Mechanisms of Mitochondrial Holocytochrome c Synthase and the Key Roles Played by Cysteines and Histidine of the Heme Attachment Site, Cys-XX-Cys-His*

Background: Cytochrome c is covalently attached to heme at a conserved CXXCH motif by holocytochrome c synthase (HCCS). Results: The residues in the heme attachment motif mechanistically contribute to the HCCS-mediated maturation of cytochrome c. Conclusion: Efficient heme attachment is coordinated by the conserved histidine residue in the motif. Significance: Insights into the mechanism of cytochrome c biogenesis broadens our understanding of mitochondrial biology. Mitochondrial cytochrome c assembly requires the covalent attachment of heme by thioether bonds between heme vinyl groups and a conserved CXXCH motif of cytochrome c/c1. The enzyme holocytochrome c synthase (HCCS) binds heme and apocytochrome c substrate to catalyze this attachment, subsequently releasing holocytochrome c for proper folding to its native structure. We address mechanisms of assembly using a functional Escherichia coli recombinant system expressing human HCCS. Human cytochrome c variants with individual cysteine, histidine, double cysteine, and triple cysteine/histidine substitutions (of CXXCH) were co-purified with HCCS. Single and double mutants form a complex with HCCS but not the triple mutant. Resonance Raman and UV-visible spectroscopy support the proposal that heme puckering induced by both thioether bonds facilitate release of holocytochrome c from the complex. His-19 (of CXXCH) supplies the second axial ligand to heme in the complex, the first axial ligand was previously shown to be from HCCS residue His-154. Substitutions of His-19 in cytochrome c to seven other residues (Gly, Ala, Met, Arg, Lys, Cys, and Tyr) were used with various approaches to establish other roles played by His-19. Three roles for His-19 in HCCS-mediated assembly are suggested: (i) to provide the second axial ligand to the heme iron in preparation for covalent attachment; (ii) to spatially position the two cysteinyl sulfurs adjacent to the two heme vinyl groups for thioether formation; and (iii) to aid in release of the holocytochrome c from the HCCS active site. Only H19M is able to carry out these three roles, albeit at lower efficiencies than the natural His-19.

Respiration in eukaryotes depends on the mitochondrial electron transport chain, composed of proteins and co-factors that convert electron flow (oxidation/reduction) into a proton gradient (1). This proton gradient is used to synthesize ATP, for transport, and for other processes essential to the cell (1). Two of the proteins in the chain are cytochrome c (cyt c) and cytochrome c 1 (cyt c 1 ) (of the bc 1 complex, complex III). These cytochromes have heme covalently attached to the proteins at the heme binding site CXXCH, where cysteines form thioether bonds to the two heme vinyl groups (2). The histidine in the CXXCH motif acts as an axial ligand to the heme iron (with Met-81 in human cyt c supplying the second axial ligand) (2).
Over two decades ago, studies in yeast showed that the gene encoding holocytochrome c synthase (HCCS) 3 is required for this covalent attachment in mitochondria (3). However, because the enzyme has been refractory to recombinant overexpression and purification, little has emerged on the mechanisms underlying HCCS function. Although fungi have two related HCCS proteins, one for attaching heme to cyt c (3) and one for cyt c 1 (4,5), animals possess a single HCCS that recognizes both c-type cytochromes (6,7). In addition to respiration, cyt c and HCCS are critical to programmed cell death (i.e. apoptosis) in animals (8,9). Recently, the human HCCS gene was shown to be mutated in the genetic disease microphthalmia with linear skin defects (10). Thus, it is important to understand the mechanisms of HCCS function and cyt c assembly in mitochondria (which is also referred to as the system III cyt c biogenesis pathway (11)).
Very recently, the system III pathway was reconstituted (12) in Escherichia coli ⌬ccm, a strain lacking the endogenous system I genes (ccmA-H) (13). The human HCCS, engineered as an N-terminal GST fusion protein, was able to attach heme to its cognate human cyt c when both were recombinantly expressed in this background. It was demonstrated that human GST-HCCS is membrane localized (12), although no transmembrane helices are predicted in HCCS (6,14). HCCS was purified with endogenous heme, and this heme required HCCS residue His-154 for binding. A model describing four steps in the HCCS pathway for cyt c assembly was proposed (Fig. 1A): step 1 is the binding of heme with His-154 as an axial ligand; step 2 is the binding of apocyt c substrate; step 3 is formation of the two thioether bonds; step 4 is the release of holocyt c from the active site of HCCS. The model is based on specific observations in the recombinant system. For example, when expressed with its cognate cyt c, recombinant WT HCCS co-purified as a complex with its holocyt c, but the HCCS H154A mutant did not co-purify (step 2). When both thioether bonds are formed (step 3), the holocyt c is released to the E. coli cytoplasmic fraction (step 4). If only one thioether bond is formed (e.g. in single cysteine variants of the CXXCH), the cyt c is trapped as a complex and not released (Fig. 1A). A structural schematic indicating where residues such as Cys-15, Cys-18, and His-19 (of CXXCH) might be positioned in the HCCS active site in the complex is shown in Fig. 1B.
One of the proposals for step 2 is that His-19 of the apocyt c binds to the heme iron, replacing an unknown ligand from HCCS (Fig. 1, A and B). In the present study we have characterized the multiple roles of His-19 in our 4-step model. We have also further characterized the roles of the two cysteines, including a comprehensive analysis of HCCS complexes with cyt c substituted at Cys-15, Cys-18, His-19, and double and triple substitutions in the conserved heme attachment site. Resonance Raman (RR) spectroscopy was used to interrogate the heme environment in the HCCS complexes. The RR and UVvisible spectra are consistent with His-19 acting as the second axial ligand in the bis-His HCCS⅐cyt c complex (step 2). These data support the proposal that perturbations of heme (e.g. puckering) caused by the formation of both thioether bonds (step 3) lead to release of holocyt c from the HCCS active site (step 4).
Imidazole Complementation and Bacterial Protein Extraction Reagent Functional Assay-E. coli strains were grown overnight and used to inoculate 5 ml of LB broth supplemented with appropriate antibiotics. These cultures were grown in the presence or absence of 10 mM imidazole (pH 7) at 37°C with shaking at 200 rpm for 3 h, followed by induction with 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside and 0.8% arabinose (w/v) for an additional 3 h. Cells were harvested by centrifugation at 4,500 ϫ g, and the cell pellet was lysed in 200 l of B-PER reagent (Thermo Scientific). Total protein was quantified by absorbance at 280 nm using a Nanodrop 1000 spectrophotometer (Thermo Scientific) and 100 g of extracted protein was resolved by SDS-PAGE, transferred to nitrocellulose membranes, and analyzed by heme stain.
Heme Stains and Sypro Ruby Protein Blot Staining-Heme stains were performed as described previously (16). Briefly, to preserve the heme signal, protein samples were prepared for SDS-PAGE with loading dye at 1:1 (v/v) that did not contain reducing agents and the samples were left unboiled. Following electrophoresis, proteins were transferred to nitrocellulose membranes and the chemiluminescent signal for the heme stain was developed using the Supersignal Femto kit (Thermo Scientific) and detected with the ImageQuant LAS4000 Mini detection system (Fujifilm-GE Healthcare). Following heme staining, membranes were washed in PBS and treated with fixing solution (7% acetic acid, 10% methanol (v/v)) for 15 min. The membranes were washed in deionized water, stained with Sypro Ruby protein blot reagent (Molecular Probes) for 15 min, and washed again in deionized water. Sypro-stained proteins were visualized with the ImageQuant LAS4000 Mini detection system using the Y515-Di filter.
Protein Expression and Purification-GST-HCCS proteins were expressed (with or without cytochrome c variants) and purified from the E. coli ⌬ccm strain RK103 as described previously (12). Briefly, 100-ml starter cultures were grown overnight at 37°C with shaking and used to inoculate 1 liter of LB broth supplemented with the appropriate antibiotics. Following 1 h growth of the 1-liter cultures at 37°C with shaking at 120  TTA TGA AGA GCT CCC AGA GCC ACA CCG TTG AAA AG  pBAD CYCS (C15S,C18A)  CYCS_C15S_C18A_Rev  CTT TTC AAC GGT GTG GCT CTG GGA GCT CTT CAT AA  2 CYCS_C15S_C18A_H19A_Fwd GAA GAG CTC CCA GAG CGC CAC CGT TGA AAA GGG A  pBAD CYCS (C15S,C18A,H19A)  CYCS_C15S_C18A_H19A_Rev TCC CTT TTC AAC GGT GGC GCT CTG GGA GCT CTT C  3 CYCS_H19C_Fwd  ATG AAG TGT TCC CAG TGC TGC ACC GTT GAA AAG GGA GG  pBAD CYCS (H19C)  CYCS_H19C_Rev  CCT CCC TTT TCA ACG GTG CAG CAC TGG GAA CAC TTC AT  4 CYCS_H19G_Fwd TGA AGT GTT CCC AGT GCG GCA CCG TTG AAA AGG GAG pBAD CYCS (H19G) CYCS_H19G_Rev CTC CCT TTT CAA CGG TGC CGC ACT GGG AAC ACT TCA 5 CYCS_H19K_Fwd TGA AGT GTT CCC AGT GCA AGA CCG TTG AAA AGG GAG G pBAD CYCS (H19K) CYCS_H19K_Rev CCT CCC TTT TCA ACG GTC TTG CAC TGG GAA CAC TTC A  6 CYCS_H19M_Fwd ATT ATG AAG TGT TCC CAG TGC ATG ACC GTT GAA AAG GGA GGC AAG pBAD CYCS (H19M) CYCS_H19M_Rev CTT GCC TCC CTT TTC AAC GGT CAT GCA CTG GGA ACA CTT CAT AAT 7 CYCS_H19R_Fwd AGT GTT CCC AGT GCC GCA CCG TTG AAA AGG G pBAD CYCS (H19R) CYCS_H19R_Rev CCC TTT TCA ACG GTG CGG CAC TGG GAA CAC T 8 CYCS_H19Y_Fwd ATG AAG TGT TCC CAG TGC TAT ACC GTT GAA AAG GGA GGC pBAD CYCS (H19Y) CYCS_H19Y_Rev GCC TCC CTT TTC AAC GGT ATA GCA CTG GGA ACA CTT CAT rpm, the cultures were induced with 0.1 mM isopropyl ␤-D-1thiogalactopyranoside for expression of pGEX-HCCS for 5 h. For co-expression of the pBAD-cycS (cytochrome c variants), the cultures were induced with 0.2% arabinose (w/v) 2 h after the induction of HCCS expression. Cells were harvested by centrifugation at 4,500 ϫ g, resuspended in PBS with 1 mM PMSF, and sonicated. The crude sonicate was cleared by centrifugation at 24,000 ϫ g for 20 min, and the membrane fraction was isolated by ultracentrifugation at 100,000 ϫ g for 45 min. Membrane pellets were solubilized in 50 mM Tris (pH 8), 150 mM NaCl, 1% Triton X-100 on ice for 1 h. Solubilized membranes were loaded onto glutathione-agarose (Pierce) for an overnight batch pull-down of GST-HCCS protein (with or without the co-purified cytochrome c variants). Bound GST-HCCS protein or co-complexes were eluted with 20 mM reduced glutathione in 50 mM Tris (pH 8), 150 mM NaCl, 0.02% Triton X-100, concentrated in an Amicon Ultra Centrifugal Filter (Millipore) with either a 100,000 or 30,000 cutoff, and the total protein concentration was determined using the Bradford reagent (Sigma). Cytochrome c Purification-⌬ccm E. coli carrying plasmids for GST-HCCS and cytochrome c were inoculated into 100 ml of LB supplemented with the appropriate antibiotics, grown overnight at 37°C with shaking, and used to inoculate 1 liter of LB broth. Following 1 h growth of the 1-liter cultures at 37°C with shaking at 120 rpm, the cultures were induced with 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside for expression of pGEX-HCCS. 2 h after the induction of HCCS expression, arabinose was added to 0.2% (w/v) to induce the expression of pBAD-cycS (cytochrome c variants) overnight. Cells were harvested by centrifugation at 4,500 ϫ g, resuspended in 50 mM Tris (pH 8) with 1 mM PMSF, and sonicated. The crude sonicate was cleared by centrifugation at 24,000 ϫ g for 20 min, and the soluble fraction was isolated by ultracentrifugation at 100,000 ϫ g for 45 min. The supernatant was loaded onto CM-Sepharose Fast Flow resin (GE Healthcare) for an overnight batch pull-down of positively charged proteins (including cytochrome c). Bound proteins were eluted with 50 mM Tris (pH 8), 500 mM NaCl, concentrated in an Amicon Ultra Centrifugal Filter (Millipore) with a 10,000 cutoff, and the total protein concentration was determined using Bradford reagent (Sigma).
UV-visible Absorption Spectroscopy-UV-visible absorption spectra were recorded with a Shimadzu UV-2101 PC UV-visible scanning spectrophotometer at room temperature as described previously (17). All spectra were obtained in the same buffer in which the proteins were purified (for membrane protein complexes, 50 mM Tris (pH 8), 150 mM NaCl, and 0.02% Triton X-100; for cytochrome c, 50 mM Tris (pH 8) and 500 mM NaCl). Chemically reduced spectra were generated upon the addition of solid sodium dithionite (sodium hydrosulfite) to the purified sample. Where specified, imidazole (1 M, pH 7) was added to purified protein samples at 100 mM prior to the recording of spectra.
Reduced Pyridine Hemochrome-Pyridine extractions of purified proteins were performed as described previously (18). Briefly, 0.5 M NaOH and pyridine were added to 100 g of purified protein to yield final concentrations of 100 mM NaOH and 20% pyridine (v/v). Samples were chemically reduced with the addition of solid dithionite (sodium hydrosulfite) and UVvisible spectra were recorded from 500 to 600 nm.
Resonance Raman (RR) Spectroscopy-All samples for RR were prepared in 50 mM Tris (pH 8), 150 mM NaCl, and 0.02% dodecyl maltoside. The extent of heme loading varied in the co-purified HCCS⅐cytochrome c complexes as follows: HCCS⅐ WT cytochrome c, 11%; HCCS⅐C15S cytochrome c, 16%; HCCS⅐C18A cytochrome c, 22%; HCCS⅐H19M cytochrome c, 3%; and HCCS⅐H19A cytochrome c, 4%. The RR samples for ferric HCCS⅐WT cytochrome c, HCCS⅐C15S cytochrome c, and HCCS⅐C18A cytochrome c were 33 M in heme; the HCCS⅐ cytochrome c H19M and H19A mutants were 28 and 25 M in heme, respectively. Ferrous HCCS⅐cytochrome c complexes were prepared under nitrogen from the ferric samples by addition of a 50-fold excess of sodium dithionite in a buffered solution.
Resonance Raman spectra were recorded with 413.1-nm excitation from a Kr ϩ laser or 514.5 nm emission from a Ar ϩ laser using the 135°backscattering geometry for collection of Raman-scattered light. The spectrometer was calibrated against Raman frequencies of toluene, dimethylformamide, acetone, and methylene bromide. Spectra were recorded at ambient temperature from the samples in spinning 5-mm NMR tubes. UV-visible absorbance spectra were recorded from RR samples before and after spectral acquisition to assess whether sample integrity had been compromised by exposure to the laser beam. Laser power at ferric and ferrous samples ranged from 6 to 9 milliwatt; no spectral artifacts due to photoinduced chemistry were observed with these irradiation powers.

HCCS Recognition of Cytochrome c Variants Possessing Single Cys, Double Cys, and Triple Cys/His Substitutions, and Analysis of Complexes Formed-Single
Cys and His substitutions in the CXXCH site of cyt c still form complexes with the human HCCS. Using various approaches, we and others have suggested that Phe-11, Ile-10, and Lys-8 ( Fig. 1B) need to be properly positioned within the predicted ␣-helix for HCCS-mediated synthesis of cyt c (12,19,20). We hypothesized that these residues along with the CXXCH motif comprise important contacts for apocyt c binding to HCCS and subsequent heme attachment. We wanted to further characterize the HCCS⅐cyt c complexes with single cysteine substitutions and examine the recognition and other properties of cyt c with multiple substitutions. We focus in this report on the cysteines and histidine.
Human HCCS was co-expressed and purified in the presence of its cognate cyt c variants (wild type (WT), single Cys, double Cys, and triple Cys/His). The WT cyt c, single Cys and double Cys variants all co-purify with HCCS, as shown by Sypro Ruby staining ( Fig. 2A) and heme staining (Fig. 2B). As shown previously the single Cys substitutions have ϳ2-fold more cyt c trapped in the complex than WT cyt c (12). The double Cys mutant (which would not be detected by the heme stain because it cannot covalently link to the heme) exhibits the highest level of cyt c in the complex. We estimate this to be at least 4-fold higher than WT (see Fig. 2A, lane 4). This suggests that the double Cys mutant is fully trapped in complex with HCCS,  (12). A, a four-step model for assembly, as discussed in the text. B, diagram of a PEP-FOLD generated structure for the first 20 residues of the human cytochrome c, with key residues described in the text, manually modeled with heme (in pink), and displayed using PyMol. The blue oval represents HCCS with the heme iron ligand His-154. more so than the single Cys mutants or WT. We previously showed that small amounts of Cys-15 (ϳ8%) and Cys-18 (ϳ3%) substituted cytochromes were released from the complex into the cytoplasmic fraction (12). It is possible that the double mutant is not released, facilitating the trapping of more complex. The triple mutant could not be co-purified ( Fig. 2A, lane 5), showing no apocyt c upon purification of HCCS, similar to HCCS expressed and purified in the absence of cyt c expression (lane 6). This suggests that multiple residues (e.g. Lys-8, Ile-10, Phe-11, Cys-15, Cys-18, and His-19 (12,19,20)) form contacts with HCCS and that in this case, it was necessary to remove three of the contacts to prevent stable binding to HCCS.
UV-visible absorption spectra of the complexes (Fig. 2, C-G) and HCCS alone (Fig. 2H) provided information on the state of heme at the active site of HCCS. Previously we have shown that HCCS alone shows a Soret maximum at ϳ423 nm (12), and that this spectrum does not change (i.e. red-shift toward a more reduced state) when the reductant sodium dithionite is added under ambient atmospheric conditions (Fig. 2H, red line). The HCCS complexes with cyt c WT, C15S, and C18A variants each are purified in the oxidized state with a blue-shifted Soret maximum (412-415 nm). Upon reduction, these show red-shifted Soret maxima (420 -424 nm) and ␣ maxima between 553 and 562 nm (Fig. 2, C-E), as shown previously (12). HCCS complexes with the double Cys variant were again purified in the oxidized state and when reduced, a red-shifted Soret (426 nm) and ␣ maximum at 561 nm were observed (Fig. 2F). The changes observed upon chemical reduction in WT, Cys-15, Cys-18, and double Cys are consistent with the proposal that in each of these complexes His-19 provides the second axial ligand to the heme iron. HCCS complexes with the triple substitution, which does not appear to bind the apocyt c substrate, shows spectra ( Fig. 2G) nearly identical to HCCS only (Fig. 2H). This is consistent with a lack of apocyt c substrate binding.
Heme stains of the single Cys complexes, as shown previously (12), exhibit evidence of covalent bonding to the heme within the complex (Fig. 2B), but spectra were unlike those expected with a single covalent linkage. C15S complexes showed split ␣ maxima (555 and 560 nm) and C18A showed an ␣ maximum at 562 nm (Fig. 2, D and E). Single Cys covalent linkages often, but not always, show ␣ maxima around 555 nm in their reduced native state (21,22). Pyridine hemochrome spectra (Fig. 3, A and B) of the complexes, reduced with dithionite, show that heme in the WT complex likely has two covalent linkages (549 nm), the single cysteines have one covalent linkage (553 nm), and that the double Cys, triple substitution, and HCCS only (no cyt c) show no covalent linkages and thus are b-type heme (556 nm). The unique spectra of the purified complexes are most likely a reflection of the unique heme environment at the active site in each HCCS complex.  OCTOBER 17, 2014 • VOLUME 289 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 28799
We have shown previously that HCCS has an axial ligand provided by residue His-154 and an unidentified second axial ligand (12). This second ligand can be replaced in vitro by imidazole, whereupon the HCCS is air oxidized, resulting in a blueshifted Soret maximum, from 424 to 414 nm (Fig. 3H). To further examine whether the His-19 residue is replacing one of the axial ligands (from HCCS) in all complexes where His-19 is present, we added 100 mM imidazole to the complexes. Soret maxima in the HCCS complexes with WT, C15S, C18A, and double Cys mutants were not shifted upon imidazole addition (Fig. 3, C-F), consistent with the proposal that the His-19 imidazole side chain already occupies the second axial ligand position. Soret maxima of the triple Cys/His mutant complex (Fig.  3G) and HCCS alone (Fig. 3H) were both blue shifted 8 -10 nm upon addition of imidazole. Thus the second axial ligand to heme in these two proteins can be provided by exogenous imidazole, thereby facilitating oxidation.
Mechanisms of Recognition, Attachment, and Release Orchestrated by the Cytochrome c His-19 Residue (of CXXCH)-In heme proteins like myoglobin, when the proximal histidine ligand (His-93) is replaced with a glycine, heme no longer binds to the recombinant H93G protein in vivo (23). However, adding 10 mM exogenous imidazole during growth allows the H93G myoglobin to again bind heme in vivo (23). Imidazole binds to the H93G "cavity" and facilitates the binding of heme. Our group has used this "imidazole correction" to analyze many of the heme-binding proteins (with histidines) in cyt c assembly systems I, II, and III (12,17,24). (System III is the HCCS, whereas systems I and II are in prokaryotes.) These proteins with conserved histidines, shown to bind heme, are no longer functional when substituted with alanine or glycine. Indeed, the HCCS H154A mutant was restored by exogenous imidazole for its cyt c assembly function in recombinant E. coli (12). Likewise, if the sole function of the cyt c His-19 is to bind heme iron in HCCS, then imidazole should correct the function of cyt c H19A or H19G variants. (That is, exogenous imidazole should bind to the heme iron and prepare it for thioether bond formation.) Neither the H19A nor H19G variant was corrected for heme attachment (cyt c assembly) by exogenous imidazole (Fig.  4). As shown previously (12), the HCCS H154A variant is corrected for function (Fig. 4). We propose that in addition to providing the ligand to heme in complex with HCCS, an additional role of His-19 is to spatially position the two cysteine thiols adjacent to the heme vinyls for thioether formation. Free imidazole would serve the first role (as it does with HCCS H154A) but would be unable to position the adjacent CXXC motif because, not being bonded to the CXXC motif, free imidazole binding cannot constrain the CXXC conformation to template its regiospecific cross-linking reactions with the heme vinyl groups.
Whereas the apocyt c substrates are still recognized by HCCS when cysteines or histidine are substituted, the specific roles of His-19 clearly involve more than just recognition. That is, His-19 is required for (i) liganding heme iron (in complexes with HCCS) and (ii) positioning the CXXC for attachment (Fig.  1, steps 2 and 3). We wanted to determine whether other possible side chains that can coordinate to heme iron (e.g. Met, Tyr, and Cys) would substitute for these two roles of His-19 in the cyt c assembly. We examined HCCS complexes with cyt c containing H19G, Ala, Met, Cys, Tyr, Arg, and Lys substitutions. Each complex was purified from membrane fractions using Triton X-100 and glutathione columns. SDS-PAGE and Sypro Ruby staining for protein showed approximately equal levels of GST-HCCS and cytochrome c (Fig. 5A). The staining for heme in these complexes showed that for both WT (Fig. 5B, lane 1) and H19M cyt c (lane 6), significant levels of heme was covalently attached to the cyt c. For all other His-19 substitutions, the majority of heme remained with the HCCS. (Although SDS-PAGE typically dissociates heme that is non-covalently bound to a heme protein (e.g. with b-hemes), the SDS-PAGE conditions used here employ room temperature solubilization in SDS sample buffer, no DTT, and it has been shown previously that with these conditions some b-heme remains with the HCCS polypeptide (12).) UV-visible spectra of complexes were also recorded. Under ambient atmospheric conditions, it was possible to chemically reduce only the H19M complex to yield an ␣ and ␤ absorption (Fig. 5D) like the WT (Fig. 5C). However, the Soret regions of complexes with all substitutions looked similar, at ϳ424 nm, regardless of oxidation state. Because these Soret absorption results were similar to the HCCS only preparations, we hypothesized that possibly the H19M is a mixture of c-heme and b-heme, whereas the other substitutions are largely b-heme (i.e. non-covalently linked). In this case, the Soret of H19M reflects the b-heme species. We recorded pyridine hemochrome spectra from heme extractions of the reduced complexes to identify the heme types. Although the WT complex yielded a hemochrome maximum at 551 nm (Fig. 5C, inset), indicative of mostly c-type, the H19M was 553 nm (Fig. 5D,  inset), thus a mixture, and the other His-19 substitutions at 555 nm (Fig. 5, E-J), thus mostly b-hemes. Consistent with H19M complexes representing a mixture of b-and c-type hemes, the resonance Raman spectra also suggested it has similarities to H19A (b-heme) and WT (c-heme) (see below). We conclude that the methionine substitution may satisfy the two roles of His-19 described above, albeit with less efficiency. Thus, the H19M was able to at least partially accomplish steps 2 and 3 (Fig. 1A), therefore we determined whether it also could fulfill step 4, the "release" step, yielding a biosynthesized cyt c with axial bis-met ligation. Recombinant cultures, expressing the HCCS and cyt c genes were harvested and fractionated, and the soluble fractions (containing released cyt c) were studied further. Proteins were partially purified over cation exchange columns because the isolectric point (pI) of the human cyt c is nearly at pH 10 (based on pI calculations of the full-length protein). Fractions eluted with high salt were analyzed by UV-visible spectra (Fig. 6A) and by SDS-PAGE and heme stains (Fig. 6B) to quantitate levels of released holocyt c.
Only the H19M substitution yielded biosynthesized cyt c, with a spectrum distinct from the WT, including a split Soret band. Both the WT and H19M are purified in the reduced state, yielding nearly identical ␣ (550 nm) and ␤ (520 nm) absorption maxima (Fig. 6A). The amounts of H19M purified ranged from 1 to 5% of the WT holocyt c yield, whereas all other His-19 variants yielded less than 0.1%, the limit of detection in these experiments. A previous report has suggested that arginine can replace the His-19 axial ligand to some extent (25). We did not produce any H19R product in our study. We propose that the Met-19-iron interaction is less effective (than His-19-iron) at "pulling" the heme from the HCCS His-154 ligand (step 4). Therefore, these results suggest that His-19 also plays a key role in the release step 4 (Fig. 1A).
Resonance Raman Spectroscopy of the HCCS Complexes-To provide further insight into the heme environments in its purified ternary complexes with HCCS and each of the cyt c constructs, we examined the complexes by RR spectroscopy. The RR signature of heme is responsive to structural and conformational changes that it undergoes during biochemical transformations, such as those associated with the biosynthesis of holocyt c. Cross-linking of the heme 2-and 4-vinyl groups to the Cys residues of the CXXCH motif in cyt c lowers the heme symmetry by inducing equilibrium out-of-plane distortion of the heme and modifying the vinyl substituents (28) (Scheme 1).
We have recorded high and low frequency, Soret (or B)-excited RR spectra of the complexes in both their ferrous and ferric states. Spectra of the ferrous form of HCCS, which is thought to be the active form, are shown in Fig. 7, along with the Q-excited spectrum of the WT complex in the core size marker region. Interpretation of the spectra of the ferric complexes, shown in Fig. 8, is consistent with that of the ferrous complexes. To the best of our knowledge, the RR spectra reported here for C15S and C18A mutants provide the first RR spectroscopic signatures of hemes cross-linked to cyt c through a single thioether bridge.
In-plane Skeletal and Substituent Modes-High-frequency RR spectra of HCCS in complex with heme and WT cyt c, C15S, C18A, H19M, or H19A cyt c variants are shown in Fig. 7A. This region of the heme b spectrum is dominated by bands arising from totally symmetric skeletal vibrations that are characterized by in-plane motions of the porphine core atoms. The largest band in this range of the spectrum is 4 , which is sensitive to the porphyrin * electron density. The heme * density is well correlated with the oxidation state of iron, except in its CO and NO complexes. The 4 frequencies in Fig. 7A range from 1355 to 1359 cm Ϫ1 , consistent with the presence of ferrous heme in these complexes (29).
The bands near 1490 cm Ϫ1 arise from 3 , a mode comprising significant C ␤ -C ␤ stretching and known to be particularly sensitive to the spin state of the iron center. This frequency is definitive for low spin (LS) ferrous hemes. Note that the spectra of the complexes comprising the H19A and H19M cyt c variants exhibit two 3 bands, one attributable to hexacoordinate (6c) LS heme at 1490 cm Ϫ1 and another at 1469 cm Ϫ1 , typical of pentacoordinate (5c) high spin (HS) ferrous heme. The UV-visible spectra in Fig. 2H are typical of a 6c-LS ferrous heme. Thus, the b-heme in HCCS is expected to exhibit one 3 band near 1490 cm Ϫ1 . Even though evidence suggests that reaction of HCCS with H19A cyt c does not yield holocyt c, the RR spectrum makes it clear that interaction with this unproductive cyt c variant triggers scission of an iron-axial ligand bond in holo-HCCS. Interestingly, the relative 3 intensities of the 5c-HS and 6c-LS hemes invert for the cyt c H19M mutant. This is attributed to an increase in the population of the 6c-LS heme due to a small extent of heme coordination by Met-19.
The next highest frequency band in HCCS⅐cyt c WT occurs at 1591 cm Ϫ1 . This is actually an envelope of overlapping bands that comprise contributions from the 2 (ϳ1597 cm Ϫ1 , A 1g in D 4h ) and 19 (ϳ1580 cm Ϫ1 , A 2g in D 4h ) modes. The assignments of these overlapping bands were disentangled by Q-band excitation, which selectively enhances Raman scattering by the non-totally symmetric, including A 2g , modes. A 2g modes give rise to anomalously polarized bands. The inset of Fig. 7A shows the parallel and perpendicular polarized, Q-excited RR spectra of the HCCS⅐cyt c WT complex. These spectra clearly reveal three anomalously polarized bands at 1582, 1398, and 1313 cm Ϫ1 , corresponding to the 19 , 20 , and 21 assignments, respectively, in the Q-excited RR spectrum of cyt c (30). The depolarized band at 1538 cm Ϫ1 corresponds to 11 , a B 1g mode, which also appears in the B-excited spectra at a frequency consistent with a 6c-LS heme. B-excited RR spectra of D 4h metalloporphyrins do not typically contain a prominent 19 band because, due to its A 2g symmetry, its scattering is not enhanced. However, the prominence of 19 , 20 , and 21 in the HCCS complexes examined here is consistent with a symmetry lowering due to an equilibrium out-of-plane distortion similar to that observed in holocyt c (30). Interestingly, the relative intensities of the 19 band is substantially greater in the C15S and C18A complexes. We suggest that this intensity difference is attributable to a strong S 4 distortion imposed on the singly cross-linked heme by the HCCS⅐cyt c complex to conformationally poise it for formation of the second thioether cross-link. The bands in this frequency range of the spectra recorded from the His-19 variants show two bands at frequencies of ϳ1600 and 1581 cm Ϫ1 , typical of 6c-LS and 5c-HS heme b, respectively. These signatures are consistent with significant amounts of hemes in the H19A and H19M complexes not having undergone crosslinking with cyt c. Fig. 7A shows that 10 , which is of rather low intensity in cyt c (30), is not apparent in the spectra of these ferrous complexes. Although the reason(s) for this is not clear, the 6c-LS ferric system (Fig. 8A) does exhibit bands attributable to 10 at frequencies ranging from 1633 to 1636 cm Ϫ1 , consistent with the 6c-LS hemes expected for holo-HCCS (4) and holocyt c. The bands between 1614 and 1619 cm Ϫ1 in Figs. 7A and 8A occur at frequencies atypical of either ferrous or ferric cyt c. We suggest that these bands arise from the C ϭ C stretching modes of the vinyl substituents on heme b, in the case of the uncross-linked H19A and H19M mutants, and a single vinyl substituent in the C15S and C18A mutants. This band is small in the HCCS⅐WT cyt c complex, presumably due to only a residual population of uncross-linked vinyl substituents trapped in the complex with HCCS.
Low-frequency RR spectra of HCCS in complex with heme and the aforementioned cyt c constructs are shown in Figs. 7, B and C, and 8B. In heme b spectra, 7 gives rise to the most resonance enhanced band in this region of the B-excited RR spectra. That is not the case for cyt c. By contrast with heme b, the nearby Ca-S bands (see Scheme 1 for atom identities) are among the most strongly resonance enhanced bands in the B-excited RR spectrum of cyt c (30). The physical basis for this striking resonance enhancement is currently not well understood, although, given the extent to which both C a -S bonds are oriented away from the mean porphyrin plane (31), it likely involves a combination of symmetry lowering and hyperconjugation of that bond into the a 1u (in D 4h ) porphyrinate HOMO (26). Based on the breadth and asymmetry of the RR feature at 681 cm Ϫ1 (Fig. 7B), and by analogy with the cyt c spectrum, that feature is an envelope of bands. Based on the positions of its shoulders, it comprises two bands assigned to Ca-S at ϳ681 and ϳ690 cm Ϫ1 and a third, relatively weak shoulder/band assigned to 7 at ϳ700 cm Ϫ1 . These frequencies and relative intensities are consistent with those of the corresponding modes in the B-excited spectrum of holocyt c, all of which give rise to polarized bands (30). In the spectra of HCCS⅐cyt c C15S and C18A, there is only one band in the Ca-S region at 676 cm Ϫ1 , consistent with their inability to form thioether cross-links to both the 2-and 4-heme vinyl groups. The 676 cm Ϫ1 bands could arise from 7 in the non-fully cross-linked complexes. However, in the spectra of both Cys mutants, weak bands occur near 700 cm Ϫ1 , in the frequency range expected for 7 , suggesting that symmetry lowering of the porphyrin system is sufficient with either one or two cross-links to diminish its resonance enhancement in the B-excited RR spectra. Moreover, we sug-gest that the aforementioned resonance enhancement of Ca-S scattering indicates that, like cyt c, the single C a -S bonds in these mutants are oriented significantly out of the mean porphyrin plane.
The nontotally symmetric 15 bands at 747 cm Ϫ1 (B 1g in D 4h ) are stronger than the 7 bands in these spectra. This has been attributed to vibronic coupling of B x and B y states thought to result from symmetry lowering in the cross-linked heme c (30). By contrast, the H19A and H19M mutants both exhibit strong 7 bands near 675 cm Ϫ1 with I( 7 ):I( 15 ) ratios typical of heme b spectra (32). These features are consistent with minimal crosslinking between cyt c and the heme b in HCCS⅐heme⅐cyt c H19A and H19M.
In D 4h iron porphyrinates, 48 is a Raman-forbidden mode of E u symmetry. In cyt c, it is activated by symmetry-lowering porphyrin distortions that give rise to puckering, in which the cross-linked edge of the porphyrin is folded about the meso carbon atom between the cross-linked pyrrole rings (see Scheme 1). This C s distortion gives rise to bands near 640 cm Ϫ1 in the spectra of the HCCS⅐heme⅐cyt c WT, C15S and C18A complexes (Fig. 7B). Interestingly, the single 48 frequency in the D 4h system is split into two frequencies in the WT (632 and 643 cm Ϫ1 ) and C15S (631 and 647 cm Ϫ1 ), consistent with the aforementioned C s porphyrin distortion. By contrast, 48 is activated in HCCS⅐heme⅐cyt c C18A, but splitting is not detected. This suggests diminished puckering of its 2,4-␤-pyrrole edge relative to that imposed in the C15S complex. This, we hypothesize, has implications for the role of the residues at positions 15 and 18 in the assembly mechanism (see below).
The 400 cm Ϫ1 region of the spectra in Fig. 7C is revealing with regard to thioether cross-links, as the substituent bending modes (designated as ␦) at ␤-pyrrole positions 2 and 4 give rise to bands near this frequency. In the B-excited RR spectra of heme b, there are typically two such ␦ C␤CaCb modes in the 415-420 cm Ϫ1 range (32). However, the C a -S Cys cross-links in cyt c increase the complexity of this spectral region due to the appearance of two ␦ C␤CaS modes at 394 and 401 cm Ϫ1 (30,33). In our spectrum of HCCS⅐cyt c WT, the ␦ C␤CaS and ␦ C␤CaCb bands are centered around 400 and 414 cm Ϫ1 , respectively. Based on the B-excited RR spectra of cyt c (30,33), we assign these bands to envelopes that comprise both bands from each type of substituent bending mode. The broadening that precludes resolution of both bands from each mode type is likely attributable to inhomogeneous broadening in the detergentsolubilized HCCS⅐cyt c complex. However, in Cys-15 and Cys-18 mutants complexes, it is clear that the frequencies of the vinyl and thioether substituents are different when they occur at the 2-and 4-␤-pyrrole positions. Thus, the RR spectra of the HCCS⅐cyt c C15S and C18A mutants support the formation of single thioether cross-links between the heme and cyt c at the vinyl position alternate to the mutation.
The region of the spectrum near 350 cm Ϫ1 exhibits bands arising from modes having Fe-N pyrrole character. In the B-excited RR spectrum of a D 4h iron porphyrinate, this region contains a single strongly enhanced band near 345 cm Ϫ1 corresponding to the totally symmetric Fe-N stretching mode, 8 (26,27,32). The non-totally symmetric Fe-N stretches are designated as 50 and, because of their noncentrosymmetric E u sym- metry, they are Raman forbidden. Protein bridging of the 2-and 4-␤-pyrrole carbons through the thioether bridges puckers the porphine ring, thereby lowering its symmetry to approximately C s (31). In this point group, 50 mode becomes Raman allowed and resonance enhanced in the B-excited RR spectrum. Accordingly, the spectrum of HCCS⅐cyt c WT in Fig. 7C exhibits two overlapping bands at 343 and 353 cm Ϫ1 , attributable to the 8 and 50 modes, respectively. Interestingly, these modes are differentially resonance enhanced in C15S and C18A mutant complexes. We attribute that enhancement pattern to differences in the porphyrin distortion between the C15S and C18A complexes. Specifically, substitution of the methyl side chain in the Ala variant for the longer ethylthiol side chain of Cys relaxes the steric constraints that the complex can impose on the porphyrin conformation. Although Ser-15 cannot be cross-linked to the 2-␤-pyrrole position, the activation of 50 and several out-of-plane modes (see below) in the RR spectrum of the C15S complex suggests that it is capable of imposing the puckered conformation of holocyt c through steric interactions that mimic those of Cys-15 in HCCS⅐heme b⅐cyt c WT. Thus, the RR spectra of the WT, C15S, and C18A complexes suggest that the assembly mechanisms require the HCCS⅐heme b⅐cyt c complex to pucker the 2,4 edge of the porphine ring for release. This region of the C18A spectrum is dominated by the band at 343 cm Ϫ1 , the same frequency observed for 8 in the His-19 mutant complexes wherein the heme remains unattached (i.e. heme b).
Finally, in D 4h iron porphyrinates, 51 is an E u mode and, therefore, Raman forbidden (26,27). However, it is activated by the symmetry-lowering porphyrin distortions in the HCCS⅐ heme⅐cyt c WT and C15S complexes, giving rise to RR bands at 306 and 300 cm Ϫ1 , respectively (Fig. 7C). The inactivity of 51 in the C18A mutant is further evidence that its complex cannot enforce significant heme puckering. The inactivity of 51 in H19A and H19M mutants is consistent with the lack of other distortion-sensitive bands and their lack of significant holocyt c production.
Out-of-plane Porphyrin Deformation Modes-The out-ofplane deformation modes, ␥ 5 and ␥ 15 have A 2u and B 2u symmetry, respectively, in D 4h metalloporphyrins. As non-centrosymmetric u modes, they are both Raman forbidden. Thus, as expected, there are no bands attributable to these modes in the B-excited RR spectra of the HCCS⅐heme⅐cyt c C18A, H19A, or H19M. Consistent with their activation of other Raman forbidden modes (see below), Fig. 7B shows bands arising from ␥ 5 and ␥ 15 at 735 and 716 cm Ϫ1 in the spectrum of the WT complex. The spectrum of the C15S complex only contains ␥ 15 at 714 cm Ϫ1 . The reason for this selectivity is not clear. However, that these modes are activated in the spectra of the WT and C15S complexes clearly show that they impose an equilibrium outof-plane distortion that removes the iron porphyrinate center of symmetry.
The centrosymmetric ␥ 21 mode is doubly degenerate (E g ) in D 4h metalloporphyrins. As such, it is Raman allowed, but due to its out-of-plane coordinates, it is only poorly enhanced with -* excitation (27). The low RR intensities of ␥ 21 in the H19A and H19M spectra of Fig. 7C suggest that the hemes in these complexes exhibit only moderate out-of-plane distortion. By contrast, the ␥ 21 bands in the spectra of HCCS⅐heme⅐cyt c WT, C15S and C18A are noticeably more intense, suggesting that the complex is able to increase the out-of-plane distortion when the His-19 residue of cyt c is available to coordinate the heme. Moreover, the double degeneracy is lost in the WT complex, as evinced by the splitting of ␥ 21 into two bands at 549 and 564 cm Ϫ1 , clearly showing that the rotational symmetry of the heme has been lost. We take this as further evidence that the WT complex imposes C s puckering of the heme, similar to that seen in the crystal structure of holocyt c (31).
Possible Fe-S Met Mode in HCCS⅐Heme⅐Cytochrome c H19M-The 3 frequencies in Fig. 7A reveal that heme exists as a mixture of 5c-HS and 6c-LS Fe(II) in HCCS complexes with H19A and H19M. Assuming that the RR cross-sections for 3 are the same for these two mutant complexes, the H19M complex has a larger fraction of its heme in the 6c-LS state. One interpretation of this difference is that the thioether side chain of Met-19 is coordinated to the Fe(II) center in a small fraction of the complexes and that, by virtue of this coordination to cyt c, those complexes could proceed beyond step 2 of the proposed mechanism to the cross-linking and release steps.
The 57 Fe nuclear resonance vibrational spectrum of ferrous cyt c provides strong evidence for a band at 372 cm Ϫ1 having significant Fe-S Met-80 stretching character (33). This out-ofplane mode is also present in the B-excited RR spectrum. Thus, it is reasonable to speculate that, if Met-19 were to coordinate to the heme iron in ferrous HCCS⅐heme⅐cyt c H19M, it would give rise to a RR band of similar frequency having Fe-S Met-19 stretching character. The spectrum of the H19M complex in Fig. 7C clearly exhibits a band in this Fe-S Met stretching region at 385 cm Ϫ1 . Although it occurs at a frequency 12 cm Ϫ1 higher than the band in cyt c, this frequency difference could be accounted for by differences in the trans effect of their His ligands and/or differences in the coupling of the Fe-S Met bond stretch and porphyrin out-of-plane coordinates. This possible coordination is of mechanistic interest, as it could, within the mechanistic model emerging from this work, account for the ability of the H19M complex to generate small amounts of holocyt c. However, this region of the spectrum is also home to the propionate bending bands, ␦ C␤CcCd . Therefore, based on the data at hand, it is not possible to unambiguously assign the 385 cm Ϫ1 band to a Fe-S Met-19 stretching mode. That assignment would require 35 S isotope labeling of the cyt c H19M.

DISCUSSION
The following conclusions concerning the mechanisms of HCCS function can be made from this study. The cysteines and histidine within the cyt c heme binding (attachment) site play more active roles in assembly than just as recognition determinants by the HCCS. Single Cys, His, and double Cys variants of cytochrome c are each still recognized by the HCCS and co-purify. A triple mutant (both Cys and His) does not co-purify. UV-visible and RR spectroscopy of purified complexes revealed that each trapped cyt c (C15S, C18A, H19A, and WT) showed very distinct spectroscopic profiles. To our knowledge, these are the first published RR spectra of single thioether-attached cytochromes. These data support the contention that such spectra reflect the active site of HCCS with distinct heme envi-ronments for each variant, and that His-19 is an axial ligand (in the complexes where His-19 is present). We suggest that the single cysteine substitutions, each with a single thioether formed at the HCCS active site, do not efficiently release from this complex because both thioethers are needed to alter the conformation of the heme to weaken binding by the HCCS enzyme (for release). The RR spectra of HCCS complexes with WT, C15S, and C18A cyts c indicate that the WT exhibits more puckering than C15S, and C15S exhibits more puckering than the C18A variant. This is consistent with the results that C15S is released (step 4) at ϳ8% WT levels and C18A at only 3% WT levels (12). We propose that puckering induced by spontaneous thioether bond formation reduces the interaction of heme with HCCS, leading to release (step 4).
As depicted in the model (Fig. 1), His-19 (of the CXXCH site) is critical for (i) providing the second axial ligand to the heme at the HCCS active site (step 2); (ii) positioning the adjacent thiols of CXXC in close proximity to the two vinyl groups of heme for attachment (step 3); and (iii) once the two thioethers are formed (step 3), for pulling the heme away from the HCCS active site (and the His-154 axial ligand), thus releasing the holocytochrome c (step 4) for final folding. During this folding process cyt c Met-81 would replace the HCCS His-154 ligand. We discovered that methionine can partially replace the roles played by the His-19 residue, thus it is possible to biosynthesize a bis-methionine cyt c. We currently are optimizing the biosynthesis of this H19M, bismet cyt c for future studies.