is typical for cyanobacteria, argues for a relevant functional role of this encoded Rieske protein in the cytochrome b 6 f complex of Synechocystis PCC6803. Sequence alignment of the three encoded proteins ( Fig. 1) clearly shows a higher similarity of PetC1 with PetC2 (44% identity) than with PetC3 (34% identity). PetC1 and PetC2 are similar in size with a calculated molecular mass of about 19 kDa. Due to three deletions in the PetC3 protein sequence, its calculated molecular mass is only 14 kDa. However, all three amino acid sequences show the sequence motifs for binding of the [2Fe-2S] cluster (Box I and Box II in Fig. 1) with an identical number of amino acid residues in between. Although up to now no information is available on the three proteins, their sequences indicate binding sites for the typical Rieske FeS cluster.
Also, the N-terminal part of all three protein sequences consists of hydrophobic regions of about 20 amino acids, which are predicted to form a ␣-helical transmembrane anchor.
Production of Specific Peptide Antibodies-Short amino acid sequences of the individual Rieske proteins were selected for the production of specific peptide antibodies in rabbits (Fig. 1), and the raised antibodies were tested for their specificity. Each antibody cross-reacts with the individual Rieske protein containing the peptide sequence used for immunization. Whereas the anti-PetC1 and anti-PetC2 antibodies showed no crossreactivity with any of the other two Rieske proteins, the anti-PetC3 antibody also cross-reacts with the PetC1 protein. Because of the molecular mass difference between these two proteins, the identification of the cross-reacted protein was straightforward.
Heterologous Overexpression and Characterization of the Isolated Putative Rieske Proteins-To facilitate more detailed studies on the three isolated Synechocystis proteins, plasmids for the heterologous overexpression of the three full-length proteins in E. coli were constructed as described under "Materials and Methods." All three E. coli strains carrying the expression plasmids for PetC1, PetC2, and PetC3, respectively, show a predominantly expressed protein with a apparent molecular mass of 19 (Fig. 2, A and B) or 14 kDa (Fig. 2C); these masses agree well with the molecular weight derived from the amino acid sequences of the individual Rieske proteins. The different Rieske antibodies cross-reacted as expected with the overexpressed proteins from E. coli (Fig. 3), and they showed no cross-reactivity with fractions of E. coli that did not contain these overexpressing constructs. Whereas most of the heterologous protein was deposited in inclusion bodies (low speed pellet), a minor fraction of the overexpressed protein was found to be associated with E. coli membranes (high speed pellet).
Although the inclusion body fractions showed only a very weak Rieske-type EPR signal (results not shown), membranes isolated from cells in which PetC2 and PetC3 were overexpressed yielded a typical Rieske-type EPR signal as shown in Fig. 4, A and B, respectively. This signal was not observed with membranes from control cells (Fig. 4C). Interestingly, in the control membranes an EPR signal of unknown origin with a g value of 2.02 was detected, most likely caused by the presence of small amounts of copper. Besides this signal, membranes of the E. coli strains overexpressing PetC2 showed additional signals with g xyz ϭ not detected, 1.90, and 2.03 (Fig. 4A). In membranes of cells overexpressing PetC3, a Rieske-type EPR signal with a greater amplitude was detected, and the g z value overlaps considerably with the signal of the unknown origin. However, the detected g xyz values of 1.78, 1.90, and 2.02 (Fig.  4B) also indicate the presence of a typical Rieske iron-sulfur cluster. Because PetC1 could not be obtained in a native form with an incorporated iron-sulfur cluster, this protein was purified from inclusion bodies of E. coli cells overexpressing this protein. After solubilization by urea, the inclusion body fraction was further purified by anion exchange high pressure liquid chromatography as described previously (10). Incorporation of the iron-sulfur cluster into the refolded protein was achieved enzymatically using IscS, which is known to mediate the incorporation of iron-sulfur clusters, and the reconstituted protein was further characterized (10). The isolated PetC1 protein showed g values of g xyz ϭ not detected, 1.90, and 2.04, which are identical to the EPR characteristics of the isolated cytochrome b 6 f complex (10), whereas the g values of the isolated PetC2 and PetC3 proteins show slight deviations.
Determination of the Midpoint Potential of Three Synechocystis Rieske Proteins by EPR Spectroscopy-In contrast to PetC1 and PetC2, which were isolated in the fully reduced form after overexpression, PetC3 was almost completely oxidized under aerobic conditions when no reducing agents were added. This indicates a lower redox midpoint potential (E m ) of this protein in comparison with the other two proteins. Determination of the midpoint potentials of the two Rieske proteins PetC1 and PetC3 by redox titration (Fig. 5, A and B) yielded a high E m value of ϩ320 mV for PetC1, which is in good agreement with values of other Rieske proteins from cytochrome b 6 f and bc 1 complexes. Interestingly, the midpoint potential of the third Rieske protein, PetC3, was determined to ϩ135 mV (Fig. 5B), i.e. about 200 mV lower than the midpoint potential of the PetC1 protein of Synechocystis. This fact might explain the observed oxidized form of PetC3 during the EPR measurements under aerobic conditions in the absence of reductants. All attempts to determine the exact value of the midpoint potential of PetC2 by EPR failed. The purified protein was not stable after addition of the mediators, as already observed for other redox active proteins. 3 However, based on other methods the midpoint potential of PetC2 could be roughly estimated (see "Discussion").
Phylogenetic Analysis-Phylogenetic analysis of the three Rieske proteins showed that PetC1 is very closely related to most other cyanobacterial Rieske proteins (Fig. 6). Although PetC2 is more distant to most of the cyanobacterial proteins, it also clearly belongs to the group of cytochrome b 6 f complex Rieske proteins. In contrast, PetC3 shows no particular similarities to other cytochrome b 6 f complex Rieske proteins. This protein segregates into a heterogeneous group of proteins, which includes the Rieske proteins from Heliobacillus, Mycobacterium, and Streptomyces as well as the archaeal Rieske proteins (Fig. 6). This phylogenetic classification and the detected low midpoint potential argue against a functional role of this protein in the cyanobacterial cytochrome b 6 f complex.
Translational Analysis-Although the characterization of the heterologously overexpressed proteins showed that all three proteins are Rieske iron-sulfur proteins, the expression of all three gene products in Synechocystis had to be demon-strated. Polyclonal anti-peptide antibodies raised against unique regions of the proteins (Fig. 1) were used to test Synechocystis membranes for the presence of the three gene products. Fig. 7 shows an immunoblot of membranes from Synechocystis WT, 4 ⌬petC1, ⌬petC2, and ⌬petC3. The anti-PetC1serum reacted with a protein of 19 kDa in WT, ⌬petC2, and ⌬petC3 membranes. By using membranes from the ⌬petC1 strain no cross-reaction was observed, indicating the monospecificity of this antibody. The anti-PetC3-antibody reacted with proteins with an apparent molecular mass of 19 and 14 kDa caused by the cross-reaction with the PetC1 protein as outlined above. The presence of PetC1 degradation products could be excluded because (a) the PetC1 antibody failed to detect a protein with a molecular mass of 14 kDa, and (b) using membranes from the Synechocystis ⌬petC3 mutant 1 no crossreaction with a 14-kDa protein could be detected with the anti-PetC3-antibody (Fig. 7).
By using the anti-PetC2 antibody, no gene product could be detected in the isolated membranes of Synechocystis.
To determine whether PetC1 and PetC3 are integral subunits of the cytochrome b 6 f complex, immunoblots were done with the isolated cytochrome b 6 f complex from Synechocystis (Fig. 7). From these blots it is evident that the two different Rieske proteins are not only associated with Synechocystis membranes, but also with the isolated cytochrome b 6 f complex. When using an identical protein concentration for the immunoblot analysis the cross-reaction of the anti-PetC1 antibody appeared much faster than with the anti-PetC3 antibody. Because both peptide antibodies show a similar cross-reactivity with the overexpressed Rieske proteins, this result indicates a much lower content of PetC3 in our cytochrome b 6 f complex preparation than of PetC1. This is supported by the fact that the isolated cytochrome b 6 f complex showed, even in a silverstained polyacrylamide gel, no band at 14 kDa corresponding to PetC3, whereas all other subunits are present in a ratio of about 1:1 (Fig. 8).

Heterologous Overexpression of Functional Rieske Proteins-
Although E. coli is the most popular host for the heterologous overexpression of foreign proteins, the overexpression of membrane proteins is often toxic and can have lethal consequences (8). There are, however, several examples in which successful overexpression of functional membrane proteins in E. coli cell membranes has been achieved. Specifically, the functional overexpression of Rieske proteins from hyperthermophilic organisms with incorporated iron-sulfur cluster has been reported (11,12).
To our knowledge the overexpression of full-length Rieske proteins with incorporated iron-sulfur cluster from a mesophilic 3 S. Anemü ller, unpublished observations. 4 The abbreviation used is: WT, wild type. organism in E. coli has never been reported. Here we show the overexpression of PetC proteins from mesophilic organisms in their native form with incorporated iron-sulfur cluster. The fact that all three overexpressed PetC proteins were localized in inclusion bodies or cell membranes of E. coli but not in the soluble cell fraction argues for the existence of a membrane anchor in these proteins similar to other Rieske proteins (4).
Three Rieske Proteins Are Encoded in the Genome of Synechocystis-The three overexpressed proteins were characterized by EPR measurements and typified as Rieske proteins by the presence of the Rieske [2Fe-2S] cluster g y ϭ 1.90 value. Although the g values of PetC1 and PetC2 were almost identi-cal, the EPR spectrum of PetC3 differs slightly. The g z value is shifted to a higher magnetic field, and the g x value, which was not detectable for the other two Rieske proteins, could be easily recognized.
The midpoint potential determined by EPR-monitored redox titration for PetC1 (ϩ320 mV) is typical for cytochrome bc complex Rieske proteins from ubi-and plastoquinol-containing organisms that show a highly positive midpoint potential of around ϩ300 mV (12). In contrast, proteins with a low midpoint potential of about ϩ135 mV, as determined for PetC3, are not found in the cytochrome bc complexes of any of these organisms.
petC3 Encodes an Unusual Rieske Protein-The sequence alignment of the three Synechocystis Rieske proteins in Fig. 1 clearly showed the two cluster binding domains in petC3, which are typical for Rieske proteins. Although the so-called Box II is formed by the CPCHGSXY motif typical of plasto-or ubiquinolcontaining organisms, the sequence of PetC3 shows amino acid exchanges resulting in the sequence CPCHGAXF. The replacement of Ser 6 of the binding motif as well as the Tyr 8 3 Phe substitutions can also be found in some other organisms (12), with both single amino acid exchanges resulting in a lowered midpoint potential of the expressed protein. In Saccharomyces cerevisiae the Ser 6 3 Ala mutation lowered the midpoint potential of the protein from ϩ285 to ϩ155 mV and the mutation Tyr 8 3 Phe from ϩ285 to ϩ217 mV (7). After double substitution, the expressed protein showed a low E m value of ϩ105 mV, and the resulting sequence of the Box II was CPCHGZXF as in the sequence of the PetC3 protein. The midpoint potential of the mutated protein from S. cerevisiae correlates well with the determined E m ϭ ϩ135 mV of PetC3. In conclusion, besides the exact determination of the E m by spectroscopic methods, a low midpoint potential of PetC3 could be predicted by comparison with other PetC sequences and the characteristics of the encoded Rieske proteins. Heliobacillus mobilis also contains a   Rieske protein with a Box II sequence identical to the PetC3 protein (24) and with a similar midpoint potential of ϩ150 mV. Interestingly, all organisms with the indicated single amino acid exchanges as well as organisms with a double mutation like H. mobilis use menaquinol instead of ubi-or plastoquinol as electron donor for their cytochrome bc complex (12).
The fact that the amino acid sequence of the two boxes ligating the iron-sulfur cluster is identical in PetC1 and PetC2 argues for a midpoint potential of about ϩ300 mV for the PetC2 protein, which is in good agreement with the observed completely reduced PetC1 and PetC2 proteins under aerobic conditions and in contrast with the oxidized PetC3.
On the Presence and Function of Three Rieske Proteins-Immunoblotting analysis clearly showed the presence of at least two petC gene products, PetC1 and PetC3, associated with membranes of the cyanobacterium Synechocystis PCC6803. The presence of the two proteins is the first evidence for the functional expression of various Rieske proteins in cyanobacteria. Although the functional expression of PetC2 was not indicated by immunoblot analysis, preliminary results with a promoter probing vector showed that PetC2 is also expressed to a certain degree in Synechocystis. 5 This suggests that the anti-PetC2 antibody used in the immunoblot analysis showed an insufficient cross-reactivity.
Our investigations on the isolated cytochrome b 6 f complex of Synechocystis clearly showed that at least two Rieske genes are continuously expressed, and the resulting subunits are associated with this complex in vivo, although apparently in different amounts. Consequently, the purified cytochrome b 6 f complex from Synechocystis PCC6803 is heterogeneous with respect to its Rieske subunit. As the gene petC1 is organized with petA in one operon, it was likely that the encoded protein is the predominant form in the complex.
Mutagenesis experiments suggested that the gene encoding PetC2 cannot be completely deleted if the expression of PetC1 is strongly diminished, whereas the petC2 gene can easily be deleted without a resulting phenotype in the WT strain. 1 These results are in good agreement with the determined characteristics of the isolated proteins. Apparently, PetC1 and PetC2 not only show a high sequence homology but display similar midpoint potentials, suggesting that PetC2 should be able to replace PetC1. In contrast, other deletion experiments suggested that PetC3 is not able to replace the main Rieske protein PetC1. This is in agreement with the unusual characteristics of PetC3 reported here, especially with the low midpoint potential, which may not allow this protein to act as a Rieske protein in a PQ-oxidizing cytochrome b 6 f complex. However, in contrast to the petC2 gene, the deletion of petC3 resulted in a phenotype indicating a physiological function of the encoded protein in Synechocystis. 1 Role of the Midpoint Potential for the Cytochrome b 6 f Complex Activity-Up to now, no Rieske protein with a similar low midpoint redox potential as PetC3 has been reported for any cytochrome b 6 f complex. As shown in Ref. 7, the activity of the cytochrome bc 1 complex of S. cerevisiae decreases with decreasing midpoint potential of the Rieske protein. Nevertheless, even cells expressing mutant Rieske proteins with a single substitution (Ser 6 or Tyr 8 ), which lowered the midpoint potential to about ϩ200 mV, were able to grow on ethanol/glycerol medium, although with a 30 -40% lowered cytochrome bc 1 complex activity. These results argue for the general compatibility of this single amino acid exchange with ubiquinol as the electron donor for the Rieske protein. In contrast, a mutant strain with an Ser 6 3 Ala exchange in the Rieske protein resulting in an E m ϭ ϩ155 mV was unable to grow on ethanol/glycerol medium, i.e. a Rieske protein with such a low E m value seems not to be able to oxidize ubiquinol efficiently. Because the E m values of ubi-and plastoquinol are similar (25), it seems appropriate to assume similar data as those obtained with a bc 1 complex-containing organism also for the plastoquinol-oxidizing cytochrome b 6 f complex. Results obtained from the mutagenic studies with the S. cerevisiae Rieske protein argue against a possible role of PetC3 in a plastoquinol-oxidizing reaction. However, some bacteria have been reported to contain cytochrome bc complexes with Rieske proteins of low redox potentials, i.e. between ϩ100 and ϩ200 mV; these were found to use menaquinol with a low midpoint potential of about Ϫ60 mV instead of plastoquinol/ubiquinol as electron carrier (26). From this point of view and considering the fact that PetC3 has been identified in this report in highly purified cytochrome b 6 f complexes, a conceivable function of PetC3 seems to be part of a cytochrome b 6 f complex that oxidizes menaquinol instead of plastoquinol. Although the metabolism for naphthoquinol biosynthesis has already been investigated for Synechocystis and all enzymes involved have been identified (27), nothing is known yet about free naphthoquinone in this cyanobacterium. The present work suggests the existence of a naphthoquinol-oxidizing cytochrome b 6 f complex; its location and possible physiological function remains to be determined in future experiments.