Recombinant cytochrome rC557 obtained from Escherichia coli cells expressing a truncated Thermus thermophilus cycA gene. Heme inversion in an improperly matured protein.

Cytochrome rC(557) is an improperly matured, dimeric cytochrome c obtained from expression of the "signal peptide-lacking" Thermus thermophilus cycA gene in the cytoplasm of Escherichia coli. It is characterized by its Q(00) (or alpha-) optical absorption band at 557 nm in the reduced form (Keightley, J. A., Sanders, D., Todaro, T. R., Pastuszyn, A., and Fee, J. A. (1998) J. Biol. Chem. 273, 12006-12016). We report results of a broad ranging, biochemical and spectral characterization of this protein that reveals the presence of a free vinyl group on the porphyrin and a disulfide bond between the protomers and supports His-Met ligation in both valence states of the iron. A 3-A resolution x-ray structure shows that, in comparison with the native protein, the heme moiety is rotated 180 degrees about its alpha,gamma-axis; cysteine 14 has formed a thioether bond with the 2-vinyl of pyrrole ring I instead of the 4-vinyl of pyrrole ring II, as occurs in the native protein; and a cysteine 11 from each protomer has formed an intermolecular disulfide bond. Numerous, minor perturbations exist within the structure of rC(557) in comparison with that of native protein, which result from heme inversion and protein-protein interactions across the dimer interface. The unusual spectral properties of rC(557) are rationalized in terms of this structure.

Cytochrome c biosynthesis in Gram-negative, prokaryotic cells occurs from the cycA structural gene that normally encodes a preapoprotein having a short N-terminal signal peptide to direct the protein to the periplasm. The process occurs largely in the periplasm, where its completion requires a host of cytochrome c maturation factors (Ccm) 1 (see Refs. 1-4 for a review). Interestingly, two reports describe expression in Escherichia coli of truncated cycA genes (those lacking code for the signal peptide) from thermophilic organisms, Hydrogenobacter thermophilus (5) and T. thermophilus (6), to obtain significant amounts of cytochrome c-like material. Because truncated cycA genes from mesophilic organisms are generally expressed quite inefficiently in E. coli (see Ref. 6 for a review), it was suggested that thermophilic apoproteins fold properly in the cytoplasm, thereby permitting spontaneous heme insertion and thioether formation (7). In the case of the truncated cycA gene from Thermus, it was evident that the resulting product was actually a complex mixture of cytochrome c-like material, which, when examined carefully by a number of spectral techniques, was found to differ significantly from the native protein (Ref. 6 and see below).
From spontaneous maturation of "signal peptide-deficient" Thermus CycA in the cytoplasm of E. coli, two major fractions are obtained, both of which are flawed. The majority product, cytochrome rC 552 , is an ensemble of molecules, some of which have the chemical potential, in a yet unknown form, to spontaneously convert to a novel cytochrome pigment, p572, which, in its reduced form, has optical absorption bands at ϳ430 and 572 nm (cf. Ref. 8 for a brief description of p572). The minority product, cytochrome rC 557 , representing ϳ30% of the cytochrome, is a dimer having its Q 00 band (also called the ␣-band) at 557 nm (6). It has recently been shown that these "errors" in c 552 maturation can be avoided by coexpressing a mature, chimeric cycA gene along with the ccmABCDEFGH genes on a separate plasmid. The resulting recombinant cytochrome c 552 , denoted rsC 552 , obtained in this manner has been examined in functional and structural detail and is identical to native cytochrome c 552 except for having lost two amino acid residues, Gln-Ala, from its N terminus (8).
Here we describe the nature of cytochrome rC 557 and offer speculation on the mechanism of its in vivo formation. This protein was originally thought to be an improperly folded cytochrome c in which both cysteine thiols had reacted with the porphyrin, histidine rather than methionine was bound in the sixth coordination position, and the dimer was stabilized by * This work was supported by National Institutes of Health Grants GM35342 (to J. A. F.) and GM48495 (to D. E. M.). Portions of this work are based upon research conducted at the Stanford Synchrotron Radiation Laboratory, which is funded by the Department of Energy, Office of Basic Energy Sciences. The Biotechnology Program is supported by the National Institutes of Health, National Center for Research Resources, Biomedical Technology Program and the Department of Energy, Office of Biological and Environmental Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The noncovalent interactions. At the time, this idea accounted for the unusual spectral features of the as-isolated protein, its normal heme C pyridine hemochrome spectrum and the absence of both Cys-11 and Cys-14 during standard N-terminal sequencing (6). The current results support a different model.
Molecular Genetics-Procedures were carried out generally as described by Sambrook et al. (11). Oligonucleotides were purchased from Life Technologies, Inc. The polymerase chain reaction was performed using ready-to-go polymerase chain reaction beads (Amersham Pharmacia Biotech). The polymerase chain reaction products were purified using Qiagen purification kits to remove the mineral oil and excess primer. The QiaExII kit was used for isolating restriction fragments from agarose gels. DNA sequencing was done at the UCSD Molecular Pathology Shared Resource in the UCSD Cancer Center.
Construction of the cycA Gene Encoding Cys-11 to Ala-11-Mutation of the cycA gene to encode for alanine rather than cysteine at position 11 (mature, native numbering) was carried out using the Stratagene (San Diego CA) QuikChange TM site-directed mutagenesis kit. This procedure requires two polymerase chain reaction primers, both having the desired mutation. Plasmids pRC552 and pRSC552 were isolated from E. coli DH10B. The sense primer is 5Ј-GATCTACGCCCAGGCCGCG-GGGTGCC-3Ј, and the antisense primer is 5Ј-GGCACCCCGCGGCCT-GGGCGTAGATC-3Ј (the Cys-11 to Ala-11 mutation is underlined). The manufacturer's instructions were followed to introduce the mutation into both plasmids, and the mutation was confirmed by DNA sequencing. We denote these plasmids pRC552(C11A) and pRSC552(C11A).
Protein Expression and Purification-E. coli cells were grown, and cytochrome rC 557 was isolated as described by Keightley et al. (6), except that residual rC 552 was removed by treatment with hydroxyapatite. 2 As predicted from the work of Sambongi et al. (12), cytochrome rsC 552 (C11A) was not synthesized in E. coli cells bearing plasmids pRSC552(C11A) and pEC86, probably because of obligatory formation of a Cys-11-Cys-14 disulfide bond during cytochrome synthesis (cf. Ref. 12). However, cytochrome rC 552 (C11A) was synthesized in good quantity by E. coli cells bearing the plasmid pRC552(C11A), presumably by a spontaneous route that occurs in the cytoplasm. The resulting cytochrome rC 552 (C11A) was purified by the same procedure as described for rsC 552 (6). Optical absorption spectra of cytochrome solutions were recorded using a SLM/AMINCO model DB3500 spectrophotometer. The standard, recombinant cytochrome rsC 552 has a reduced minus oxidized extinction coefficient, ⌬⑀ 552 ϭ 14.26 mM Ϫ1 cm Ϫ1 (8). Second derivative optical absorption spectra were obtained using the vector manipulation program, IGOR (WaveMetrics, Lake Oswego, OR). Pyridine hemochromes were prepared and quantified by the methods of Berry and Trumpower (13). Electrospray mass spectrometry was carried out at the Scripps Research Institute of Mass Spectrometry Facility (La Jolla, CA) using a PerkinElmer Life Sciences SCIEX API III mass analyzer with the orifice potential set at 100 V (14). Mass spectra were obtained from ion spectra using the PerkinElmer Life Sciences program for Macintosh, BioMultiView. Electron transfer activity with cytochrome ba 3 was carried out as described in Ref. 8.
Chemical Methods-N-terminal amino acid sequencing was carried out at the Protein Chemistry Laboratory (Department of Biochemistry and Molecular Biology, University of New Mexico) as described (15). Reduction and alkylation of the cysteine residues in rC 557 was carried out by dissolving 8 nmol of the protein into 100 l of 6 M guanidine hydrochloride containing 10 mol of dithiothreitol in Tris buffer, pH 8.5 (16). This solution was incubated at 37°C for 20 h under a blanket of nitrogen. The protein thus reduced was reacted with 20 mol of iodoacetamide for 20 min at room temperature, after which the pH of the solution was adjusted to pH 2-3 with trifluoroacetic acid, and the protein was recovered by passage through a Vydac C-4 reversed phase column. As a positive control, 8 nmol of ␣-lactalbumin were processed in parallel. Amino acid analysis was performed using the PicoTag method (17).
Hydrazine hydrate was used to selectively reduce heme vinyl to ethyl groups as described by Fischer and Gibian (18). The cytochromes were treated as follows. 100 l of ϳ20 M cytochrome c, 400 l of 95% hydrazine hydrate was diluted to 900 l with 100 mM Tris-HCl buffer at pH 8. This solution was heated at 90°C for 5 min and cooled to room temperature before its optical absorption spectrum was recorded; hydrazine also reduces the iron of the heme.
High Resolution NMR Spectra-Protein samples for 1 H NMR spectroscopy consisted of 1.0 -2.0 mM protein in 100 mM potassium phosphate, pH 7.0, in 90% H 2 O, 10% D 2 O or in 99% D 2 O. Samples were oxidized with a 3-fold molar excess of potassium ferricyanide and concentrated using a Centricon-10 device prior to data collection. 1 H NMR spectra were collected on a 500-MHz Varian INOVA instrument. Onedimensional NMR spectra were collected with recycle times of 250 ms, 350 ms, or 2 s, a sweep width of 40,000 or 50,000 Hz, and presaturation to suppress the solvent resonance.
X-ray Crystallography-Crystals of recombinant Thermus cytochrome rC 557 were obtained against 42% MPEG 5K (w/w), 0.1 M imidazole-malate buffer, pH 6.3, 0 -50 mM NaCl, using ϳ12 mg/ml protein suspended in 25 mM Tris/HCl, pH 8.0, and a drop size of 1 l of protein, 1 l of mother liquor. The crystals belong to space group P2 1 2 1 2, with cell dimensions 98.72, 69.05, and 36.58 Å. Diffraction data to 3.0-Å resolution were collected at two wavelengths, at the white line peak (1.7389 Å) and at a remote high wavelength (1.4586 Å) at the Stanford Synchrotron Radiation Laboratory, beamline 9.2. The data were processed using MOSFLM (19), scaled, and further reduced using the CCP4 suite of programs (20) (see Table I). Anomalous Patterson maps featured two prominent peaks, attributable to two heme groups in the asymmetric unit. Experimental phases were calculated using Xheavy (21) and solvent-flattened and 2-fold averaged using the automask averaging procedure of Ref. 22. Two copies of the previously determined cytochrome c 552 (PDB accession 1C52) were positioned in the experimental electron density maps. A bijvoet difference Fourier map made using the 1.7389-Å Bijvoet differences and the experimental phases showed two large peaks at the positions of the heme. The two copies were then subjected to rigid body minimization, rebuilt using Xfit/ XtalView (21), and further refined using CNS (23) (cf. Table I). Atomic coordinates have been deposited with the Protein Data Bank (1FOC).

RESULTS
Chemical and Optical Absorption Studies-Cytochrome rC 557 's activity as an electron transfer partner with cytochrome ba 3 (6) was reported earlier to be ϳ8% of native Thermus cytochrome c 552 , but subsequent comparisons in this system indicate activity ranging from 0 to 5% (data not shown). Because the current purification procedure 2 removes contaminating rC 552 , dimeric rC 557 is judged to be inactive, although some low level, nonspecific electron transfer may occur.  tural properties virtually identical to those of native Thermus cytochrome c 552 (8). Additionally, its reduced pyridine hemochrome spectrum has the Q 00 band at 551 nm, typical of heme C, and is not affected by heating with hydrazine, indicating the expected absence of vinyl groups (18). The Cys-11 to Ala mutant cytochrome rC 552 (C11A) has not been described previously. 3 Its Q 00 band appears at 553 nm, as may be expected from a monothioether c-type cytochrome (24 -27), and it also exhibits a red-shifted, reduced pyridine hemochrome spectrum having a maximum at 553 nm. Upon heating with hydrazine, the Q 00 band of rC 552 (C11A) is shifted to 551 nm, indicating the presence of a free vinyl group. Both observations are consistent with the fact that this mutant form of the cytochrome can possess only one heme thioether linkage. Spectra of cytochrome rC 557 are the lower traces in Fig. 1, A and B. It was reported previously (6) that the Q 00 peak of its pyridine hemochrome is at 551 nm (see "Discussion"). Upon hydrazine treatment, Q 00 of rC 557 is shifted to 550 nm, indicating the presence of at least one free vinyl group on each heme. Note in Fig. 1B that each of the recombinant cytochromes c exhibits splitting of its Q 00 band at room temperature. 4 The chemical evidence indicating a free vinyl group in rC 557 prompted us to search for a cysteine residue having a free thiol group. Accordingly, cytochrome rC 557 was dissolved into 6 M guanidine-HCl, treated with an excess of dithiothreitol under anaerobic conditions, alkylated with iodoacetamide, and subjected to quantitative amino acid analysis (see "Experimental Procedures"). In two separate experiments, 0.8 and 1.04 mol of carboxymethyl-cysteine were found per mol of heme C.
High Resolution NMR Studies-NMR spectroscopy is a sensitive tool with which to probe the environment of the heme in cytochromes. For example, due to the presence of the S ϭ 1 ⁄2, low spin ferric ion, oxidized cytochromes c show a rich spectrum of 1 H NMR resonances shifted both downfield and upfield of the diamagnetic region. Strongly upfield shifted 1 H NMR resonances may also be observed in the S ϭ 0, low spin ferrous cytochromes, as a result of ring current effects. Fig. 2 shows the upfield shifted 1 H NMR spectra of reduced rsC 552 (native structure), rC 552 (C11A) and rC 557 . Resonances in the region Ϫ2 to Ϫ3 ppm are diagnostic for Fe(II)-S(Met) coordination in the reduced form of cytochromes, where the single three-intensity peak near Ϫ3 ppm is attributed to the axial Met ⑀-CH 3 (28,29). Moreover, a weak optical absorption band attributed to a Met(S)-to-Fe(III) charge transfer occurs at ϳ690 nm (29,30) in the oxidized form of the protein (data not shown). These observations support His-Met axial ligation for rC 557 in both valence states and obviate the previous suggestion of bis-histidine ligation (6).
While already indicated in the optical spectrum of rC 557 , the paramagnetically shifted 1 H NMR spectrum of oxidized rC 557 reveals that the heme environment in this cytochrome is distinctly different from that in native protein. Downfield shifted 1 H NMR spectra of oxidized rsC 552 , rC 552 (C11A) and rC 557 are shown in Fig. 3. The heme methyl resonances in these proteins (see Fig. 4 for numbering) were identified by analysis of oneand two-dimensional NOESY spectra, and were assigned for some of the protein forms. 5 The monomeric proteins, rsC 552 and rC 552 (C11A), have downfield-shifted heme 3-CH 3 and 8-CH 3 resonances with shifts of 33-35 ppm, characteristic of c-type cytochromes with His-Met ligands oriented approximately as shown in Fig. 4 (31,32). As expected, rsC 552 (Fig. 3A) and rC 552 (C11A) (Fig. 3B) also have 5-CH 3 and 1-CH 3 resonances between 10 and 20 ppm, with the shift of 5-CH 3 greater than that of 1-CH 3 . In contrast, rC 557 displays a significantly perturbed 1 H NMR spectrum (Fig. 3C), with heme methyl shifts of 39.4, 34.0, and 17.3 ppm (the fourth methyl resonance has not been identified). Although significantly broadened under these recording conditions, the presence of only three well resolved, hyperfine-shifted three-intensity peaks indicates that the two hemes in the rC 557 dimer are in similar environments. Efforts to assign these three-intensity peaks, however, failed to yield reliable information. 6 The difficulties in assigning the NMR resonances of rC 557 are attributed to conformational heterogeneity exhibited by the nonnatively folded protein in solution (see "Discussion").  FIG. 2. Upfield-shifted 1 H NMR resonances for reduced T. thermophilus cytochromes rsC 552 (A), rC 552 (C11A) (B), and rC 557 (C). The three-proton intensity peaks at Ϫ2.7 to Ϫ2.9 ppm indicate methionine-sulfur coordination to Fe(II).
Crystallographic Studies-We determined the structure of cytochrome rC 557 to a resolution of 3.0 Å using multiwavelength anomalous dispersion. The dimer of cytochrome c protomers is formed by a disulfide bond between the Cys-11 residue of each cytochrome c. The overall structure is shown in Fig.  5, which is a view down an approximate 2-fold axis that relates the two halves of the dimer. This is a noncrystallographic symmetry axis and does not apply to the entire molecule because the structure in the region of the disulfide bond is significantly distorted. Fig. 6 provides a stereo view of electron density around one of the hemes that shows the thioether linkage between the heme and Cys-14 and continuous density connecting Cys-11 on one molecule with Cys-11 on the other molecule of the dimer.
Close examination of the electron density accounting for the heme cofactor provides evidence that the heme is inverted about its ␣-␥ axis (see Fig. 4). Flipping the heme in such a manner also changes the relative positions of the methyl and vinyl groups on the I and II rings of the heme. Further evidence for this comes from the fact that the density for Cys-14 clearly shows it linking to what would have been the methyl position if the heme were not flipped, which is chemically unreasonable. Flipped over, however, the bond forms to the new vinyl position. This can be seen in Fig. 7, which provides a stereo view of heme electron density in rC 557 into which an inverted heme is fitted (yellow model) and overlaid with the model of the heme as it is found in cytochrome rsC 552 (red model). Compared with rsC 552 , the heme in rC 557 is shifted "up" about 1 Å toward the propionates in Fig. 8. Cys-14 has formed a thioether linkage in both molecules, while the C ␣ of Cys-11 has moved ϳ4.5 Å from its position in rsC 552 ; note S␥ shown as the isolated yellow ball in Fig. 7. The fit of the inverted heme (yellow model) into the electron density from rC 557 is good, whereas both the Cys-11 and Cys-14 thioether linkages observed in rsC 552 fall outside the rC 557 electron density (red). By rotating the heme about its ␣-␥ axis, the free vinyl group of heme ring II in rC 557 fits nicely into the rC 557 electron density. Further, the thioether linkage between the vinyl group on ring II and Cys-14 in rsC 552 (red) clearly falls outside the electron density of rC 557 . These observations indicate that the heme is indeed rotated 180°between rC 557 and rsC 552 structures. Overlays of the heme axial ligand regions of rC 557 with the corresponding parts of rsC 552 reveal that neither the bond lengths nor the orientations of the imidazole ring and CH 2 -S-CH 3 planes of the histidine and methionine ligands, respectively, differ in the two molecules (not shown). Thus, the chirality of methionine ligation is preserved in the two molecules and corresponds to that found in mitochondrial cytochromes c (cf. Ref. 30).
Additional evidence for incorrect insertion of the heme in rC 557 can be found in global changes in the molecule. Fig. 8 shows a partial view of cytochrome rsC 552 (red) overlaid on cytochrome rC 557 (yellow). Cytochromes rC 557 and rsC 552 were aligned by fitting C ␣ atoms in the C-terminal two-thirds of the molecule. The two molecules are well matched in the region of residues 60 -131, root mean square difference of 0.39 Å, whereas in the region of residues 4 -59, the structures deviate significantly with a root mean square difference of 1.13 Å. It can be seen that the position of the sulfur atom of the axial ligand Met-69 is similar in the two structures but that the heme iron in rC 557 has moved about 1 Å to the left in Fig. 8, appearing to have pivoted about the sulfur atom of Met-69. This movement is accompanied by a slight opening of a cleft in the molecule denoted by the arrow in the Fig. 8. While numerous small differences exist between the rsC 552 and rC 557 structures, the only residues in the rC 557 monomer that have greatly different positions relative to rsC 552 are Ala-12 and Cys-11. This can be understood if the disulfide bond between the protomers of rC 557 serves to pull this region of the molecule out of the conformation it would take in rsC 552 .
The mean B factor is 75 with a root mean square deviation of 28, which is consistent with the low resolution limit of the diffraction from the crystal and may be compared with the 1.2-Å structure of Than et al., where the mean B is 11.2 and has a root mean square deviation of 12 (33). Subjective evaluation of the electron density maps indicates that the electron density is weak or significantly smeared in the sequence regions 10 -13, 20 -23, 30 -39, 74 -82, and 94 -118; in addition, there is no density for residues 92 and 93, and the C terminus is disor- In addition to the disulfide, the interface between the two protomers is made up of several hydrophobic residues: Ala-10, Cys-11 (as a disulfide), Ala-12, Cys-14 (as a thioether), Ile-22, Ala-25, Phe-26, Val-68, and Phe-72; the hemes are also in van der Waals contact. It is estimated that ϳ10% of each protomer's surface is occluded, with the area of the interface being 1330 Å 2 . The surface area was computed using a 1.4-Å radius probe.
Returning to the higher than average B value in specific regions of the molecule, the distortion in the 74 -82 region most likely results from unfavorable interaction between Gln-74 across the dimer interface with Cys-11 and Ala-12, while the distortion in the 94 -118 region probably results from unfavorable interaction of Asp-77 across the interface with Ala-105. It is noteworthy that these regions of higher than average B value are part of the C-terminal "thermo helices" region (33) that is wrapped around the inner, canonical cytochrome c structure. Residues 30 -39 form a helix that packs on one side of the heme, and smearing of its density probably results from an unfavorable interaction of the helical unit with the flipped, displaced heme.   Fig. 4B), the chain tracing from Cys-14 to Cys-11, and the electron density arising from the disulfide bond to the Cys-11 of the adjacent protomer (in gray). Sulfur atoms are in light blue, and nitrogen atoms are in dark blue. Note the close proximity of the disulfide sulfur atom to the heme. tion about the relative stability of the A and B isomers of recombinant rabbit liver cytochrome b 5 and their rate of interconversion. Thus, when the apo form of cytochrome b 5 is reconstituted with heme, the A and B isomers form in ϳ1:1 ratio, and over a period of hundreds of minutes, depending on pH, the A to B ratio increases to ϳ8:1. Recently, Banci et al. (38) and Arnesano et al. (41) determined the solution structures of the oxidized and reduced forms of the A and B conformers of recombinant rat liver cytochrome b 5 . In these examples, the free energy differences between the A and B forms are very small, on the order of 1 k B T.
A major distinction between heme binding in cytochromes b 5 and cytochromes c is that the "hydrophobic end" of the heme penetrates the protein and the propionate side chains are exposed to solvent in the b-type cytochromes, whereas the opposite holds for heme binding in the c-type cytochromes (see Fig.  4). For Thermus cytochrome c 552 , in which the heme is present in the A conformation, the carboxyl group of propionate 7 forms a salt bridge with the side chain of Arg-125, while the carboxyl group of propionate 6 is in hydrogen bonding distance to main chain NH and/or oxygen of residues Gln-55, Gly-56, Asn-66, and Gly-67. Although the geometry of these hydrogen bonds is poor in the current structural model, this may partially reflect the relatively low resolution of the data and the B factors in the range 58 -98 in these regions. Because the "hydrophilic" portion of the heme is indistinguishable in the two conformations (see Fig. 4), any energetic difference between the A and B isomers must result from different interactions between the protein and porphyrin substituents at positions 1-4. These are likely to be small, making it reasonable to suggest that, prior to thioether bond formation, the protein accommodates A and B conformations of the heme with similar affinity. Trapping the B conformation probably occurs upon formation of the Cys-14/2vinyl thioether bond, while Cys-11 subsequently forms a disulfide link with another pre-rC 557 molecule. Because these reactions are occurring without intervention of cytochrome c maturation factors, the final A to B ratio of ϳ3 most likely depends on relative rates of thioether formation in the noncovalent A and B conformations.
Structural Perturbations Caused by Heme Inversion and Disulfide Formation-By whatever mechanism, once the heme is fixed in the B conformation and the disulfide bond has formed, the molecule differs in minor but significant ways from the native structure. Small structural differences are distributed The two sulfur atoms on the right of the rC 557 model correspond to the disulfide bridge between the Cys-11 residues from the two protomers. The two molecules were superimposed using the best fit of C ␣ atoms between residues 60 and 131 (see "Results"). The black arrow points to a cleft between the domains "above" and "below" the hemes that is opened slightly in rC 557 . throughout the molecule, while the larger ones are found in those regions affected by interactions across the dimer interface, as noted from the comparison of local B factors. Although we are unable to distinguish changes due to accommodation of the inverted heme within the heme pocket and those induced by protein-protein interactions across the dimer interface, it is likely that the protomer, as it is captured in the rC 557 dimer, is unable to achieve the global free energy minimum represented by the native structure. One may imagine this as a perturbation of the energy landscape near the energy minimum of the folding funnel (42). 7 Our NMR data are consistent with this interpretation (cf. Footnote 6).
Novel Spectral Features of Cytochrome rC 557 -Now that we know the structure of cytochrome rC 557 , it is possible to provide some rationalization of its unique optical absorption and hyperfine shifted 1 H NMR spectra. The energy and shape of the Q 00 band contains information about the structure and environment of the heme. This band is actually composed of two normally degenerate xy polarized transitions, which occur at different energies only when perturbations on the heme significantly lower its normal D 2h symmetry (43,44). The average energy of the Q 00 transitions depends on the nature of the heme substituents, to a lesser extent on the axial ligands of the iron, and to a much lesser extent on the dielectric environment. The energy separating the two Q 00 transitions depends on a variety of factors including distortion of the heme from planarity and the dispositions of charges around the heme pocket (cf. Rasnik et al. 8 for additional references). Of interest here are the vinyl groups of the heme and their ability to form thioether linkages with the protein. A general rule of thumb in b-and c-type cytochromes is that the presence of one vinyl group shifts the absorption ϳ5 nm to the red, while two free vinyl groups shift the absorption an additional ϳ5 nm (cf. Ref. 24). Accordingly, Q 00 of a typical b-type cytochrome containing Fe(II)-protoporphyrin IX occurs around 560 nm, at ϳ555 in a monothioether cytochrome c, and close to 550 nm in a dithioether cytochrome c. In naturally occurring (24) and synthetic monothioether cytochromes c (26,46), the Q 00 bands occur over the range 553-561 nm. Because the Q 00 band of cytochrome rC 557 lies within this range, why was the possibility of a monothioether arrangement not suggested earlier? The principal reason was that Q 00 of its pyridine hemochromogen of rC 557 occurs at 551 nm (6), which is identical to that of rsC 552 and is diagnostic of a dithioether hemochromogen. We speculate here that under the strongly basic and reducing conditions required to form the pyridine hemochromogen, the disulfide is reduced to yield a thiolate group that reacts with the free vinyl to form dithioether heme. Thus, it is likely that the one free vinyl group in "as-isolated" rC 557 accounts for the bulk of the observed red shift of its Q 00 band.
Heme methyl 1 H NMR shifts of c-type cytochromes with ligands oriented as shown in the A conformation (Fig. 4) show little variation across species, even for proteins with significantly different folds (32). Thus, the differences between the pattern of hyperfine shifts of rC 557 compared with the monomeric proteins indicates a significant change in heme electronic structure. One possible explanation for this could be a change in the heme axial ligands. However, as noted above, structurally similar His-Met ligation occurs in both valence states of rC 557 . Another possible explanation for the perturbation of the rC 557 NMR spectrum could be the breakage of the thioether bond to the heme. However, rC 552 (C11A) necessarily has one vinyl in place of a thioether, and its heme methyl shift pattern is similar to that of rsC 552 Other modifications of the porphyrin, such as oxidation, also can be eliminated by examination of the optical absorption spectrum of rC 557 and its pyridine hemochrome. The presence of a second, nearby heme group in rC 557 is not expected to significantly change chemical shifts through dipolar effects because the shortest distance between the iron of one heme and a proton on the other is Ͼ10 Å (47) This leaves the possibility of the heme being oriented differently in the rC 557 protein compared with the wild-type monomer. Rotation of the heme about its ␣,␥-meso axis leads to significant perturbation of the heme methyl shift pattern (Ref. 48; see Refs. 34 -41). Interestingly, we observe resonances attributed to a minor form of rC 552 (C11A) with similar shifts as rC 557 (resonances marked with asterisks in Fig. 2B); these may arise from a minority conformer in which the heme is flipped in this mutant.
According to recent theories explaining the paramagnetically shifted 1 H NMR resonances, the magnitude of the shift of each methyl group resonance is determined by the orientations of the filled p-orbital (or sp 3 -orbital) on the sulfur atom of the liganding methionine and the -bonding p-orbital of the ⑀-N atom of the liganding imidazole ring (31,32). The average value of these angles in both rsC 552 and rC 557 is ϳ95-100°. From Fig. 6 of Shokhirev and Walker (32), the predicted order of the methyl shifts is C 3 Ͼ C 8 Ͼ C 5 Ͼ C 1 . This order corresponds to our assignments in rsC 552 (see Fig. 3A). Further, rotating the heme about its ␣,␥-axis should not change the orientation of the magnetic axes with respect to the protein; however, the positions of the methyl groups with respect to the magnetic axes will change, and a different order of methyl group shifts is predicted. Unfortunately, probably because of microheterogeneity in the rC 557 structure, we have not been able to assign the three-intensity lines shown in Fig. 3C. 6 Given the great similarity of the axial ligand orientations in rsC 552 and rC 557 , assigning the methyl resonances in rC 557 would provide a valuable test of the theory, but these must await further experimentation.
As a substrate of cytochrome ba 3 , cytochrome rC 557 is essentially devoid of activity. The likely reason for this is that the surface around the exposed heme edge, which is most likely the binding interface with cytochrome ba 3 , is largely occluded in dimer interface of rC 557 . Cursory comparison of this surface with the analogous region of horse heart cytochrome c reveals that this face of Thermus cytochrome c 552 is largely neutral, as already noted in Ref. 33, whereas that of the horse heart cytochrome c has several exposed positive residues in this same area. The latter explains why horse heart cytochrome c is only a weakly active substrate for ba 3 (45,49).