The past 10 years have heralded remarkable progress in the understanding of the biogenesis of c-type cytochromes. The hallmark of c-type cytochrome synthesis is the covalent ligation of haem vinyl groups to two cysteinyl residues of the apocytochrome (at a Cys–Xxx–Yyy–Cys–His signature motif). From genetic, genomic and biochemical studies, it is clear that three distinct systems have evolved in nature to assemble this ancient protein. In this review, common principles of assembly for all systems and the mmicular mechanisms predicted for each system are summarized. Prokaryotes, plant mitochondria and chloroplasts use either system I or II, which are each predicted to use dedicated mechanisms for haem delivery, apocytochrome ushering and thioreduction. Accessory proteins of systems I and II co-ordinate the positioning of these two substrates at the membrane surface for covalent ligation. The third system has evolved specifically in mitochondria of fungi, invertebrates and vertebrates. For system III, a pivotal role is played by an enzyme called cytochrome c haem lyase (CCHL) in the mitochondrial intermembrane space.
During the past half-century few macrommicules have received more attention than the electron transport protein cytochrome c. Complementing the rich body of literature on the electron transfer functions and evolution of cytochrome c is a wealth of high-resolution three-dimensional structures. These studies have detailed the key roles that cytochrome c plays in both aerobic and anaerobic respiration and in photosynthesis (Fig. 1A). Recently, renewed interest in this class of mmicule has stemmed from the discovery that programmed cell death in eukaryotes (apoptosis) requires the release of cytochrome c from the mitochondrion (Kluck et al., 1997; see review by Reed, 1997; Yang et al., 1997). Less publicized is the evidence that in certain prokaryotes, a defect in the biogenesis of cytochrome c results in a dramatic 100-fold increase in synthesis and excretion of haem biosynthetic intermediates (Biel and Biel, 1990; Goldman et al., 1997). In addition, defects in some of the cytochrome c biogenesis genes can result in loss of copper resistance (Yang et al., 1996) and pyoverdine production (Gaballa et al., 1996) in Pseudomonas fluorescens, as well as high-affinity iron acquisition in Rhizobium leguminosarum (Yeoman et al., 1997). Results like these highlight the surprises that remain to be uncovered with respect to the roles of cytochrome c for signalling as well as the pathways of their biogenesis.
The cytochromes remain at the frontiers of science owing to the remarkable progress in solving the three-dimensional structures of cytochrome c oxidases and the cytochrome bc1 complex (Iwata et al., 1995; Tsukihara et al., 1996; Xia et al., 1997). Electron transport from cytochrome bc1 to cytochrome c to cytochrome c oxidase to oxygen is now envisaged down to the atomic level.
The biogenesis of these haem proteins has evolved as a new frontier in the field of bioenergetics. This field of study is concerned with the transport of apocytochromes to the appropriate compartment and their assembly into native conformations containing haem. For example, c-type cytochromes are located outside the cytoplasmic membrane in prokaryotes, in the intermembrane space in mitochondria and in the lumen in chloroplasts (Fig. 1A). A key feature of c-type cytochromes, which distinguishes them from the b-type cytochromes, is the covalently attached haem that is linked via thioether bonds between two vinyl groups of the haem and two cysteinyl residues. These cysteinyl residues are part of a signature C–X–X–C–H motif of the cytochrome in which the histidinyl residue acts as an axial ligand to the iron of haem (Fig. 1B).
An initial understanding of factors necessary for the assembly of c-type cytochromes has developed from genetic analyses of Saccharomyces, Neurospora, model bacteria and Chlamydomonas. These studies and recent genomic sequence analyses have led to the view that three separate systems (denoted I, II and III) for the specific biogenesis of c-type cytochromes have evolved (see Fig. 2 for three models). These three systems will be compared and summarized in this review. The rationale for designating these systems I, II, and III stems from the possible evolutionary progression and decreasing order of complexity from system I to system III.
Common principles and requirements for the three systems
It is commonly accepted that haem ligation precedes the natural folding of cytochrome c into its native conformation. However, the specific subcellular location and the covalent attachment of the haem group pose distinct problems for the assembly of this protein. For example, if assembly occurs at the site of function (see below), how do cells transport haem and apocytochrome c to their ultimate destination? As both haem and apocytochrome cysteinyl side-chains must be reduced for ligation, what factors prevent oxygen and other reactants from disrupting the process? What is the mmicular mechanism of covalent haem attachment? What accessory factors are necessary in each energy-transducing membrane and what are their exact functions?
A common principle that has emerged on the biogenesis of c-type cytochromes is the transport of apocytochrome c to its sites of function before haem attachment. As haem is not synthesized at this location, a logical corollary is that haem must also be transported or diffuse to these compartments. Within these compartments haem must find the two apocytochrome c cysteinyl residues, and, in a stereospecific manner, the thiol groups must attack carbons of the vinyl side chains of haem (Fig. 1B). Thus, the cysteines must be reduced. Nicholson and Neupert (1989) further demonstrated that the iron of haem must also be reduced for ligation to occur in fungal mitochondria. The chemical basis for this requirement was evaluated by Barker et al. (1993), who showed that different thiol adducts are produced when the iron is oxidized. Based on these chemical principles, it is now generally assumed that ferrohaem is the substrate for holocytochrome c formation in all three systems. Although system III (e.g. fungal mitochondria) may not adhere to this rule, one hypothesis emerging from this observation is that accessory factors are not directly involved in the formation of the thioether bonds. Rather, they function to bring together the two substrates in the appropriate conformation and chemical state so that ligation can occur. Indeed, when cysteinyl residues are engineered near the vinyl groups of the haems of the b-type cytochromes (e.g. cytochrome b5 and b562; Barker et al., 1993; 1995), ligation occurs spontaneously. The cytochrome c haem lyase reaction therefore simply obeys Markovnikov's rule of thioether formation occurring to the carbon with the fewest hydrogens.
The last step of haem biosynthesis is the insertion of reduced iron into protoporphyrin IX by the enzyme ferrochelatase. This enzyme resides in the cytoplasm of bacteria, the matrix of mitochondria or the stroma of chloroplasts (Smith and Kohorn, 1994). In most cases ferrochelatase appears to be associated with the membrane (e.g. Harbin and Dailey, 1985; Smith and Kohorn, 1994). The iron incorporated into haem is therefore already reduced in the ultimate step of haem synthesis. As the cysteinyl residues of apocytochrome c are reduced as they exit the ribosome, the reduction status of the two substrates only has to be maintained before ligation. Nevertheless, the milieu of the compartment to which the substrates are secreted may be oxidizing or reducing, and each of the three systems must cope with this environment. It is within these contexts that the three systems have evolved and that the mmicular mechanisms of biogenesis are discussed.
Studies in model Gram-negative bacteria first suggested that cytochrome c biogenesis in prokaryotes occurs by a more complicated pathway than that worked out for fungal mitochondria (Fig. 2). For system I, the present review will introduce the concepts of two functional subpathways, one with proteins involved in haem delivery (HelABCD, Ccl1, and CycJ) and another in apocytochrome presentation and thioreduction (Ccl2, HelX and CycH). These nine proteins have been shown or proposed to be specifically necessary for c-type cytochrome biogenesis in α and γ proteobacteria such as Rhodobacter capsulatus (Kranz, 1989; Beckman et al., 1992; Beckman and Kranz, 1993; Lang et al., 1996), Bradyrhizobium japonicum (Ramseier et al., 1991; Ritz et al., 1993; 1995), and Paracoccus denitrificans (Page and Ferguson, 1995; 1997; Page et al., 1997a). (Note that these genes have also been called ccm, cyc or cyt, but the above nomenclature is based on the first designations published for each gene and will be retained for simplicity in this review. See Thöny-Meyer, 1997; Zumft, 1997 for a compilation of equivalent nomenclature and gene organization for different organisms and full reviews on haem protein assembly in general.) Genomics studies have recently provided an excellent evolutionary framework for discussion of the three systems. Genes for Hel and Ccl proteins are present in the Archae, Archaeoglobus (TIGR database) and Pyrobaculum (S. Fitz-Gibbon, personal communication). Deinococcus, which diverged from the eubacterial branch leading to Gram-positive and -negative organisms, contains at least six of the genes of system I, again suggesting an ancient origin for system I (Goldman and Kranz, 1998). As a relative of the most primitive eukaryote, the protozoan Reclinomonas, contains the helABC and ccl1 genes in its mitochondrial genome (Lang et al., 1997), it is also likely that the first mitochondrion used system I. Plant mitochondria have retained much of system I, with a few interesting modifications noted below. At some point in evolution, however, mitochondria of certain eukaryotes lost these components and evolved a new mechanism of generating c-type cytochromes (system III). Except when relevant, the present review will not involve discussion of evolution or horizontal transfer of the three systems. Rather, an overview of the mmicular mechanisms will be stressed.
A putative haem delivery pathway
Initial mmicular genetic studies on helABC indicated that these genes are required for biogenesis (Kranz, 1989), whereas sequence analysis suggested that they encode the subunits of an ABC (ATP-binding cassette) transporter (Ramseier et al., 1991; Beckman et al., 1992). An important challenge for the field has been to define the substrate transported as well as the macrommicular structure of the transporter. Because apocytochrome c was known to be transported in hel mutants, haem became a likely candidate for the substrate (Beckman et al., 1992). Using immunological methods, the R. capsulatus HelABCD proteins have been shown to form a four-subunit membrane complex of the exporter family (Goldman et al., 1997). Recently, topological analysis of all four subunits using a phoA and lacZ fusion approach has demonstrated that HelB and HelC each comprise six transmembrane helices, whereas the 52 residue HelD protein has a single transmembrane domain (Goldman et al., 1998). The ATP-binding cassette subunit, HelA, is soluble when overproduced but requires both HelB and HelC (but not HelD) for membrane localization (Goldman et al., 1997). These studies have provided the framework to begin the investigation of the structure–function relationships of specific domains and residues in the transporter. For example, within the HelC protein, a highly conserved tryptophan-rich motif, designated the ‘WWD’ domain, is exposed to the periplasm (Goldman et al., 1998). This WWD domain represents a signature sequence that was initially identified in HelC, Ccl1 and a chloroplast ORF (now designated CcsA — see system II below) (Beckman et al., 1992). A central theme for this family of proteins is the hypothesis that this motif is involved in the binding and presentation of haem. In fact, the experimentally determined topology of Ccl1 (11 transmembrane domains) and the Mycobacterium CcsA (six transmembrane domains — see system II) demonstrate that the WWD domains are exposed to the periplasm in these family members as well (Goldman et al., 1998). A second feature common to all three family members (HelC, Ccl1, CcsA) is the conservation of two histidinyl residues within periplasmic domains that bracket the WWD domain. The histidinyl residues could serve as ligands to the haem iron.
Thus, it is proposed that haem is presented by the WWD domain in an oriented fashion with two histidinyl ligands supplied by the protein. Such ligands would prevent reaction with oxygen or oxidation as haem passes through the pathway. This idea has been tested in R. capsulatus by mutagenesis of highly conserved residues in helC and ccl1 (Goldman et al., 1998). Briefly, histidinyl residues are essential for the function of the proteins, whereas tryptophan residues are not. These data support the view that, at least in R. capsulatus, histidine ligation to the iron of haem is required, but that the WWD domain may provide multiple surfaces for interaction/orientation (i.e. robustness). The theme of adjacent histidines as ligands to the iron of haem, in combination with other sites of contact that are often provided by aromatic residues, is observed in many haem proteins, including haemopexin and haemoglobin (e.g. Perutz, 1990; Morgan et al., 1993). Based on the above results, it is proposed that haem travels through HelABCD to Ccl1 where it is tethered to the periplasmic surface with the haem vinyl side-chains exposed. Another gene, called cycJ, has recently been shown to be required for biogenesis in R. capsulatus (M. Sawant and F. Daldal, personal communication). The cycJ gene is linked to the aforementioned genes in some organisms, but not in R. capsulatus (Delgado et al., 1995; Kereszt et al., 1995; Ritz et al., 1995). The CycJ protein is proposed to have a periplasmic, conserved histidinyl residue that could be involved in the relay of haem from HelABCD to Ccl1, but this has not been tested. In fact, there is no direct biochemical evidence that the functions of these proteins are in haem transport (see Future perspectives). Nevertheless, the evidence cited above still makes the haem transport hypothesis the most compelling. This hypothesis is also consistent with the fact that suppressors of R. capsulatus helABCD or ccl1 deletion strains have not yet been isolated (Gabbert et al., 1997; Goldman et al., 1997). These five proteins must play a critical role in biogenesis that cannot be replaced by other gene products.
After the secretion of apocytochrome c, the periplasmically oriented and membrane-bound Ccl1/Ccl2/CycH/HelX proteins are proposed to continue the assembly process of the oxidized apocytochrome (Monika et al., 1997). Ccl2 is anchored to the membrane by a single transmembrane region in the C-terminus and is proposed to reduce the cysteinyl residues of apocytochrome c while presenting it to the haem, which is tethered to Ccl1. Subsequently, the thioredoxin-like protein HelX reduces oxidized Ccl2 (Monika et al., 1997). Evidence that a general thioreduction protein, called DipZ, is involved in biogenesis has come from studies in E. coli (Sambongi and Ferguson, 1994; Crooke and Cole, 1995). It is presently thought that DipZ may reduce HelX either directly or indirectly. Another general thiol protein in the periplasm, DsbC, which is reduced by DipZ (Rietsch et al., 1996), is not required for cytochrome c biogenesis and thus is not an intermediate (Metheringham et al., 1996). The HelX protein is a periplasmically oriented thioredoxin-like protein that is anchored to the membrane via its uncleaved signal sequence (Beckman and Kranz, 1993; Fabianek et al., 1997; Monika et al., 1997; Page and Ferguson, 1997). Each of the two cysteinyl residues on Ccl2 and HelX is required for function and in vivo and in vitro studies demonstrate that HelX reduces Ccl2, which subsequently reduces apocytochrome c (Monika et al., 1997). This scenario represents the specific ‘thioreduction branch’ for biogenesis and requires that the apocytochrome c cysteinyl residues are initially oxidized after secretion through the general secretory components (see above). Thus, a role for the periplasmic oxidoreductase, DsbA, may be to directly oxidize cytochrome c cysteinyl residues as they are secreted (Metheringham et al., 1995). The DsbB protein, which reoxidizes DsbA, is also required for cytochrome c biogenesis (Metheringham et al., 1996). In addition to its role in reduction of apocytochrome c, the Ccl2 protein probably provides for specific recognition and presentation of the apocytochrome (i.e. to haem that is tethered to the Ccl1 protein). That these last functions exist is consistent with the observation that dithiothreitol-dependent suppressors of an R. capsulatus helX deletion mutant occur at high frequency, whereas suppressor mutations do not arise in a ccl2 mutant (Monika et al., 1997). Thus, HelX appears to fulfil a more restricted role than Ccl2.
For CycH, both membrane-spanning domains are necessary for production of all c-type cytochromes. Surprisingly, the C-terminal periplasmic domain of CycH is needed only for the synthesis of c holocytochromes that are soluble or anchored by their signal sequences (Lang et al., 1996). The mmicular basis for this bipartite nature is unknown but it is feasible that CycH assists apocytochrome c polypeptides in their passage to Ccl2/Ccl1, with the periplasmic domain not required for the relay of the C-terminally anchored apocytochrome c1. This result is consistent with the observation that overproduction of the Ccl1 and Ccl2 proteins suppress the cycH phenotype (R. Kranz and F. Daldal, unpublished). That is, CycH may usher apocytochromes to the ultimate step involving Ccl1 and Ccl2. A specific interaction between Ccl1, Ccl2 or CycH has not yet been demonstrated.
Modifications of system I in plant mitochondria
In plants, the apocytochrome c1 precursor protein contains a bipartite presequence characteristic of those involved in targeting to the intermembrane space of fungal mitochondria (Braun et al., 1992). Little is known about the import of apocytochrome c to the intermembrane space, although, as in fungi, no signal appears to be present and only the N-terminal methionine is processed. The mitochondrial genomes of land plants encode proteins showing similarities with the biogenesis proteins of system I that are proposed to be involved in haem delivery. Wheat mitochondria contain orthologues of the two transmembrane proteins of the proposed haem ABC transporter, HelB and HelC, whereas Ccl1N and Ccl1C show similarities with the N-terminal and C-terminal parts of Ccl1 (Gonzalez et al., 1993, G. Bonnard, unpublished results; Bonnard and Grienenberger, 1995). In addition, a CycJ counterpart with a putative mitochondrial targeting sequence is encoded by a nuclear gene in Arabidopsis thaliana (Sadowski et al., 1996). Analysis of the sequenced mitochondrial genomes of land plants Marchantia polymorpha (Oda et al., 1992) and A. thaliana (Unseld et al., 1997) does not reveal genes other than helB, helC and ccl1 that are similar to bacterial ones. This suggests that the other genes might have been transferred to the nuclear genome as has been proposed for the helA counterpart. In all land plants studied to date, HelB and HelC have mmicular mass and hydrophobicity profiles similar to the bacterial proteins and share an overall sequence similarity of 50% with bacterial Hel proteins (Schuster, 1994; Bonnard and Grienenberger, 1995; Jekabsons and Schuster, 1995; Nakazono et al., 1996; Shikanai et al., 1996). In HelC, this similarity increases to 86% in the 43 residue WWD domain. The organization of genes coding for Ccl1 counterparts is variable in plant mitochondria. In contrast to the situation found in wheat and Oenothera berteriana (Schuster et al., 1993; Gruska et al., 1995), three genes encode proteins showing similarity with Ccl1 in the lower plant M. polymorpha (Oda et al., 1992). In A. thaliana and rapeseed, the expression of a split ccl1N gene has been discussed (Handa et al., 1996; Menassa et al., 1997). A number of transmembrane domains can be predicted for the mitochondrial proteins, indicating a membrane topology similar to the bacterial Ccl1. However, conserved domains are interspersed by plant-specific sequences. Similar to what is found in the HelC protein, the highest conservation in Ccl1N is observed for the WWD domain and the surrounding 120 residues with 78% similarity, emphasizing their functional significance. C→U mRNA editing is crucial for the preservation of the WWD motif in both HelC and Ccl1N. As described above, histidinyl residues in the HelC, Ccl1 and CycJ counterparts from plants are conserved. The thio–redox pathway, and in particular a specificity-conferring protein similar to Ccl2, has not yet been discovered in plants.
A model for a second pathway of cytochrome c biogenesis has emerged from studies on chloroplasts of Chlamydomonas reinhardtii, a unicellular green alga, and Bacillus subtilis. Photosynthesis in chloroplasts and cyanobacteria requires the cytochrome b6/f complex, in which cytochrome f is a c-type cytochrome (Fig. 1A). In some organisms, the soluble cytochrome c6 replaces plastocyanin function under copper-deficient conditions. In B. subtilis, as in other bacteria, cytochrome c functions in respiratory pathways.
A putative haem delivery pathway
The C-terminal third of the chloroplast CcsA protein (formerly called Ycf5) was noted to show striking sequence identity to the tryptophan-rich WWD domain of Ccl1 and HelC (see above and Beckman et al., 1992). The corresponding ccsA locus in C. reinhardtii was recently shown to be required for haem attachment to both apocytochromes c6 and f (Xie and Merchant, 1996). Besides ccsA, mutations in at least five nuclear genes yield the same pleiotropic c-type cytochrome deficiency in the chloroplast (Xie et al., 1998). One of the nuclear genes, Ccs1, has been cloned, but a putative function is not obvious from the sequence (Inoue et al., 1997). The gene product, found in some algal plastomes (Ycf44), is not as well conserved as is CcsA. The Ccs1 protein is predicted to contain at least one transmembrane domain with a large C-terminal soluble domain. The CcsA and Ccs1 proteins are suggested to function together in a complex as ccsA mutant strains do not accumulate Ccs1 and ccs2, ccs3 and ccs4 strains accumulate reduced amounts (B. Dreyfuss, unpublished). Analyses of the genome databases (12/97) suggests that the CcsA–Ccs1 components are more related to open reading frames in Gram-positive bacteria (B. subtilis, Mycobacterium tuberculosis and M. leprae) than to cytochrome assembly components in α and γ proteobacteria or fungal and mammalian mitochondria.
Orthologues of Ccs1 and CcsA in B. subtilis are called ResB and ResC, respectively, owing to their occurrence in an operon containing regulators of the pet/qcr genes encoding the respiratory bc1 complex (Sun et al., 1996). The specific biochemical function of the Ccs factors has not been deduced, but recently the membrane arrangement of the Mycobacterium CcsA protein has been determined experimentally (Goldman et al., 1998). It contains a canonical pattern for a transporter with six transmembrane domains. As noted above, the highly conserved WWD domain and adjacent histidines are positioned at the surface of the membrane, as might be predicted if CcsA were replacing the function of the HelABCD→Ccl1 haem delivery pathway of system I. The membrane domain of Ccs1 might conceivably be involved in transporter function, or, alternatively, it might simply serve a structural role for the interaction of other assembly components.
A thioreduction–oxidation pathway
For bacteria with system II, all c-type cytochromes possess typical Sec-dependent signal sequences and are presumably translocated as described for prokaryotes using system I. Proteins that are destined for the chloroplast lumen contain one or two targeting signals depending on whether they are plastid or nuclear encoded (for review see Kouranov and Schnell, 1996). Nuclear-encoded thylakoid proteins use a stroma-targeting domain that is recognized by the translocation complex in the chloroplast envelope membrane. This is cleaved off upon entry into the stroma. Multiple protein translocases function in the thylakoid membrane, including one related to the bacterial Sec machinery, another related to the signal recognition particle-dependent machinery, and a third, which is called the pH-dependent pathway (for review, see Cline and Henry, 1996; Settles et al., 1997). Cytochrome f is encoded as a precursor by the plastid petA gene, and is co-translationally inserted into the thylakoid membrane (for review, see Howe and Merchant, 1994a; Perret et al., 1998)). Biochemical and genetic evidence suggest that pre-cytochrome f uses the Sec machinery (Nohara et al., 1996; Voelker et al., 1997); however, the SRP and pH pathways may also be used in parallel (High et al., 1997; Settles et al., 1997). The N-terminal signal sequence, which is required for insertion of the protein into the thylakoid membrane (Smith and Kohorn, 1994), is cleaved on the luminal side of the thylakoid membrane at a consensus Ala–Xxx–Ala-processing site to generate the mature N-terminus. The processing site is also recognized by bacterial leader peptidase (Anderson and Gray, 1991). The amino group of the N-terminal tyrosine serves as one of the axial ligands to the haem iron (Martinez et al., 1994). Processing is therefore a prerequisite for the assembly of functional holocytochrome f. In contrast to the situation for cyt c1 maturation in mitochondria (see system III below), processing of precytochrome f does not depend on haem attachment (Howe et al., 1995; Kuras et al., 1995). As in most other systems, the apoproteins are degraded and therefore do not accumulate. Cytochrome c6 is encoded in the nucleus with a typical two-domain targeting sequence. The preprotein is processed sequentially as the precursor is translocated across the chloroplast envelope and then the thylakoid membrane (Howe and Merchant, 1993; reviewed by Cline and Henry, 1996). As for cytochrome f, the import and processing of preapocytochrome c6 occur independently of haem attachment. If the apoprotein is not converted to the holoform it is degraded rapidly (Howe and Merchant, 1994b).
The ResA protein encoded by the Bacillus resABC operon is related to HelX and the thioredoxin family of thiol-disulphide oxidoreductases, which suggests that it might be required for reduction of apocytochrome c making it functionally equivalent to HelX/Ccl2 from system I. Recently, the ccdA gene (not linked to the res operon in B. subtilis) was identified as being required at a late stage in the maturation of all c-type cytochromes of B. subtilis (Schiött et al., 1997). CcdA is related to the disulphide isomerase protein DipZ discussed under System I and has been hypothesized as being involved in the transfer of reducing equivalents to the p-side of the energy-transducing membrane. Homologues of CcdA are present in some chloroplast genomes (in one case linked to Ccs1), a cyanobacterial genome, and other Gram-positive bacteria (in which they are sometimes linked to ResABC homologues). Although no evidence for thioreduction activities of CcdA or ResA has been published, a CcdA/ResA/apocytochrome c pathway seems plausible based on analogies to the system I pathway described above. Topological, mutagenesis and in vitro analysis will be necessary to substantiate these ideas.
Taken together, these results suggest that at least these four proteins, CcsA, Ccs1, ResA and CcdA, are used in the Gram-positive bacteria and plastid system for cytochrome c biogenesis. Surprisingly, analysis of the genome of Helicobacter pylori (Tomb et al., 1997), a Gram-negative ε-proteobacterium, has uncovered candidate homologues of the system II genes but not system I genes. As in B. subtilis, Ccs1, CcsA and ResA are linked; in fact, candidate Ccs1 and CcsA orthologues in Helicobacter are fused in a single open reading frame, which supports the model that they form a functional complex, perhaps via interactions through the transmembrane domain of Ccs1 (Goldman and Kranz, 1998).
Cytochrome c haem lyases as central components of c-type cytochrome biogenesis
In fungal, vertebrate and invertebrate mitochondria (system III), the biogenesis of c-type cytochromes appears to be far less complex than in the previously mentioned systems. Intensive genetic and biochemical studies have been performed in the last decade using the fungi Saccharomyces cerevisiae and Neurospora crassa. Cytochrome haem lyases (also known as holocytochrome synthases) are the central components of biogenesis in system III. Genetic studies have identified two types of related enzymes differing in their substrate specificity. Cytochrome c haem lyase (CCHL) of yeast (Dumont et al., 1987) and N. crassa (Drygas et al., 1989) are required for covalent haem attachment to apocytochrome c, whereas cytochrome c1 haem lyase (CC1HL) of yeast (Zollner et al., 1992) was found to be involved in the maturation of cytochrome c1, an integral component of the bc1 complex of the electron transport chain. Orthologues of fungal cytochrome haem lyases have been identified in genome sequencing projects from various species including Homo sapiens and nematodes (see Steiner et al., 1996). These lyases are essential for covalent haem linkage (Dumont et al., 1988; Nicholson et al., 1988) and have been shown to interact directly with both haem and the apocytochromes (Mayer et al., 1995; Steiner et al., 1996). Therefore, these proteins are believed to function as enzymes catalysing the covalent attachment of haem to the apocytochromes. However, biochemical proof for a direct enzymatic function of the purified cytochrome haem lyases is still lacking. It is also unknown whether other accessory factors are required for the conversion to the holoproteins.
One such accessory factor was proposed to be Cyc2p, a protein of the mitochondrial inner membrane of Saccharomyces (Dumont et al., 1993). However, the role of this genetically identified component in holocytochrome formation is unclear at present. Although deletion of a C-terminal segment of CYC2 resulted in a specific decrease in the content of holocytochrome c (Dumont et al., 1993), removal of the entire CYC2 gene had a pleiotropic effect on various mitochondrial functions including the biogenesis of cytochrome c (Dumont, 1996, F. Sherman, personal communication). Therefore, Cyc2p is now believed to play a more general role in mitochondrial biogenesis and to influence cytochrome synthesis only indirectly.
Pioneering biochemical investigations carried out by the groups of Neupert and Sherman have elucidated important aspects of the mechanism of covalent haem attachment to the apocytochromes (reviewed in Stuart and Neupert, 1990; Dumont, 1996). Haem linkage to apocytochrome c could be reconstituted with isolated mitochondria after importing the apoprotein into the intermembrane space. Holocytochrome c was not formed in mutants defective in CCHL emphasizing the importance of this protein for haem attachment. Similar findings were made for detergent extracts of mitochondria, suggesting that the haem attachment reaction does not depend on the integrity of the organelle. The biogenesis of cytochrome c1 is far more complex (see, e.g., Glick et al., 1992; Arnold et al., 1998). Processing of the intermediate form of cytochrome c1 to the mature protein by inner membrane protease 2 (Imp2), requires preceding haem attachment (Ohashi et al., 1982; Nicholson et al., 1989) that is mediated by CC1HL (Zollner et al., 1992). Conceivably, signal cleavage by Imp2 depends on a conformation of the intermediate form of cytochrome c1, which is adopted after covalent modification with haem. The formation of holocytochromes c and c1 requires the reduction of haem (Nicholson and Neupert, 1989; Nicholson et al., 1989). In crude extracts, this reduction can be mediated by NADH and reduced flavin nucleotides (Nicholson and Neupert, 1989). The mechanism for reduction in vivo is not yet determined.
Membrane transport of apocytochrome c
Import of apocytochrome c, unlike all other known nucleus-encoded mitochondrial preproteins, does not require the translocase of the outer membrane of mitochondria (the TOM complex, see Lill and Neupert, 1996) or other protease-sensitive factors of the outer membrane (Mayer et al., 1995). Acidic phospholipids are necessary for membrane binding and transport (see review by de Kruijff et al., 1992). Wang et al. (1996) have elucidated some of the apocytochrome c sequence requirements for mitochondrial import. Earlier studies had suggested a crucial role for CCHL in the import from the cytosol into the intermembrane space. The levels of apocytochrome c transport correlated with the levels of CCHL in the cell (Dumont et al., 1988; 1991; Nargang et al., 1988). The import mechanism of apocytochrome c was further clarified using purified outer membrane vesicles that lack CCHL (Mayer et al., 1995). Import of apocytochrome c into vesicles is not observed unless a mechanism to trap the apocytochromes is provided. This was accomplished by enclosing apocytochrome c antibodies in the lumen of the vesicles. In intact mitochondria, the role of this binding partner on the trans side of the outer membrane is performed by CCHL, which was found to directly associate with the translocated apoprotein (Mayer et al., 1995).
Besides direct binding to CCHL, another factor shifting the equilibrium of the import reaction involves folding of apocytochrome c in the intermembrane space. Subsequent to synthesis on cytoplasmic ribosomes, the apoprotein has a protease-sensitive unfolded conformation that may be important for penetration of the outer membrane (Jordi et al., 1992). After the binding of apocytochrome c to CCHL within the intermembrane space (Nicholson et al., 1988), a protease-resistant, native form appears after covalent attachment of haem. This might be the trigger for dissociation of the cytochrome-CCHL complex.
A reversible haem-binding motif in haem lyases
The members of the ‘haem lyase family’ share about 35% identical amino acid residues (50% similarity, see Steiner et al., 1996). The homology between the enzymes is particularly striking in the C-terminal two-thirds that are essential for enzymatic function. In the N-terminal third, there is no detectable similarity except for a highly conserved sequence, the so-called ‘CPV motif’ that is present in one (CC1HL) to three (CCHL) copies. This short sequence motif has been shown recently to serve as a reversible haem-binding site of these lyases (Steiner et al., 1996) as well as several other haem-binding proteins such as the transcription factor Hap1p and haem oxygenase-2 (Zhang and Guarente, 1995; McCoubrey et al., 1997). Surprisingly, CC1HL retained residual catalytic activity when the CPV motif was deleted or mutated, leading to the suggestion that a second haem-binding site exists on CC1HL (Steiner et al., 1996). Haem oxygenase-2 is also proposed to contain two different haem-binding sites (McCoubrey et al., 1997). Understanding the regions within the CCHL that are important for specific functions provide interesting objectives for further study of system III.
Key features of the biogenesis of c-type cytochromes in system III are different from those in systems I and II. However, the lower complexity of system III with just the CCHL as the central component of biogenesis may be deceptive. Some components may have escaped genetic identification because they may not be specific for the biogenesis of cytochromes and may also participate in other aspects of mitochondrial biogenesis. Mutation of these proteins will probably be associated with pleiotropic phenotypes. Examples for such components yet to be identified are (i) transporters for haem and iron in the mitochondrial inner membrane, (ii) proteins that ensure diffusion of hydrophobic haem through the soluble compartments, and (iii) disulphide and haem reduction components in the intermembrane space.
Future perspectives on mmicular mechanisms and evolution
Although the possible evolutionary divergence and horizontal transfer of the three systems has been addressed recently (Goldman and Kranz, 1998), the environmental and physiological constraints that select for the three systems have not been determined. In the Gram-negative α and γ proteobacteria it is well accepted that directed pathways are needed in the periplasm for the formation of disulphide-bonded proteins (see Raina and Missiakis, 1997). We now know that pathways also exist for the specific reduction of thiol-proteins such as cytochrome c (system I). It is unclear whether system II evolved from system I or vice versa, but the common elements for these two systems (e.g. haem delivery/protection and thioreduction) may indicate that the major physiological constraints are similar for each. Given that system I and/or II were present before the formation of oxygen on earth, it is worth considering that these specific pathways also protect against other haem- and thioreactive compounds (e.g. HCN, metals, organic oxidants). Ferguson and colleagues have recently suggested that a need for system I compared to system II may be based on the disulphide bond formation of cytochrome c by DsbA/B before reduction (Metheringham et al., 1995; Page et al., 1997b). This view may be oversimplified as, for example, Deinococcus contains system I but not a DsbA/B pathway as we know it. In fact, we have recently discovered that the β proteobacteria Neisseria meningitidis (Sanger Centre Projects database) has system II and DsbA/B proteins (B. Goldman and R. Kranz, unpublished). It is possible that a more complex system may be required to ligate haem to many different, and in some cases multihaem, types of c-type cytochromes. For these multihaem proteins, the function of sequential oxidation and reduction might be to ensure that correct pairs of cysteines are presented to the haem attachment complex. More biochemical, physiological, and genomics information are needed before the advantages or disadvantages of each system become apparent.
If we assume that the evolution and retention of the complex pathways represented by systems I and II are due to a need to protect biogenesis from oxidizing and rapidly changing environments, then predictions can be made on the evolution of system III. That is, organisms using system III may have developed mechanisms to generically buffer the intermembrane space from oxidizing mmicules, at least within the time frame that such mmicules may be detrimental to cytochrome c biogenesis. However, much more work is needed on the oxidation/reduction status of the compartments in which c-type cytochromes reside to make definitive conclusions on this matter (Fig. 1A). A particular dearth of knowledge concerns the mitochondria from microbial eukaryotes (other than fungi and Reclinomonas). Lower eukaryotes represent tremendous evolutionary and physiological diversity. It is possible that determining the system present in these organisms will offer clues to the milieu within their intermembrane space and the evolution of system III.
A topic of future interest is the specificity that is built into the three systems. For example, systems I and II appear to recognize a variety of secreted apoproteins containing the C–X–X–C–H motif, including heterologous cytochromes c and multihaem c-type cytochromes. Recently, Cole and colleagues have analysed a multihaem c-type cytochrome (NrfA or cytochrome c552) involved in dissimilatory nitrite reduction in E. coli (Eaves et al., 1998). Cytochrome c552 contains four typical C–X–X–C–H motifs with haems attached through the HelABCD/CycJ/Ccl12/CycH system I pathway (CcmABCDEFGH, see Grove et al., 1996). Surprisingly, a fifth motif with residues C–W–S–C–K also has covalent haem, probably representing the site of nitrite reduction. It was further shown that a second set of genes that includes nrfEFG (similar to the ccl1, ccl2 and cycH genes) are required for ligation of haem to the C–W–S–C–K motif. Thus, a subset of dedicated system I proteins (NrfEFG) may have evolved to specifically recognize this unique C–W–S–C–K apocytochrome c motif. It has been known for some time that the CCHL and CC1HL enzymes have a narrow substrate specificity (see System III above). Defining the recognition elements in both the substrates and the assembly factors (e.g. Ccl2 or CCHL) and the exact mechanisms for achieving specificity remain important goals.
The potential existence and role of intracellular haem delivery proteins is now better appreciated. Nevertheless, the isolation and biochemical reconstitution of members of the WWD domain family is necessary to directly prove a role in haem delivery. Although the putative transmembrane delivery proteins discussed for system I and II appear to be specific for the biogenesis of pathways of c cytochromes, it is usually assumed that haem required for the synthesis of other haem proteins diffuses directly to the apoprotein, either before folding (e.g. fungal mitochondria via CCHL) or after folding (e.g. haemoglobin and b-type cytochromes) (Goldman et al., 1996). Nevertheless, it is likely that other specific or general transmembrane haem delivery systems are yet to be discovered (e.g. Kuras et al., 1997). The detailed mmicular basis for the delivery of haem in all three systems will be fertile ground for future research.
Finally, in certain systems it is likely that even more factors will be discovered. In fact, the minimal set of factors necessary for the in vitro synthesis of a c-type cytochrome has never been determined. As with many cellular processes, the development of rapid, highly sensitive in vitro assays to study biogenesis mechanisms could be considered the Holy Grail for this field. Only with extracts from Neurospora and Saccharomyces (system III) have such assays been successful. For system I (and probably system II) it is particularly noteworthy that all of the factors required specifically for biogenesis are integral membrane or membrane-associated proteins. Although this presents technological challenges for the future, it can be rationalized from a physiological perspective — by tethering a reaction pathway to the surface of a membrane, the likelihood of interaction is increased by many orders of magnitude (cited in Gelb, 1997). In vitro reconstitution will be invaluable for the mmicular dissection of the haem linkage reaction and for the eventual definition of all the necessary factors. Questions concerning the chemistry and substrate recognition determinants of the reaction will then be resolvable.
We would like to thank Scott Hultgren and Gabe Soto for help with the graphic interfaces and Fevzi Daldal for personal communications and discussions. R.G.K. is supported by NIH GM 47909, B.S.G. is supported in part by a Monsanto post-doctoral fellowship, S.M. by NIH GM 48350, R.L. by Sonderforschungsbereiche 184 and 286 of Deutsche Forschungsgemeinschaft, G.B. by CNRS.