Dynamic Features of a Heme Delivery System for Cytochrome c Maturation*

In Escherichia coli, heme is delivered to cytochrome c in a process involving eight proteins encoded by the ccmABCDEFGH operon. Heme is transferred to the periplasmic heme chaperone CcmE by CcmC and from there to apocytochrome c. The role of CcmC was investigated by random as well as site-directed mutagenesis. Important amino acids were all located in periplasmic domains of the CcmC protein that has six membrane-spanning helices. Besides the tryptophan-rich motif and two conserved histidines, new residues were identified as functionally important. Mutants G111S and H184Y had a clear defect in CcmC-CcmE interaction, did not transfer heme to CcmE, and lacked c-type cytochromes. Conversely, mutants D47N, R55P, and S176Y were affected neither in interaction with nor in delivery of heme to CcmE but produced less than 10% c-type cytochromes. A strain carrying a CcmCE fusion had a similar phenotype, suggesting that CcmC is important not only for heme transfer to CcmE but also for its delivery to cytochrome c. Co-immunoprecipitation of CcmC with CcmF was not detectable although CcmE co-precipitated individually with CcmC and CcmF. This contradicts the idea of CcmCEF supercomplex formation. Our results favor a model that predicts CcmE to shuttle between CcmC and CcmF for heme delivery.

In Escherichia coli, heme is delivered to cytochrome c in a process involving eight proteins encoded by the ccmABCDEFGH operon. Heme is transferred to the periplasmic heme chaperone CcmE by CcmC and from there to apocytochrome c. The role of CcmC was investigated by random as well as site-directed mutagenesis. Important amino acids were all located in periplasmic domains of the CcmC protein that has six membranespanning helices. Besides the tryptophan-rich motif and two conserved histidines, new residues were identified as functionally important. Mutants G111S and H184Y had a clear defect in CcmC-CcmE interaction, did not transfer heme to CcmE, and lacked c-type cytochromes. Conversely, mutants D47N, R55P, and S176Y were affected neither in interaction with nor in delivery of heme to CcmE but produced less than 10% c-type cytochromes. A strain carrying a CcmCE fusion had a similar phenotype, suggesting that CcmC is important not only for heme transfer to CcmE but also for its delivery to cytochrome c. Co-immunoprecipitation of CcmC with CcmF was not detectable although CcmE co-precipitated individually with CcmC and CcmF. This contradicts the idea of CcmCEF supercomplex formation. Our results favor a model that predicts CcmE to shuttle between CcmC and CcmF for heme delivery.
The covalent attachment of heme to the CXXCH signature motif of apocytochromes is a critical step during cytochrome c biosynthesis. The process of cytochrome c maturation involves the formation of two thioether bonds between the vinyl groups of heme and the cysteinyl residues of apocytochrome c. Three distinct systems for the biogenesis of c-type cytochromes have evolved in different classes of organisms, which are annotated as systems I, II, and III (reviewed in Refs. [1][2][3][4]). System I is present in ␣and ␥-proteobacteria, Deinococcus, archaea, and protozoal and plant mitochondria. System II prevails in Grampositive bacteria, cyanobacteria, some ␤-, ␦-, and ⑀-proteobacteria, and plant and algal chloroplasts. System III cytochrome c maturation operates in fungal and metazoal mitochondria. The ␥-proteobacterium Escherichia coli requires eight cytoplasmic membrane proteins encoded by the ccmABCDEFGH operon (5,6). These proteins catalyze the covalent attachment of heme to the conserved CXXCH motifs of apocytochromes c. In E. coli, c-type cytochromes are synthesized exclusively under anaerobic respiratory conditions, under which the expression of the ccm genes is also induced (7)(8)(9). One of the key steps in the maturation pathway is the transfer and covalent attachment of heme to the cytochrome c maturation-specific heme chaperone, CcmE (10); subsequently, heme is transferred from the holo-CcmE intermediate to apocytochrome c. It has been shown earlier that covalent binding of heme to an essential histidine (His 130 ) of apo-CcmE requires the activity of the CcmC maturation factor (11). CcmC is an integral membrane protein whose topology was derived from that of its homologues in Rhodobacter capsulatus and Pseudomonas fluorescens (12,13). E. coli CcmC shares 41% identity and 58% similarity with its R. capsulatus homologue HelC and 52% identity and 70% similarity with its P. fluorescens homologue. According to the experimentally established membrane topology of these proteins, CcmC contains six transmembrane domains, separated by two cytoplasmic and three periplasmic domains (14) (Fig. 1).
An E. coli ccmC in-frame deletion mutant is deficient in the production of c-type cytochromes (15), most likely because heme cannot be transferred to CcmE (11). Several ccmC mutants from different organisms have been described, which were not only affected in cytochrome c maturation but also exhibited other phenotypes. Particularly interesting was the analysis of P. fluorescens ccmC mutants, which showed that CcmC is required for the production of pyoverdines, a fluorescent siderophore (13,16). In Paracoccus denitrificans, a ccmC mutant in addition to deficiencies in cytochrome c maturation and siderophore production showed intolerance to rich medium (17). The connection between these processes is not known. In the pathogenic bacterium Legionella pneumophila, CcmC was found to be required for cytochrome c production, for growth in low iron conditions, and for at least some forms of intracellular infection of eukaryotic hosts (18).
CcmC contains a conserved tryptophan-rich signature motif (WGXWXWDXRLT, where represents an aromatic amino acid residue) (1,3,19,20) that resides in the second periplasmic domain (Fig. 1) and shares a high degree of similarity with the tryptophan-rich motif of CcmF, NrfE, and the CcsA homologues of type II cytochrome c maturation (21). In addition, two absolutely conserved histidines are present in the first and third periplasmic domains of CcmC. Mutational analysis of these motifs of the E. coli CcmC protein revealed that they are functionally important (22). The accumulation of hydrophobic residues within the tryptophan-rich motif was initially thought to provide a platform for the binding of heme with the two conserved histidines serving as axial heme ligands (12). Only recently it was shown that CcmC can bind heme and that the tryptophan-rich motif is involved in a direct interaction with CcmE (14). Apart from the amino acids in this motif, CcmC contains several other highly conserved residues, which have never been studied so far in regard to their role in heme binding and/or delivery during cytochrome c maturation. In this study, we asked whether additional domains or amino acids are required for the function of CcmC. This is important because CcmC undergoes protein-protein interaction with at least CcmE but most likely also with CcmD, a small cytoplasmoriented membrane protein (22), and perhaps also with the ABC transporter subunits CcmAB (23). Although CcmC seems to bind heme (14), no mutants with defects in heme binding have been described so far. If CcmC is involved in heme transport across the membrane, one might find residues with membrane internal or even cytoplasmic location that are necessary for heme binding. Here, we explored on the one hand a strategy to generate random ccmC mutants deficient in cytochrome c maturation. These were identified in a genetic screen based on the colony phenotype under anaerobic respiratory growth conditions, when E. coli synthesizes up to six different c-type cytochromes (24,25). Their expression depends on the availability of terminal electron acceptors such as nitrate, nitrite, and trimethylamine N-oxide (TMAO) 1 in the growth medium. Any block of cytochrome c maturation would affect the cells' ability to respire anaerobically on such terminal electron acceptors. Our screen allowed the identification of cytochrome c maturation-deficient mutants in a background strain that lacks Me 2 SO reductase (26). This approach to identify new and important residues in CcmC was unbiased, which is important because of the high number of highly conserved residues. On the other hand, we also introduced alanines at invariant and highly conserved positions that had neither been mutated before nor hit by the random mutagenesis approach.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Bacterial strains and plasmids used in this study are listed in Table I. Bacteria were grown aerobically in LB medium or anaerobically in minimal salts medium (25) supplemented with 0.4% glycerol, 40 mM fumarate, and 5 mM sodium nitrate as the terminal electron acceptor. Antibiotics were added at the following final concentrations: ampicillin, 200 g/ml; chloramphenicol, 25 g/ml. If necessary, cells were induced with 0.1% arabinose at midexponential growth phase.
Screening under Anaerobic Respiratory Growth on TMAO-EC28⌬dms cells carrying wild-type or mutant ccmC on a plasmid were screened on modified minimum salt medium containing 33 mM KH 2 PO 4 , 60 mM K 2 HPO 4 , 7.5 mM (NH) 2 SO 4 , 1.7 mM trisodium citrate, 0.4 mM MgSO 4 , 67.6 M Na 2 SeO 3 , 100 M Na 2 MoO 4 , 10% LB medium, 0.4% glycerol, and 10 mM TMAO as the terminal electron acceptor. Chloramphenicol was added at a final concentration of 25 g/ml. The plates were incubated for 48 h at 37°C inside an anaerobic jar under argon.
Construction of Plasmids and Site-directed Mutagenesis-E. coli strains DH5␣ (28) and XL1-Blue (29) were used as host for clonings. The plasmids pEC816, pEC817, pEC818, pEC819, pEC820, pEC828, and pEC829 expressing CcmC with the mutations W114A, R128A, L82A, W180A, W125I, and D126A, respectively, were constructed by QuikChange TM site-directed mutagenesis (Stratagene) using pEC486 as a template and the primers listed in Table II. pEC821 expressing a C-terminally truncated CcmC ⌬N223-K245 -His 6 was constructed by inverse PCR, using pEC486 as a template and the following set of divergent primers. The 5Ј primer contained the coding sequence for six histidines upstream of the stop codon and a KpnI restriction site, whereas the 3Ј primer contained a KpnI restriction site. The PCR product was digested with KpnI and self-ligated to obtain pEC821. pEC799 expressing CcmD and CcmE was constructed by amplifying a 680-bp NdeI-EcoRI by PCR fragment with the primers ccmDnNdeI and ccmEcEcoRI, using pEC86 as a template. This fragment was digested with NdeI and EcoRI and ligated into the 5.4-kb NdeI-and EcoRIdigested pISC2. pEC827 expressing hexahistidine-tagged CcmC along 1 The abbreviation used is: TMAO, trimethylamine N-oxide. with CcmD and CcmE was constructed by the QuikChange TM sitedirected mutagenesis (Stratagene) using pEC406 as a template. For the construction of a plasmid expressing CcmF, the ccmF gene was amplified by PCR using a forward primer containing the NdeI site and a reverse primer with an EcoRI site. A 2-kb NdeI-EcoRI fragment containing ccmF was then cloned behind the arabinose promoter of pISC2 (30), resulting in pEC825. Plasmid pEC826 expressing the CcmCE fusion protein was constructed by QuikChange TM site-directed mutagenesis with a primer pair that eliminates the ccmC stop codon in pEC409 by the addition of an adenine nucleotide. In plasmid pEC409, the ccmC stop codon overlaps with start codon of ccmE. The plasmids pEC803, pEC804, pEC805, pEC806, pEC807, pEC808, pEC812, pEC815, and pEC824 expressing CcmC with V177G, D47N, R55P, W119R, S176Y, L183P, D126G/H184Y, H184Y, and G111S, respectively, were obtained from the random mutagenesis as described below. Random Mutagenesis by Error-prone PCR-Mutations were introduced randomly into the 756-bp ccmC gene encoding a C-terminal histidine tag (-His 6 ). PCR was performed with 10 ng of DNA template and the forward and reverse primers pACYC184 1613-1629 and pA-CYC184 4011-3941, respectively. The reaction was carried out in the presence of 200 M dATP, dCTP, dTTP, 20 M dGTP, and 180 M dITP or in the presence of 200 M dATP, dGTP, dTTP, 20 M dCTP, and 180 M dITP for 25 cycles. The PCR products were digested with NcoI and BfmI and ligated into NcoI-and BfmI-digested pEC486. The ligated product was transformed into the EC28⌬dms strain.
Random Mutagenesis by Using a Repair-deficient Strain-Random mutagenesis of the ccmC gene was carried out in the Stratagene XL1-Red mutator strain (Stratagene). Plasmid pEC486 containing wild-type ccmC was transformed into E. coli XL1-Red. This strain lacks three key enzymes for DNA repair, which leads to a 5000-fold increase in the mutation rate. DNA extracted from approximately 1000 colonies was extracted and electroporated into the EC28⌬dms strain.
DNA Sequencing-All constructs and mutants derived from PCR products were confirmed by DNA sequence analysis (Microsynth, Balgach, Switzerland). In plasmid pEC820 and pEC829, spontaneous silent mutations were found at nucleotide positions 678 and 360 of ccmC, resulting in conservative L226L and G120G changes, respectively.
Cell Fractionation-For the preparation of subcellular fractions, bacterial cultures were grown anaerobically in the presence of nitrate. Periplasmic proteins obtained from 1200-ml cultures were treated with polymyxin-B sulfate as described previously (14). For the preparation of membrane proteins, 600 ml of anaerobically grown bacterial culture were harvested, and membrane proteins were extracted as described previously (14).
Biochemical Methods-Protein concentrations were determined using the Bradford assay (Bio-Rad) and bovine serum albumin as a standard. Holo-cytochrome c 550 and holo-CcmE formation was analyzed qualitatively by immunoblot and heme staining (22). Soluble cytochrome c (c 550 ) in the periplasmic fractions was quantified by absorption difference spectroscopy using a ⑀ 550 -536 nm of 23.2 mM Ϫ1 cm Ϫ1 (30). Immunoblot analysis of the His-tagged CcmC versions and of cytochrome c 550 -His 6 was performed using monoclonal tetra-His antibodies (Qiagen) at a dilution of 1:2000. Antibodies directed against a CcmE peptide (10) or a combination of the three CcmF peptides 410 PLVR-WGRDRPRKIRN 424 , 508 SVERDVRMKSGDSVD 522 , and 630 DPRY-RKRVSPQKTAPE 645 were used at a dilution of 1:5000 and 1:10,000, respectively. Signals were detected using goat anti-mouse IgG (for His tag detection) or goat anti-rabbit IgG (for detection of other antigens) conjugated to alkaline phosphatase (Bio-Rad) as secondary antibodies and disodium 3- Heme Binding Assays-150 g of total membrane proteins were solubilized at 4°C for 1 h in 1 ml of solubilization buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5% lauroyl sarcosine). After centrifugation, the supernatant was treated with hemin-agarose (Sigma) or aminododecyl-agarose (Sigma) as described previously for CcmC (14). Agarose beads were washed three times with solubilization buffer and twice with PBS to remove unbound material and then treated with 2ϫ SDS loading dye (containing 50 mM dithiothreitol) and incubated at room temperature for 20 min. The mixture was subjected to 15% SDS-PAGE. Proteins in the gel were visualized by immunoblot analysis.
Co-immunoprecipitation-0.5 mg of membrane proteins were solubilized and precipitated with 5 l of undiluted anti-CcmE, 5 l of undiluted anti-CcmF, or 10 l of anti-His tag antibodies according to the method described previously (14). The proteins were subjected to SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and probed with antiserum against the tetra-His tag, CcmE, or CcmF peptides as indicated.

RESULTS
Rationale of the Screen-In the ccmC in frame deletion mutant EC28⌬dms (⌬A63ϪG156; ⌬dmsABC::kan), the predicted trans-membrane helices II, III, and IV of CcmC together with the periplasmic domain II containing the tryptophan-rich motif are deleted. A ⌬ccmC⌬dmsABC double mutant was constructed by P1 transduction of a ⌬dmsABC::kan allele from a mutant kindly provided by Dr. J. Weiner. DmsABC constitutes an Me 2 SO reductase that catalyzes the reduction of Me 2 SO and TMAO under anaerobic conditions. The dms mutant cannot grow anaerobically on minimal medium with glycerol and Me 2 SO (26). However, the TMAO reductase allows E. coli to grow by anaerobic respiration with TMAO as a terminal electron acceptor. TMAO reductase is a hetrodimeric enzyme composed of TorA, a periplasmic reductase containing molybdenum as cofactor and TorC, a membrane-bound pentaheme c-type cytochrome. During anaerobic TMAO respiration, electrons are transferred from the menaquinone pool to TorC and then delivered to TorA, which in turn reduces TMAO to trimethylamine (31). TorC requires the cytochrome c maturation (ccm) system composed of the CcmABCDEFGH proteins to insert the hemes covalently. Any block of TorC maturation is expected to abolish anaerobic growth on TMAO, when the alternative pathway for electron transfer via Me 2 SO reductase is excluded. Hence, the ⌬ccmC⌬dms mutant had a strong anaerobic growth phenotype; in contrast to the wild type, which formed large red colonies on TMAO plates, the mutant grew poorly and was white. The remaining growth capability of this mutant may be due to yet another Me 2 SO reductase encoded by a second set of dmsABC genes that are expressed constitutively (32). When the mutant was transformed with a plasmid expressing ccmC-H 6 encoding wild-type, His-tagged CcmC, the cells were red and formed normal size colonies, but when an empty vector was used as control, they were small and white. This phenotype was the basis for a color screen for ccmC mutants.
Generation and Analysis of Random ccmC Mutants-Random mutations were introduced into plasmid-borne ccmC described above by (i) error-prone PCR and (ii) the use of a repair-deficient strain for plasmid propagation. Based on the red/white phenotypes, we screened 14,905 colonies. 241 white candidates were tested for expression of full-length CcmC polypeptide by immunoblot using anti-His tag antibodies. 35 mutants were found to produce normal levels of the protein. This step allowed the exclusion of all mutants in which stop codons had been introduced accidentally or that produced unstable protein. The plasmids were isolated and sequenced. We found 14 single, 10 double, and four triple mutants. The following single mutations were identified (Table III): D47N, R55P (twice), G111S (twice), W119R (twice), S176P, S176Y, V177G (three times), L183P, and H184Y. Despite the high number of redundant hits at individual amino acids from independent experiments, some residues that had been altered previously were not mutated, suggesting that we had not reached saturation. Redundant mutants were found for both mutagenesis procedures. Surprisingly, in seven cases a wildtype sequence was found along with a high plasmid copy number, as was evident from the increased amount of plasmid DNA and also from particularly strong signals in the immunoblot (data not shown). These mutants were generated from the repair-deficient strain. It is possible that the mutant phenotype imparted to the strains harboring high copy plasmids was due to a dominant negative effect exerted by overexpressed CcmC on cytochrome c maturation. Such mutants were not pursued for further analysis. However, when wild-type ccmC was overexpressed from the pACYC184 vector relative to the other chromosomally expressed ccm genes, a dominant negative effect was not observed (data not shown).
Eight single mutants (Table III) were selected for a detailed analysis; among them, all except one (H184Y) mapped to residues that had not been investigated previously. H184Y was included in our analysis for a comparison with the previously described mutant H184A (22). The seven newly identified, important amino acids Asp 47 , Arg 55 , Gly 111 , Trp 119 , Ser 176 , Val 177 , and Leu 183 were all located in periplasmic domains of CcmC (Fig. 1). To test the mutants for loss of cytochrome c maturation, they were grown anaerobically with nitrate as the sole electron acceptor. E. coli has multiple respiratory nitrate reductases (33)(34)(35), and one of them, the NarGHI enzyme, does not contain c-type cytochromes. Hence, the organism can facultatively produce holo-cytochrome c as part of an electron transport system to the periplasmic nitrate reductase, the NapABC system, and ccm mutants can grow without suffering from a selective pressure when cytochrome c maturation is blocked. This implies that formation of revertants or second site suppressors is not enforced. In our complementation experiments, we overexpressed exogenous Bradyrhizobium japonicum cytochrome c 550 to better monitor the effect of a mutation in CcmC. The ⌬ccmC⌬dmsABC mutant strain was co- transformed with a plasmid expressing from the vector promoter a His-tagged B. japonicum cytochrome c 550 plus a plasmid encoding either the His-tagged wild-type or mutant CcmC protein ( Fig. 2A). Cytochrome c formation was assayed by heme staining (Fig. 2A, upper panel) and immunoblot (lower panel) of periplasmic fractions. The apocytochrome is degraded rapidly when heme is not incorporated (30). Quantification of c-type cytochromes was done by absorption difference spectroscopy. None of the mutants produced wild-type levels of c-type  5, 7, 10, and  11). The H184Y phenotype matched that of the previously described mutant H184A (22). In contrast, in comparison with the D126G/H184Y double mutant, the single change of D126A in the tryptophan-rich motif had resulted in functional CcmC (22). The reduced minus oxidized spectra were recorded for the quantification of soluble type c cytochromes. The value for the wild type was assigned 100%, and relative values of the various ccmC mutants were calculated accordingly. Compared with the wild-type cytochrome c level (100%), mutants D47N, R55P, W119R, and V177G produced only 6.9, 6, 0.3, and 1.2% cytochrome c, respectively, whereas cytochrome c formed in mutant L183P was below the limit of detection. Since the random mutants affected cytochrome c maturation in varying degrees, it was necessary to investigate whether or not the mutations had (i) destabilized the protein in the membrane, (ii) affected CcmC-CcmE interaction, or (iii) affected the ability of CcmC to bind heme. The known function of CcmC is the delivery of heme to the heme chaperone CcmE, from where heme then is transferred in a second step to the apocytochrome. Hence, we isolated membranes from cells deleted in the entire ccm operon (EC06) and containing the plasmid pEC799 encoding CcmD and CcmE plus a plasmid expressing either the wild-type or a mutant ccmC allele. The levels of the CcmC (Fig. 2B, first panel) and CcmE polypeptides (Fig. 2B, second panel) were probed by immunoblot analysis and found to be the same for all strains; the control strain containing only CcmE but no CcmC (lane 2) showed no reaction. Holo-CcmE formation determined  (40). It predicts six transmembrane helices, four cytoplasmic domains, and three periplasmic domains. Conserved histidines and the Trp-rich motif are located in the periplasm. Amino acid residues characterized in this study are boxed. Black boxes represent amino acids for which a new phenotype was found (class II mutants). Gray boxes represent amino acid residues important for heme delivery to CcmE. Open boxes represent amino acid residues that were altered without an effect. Residues shown in boldface type are 100% conserved in CcmC homologues from the following species, derived from whole genome sequences deposited at NCBI: by heme stains (Fig. 2B, third panel) was normal in mutants D47N (lane 3) and R55P (lane 4), whereas mutants W119R, S176Y, V177G, and L183P (lanes 6 -9) contained less holo-CcmE than the wild type (lane 1). Mutants G111S (lane 5) and H184Y (lane 10) and the double mutant D126G/H184Y (lane 11) lacked holo-CcmE completely. We also investigated the effect of the mutations on the physical interaction between CcmC and CcmE (Fig. 2B, bottom panel). A co-immunoprecipitation experiment revealed that mutants D47N (lane 3) and R55P (lane 4) co-precipitated CcmC and CcmE similar to the wild type (lane 1). Mutants S176Y, V177G, and L183P (lanes 7-9) showed slightly reduced and mutant W119R drastically reduced levels of co-precipitated CcmE. For mutants G111S (lane 5) and H184Y (lane 10) and the double mutant D126G/ H184Y (lane 11), no CcmC-CcmE co-precipitation was found. We also tested the effect of the mutations on affinity binding to hemin-agarose, since it has been described previously (14), but no significant differences were observed (data not shown).
Analysis of ccmC Site-directed Mutants-CcmC contains a number of highly conserved residues, of which those localized in the core of the tryptophan-rich motif (Trp 119 -Gly 120 -Xaa-Xaa-Trp 123 -Xaa-Trp 125 -Asp 126 ) and the histidines at positions 60 and 184 have been altered previously (22). Here, we focused on additional conserved residues of the CcmC protein that had failed to be picked up in our random screen; the conserved amino acids Leu 82 and Tyr 139 in transmembrane segment II and IV, respectively, Trp 114 and Arg 128 in an extension of the Trp-rich motif, and Trp 180 in the third periplasmic domain were changed to alanine. In addition, the C-terminal 23 residues were deleted to test the role of this cytoplasmic domain. The ⌬ccmC⌬dmsABC mutant strain was complemented for cytochrome c biosynthesis in the presence of plasmid pRJ3290 encoding the B. japonicum cytochrome c 550 plus a plasmid encoding either the wild-type or a mutant ccmC allele, and cytochrome c formation was tested by heme staining, immunoblot (Fig. 3A), and absorption difference spectra (Table III). Most affected in cytochrome c maturation were the mutants R128A (lane 5) and W114A (lane 4) on either side of the Trprich motif, whereas the other mutants in residues predicted to be in transmembrane domains (lanes 3 and 6) were almost like the wild type. When the C-terminal 23 residues were deleted, a drastic effect on cytochrome c formation was observed (lane 8).
In the next experiment, the presence of CcmC and CcmE in the membrane, their interaction, and the efficiency of heme delivery to CcmE were investigated as described for the random mutants. The levels of the CcmC and CcmE polypeptides in the membrane and the formation of holo-CcmE in cells deleted in the entire ccm operon (EC06) and containing the plasmid pEC799 expressing ccmDE plus a plasmid with either the wild-type or a ccmC-His 6 mutant allele were determined. The results presented in Fig. 3B (first and second panels) show that all mutants produced normal levels of CcmC and CcmE except for the C-terminal CcmC deletion, in which the levels of CcmC but not CcmE were decreased. Although this cytoplasmic domain is not well conserved, it seems to be important for the stability of the protein. Mutants L82A (Fig. 3B, third panel The physical interaction between CcmC and CcmE was also investigated (Fig. 3B, bottom panel). None of the site-directed mutants except for the C-terminal deletion mutant was affected in the physical interaction between the two proteins. We also tried with a 2-fold higher concentration of the C-terminally truncated CcmC mutant protein to co-precipitate CcmE but failed to detect it (data not shown).
Next we investigated the effect of the mutations on the affinity of CcmC to hemin-agarose. No significant differences between the wild-type and the mutants were observed (data not shown). We also detected some heme binding of the truncated CcmC, but the signal was comparatively weak, which could be due to the instability of the protein. In summary, cytochrome c maturation of the site-directed mutants was not affected drastically in any case, and only the C-terminal deletion mutant had a clear deficiency. The site-directed mutant with the strongest effect on cytochrome c maturation (R128A) still produced more cytochrome c than the least affected random mutant (D47N). This could explain why mutations in the strongly conserved residues had not been picked up in our screen. Thus, our screen preferentially detects mutants severely affected in cytochrome c maturation. In fact, when the site-directed mutants were grown under the conditions of our screening procedure, none of them were white, confirming that they produced intermediate levels of c-type cytochromes.
Analysis of ccmC Mutants in the Tryptophan-rich Motif-The importance of the tryptophan-rich motif for CcmC function has been well documented. It is involved in the physical interaction between CcmC and CcmE (14). In previous work, mutants in residues Trp 119 , Trp 125 , and Asp 126 had affected the cytochrome c formation slightly, whereas mutants in Gly 120 , Thr 121 , Trp 122 , and Trp 123 had produced wild-type levels of cytochromes c (22). In this work, we were interested in investigating the individual contribution of the conserved amino acids in the extended Trp-rich motif to cytochrome c maturation. We thus included the new mutants W114A, W119R, and R128A and compared them with W125I and D126A in a detailed analysis. The mutants were used to complement cytochrome c biosynthesis in the ⌬ccmC⌬dmsABC strain (Fig. 4A). Whereas the changes W114A and W125I affected cytochrome c maturation only slightly, the R128A, D126A, and in particular W119R had more severe consequences, with W119R producing less than 1% of the c-type cytochromes seen in the wild type (Table III). The influence of the mutation on the presence of the CcmC and CcmE polypeptides in the membrane as well as the ability of the mutants to interact with CcmE and to deliver heme was investigated (Fig. 4B). We analyzed the levels of CcmC peptide, CcmE peptide, and holo-CcmE in cells deleted in the entire ccm operon (EC06) and containing the plasmid pEC799 expressing ccmDE plus a plasmid encoding either wild-type or a mutant CcmC-His 6 . The results (Fig. 4) show that all mutants produced normal levels of CcmC and CcmE. Mutants W119R and D126A (lanes 3 and 5, respectively) produced lower levels of holo-CcmE, although the polypeptide was synthesized normally, a finding that is in agreement with an earlier study (22). However, in a co-precipitation experiment, the interaction of CcmE with mutant W119R is clearly disturbed (Fig. 4B, bottom panel, lane 3).
Analysis of ccmC Mutants in the Absence of ccmD-In all of our experiments, holo-CcmE formation was investigated in the presence of CcmD. CcmD is a small monotopic membrane protein, which had been shown to influence heme ligation to apo-CcmE (22). In particular, it had been observed that the presence of CcmD can suppress a ccmC mutant phenotype prevailing in the absence of CcmD. Here, we investigated the ccmC mutants that had affected cytochrome c maturation drastically without apparently affecting holo-CcmE formation. We analyzed levels of CcmC polypeptide, CcmE polypeptide, and holo-CcmE in cells deleted in the entire ccm operon (strain EC06) and containing the plasmid pEC412 encoding CcmE plus a plasmid encoding either the wild-type or a mutant CcmC-His 6 . In this situation, CcmD was absent. The immunoblots presented in Fig. 5 show that the mutants produced normal levels of CcmC and CcmE protein (top and middle panel, respectively). In mutant W114A (lane 2), D47N (lane 8), and R55P (lane 9), small amounts of holo-CcmE were detectable, whereas none of the other point mutants produced visible amounts of holo-CcmE (Fig. 5, bottom panel). We included mutant W119A (lane 7), which had been already described previously (22) and does not carry a His tag, not only as an internal control but also for a better comparison with our new point mutant W119R. Our results confirm the earlier observation that CcmD is indeed required for the efficient charging of apo-CcmE with heme and that certain residues within CcmC contribute to this activity in specific ways.
CcmC Does Not Bind CcmF-One of the most striking findings of this mutant analysis was the discovery of a new phenotype of mutants D47N, R55P, and S176Y with a deficiency in cytochrome c formation despite normal levels of apo-and holo-CcmE and CcmC-CcmE interaction (Table III). One plausible explanation for this might be a direct interaction with another factor catalyzing the detachment of heme from holo-CcmE and its subsequent transfer to apocytochrome c. CcmF had previously been shown to directly interact with CcmE, and it was postulated to be a cytochrome c heme lyase (36). Since CcmE interacts with both CcmC and CcmF, we were interested in testing the formation of a ternary complex between CcmC, CcmE, and CcmF. We tested by co-immunoprecipitation whether or not CcmC could physically interact with CcmF in the presence or absence of CcmE. Proteins from membrane extracts were precipitated with anti-CcmF antibodies. Subsequent immunoblot analysis of the precipitates revealed that the CcmC polypeptide did not co-precipitate with CcmF, no matter whether or not CcmE was present in the system. The inverse precipitation (i.e. using anti-histidine tag antibodies for precipitation of His-tagged CcmC and detection for co-precipitating CcmF) revealed no specific signals (data not shown).
Analysis of a CcmCE Fusion Protein-To investigate the idea that CcmE shuttles between CcmC and CcmF and that certain CcmC mutants might prevent dissociation of CcmE from CcmC, we genetically fused the N terminus of CcmE to the C terminus of CcmC. We expected to obtain a functional fusion protein, because both termini are predicted to be in the cytoplasm. We thus tested whether we could find the same phenotype as we had obtained in mutants with a defect in the cytochrome c maturation but not in the formation of holo-CcmE. We first tested heme delivery to CcmE in the fusion protein. The CcmCE fusion protein was charged with heme, but in addition, accumulation of a degradation product corresponding in size to the wild-type holo-CcmE was also observed (Fig. 6A). Additionally, we found that heme was preferentially delivered to free apo-CcmE rather than to the apo-CcmCE fusion protein (lane 2). To assess cytochrome c formation, ⌬ccmC (EC28) and ⌬ccmE (EC65) cells were complemented with the plasmid pEC826 encoding the CcmCE fusion or plasmid pEC409 encoding CcmC and CcmE individually. We included plasmid pRJ3290 encoding exogenous B. japonicum cytochrome c 550 for a better detection of low amounts of holocytochrome c. The fusion protein could efficiently complement neither the ⌬ccmC nor the ⌬ccmE mutant background as compared with the controls where CcmC and CcmE were expressed individually in the respective backgrounds. The low amounts of holocytochrome c formation in the mutant strains expressing the fusion protein might be attributed either to a low activity of the fusion protein or to the activity exerted by the cleaved product obtained from the fusion protein (Fig. 6B). Since the CcmCE fusion protein cannot efficiently complement cytochrome c maturation, this represents another line of support for the idea that CcmE must be mobile for efficient heme delivery. DISCUSSION Heme delivery is an essential step of cytochrome c maturation. In E. coli, the heme chaperone CcmE plays a crucial role in this process by binding heme transiently in a covalent manner that is still not understood in detail. However, it is known that CcmC, the subject of the present study, is needed for heme transfer to CcmE at the periplasmic side of the membrane. The signature motif WGXXWXWD of CcmC and the two conserved histidines His 60 and His 184 have been shown to play an important role for the interaction with CcmE (14). However, the number of additional highly conserved residues within the 245-amino acid polypeptide was too high for a rational selection of specific residues for site-directed mutagenesis. Strikingly, many of the best conserved residues map to the three periplasmic loops, whereas the six-transmembrane helices are generally hydrophobic, and the small cytoplasmic parts are the least conserved. The genetic and biochemical analysis of CcmC has confirmed the role of the protein in heme delivery to CcmE, but it is still unclear whether CcmC translocates heme across the membrane or picks it up on the periplasmic side of the membrane, either from the outer leaflet or from a heme transporter, for docking it specifically to the heme chaperone CcmE. A random mutagenesis of ccmC was initiated to answer two questions. (i) Can cytoplasmic or membrane-integral amino acids be identified as essential structures for CcmC function? If CcmC were a heme permease, an inward-oriented heme binding site might be identified. In our study, we could not identify essential residues in these domains. Site-directed mutations of Leu 82 on the cytoplasmic side of the transmembrane helix II and Tyr 139 in the middle of transmembrane helix IV to alanines had no effect on CcmC function. A deletion of the C-terminal cytoplasmic extension led to an assembly or stability problem, and only low levels of truncated CcmCЈ were obtained. However, the protein was useful for heme transfer to CcmE but not for cytochrome c maturation. (ii) Are additional, not strictly conserved amino acids essential for CcmC function? Expected mutant phenotypes ranged from a complete block in heme binding, heme delivery, and CcmC-CcmE interaction to partially functioning proteins. For example, was it possible to obtain a mutant that interacted with heme and CcmE but was unable to transfer heme to CcmE? This could indicate an active participation of CcmC in the chemistry of heme attachment to CcmE. This phenotype was not found among our mutants.
The genetic screen applied to learn more about CcmC was designed such that it can be used later for other ccm genes as well. The combination of a dms with a ccm mutant led to a colony phenotype that could be easily distinguished from the wild type. In fact, the colony phenotype in our screen proved to be a very sensitive way to test cytochrome c maturation.
An important finding of this study is that all mutants with a cytochrome c-negative phenotype mapped to periplasmic domains. New residues were shown to be functionally important: in domain I by mutants D47N and R55P, in domain II by mutant G111S, and in domain III by mutants S176Y, V177G, and L183P. Other residues had previously been changed to alanine by site-directed mutagenesis (22). One of the important histidines, His 184 , and Trp 119 within the Trp-rich motif were changed to Tyr and Arg, respectively, during the random mutagenesis. In addition, some highly conserved residues that had not been hit by the random mutagenesis were changed to alanine; W114A and R128A, representing CcmC-specific extensions of the Trp-rich motif, turned pink, and L82A close to the cytoplasmic side of the membrane and Y139A in the middle of transmembrane helix IV formed red colonies like the wild type.
Truncation of the C-terminal, cytoplasmic domain led to low levels of CcmCЈ protein in the membrane. The truncated protein was able to transfer some heme to CcmE and affected cytochrome c maturation even more. The physical interaction with CcmE was not detectable, perhaps due to the low level of protein.
G111S and H184Y were the only mutants that lacked interaction with CcmE. An interaction between CcmC and CcmE, to which residues in the periplasmic domains II and III contrib-  ute, seems to be an obligatory step in heme delivery to CcmE.
We have classified mutants according to their ability to transfer heme to apo-CcmE and to form holocytochrome c (Table III). Class I mutants had a full (in mutants G111S and H184Y) or partial (in mutants W119R, D126A, V177G, and L183P) deficiency in holo-CcmE formation and a white colony phenotype, and they produced less than 10% of the c-type cytochromes found in the wild type. Class II mutants (D47N, R55P, W114A, W125I, R128A, and S176Y) showed almost normal heme transfer to apo-CcmE but were deficient in cytochrome c maturation. In this class, the most interesting mutant was S176Y. It produced wild-type levels of CcmC, CcmE, and CcmC-CcmE co-precipitates, and heme transfer was almost normal, but cytochrome c maturation failed. A similar but less prominent behavior was seen with mutants D47N and R55P and to some extent even with W114A, W125I, and R128A that produced intermediate pink colony phenotypes. Class III mutants (L82A, Y139A, and W180A) were affected neither in holo-CcmE nor in holo-cytochrome c formation. We conclude that besides its role in heme delivery, CcmC can affect cytochrome c maturation after heme is bound to CcmE. Prompted by this hypothesis, the idea that CcmC(D)E form a stable complex with CcmF was tested and rejected because co-precipitation between CcmC and CcmF was not observed, although CcmE precipitated with CcmC and CcmF individually. Our findings imply that CcmC does not interact with CcmF, neither directly nor indirectly via CcmE, and that CcmC and CcmF are involved in different, physically separable steps of cytochrome c maturation. Formally, it is also possible that CcmC interacts not only with CcmE but also with apocytochrome c directly. However, we could not find signals in co-precipitation experiments between CcmC and apocytochrome c (data not shown). From this work, we have no evidence for a "maturase supercomplex" present in E. coli membranes as was hypothesized previously (1). As a more likely alternative explaining the cytochrome c deficiency of heme delivery-positive ccmC mutants, we consider CcmE to shuttle between CcmC and CcmF for heme transfer to apocytochrome c (Fig. 7). Subsequent dissociation of CcmC from CcmE may fail in these mutants, thereby inhibiting a proficient follow-up reaction of CcmE with CcmF.
If a dynamic movement of CcmE between other components of the cytochrome c maturation machinery is essential, the irreversible binding of the two proteins in a CcmCE fusion polypeptide might stop the process. Therefore, a CcmCE fusion protein was constructed and functionally characterized. Heme delivery to the CcmE part of the fusion proteins was quite efficient, but a wild-type CcmE protein was the preferred heme acceptor. Unfortunately, the fusion protein produced a degradation product approximately the size of wild-type CcmE, suggesting that a protease-sensitive site close to the fusion site was present. Nevertheless, it was obvious that the fusion protein was less efficient in cytochrome c formation than two separate molecules, again supporting the idea of CcmE shuttling between CcmC and CcmF.
It had been noted previously that the small membrane protein CcmD is involved in heme transfer to CcmE by CcmC. It was not surprising to see that our mutants were all defective in heme delivery in the absence of CcmD. How CcmD influences the efficiency of heme transfer is unclear. We speculate that a direct interaction of CcmD with CcmC, CcmE, or both takes place. The role of CcmD will be addressed in future work.