In chloroplasts, binding of a c′-heme to cytochrome b6 on the stromal side of the thylakoid membranes requires a specific mechanism distinct from the one at work for c-heme binding to cytochromes f and c6 on the lumenal side of membranes. Here, we show that the major protein components of this pathway, the CCBs, are bona fide transmembrane proteins. We demonstrate their association in a series of hetero-oligomeric complexes, some of which interact transiently with cytochrome b6 in the process of heme delivery to the apoprotein. In addition, we provide preliminary evidence for functional assembly of cytochrome b6f complexes even in the absence of c′-heme binding to cytochrome b6. Finally, we present a sequential model for apo- to holo-cytochrome b6 maturation integrated within the assembly pathway of b6f complexes in the thylakoid membranes.
Heme (Fe-protoporphyrin IX) is an essential cofactor for many fundamental biological processes. It is a hydrophobic and cytotoxic macrocycle, which therefore requires specific pathways to be safely delivered to its subcellular destinations. The c-type cytochromes (cyts) are most frequently characterized by the covalent attachment of heme to the protein by two thioether bonds formed between the heme vinyl groups and the thiols of two cysteines in a conserved CXXCH motif, with histidine as one of the axial ligands to the heme iron. Binding of c-heme comprises several critical steps including the reduction of cysteine residues, ferrous heme-supply, and heme-binding by itself. Factors involved in covalent heme binding have been extensively studied in prokaryotes and eukaryotes, and this maturation process usually occurs through one of three multifactorial systems (I, II, and III), depending on the biological organism and cellular compartment that are known to operate on the positive side (p side) of the membrane facing the periplasmic/intermembrane/lumenal space (Kranz et al., 1998; Thöny-Meyer, 2002; Allen et al., 2008; Ferguson et al., 2008; Giege et al., 2008; Hamel et al., 2009). It has been recently recognized that there are in fact two independent c-heme maturation pathways in the chloroplast: system II, also referred to as cyt c synthesis (CCS) on the p side of membranes (Xie and Merchant, 1996; Inoue et al., 1997); and the newly characterized system IV, also referred to as cofactor assembly on complex C subunit B (CCB) on the negative side (n side) of photosynthetic membranes facing the chloroplast stroma or cyanobacterial cytoplasm (Kuras et al., 2007).
Cyt bc complexes that associate c-type hemes, b-type hemes, and iron–sulfur centers are ubiquitous in energy-transducing membranes, where they transfer electrons from a lipid-soluble carrier to a water-soluble protein carrier, while building up a transmembrane proton gradient. Although quite similar to the bc1 complex, its homologue found in bacteria and mitochondria, the b6 f complex contains additional cofactors. Six years ago, the discovery of an additional heme bound to the protein backbone of cyt b6 (Kurisu et al., 2003; Stroebel et al., 2003) came as a major surprise. This new c-heme showed unusual features. It is covalently attached by a unique thioether bond to the conserved Cys-35 of the b6 subunit (Chlamydomonas reinhardtii numbering), and thus belongs to the c-type family. It is a high spin c′-heme, with a pentacoordinated central iron. The usual amino acid axial ligands of the heme iron are missing, but a water molecule (or hydroxide ion) acts as a unique axial ligand and bridges the heme iron to a propionate side chain of bH-heme. Furthermore, ci-heme is attached on the membrane n side in the quinone-binding pocket Qi in very close proximity to bH-heme, opposite to the location of f-heme and c6-heme that are on the membrane p side.
Our previous genetic studies in C. reinhardti led to the identification of four nuclear CCB loci that control the process of apo- to holo-cyt b6 conversion that involves binding of ci-heme (Gumpel et al., 1995; Kuras et al., 1997, 2007; de Vitry et al., 2004). The signature of cyt b6 impaired in ci-heme binding was characterized in the petB-C35V mutant (the petB gene codes for cyt b6), where the unique covalent cysteine ligand to the heme was substituted by a valine (de Vitry et al., 2004). It consists of a double band on urea-SDS-polyacrylamide gels or a single band of lower apparent molecular mass than in the wild type (WT) on SDS-polyacrylamide gels that shows no peroxidase activity. The same signature is observed in all ccb mutants, which indicates that they do not undergo binding of ci-heme to cyt b6. This entirely new biogenesis pathway comprises at least the four CCB factors, which are present in all photosynthetic organisms containing b6 f complexes (available under GenBank/EMBL/DDBJ accession nos. EF190472, EF190473, EF190474, and EF190475), for which no function had been previously ascribed (Kuras et al., 2007). CCB1 is ubiquitous among organisms performing oxygenic photosynthesis, where it is present as a unique orthologue. CCB3, which is the most conserved member of the CCB pathway, belongs to the large YGGT protein family (InterPro accession no. IPR003425) found in both plastids and bacteria. CCB2 and CCB4 are encoded by paralogue genes in C. reinhardtii with a low similarity level and derived from a unique cyanobacterial ancestor. We recently demonstrated the conserved function of the CCBs in Arabidopsis thaliana chloroplasts, where they are also involved in ci-heme binding to cyt b6 (Lezhneva et al., 2008).
Here, using C. reinhardtii as an experimental system for the investigation of the CCB maturation pathway, we extended our previous biochemical work that established the chloroplast membrane localization of CCBs (Kuras et al., 2007). We examined the topology and interactions between the CCBs as well as with cyt b6 and demonstrate that the CCB pathway is organized within the thylakoid membrane in a series of distinct protein complexes that control the covalent binding of ci-heme to cyt b6.
CCBs are transmembrane proteins that resist extraction with chaotropic agents
In our previous work, the CCB1–4 proteins were immunodetected as chloroplast membrane proteins after subcellular fractionation in C. reinhardtii (Kuras et al., 2007). Using fluorescent protein reporters, we also demonstrated that the CCB1–4 proteins are targeted to the chloroplast in A. thaliana (Lezhneva et al., 2008).
We further analyzed the transmembrane topologies of these proteins in C. reinhardtii by membrane treatment with chaotropic agents and by subjecting oriented thylakoid membranes vesicles to protease treatments followed by urea/SDS-PAGE and immunodetection with anti-CCB peptide antibodies. When membranes were treated with various chaotropic agents or incubated at high ionic strength or alkaline pH, a typical peripheral membrane polypeptide (oxygen evolution enhancer 2 protein of photosystem [PS] II [OEE2]) was released in the supernatant (Fig. 1 a). In contrast, cyt f was not extracted by these treatments, as expected from its transmembrane domain (TMD). All CCB factors behaved as bona fide transmembrane proteins. Although CCB3 resisted extraction at high ionic strength and alkaline pH, we noted that some CCB3 dissociated from the membrane with the chaotropic agent KSCN, which is intriguing for such a highly hydrophobic protein.
The transmembrane topology of the CCBs was assessed by protease treatment of French press–derived membrane vesicles. Membrane vesicles subjected or not subjected to sonication were incubated in the presence of trypsin (Fig. 1 b). Before sonication, most of the thylakoid vesicles were in a right-side-out orientation, and exogenous proteases have no access to their lumenal side. In contrast, upon sonication, the vesicles burst, and exogenous proteases gain access to both sides of the membranes. Typically, the PSII subunit OEE3, used here as a control for peripheral proteins on the lumenal side of the membranes, is degraded only after sonication, when using trypsin concentrations below 10 μg/ml. In contrast, CCB1, CCB2, and CCB4 proteins are degraded by trypsin treatments below 10 μg/ml even in the absence of sonication. Only CCB3 required higher trypsin concentration (100 μg/ml) for degradation, confirming its deep embedment in the membrane. The lumenal localization of the C terminus of CCB3 was supported by the lack of detection of a CCB3 digestion product corresponding to the mature protein missing the three C-terminal amino acid residues. Therefore, we conclude that the four CCB factors have stromal-exposed segments containing Lys and Arg residues as schematized in the transmembrane topology of the CCBs in Fig. 1 c (which also shows the localization of the epitopes against which our antibodies were prepared).
Concerted accumulation of CCB2 and CCB4 in ccb mutants
Each of the CCBs—CCB1 (24 kD), CCB2 (27 kD), CCB3 (13 kD), and CCB4 (28 kD)—is absent in the corresponding ccb mutant (Kuras et al., 2007). In a first attempt to identify possible interactions between CCB factors, we wondered whether the absence of one CCB factor in the corresponding ccb mutant would alter the accumulation pattern of the other CCBs. CCB1 and CCB3 accumulated to the WT level, independently of the presence or absence of the other CCBs (Fig. 2). The accumulation of CCB2 and CCB4 was similar in WT, ccb1, and ccb3 mutants, and therefore independent of CCB1 and CCB3 accumulation (Fig. 2). However, CCB4 was barely detectable in the ccb2 mutant, and, conversely, CCB2 was barely detectable in the ccb4 mutant (Fig. 2). Thus, CCB2 and CCB4 show a concerted accumulation that suggests they form a heterodimer. We then looked at whether the presence of cyt b6, as a substrate for ci-heme binding, controlled the accumulation of the CCBs. We observed that the four CCBs still accumulated to the WT level in a mutant lacking cyt b6 (ΔpetB) or in the petB-C35V mutant that bears a point mutation in the sequence of cyt b6 that prevents ci-heme binding (Fig. 2; de Vitry et al., 2004).
Direct interaction of CCB2 and CCB4 in yeast split ubiquitin two-hybrid assays
We further investigated CCB2–CCB4 interactions using the yeast split ubiquitin system that is appropriate for testing interactions between two transmembrane proteins (as chloroplast membrane proteins [Pasch et al., 2005] and components of the cyt c maturation system I [Rayapuram et al., 2008]). Their interaction allows reconstitution of an active ubiquitin from its N and C terminus domains (Nub and Cub), fused, respectively, to each protein. The restored ubiquitin activity cleaves a transcription factor fused to Cub, thereby releasing auxotrophy for some particular nutrients.
Yeast cotransformed with CCB2 and CCB4 were able to grow on a very stringent medium for two reciprocal couples, either CCB2 fused to NubG with CCB4 fused to Cub (CCB2-N/CCB4-C) or CCB4 fused to NubG with CCB2 fused to Cub (CCB4-N/CCB2-C; Fig. 3 a). The interaction between CCB2 and CCB4 was confirmed by β-galactosidase activity assays (Fig. 3 b). The strength of the interaction between CCB2 and CCB4 was similar for the two reciprocal sets of constructs. In contrast, CCB2 fused both to NubG and to Cub or CCB4 fused both to NubG and to Cub (respectively, CCB2-N/CCB2-C and CCB4-N/CCB4-C) did not allow better growth than observed in control strains (yeast cotransformed with noninteracting Alg5 fused to NubG or Cub). These results argue for hetero-oligomerization of CCB2 with CCB4 with no homo-interaction between two copies of CCB2 or of CCB4. We could not find any evidence for a hetero-oligomerization between separate domains of CCB2 and CCB4, whether we tested soluble or TMDs of each protein (not depicted). With this technique, we also found no evidence for an interaction between CCB1 and CCB2 or CCB4 (not depicted). Finally, no interactions between any of these three CCBs and cyt b6 could be detected in split ubiquitin assays (unpublished data). We note, however, that a positive control for the expression of CCB1, using cotransformation of the C terminus of CCB1 fused to Cub and Alg5 fused to NubI, further supported the localization of the C terminus of CCB1 in the cytosol (not depicted). The transmembrane configuration of CCB3 with its N and C termini in the lumen prevented us from studying CCB3 interactions in the split ubiquitin system, which requires an N or C terminus of the protein of interest to be present in the yeast cytoplasm.
CCB complexes analyzed on blue native (BN)-PAGE
To get a broader picture of the variety of membrane protein complexes that may contain CCBs, we separated oligomeric proteins on BN-PAGE, after a mild digitonin solubilization of thylakoid membranes, using a procedure similar to the one that has been developed for membrane protein complexes from mitochondria (Schägger and von Jagow, 1991; Meyer et al., 2005) and from chloroplasts (Lennartz et al., 2006). The constitutive subunits of the oligomers were separated in a second dimension on SDS-PAGE, followed by immunoblotting with antibodies raised against each CCB or against some of the major b6 f complex subunits. Fig. 4 compares BN-PAGE assays in the WT and in a series of mutant strains defective for ci-heme binding to cyt b6. In agreement with the concerted accumulation and split ubiquitin experiments that demonstrated extensive interactions between CCB2 and CCB4, we observed that these two CCB components comigrate at 70 kD by BN-PAGE (Fig. 4, circles) in all strains except in the ccb2 mutant (Fig. 4) and ccb4 mutant (not depicted), which lack these two proteins. Therefore, we attribute this 70-kD complex to the CCB2/CCB4 heterodimer. Monomeric forms of CCB2 and CCB4 that probably result from dissociation of the heterodimer were also detected in these experiments, with an apparent molecular mass of ∼45 kD and 50 kD, respectively.
Among the various ci-heme–binding mutants that we assayed in BN-PAGE, the petB-C35V mutant, which is specifically blocked at the last stage of the binding process due to the replacement of the cysteine covalent ligand of the heme by an inactive valine, is the only one that has the same CCB complement as in the WT. A conspicuous feature of the immunodetection pattern in petB-C35V is the increased detection of several CCB-containing oligomers that were present only in trace amounts in the WT. This applies to oligomeric forms of CCB1 in the 70–110 kD region (Fig. 4, dashed boxes) and to oligomeric forms of CCB2/4/3 around 170 kD (Fig. 4, solid boxes). These observations suggest that these protein complexes are transiently formed in the process of heme delivery to cyt b6 in the WT but accumulate more extensively in the petB-C35V mutant because the last step of heme binding is impaired. The CCB1-containing oligomer at 70–110 kD remained visible in ccb2, ccb4, and ccb3 mutants but was lost in the ΔpetB strain that lacks cyt b6 (apocytochrome b6 24 kD), where CCB1 migrated only as a monomer of ∼45 kD of apparent molecular mass. Interestingly, an oligomeric form of cyt b6, which was visible in the same 70–110-kD region in the WT and in petB-C35V, ccb2, ccb4, and ccb3 mutants, was depleted in the ccb1 mutant. These migration patterns support the existence of a CCB1/b6 oligomer. Oligomeric forms of CCB2, CCB3, and CCB4 at 170 kD disappear altogether in the ccb2, ccb3, and ccb4 mutants. This behavior is consistent with their association in a 170-kD complex. This 170-kD oligomer is also absent in the ΔpetB mutant, an observation that is consistent with an interaction between the CCB2/CCB4/CCB3 oligomer and cyt b6. We note that this complex is also absent in the ccb1 mutant, although we found no evidence for the presence of CCB1 in the 170-kD region. Because formation of the CCB1/b6 oligomer still occurs in the ccb2, ccb3, and ccb4 mutants, we suggest that formation of the CCB1/b6 oligomer is a prerequisite for the formation of the CCB2/CCB4/CCB3 oligomer in the process of ci-heme binding to cyt b6.
Interactions between the CCBs and cyt b6, as analyzed by coimmunoprecipitation
To gather independent evidence for the association of several CCB factors in these oligomeric complexes, some of which likely contain cyt b6, we undertook complementary coimmunoprecipitation assays using antibodies raised against CCB3 peptides (α-CCB3), the N-terminal domain of subunit IV (α-IV), and the N-terminal (α-b6N) and C-terminal domains (α-b6C) of cyt b6 (Fig. 5). These antibodies detect the proteins against which they were prepared and not the other b6 f subunits or CCBs (controls not depicted). We used digitonin-solubilized membranes (as prepared for BN-PAGE) from the petB-C35V strain, as it shows the most abundant CCB oligomeric complexes in BN-PAGE. Coimmunoprecipitations were then assayed with polyclonal antibodies against peptides from cyt f, cyt b6, subunit IV, CCB1, CCB2, CCB3, and CCB4.
Fig. 5 a shows that α-CCB3 coimmunoprecipitates CCB2, CCB4, and cyt b6 but not CCB1, subunit IV, and cyt f. These results confirm that CCB3 indeed interacts with CCB2, CCB4, and cyt b6, as suggested by the existence of the 170-kD oligomeric form of these three CCB components in BN-PAGE. That α-CCB3 pulls down cyt b6 but not the other b6 f subunits and that α-IV pulls down b6 f subunits but no CCB components (Fig. 5 a) indicates that CCBs interact with cyt b6 before its assembly within the b6 f complex.
Coimmunoprecipitations with α-b6C and α-b6N pulled down subunit IV and cyt f as expected from the formation of b6 f complexes in the petB-C35V mutant (see the following results section). In addition, α-b6C coimmunoprecipitated the four CCBs, whereas α-b6N coimmunoprecipitated only CCB3 out of the four CCBs (Fig. 5 b). These results support the interaction of the CCB2/CCB4/CCB3 oligomer with cyt b6, as suggested by our BN-PAGE study. They also confirm the formation of CCB1/b6 and CCB3/b6 complexes. When using WT membranes, α-b6N and α-b6C coimmunoprecipitated subunit IV and cyt f but little if any CCBs (unpublished data), which is consistent with the low amount of CCB hetero-oligomers interacting with cyt b6, as detected in BN-PAGE experiments in the WT.
The difference between α-b6C, which coimmunoprecipitates the four CCBs, and α-b6N, which selectively coimmunoprecipitates CCB3, suggests that the binding site of α-b6N on cyt b6 is masked whenever this cyt interacts with CCB1 or CCB2/CCB4/CCB3, thereby preventing α-b6N access to its epitope. This steric constraint is consistent with the folding of polypeptide chains around the Qi site in the vicinity of the ci-heme–binding site, as revealed by the three-dimensional structure of b6 f complexes. Indeed, a heme ligation complex is expected to interact with cyt b6 on the stromal end of its first transmembrane helix in order to covalently attach ci-heme by a thioether bond to Cys35 (Fig. 5, c and d). In the b6 f complex structure (Stroebel et al., 2003), the N terminus of cyt b6 forms an amphipathic helix lying parallel to the membrane plane on the stromal face of the membrane near the quinone binding site Qi. The Qi site opens onto the hydrophobic groove at the dimer interface.
α-b6C coimmunoprecipitation before BN-PAGE depleted the digitonin extract in most of the CCB2/CCB4/CCB3-containing complex at 170 kD as well as in the CCB1-containing complexes found in the 70–110-kD region. In contrast, it did not modify the level of CCB2/CCB4-containing oligomers of 70 kD. These results confirm that cyt b6 is associated with the two oligomeric complexes of higher molecular mass but not with the CCB2/CCB4 heterodimer (Fig. 6).
To identify heme-containing intermediates among the CCBs, we attempted to detect peroxidase activity caused by covalent heme binding on these proteins after SDS-PAGE. We used the whole membrane fraction as well as digitonin-solubilized membranes followed by α-b6C coimmunoprecipitation from the petB-C35V strain, which accumulates assembly intermediates. We failed to detect any sign of peroxidase activity in either case. It is of note that we were able to detect peroxidase activity of 1% but not of 0.1% of cyt f using serial dilutions of SDS-solubilized WT membranes (Fig. S1). Thus, assuming that the mRNA ratios would reflect protein ratios for nucleus-encoded products (i.e., 70-fold less mRNA accumulation for CCB genes than for nuclear genes encoding b6 f complex subunits; Karpowicz, S., and S. Merchant, personal communication) and that assembly intermediates recruit most of the CCBs, we would tentatively conclude the absence of covalent heme binding to any of the CCBs.
b6f complex assembles in the absence of ci-heme binding but is highly protease sensitive
By BN-PAGE, we could also investigate the state of b6 f complex assembly (Fig. 7 a). In the WT, dimers and monomers of b6 f complex were easily detected by comigration of cyt f, cyt b6, and subunit IV at 320 kD and 140 kD, respectively (Fig. 7 a), which is in agreement with previous studies (Breyton et al., 1997; Lennartz et al., 2006). Despite the low abundance of b6 f subunits in the petB-C35V strain, as well as in all ccb mutant strains (see cyt b6 in Fig. 4, other b6 f subunits not shown), dimeric and monomeric forms of b6 f complex could still be detected, which indicates that covalent binding of the ci-heme is not a prerequisite for assembly of cyt b6 within b6 f complexes. However, the low accumulation of the b6 f complex in petB-C35V and ccb mutant strains (Kuras et al., 1997; de Vitry et al., 2004) suggests that ci-heme binding contributes to the folding of b6 f complexes in a protease-resistant form. This lower accumulation is not visible in Fig. 4 and Fig. 7 a because we overexposed the immunoblots for the b6 f subunits in the ccb and petB-C35V mutants, as compared with that in the WT (exposure time of 2 min instead of 20 s). In contrast, the assembly of b6 f complex was totally impaired in substitution mutants lacking either one of the four conserved histidines in the sequence of cyt b6, which are axial ligands of bH- or bL-hemes (Fig. 7 a).
We wondered whether these assembled forms of b6 f complex retained any functional properties. To this end, we closely examined the fluorescence induction patterns in vivo of ccb and petB-C35V mutants. In vivo fluorescence induction kinetics were measured using low actinic light followed by a saturating pulse to determine Fmax. As shown in Fig. 7 b, the stationary fluorescence in the petB-C35V strain was higher than in the WT but remained well below Fmax in these experimental conditions. Similar fluorescence data were obtained with the four ccb strains (not depicted). In contrast, both the ΔpetB strain (Fig. 7 b) and the substitution mutants of the axial ligands of bH- and bL-hemes (not depicted) showed a stationary fluorescence very close to Fmax, as expected from the block in electron transfer between PSII and PSI. These results indicate that the plastoquinone pool is not fully reduced in the stationary state in petB-C35V and ccb strains under our light intensity conditions. Thus, the low amount of assembled b6 f complex lacking covalent binding of the ci-heme contributes to some reoxidation of the plastoquinol pool at low light intensity.
Polytopic transmembrane topologies of CCB1–4
The CCBs are required for covalent attachment of ci-heme to cyt b6 in the thylakoid membrane. These proteins are encoded by the nuclear genome, translated as precursors with N-terminal transit peptides to the chloroplast, where they are inserted within membranes (Kuras et al., 2007; Lezhneva et al., 2008). Based on the analysis of the CCB sequences, we previously proposed that CCB1–4 proteins had a similar topology in C. reinhardtii (Kuras et al., 2007) and in A. thaliana (Lezhneva et al., 2008). Our protein folding model (Kuras et al., 2007) took into account the prediction of TMDs using multiple sequence alignments (TMAP; Persson and Argos, 1994) and the prediction of the localization of positively charged segments by the “positive inside rule” at the borders of the putative TMDs (Gavel et al., 1991). This transmembrane organization is supported here by our biochemical study, based on chaotropic extraction, protease accessibility, and split ubiquitin data (Fig. 1 c). In particular, the high protease sensitivity in the absence of vesicle sonication of the two homologues CCB2 and CCB4 is easily explained by their large C-terminal domains exposed to the stromal side of the membranes. Split ubiquitin experiments are consistent with a localization of N and C termini of CCB2 and CCB4 in the yeast cytosol (the compartment corresponding to the chloroplast stroma), thereby confirming our folding model with two TMDs. CCB1 is degraded at a higher trypsin concentration than CCB2 and CCB4, as expected for a protein with three TMDs and short stromal exposed domains. Split ubiquitin experiments further support the localization of the C terminus of CCB1 in the cytosol. CCB3 is very resistant to trypsin treatments, as expected for a hydrophobic protein with two TMDs and short N and C termini exposed to the lumen. That we did not detect a CCB3 digestion product corresponding to the mature protein missing the three C-terminal amino acid residues further supports the lumenal localization of the C terminus of CCB3.
CCB2 and CCB4 form a stable heterodimer
This study provides extensive evidence for the association of CCB2 and CCB4 in a stable heterodimer in vivo. They accumulate in a concerted manner in C. reinhardtii and they interact in two hybrid experiments. Moreover BN-PAGE showed that they are part of a 70-kD complex in all C. reinhardtii strains we examined, except for ccb2 and ccb4 mutants, which lack these two proteins. CCB2 and CCB4 are derived from a unique cyanobacterial ancestor likely duplicated upon endosymbiosis (Kuras et al., 2007). It is likely that cyanobacteria contain a CCB2/4-like homodimer. The molecular information born by a single CCB2/4-like protein in cyanobacteria is split between the eukaryotic CCB2 (about 30 residues after the second TMD) and CCB4 (mainly in the protein C terminus) with little redundancy (Fig. S2), which further substantiates the need for heterodimer formation in the latter case. The photosynthesis apparatus offers two other remarkable examples of heterodimer formation between two homologous proteins: the two core proteins of PSII (D1 and D2) and the two core subunits of PSI (PsaA and PsaB). In the case of type I reaction centers, a homodimer of pshA, still found in some bacteria like heliobacteria, is reminiscent of the ancestral PSI (for review see Heinnickel and Golbeck, 2007).
Distinct CCB hetero-oligomers interact with cyt b6
The presence or absence of cyt b6 had no influence on the accumulation level of any of the CCBs. However, based on BN-PAGE and coimmunoprecipitation experiments, we could identify several hetero-oligomers that interact with cyt b6: a CCB1/b6 complex of ∼70-110 kD and a CCB2/CCB4/CCB3 protein complex of 170 kD in BN-PAGE, the existence of which are confirmed by their coimmunoprecipitation with cyt b6. The accumulation of these protein complexes is particularly visible in the petB-C35V strain. Therefore, we consider that these CCB/b6 hetero-oligomers are transiently formed in the process of heme delivery to the cyt b6 substrate and are trapped in the petB-C35V mutant, which contains the same CCB protein complement as in the WT but is blocked at the very last step of ci-heme binding to cyt b6. We noted that α-b6C coimmunoprecipitates the four CCBs, whereas α-b6N selectively coimmunoprecipitates CCB3. Collectively, these observations argue for the formation of an additional CCB3/b6 complex, distinct from CCB1/b6 and CCB2/CCB4/CCB3/b6 complexes, a conclusion further supported in BN-PAGE by the shift to a lower apparent molecular mass of the CCB3-containing band in the ΔpetB strain. The b6 f monomer and dimer were estimated to contain ∼60% protein and prosthetic groups and 40% detergents and lipids (Breyton et al., 1997). Assuming (1) the same protein to lipid/detergent ratio in each CCB-containing oligomer and (2) that they would contain only one copy of their constitutive subunits, the 170-kD CCB2/CCB4/CCB3/b6 oligomer would have a calculated mass of 155 kD (protein mass of 93 kD), and the 70–110-kD CCB1/b6 oligomer would have a calculated mass of 82 kD (protein mass of 49 kD). Although these calculated values compare well with those from BN-PAGE, we cannot exclude that these three hetero-oligomers would contain additional protein subunits, as yet unknown CCB factors, or several copies of some of their subunits.
Tentative model for a CCB-mediated apo- to holo-cyt b6 conversion
Transient and stable complexes are required for cyt c biogenesis on the p side of membranes. In bacterial/mitochondrial system I, cyt c maturation E (CcmE) binds heme covalently (Schulz et al., 1998) upon interaction with CcmC and CcmD (Ren et al., 2002; Ahuja and Thöny-Meyer, 2005). Holo-CcmE is then released by the CcmA/CcmB ABC transporter (Christensen et al., 2007; Rayapuram et al., 2007). CcmE then interacts with CcmF and CcmH to perform heme ligation to the apocytochrome (Ren et al., 2002; Meyer et al., 2005; Rayapuram et al., 2008; Sanders et al., 2008). In bacterial–cyanobacterial–chloroplast system II, CCSA and CCS1 are part of a 200-kD heme delivery complex (Dreyfuss et al., 2003; Hamel et al., 2003). In mitochondrial system III, cyt c and c1-heme lyases have been identified, and interactions between heme lyase and apocytochrome c have been demonstrated (Mayer et al., 1995).
We expect that system IV, composed of CCBs, should have features distinct from the three other systems because it develops on the opposite side of the membranes. Furthermore, the ferrochelatase produces the heme and the ci-heme–binding site, with both on the stromal side of the membrane, and we do not expect the need for a transmembrane protein that would translocate the heme from one side of the membrane to the other, as it must occur for cyt biogenesis systems on the p side of the membranes. For example, the CCSA–CCS1 complex is believed to perform this initial step before heme binding on the apocytochrome. Indeed, the experimentally tested transmembrane topologies of CCB1–4 protein confirm that most of the conserved tryptophane and/or tyrosine residues that are expected to interact with heme from what is currently known in other maturation systems are all located on the n side of the thylakoid membranes. These conserved residues might participate in heme chaperoning and delivery to cyt b6 on the stromal side of the membranes. We also note that none of the four CCB factors has conserved cysteine residues, which are critical for apocytochrome and/or heme reduction in system I–II, as exemplified by the membrane-anchored thioredoxin-like protein HCF164 (Lennartz et al., 2001; Motohashi and Hisabori, 2006) that would interact with the transmembrane protein CCDA (Page et al., 2004) as a redox relay component required for reduction of the two cysteines at the heme-binding site of apocytochromes f and c6. We cannot exclude that it is merely due to the existence of additional protein factors that have escaped our genetic screen. However, one should consider the possible absence of catalytic cysteines in the process of heme binding on the n side of membranes because of numerous chemical species that have the potential to maintain the protein thiols in the stroma and cyanobacterial cytoplasm, such as the presence of thioredoxins that maintain largely reduced cysteine residues (for review see Herrmann et al., 2009). We cannot exclude either an unusual heme-binding process by a unique thioether bond or the easier access of heme-delivery complexes to the ferrous heme produced by the neighboring ferrochelatase (van Lis et al., 2005). However, none of the four CCBs displays a conserved histidine residue, which is involved in covalent heme binding to CcmE in system I (Schulz et al., 1998). Collectively with our inability to detect any peroxidase activity associated with the CCBs, this argues for the absence of a CCB intermediate carrying a covalently bound heme.
The various CCB-containing complexes that we identified in C. reinhardtii have given us access to some supramolecular traits of the biogenesis pathway for cyt b6 on the n side of membranes. First, covalent binding of the ci-heme to cyt b6 occurs independently of its assembly within the b6 f complex. This is demonstrated by the diffuse aspect of the electrophoretic band for cyt b6 in deletion mutants lacking other b6 f subunits, which is diagnostic of ci-heme binding, instead of a double band in the absence of heme binding (Kuras et al., 1997). This is further supported by direct evidence for an interaction of cyt b6 with CCB components, with the observation of a 70–110-kD CCB1/b6 complex in BN-PAGE at much lower apparent molecular masses than assembled forms of b6 f complexes. The fact that interactions between cyt b6 and the CCBs occur before its assembly within b6 f complexes is demonstrated by our coimmunoprecipitation experiments in which α-CCB3 pulls down only cyt b6 among b6 f subunits, whereas α-IV did not pull down any CCB. That unassembled cyt b6 but not the subunit IV–cyt b6 complex is the substrate for the CCB machinery is consistent with the view that a split cyt b would be a prerequisite for ci-heme insertion (Stroebel et al., 2003).
The full maturation of cyt b6 through the CCB-heme–delivering pathway is not required for its assembly within b6 f complexes, as we detected both monomeric and dimeric forms of the protein complex in the absence of ci-heme binding. A similar observation was recently reported for a ccb2 mutant in A. thaliana. This mutant had a high chlorophyll fluorescence phenotype and showed no evidence for photosynthetic electron transport (Lyska et al., 2007). This is in contrast to our observation that the ci-heme–binding defective strains in C. reinhardtii retain some activity in plastoquinol oxidation, as suggested by their fluorescence properties when illuminated at low light intensity. However, this activity in plastoquinol oxidation did not allow phototrophic growth in standard conditions (growth medium lacking reduced carbon sources, light intensities of 40–100 μE/m2/s, ambient air and room temperature), which allowed us to recover the CCB genes by complementation with a selection scheme based on the restoration of phototrophy (Kuras et al., 2007).
That b6 f complexes can assemble in the absence of ci-heme binding is in marked contrast to the required coordination of hemes bH and bL for cyt b6 assembly within b6 f complexes, as shown by our comparative BN-PAGE experiments. As we observed b6 f complex assembly in each ccb mutant, we can exclude that either of these CCBs is required for the formation of an assembly competent form of a b-heme–containing cyt b6.
On the basis of heme depletion upon gabaculine treatment and after analysis of mutants lacking axial ligands of bL-heme or bH-heme, we previously proposed that formation of a bL-heme–dependent intermediate can be prevented by gabaculine treatment (Kuras et al., 1997). Our sequential model for bL integration as a prerequisite for bH integration within apocytochrome b6 has recently been supported by heterologous expression studies of mutant variants of cyt b6 in Escherichia coli (Dreher et al., 2008). The biochemical data gathered in the present study allowed us to further enrich our original model for cyt b6 maturation (Kuras et al. 1997), with sequential binding of bL-heme, bH-heme, and ci-heme before assembly with other b6 f subunits (Fig. 8). In ccb and petB-C35V mutants, assembly of the b6 f complex continues despite the absence of covalent binding of ci-heme. The resulting complex shows limited accumulation in the thylakoid membranes, probably because it displays an increased protease sensitivity.
In our model for a CCB-mediated apo- to holo-cyt b6 conversion (Fig. 8), formation of a transient CCB1/b6 oligomer that suggests a role of CCB1 in cyt b6 chaperoning is followed by formation of a transient CCB3/b6 oligomer. The latter then recruits the stable CCB2/CCB4 heterodimer to form the membrane-integral multisubunit heme ligation complex comprised of CCB2, CCB4, and CCB3, interacting with the cyt b6. Whether CCB3/b6 is an intermediate before formation (as shown in Fig. 8) or after dissociation of CCB2/CCB4/CCB3/b6 complex is still an open question; indeed, CCB3 is mainly monomeric in the ΔpetB mutant but surprisingly still oligomerizes in ccb1, ccb2, and ccb4 mutants. Concerning the specific roles of CCB2 and CCB4, one could speculate that CCB2, the less conserved CCB protein, would have evolved to adapt for heme scavenging in different organisms performing oxygenic photosynthesis, whereas the better conserved CCB4 protein would pack closely with CCB3 and cyt b6, whose sequences are also well conserved. The binding of b-hemes to apocytochrome b6 before ci-heme binding makes it difficult to assess unambiguously the possible association of a heme with the CCB moiety of these transient oligomers. Once ci-heme is thioether bound to Cys35 of cyt b6, this membrane-integral multisubunit heme ligation complex dissociates, freeing CCB3 and the stable CCB2/CCB4 heterodimer, now available for a new round of catalysis.
The photosynthetic apparatus requires a diversity of biogenesis pathways for cofactor delivery to their protein substrates. With the present biochemical dissection of the CCB pathway, we now have a refined view of the protein scaffold involved in heme delivery to the photosynthetic membranes. In that respect, it is of note that the two machineries for heme delivery on the stromal and lumenal side of the membranes appear not to share any common protein components: mutants defective for the CCS pathway (heme binding on the lumenal side) showed unaltered heme binding to cyt b6, and, similarly, CCB mutants show unaltered heme binding to cyt f (Kuras et al., 2007). How neosynthesized heme is distributed between these two pathways remains to be explored.
Materials and methods
Strains and culture conditions
C. reinhardtii mutants have been described previously and were as follows: ccb, petB-C35V, petB-H202Q, petB-H137G, and ΔpetB (Kuras et al., 1997, 2007; de Vitry et al., 2004). Cells were grown at 25°C in Tris acetate phosphate (TAP) medium, pH 7.2, under dim light (6 μE/m2/s) and collected during the exponential phase at 2 × 106 cells per milliliter.
Protein preparation and analysis
Cell preparation and chaotropic extractions of membrane proteins were performed as described previously (Breyton et al., 1994; de Vitry et al., 2004), except that supernatants from NaCl and NaI extractions were washed on microcon (YM-3; Millipore) in DTT-Na2CO3 0.1 M. For proteolysis experiments, French press–derived thylakoid membrane vesicles were prepared extemporaneously as in Chua and Bennoun (1975) but without protease inhibitors, and finally washed twice with 20 mM Tris-HCl, pH 8.0. Various concentrations of trypsin were added to thylakoid membranes at a final chlorophyll concentration of 270 μg/ml. Half of each sample was sonicated three times for 10 s on ice. Trypsin treatments were performed at room temperature for 30 min, and the reaction was arrested by adding 2 mM phenyl methylsulfonylfluoride as an inhibitor. Polypeptides were separated on 12–18% SDS-polyacrylamide gels in the presence of 8 M urea (de Vitry et al., 1999). Proteins were electrotransferred onto Immobilon NC membranes (Westran S membranes; GE Healthcare) in a semidry blotting apparatus at 0.8 mA/cm2 for 45 min. For immunodetection by the chemoluminescence method (GE Healthcare), we used antisera against cyt b6, cyt f, subunit IV, PSII OEE subunits, and ATP synthase β subunit at a 1:10,000 dilution; CCB3 at a 1:2,500 dilution; and CCB1, CCB2, and CCB4 at a 1:300 dilution.
Yeast split ubiquitin two-hybrid assays
The principle is based on the in vivo split ubiquitin reconstitution caused by the strong interaction of the two ubiquitin moieties. The WT N-terminal half of ubiquitin (NubI) interacts with the C-terminal half of ubiquitin (Cub) even if these domains are fused to transmembrane proteins that do not interact; it is used as a positive control of transmembrane protein expression. Because the mutated N-terminal half of ubiquitin with a substituted isoleucine at position three for a glycine (NubG) has a lower affinity than NubI for Cub, it allows split ubiquitin reconstitution only when the two domains are fused to transmembrane proteins that interact. The transcription factor LexA-VP16 is fused to the Cub domain. Upon interaction of Cub and NubG or NubI, the polypeptide chain between Cub and LexA-VP16 is cleaved. It can translocate to the nucleus and activate transcription of genes, allowing growth in the absence of adenine or histidine or allowing β-galactosidase activity. The cleavage and translocation of the transcription factor require the two ubiquitin moieties located in the yeast cytoplasm. We used Alg5, a yeast transmembrane protein that is not expected to interact with our proteins of interest, fused to NubI as an expression control, or fused to NubG or Cub as an interaction test control. cDNA fragments coding for mature CCB1, CCB2, CCB4, and cyt b6 were amplified by polymerase chain reaction (TripleMaster; Eppendorf) using primers listed in Table S1 that all contained a SfiI site used to fuse in frame the coding sequences to the ubiquitin NubG domain (pPR3-STE vector) and the ubiquitin Cub domain (pBT3-STE vector; DUALmembrane kit3; Dualsystems Biotech). The yeast NYM51 strain (Mata, leu2, trp1, his3, ade2, LYS2::(lexAop)4-HIS3, ade2::(lexAop)8-ADE2, ura3::(lexAop)8-lacZ) was cotransformed with combinations of constructs expressing proteins fused to NubG and Cub domains. Cotransformants were allowed to grow for 3 d at 30°C on synthetic medium (SD) lacking leucine and tryptophan. The interaction between two proteins at the yeast membrane results in the activation of reporter genes ADE2, HIS3, and lacZ. The expression of auxotrophic growth markers ADE2 and HIS3 was monitored by the growth on a medium lacking adenine and histidine in the presence of 10 mM 3-aminotriazole. The expression of lacZ was followed by measuring at OD417nm the accumulation of the product metabolized by β-galactosidase with 2.2 mM o-NPG (2-nitrophenyl β-D-galactopyranoside; Sigma-Aldrich) as a substrate (Rayapuram et al., 2008). A construct expressing Cub fused to the protein of the yeast endoplasmic reticulum Alg5 (pCCW-Alg5 vector) was used as negative control for the NubG fusions. In the same way, a construct expressing NubG fused to Alg5 (pDL2-Alg5 vector) was used as negative control for Cub fusions. Transformations in yeast were controlled by growth on a medium lacking both leucine and tryptophan.
BN-PAGE and immunoprecipitations
Membrane fractions were prepared from sonicated cell lysates as in Kuras et al., (2007). All steps up to the second dimension analysis were performed at 4°C. Membrane fraction was pelleted in a TLA-100.3 rotor (Beckmann Coulter) at 270,000 g for 10 min and resuspended in 50 mM Bis-Tris, pH 7.0, supplemented with 1× complete EDTA-free protease inhibitor mixture (Roche). Immediately after preparation, membranes were solubilized for 30 min in 50 mM Bis-Tris, pH 7.0, and 416 mM aminocaproic acid by 2.5% freshly prepared digitonin at a final concentration of chlorophyll of 0.75 mg/ml (corresponding to 6.68 g of detergent per gram of protein). Taurocholic acid (Sigma-Aldrich) was added to supernatant recovered after centrifugation in a TLA-100 rotor (Beckmann Coulter) at 250,000 g for 30 min, to a final concentration of 0.25% (wt/vol). Supernatant and molecular weight markers (gel filtration calibration kit HMW; GE Healthcare) in 50 mM Bis-Tris, pH 7.0, and 416 mM aminocaproic acid were loaded on a BN polyacrylamide gel (Schägger and von Jagow, 1991) with a 3% acrylamide stacking gel and a 6–18% acrylamide separating gel. Electrophoresis was initiated with cathode buffer containing 0.02% Coomassie blue for 1 h at 150 V, then overnight at 380 V. The buffer was then exchanged with new cathode buffer containing 0.002% Coomassie blue, and electrophoresis was continued for 6 h at 600 V. For the denaturizing second dimension analysis, strips cut out from the first BN gel were incubated at room temperature twice for 15 min in 1% SDS, 1 mM EDTA, 10 mM DTT, 100 mM Na2CO3, and protease inhibitors (200 μM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 5 mM aminocaproic acid), then washed five times for 5 min in 0.25 M Tris, 1.92 M glycine, 0.01% SDS, and 1 mM EDTA. The strips of BN-PAGE were then embedded in a stacking gel on top of a resolving 12–18% SDS-polyacrylamide gel containing 8 M urea, then electrotransferred and analyzed by immunoblotting. For immunoprecipitations, antibodies were coupled to protein A–Sepharose beads as in Schroda et al., (2001). The coupled beads were washed twice in 50 mM Bis-Tris, pH 7.0, 416 mM aminocaproic acid, and 0.2% digitonin, and incubated with digitonin-solubilized membranes (prepared as for BN-PAGE) for 1 h at 4°C. After immunoprecipitation, the coupled beads were washed three times in the previous solution and then twice in 10 mM Tris-HCl, pH 7.5. Immunoprecipitated proteins were recovered by resuspending the beads in 2% SDS and 12% sucrose and boiling for 50 s. After centrifugation, the supernatant was complemented to 0.1 M DTT and 0.1 M Na2C03, and analyzed by SDS-PAGE. Peptide sequences recognized for immunoprecipitation and/or immunodetection experiments were SVTKKPDLSPDVLKY for α-IV (Pierre et al., 1995), SKVYDWFEERLEIQ for α-b6N, and LMIRKQGISGPL for α-b6C (de Vitry et al., 2004), which are localized in cyt b6 structure in Fig. 5 (c and d); and RYYFKRYNYDVKETGE for α-CCB1, FSERPRALADRERGW for α-CCB2, GSCLAADSPENAEQTV and GPQGLLTLIQQRGGL for α-CCB3, and TVDFVEPGLKPYA for α-CCB4 (Kuras et al., 2007), which are localized in CCBs transmembrane topology in Fig. 1 c.
Fluorescence kinetics were measured with a JTS10 spectrometer (Bio-Logic). Cells were dark-adapted for 1 min before each experiment. A low intensity (150 μE/m2/s) continuous illumination was provided with an orange electroluminescent diode (620 nm), a light condition which generates a photosynthetic linear electron transfer of 100 electrons per second in the WT. A saturating pulse was added in the end of the low-light period to fully reduce the quinone pool. Fluorescence was detected with high-pass colored RG695 filters (Schott).
Online supplemental material
Fig. S1 shows the absence of detectable covalent heme binding on CCBs. Fig. S2 shows sequence alignments of CCB2 and CCB4 orthologues with conserved domains repartition highlighted, and schematizes their transmembrane topology. Table S1 describes primers, templates, and vectors used for split-ubiquitin experiments.
We acknowledge the advice of G. Bonnard and C. Raynaud for two-hybrid experiments, C. Breyton and O. Vallon for BN-PAGE, D. Drapier for immunoprecipitation, J. Alric for fluorescence, and Y. Choquet, X. Johnson and R. Kuras for critical reading of the manuscript.
This work was supported by Centre National de la Recherche Scientifique/Université Pierre et Marie Curie (UMR 7141) and Agence Nationale de la Recherche (ANR-07-BLAN-0114). D. Saint-Marcoux was supported by a Ministère de l'Enseignement Supérieur et de la Recherch fellowship and is attaché temporaire d'enseignement et de recherche at Université Pierre et Marie Curie Paris 06.
Abbreviations used in this paper: BN, blue native; CCB, cofactor assembly on complex C subunit B; Ccm, cytochrome c maturation; CCS, cytochrome c synthesis; cyt, cytochrome; n side, negative side; OEE, oxygen evolution enhancer; PS, photosystem; p side, positive side; TMD, transmembrane domain; WT, wild type.