CcmI Subunit of CcmFHI Heme Ligation Complex Functions as an Apocytochrome c Chaperone during c-Type Cytochrome Maturation*

Background: Cytochrome c maturation (Ccm) is the covalent ligation of heme b to an apocytochrome c by the Ccm apparatus. Results: CcmI subunit of CcmFHI heme ligation complex recognizes and binds specifically apo- and not holocytochrome c2. Conclusion: CcmI and its homologues are apocytochrome c chaperones. Significance: A first glimpse to how the heme ligation complex recognizes an apocytochrome c before, and releases a holocytochrome c after, cofactor addition. Cytochrome c maturation (Ccm) is a sophisticated post-translational process. It occurs after translocation of apocytochromes c to the p side of energy transducing membranes and forms stereo-specific thioether bonds between the vinyl groups of heme b (protoporphyrin IX-Fe) and the thiol groups of cysteines at their conserved heme binding sites. In many organisms this process involves up to 10 (CcmABCDEFGHI and CcdA) membrane proteins. One of these proteins is CcmI, which has an N-terminal membrane-embedded domain with two transmembrane helices and a large C-terminal periplasmic domain with protein-protein interaction motifs. Together with CcmF and CcmH, CcmI forms a multisubunit heme ligation complex. How the CcmFHI complex recognizes its apocytochrome c substrates remained unknown. In this study, using Rhodobacter capsulatus apocytochrome c2 as a Ccm substrate, we demonstrate for the first time that CcmI binds apocytochrome c2 but not holocytochrome c2. Mainly the C-terminal portions of both CcmI and apocytochrome c2 mediate this binding. Other physical interactions via the conserved structural elements in apocytochrome c2, like the heme ligating cysteines or heme iron axial ligands, are less crucial. Furthermore, we show that the N-terminal domain of CcmI can also weakly bind apocytochrome c2, but this interaction requires a free thiol group at apocytochrome c2 heme binding site. We conclude that the CcmI subunit of the CcmFHI complex functions as an apocytochrome c chaperone during the Ccm process used by proteobacteria, archaea, mitochondria of plants and red algae.

The c-type cytochromes are widespread electron carrier proteins found in all domains of life (1). They have conserved fea-tures despite their diverse three-dimensional structures, and their functions extend from electron transport in biological systems (2) to apoptosis in eukaryotic cells (3). All cytochromes c have at least one protoporphyrin IX-Fe (heme b) cofactor that is stereo-specifically attached to the polypeptide chain via two thioether bonds (4). These bonds are formed between the vinyl groups at positions 2 and 4 of the heme tetrapyrrole ring and the two thiol groups of Cys 1 and Cys 2 of apocytochrome c (C 1 XXC 2 H) heme binding site, respectively. In ␣and ␥-proteobacteria like Rhodobacter capsulatus, cytochrome c maturation (Ccm) 4 occurs in the periplasm after translation and translocation of apocytochromes c from the cytoplasm via the Sec secretion pathway (see Fig. 1). It is a complex process that involves up to 10 membrane proteins: CcmABCDEFGHI and CcdA (5,6). These components are responsible for (i) transport and relay of heme (CcmABCDE), (ii) preparation and chaperoning of ligation-competent apocytochromes c (CcdA and CcmG), and (iii) ligation of heme to apocytochromes c (Ccm-FHI) to produce mature holocytochromes c (5,6). The final step of this intricate process is the ligation per se of heme b to apocytochrome c to form mature cytochrome c, and this is attributed to the multisubunit membrane complex CcmFHI (7) (see Fig. 1). CcmF contains heme b and belongs to the heme handling protein family (8,9). It has a conserved periplasmic WWD motif and two conserved His residues that are proposed to acquire heme from holoCcmE (9). CcmH has a single transmembrane helix and a periplasmic domain with a non-canonical thioredox-like motif (10). Last, CcmI is a bipartite protein with an N-terminal membrane integral part (CcmI-1) composed by two transmembrane helices linked through a cytoplasmic loop with a leucine zipper-like motif (11,12). Its second part is a large C-terminal periplasmic domain (CcmI-2) with three tetratricopeptide repeats (TPR) (see Fig. 2A). Earlier genetic studies inferred that the two parts of CcmI play different roles during Ccm (11,(13)(14)(15)(16). In R. capsulatus, the absence of CcmI differently affects the production of membrane-bound (i.e. cytochrome c 1 subunit of the cytochrome bc 1 complex, cytochrome c y , and cytochromes c o and c p subunits of cbb 3 -type cytochrome c oxidase) and periplasmic (i.e. cytochromes c 2 and cЈ) electron carrier c-type cytochromes (17)(18)(19)(20). As expected, CcmI-null mutants produce no c-type cytochromes (11,12). However, mutants containing only the CcmI-1 domain still produce cytochrome c 1 . In contrast, mutants lacking CcmI-1 do not produce any c-type cytochromes (11,12). These findings suggested that CcmI-1 is required for maturation of all c-type cytochromes, whereas CcmI-2 is unnecessary for that of the C-terminally anchored cytochrome c 1 . Overall, studies suggested that CcmI-1 might be involved in heme ligation together with CcmF and CcmH, whereas CcmI-2 cooperates with CcmG for delivery of apocytochromes c to the heme ligation complex (12,21,22). In fact, recent data established that R. capsulatus CcmFHI forms a membrane-integral complex (7) (Fig. 1) and suggested that CcmG might function as a apocytochrome c 2 "holdase" (23). How CcmFHI recognized apocytochromes c remained unknown.
In this study we chose the apocytochrome form of R. capsulatus cytochrome c 2 as a substrate to test the putative chaperone role of CcmI during Ccm. Cytochrome c 2 is a periplasmic electron transfer protein involved in both respiration and photosynthesis and belongs to the Class I c-type cytochromes (24). These c-type cytochromes are characterized by an N-terminallocated heme binding motif (C 1 XXC 2 H) and a C-terminal-located methionine residue acting as the sixth axial ligand of heme iron. We heterologously expressed in Escherichia coli and affinity-purified R. capsulatus CcmI and apocytochrome c 2 .
Using a reciprocal in vitro protein-protein interaction assay, we demonstrated for the first time that R. capsulatus CcmI tightly binds apocytochrome c 2 but not holocytochrome c 2 . This binding occurs via strong interactions between the periplasmic CcmI-2 domain of CcmI and the C-terminal helical portion of apocytochrome c 2 . Other salient structural elements of apocytochrome c 2 , such as its heme binding site Cys residues or its heme iron axial ligands, and the membrane-integral CcmI-1 domain of CcmI play less critical roles in these interactions. We conclude that the CcmI subunit of the CcmFHI complex functions as an apocytochrome c chaperone during the Ccm process.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-The bacterial strains and plasmids used in this work are described in Table 1. R. capsulatus strains were grown chemo-heterotrophically (i.e. by respiration) at 35°C on enriched (mineral-peptone-yeast extract medium) medium supplemented as needed with tetracycline at 2.5 g per ml final concentration (25). E. coli strains were grown aerobically at 37°C in Luria-Bertani (LB) broth medium supplemented with ampicillin at a final concentration of 100 g/ml. The E. coli RP4182, which lacks supE, was used as an expression strain for apocytochrome c 2 to avoid any undesired C-terminal extension via translational read-through across its stop codon.
Molecular Genetic Techniques-Molecular genetic techniques were performed according to Sambrook et al. (26), and all constructs were confirmed by DNA sequencing and ana-FIGURE 1. Overall organization of Ccm system I. After translocation to the periplasm via the general secretory pathway (SEC), a pre-apocytochrome is processed to an apocytochrome that undergoes a series of thio-redox reactions. CcdA and CcmG (not shown for the sake of simplicity) keep reduced the heme binding site cysteines of apocytochrome to render it ligation-competent (left, apocytochrome c translocation, preparation, and delivery). CcmABCDE proteins mediate translocation and relay of heme b to the ligation site (right, heme b translocation, preparation, and delivery). The final step of Ccm is attributed to the CcmFHI proteins that form a multisubunit membrane complex responsible for apocytochrome c-heme b ligation reaction per se (middle, apocytochrome c and heme b ligation). lyzed using the Serial Cloner 2.1 and BLAST software. Nucleotide sequences of the primers used are described in supplemental Table S1). Plasmid pAV1 was generated from pCS1726 (7), carrying an in-frame Strep-tag II epitope (WSHPQFEK) fused at the 5Ј-end of cycA encoding R. capsulatus cytochrome c 2 without its signal sequence by creating an additional stop codon (TAA) 15 bp downstream from the native TAG codon of cycA (Fig. 2B). Plasmids pAV1C13S, pAV1C16S, and pAV1C13S/ C16S containing the single and double Cys to Ser substitutions at the 13 CKTCH 17 heme binding motif of apocytochrome c 2 were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) following the manufacturer's instructions and plasmid pAV1 as template. Similarly, pAV1H17S or pAV1M96S with the apocytochrome c 2 heme iron axial ligands His-17 or Met-96 to Ser substitutions, respectively, were obtained. The truncated versions of apocytochrome c 2 were generated by inserting a stop codon (TAA) 78 or 42 bp upstream the native TAG stop codon of cycA using pAV1 as template, yielding pAV2 or pAV3, respectively. The C-terminal apocytochrome c 2 truncation in pAV2 eliminated the last C-terminal 26 amino acids of holocytochrome c 2 including Met-96, whereas that in pAV3 eliminated only the last 15, excluding Met-96 (Fig. 2B). Plasmid pAV2C13S/C16S was constructed using pAV2 as a template and the primers used before for the full-length Cys-less mutant. Apocytochrome c 2 derivatives contained the N-terminal in-frame Strep-tag II epitope sequence followed by the Factor Xa cleavage site.
PCR amplification using plasmid pSVEN containing ccmI (previously called cycH) (11) and primers listed in supplemental Table S1, adding a 5Ј-NdeI and 3Ј-BamHI restriction sites, yielded a 1.3-kb fragment that was phosphorylated and cloned into the EcoRV site of pBSK to yield pMADO3. This plasmid was digested with NdeI and BamHI sites, and the fragmentcontaining ccmI was cloned into the same sites of pCS1303 (a derivative of pCS905 (27)) to obtain pMADO5 ( Fig. 2A, CcmI). Plasmid pMADO5 contained an in-frame 10-histidine-long (His 10 ) epitope tag and a Factor Xa cleavage site fused at the N-terminal end of CcmI. An appropriate derivative (i.e. pCS1587, Table 1) carrying the His 10 -CcmI complemented fully in vivo an R. capsulatus mutant strain (MT-SRP1, Table 1) lacking CcmI, establishing that it was functional. The CcmI-1 portion of CcmI, corresponding to the N-terminal 360 bp of ccmI, was cloned from a wild type R. capsulatus (MT1131) genomic DNA by PCR again using primers with NdeI and BamHI restriction sites. The PCR product thus obtained was digested with NdeI and BamHI and cloned into pCS1303 using the same sites to yield pMADO5-1. This plasmid contained the N-terminal 121 amino acids of CcmI with its two transmembrane helices and its cytoplasmic loop ( Fig. 2A, CcmI-1). Plasmid pMADO5 was used as a template for site-directed mutagenesis to delete the first in-frame 249 bp of ccmI to yield pMADO5-2, producing CcmI-2 lacking the N-terminal 83 amino acids forming the first transmembrane helix and the cytoplasmic loop of native CcmI ( Fig. 2A, CcmI-2). All CcmI derivatives contained the N-terminal His 10 epitope tag followed by the Factor Xa cleavage site.
Preparation of R. capsulatus or E. coli Detergent-solubilized Membrane Proteins-R. capsulatus MT1131 cells were resuspended in buffer A (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM EDTA, 1 mM PMSF) and intracytoplasmic membrane vesicles (chromotophores) prepared using a French pressure cell (25). After a 30-min low speed centrifugation (4°C and 14,000 ϫ g) the supernatant was further centrifuged for 2 h at 4°C and 138,000 ϫ g to pellet chromatophores. Pellets were resuspended in buffer A, and membrane protein concentrations were determined according to the bicinchoninic acid kit (Sigma, Inc.) using bovine serum albumin as a standard. Membranes were dispersed by the addition of n-dodecyl-␤-D-maltoside (DDM) (Anatrace, Inc.) at a protein:detergent ratio of 1:1 (w/w) under continuous stirring for 1 h at 4°C, and non-solubilized membranes were removed by centrifugation at 138,000 ϫ g for 1 h to yield "clear" supernatants that were stored at Ϫ20°C until further use. In the case of E. coli, cells expressing CcmI-1 were resuspended in buffer A without EDTA, and dispersed membrane fractions were obtained as described for R. capsulatus cells.
Heterologous Expression and Purification of His 10 -CcmI and Strep-apocytochrome c 2 and Their Derivatives-E. coli strains were grown at 37°C (with 200 rpm shaking) in 1 liter of LB medium supplemented with 100 g/ml ampicillin until an A 600 of 0.6 -0.8. Expression of the desired proteins in E. coli was controlled by a Ptac-lac promoter-operator system. Cultures were induced by the addition of 1 mM isopropyl ␤-D-1-thiogalactopyranoside for 4 h. For purification of His 10 -CcmI, His 10 -CcmI-2, and Strep-apocytochrome c 2 as well as its mutant derivatives, cells were lysed using 5 ml of CelLytic™ B 2ϫ Cell Lysis Reagent (Sigma) per g of wet cells supplemented with lysozyme (0.2 mg/ml), DNase (0.05 mg/ml), MgCl 2 (10 mM), and EDTA-free protease inhibitor mixture (Sigma) for 1 h at 4°C with gentle shaking. Crude lysates were centrifuged for 30 min at 20,000 ϫ g, and cleared supernatants were filtered through a 0.45-m filter.
For purification of His 10 -CcmI-1, solubilized membranes were loaded initially onto a Ni 2ϩ -Sepharose high performance column. However, because of poor binding efficiency of His 10 -CcmI-1, the flow-through was re-loaded onto a Q-Sepharose Fast Flow column (GE Healthcare) previously equilibrated with a buffer containing 50 mM Tris-HCl, pH 8, and 0.01% DDM. In this way His 10 -CcmI-1 eluted before applying a salt gradient and purified away from the majority of E. coli membrane proteins. After concentration using Amicon YM-10 filters (Millipore Inc.) and desalting as described above, partially purified His 10 -CcmI-1 was identified with anti-CcmI antibodies (see below).
For purification of Strep-apocytochrome c 2 and its mutant variants, cell lysates were diluted 5-fold with "wash buffer" composed of 100 mM Tris-HCl, pH 8, and 150 mM NaCl, loaded onto a 5-ml Strep Tactin-Sepharose (IBA, Inc.) column, washed extensively, and eluted by adding 2.5 mM desthiobiotin to the wash buffer following the manufacturer's instructions. Eluted proteins were concentrated via Amicon YM-3 filters (Millipore Inc.) and desalted using a buffer with 50 mM Tris-HCl, pH 8, 50 mM NaCl. Purification of R. capsulatus cytochrome c 2 was done as in Holden et al. (28).
Production of Anti-CcmI Polyclonal Antibodies-Nickel affinity chromatography-purified R. capsulatus His 10 -CcmI (ϳ3 mg) was subjected to SDS-PAGE, electro-eluted from the gel matrix, and used as an antigen for rabbit polyclonal antibodies production, which was carried out by Open Biosystems, Inc.
Protein-Protein Interaction Studies Using Reciprocal Co-purification Assays-Protein-protein interactions between CcmI and its derivatives and apocytochrome c 2 and its mutant variants were assayed as follows. His 10 -CcmI (5 or 10 g) and Strepapocytochrome c 2 (15 or 30 g) were mixed in binding/wash buffer containing 50 mM Tris-HCl, pH 8, 50 mM NaCl, 50 mM imidazole, and 0.01% DDM (final volume of 400 or 800 l, respectively) and incubated for 2 h at 25°C with gentle shaking. After incubation, epitope-tagged components of the assay mixture were re-purified by using as appropriate 200 l of Strep Tactin or Ni 2ϩ -Sepharose resins packed into small columns equilibrated with wash buffer and washed extensively with at least 10 times the column volume. Elution with 2.5 times the column volume was performed by adding 2.5 mM desthiobiotin or 500 mM imidazole to the wash buffer for the Strep-tag or His-tag purifications, respectively. Flow-through and elution fractions were precipitated with methanol:acetone (7:2, v/v) overnight at Ϫ20°C and analyzed by SDS-PAGE. For experiments using solubilized membranes of R. capsulatus MT1131 cells, 10, 20, and 30 g of Strep-apocytochrome c 2 were incubated with 400 g of solubilized membrane proteins for 16 h at 4°C with gentle shaking. After incubation, the mixtures were loaded onto Strep Tactin columns as described above. Elution fractions were concentrated using Amicon YM-3 filters (Millipore Inc.) and analyzed by SDS-PAGE. Protein-protein interactions between His 10 -CcmI and Strep-apocytochrome c 2 were also tested by size exclusion chromatography using a Sephacryl S200 column (GE Healthcare). The two proteins were co-incubated under the conditions used for the co-purification assays (see above) and loaded onto the S200 column equilibrated with the binding/wash buffer, and elution profiles were determined using SDS-PAGE.
SDS-PAGE and Immunoblot Analyses-SDS-PAGE was performed using 15% polyacrylamide gels according to Laemmli (29). For CcmI, CcmI-1, and CcmI-2 immunodetection, gelresolved proteins were electroblotted onto Immobilon-PVDF membranes (Millipore, Inc.) and probed with rabbit polyclonal antibodies that were raised against R. capsulatus CcmI. Horseradish peroxidase conjugated anti-rabbit IgG antibodies (GE Healthcare) were used as secondary antibodies, and detection was performed using the SuperSignal West Pico Chemiluminescent Substrate from Thermo Scientific, Inc.
Reduction and Alkylation of Apocytochrome c 2 , Its Mutant Derivatives, and CcmI-1-Approximately 1 mg (70 M) of purified Strep-apocytochrome c 2 or its appropriate mutant derivatives in 50 mM Tris-HCl, pH 8, 50 mM NaCl were incubated with 10 mM dithiothreitol (DTT) for 1 h at 25°C followed by the addition of freshly prepared 50 mM iodoacetamide (IAM) for another 1 h in the dark. After incubation, excess of DTT and IAM was removed using a PD-10 desalting column previously equilibrated with the same buffer. Reduction/alkylation of CcmI-1 was conducted in a similar fashion using ϳ400 g of partially purified CcmI-1 except that all buffers contained 0.01% DDM.
Circular Dichroism Spectroscopy-The far-UV circular dichroism (CD) spectra (195-250 nm) were recorded with a Model 202 spectropolarimeter (AVIV Instruments, Inc.) using a 2-mm path length cuvette (Hellma, Inc.). CD spectra of proteins were recorded at 25°C using a 3-nm bandwidth and a step size of 1 nm. A time constant of 10 s was used to improve the signal to noise ratio and to decrease the contribution of the solvent at lower wavelengths. CD spectra were recorded using protein concentrations of 1-100 M depending on the number of amino acid residues of proteins examined in 20 mM sodium phosphate buffer, pH 7.5, and corrected by subtracting the spectrum of the buffer alone. Raw CD data were converted into the mean residue ellipticity [] (deg cm 2 dmol Ϫ1 ) at each wavelength using the relation, [] ϭ /(10CNl), where is the observed ellipticity in millidegrees at wavelength , C is the molar protein concentration, N is the number of amino acids of the protein, and l is the path length of the cuvette in cm. To monitor the conformational changes induced by CcmI-apocytochrome c 2 interactions, protein concentrations identical to those used for acquisition of individual CD spectra were incubated using the sodium phosphate buffer described above for 2 h at 25°C. The CD spectrum of the protein mixtures was then compared with the sum of the spectra of the individual proteins. As a negative control, CD spectrum of CcmI incubated with bovine serum albumin was also recorded.
Chemicals-All chemicals and solvents were of high purity and HPLC spectral grades and purchased from commercial sources.

Overproduction and Purification of Epitope-tagged CcmI and Apocytochrome c 2
Defining the molecular nature of specific interactions between various apocytochromes c and Ccm components is important for our understanding of how Ccm proceeds. Earlier genetic studies (11,12) led us to initiate a detailed biochemical analysis of the interactions between CcmI and apocytochrome c 2 . The in vivo accumulation of apocytochromes c is not readily observed in R. capsulatus wild type or Ccm-defective strains, and the purification of a large amount of native membrane-integral Ccm components is also difficult. Thus, we used in vitro approaches with heterologously expressed, N-terminally epitope-tagged apocytochrome c 2 and CcmI (Table 1, Fig. 2). These proteins were over-produced in E. coli and purified in large amounts (ϳmg) by affinity chromatography as described under "Experimental Procedures." The purity of His 10 -CcmI and Strep-apocytochrome c 2 was assessed by SDS-PAGE (Fig. 3A) as well as by nanoLC-MS/MS and immunoblot analyses using anti-CcmI and anti-cytochrome c 2 polyclonal antibodies, respectively (not shown). Purified His 10 -CcmI samples contained a major band (Ͼ90%) of 50 kDa (i.e. fulllength CcmI) and trace amounts of three minor bands of 33, 27, and 21 kDa molecular masses, whereas Strep-apocytochrome c 2 samples had a major band (Ͼ90%) of 13.5 kDa and a minor band of 12.9 kDa molecular mass. In all cases nanoLC-tandem mass spectrometry analyses (data not shown) confirmed that the minor bands corresponded to N-terminal-tagged, C-terminal-truncated derivatives of CcmI or apocytochrome c 2 , produced and co-purified under the conditions used.

CcmI Interacts Directly with Apocytochrome c 2 in Vitro
To probe whether CcmI interacts directly with apocytochrome c 2 during Ccm, in vitro reciprocal protein-protein interaction assays were performed. Purified His 10 -CcmI and Strep-apocytochrome c 2 were co-incubated and subjected to affinity chromatography using Strep Tactin or Ni 2ϩ -Sepharose resin columns, as described under "Experimental Procedures." Different column fractions were analyzed by SDS-PAGE and immunoblots as needed. Control experiments revealed neither unspecific binding of His 10 -CcmI to the Strep Tactin resin nor Step-apocytochrome c 2 to Ni 2ϩ -Sepharose resin columns (Fig.  3B, lanes 1 and 3). In contrast, when His 10 -CcmI and Strepapocytochrome c 2 were co-incubated, they were co-eluted from either of the tag-affinity columns (Fig. 3B, lanes 2 and 4), indicating that CcmI and apocytochrome c 2 interact with each other to form a stable binary complex. We note the presence of minor, C-terminal-truncated forms of CcmI in lanes 2 and 4 (not shown) and the absence of the C-terminal-truncated form of apocytochrome c 2 in lane 4 only.
The co-elution assays were repeated with DDM-dispersed membranes of wild type R. capsulatus strain MT1131 to assess the specificity of CcmI and apocytochrome c 2 interactions. Different amounts of purified Strep-apocytochrome c 2 were incubated, with DDM-dispersed membrane proteins under our assay conditions. Strep Tactin elution fractions analyzed by SDS-PAGE and immunoblots with anti-CcmI specific antibodies showed that native CcmI and Strep-tagged apocytochrome c 2 co-eluted in quantities proportional to the amounts of apocytochrome c 2 used (Fig. 3C). Thus, apocytochrome c 2 interacts specifically with CcmI. Furthermore, formation of a stable complex between His 10 -CcmI and Strep-apocytochrome c 2 was also confirmed by size exclusion chromatography with Sephacryl S-200 column (data not shown).

CcmI Interacts with Apocytochrome c 2 but Not with Holocytochrome c 2 in Vitro
The finding that CcmI binds to apocytochrome c 2 led us to investigate if it also interacted with holocytochrome c 2 . We co-incubated purified His 10 -CcmI and R. capsulatus holocytochrome c 2 (28) and subjected the mixture to the co-purification assay using Ni 2ϩ -Sepharose resin under the conditions that evidenced CcmI-apocytochrome c 2 interactions. The data showed that holocytochrome c 2 does not co-purify with His 10 -CcmI under the conditions used. Thus, CcmI binds only apocytochrome c 2 , and not holocytochrome c 2 , in agreement with its role as an apocytochrome c chaperone during Ccm (Fig. 4).

Co-incubation of Apocytochrome c 2 and CcmI Induces Conformational Changes in Interacting Partners
CD spectroscopy in the far UV region (178 -260 nm) is a powerful technique to probe conformational changes that occur during protein-protein interactions in solution (30,31). As expected, the CD spectrum of purified Strep-apocytochrome c 2 showed low ellipticity above 215 nm and a negative peak around 202 nm, typical of proteins with "random coil"    Fig. 3, panel B, except that His 10 -CcmI (5 g) was incubated with holocytochrome c 2 (15 g) instead of apocytochrome c 2 . Flow-through (FT) and elution (E) fractions from the Ni 2ϩ -Sepharose resin were analyzed by SDS-PAGE as above. Note that holocytochrome c 2 was not retained by Ni 2ϩ -Sepharose resin either when alone or when incubated together with CcmI. Molecular markers (in kDa) are shown on the left. conformations (Fig. 5A) (32,33). Similar spectra were also obtained for the heme binding site cysteine and heme iron axial ligands mutants, the C-terminal-truncated derivatives of apocytochrome c 2 , and the synthetic peptide (see below) (data not shown). On the contrary, the CD spectrum of holocytochrome c 2 and purified His 10 -CcmI showed the typical characteristics of highly ␣-helical proteins, with negative bands at 222 and 208 nm (Fig. 5A) (32-37).
Next, purified His 10 -CcmI and Strep-apocytochrome c 2 were coincubated, and their CD spectra between 195 and 250 nm were acquired ("Experimental Procedures") to assess whether their conformations changed upon their interactions. When the CD spectrum of the protein mixture was compared with the sum of the individual spectrum of each interacting partner, an increase in the helical content of the mixture, as diagnosed by the increase in intensity of the negative bands at 208 and 222 nm, was observed (Fig. 5B). As expected, because CcmI is a highly helical protein, this change is small, but it provides additional evidence for direct association between CcmI and apocytochrome c 2 (31). As controls, His 10 -CcmI incubated either with holocytochrome c 2 or bovine serum albumin instead of Strep-apocytochrome c 2 exhibited no CD spectral changes (Fig.  5, C and D, respectively). Thus, CcmI interacts directly only with the apo form of cytochrome c 2 and not with its holo form.

Structural Determinants Responsible for CcmI and Apocytochrome c 2 Interactions
Considering the putative role of CcmI as an apocytochrome c chaperone (11,12,15,16), defining the molecular basis of their interactions is of paramount importance. R. capsulatus cytochrome c 2 , like its mitochondrial counterpart horse heart cytochrome c, belongs to Class I of c-type cytochromes (24). These proteins have a heme binding (C 1 XXC 2 H) motif located at the N-terminal, a methionine residue acting as the sixth axial ligand of heme iron close to the C-terminus (Fig. 2B), and a globular fold with interacting N-and C-terminal helices (38 -40).
Heme Binding Site Cys Residues-Participation of the heme binding site ( 13 CXXCH 17 ) Cys residues of apocytochrome c 2 in CcmI interactions was examined by using single and double Cys to Ser substitutions as well as by alkylating with IAM the free Cys-thiol groups of Strep-apocytochrome c 2 and its mutant derivatives. Binding of purified apocytochrome c 2 (-C13S, -C16S, and -C13S/C16S) mutants to His 10 -CcmI were probed via affinity co-purification as for the native Step-apocytochrome c 2 . The data indicated that apocytochrome c 2 mutants lacking either or both of the heme binding site Cys residues still co-purified with CcmI. Image analyses of Coomassie Bluestained gels indicated that the amount of CcmI that co-purifies with the -C13S, -C16S, and -C13S/C16S apocytochrome c 2 mutants decreased by only ϳ 20, 10, and 15%, respectively, relative to that observed with the native apocytochrome c 2 (Fig.  6, lanes 4, 6, and 8). Thus, the heme binding site Cys residues have a small contribution to CcmI-apocytochrome c 2 interactions.
During Ccm, heme binding site Cys-thiols are thought to undergo oxidation reduction via the periplasmic thio-redox pathways (5,6,23). Any requirement for a defined redox state of the heme binding site Cys residues for CcmI-apocytochrome c 2 interaction was tested by DTT/IAM treatments of native and mutated apocytochrome c 2 proteins. Co-purification assays revealed that chemical modification of the Cys residues decreased by ϳ 20 -30% that of the amount of co-purifying CcmI as compared with that seen with untreated, native apocytochrome c 2 (Fig. 6, lanes 3, 5, and 7). Thus, the presence of a disulfide bond at the heme binding site of native apocyto-chrome c 2 only slightly favored its interaction with CcmI. We note that similar CcmI co-purification patterns observed with the native or DTT/IAM-treated or Cys to Ser mutant proteins exclude the occurrence of excessive intermolecular disulfide bonds forming apocytochrome c 2 polymers under the conditions used.
Axial Ligands His-17 and Met-96 of Heme Iron Atom-The His-17 to Ser and Met-96 to Ser variants of Strep-apocytochrome c 2 were constructed and purified, and their binding ability to His 10 -CcmI was tested. The data indicated that apocytochrome c 2 mutants lacking the heme iron axial ligands His-17 or Met-96 co-purified with CcmI at amounts similar to those observed with the native apocytochrome c 2 . Thus, the heme axial ligands are not important for CcmI-apocytochrome c 2 interactions (Fig. 6, lanes 9 -11).
The C-terminal ␣-Helix of Apocytochrome c 2 -Minor contributions of the residues involved in heme ligation of apocytochrome c 2 to its interactions with CcmI suggested that additional structural elements might be involved in apocytochrome c-chaperone interactions. Strong hydrophobic interactions between the N-and C-terminal ␣-helices of apocytochrome c are important for folding of Class I c-type cytochromes (41). Thus, two apocytochrome c 2 derivatives lacking their C-terminal helices (t26-and t15-apocytochrome c 2 , missing the last 26 and 15 amino acids of apocytochrome c 2 , respectively) were constructed. Based on R. capsulatus cytochrome c 2 three-dimensional structure, the t15-apocytochrome c 2 mutant lacked only the C-terminal helix that interacts with the N-terminal region but retained the axial heme iron ligand Met-96, whereas the t26-apocytochrome c 2 lacked both of these features (Table 1 and Fig. 2B). A Cys-less derivative of the truncated t26-apocytochrome c 2 was also produced to further assess the contribution of the heme binding site cysteine residues to these interactions. Again, purified truncated apocytochrome c 2 derivatives were subjected to co-purification assays similar to those carried out with native apocytochrome c 2 (Fig. 7A). The amount of CcmI that co-purified with t26-apocytochrome c 2 (apparent molecular mass of 11.5 kDa) decreased significantly to ϳ20% of that observed with native apocytochrome c 2 (Fig. 7A, lanes 1  and 2). A similar situation was also observed with the t15-apocytochrome c 2 (apparent molecular mass of 13 kDa) variant (Fig. 7A, lanes 4 and 5). Moreover, the amount of CcmI retained by the truncated t26-apocytochrome c 2 became barely detectable (Ͻ5%) when its Cys-less derivative was used (Fig. 7A, lane  3). These data indicated that the C-terminal portion of apocytochrome c 2 was a major determinant for binding CcmI, whereas its heme binding site Cys residues provided a small but additive contribution. Furthermore, the comparable binding abilities seen with the t15-and t26-apocytochrome c 2 derivatives confirmed that the heme iron sixth axial ligand is not involved in this interaction.

A Synthetic Peptide Corresponding to the C-terminal Helix of Apocytochrome c 2 Is Sufficient to Recognize and Bind CcmI
CcmI binding ability to the C-terminal helix of apocytochrome c 2 was assessed using a synthetic peptide (NH 2 -WSHPQFEKIEGRKAKTGMAFKLAKGGEDVAAYLASVVK-COOH) corresponding to the C-terminal 26 amino acid FIGURE 6. Co-purification of CcmI with apocytochrome c 2 and its derivatives lacking the heme binding site Cys residues Cys-13 and C-16 or the heme iron axial ligands His-17 and Met-96. Co-purification experiments were conducted as in Fig. 3, panel B, except that His 10 -CcmI (5 g) was incubated with apocytochrome c 2 (15 g) or its appropriate derivatives. Lane 1 shows that CcmI does not exhibit unspecific binding to Strep Tactin resin. Lanes 2 and 9 show co-purification of CcmI with native apocytochrome c 2 , and the amount of CcmI present in the elution fraction was taken as 100% for image quantification using Image J program and used for comparison with the amounts seen in lanes 3-8, 10, and 11. Lanes 4, 6, and 8 show slightly lower amounts of CcmI present in the elution fractions when apocytochrome c 2 (Apocyt c 2 )-C13S (80%), -C16S (90%), and -C13S/C16S (85%), respectively, were used. Lanes 3, 5, and 7 also show lower amounts of CcmI present in the elution fractions when wild type apocytochrome c 2 (80%) and single Cys mutants (70% for C13S and 75% for C16S) treated with DTT and IAM, respectively, were used. Lanes 10 and 11 show similar amounts of CcmI present in the elution fractions when incubated with apocyt-H17S (97%) and -M96S (102%) mutants, respectively, as compared with native apocytochrome c 2 (lane 9).
The asterisks indicate that apocytochrome c 2 Cys thiols were treated with DTT/IAM, and the molecular markers (in kDa) are indicated on the left. residues of native apocytochrome c 2 (deleted in t26-apocytochrome c 2 mutant) (Fig. 2B). Incubation of increasing amounts (5, 10, and 20 g) of this peptide instead of apocytochrome c 2 , with purified His 10 -CcmI led to concentration-dependent copurification of His 10 -CcmI-peptide complex (Fig. 7B). Thus, the last 26 C-terminal amino acids of apocytochrome c 2 are sufficient for its interaction with CcmI in vitro. Based on overall data we concluded that the C-terminal helix of apocytochrome c 2 promotes its binding to CcmI during Ccm.

Periplasmic Domain of CcmI Is Responsible for Binding of Apocytochrome c 2 C-terminal Helix
To establish which functional domain of CcmI interacts with apocytochrome c 2 , two truncated versions of His 10 -CcmI were produced by site-directed mutagenesis ( Fig. 2A). The His 10 -CcmI-2 derivative of CcmI (apparent molecular mass of 42 kDa) contained only the second transmembrane helix followed by the large periplasmic portion containing the three TPR motifs of CcmI ( Fig. 2A, CcmI-2). Co-purification assays with purified His 10 -CcmI-2 were conducted as done with the intact CcmI using native apocytochrome c 2 and its derivatives. The data indicated that CcmI-2 co-purified with apocytochrome c 2 and its derivatives in amounts similar to those seen with intact CcmI (supplemental Fig. S1, A and B). Finally, like intact CcmI, CcmI-2 also bound quantitatively to the synthetic peptide mimicking the C-terminal portion of apocytochrome c 2 (supplemental Fig. S1C). We therefore concluded that the periplasmic CcmI-2 domain of CcmI is sufficient to bind apocytochrome c 2 via its C-terminal helical portion.

Membrane-embedded CcmI-1 Domain of CcmI Binds Weakly to Apocytochrome c 2
A His 10 -CcmI-1 derivative (apparent molecular mass of 14 kDa) that contains only the N-terminal 121 amino acid residues corresponding to the two transmembrane helices and the cytoplasmic loop of CcmI was also produced ( Fig. 2A, CcmI-1). However, this membrane-integral protein was present in low amounts in E. coli membranes, and its purification was very difficult. We succeeded in obtaining His 10 -CcmI-1-enriched fractions and used them for co-purification assays with Strepapocytochrome c 2 and derivatives as done before. Identification of His 10 -CcmI-1 in the flow-through and elution fractions was accomplished using anti-CcmI antibodies. Incubation with apocytochrome c 2 allowed retention and co-elution of a small amount of CcmI-1 by the Strep Tactin column, as detected in the elution fractions with anti-CcmI antibodies. Most of His 10 -CcmI-1 was present in the flow-through (Fig. 8, lane 2) and wash fractions (not shown), indicating that binding of CcmI-1 to apocytochrome c 2 was poor, sharply contrasting that seen with intact CcmI or CcmI-2 proteins. When in the previous experiments a 10-fold excess of apocytochrome c 2 over His 10 -CcmI or His 10 -CcmI-2 was used, the amounts of CcmI or CcmI-2 that co-eluted with apocytochrome c 2 were virtually close to 100%, and these proteins were undetectable in the flowthrough fractions (data not shown). The ratio of apocytochrome c 2 to CcmI-1 was much larger in the latter experiments, and only a small fraction of CcmI-1 co-purified with apocytochrome c 2 (Fig. 8, top and bottom panels), suggesting that it interacted very weakly with apocytochrome c 2 . . CcmI interacts directly with the C-terminal portion of apocytochrome c 2 . Panel A, co-purification of His 10 -CcmI with full-length and C-terminal-truncated derivatives t15-and t26-apocytochrome c 2 and appropriate double Cys mutants is shown. Co-purification was done as in Fig. 3, panel B, except that His 10 -CcmI (5 g) was incubated with apocytochrome c 2 (Apocyt c 2 , 15 g) or the appropriate derivatives as before. Lanes 1 and 4 show co-purification of CcmI with full-length apocytochrome c 2 , and the amount of CcmI present in the elution fraction was taken as 100% for image quantification as in Fig. 6. The amount of CcmI present in the elution fraction after incubation with t26-apocytochrome c 2 was decreased to 20% as compared with the full-length apocytochrome c 2 (lane 2). Lane 3 shows that after mutation of the Cys residues of t26-apocytochrome c 2 , the amount of CcmI that co-purifies is barely detectable. The amount of CcmI present in the elution fraction after incubation with t15-apocytochrome c 2 (lane 5) was also decreased to 17% as compared with the full-length apocytochrome c 2 shown in lane 4, similar to that seen with the t26-apocytochrome c 2 (lane 2). Panel B, co-purification of His 10 -CcmI using different amounts of a synthetic peptide corresponding to the C-terminal 26-amino acid residues of apocytochrome c 2 is shown. Lane 1 shows that His 10 -CcmI is not retained by the Strep Tactin resin in the absence of the synthetic peptide, and increasing amounts of His 10 -CcmI co-purified with 5, 10, and 20 g (lanes 2, 3, and 4, respectively) of the synthetic peptide.  Fig. 3, panel B, except that partially purified His 10 -CcmI-1 (20 g) was incubated with apocytochrome c 2 (apocyt c 2 , 15 g) or single or double Cys apocytochrome c 2 mutants. The assay mixtures were then subjected to affinity chromatography using Strep Tactin-Sepharose, and elution and flow through fractions were analyzed by SDS-PAGE/immunoblot using anti-CcmI antibodies. Lane 1 shows that CcmI-1 does not exhibit any unspecific binding to the Strep Tactin resin. Lane 2 shows co-purification of CcmI-1 with full-length apocytochrome c 2 . Lane 3 shows similar amounts of DTT/AIMtreated CcmI-1 co-purified with apocytochrome c 2 . Lanes 4, 5, and 7 show CcmI-1 co-purified with the single Cys mutants apocytochrome c 2 -C13S and -C16S and the truncated t26-apocytochrome c 2 , respectively. Lanes 6 and 8 show lower amounts of CcmI-I being co-purified with apocytochrome c 2 and the truncated variant double Cys mutants apocytochrome c 2 -C13S/C16S and t26-apocytochrome c 2 -C13S/C16S, respectively. Lane 9 shows that CcmI-1 does not bind or recognize the synthetic peptide corresponding to the 26 C-terminal amino acids of apocytochrome c 2 . Molecular mass markers (in kDa) are indicated on the left, and CcmI-1* indicates that it was treated with DTT/IAM as described under "Experimental Procedures."

Heme Binding Site Cys Residues of Apocytochrome c 2 Are More Important Than Its C-terminal Helix for CcmI-1 Interactions
Comparison of the amounts of CcmI-1 that co-purified with the native, the heme binding site single (C13S or C16S) mutants, and the double (C13S/C16S) mutant of apocytochrome c 2 indicated that those that retained at least one Cys residue at their heme binding site interacted more strongly with CcmI-1 than those that lacked both of them (Fig. 8, lanes 2, 4,  5,and 6). As R. capsulatus CcmI has a unique Cys residue at its position 7, DTT/IAM-treated CcmI-1 was tested for its interactions with the native apocytochrome c 2 . The data indicated that DTT/IAM-treated CcmI-1 still bound apocytochrome c 2 in amounts similar to its untreated form (Fig. 8, lanes 2 and 3). Similarly, co-purification assays using the truncated t26-apocytochrome c 2 and its double Cys mutant derivative confirmed that the C-terminal portion of apocytochrome c 2 was not important, but the absence of the Cys residues decreased the CcmI-1-apocytochrome c 2 interactions (Fig. 8, lanes 7 and 8). Therefore, of the two domains of CcmI, the membrane integral CcmI-1 domain interacts poorly with the N-terminal heme binding region, whereas the periplasmic CcmI-2 domain binds strongly the C-terminal helical portion of apocytochrome c 2 .

DISCUSSION
In most Gram-negative bacteria like R. capsulatus that use Ccm (System I) for cytochrome c production, CcmI is an important subunit of the CcmFHI complex thought to perform the final heme-apocytochrome c ligation step (7). Earlier findings suggested that CcmI acts as a chaperone for apocytochromes c to facilitate their delivery to the heme ligation complex (11,12,21,22). In this work, by probing in vitro protein-protein interactions between purified R. capsulatus CcmI and apocytochrome c 2 , we demonstrated unequivocally that CcmI tightly binds apocytochrome c 2 but not holocytochrome c 2 . This conclusion is supported by reciprocal co-elutions, CD spectroscopy, and size exclusion chromatography. In particular, CD spectra indicated that conformational changes occur in CcmI upon its incubation with only apocytochrome c 2 , and not holocytochrome c 2 or bovine serum albumin, corroborating that CcmI binds the apo-but not the holo-form of cytochrome c 2 in solution. Moreover, experiments using DDM-dispersed total membrane proteins indicated that these interactions are specific.
Current knowledge about the protein-protein interactions between the Ccm components and apocytochromes c is limited. Earlier, the ability of E. coli CcmH to reduce in vitro a disulfide bond formed at the heme binding site of an apocytochrome c-like peptide was reported (10,42). Also, using a yeast two-hybrid system, the interactions in vivo between Arabidopsis thaliana CcmH or CcmFN2 and apocytochrome c were described (43). More recently, in vitro weak binding of apocytochrome c 2 to a Cys-less CcmG protein was detected (23). This work provides the first biochemical evidence that the CcmI subunit of the CcmFHI complex acts as an apocytochrome c chaperone for R. capsulatus apocytochrome c 2 and possibly for other c-type cytochromes during Ccm.
The observation that CcmI binds specifically apocytochrome c 2 and the well known structural changes that apocytochromes c undergo during their folding process (41,44,45) allowed identification of the structural elements that mediate interactions between these partners. A major finding is that CcmI recognizes and binds the C-terminal helix of apocytochrome c 2 rather than the heme binding ( 13 CXXCH 17 ) site at the N terminus or the 5th (i.e. His-17) or 6th (i.e. Met-96) axial ligands of heme iron. Moreover, under our in vitro assay conditions, CcmI-apocytochrome c 2 interactions were unaffected by the redox state of heme binding site Cys residues. Future quantitative measurements or in vivo assays may reveal different CcmI binding affinities for various derivatives of apocytochrome c 2 .
The weakened binding of CcmI to the C-terminal-truncated apocytochrome c 2 derivatives as well as the binding of a synthetic peptide corresponding to the last 26 residues demonstrate clearly that this portion of apocytochrome c 2 is necessary and sufficient for recognition and binding of CcmI. In class I-type cytochromes c, like the horse heart cytochrome c or R. capsulatus cytochrome c 2 , interactions between apocytochrome c and heme are well studied (24,32,45,46). Upon the addition of a tetrapyrrole ring, the N-and C-terminal helices of cytochrome c pair together in a tight perpendicular arrangement. These helix-helix interactions are mediated via Gly-6 and Phe-10/Leu-94 and Tyr-97 in horse heart mitochondrial cytochrome c and via Gly-6 and Phe-10/Val-107 and Tyr-110 in R. capsulatus cytochrome c 2 . Moreover, a third helix adjacent to the sixth axial Met ligand provides conserved hydrophobic contacts to the heme cofactor (24,45,46). Thus, like any class I-type cytochrome c in the absence of heme, the C-terminal helix of R. capsulatus apocytochrome c 2 is available to interact with CcmI, whereas this helix is packed against the N-terminal helix together with heme in holocytochrome c 2 .
The fact that purified CcmI-2 binds apocytochrome c 2 as well as the intact CcmI indicates that the periplasmic C-terminal domain with its TPR motifs is responsible for these interactions. The TPR motifs are widely spread modules composed by 34 hydrophobic residues forming two antiparallel ␣-helices. Tandem arrays of TPR motifs form righthanded super helical structures with concave and convex surfaces for protein-protein interactions (47,48). Interactions similar to those seen here were also observed between the E. coli TPR-containing NrfG, belonging to the nitrite reductase specific NrfEFG heme ligation complex, and the C-terminal helical portion of NrfA apocytochrome c (49). Remarkably though, when both the C-terminal portion of apocytochrome c 2 and its heme binding site Cys residues are eliminated, the CcmI-apocytochrome c 2 interactions are completely abolished. Thus, the heme binding site Cys residues of apocytochrome c 2 also contribute, although to a lesser extent, to its interaction with CcmI.
Unexpectedly, we observed that apocytochrome c 2 also interacts in vitro with the membrane integral CcmI-1 domain of CcmI. Under our experimental conditions, these interactions were much weaker than those seen with CcmI-2 and involved different structural elements of apocytochrome c 2 , including a free thiol group at its heme binding site. Earlier studies indicated that deletion of CcmI-1 abolished the production of all together with heme under reducing conditions. Ongoing studies will address these and other related questions.