Studies of the Cytochrome Subunits of Menaquinone:Cytochromec Reductase (bc Complex) of Bacillus subtilis

The menaquinone:cytochrome creductase, or bc complex, of Bacillus subtilisbelongs to a third class of bc-type complex, distinct from the bc 1 andb 6 f classes. Using a mutagenesis approach, we demonstrate that the cytochrome b (QcrB) andc (QcrC) subunits of the complex give rise to bands at 22 and 29 kDa, respectively, after denaturing electrophoresis; that both subunits are required for proper complex assembly and/or stability; and that both subunits retain one heme molecule under denaturing conditions. This unusual property of a b-type cytochrome was investigated further. We present evidence for the existence of a covalent linkage between the polypeptide and hemeb H and of an important role for Cys43 in binding of heme b H. It is proposed that heme is also covalently attached to the cytochromeb subunit of b 6 fcomplexes of chloroplasts and cyanobacteria.

The cytochrome bc 1 complex (quinol:cytochrome c oxidoreductase) is an integral membrane protein complex that functions as part of an electron transfer chain by passing electrons from quinol in the membrane to a c-type cytochrome. Coupled to electron transfer is the transport of protons across the membrane, and consequently, the enzyme contributes to the proton motive force. The complex has been isolated from mitochondria and several bacteria, and a similar complex, called the b 6 f complex, has been isolated from plant chloroplasts and from cyanobacteria. Recently, crystal structures of the soluble part of the Rieske protein (1) and of the intact complex (2) from bovine heart mitochondria have been solved.
There is considerable variation, depending on the source, in the number of subunits making up the complex, but three subunits are always present: an iron-sulfur (Rieske) protein containing a high potential [2Fe-2S] cluster with cysteine and histidine ligation; a cytochrome b containing two low spin btype hemes, b L and b H , both with bis-histidine ligation but with different electrochemical properties; and a cytochrome c containing one c-type heme with histidine/methionine axial ligation in the case of c 1 and histidine/tyrosine (␣-amino group) in the case of f (3). The cytochrome b of bc 1 complexes consists of approximately 400 amino acid residues, arranged as eight transmembrane helices, while the cytochrome b of b 6 f complexes may be regarded as being split into two subunits: one of ϳ220 residues, containing four transmembrane helices, which house the two b-type hemes; and the other, called subunit IV, of ϳ160 residues, which contains three transmembrane helices.
Relatively little is known about the bc-type complexes of Gram-positive bacteria. Only one complex, that from the thermophilic bacterium Bacillus sp. PS3, has been purified to date (4). Genes (called qcrABC) encoding bc-type complexes have recently been identified in Bacillus subtilis (5) and Bacillus stearothermophilus (6). Their predicted protein products appear to constitute a distinct, third class of complex, which bears greater similarity to b 6 f than to bc 1 complexes. The predicted 224-amino acid residue cytochrome b subunit (QcrB) is similar to that of b 6 f complexes. However, the third subunit of the complex, QcrC, can be regarded as a fusion protein consisting of subunit IV and a cytochrome c. The latter bears little similarity to either cytochrome c 1 or cytochrome f but, rather, resembles small Bacillus c-type cytochromes.
Cytochromes of c-type are distinct from other cytochromes in that the heme molecule is covalently attached, generally via two thioether linkages, to the polypeptide. SDS-PAGE 1 of Bacillus subtilis membrane preparations from cells grown in the presence of the heme biosynthetic precursor, 5-aminolevulinic acid, 14 C-labeled, has previously been used to identify membrane-bound c-type cytochromes because noncovalently attached heme (presumably, all heme except heme c) is lost from its protein ligands under the denaturing conditions of the gel (7). Such an analysis of wild-type B. subtilis reveals four major radioactive bands at 36, 29, 22, and 16 kDa, respectively. Two of the four bands have been identified unambiguously; the 36-kDa band is due to subunit II of the cytochrome c oxidase, caa 3 , and the 16-kDa band is due to cytochrome c 550 (8). The 16-kDa c 550 band masks an additional small cytochrome c, which has been recently identified. 2 The two remaining bands, of 29 and 22 kDa, are associated with the qcr operon, since deletion of the promoter region results in the loss of both bands (5). However, it is unclear why the qcr operon, which encodes one c-type cytochrome, should give rise to two bands. It was proposed that the 29-kDa band may be QcrC (predicted mass ϭ 28 kDa) and that the 22-kDa band might be a degradation product of QcrC. Studies using radioactively labeled heme in place of aminolevulinic acid confirmed that both the 22 and 29 kDa bands contain heme and not an alternative prosthetic group for which aminolevulinic acid is a precursor (9). Here we describe an investigation of the identities of the bands, using a combination of qcr gene deletion, insertion, and site-directed mutagenesis in B. subtilis, which shows that the 29-and 22-kDa bands correspond to the cytochrome c (QcrC) and cytochrome b (QcrB) subunits of the bc complex, respectively. Normally, b-type heme dissociates from its ligating polypeptide under denaturing conditions, but the case of QcrB is clearly an exception. Resistance to extraction by acidified acetone suggests that one heme of QcrB is covalently linked to the polypeptide. We present evidence from site-directed mutagenesis studies of QcrB that indicates that Cys 43 is involved in binding heme b H ; and we propose that heme may also be covalently bound to cytochrome b subunits of b 6 f-type complexes.

EXPERIMENTAL PROCEDURES
Strains and Growth Media-The bacterial strains and plasmids used are presented in Table I. The B. subtilis 168 strain used in this study was found to be oligosporogenic. We have observed that c-type cytochromes in sporulating strains undergo much greater proteolytic degradation than those of strains defective in sporulation. Escherichia coli strains were grown at 37°C in LB broth (10) or on LA plates consisting of LB with 0.9% (w/v) agar, or on TBAB (Difco) plates. B. subtilis strains were grown at 37°C in NSMP (11) or on TBAB plates. TBAB plates containing 1.5% (w/v) starch were used to test strains for amylase production. Antibiotics, where appropriate, were added to liquid media and plates at the following concentrations: chloramphenicol, 4 mg/liter Genetic Techniques-General molecular genetics techniques were used as described in Sambrook et al. (10). Plasmid DNA was isolated using Bio-Rad or Promega Wizard miniprep kits. Chromosomal DNA from B. subtilis strains was isolated as described by Marmur (12). E. coli strains were transformed by electroporation (13) using a Bio-Rad Gene pulser. B. subtilis-competent cells were prepared and transformed essentially as described previously (14). DNA sequence analysis was carried out using the dideoxynucleotide termination method (15), with Sequenase II (Stratagene/Amersham Pharmacia Biotech) and [␣- 35

S]ATP (Amersham Pharmacia Biotech).
Construction of Plasmids pPP503, pPP502, and pPP491-Plasmid pPP503 ( Fig. 1) was constructed as follows. The 2.8-kb XbaI-ClaI fragment of plasmid pPP499 containing qcrABC was ligated into similarly digested pDH88. A 63-base pair XbaI-NotI fragment of the resulting plasmid was removed, and the plasmid was recircularized by ligation after the ends had been made blunt by Klenow enzyme treatment. A 4.4-kb EcoRI-BamHI fragment of the plasmid containing qcrABC under P spac was then ligated into similarly digested pVK47, a derivative of pDH32 lacking a 2.6-kb PvuII fragment of lacZ, resulting in pPP503.
Plasmid pPP503 was digested with HindIII, which results in the vector part and two fragments. The 1.6-kb qcrAB fragment was religated to the vector part, resulting in pPP502 in which the qcrAB genes have the same orientation as in pPP503.
Plasmid pPP491 (used to construct LUH58 and LUH59) was obtained by ligating a 1.8-kb DraI fragment from pPP453 containing qcrC to HindIII-digested and Klenow enzyme-treated pVK48. The junctions between the insert and vector were confirmed by DNA sequence analysis. The plasmid pVK48 was kindly provided by Dr. V. K. Chary (Temple University School of Medicine); it was constructed by cloning a 1.7-kb EcoRI-BamHI fragment of plasmid pDG148 (containing the P spac promoter, the multiple cloning cassette, and the lacI gene) (16) into an EcoRI-BamHI-digested pVK47.
Mutagenesis was carried out on the pALTER-1 derivative pPP511. The desired mutations were confirmed by DNA sequence analysis using custom primers. Plasmids for insertion of mutant qcrB variants under the P spac promoter at the amyE locus on the B. subtilis chromosome were obtained by digesting the mutant pPP511 plasmids with HindIII and ligating the 1.6-kb qcrAB fragment into HindIII-digested pPP502.
Heme-specific Labeling-Growth of strains, heme-specific labeling, and preparation of membranes were carried out as described previously (5). Briefly, strains were grown overnight at 37°C in 25 ml of NSMP supplemented with 2 M 5-[4-14 C]aminolevulinic acid (51 mCi/mmol). Cells were collected by centrifugation and treated with lysosyme, and the particulate fraction was collected, washed, and suspended in 20 mM MOPS buffer, pH 7.4. The particulate fraction was incubated at 40°C for 30 min or at 100°C for 5 min in the presence of SDS. SDS-PAGE and subsequent treatment of the gel was carried out as described previously except that the gel was not treated with salicylic acid prior to drying (5). The dried gel was exposed to a storage phosphor screen for 1 week, and the screen was scanned using a PhosphorImager SI (Molecular Dynamics). The resulting digitized autoradiogram was analyzed using the program ImageQuant (Molecular Dynamics).
Heme Extraction-Isolated B. subtilis membranes (75 l, protein concentration approximately 14 mg/ml) were precipitated by the addition of 1 ml of ice cold acetone and mixing. The suspension was centrifuged at 12,000 rpm for 10 min at 4°C, and the supernatant was removed. The pellet was twice extracted with 1 ml of ice-cold acidified acetone (24 mM in HCl) and subsequently twice more with ice-cold acetone, also at 4°C. Finally, the pellet was resuspended in 20 l of 40 mM Tris borate buffer containing 3.5 mM SDS, pH 8.64.

Analysis of qcrB and qcrC Mutants-Mutants of strain 168
containing an insertion in qcrB (LUH51) and a deletion of qcrC (LUH52), respectively, were constructed and grown in the presence of 5-[4-14 C]aminolevulinic acid. SDS-PAGE of resulting membrane preparations followed by autoradiography revealed that both mutants lacked the 29-kDa band. LUH51 also lacked the 22-kDa band, while LUH52 contained a faint band at 22 kDa (autoradiogram not shown). The latter result is consistent with the assignment of the 29-kDa band as QcrC, while the absence of the 29-kDa band from the qcrB mutant, LUH51, might arise from a polar effect on transcription of qcrC resulting from the insertion of the neo gene cassette into qcrB. To investigate this further, two additional mutants were constructed. These strains, LUH58 and LUH59, contained the same qcrB and qcrC mutations as LUH51 and LUH52, respectively, and in addition an intact copy of qcrC under the control of the inducible P spac promoter, inserted elsewhere on the chromosome, at the amyE locus. The strains were grown in the presence of 5- (17), so we relied upon data available from studies of bc 1 -type cytochrome b proteins but considered also the possibility that the replacement of residues important for heme binding and subunit folding in the bc 1 -type protein might affect the b 6 f-type protein differently. The mutations selected were as follows: Gly 45 3 Asp, Arg 91 3 Gln, His 94 3 Asp, Arg 111 3 Gln, and Trp 126 3 Ala. Three of the residues, Gly 45 , His 94 , and Trp 126 , are invariant, and the remaining two are highly conserved, among cytochrome b sequences (18). Gly 45 is believed to form part of the heme b H pocket (19) but is also proposed to be important for packing of heme b L , while Arg 111 and Trp 126 are predicted to be close to heme b H at the cytoplasmic boundary (20). His 94 serves as a ligand to heme b L (19), and Arg 91 is predicted to be lying close to it (18).
The mutated qcrB genes were constructed in E. coli and subsequently moved into B. subtilis strain LUH61 containing a neo gene cassette inserted into qcrB, with expression of qcrC directed from the neo promoter. The mutated qcrB genes were inserted at the amyE locus on the chromosome under the control of the P spac promoter. Strains were grown in the presence of 5-[4-14 C]aminolevulinic acid, and membranes were prepared and analyzed by SDS-PAGE. The resulting autoradiogram is shown in Fig. 3. Lanes 2 and 3 result from the control strain LUH62, which contained an inactivated qcrB gene in the qcr operon and functional qcrB inserted at amyE under the control of the P spac promoter, grown in the absence and presence of IPTG, respectively. In the absence of the inducer, neither band is observed, consistent with our earlier results. In the presence of the inducer, bands at 29 and 22 kDa are observed, although with an intensity significantly reduced from that observed in wild-type strains. The reduction in intensity is considered to be a consequence of the requirement for the presence of both QcrB and QcrC in order that either subunit be observed, together with transcription of either qcrB or qcrC occurring at a reduced level compared with the wild-type situation. Data presented above indicate that transcription from the P spac promoter is at least as efficient as that from the native qcr promoter (Fig. 2,  lane 3), so we conclude that transcription of qcrC from the neo promoter is less efficient than in the wild-type situation, and consequently the concentration of stable complex is reduced.
Four of the five strains with amino acid substitutions in QcrB (grown in the presence of IPTG) lack both 29-and 22-kDa bands (lanes 5-7 and 9), indicating a lack of bc complex. We conclude that the substitutions Gly 45 3 Asp (LUH64), Arg 91 3 Gln (LUH65), His 94 3 Asp (LUH66), and Trp 126 3 Ala (LUH68) all result in the inability of the QcrB protein to fold properly, which in turn affects the assembly and/or stability of the complex. The mutant containing the substitution Arg 111 3 Gln (LUH67) gives rise to a single band at 29 kDa (lane 8). In this case, the substitution probably affects the affinity of QcrB for heme but does not significantly affect the assembly or stability of the complex, and consequently, the other cyto-  (37). Approximately 150 g of protein was loaded in each lane. After electrophoresis, the gel was incubated in methanol/acetic acid. An autoradiogram of the dried gel is shown. Lanes 1 and 2, membrane preparations from strain LUH58 grown in the presence (1 mM) and absence of IPTG, respectively. Lanes 3 and 4, membrane preparations from strain LUH59 grown in the presence (1 mM) and absence of IPTG, respectively. The radioactivity running at the gel front is due to dissociated heme. chrome subunit is observed. Thus, we conclude that the 22-kDa band corresponds to QcrB and the 29-kDa band corresponds to QcrC.
The 29-and 22-kDa bands resulting from preparations of membranes from strain 168, in all cases (Figs. 3 and 4), showed an intensity ratio close to 1:1, with a variation of Ͻ15%, as determined using the program ImageQuant (Molecular Dynamics) to analyze digital scans of storage phosphor screens. This observation and the presence of one covalently bound heme in QcrC suggest that one of the two hemes of QcrB is tightly bound to the protein. The substitution Arg 111 3 Gln alters the environment predicted to be close to heme b H and causes loss of heme from QcrB, indicating that b H is likely to be the heme that remains bound in the wild-type protein. Attempts to alter the heme b L binding pocket resulted in the absence of both QcrB and QcrC, suggesting that maintenance of the heme b L binding pocket is essential for the proper folding and stability of the complex. Similar conclusions were drawn from mutagenesis studies of residues in the heme b L pocket of the bc 1 complex of Rhodobacter sphaeroides (19).
Heme Attached to QcrB Survives Extraction with Acidified Acetone-In b-type cytochromes, heme is bound to the protein via noncovalent interactions, which are usually easily disrupted when the protein is denatured. That heme remains associated with QcrB under denaturing conditions indicates that heme may be covalently attached to QcrB. To investigate this possibility further, membranes from B. subtilis strain 168 with 14 C-labeled heme were treated with acidified acetone, according to the procedure described by Rieske (21) for the extraction of noncovalently bound heme from cytochromes. The autoradiogram resulting from SDS-PAGE analysis of extracted membranes is shown in Fig. 4. Lanes 2 and 3 correspond to treatment with acetone and acidified acetone, respectively. The lanes differ only at the gel front, where the band due to dissociated heme is significantly reduced in intensity after acidified acetone extraction. Clearly, the intensity of the band due to QcrB is unaffected relative to QcrC, consistent with the covalent attachment of heme to QcrB.
Investigation of the Role of Cys 43 in Heme Binding in QcrB-Tight binding of heme to QcrB is also found in some other bc-type complexes. Studies of cytochrome b 6 f complex from pea and spinach chloroplasts indicated, by use of peroxidase activity as a stain for heme, that heme remains associated with the b 6 subunit, even under denaturing conditions (22,23). A similar observation was made from SDS-PAGE analysis of purified bc complex from Bacillus sp. PS3 (which is similar to B. stearothermophilus) (4). However, early studies of a bacterial bc 1 complex by SDS-PAGE analysis showed, also through peroxidase activity staining, that the heme groups of cytochrome b are very easily released from the protein in the presence of ␤-mercaptoethanol (24).
The bc/b 6 f complexes of Bacillus species, cyanobacteria, and chloroplasts are similar in that they contain a short cytochrome b subunit relative to that of mitochondrial and bacterial bc 1 complexes. An alignment of cytochrome b subunit N-terminal sequences from bovine heart mitochondria, Saccharomyces cerevisiae mitochondria, R. sphaeroides, Synechocystis sp. PCC6803, and B. subtilis is shown in Fig. 5A. Several residues in this region are invariant among the sequences including Gly 45 , Arg 91 , and Trp 126 (B. subtilis QcrB numbering). B. subtilis QcrB contains a single cysteine residue (Cys 43 ), and it is noted that this residue is invariant in the bc/b 6 f-type cytochrome b polypeptides but is not conserved among the bc 1 -type cytochrome b polypeptides. Reference to structural models of cytochrome b (25,26) and to the crystal structures of bovine (2) and chicken heart 3 mitochondria bc 1 complex indicates that Cys 43 of the Bacillus protein lies at the N terminus of helix A, close to the b H binding pocket (Fig. 5B). It is possible that Cys 43 of B. subtilis QcrB may lie close enough to b H to form a covalent attachment to it. To investigate further whether Cys 43 is involved in heme binding in QcrB, a Cys 43 3 Ser variant protein was generated. SDS-PAGE analysis of membranes from strain LUH63, containing this variant QcrB, grown in the presence of 5-[4-14 C]aminolevulinic acid and subsequent autoradiography reveals a low intensity band at 29, but not at 22, kDa (Fig. 3,  lane 4). Thus, Cys 43 is required for QcrB to retain heme under denaturing conditions.

DISCUSSION
The work described here completes the assignment of all heme-containing protein bands observed after SDS-PAGE analysis of membrane preparations from B. subtilis. Data from deletion studies and the site-directed QcrB variants Arg 111 3 Gln and Cys 43 3 Ser show clearly that the bands at 29 and 22 kDa belong to the QcrC and QcrB subunits of the bc complex, respectively.
Relatively little is known about the synthesis and assembly of bc 1 /b 6 f complexes. A tentative model for maturation of bacterial bc 1 complexes was recently proposed, in which cytochromes b and c 1 form a protease-resistant primary complex to which the Rieske iron-sulfur protein then associates (27). In B. subtilis, both QcrB and QcrC are required in order that either subunit be observed at normal intensity, indicating that the presence of both subunits is essential for the correct folding and 3 E. A. Berry, personal communication. assembly of the complex. Lane 4 of Fig. 2, arising from LUH59 (a qcrC deletion mutant), contains a faint band at 22 kDa, indicating a small amount of holo-QcrB. In the absence of QcrB, however, no trace of QcrC is detected (Fig. 2, lanes 1 and 2). These findings suggest that QcrB can be inserted into the membrane and bind heme to some degree in the absence of QcrC but that in the absence of QcrB no holo-QcrC is formed.
The substitutions Gly 45 3 Asp, Arg 91 3 Gln, His 94 3 Asp, and Trp 126 3 Ala did not result in a detectable complex, and we conclude that these substitutions result in a QcrB protein that is unable to fold correctly. Gly 45 is one of several invariant glycine residues that are believed to be important for the structure of cytochrome b and particularly for the packing of the two hemes. Replacement of the equivalent glycine residue of R. sphaeroides (Gly 48 ) and S. cerevisiae (Gly 33 ) with aspartate also resulted in the failure of the complex to assemble (26,28). His 94 is one of the four invariant histidine residues that serve as ligands to the two b-type hemes. Mutagenesis studies of R. sphaeroides cytochrome b showed that this residue (His 97 in R. sphaeroides) ligates heme b L (19), a conclusion that was recently confirmed by the x-ray structure of the bc 1 complex from bovine heart (2). Mutagenesis studies also indicated that assembly and stability of the Rhodobacter protein is more sensi-tive to changes in the heme ligands of b L than to those of b H . An assembled complex deficient in heme b H could be obtained, but one deficient in heme b L could not (19). Data presented here indicate that the structure/assembly of B. subtilis QcrB is similarly sensitive to changes in the b L heme pocket, since neither Arg 91 3 Gln nor His 94 3 Asp gave a detectable complex. Arg 111 and Trp 126 are predicted to be near the cytoplasmic surface of the QcrB subunit, close to heme b H (20). The equivalent residues (Arg 114 and Trp 129 ) of R. sphaeroides are not absolutely essential for assembly/stability of the complex or for binding of heme b H , although Arg 114 could be replaced only with another positively charged residue, i.e. lysine. The data reported here indicate that the b H site in the cytochrome b subunit of B. subtilis must be somewhat different from that of R. sphaeroides. The location of a positive charge at position 111 was not found to be essential for assembly/stability of the B. subtilis bc complex, since the replacement Arg 111 3 Gln resulted in an assembled complex, albeit with altered b H binding properties. It is not known whether heme b H is bound to this variant enzyme complex in vivo and subsequently lost during electrophoresis or whether it fails to insert at all. The Trp 126 3 Ala substitution did not lead to assembled complex, and we conclude that this residue is more critical for assembly/stability of the bc complex from B. subtilis than it is for the bc 1 complex from R. sphaeroides.
Although it appears that proper assembly of QcrB will not survive substitutions in the region of heme b L , previous studies of bc 1 -type complexes indicate that this heme may be more easily lost than heme b H during purification (29,30). However, under denaturing conditions both hemes are readily lost from the cytochrome b subunit of bc 1 -type complexes. Here, we have demonstrated that QcrB from B. subtilis retains heme under these conditions. Acidified acetone extraction of noncovalently bound heme from membrane preparations of B. subtilis, prior to SDS-PAGE analysis, failed to affect the intensity of the QcrB band, consistent with the covalent attachment of heme to QcrB. Studies of the purified bc complex from Bacillus PS3 indicate that this is likely to be a common property of Bacillus QcrB subunits (4). In addition, studies indicate that Bacillus firmus OF4 has the same pattern of heme-containing bands as B. subtilis after electrophoresis of membrane preparations, although with some variation in apparent size (31). The ability to heme-stain cytochrome b of the b 6 f complex of chloroplasts after SDS-PAGE suggests that this may be a property of all b 6 f-type cytochrome b proteins (22,23). Mutagenesis studies reported here indicate that it is likely that heme b H , rather than b L , remains bound, and sequence alignment studies show further that there is a cysteine residue (Cys 43 ) lying close to the b H site, which is invariant among b 6 f-type QcrB proteins but not among bc 1 -type proteins. The substitution Cys 43 3 Ser abolished the ability of QcrB to retain heme under denaturing conditions, although the complex was still assembled. This could simply be due to an alteration of the b H heme binding pocket that prevents heme insertion. However, the replacement residue, serine, is of similar size to cysteine, has similar electrostatic properties, and would not be expected to cause a significant structural rearrangement. The equivalent residue in other cytochrome b proteins varies considerably, from asparagine in most mitochondrial proteins to isoleucine in Rhodobacter to threonine in Bradyrhizobium (18), suggesting that it is not essential for protein folding. Indeed, substitution of the equivalent residue of S. cerevisiae cytochrome b, Asn 31 ( Fig. 5A) with lysine did not affect the heme binding properties of the cytochrome b subunit but affected the sensitivity of the complex to inhibitors of electron transport that bind at the quinone reduction (Q i or Q N ) site (32). Thus, the data are consistent with a direct role for Cys 43 in heme b H binding to B. subtilis QcrB.
Recently, several examples of unusual covalent bonds within metalloproteins have been reported, including the covalent attachment of heme to tyrosine (33) and to lysine residues (34). Therefore, although we have shown that Cys 43 is important for heme b H binding, it is possible that a residue other than Cys 43 is the site of covalent attachment. If covalent attachment between Cys 43 and heme b H does occur, it is not clear whether such a bond is of functional importance or what the nature of the attachment is. Cysteine residues of c-type cytochromes form thioether bonds with the heme vinyl side chains. However, these cysteines occur in the conserved motif Cys-X-X-Cys-His, which is not present in QcrB. There are examples of c-type hemes that are attached by only one cysteine instead of the usual two (e.g. cytochrome c 1 of Euglena gracilis (35)), but in these cases a shorter Cys-His motif is found. What is certain, however, is that the covalent attachment is not formed by the general cellular machinery for cytochrome c biosynthesis; holo-QcrB is formed in a B. subtilis strain deleted for ccdA, a gene essential for the synthesis of c-type cytochromes (36).