Molecular Genetic Identification of a Pathway for Heme Binding to Cytochrome b 6 *

Heme binding to cytochromeb 6 is resistant, in part, to denaturing conditions that typically destroy the noncovalent interactions between the b hemes and their apoproteins, suggesting that one of two b hemes of holocytochrome b 6 is tightly bound to the polypeptide. We exploited this property to define a pathway for the conversion of apo- to holocytochrome b 6, and to identify mutants that are blocked at one step of this pathway.Chlamydomonas reinhardtii strains carrying substitutions in either one of the four histidines that coordinate the bh or bl hemes to the apoprotein were created. These mutations resulted in the appearance of distinct immunoreactive species of cytochrome b 6, which allowed us to specifically identify cytochrome b 6 with altered bh or bl ligation. In gabaculine-treated (i.e. heme-depleted) wild type and site-directed mutant strains, we established that (i) the single immunoreactive band, observed in strains carrying the bl site-directed mutations, corresponds to apocytochrome b 6 and (ii) the additional band present in strains carrying bhsite-directed mutations corresponds to a bl-heme-dependent intermediate in the formation of holocytochrome b 6. Five nuclear mutants (ccb strains) that are defective in holocytochromeb 6 formation display a phenotype that is indistinguishable from that of strains carrying site-directed bh ligand mutants. The defect is specific for cytochromeb 6 assembly, because the ccbstrains can synthesize other b cytochromes and allc-type cytochromes. The ccb strains, which define four nuclear loci (CCB1, CCB2,CCB3, and CCB4), provide the first evidence that a b-type cytochrome requires trans-acting factors for its heme association.

Quinol oxidizing complexes, the cytochrome bc 1 complex of mitochondria and bacteria, and its chloroplast counterpart, the cytochrome b 6 f complex, couple translocation of protons across the membrane to oxidation of lipophilic electron carriers (quinols) and reduction of small hydrophilic proteins (reviewed in Refs. 1 and 2).
The cytochrome b 6 f complex binds five cofactors: two b hemes (high potential b h and low potential b l ), one c heme, a Fe 2 S 2 cluster, and a chlorophyll a (13)(14)(15)(16). In c-type cytochromes, the heme is covalently attached to the protein by thioether linkages between the sulfhydryl groups of (usually) two cysteine residues and the vinyl groups of the tetrapyrrole ring. In b-type cytochromes, the heme is believed to be noncovalently bound to the protein through pairs of histidines, which serve as axial ligands for the iron atom.
Cytochrome b 6 and subunit IV are homologous, respectively, to the first four ␣-helices and next three ␣-helices of cytochrome b of the cytochrome bc 1 complex (17). b h and b l hemes are associated with cytochrome b 6 , b l heme on the lumenal side of the thylakoid membrane (close to the quinol oxidizing site) and b h heme on the stromal side (close to the quinone reducing site).
The crystallographic structure of the bc 1 complex from bovine heart mitochondria is being resolved (18,19). The two hemes are bis-histidine-coordinated (20) and span the membrane bilayer approximately perpendicularly to the membrane plane (21). The two pairs of histidines that coordinate the central iron atoms are conserved in all cytochrome b (22) and located on helices B and D as follows in C. reinhardtii: (B)His 100 -b h -His 202 (D) and (B)His 86 -b l -His 187 (D) (5).
Numerous mutations of cytochrome b have been obtained in photosynthetic bacteria (Rhodobacter species) and mitochondria (reviewed in Ref. 23). By contrast, only a few mutations in petB that alter cytochrome b 6 are known (24 -26). These mutations were generated in C. reinhardtii, which is a unique system for mutational studies of the cytochrome b 6 f complex owing to the availability of chloroplast gene replacement methodology coupled with the fact that the cytochrome b 6 f complex is dispensable for growth in C. reinhardtii.
In contrast to knowledge accumulated on the biosynthetic pathway of the tetrapyrrole cofactors and on the process of covalent attachment of c hemes to their apoproteins (reviewed in Refs. [27][28][29][30], conversion of b-type apocytochromes to their holo-form has received little attention, possibly due to the difficulty in distinguishing heme association defects from defects in other aspects of cytochrome b assembly. Since hemin can bind, without catalysis, to synthetic peptides that mimic helices B and D of cytochrome b (31), one view is that heme binding to cytochrome b should also be uncatalyzed in vivo. This view is uncontested by virtue of the fact that protein factors involved in the catalysis of heme binding to cytochrome b or b 6 have not been identified. The hemes of mitochondrial and bacterial cytochrome b are not detected after SDS-PAGE 1 of membranes or purified complexes (32,33), and this undoubtedly accounts for the dearth of information on cytochrome b assembly. By contrast, heme(s) of cytochrome b 6 can be visualized after electrophoresis (3, 34 -36). Here, we demonstrate that tight binding of heme is a unique aspect of the chloroplast cytochrome b 6 (which may reflect a different mode of association, perhaps covalent, of at least one of the two hemes), and we exploit this property to (i) establish a temporal order of heme binding to apocytochrome b 6 in a multistep pathway in vivo and (ii) identify mutants that are blocked specifically at the step of heme insertion into the b h site. This has been accomplished by monitoring the synthesis of cytochrome b 6 in strains carrying site-directed alterations of b l or b h ligands and in nuclear mutants with similar phenotypes. These nuclear mutants define four loci.

MATERIALS AND METHODS
Strains-Wild type (WT) strain CC125 mtϩ was used for UV mutagenesis, and 137c was used for chloroplast transformation and genetic analysis. Strains were grown on Tris acetate phosphate (TAP) medium, pH 7.2, at 25°C under dim light (5-6 E).
Isolation of Nuclear Mutants Deficient in Cytochrome b 6 f Complexes-WT cells grown to a density of 3-5 ϫ 10 6 cells⅐ml Ϫ1 were transferred to Petri dishes exposed to 260-nm UV irradiation for 1-5 min with constant agitation. The cells' viability was approximately 20 -30%. Irradiated cells were transferred to the dark for 2 h to minimize photoreversion. The cells were then plated on thin agar slabs (2% TAP agar poured over Miracloth circles) placed over TAP agar plates. Plates were returned to the dark for 24 -48 h. Agar slabs were transferred to TAP agar plates containing 20 mM metronidazole and placed in high light (50 -60 E). After 24 -48 h, agar slabs were transferred to fresh TAP agar plates and maintained under dim light for 2-3 weeks until colonies were apparent (50 -150 colonies/plate). Surviving colonies were tested for fluorescence induction kinetics (3,38); mutants showing a decreased cytochrome f by immunoblotting and a fluorescence of cytochrome b 6 f-deficient mutants (26) were chosen.
Mutagenesis and Plasmids-Site-directed mutagenesis was performed in Escherichia coli as by Kunkel (39) on plasmid pdWB (24), which encompasses the whole petB coding sequence and its flanking regions. Mutated products were verified by sequencing. Plasmid pWFA was constructed by introducing the 1.9-kilobase pair EcoRV-SmaI fragment of plasmid pUC-atpX-AAD (40), bearing the aadA cassette, at the unique EcoRV site of plasmid piWF (24). In the resulting pWFA plasmid, the aadA cassette is located 309 base pairs downstream from the end of the petA coding region and is transcribed from the same strand as the petA gene.
Chloroplast Transformation Experiments-C. reinhardtii cells were transformed by particle bombardment as by Boynton et al. (41). We attempted to complement the ⌬petB strain, deleted for the petB gene (24), with the mutated petB genes (Table I); after bombardment, transformant cells were plated on minimum medium (MM) under high light. Since no phototrophic transformants were recovered by the above procedure, we used the WT strain in co-transformation experiments with plasmid pWFA, which confers resistance to spectinomycin. Cells were grown for approximately six generations in the presence of 0.5 mM 5-fluorodeoxyuridine before transformation (42) and plated after transformation on TAP-spectinomycin (100 g⅐ml Ϫ1 ) under dim light to select for spectinomycin-resistant clones. Resistant clones were then screened by fluorescence to choose those that were defective in cytochrome b 6 f activity. The transformants were confirmed to be homoplasmic for the petB mutation by restriction fragment length polymorphism analysis and DNA filter hybridization with specific probes (data not shown). The introduced mutations were further confirmed by direct sequencing of the mutated petB genes in these transformants (data not shown).
Genetic Analysis-Crosses were done according to Harris (43), and complementation analysis between nuclear mutants was done according to Goldschmidt-Clermont et al. (44). For reversion tests, mutant strains were grown in TAP to a density of 2 ϫ 10 6 cells⅐ml Ϫ1 ; the cells were collected by centrifugation and resuspended in MM to a density of 2 ϫ 10 8 cells⅐ml Ϫ1 . One-half ml was spread onto MM agar plates (10 plates) and maintained under high light for 2-3 weeks, at which time colonies were counted. Recombination tests were performed to detect tight linkage; at the same time, at least 30 zygotes isolated from crosses between different mutant strains were transferred separately to either TAP or MM agar plates. They were kept under dim light on TAP plates during 2 weeks or high light on MM plates for 3-4 weeks; the number of zygotes giving rise to colonies was estimated on each plate; the number of tetratype and nonparental ditype tetrads was estimated as a 1 /(b 1 ϫ a 2 /b 2 ), where a 1 is the number of zygotes that give rise to colonies on minimal medium, a 2 is the number of zygotes transferred to minimal medium, b 1 is the number of zygotes that give rise to colonies on TAP, and b 2 is the number of zygotes transferred to TAP.
Protein Isolation, Separation, and Analysis-Biochemical analyses were carried out on cells grown to a density of 2 ϫ 10 6 cells⅐ml Ϫ1 . For analysis of polypeptide contents, samples were resuspended in 100 mM DTT, 100 mM Na 2 CO 3 and solubilized in the presence of 2% SDS at 100°C for 50 s. When indicated, nonheated samples were solubilized in the presence of 2% SDS at room temperature. Polypeptides were separated in the Laemmli system (45) using 12-18% acrylamide gels in the presence of 8 M urea or using 15% acrylamide gels containing 0.1% SDS. Heme staining was detected by peroxidase activity of heme binding subunits using 3,3Ј,5,5Ј-tetramethylbenzidine (TMBZ) as in (46). Immunodetection was carried out as in Kuras and Wollman (24) using a cytochrome b 6 antibody against an N terminus peptide alone that strongly labels a contaminant (Figs. 2A and 5A) or in combination with an antibody against a C terminus peptide that attenuates contamination (Figs. 3B, 6B, 7, and 8). Thylakoid membrane proteins were purified as in Ref. 47. Cytochrome complexes were extracted by hecameg solubilization of thylakoid membranes (4). For simplified membrane preparations, cells were harvested by centrifugation, concentrated to 15 ϫ 10 6 cells/ml in 5 mM HEPES-KOH, pH 7.5, 10 mM EDTA, 0.3 M sucrose containing protease inhibitors (200 M phenylmethylsulfonyl fluoride, 1 mM benzamidine, 5 mM ⑀-amino-n-caproic acid), broken in a French press at 4000 p.s.i., and collected by centrifugation at 15,000 ϫ g for 10 min. For phosphate/acetone extraction, the supernatant from hecameg-solubilized thylakoid membranes (4) was incubated in 500 mM ammonium phosphate, pH 8.0 (50 l of supernatant added to 50 l of 1 M NH 4 HPO 4 , pH 8.0) at 4°C for 30 min and then extracted with 10 volumes of chilled 100% acetone for 10 min. For acid/acetone extraction, the supernatant from hecameg-solubilized membranes was incubated at 4°C for 30 min in 20 volumes of chilled acetone containing 0.2% of HCl (12 N). The precipitating proteins after phosphate/acetone or acetone/acid treatments were collected by centrifugation and prepared for electrophoresis as above.
Pulse-labeling Experiments-Whole cells (2 ϫ 10 6 cells⅐ml Ϫ1 ) were pulse-radiolabeled for 5 min as by Delepelaire (48). 14 C-Acetate was used in the presence of 8 g⅐ml Ϫ1 cycloheximide (an inhibitor of cytosolic protein synthesis) at 10 Ϫ4 M (5 Ci⅐ml Ϫ1 ). To end the labeling period, the isotope was diluted by the addition of 10 volumes of chilled sodium acetate (50 mM).
Gabaculine Inhibition of the Tetrapyrrole Biosynthetic Pathway-Cells grown to a density of 1.5 ϫ 10 6 ml Ϫ1 were treated with 2 mM gabaculine for 6 h as by Howe et al. (49). Cells were harvested by centrifugation, washed, and concentrated to 1.5 ϫ 10 7 cells ml Ϫ1 in minimum Tris medium with 1-2 mM gabaculine and agitated for 1 h. Cells were then pulse-labeled for 10 min with 14 C-acetate at a final concentration of 2 ϫ 10 Ϫ4 M (10 Ci⅐ml Ϫ1 ) in the presence of cycloheximide (8 g⅐ml Ϫ1 ) and 2 mM gabaculine. For immunodetection of cytochrome b 6 after gabaculine treatment, cells were not pulse-labeled, and simplified membrane preparation was used.

RESULTS
Heme Binding to Cytochrome b 6 Is Partly Resistant to Denaturing Treatments-Since cytochrome b 6 can be heme-stained after electrophoresis (3), while the related cytochrome b cannot be, we tested denaturing treatments known to extract noncovalently bound hemes to determine the chemical basis for this unique property of cytochrome b 6 . A cytochrome b 6 f-enriched fraction obtained by hecameg solubilization of WT thylakoid membranes was treated with phosphate/acetone, with acid/ acetone (50), or by boiling for 50 s in the presence of 2% SDS and analyzed for heme content after electrophoresis in ureacontaining polyacrylamide gels. Surprisingly, cytochrome b 6 continued to stain for heme by the TMBZ assay even after these treatments (Fig. 1A). To confirm that this property is unique to cytochrome b 6 , we used a membrane preparation from bovine heart mitochondria to analyze the heme-cytochrome b interaction after electrophoresis on a 15% acrylamide gel containing 0.1% SDS (Fig. 1B). This gel system resolves the closely migrating mitochondrial cytochrome b and c 1 (33). As noted in that work, TMBZ staining revealed only one slowly migrating band, attributed to cytochrome c 1 , and a high mobility band, typical of cytochrome c, with no evidence for cytochrome b staining. By contrast, cytochrome b 6 was heavily stained with TMBZ under these conditions (Fig. 1B). These experiments point to unusually tight binding of heme(s) to cytochrome b 6 , i.e. resistant to denaturation. Since cytochrome b 6 did not stain with TMBZ more heavily than cytochrome f, which binds only one heme, we suggest that only part of the b heme complement of cytochrome b 6 is tightly associated with the polypeptide.
Mutation of the Heme-binding Histidines Revealed Distinct Forms of Cytochrome b 6 That Are Resolved by Gel Electrophoresis and Correspond to b l Site and b h Site Mutants-To determine whether we could distinguish which of the two hemes bound tightly to apocytochrome b 6 and whether both heme binding sites were required for cytochrome b 6 assembly, we used site-directed mutagenesis to construct petB genes in which one of the four heme-liganding histidines, either His 100 and His 202 to b h or His 86 and His 187 to b l , were substituted individually (Table I). At least three independent transformants for each modification were characterized for their biochemical properties. None of the transformants displayed a heme-stainable cytochrome b 6 after SDS-PAGE of solubilized membranes, and the level of heme-stained cytochrome f was reduced to that observed in a ⌬petB deletion strain (data not shown). Immunodetection indicated that in each transformant cytochrome f accumulated to 10% of the WT level, and cytochrome b 6 and subunit IV were present in trace amounts ( Fig. 2A).
Analysis of the immunoreactive species in the transformants versus the WT strain indicated a striking change in the migration pattern of cytochrome b 6 ( Fig. 2A). Holocytochrome b 6 migrated as a broad diffuse band in the WT ( Fig. 2A, lane WT, and dilution series in bottom part), as it does in ⌬petA and in ⌬petD mutants (bottom part of Fig. 2A), which synthesize holocytochrome b 6 , while it migrated as discrete sharp bands in the petB site-directed heme-binding mutants. (The diffuse pattern in the WT strain does not result from overloading, because the same pattern is noted in dilute samples or in the ⌬petA and ⌬petD strains, which accumulate much less holocytochrome b 6 .) The ⌬petB lane serves as a control to indicate that the diffuse band in WT indeed corresponds to cytochrome b 6 and that a band, noted with an asterisk, is an unrelated crossreacting signal. The contaminating polypeptide appears in the region of cytochrome b 6 only when electrophoresis is carried out in the presence of 8 M urea ( Fig. 2A, second part). This crossreacting band is not visible in the absence of urea ( Fig. 2A,  third part). In the four transformants bearing substitutions of the b l heme ligands, cytochrome b 6 migrated as a single band below the contaminating asterisk band ( Fig. 2A, second part,    as a doublet on both sides of the asterisk band ( Fig. 2A, second part, b h transformants); small variations in electrophoretic mobility of these bands were observed, depending on the substituting residue ( Fig. 2A, see for instance b h transformant TB100L). Thus, mutations preventing histidine ligation of the b h hemes can be distinguished from those preventing b l heme ligation, according to their migration pattern on SDS-urea gels. This suggested that it might be possible to resolve intermediates in the in vivo synthesis of holocytochrome b 6 from apocytochrome b 6 on the basis of their migration pattern in SDS and urea-containing polyacrylamide gels.
The synthesis of cytochrome b 6 was monitored by a brief period of labeling. In the WT strain, cytochrome b 6 appears as a broad and diffuse band similar in shape to that detected by immunoblotting or TMBZ staining (Fig. 2B). By contrast, the transformants display sharp and heavily labeled bands in the region of cytochrome b 6 migration, indicating that the rate of petB mRNA translation is not reduced greatly in the hemebinding site mutants. The absence of a comparable signal in the ⌬petB strain confirms that the signal in the mutants corresponds to the petB translation product. The decreased accumulation of cytochrome b 6 in the heme-binding site mutants ( Fig.  2A) must result from degradation of a protease-susceptible apoprotein. Accordingly, the rate of synthesis of cytochrome f in the mutants is reduced to about 10% of that observed in the WT strain, but no change in the rate of subunit IV synthesis is noted, which is consistent with the phenotype of the ⌬petB deletion strain (24). In summary, the pattern of the cytochrome b 6 bands in the mutant strains were similar whether probed for accumulation ( Fig. 2A) or synthesis (Fig. 2B).
Synthesis of Cytochrome b 6 in the Presence of Gabaculine, an Inhibitor of the Tetrapyrrole Biosynthetic Pathway, Reveals a b l Heme-dependent Intermediate in the Biogenesis of Cytochrome b 6 -To order the electrophoretic species identified in the strains carrying b l versus b h ligand mutations in the context of holocytochrome b 6 maturation, synthesis of cytochrome b 6 was monitored in cells depleted of heme. This was accomplished by treating cells with gabaculine, which prevents tetrapyrrole synthesis, for 6 h prior to and during a pulse-labeling experiment. Radiolabeled cells were solubilized and analyzed by urea/SDS-PAGE. In gabaculine-treated WT cells, neosynthesized cytochrome b 6 migrated as a discrete band at the front of the diffuse band observed in untreated cells (Fig. 3A) in a pattern reminiscent of that of radiolabeled b l site mutants (see Fig. 2B). By contrast, the pool of cytochrome b 6 that had accumulated in the membranes and was preexisting was still visualized as a broad, diffuse band whether or not the cells were treated with gabaculine (Fig. 3B). Again, the absence of a signal in the ⌬petB strain authenticates the identity of the bands in the WT and TB202Q strains. Thus, heme depletion by gabaculine treatment prevents heme assembly into neosynthesized apocytochrome b 6 but has no effect on preexisting cytochrome b 6 . In gabaculine-treated b l transformants (e.g. strain TB187G, Fig. 3B) we observed no change in the electrophoretic position of the cytochrome b 6 band. In contrast, in gabaculinetreated b h transformants (e.g. strain TB202Q), most of the newly synthesized cytochrome b 6 doublet is converted to a discrete single band migrating in the position of the lower band  Table I). An unrelated cross-reaction of the b 6 anti-peptide antibody (also present in the mutant strain deleted for the petB gene (⌬petB)) is observed when the gel is run in the presence of urea, and its position is noted (*). When cytochrome b 6 migrated as a doublet of the doublet (Fig. 3A). Since the pool size of cytochrome b 6 accumulated in the mutant is small, reflecting the shorter half-life of the mutant protein, the effect of gabaculine treatment on the upper band is apparent even when unlabeled extracts are examined in immunoblotting experiments (Fig.  3B); specifically, the proportion of the upper band of the doublet is highly decreased relative to that of the lower band. These results confirmed that (i) the lower band of the doublet occurring in b h site mutants corresponds to apocytochrome b 6 and (ii) the upper band, whose synthesis is strongly reduced upon gabaculine treatment, is a heme-dependent intermediate in the formation of holocytochrome b 6 . We suggest a temporal order of heme association with apocytochrome b 6 such that the b l heme is assembled prior to the b h heme.
A Set of Nuclear Mutants of C. reinhardtii Presents the Same Phenotype as the b h Transformants-To characterize the nuclear factors involved in the biogenesis of cytochrome b 6 f complexes, we have isolated a number of nuclear mutants deficient in cytochrome b 6 f activity (3,51), all of which, including the ccb strains described below, displayed fluorescence characteristics typical of mutants with impaired cytochrome b 6 f activity. In contrast to the WT strains, their fluorescence yield (26) rose continuously to an F max level (Fig. 4). They were not altered in photosystem II, since this defect would cause a loss in variable fluorescence and a block at the F max level (see the flat fluorescence trace that is observed in mutants lacking photosystem II reaction centers in Fig. 4). The collection of mutants was screened by TMBZ staining, for the loss of cytochrome b 6 and reduced accumulation of cytochrome f, and by immunoblotting.
The set of ccb nuclear mutants accumulated low amounts of distinct forms of cytochrome b 6 , which are visualized after electrophoretic separation of solubilized membrane proteins (Fig. 5). These forms of cytochrome b 6 , revealed either by immunodetection (Fig. 5A) or by autoradiography (Fig. 5B), correspond in all ccb mutants to the doublet, which is similar to that observed in transformants with altered b h axial ligands. This suggests that the ccb strains are blocked at the same step at which the engineered b h site mutants are blocked, viz. at the step of conversion of the b l heme-dependent intermediate to the holocytochrome (see Fig. 9). The nuclear Ccb genes would then be proposed to encode factors required for the proper interaction of the b-type hemes with apocytochrome b 6 to produce holocytochrome b 6 . To support this model, representative ccb mutants were treated with gabaculine to test whether the ccb strains would behave exactly like the b h site mutants (representative example in Fig. 6). Indeed, gabaculine treatment inhibited the formation of the b l heme-dependent intermediate (upper band of doublet), and this is apparent both in "pulse" radiolabeling experiments, where de novo synthesis of cytochrome b 6 is monitored (Fig. 6A, compare plus lane to minus lane for ccb4 -2), and also when cytochrome b 6 accumulation is assessed by immunoblotting (Fig. 6B, note the depletion of the upper band corresponding to the b l heme-dependent intermediate).
The b l Heme-dependent Intermediate Is Resistant to Denaturing Treatments-In an attempt to assess the nature of the modification that gives rise to the b l heme-dependent intermediate (upper band), crude membrane preparations from WT, ⌬petB, TB187G, TB202Q, ccb1-1, and ccb4 -2 strains were treated with phosphate/acetone, which should release noncovalently bound hemes. The migration pattern of cytochrome b 6 species remained unchanged in all strains after phosphate/ acetone treatment (Fig. 7). Therefore, neither the diffuse aspect of cytochrome b 6 in the WT strain nor the upper band of the doublet in b h site-directed mutants and nuclear ccb mutants can be accounted for by noncovalent association of heme with the polypeptide. To test whether the b l heme-dependent intermediate (upper band) was associated with heme, heme staining assays were conducted after separation of membrane proteins under less denaturing gel conditions, such as solubilization on ice with 1% SDS or 0.88% octyl-glucoside, 0.22% SDS and electrophoresis at 4°C in the absence of urea. Nevertheless, we could not detect a heme-staining band in the region of cytochrome b 6 in any of the b h transformants or the ccb mutants after SDS-PAGE, despite attempts to use more sensitive heme detection methods sensitive to 1% of the WT content in cytochrome b 6 (data not shown).
Genetic Analysis Shows That Four Nuclear Loci Are Required for the Conversion of the b l Heme-dependent Intermediate to Mature Holocytochrome b 6 -The number of loci represented by the ccb strains was determined by genetic analyses. Two nuclear mutants with similar phenotypes, M⌽30 and M⌽35, obtained by random integration of transforming DNA (37), were also included in this analysis, although they displayed a much lower fertility. In recombination tests, it is assumed that mutations in the same gene should be closely linked, thereby preventing a high frequency of recombination events. The frequencies of tetratype and nonparental ditype tetrads were estimated from crosses between the various mutants (Table II, upper part). Three mutant strains gave rise to WT progeny when crossed with all of the other mutants, therefore each defining a locus, respectively CCB1, CCB2, CCB3 (C for cofactor binding, C for cytochrome b 6 f complex, and B for subunit PetB). Another three strains (ccb4 -1, 4 -2, 4-M⌽35) failed to recombine with each other but were able to recombine with the three mutants above. (The apparently lower recombination frequency for M⌽35 is probably due to its lower fertil-ity.) Therefore, the three strains define alleles at a single locus, CCB4. We were unable to characterize further the locus of mutant M⌽30 because to its very low fertility. The conclusions from the recombination analysis were confirmed by complementation analysis (44). Only the two closely linked mutations in strains ccb4 -1 and ccb4 -2 failed to complement each other (Table II, lower part). The complementation tests confirmed that the two strains carried mutated alleles of the same gene. Thus, the genetic analysis of these six nuclear mutants define four different nuclear genes. Since the five mutations complement with at least three other mutations, they are all recessive mutations.

Analysis of Double Mutants Confirms That the ccb Strains Are Affected at the Same Step as Are the b h Site Mutants, Which
Follows the Step Affected in the b l Site Mutants-To confirm the temporal order of cytochrome b 6 assembly suggested above (see Fig. 9 also), we generated double mutants by crossing the ccb strains with the chloroplast transformants lacking either a b h or a b l liganding histidine. The b h TB202Q mt ϩ transformant was crossed with each of five nuclear ccb mutant strains (ccb1-ccb4) mt Ϫ , while the b l TB187G mt ϩ transformant was crossed with the nuclear mutants ccb1-1 and ccb4 -2 mt Ϫ . The resulting tetrads were dissected to recover the four progeny of the zygotes. All progeny had inherited the chloroplast His 202 3 Gln or His 187 3 Gly mutations, transmitted uniparentally by the mt ϩ parent, while only two members of the tetrad inherited the nuclear mutant allele transmitted by the mt Ϫ parent, the two other members having a WT nuclear genome. We analyzed the cytochrome b 6 content in the progeny from these crosses by immunoblotting. In all crosses, the four members of the tetrad presented the same phenotype, which was identical to the  Fig. 2A. The strains are identified by the alleles they carry at the nuclear CCB loci (see Table  II). B, autoradiogram of a urea/SDS-PAGE gel showing synthesis of chloroplast-encoded proteins during a 5-min labeling in the presence of 14 C-acetate and an inhibitor of cytoplasmic translation as in Fig. 2B.

TABLE II
Complementation and recombination analysis of nuclear mutants affected in heme binding to cytochrome b 6 Complementation tests are shown below the diagonal (bottom and left side). Fluorescence analysis was performed on zygotes four days after mating. A plus sign indicates complementation detected as a restoration of a WT fluorescence pattern, while a minus sign indicates the absence of complementation in zygotes that retained a mutant fluorescence pattern. Recombination tests are shown above the diagonal (top and right side). Recombination was scored by measuring germination of zygotes on minimal medium after correcting for the viability of each cross (for details see "Materials and Methods"). Reversion frequencies are indicated in the boldface diagonal.
a Mutant strain ccb4-M⌽35 has lower frequencies of tetratype and nonparental ditype tetrads because of a strong mortality.
b Mutant strain ccb1-1 has a tendency to revert. c ND, not determined.
phenotype of the chloroplast parent. Upon electrophoresis, cytochrome b 6 migrated as a doublet for the progeny of TB202Q ϫ ccb4 -1 crosses, but it migrated as a single sharp band for the progeny of TB187G ϫ ccb4 -2 crosses (Fig. 8). This confirms that the b l site mutation affects a step before that affected in the ccb strains and that the phenotype conferred by the b h site mutation is the same as that conferred by the mutations in the trans-acting CCB loci.

DISCUSSION
The unique denaturation-resistant association of heme with cytochrome b 6 suggested an unusual mode of b heme binding, and it also permitted us to dissect the b heme assembly pathway in vivo by exploiting genetic approaches in C. reinhardtii. Chloroplast mutants, obtained by site-directed mutagenesis of either one of the two histidines in each pair that coordinate b h and b l , were used as templates to define a distinctive phenotype for mutations affecting heme association with cytochrome b 6 . Upon urea/SDS-PAGE, cytochrome b 6 migrates as a broad and diffuse band in the WT strain, whereas it is replaced by a sharp band corresponding to apocytochrome b 6 in b l transformants and a distinct doublet in b h transformants. The additional species observed in b h transformants was identified as a b l heme-dependent intermediate in the assembly of holocytochrome b 6 . These characteristic features were used to identify a set of nuclear mutants that displayed the same phenotype as b h site mutants. The mutants represent four nuclear loci, which we propose encode factors required specifically for the heme assembly into apocytochrome b 6 .
Conversion of Apocytochrome b 6 to Holocytochrome b 6 Is a Multistep Process-We propose that the single, gabaculineinsensitive sharp band, detected in strains where the His ligands to b l have been substituted, most likely corresponds to apocytochrome b 6 , as noted previously for Rhodobacter spheroides bc 1 complex with substituted b l ligands (52). In the case of the doublet detected in ccb mutants and in transformants mutated for the His ligands to b h , the lower band is proposed to correspond to apocytochrome b 6 while the upper band is proposed to correspond to a b l -dependent intermediate form. These assignments are supported by the results of radiolabeling experiments in the presence of gabaculine. When the heme pools are depleted, we expect that newly synthesized cytochrome b 6 will remain in the apoform, and the species detected under these conditions should correspond to apocytochrome b 6 (see for example, Ref. 53). Indeed, gabaculine-treated cells of the WT produce, upon short pulse labeling, a sharp band of the same electrophoretic mobility as that observed in untreated b l transformants (viz. apocytochrome b 6 ). Also, the synthesis of the upper band of the doublet is hampered in ccb mutants and b h transformants treated with gabaculine, which supports the model that the upper band represents a heme-dependent (perhaps heme-associated) species. The association of heme with the upper band could not be measured by heme staining. On the basis of its much reduced accumulation relative to holocytochrome b 6 (Figs. 3B and 6B), it is likely that the heme content of the upper band remains below the sensitivity of the staining technique. However, we cannot exclude the possibility that some post-translational modification accompanies b l binding and yields this intermediate form of cytochrome b 6 . The generation of this species is not dependent upon the Ccb-encoded factors, since the progeny derived from crosses between the b h chloroplast transformant TB202Q and the various ccb mutants still display the electrophoretic doublet.
By analyzing the electrophoretic mobility of cytochrome b 6 in the WT strain versus those containing substitutions of the His ligands to b l and b h , we conclude that the diffuse aspect of WT cytochrome b 6 results from its interaction with the two hemes. Proper folding of the polypeptide in the environment of the b h site is required for formation of the diffuse migrating species. For instance, substitution of Leu 204 by a proline residue in the vicinity of His 202 (b h ligand) yields an electrophoretic doublet, presumably due to misfolding of the b h attachment site (26).
The ccb mutants and the b l /b h chloroplast transformants are nonphototrophic strains and are deficient in the various subunits of the cytochrome b 6 f complex. Since the abundance of petB transcripts and the rate of synthesis of cytochrome b 6 polypeptides were not decreased in these strains, their low content of cytochrome b 6 polypeptides must result from their increased proteolytic susceptibility because of impaired b heme binding.
Conversion of Apocytochrome b 6 to Holocytochrome b 6 Is a Specifically Assisted Process: Mutations in Four Different Nuclear Genes Result in the Same Phenotype as b h Transformants-The five ccb mutants that display a phenotype identical to that of b h transformants are altered specifically in the biosynthesis of cytochrome b 6 and not in the general pathway of b-type heme biosynthesis. The argument for this conclusion is as follows: (i) the ccb mutants are not deficient in cytochrome b 559 , since they exhibit fluorescence induction curves characteristic of cytochrome b 6 f mutants, whereas cytochrome b 559 mutants, because they do not accumulate photosystem II (54), should have no variable fluorescence; (ii) holocytochrome f formation is normal (i.e. it stains for heme) in ccb mutants although its abundance is reduced due to disruption of assembly of the cytochrome b 6 f complex; (iii) the synthesis and accumulation of cytochrome c 6 occurs normally in these strains (data not shown); and (iv) the mitochondrial cytochromes are present as evidenced by normal TMBZ staining of the c-type cytochromes and normal growth by dark respiration on TAP (data not shown). We therefore conclude that the ccb mutants are affected in nuclear factors specifically required for the proper binding of the b-type hemes to apocytochrome b 6 or for the folding of the apoprotein in a conformation suitable for heme association.
Genetic analysis indicates that the mutations belong to four nuclear loci, CCB1, CCB2, CCB3, and CCB4. Since three of the four complementation groups are represented by only one mutation, it is likely that the number of loci involved in the process may be higher. The number of nuclear genes involved in this single assembly process may seem rather high; the biochemis- try of this process is not understood in any system, and it is therefore not possible to speculate on the function of these loci. In Saccharomyces cerevisiae, at least two loci (CBP3 and CBP4) have been identified as encoding candidate cytochrome b assembly factors, but the inability to distinguish heme association mutants from those affected at an early step in bc 1 complex assembly has precluded a definitive functional assignment to these loci (55,56).
One Heme May Be Covalently Attached to Cytochrome b 6 -Surprisingly, denaturing conditions, such as acid/acetone treatment used to prepare globin from hemoglobin (50), which are generally considered to dissociate noncovalently bound hemes from proteins, did not affect the shape of the electrophoretic bands for cytochrome b 6 ; neither the WT diffuse band nor the doublet in b h transformants or in ccb mutants was converted to a sharp single band. We could show, by TMBZ staining after urea/SDS-PAGE, that WT cytochrome b 6 retained b heme after phosphate/acetone or acetone/acid extraction. Heme binding to cytochrome b 6 also resisted boiling SDS. It should be noted, however, that it did not heme-stain more than cytochrome f, which binds only one heme. This observation suggests that only part of the b heme complement of cytochrome b 6 shows high binding affinity for the apoprotein.
The stability of a hemoprotein form of cytochrome b 6 is in sharp contrast with the various reports that b hemes are lost after SDS-PAGE of bacterial cytochrome b (32), mitochondrial cytochrome b (Ref. 33 and Fig. 1B), or the b-type cytochrome from Chlorobium limicola, which shares intermediate properties between cytochrome b and cytochrome b 6 (57). The unusual stability of b heme binding to cytochrome b 6 strongly suggests some linkage that cannot be formed within cytochrome b from bc complexes. Some residues close to the His ligands of the b 6 hemes, conserved in sequences of cytochrome b 6 , but absent from cytochrome b, including the one from C. limicola, could be involved in interactions with the tetrapyrrole ring. Given the similar absorption spectra of cytochrome b 6 and cytochrome b, any covalent linkage of cytochrome b 6 hemes ought to occur through atoms that are not conjugated with the macrocycle. Site-directed mutations of residues that are candidates for providing covalent ligands to the hemes in the vicinity of b l or b h should allow us to settle this point.
Pathway from Apocytochrome b 6 to Holocytochrome b 6 -From the various electrophoretic patterns of cytochrome b 6 forms, it seems reasonable to suggest that the products of the CCB loci catalyze the association of the b h heme to the b lbinding intermediate form of cytochrome b 6 . A schematic view of the apo-to holo-conversion for cytochrome b 6 is presented in Fig. 9 with the following steps: membrane integration of apo- It may sound paradoxical that the CCB loci-assisted step is at the level of b h binding to the b l intermediate form. The latter is stable in denaturing conditions, thereby supporting a tight heme binding of b l rather than b h to cytochrome b 6 . This raises the question of whether the form of cytochrome b 6 revealed in the b h and ccb mutants is a genuine assembly pathway intermediate in cytochrome b 6 biogenesis or perhaps a dead end process that occurs only in the mutants. In this view, proper covalent binding of b l to the apocytochrome may require the concerted presence of the b h substrate and the Ccb gene products. In the absence of either of these factors, spontaneous covalent binding of b l in an inappropriate conformation may occur at low yield generating the b l -dependent intermediate. Regardless, the occurrence of this form has permitted us to resolve the two steps in holocytochrome b 6 formation.