CCME, a nuclear-encoded heme-binding protein involved in cytochrome c maturation in plant mitochondria.

The maturation of c-type cytochromes requires the covalent attachment of the heme cofactor to the apoprotein. For this process, plant mitochondria follow a pathway distinct from that of animal or yeast mitochondria, closer to that found in alpha- and gamma-proteobacteria. We report the first characterization of a nuclear-encoded component, namely AtCCME, the Arabidopsis thaliana orthologue of CcmE, a periplasmic heme chaperone in bacteria. AtCCME is targeted to mitochondria, and its N-terminal signal peptide is cleaved upon import. AtCCME is a peripheral protein of the mitochondrial inner membrane, and its major hydrophilic domain is oriented toward the intermembrane space. Although a AtCCME (Met(79)-Ser(256)) is not fully able to complement an Escherichia coli CcmE mutant strain for bacterial holocytochrome c production, it is able to bind heme covalently through a conserved histidine, a feature previously shown for E. coli CcmE. Our results suggest that AtCCME is important for cytochrome c maturation in A. thaliana mitochondria and that its heme-binding function has been conserved evolutionary between land plant mitochondria and alpha-proteobacteria.

Electron transport constituents of respiratory chains in prokaryotes and eukaryotes are composed mainly of proteins containing a variety of cofactors including metal ions, iron-sulfur clusters, nucleotides, or hemes. During the last 10 years, there has been an increasing interest in their biogenesis, which involves the transport of both the apoprotein and prosthetic groups, their folding, and their subsequent assembly into the corresponding complexes. In the case of c-type cytochromes, the covalent attachment of heme to the apoprotein occurs at the site of function, which is a different subcellular compartment from that in which synthesis of heme or apocytochromes occurs. These two characteristics add a further complexity to the posttranslational maturation pathway of these heme-containing proteins.
Genetic analysis of model organisms has led to the identification of three different systems for the maturation of c-type cytochromes (1,2) in Gram-negative bacteria (system I), chloroplast (system II), and fungal mitochondria (system III). Although system I and system II share common elements, genome sequencing projects have shown that no homology could be found between systems I and III. In Escherichia coli (system I) at least eight genes, ccmA-ccmH 1 , are essential for the production of holocytochrome c (3,4). In contrast, in yeast mitochondria (system III), only two proteins, cytochrome c and cytochrome c 1 heme lyases were shown to be the central components of the formation of holocytochromes (5,6). Orthologues of fungal heme lyases were found in vertebrate and invertebrate nuclear genomes but not in plant genomes. Unexpectedly, land plant mitochondrial genomes encode orthologues of bacterial ccm genes, namely ccmB, ccmC, and ccmF (7,8). These genes are also found in the mitochondrial genome of early diverging protists such as Reclinomonas americana (9) but not in the mitochondrial or nuclear genomes of yeast. This finding reveals that at least two distinct routes for c-type cytochrome assembly were established during mitochondrial evolution as exemplified by yeast and land plants. Plant mitochondria have evolved unique strategies for the structure and expression of their genetic information (10) and have developed specific biochemical and physiological functions as organelles belonging to photosynthetic organisms (11). The c-type cytochrome maturation pathway is an example of the specific features displayed by plant mitochondria compared with their mammalian or fungal counterparts.
All mitochondria contain two c-type cytochromes: cytochrome c 1 is part of complex III (cytochrome c reductase), whereas cytochrome c shuttles from complex III to complex IV (cytochrome c oxidase). Both apoproteins are synthesized in the cytosol and are imported post-translationally into the mitochondrial inner membrane and intermembrane space.
Heme needs to be translocated from the matrix side of the inner membrane where ferrochelatase, the last enzyme in its biosynthesis, is located (12) to the intermembrane space where ligation to apocytochrome takes place. In plants, ␦-aminolevulinate, the universal precursor of tetrapyrroles (chlorophyll and heme) is formed in chloroplasts via the C5 (glutamate) pathway (13), whereas in yeast and mammals, it is synthesized by 5-␦-aminolevulinate synthase, a mitochondrial enzyme (14). The next steps of protoporphyrin IX synthesis involve trafficking in cytosol, mitochondria, and plastids in the case of plant cells (15). The trafficking of heme precursors in eukaryotic cells is poorly understood as is the intracellular transport of heme in the cytoplasm or to the nucleus where heme is implicated in transcriptional regulation. In animals and fungi, a short sequence called the "CPV" motif is conserved in cytochrome heme lyases and in other heme-binding proteins such as the transcription factor Hap1, 5-␦-aminolevulinate synthase or heme * This work was supported in part by Centre National de la Recherche Scientifique, the Swiss National Foundation for Scientific Research, and the Eidgenössische Technische Hochschule. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a grant from the Ministère de l'Education Nationale and by a European Molecular Biology Organization short-term fellowship.
ʈ To whom correspondence should be addressed. oxygenase-2 (16). In cytochrome c 1 heme lyase, this motif has been shown to bind heme in a reversible manner (17). However, the exact mechanisms of the transient heme binding to the CPV motif is still unknown.
In bacteria, CcmE is crucial for cytochrome c maturation. This periplasmic protein is anchored in the membrane by its N-terminal hydrophobic domain (18). The CcmE protein from E. coli was shown to be a heme-binding protein (19,20). The bacterial CcmE protein is characterized by two stretches of conserved amino acids, of which one contains a strictly conserved histidine shown to bind heme covalently (19). This protein is believed to capture heme in the periplasm in the presence of CcmC (21,22) and transfer it to the apocytochrome in the presence of one or more of the CcmF, CcmG, or CcmH proteins (19). In this paper we present the characterization of Arabidopsis thaliana CCME, a plant orthologue of E. coli CcmE. We show that the nuclear gene AtCCME encodes a mitochondrial protein attached to the inner membrane and oriented toward the intermembrane space. We have tested the extent to which the plant mitochondrial CCME could substitute for its E. coli orthologue in vivo.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-AtCCME cDNA clone (GenBank TM accession number ATU72502) was obtained from F. Grellet (CNRS, Perpignan France). Plasmids expressing AtCCME under an arabinose-inducible promoter were constructed by PCR. 1 To generate the plasmid pAT1, the coding region of AtCCME corresponding to Met 79 -Ser 256 was amplified using O1 and O 2 as 5Ј and 3Ј primers respectively. The PCR fragment was cloned in the BamHI site of pUC19. Clones with the correct orientation were checked by sequencing, the NdeI/EcoRI fragment was cut out and cloned into pISC-2. The His 222 to Ala AtCCME mutant was generated by "overlap extension" PCR technique. Two overlapping complementary oligonucleotides, O3 and O4, were designed to introduce the mutation. Two amplifications were performed separately using O1-O3 and O4-O5 primer pairs. The purified PCR products were annealed and amplified with O1 and O5. The final PCR product was cut by NdeI and EcoRI and cloned into pISC-2, generating pAT2, as follows: O1, 5Ј-CG-GGATCCATATGCAGAATCGCCGTTTATGG-3Ј; O 2 , 5Ј-CTCCGGATC-CTAAGAAGCCGCAACTTCAGC-3Ј; O3, 5Ј-CTCATCAGCCTTAGCCA-AAACTTCAGTC-3Ј; O4, 5Ј-GGCTAAGGCTGATGAGAAGTATATGCC-A-3Ј; O5, 5Ј-GAAGAATTCAGGATCCTAAGAAGCCGCAACTTC-3Ј. Restriction sites are shown in italics. The E. coli ccmE gene was amplified by PCR and cloned as an NdeI-EcoRI fragment into pISC-2, resulting in pEC412, and the cloned DNA was sequenced. Plasmid pEC101, containing ccmABCD, was cloned by inserting a 1.2-kilobase pair AflII-SspI fragment containing ccmBCD into AflII-FspI-digested pEC86 (23), which provided ccmA. The information concerning strains and plasmids used is this work is summarized in Table I.
Overexpression of AtCCME and Antibody Production-A portion of the AtCCME cDNA (corresponding to Phe 111 -Ser 256 ) was amplified by PCR using the following two primers: 5Ј-ATCGGATCCTCTTCTAC-CTAACGCCA-3Ј and 5Ј-TAAAAAGAA TGAATTCTATACCAATC-3Ј. The PCR product was cut by BamHI and EcoRI and cloned into the corresponding sites of pRSET-C (Invitrogene) downstream of a His tag sequence. After induction with 2 mM isopropyl-1-thio-␤-D-galactopyranoside, the fusion protein was expressed in E. coli strain BL21 (24) and purified in a denaturing buffer by affinity column chromatography using Ni-NTA-Sepharose (Novagen). The fusion protein was injected into rabbits to raise polyclonal antibodies. The sera were purified against the fusion protein, His 6 -AtCCME, coupled to CnBr-activated Sepharose according to the supplier (Amersham Pharmacia Biotech).
Western Blot and Immunodetection-Proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred onto an Immobilon-P membrane (Amersham Pharmacia Biotech). Western blots with AtCCME purified antibodies were performed at a dilution of 1/5000. Goat anti-rabbit antibodies conjugated to horseradish peroxidase (Amersham Pharmacia Biotech) were used as secondary antibodies and revealed with ECL reagents (Amersham Pharmacia Biotech). Antibodies directed against the spinach large subunit (LSU) of ribulose bisphos-phate carboxylase (a gift from B. Camara, IBMP-CNRS, Strasbourg, France) and ␣-tubulin (Amersham Pharmacia Biotech) were used as control. Antibodies directed against potato porin (25) (provided by H.-P. Braun, Hannover University, Hannover, Germany), tobacco manganese-superoxide dismutase (26) (obtained from F. van Breusegem, Gent, Belgium), wheat subunit 9 of NADH dehydrogenase (27), and yeast cytochrome c 1 (provided by G. Schatz, Basel University, Basel, Switzerland) were used as control for outer membrane, matrix, and extrinsic or intrinsic inner membrane protein fractions, respectively, of the mitochondria. The production of antibodies against E. coli CcmE and analysis of proteins expressed in E. coli with alkaline phosphatasecoupled secondary antibodies are described elsewhere (19). The apparent molecular weight of proteins was calculated using middle range molecular weight markers (Bio-Rad) as ladder.
Southern Hybridization-A. thaliana ecotype Columbia were grown on soil in a greenhouse for ϳ4 weeks. Total DNA was extracted as described (28). 25 g of DNA were digested with each restriction endonuclease and analyzed on 0.6% agarose gels. After Southern transfer to a nylon membrane, hybridization was performed under standard conditions at 65°C with a 32 P-labeled AtCCME cDNA probe prepared by random hexamer extension. The membrane was washed in 2ϫ SSC, 0.1% SDS for 30 min at 65°C and twice in 0.2ϫ SSC, 0.1% SDS for 30 min at 65°C.
Bacterial Growth Conditions and Cell Fractionation-E. coli was grown aerobically in LB medium or anaerobically in minimal salt medium with 5 mM nitrite as the electron acceptor (29). For the expression of Bradyrhizobium japonicum cytochrome c 550 , E. coli cells were grown to midexponential phase and then induced with 0.4% arabinose. Wholecell protein analysis, isolation of periplasmic and membrane fractions, and heme staining were performed as described previously (3,21,22,30).
Purification of A. thaliana Mitochondria and Chloroplast-Arabidopsis protoplasts were prepared from 3-4-day-old suspension cell cultures as described previously (31). The washed protoplasts were resuspended in an extraction buffer (400 mM sucrose, 50 mM Tris-HCl, pH 7.5, 3 mM EDTA, 0.1% bovine serum albumin, and 2 mM dithiothreitol), and disrupted by filtrations through nylon membranes (32). The broken cells were diluted in a large volume of the extraction buffer, and differential centrifugations were carried out as described previously (33). The chloroplast-enriched fraction was loaded onto a 40/80% Percoll step gradient, and intact chloroplast were collected as described (31). Mitochondria were layered onto a 13.5-21-45% Percoll step gradient and spun at 75,000 ϫ g for 45 min. The mitochondria were collected at the 21/45% interface and washed in the extraction buffer without bovine serum albumin and dithiothreitol.
Mitoplast Preparation and Submitochondrial Fractionation-Mitoplasts were prepared as described (34) with some modifications (35). The mitochondria were resuspended in a swelling buffer, and the outer membrane rupture was achieved by Dounce homogenization. Mitoplast and outer membrane fractions were isolated after centrifugation through a bovine serum albumin-free discontinuous gradient of 22, 33, and 47% sucrose. Outer membranes were collected at the 8.6/22% interface, diluted, and recovered by centrifugation at 38,000 rpm in a Ti-75 rotor. Intact mitoplasts were collected from the 33/47% interface, washed and resuspended in 20 mM MOPS, pH 7.2, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride at a protein concentration of 3 mg/ml, and broken by three freeze/thaw cycles followed by sonication (5ϫ 10 s, 300 W, Sonic Vibra Cells). The membrane (P) and soluble (S) fraction of mitoplasts were separated by a 30-min centrifugation at 100,000 ϫ g in a Beckman TLA-100 rotor. The soluble proteins were precipitated by 10% trichloroacetic acid. The membrane fraction was subjected to alkaline treatment (0.1 M Na 2 CO 3 , pH 11.5, for 30 min at 4°C) to extract peripheral proteins (36). A 30-min centrifugation at 100,000 ϫ g in a Beckman TLA-100 rotor allows the separation of soluble (peripheral) from insoluble (integral) protein fractions. Freshly purified mitoplasts were subjected to proteinase K treatment. Mitoplasts (1 mg/ml) were incubated with 100 g/ml proteinase K in 8.6% sucrose, 50 mM Tris-HCl (pH 7.5) for 30 min at 4°C. Phenylmethylsulfonyl fluoride was added at the final concentration of 1 mM to stop the protease activity, and the mitoplasts were recovered by centrifugation through a 22% sucrose cushion at 15,000 ϫ g for 15 min.
Import of Radiolabeled Proteins into Isolated Mitochondria-Mitochondria were isolated from potato tubers (Solanum tuberosum, var. Bintje) with a juice extractor as described (37). Proteins were synthesized from the corresponding cDNA clones in pBluescript vector by coupled transcription/translation in the presence of [ 35 S]methionine according the supplier's instruction (Promega). Import assays were carried out as described (38).

RESULTS
AtCCME Is a Single-copy Gene in A. thaliana-The physical and genetic analysis of a nuclear gene cluster on A. thaliana chromosome 3 (39,40), revealed the presence of AtCCME, a gene encoding a protein showing sequence similarity with the E. coli CcmE (3) and its orthologues for Gram-negative bacteria (4,41). All cDNAs, isolated from a library prepared from growing cell suspensions, share the same 5Ј-end at only 5 nucleotides from the initiation codon, whereas two different polyadenylated ends were found at 226 and 268 nucleotides from the stop codon. 2 No transcript could be detected by Northern experiments, but reverse transcription-PCR experiments allow detection of AtCCME transcripts in the roots, rosette leaves, stems, stem leaves and flowers of A. thaliana (data not shown). A radiolabeled probe generated from AtCCME cDNA (ATU72502) was used for Southern hybridization of A. thaliana total DNA. The probe hybridized to a single DNA fragment for each of the five restriction enzymes used. The sizes of the bands correspond to those deduced from the physical map of the AtCCME genomic locus (Fig. 1, A and B). Therefore, in A. thaliana, CCME is a single-copy gene located on chromosome 3 and is expressed at a low level.

AtCCME Is a Mitochondrial Protein and Its Precursor Is Imported into Mitochondria through an N-terminal Cleavable
Targeting Sequence-AtCCME cDNA encodes a putative protein of 256 amino acids, which presents an N-terminal extension of about 70 -80 amino acids when compared with bacterial proteins (Fig. 2). An extension of a similar size is also present in the putative protein encoded by a CCME homologous gene of Oryza sativa (AC025783). In AtCCME, this extension is en-riched in positively charged (Arg) and hydroxylated (Ser) residues and contains few acidic residues. Its first 20 amino acids could form an amphiphilic ␣-helix. These features are characteristic of the import domain of mitochondrial targeting sequences (42,43). The localization of AtCCME was checked first by immunodetection using purified polyclonal antibodies generated against an overexpressed fusion protein (from Phe 111 to Ser 256 ). The anti-AtCCME antibodies recognized a 27-kDa protein in mitochondria but not in chloroplast protein fractions prepared from A. thaliana protoplasts (Fig. 3A). CCME could also be detected in cauliflower, turnip, rapeseed, and radish mitochondria, but not in potato, pea, sunflower, wheat, and maize mitochondria, indicating that anti-AtCCME antibodies are rather specific to Brassica species (data not shown). To test whether AtCCME encodes a precursor processed after import, we tried to import the radiolabeled protein in vitro into purified mitochondria. A major 32-kDa protein was obtained by in vitro coupled transcription/translation of AtCCME cDNA (Fig. 3B). After its incubation with mitochondria, a signal corresponding to a 27-kDa protein resistant to added proteinase K appeared. The protection was abolished when mitochondrial membrane proteins were extracted by Triton X-100 before proteinase K treatment. AtCCME import was inhibited in the presence of valinomycin, indicating the requirement of an electrochemical membrane potential, ⌬, to achieve AtCCME translocation (44). A 5-kDa reduction was observed between the apparent molecular weights of the precursor and the mature protein. For the mitochondrial processing peptidase domain, only a weak consensus has been found, mainly a conserved Arg in position-2 or-3 from the cleavage site. For plant mitochondrial precursors, two cleavage motifs were proposed (45), of which one, an "R-2" motif (RX(A/S)(T/S)), is present at position 47-50 (RLSS, Fig.  2). Although the actual processing site has not been determined experimentally, the R-2 motif described above is a good candidate. A cleavage at this position would shorten the precursor protein by 48 amino acids, corresponding to a 5.4-kDa peptide, which is in agreement with the shift observed in import assays. The radioactive band of the in vitro processed protein and the immunodetected A. thaliana endogenous mitochondrial protein migrate at the same position (Fig. 3C), which strongly suggests that in vitro processing in potato mitochondria reflects in vivo maturation of A. thaliana protein.
AtCCME Is a Peripheral Inner Membrane Protein-Bacterial CcmE are mainly hydrophilic proteins except for a short hydrophobic region at the N-terminal domain, which is predicted to act as a noncleavable signal sequence and to anchor the protein in the membrane. Although its amino acid sequence is not conserved, such a hydrophobic domain is present in At-CCME (Fig. 2), which could, as for its bacterial counterparts, attach the protein in a membrane. AtCCME antibody recognized a protein of 27 kDa in the mitochondria (M) and mitoplasts fraction (MP), whereas no signal could be detected in the outer membrane fraction (OM), characterized by the presence of porin, a major protein of the mitochondrial outer membrane (25) and the absence of cytochrome c 1 , a subunit of the mitochondrial inner membrane cytochrome bc 1 complex (Fig. 4A). The presence of porin in the mitoplast fraction indicated that, in the experimental conditions used, part of the outer membrane remains attached to the inner membrane, most likely at contact points between the two membranes. The mitoplasts were broken and further separated into soluble (S) and membrane (P) protein fractions; AtCCME was present in the membrane fraction (Fig. 4B). No contamination between matrix and membrane proteins could be detected, which was verified using antibodies directed against a matrix protein, manganese-superoxide dismutase, and antibodies against cytochrome c 1 . These results strongly suggest that AtCCME is located in the mitochondrial inner membrane. To further characterize the nature of the membrane interaction, extreme pH treatment was used to extract extrinsic proteins by disruption of electro-static interactions (46). After alkali treatment of mitoplast membranes, AtCCME was found in the soluble fraction (Fig.  4C). AtCCME protein behaves like NAD9, a protein located in the peripheral arm (iron-sulfur protein fraction) of the L-shaped complex I (27,47), and not like cytochrome c 1 , a protein with a C-terminal membrane helix responsible for its intrinsic behavior (Fig. 4C). Our results suggest that AtCCME is attached to the mitochondrial inner membrane by electrostatic interactions such as protein-protein interactions rather than direct contact with the lipid bilayer. AtCCME Is Oriented toward the Intermembrane Space-PhoA fusion analyses of B. japonicum CcmE (BjCcmE) and E. coli CcmE (EcCcmE) proteins have shown that the hydrophilic part of the protein is exposed to the periplasm (18,19). By analogy, AtCCME is predicted to be located at the outer face of the inner membrane. To assess this hypothesis, intact mitoplasts were treated with proteinase K to strip inner membrane proteins, which are exposed to the intermembrane space. Mitochondria and mitoplasts incubated in the same conditions in the absence of protease treatment were used as control (Fig.  4D). When proteinase K was added, the immunodetection of AtCCME was lost, whereas it was still observed in untreated mitoplast. The intactness of the inner membrane after proteinase K treatment was assayed with NAD9 antibody (27). NAD9 is located in the iron-sulfur protein fraction of complex I, which is facing the matrix (47), and therefore protected from proteinase K. The hydrophilic domain of AtCCME is most likely completely digested in treated mitoplasts because no signal corresponding to a smaller partially protected protein could be detected. Therefore in AtCCME, as in its bacterial counterpart, the main conserved motifs are localized on the external side of the inner membrane.
Assaying Complementation of a ⌬ccmE E. coli Strain for Holocytochrome c Production-In addition to its location and FIG. 2. Amino acid sequence comparison of CCME proteins. CCME proteins from different organisms are aligned: AtCCME (39); O. sativa (OsCCME) (GenBank TM accession number AC025783); ␣-proteobacteria B. japonicum (BjCcmE) (18); and ␥-proteobacteria E. coli (EcCcmE) (GenBank TM accession number U00008). The four protein sequences were aligned using the GCG PileUp algorithm (57). Residues identical in more than two sequences are highlighted in black, and the functionally conserved residues are highlighted in gray. The conserved motif 1 and motif 2 regions are indicated. The consensus R-2 plant processing site in AtCCME is boxed. The amino acids corresponding to the hydrophobic domain are underlined. The conserved histidine residue, which binds heme covalently in E. coli, is marked with an asterisk. The numbering is that of the AtCCME amino acid sequence.

FIG. 3. AtCCME is imported into mitochondria.
A, Western blot analysis of Arabidopsis protein fractions (30 g) of protoplast (P), chloroplast (C), and mitochondria (M), probed with antibodies directed against AtCCME, NAD9, LSU (large subunit of ribulose bisphosphate carboxylase), and ␣-tubulin. B, in vitro translated AtCCME precursor (Pre, p) corresponds to a 32-kDa band (lane 1). In the presence of mitochondria (M), a smaller protein corresponding to the mature form appears at 27 kDa. This mature protein is resistant to proteinase K (PK) in intact mitochondria but is degraded when mitochondria membranes are disrupted by Triton X-100 (TX). The import process is abolished in the presence of valinomycin (Val). The precursor and the mature polypeptides are indicated by arrows. Lane 1 contains one-fifth of the amount of precursor included in each import reaction. 40 g of purified potato mitochondria were used for each import assay. Proteins were separated on a 15% SDS-polyacrylamide gel; proteins labeled with [ 35 S]Met were detected by autoradiography for 2 weeks at Ϫ80°C. C, in vitro translated AtCCME precursor (lane 1), import assay of AtCCME in potato mitochondria (lane 2), and 20 g of A. thaliana total mitochondrial proteins (lane 3) were loaded on the same gel. After Western blotting, the membrane was cut in two pieces: one was exposed to a film (lanes 1 and 2), and the second probed with the purified anti-AtCCME antibody (lane 3). topology in mitochondria, the AtCCME sequence similarity with bacterial proteins suggests that the plant protein could fulfill similar functions for cytochrome c biogenesis in mitochondria. Indeed, AtCCME shares 35.5 and 40.6% identical amino acids with E. coli CcmE and B. japonicum CcmE proteins, respectively, whereas prokaryotic CcmE proteins share from 26 to 81% identity. The two major conserved motifs in bacteria, motif 1 and motif 2, are present in AtCCME, although in A. thaliana and in O. sativum, the insertion of a sequence rich in charged residues increases the distance between them (Fig. 2). In motif 2, the histidine residue, which was shown to bind heme covalently in E. coli (19), is strictly conserved in all organisms. To test whether the eukaryotic protein could complement the E. coli ⌬ccmE mutant strain (EC65, Table I), we constructed the plasmid pAT1, which expresses a truncated form of AtCCME deleted from its first 78 amino acids. AtCCME (Met 79 -Ser 256 ) best corresponds to the EcCcmE size, pI, and hydrophobicity profiles. EC65 strain was transformed with a vector carrying various ccmE genes together with pRJ3291 (Table I), carrying cycA, the soluble periplasmic cytochrome c 550 gene from B. japonicum used as a reporter for holocytochrome c maturation (3). The peroxidase activity associated with covalently bound heme was used to check the presence of holocytochromes c in periplasmic protein extract (22). The transformants were grown anaerobically to induce expression of the chromosomal genes ccmA-ccmH and promote cytochrome c production. Both CcmE and cytochrome c 550 were expressed from an arabinose-inducible promoter. When Ec-CcmE was expressed in EC65 background, three cytochromes c were detected by heme stain: NrfA and NapB, two endogenous cytochromes c, and the reporter cytochrome c 550 of B. japonicum (Fig. 5, lane 1). When the complementation was tried with pAT1, no holocytochrome c could be detected (Fig. 5, lane 2). AtCCME Covalently Binds Heme in E. coli-The cytochrome c maturation pathway was impaired when trying to comple-ment a ⌬ccmE E. coli strain with AtCCME in anaerobic respiration. We wanted to know at which step heme trafficking was blocked. For this determination, we tested whether heme transfer to CcmE was possible. Heme stain was used to check the formation of holo-CCME, i.e. a protein binding heme in a covalent way. We transformed EC06, an E. coli strain deleted in all ccm genes (3) with pAT1 alone or with pEC101 (Table I). The assays were performed under aerobic growth conditions in the presence of the CcmA-D proteins, a condition that is sufficient for heme incorporation into CcmE. We first checked whether AtCCME was correctly expressed and inserted in E. coli membranes. After induction by arabinose, a protein corresponding to the truncated form of AtCCME was expressed and detected in the membrane fraction (Fig. 6A, lane 1). Immunodetection using antibodies directed against AtCCME or Ec-CcmE suggests that the level of expression of AtCCME is reduced compared with that of EcCcmE (Fig. 6, A and B). In a ⌬ccm background, AtCCME, as its E. coli counterpart, did not bind heme (Fig. 6C, lane 1). When the E. coli CcmABCD proteins were expressed with AtCCME, the mitochondrial protein was able to bind heme in a covalent way (Fig. 6C, lane 2). This shows that the truncated mitochondrial protein is correctly inserted in the bacterial membrane and that its conserved heme-binding domain is orientated toward the periplasm, thus allowing heme attachment. In E. coli, CcmC is the only Ccm protein that is strictly required for heme transfer and binding to EcCcmE (19). Our results suggest that heme transfer is possible from E. coli CcmC to A. thaliana CCME. To check whether, in AtCCME, heme is attached to the conserved histidine of motif 2, we changed the histidine 222 to an alanine by site-directed mutagenesis. EC06, the ⌬ccm strain, was cotransformed with pAT2 expressing the His 222 Ala truncated AtC-CME and pEC101 expressing CcmABCD. Although the mutant protein was expressed, no heme-binding AtCCME could be detected (Fig. 6, A and C, lane 3). A positive control was done FIG. 4. Submitochondrial localization and topology of AtCCME. A, mitochondria (M), outer membrane (OM), and mitoplast (MP) protein extracts were analyzed for AtCCME, porin, and cytochrome c 1 (cyt c 1 ). B, mitoplasts were subjected to freeze/thaw cycles and sonication. Soluble and membrane proteins were collected after ultracentrifugation. Total mitoplast (MP), supernatant (S), and pellet (P) fractions were analyzed with the indicated antibodies. C, The mitoplast pellet fraction was treated with Na 2 CO 3 , pH 11.5, to extract peripheral proteins (S), whereas intrinsic membrane proteins remains in a 100,000-g pellet (P). D, mitoplast were treated with 100 g/ml proteinase K and analyzed for AtCCME and NAD9. NAD9, which is exposed to the matrix, is shown as the control for inner membrane integrity. The Western blots were probed with antibodies directed against the following mitochondrial proteins: AtCCME, potato porin, wheat NAD9, yeast cytochrome c 1 , and tobacco manganesesuperoxide dismutase (Mn-SOD). with a strain expressing E. coli CcmABCD and CcmE (Fig. 6, B and C, lane 4). Histidine 222 is crucial for heme binding to AtCCME, as it is for the bacterial protein. Together with the complementation assays, these results show that AtCCME is partially able to complement EcCcmE function on the heme trafficking part of cytochrome c maturation pathway. Indeed, AtCCME is able to bind heme, most likely through the conserved histidine of motif 2, in a bacterial heterologous background including EcCcmC. Our results suggest that a full complementation of the E. coli ⌬ccmE strain by AtCCME is most probably impaired by the absence of heme release to the following proteins of the pathway (CcmF, -G, or -H) rather than by the low efficiency with which AtCCME gets heme from CcmC.

DISCUSSION
In photosynthetic eukaryotes, two cytochrome c biogenesis pathways operate in the same cell although in separated compartments, mitochondria and chloroplasts. In land plants, sys-tem I and system II are proposed to perform cytochromes c maturation in mitochondria and chloroplast, respectively. Aside from organelle genes, several nuclear loci are most likely involved in cytochrome c biogenesis for each system. System I and system II share some common elements (heme delivery and thioreduction), and their evolutionary link appears through the conservation of a tryptophan-rich motif and some histidine residues (3,22,48). Therefore, the subcellular location of nuclear-encoded proteins showing similarities with bacterial proteins required for cytochromes c biogenesis must be addressed carefully. AtCCME is the first ccm orthologue found in a plant nuclear genome. In A. thaliana this gene is unique. Using different approaches, in vitro import into isolated mitochondria and immunodetection into Arabidopsis protein extracts, we have demonstrated clearly that AtCCME is a mitochondrial protein. In addition, in vivo experiments were performed using transient expression of different AtCCME-GFP fusion proteins in tobacco cells. All these fusion proteins were detected exclusively in mitochondria. 3 We can exclude a dual targeting to both organelles, which has been reported for other proteins such as glutathione reductase (49), ferrochelatase (50), and aminoacyl-tRNA synthetases (51,52).
We have shown that AtCCME is associated with the mitochondrial inner membrane. AtCCME has a typical mitochondrial targeting sequence at its N terminus, which is able to target a reporter protein to mitochondria 3 and which is cleaved upon in vitro import. Our attempts to purify the mature protein for N-terminal sequencing have been unsuccessful, mainly because of its instability when extracted. The exact N-terminal sequence of the mature protein remains unknown. Nevertheless, we propose that a domain of about 16 to 30 residues, conserved in several plant CCME proteins and rich in charged amino acids, is present at the N-terminal end of the mature mitochondrial protein, preceding the hydrophobic domain. This region would constitute a plant-specific motif not found in bacterial CcmE proteins.
In a number of plants, mitochondrial genes encoding counterparts of CcmB, CcmC, and CcmF are transcribed and their mRNAs are edited. In some cases, the putative corresponding proteins could be immunodetected in the membrane protein fraction of mitochondria (53). 4 However, no topology of any of these proteins has been described, and no functional analysis of the corresponding mitochondrial genes has been successful up to now. AtCCME, the product of a nuclear gene, comprises the relevant features that play a role in mitochondrial cytochrome c biogenesis and appears a better candidate for functional analysis based on complementation of E. coli mutant strains.
The complementation of ⌬ccmE E. coli strains has been tested at two different steps of cytochrome c maturation: heme binding to CcmE and heme transfer to apocytochrome c. At-CCME was detected in E. coli membranes and could be hemestained when overexpressed with EcCcmABCD. Because heme binding occurs only when CCME is translocated to the periplasm (19), this indicates the correct location of AtCCME in E. coli. Complementation assays of a ⌬ccmE E. coli strain by AtCCME (Met 79 -Ser 256 ) were unsuccessful for holocytochrome c production. The results of the complementation assays could be explained by the inability of the AtCCME to release heme to the following bacterial partners of the maturation pathway, although it is able to catch it from EcCcmC. Within Ccm proteins, CcmC mitochondrial orthologues are among the closest relatives of their bacterial counterparts, by their predicted topology and by the sequence conservation of the Trp-rich motif and the two flanking essential histidine residues (48,54,55). These conserved motifs were proposed to have a function in heme delivery. They are also found in the mitochondrial CcmF N orthologue to the N-terminal part of CcmF. The conserved domains are interspersed by plant-specific sequences, creating greater divergences between the plant mitochondrial and their bacterial CcmF counterparts than for CcmC ones. The potential interactions of AtCCME with EcCcmF could be less efficient in a heterologous system, explaining the absence of heme release from CCME. The knowledge of the full set of proteins for the c-type cytochromes pathway in mitochondria will help in designing design different combinations of mitochondrial/bacterial genes for holocytochrome c formation in bacteria.
Because of experimental limitations, heme binding to At-CCME in mitochondria could not be tested with the heme staining methods used for the overexpressed EcCcmE. The instability of the purified mitochondrial protein and the fact that heme binding, although covalent, is proposed to be transient were the main difficulties encountered in getting direct evidence of heme binding to AtCCME in plant mitochondria.
The cytochrome c biogenesis pathway, known as system I, is followed by ␣and ␥-proteobacteria and Archae. Genes encoding related proteins were identified by sequence similarities in mitochondrial genomes of a few protists, one red algae (56), and land plants. The maximum set of ccm genes (system I) found in mitochondrial genomes is found in protists like R. americana (9). CcmA, CcmB, CcmC, and CcmF orthologues are encoded by these mitochondrial genomes, which are among the closest relatives of the ancestral mitochondrial genome and resemble eubacterial ones. A ccmH/cycL-like gene (GenBank TM accession number AC007591) was recently found in A. thaliana (chromosome 1; 28% amino acid sequence identity with the B. japonicum CycL gene product), but the subcellular localization of its product is unknown. CCME is a new mitochondrial protein proposed to be involved in cytochrome c biogenesis in plant mitochondria. The existence of such a set of proteins strongly argues in favor of the conservation of a functional system I in plant mitochondria. In this paper, for the first time, a function has been attributed to a mitochondrial Ccm counterpart. We propose that AtCCME is a heme chaperone of the intermembrane space attached to the inner mitochondrial membrane. Because His 222 is essential for heme binding on AtCCME, we propose that heme is attached by a single covalent bond to a histidine, as found for E. coli CcmE.
In eukaryotes, AtCCME is a unique example of a mitochondrial heme chaperone. In system II, no heme chaperone has yet been described neither in Chlamydomonas chloroplast nor in Gram-positive bacteria, which are the model organisms for the study of this pathway. However, it is possible that other types of heme chaperones, perhaps with a different type of heme linkage, will be discovered in these systems. In system III, it is also unknown whether an heme chaperone is needed and, if so, whether this function is held by the cytochrome c and cytochrome c 1 heme lyase proteins or another unidentified protein.
It is striking that in mitochondria two systems have evolved for cytochrome c maturation. More detailed knowledge of each of these systems will help the understanding of why system I has been retained in the mitochondria of some species and why and from what source system III has evolved.