A Strategic Protein in Cytochrome c Maturation

CcmH (cytochromes c maturation protein H) is an essential component of the assembly line necessary for the maturation of c-type cytochromes in the periplasm of Gram-negative bacteria. The protein is a membrane-anchored thiol-oxidoreductase that has been hypothesized to be involved in the recognition and reduction of apocytochrome c, a prerequisite for covalent heme attachment. Here, we present the 1.7Å crystal structure of the soluble periplasmic domain of CcmH from the opportunistic pathogen Pseudomonas aeruginosa (Pa-CcmH*). The protein contains a three-helix bundle, i.e. a fold that is different from that of all other thiol-oxidoreductases reported so far. The catalytic Cys residues of the conserved LRCXXC motif (Cys25 and Cys28), located in a long loop connecting the first two helices, form a disulfide bond in the oxidized enzyme. We have determined the pKa values of these 2 Cys residues of Pa-CcmH* (both >8) and propose a possible mechanistic role for a conserved Ser36 and a water molecule in the active site. The interaction between Pa-CcmH* and Pa-apocyt c551 (where cyt c551 represents cytochrome c551) was characterized in vitro following the binding kinetics by stopped-flow using a Trp-containing fluorescent variant of Pa-CcmH* and a dansylated peptide, mimicking the apocytochrome c551 heme binding motif. The kinetic results show that the protein has a moderate affinity to its apocyt substrate, consistent with the role of Pa-CcmH as an intermediate component of the assembly line for c-type cytochrome biogenesis.

What is the mechanism whereby the heme is covalently attached to c-type cytochromes? Despite the crucial role played by these proteins in many cellular processes, this question remains largely unanswered. In c-type cytochromes, the heme is covalently bound to the apoprotein via two thioether bonds between the heme vinyls and the thiol groups occurring in the conserved CXXCH motif. The apoprotein and the heme group have to be in the correct stereochemical orientation for promotion of thioether bond formation; however, the chemistry of such a process is poorly understood, and only recently, in vitro formation of a c-type cytochrome has been reported (1,2). In the cell, the necessary stereospecific insertion and attachment of heme to apocytochrome c (apocyt) 2 occurs on the p-side of the membrane, i.e. the intermembrane space of mitochondria, the chloroplast lumen, or the periplasm of bacteria. This process requires an elaborate post-translational modification machinery that must also ensure translocation of both apocyt and heme from the n-side of the membrane where they are synthesized.
It is known that three different multienzymatic systems have evolved in different organisms to accomplish this task. Although System III consists of a single cyt c heme lyase (3) and is adopted exclusively by eukaryotes, Systems I and II consist of different proteins that have been discovered in Gram-negative bacteria, in plant or protozoal mitochondria (System I), and in Gram-positive bacteria, plant chloroplasts, and cyanobacteria (System II) (4 -6). Both systems comprise different membraneanchored proteins that expose their active site to the periplasm (System I) or to the exterior (System II), each performing a particular function. The CcmABC components of System I have been recently recognized as subunits of an ATP binding cassette transporter that, together with CcmE and CcmD, acts as a heme release complex (7,8). On the other hand, since in the periplasm the 2 Cys residues of the heme binding motif of apocyt (CXXCH) are oxidized by DsbA (9), a crucial bottleneck in the maturation of apocyt is the required reduction of these cysteines before covalent heme attachment can occur. In System II, this reductive step is carried out by the thioredoxin-like (TRX-like) ResA, a membrane-associated thiol-oxidoreductase (10 -12), whereas in System I, the reduction of the apocyt disulfide bond was proposed to be carried out by CcmH and/or CcmG via the formation of a mixed disulfide complex (13)(14)(15). According to this view, the mixed disulfide complex between CcmH and apocyt is subsequently resolved by CcmG, releasing reduced apocyt and oxidized CcmH (15,16). Although CcmG is a structurally well characterized TRX-like protein that receives electrons from DsbD (17)(18)(19), no structural information is available for any CcmH protein. Previous studies indicated that CcmH is a membrane-bound protein exposing the conserved redox-active LRCXXCQ motif to the periplasm; its essential role in the maturation of apocyt was suggested by the observation that loss of CcmH impairs the synthesis of c-type cytochromes in vivo (13,14,20). Moreover, in vitro evidence indicates that Rhodobacter capsulatus and Arabidopsis thaliana CcmH homologues are able to reduce the CXXCH motif of an apocyt-mimicking peptide (13,21). Taken together, these observations suggest that CcmH is redox-active and is specifically involved in the thioreduction pathway of cytochrome c maturation.
To shed light on the extracytoplasmic-heme ligation process occurring in System I organisms, we expressed, crystallized, and characterized the soluble periplasmic domain of CcmH from the opportunistic pathogen Pseudomonas aeruginosa (hereafter named Pa-CcmH*). The three-dimensional structure of Pa-CcmH* solved at 1.7 A resolution reveals that the periplasmic domain adopts a peculiar three-helix bundle fold that is different from that of canonical thiol-oxidoreductases and provides a basis for understanding its specific reductant activity in a highly oxidizing environment. Moreover, we have determined (i) the pK a value of the 2 Cys residues (Cys 25 and Cys 28 ) engaged in the disulfide bond and (ii) the redox potential of both Pa-CcmH* and its physiological partner Pa-apocyt c 551 . The interaction between Pa-CcmH* and Pa-apocyt c 551 was characterized in vitro, by fluorescence stopped-flow experiments, using a Trp-containing fluorescent variant of Pa-CcmH* and a dansylated nonapeptide mimicking the heme binding motif of P. aeruginosa cytochromes c 551 .

EXPERIMENTAL PROCEDURES
Cloning, Expression, and Purification-The soluble N-terminal domain of CcmH from P. aeruginosa (residues 21-100) was obtained by removing the signal peptide, the membrane-anchoring helix, and the cytoplasmic C-terminal stretch. The domain was amplified from the genomic DNA (strain PA01) and cloned in pET28b (Novagen). Site-directed mutants were produced by using a QuikChange mutagenesis kit (Stratagene, La Jolla, CA). Expression of the His tag fusion protein in Escherichia coli BL21 (DE3) was induced with 1 mM isopropyl-1thio-␤-D-galactopyranoside; cells were grown in LB broth at 27°C. Pa-CcmH* was purified in two steps with nickel affinity and ion-exchange chromatography. The yield of purified protein was ϳ25 mg/liter. After tag cleavage with thrombin, Pa-CcmH* was again subjected to ion-exchange purification. Selenomethionine-substituted Pa-CcmH* was expressed in BL21 (DE3) E. coli cells in minimal medium supplemented with 50 mg/liter D-L-selenomethionine following standard procedures and was purified under the same conditions as the wild type protein.
Protein Crystallization-Crystallization conditions were 0.1 M Tris-HCl buffer (pH 8.5) containing 20% polyethylene glycol 1000 at room temperature. Drops were prepared by mixing 2.5 l of precipitant solution and 1.0 l of protein (13 mg/ml) and allowed to equilibrate against 0.5 ml. High quality crystals for the Se-Met protein were obtained by mixing 1 l of protein (9 mg/ml) and 1 l of the reservoir solution consisting of 38% polyethylene glycol 6000, 0.8 M NaCl, and 0.1 M Tris-HCl, pH 8.0. Crystals appeared in 2 days and grew to full size in 1 week. Diffraction data for both native and Se-Met crystals were collected at cryogenic temperature (100 K) at the European Synchrotron Radiation Facility (ESRF) Synchrotron (Grenoble, France). Data were indexed and integrated with DENZO and SCALEPACK (22). One molecule was found in the asymmetric unit for both the native and the Se-Met derivative crystals.
Crystal Structure Determination and Refinement-Pa-CcmH* structure was solved by a combination of MAD and MR procedures. MAD was carried out on the Se-Met derivative data. The program SOLVE (23) identified one selenium site. Initial phase calculation led to an overall figure of merit of 0.61 at 35-3.5 Å resolution. Solvent flattening with RESOLVE (24) improved the quality of the map, yielding a figure of merit of 0.8 at 2.5 Å resolution. The ARP-WARP (25) automatic building procedure provided a partially built model. This model was then used as a template to perform an MR with the native data set using AMORE (26) and gave a complete and easily interpretable electron density map at 1.7 Å resolution. The model (residue 1 in the structure corresponds to Ala 21 in the full-length protein) was completely built, and water molecules were added using COOT (27). Several cycles of refinement (REFMAC5 (28)), visual inspection, and manual rebuilding led to a final R factor of 21.0% and R free of 23.5% at 1.7 Å resolution. The average atomic B-factor for the structure is 31.16 Å 2 . The model quality was checked with PROCHECK (29); all residues lie in the most favored and additionally allowed regions of the Ramachandran plot. The atomic coordinates have been deposited in the Protein Data Bank (Protein Data Bank ID code 2HL7). Analysis of the cavities was performed with the program CastP (30).
Biochemical Analyses-Pa-apocyt was obtained from purified Pa-cyt c 551 (prepared as described in Ref. 31) following the original protocol of Anfinsen and co-workers (32). Standard redox potentials (E 0 Ј) of Pa-CcmH* and Pa-apocyt were determined as described previously (33).
pK a Determination of the Active Site Cysteines-Reduced Pa-CcmH* (5 M) was mixed with an excess of 5,5Ј-dithio-bis(2nitrobenzoic acid) (DTNB; 50 -100 M) in the pH range 4.0 -9.5 in the Pi-Star stopped-flow instrument (Applied Photophysics, Leatherhead, UK), and the reaction was monitored following the absorbance at 412 nm. Absorbance data were fitted to a single exponential function to obtain an observed, pseudo-first order rate constant (k obs ). pK a values were determined by plotting k obs values as a function of pH and fitting to an appropriate equation. For single pK a processes, Equation 1 was used, where k SH and k SϪ are the rate constants for the protonated and deprotonated forms, respectively (34). For two proton dissociation events, an equation that takes into account two pK a values where k 1 is the association or on-rate constant, k Ϫ1 is the dissociation or off-rate constant, and [A 0 ] is the initial concentrations of the varied reagent (36).

RESULTS
Overall Structure of CcmH-The soluble redox-active N-terminal domain of Pa-CcmH (Pa-CcmH*), obtained by removing the N-terminal signal peptide and the C-terminal membraneanchoring helix, is composed of 80 residues. This domain was expressed in E. coli, purified, and crystallized, and its threedimensional structure was solved by a combination of MAD and MR (see "Experimental Procedures"). The high quality electron density map obtained at 2.5 Å resolution by MAD using a Se-Met derivative enabled us to build a partial model of the protein. MR was employed to position the model in a different crystal form, obtaining a complete model at higher resolution. The final structural model consisting of 4 residues derived from the His tag cleavage site (numbered from Ϫ4 to Ϫ1) plus 78 residues of the protein domain (Val 79 and Asn 80 were excluded because they were undetectable in the electron density) was refined at 1.7 Å resolution with R work ϭ 21% and R free ϭ 23.5%. The model presents a good stereochemistry with all residues located in the allowed regions of the Ramachandran plot (see "Experimental Procedures" and Table 1).
The fold of Pa-CcmH* (Fig. 1A) consists of a three-helix bundle stabilized by hydrophobic interactions, with a short N-terminal extension and a C-terminal tail connecting the bundle to the transmembrane helix in the full-length protein. Helices ␣1 (residues 10 -22) and ␣2 (residues 39 -54) are connected by a long loop (residues 23-38), whereas helices ␣2 and ␣3 are linked by a short loop (residues 58 -69). A search using DALI (37) returned significant hits only with unrelated ␣-helical proteins, indicating that the three-helix bundle fold of Pa-CcmH* is unique among redox-active proteins.
In Pa-CcmH*, the redox-active motif Cys 25 -Pro 26 -Lys 27 -Cys 28 is part of the loop connecting helices ␣1 and ␣2 (Fig. 1, A  and B). The electron density for the side chains of these 2 Cys residues is continuous, indicating that the protein is in the oxidized state (supplemental Fig. 1). Attempts to crystallize the reduced form were unsuccessful. The conformation of the disulfide bond, albeit located in a peculiar fold, resembles that of the TRX and TRX-like proteins, displaying the typical righthand hook conformation with the s-s dihedral angle ϭ 81.1°, the S␥-S␥ distance ϭ 2.02 Å, and the C␣-C␣ distance ϭ 5.34 Å (38,39). The accepted catalytic mechanism of TRX proteins (40) assumes that the solvent-exposed N-terminal cysteine of the CXXC motif performs a nucleophilic attack on the disulfide bridge of the target substrate. This process leads to a mixed disulfide complex that is subsequently resolved by the buried C-terminal Cys activated by a nearby conserved acidic residue (Asp or Glu) (41)(42)(43). Interestingly, analysis of the accessible surface area reveals that in Pa-CcmH*, the N-terminal Cys 25 residue is buried, whereas the C-terminal Cys 28 is solvent-exposed, at variance with the canonical arrangement observed in the TRX superfamily where the N-terminal Cys of the CXXC motif is always solvent-exposed. Moreover, neither the acidic Asp/Glu residue nor the conserved and functionally important cis-Pro residue (44) are present in the active site, suggesting that Pa-CcmH* may catalyze thiol-disulfide exchange by a different mechanism.
Multiple sequence alignment of CcmH homologues from various organisms (Fig. 2) shows that, in addition to the catalytic residues Cys 25 and Cys 28 , 10 other residues are fully conserved, mostly located near the active site (Fig. 1B). Gly 55 and Gly 70 act as helix-breaking residues and are located at the end of the ␣2 and ␣3 helices, respectively; Leu 23 and Val 73 side chains point to the hydrophobic core of the bundle and presumably fulfill a structural role; Arg 24 and Gln 29 are located close to the catalytic Cys and may be involved in the binding to the redox partners of CcmH, apocyt, and/or CcmG. The hydroxyl group of the fully conserved Ser 36 is at 6.2 Å distance from the S␥ of the Cys 28 , making implausible a direct interaction between these 2 residues. Interestingly, however, a water molecule found in the crystal structure is at 3.88 Å from the S␥ of Cys 28 and H-bonded to the -OH group of Ser 36 (2.59 Å). This H-bonding network involving a water molecule seems to be highly conserved in the CcmH proteins since the position of Ser 36 is secured by the presence of the fully conserved Arg 45 , whose guanidinium group is H-bonded to the backbone oxygen of Ser 36 (2.70 and 3.09 Å, respectively) (Fig. 1C). Taken together, evidence suggests a possible role for Ser 36 in the activation of Cys 28 through a water molecule (see below).
The side chain of Ser 36 is also found to line a small cavity (volume ϭ 89.8 Å 3 ). This pocket, located behind the catalytic cysteines, is surrounded by conserved or conservatively mutated polar and hydrophobic residues (Cys 25 , Cys 28 , Gln 29 , Asn 30 , Gln 31 , Ile 33 , Ser 36 , Ala 38 , Ile 40 , Ala 41 , Leu 44 ). Although the overall hydrophobic character of the pocket is suitable for the interaction with an unfolded substrate, such as the heme binding motif of apocyt, the presence of conserved polar residues may be strategic in the recognition of the conserved His of the CXXCH motif of c-type cytochromes.
Determination of the Redox Potential of Pa-CcmH* and Pa-apocyt-To provide information about the thermodynamic driving force for the flow of reducing equivalents in the redox pathway involving CcmH and apocyt in P. aeruginosa, we determined the redox potential (E 0 Ј) of Pa-CcmH* and Pa-apocyt. The difference in electrophoretic mobilities of the reduced-modified form and the oxidized form was used to estimate the E 0 Ј values of Pa-CcmH* and Pa-apocyt following Inaba and Ito (33). Fit of the data (supplemental Fig. 2) to the Nernst equation yields E 0 Ј ϭ Ϫ215 mV for Pa-CcmH* and E 0 Ј ϭ Ϫ190 mV for Pa-apocyt. These values are in agreement with the estimates previously reported for the homologous protein Ccl2 from  SEPTEMBER 14, 2007 • VOLUME 282 • NUMBER 37

JOURNAL OF BIOLOGICAL CHEMISTRY 27015
R. capsulatus and for a model peptide of apocyt c 2 (Ϫ210 and 170 mV, respectively) (45). The slightly more positive E 0 Ј value of Pa-apocyt is consistent with its reduction by Pa-CcmH*; therefore we conclude that the difference in redox potential, although small, is thermodynamically favorable.
The pK a of the Active Site Cysteines-The pH dependence of the rate of reaction between DTNB and Pa-CcmH* was used to determine the pK a of the Cys residues. Control experiments carried out under pseudo-first order conditions at selected pH values confirmed that the reaction is second order (data not shown). At variance with what was observed in the case of TRX and TRX-like proteins (43), the time course was always monophasic (Fig. 3A, inset), suggesting a one-step bimolecular reaction mechanism. The pH dependence of the pseudo-first order rate constant for DTNB reduction by Pa-CcmH* (Fig. 3A) displayed a minor systematic deviation from a model accounting for a single titratable group, showing a Hill coefficient of about 0.8 (data analysis not shown). Thus, to assess the pK a of the active site thiols individually, two single variants of Pa-CcmH* (C25S and C28S) were generated and purified. The pseudo-first order rate constant of the reaction with DTNB of these variants at various pH values (Fig. 3B) can be fitted to a one-proton dissociation process, with very similar pK a values (8.4 Ϯ 0.1 for Cys 25 and 8.6 Ϯ 0.1 for Cys 28 ; see also Ref. (45). In the case of TRX and TRX-like proteins, the pK a values of the catalytic Cys residues are widely separated, being in the range of 6.7-7.5 for the N-terminal reactive cysteine and Ͼ9 for the C-terminal one (42,46). Thus, the active site Cys residues of Pa-CcmH* have unusual pK a values when compared with canonical TRX proteins. Recently however, also for the TRX-like ResA protein from Bacillus subtilis, the two Cys residues were reported to have similarly close pK a values (12); in this case, the crystal structure of the reduced state of the protein, responsible for the reduction of apocyt in System II, reveals an unusually large separation between the two thiols, which may account for the result.
Comparison of the pH dependence of the reduction of DTNB by Pa-CcmH* (Fig. 3A) with that of the two single Cys variants (Fig. 3B) highlights faster pseudo-first order rate constants in the former case under the same conditions. This unpredicted observation can be explained on the basis of the kinetic mechanism if both Cys residues were able to reduce DTNB. Although each Cys residue can react individually with DTNB because of their similar pK a values, their close proximity in the three-dimensional structure of Pa-CcmH* prevents them from reacting simultaneously, giving rise to a partitioning effect between two alternative parallel pathways (involving either Cys 25 or Cys 28 , respectively). Such a partitioning process is clearly not available in the case of the single Cys variants, which therefore react, under the same conditions, with slower rate constants. These effects may account for the low Hill coefficient calculated for the wild type Pa-CcmH*, whose titration results from a complex combination of two different processes (36).
Interaction with Pa-apocyt-As outlined above, it is generally accepted that CcmH may act as a specific thiol-oxidoreductase

Three-dimensional Structure of CcmH and Its Binding to apocyt
in the redox pathway of the c-type cytochromes biogenesis (4,6). As a control, we observed that, similarly to its homologues Ccl2 (13) and AtCCMH (21), Pa-CcmH* is unable to reduce insulin in vitro (data not shown).
The ability of Pa-CcmH* to interact with apocyt was initially tested in an experiment whereby reduced Pa-CcmH* was labeled with DTNB and then either mixed with E. coli TRX or mixed with a nonapeptide corresponding to the heme binding motif of Pa-apocyt (KGCVACHAI). The results (data not shown) indicate that only the peptide releases thionitrobenzoic acid and thus can interact with the labeled cysteines of Pa-CcmH*.
To further characterize the interaction in solution between Pa-CcmH* and the same Pa-apocyt nonapeptide, we carried out fluorescence stopped-flow binding experiments. We followed the binding kinetics by FRET between a fluorescence donor (an engineered variant of Pa-CcmH* with the substitution Tyr 64 to Trp; Y64W) and an acceptor (a dansyl group covalently bound to the N-terminal Lys residue of the peptide). Upon mixing the peptide with fully oxidized Pa-CcmH*, no reaction was observed. On the other hand, complex formation was detected by challenging the peptide with reduced Pa-CcmH* (see "Experimental Procedures"). The time courses obtained under pseudo-first order conditions matched single exponentials at all concentrations of Pa-CcmH* (always in excess) (Fig. 4A). A plot of the observed rate constant against Pa-CcmH* concentration (Fig. 4B) shows a minor deviation from second-order kinetics; fitting the data to Equation 2 (see "Experimental Procedures") yields the two microscopic rate constants for the binding of Pa-CcmH* to the dansylated nonapeptide (k on Ӎ 0.05 M Ϫ1 s Ϫ1 and k off Ӎ 0.5 s Ϫ1 ), from which an equilibrium dissociation constant (K D ) of ϳ10 M was calculated. A similar K D value was calculated from the amplitudes of the kinetic traces (data not shown). Nevertheless, we should point out that this experiment demands that the two catalytic partners are in opposite redox states; since we were unable to isolate fully reduced fractions of Pa-CcmH*, the actual concentration of reduced Pa-CcmH* is uncertain, and thus the K D value is very likely to be overestimated. When the same experiment was carried out using the two single Cys variants (Y64W-C25S and Y64W-C28S), no binding was observed by FRET (data not shown); apparently in vitro, both cysteines are required for the recognition/binding reaction of Pa-CcmH* to the Pa-apocyt-mimicking nonapeptide. These results imply that a more complex scenario may be more appropriate to describe the binding reaction of Pa-CcmH* to the Pa-apocyt, as mirrored by the systematic deviation from linearity of the observed pseudo-first order rate constant, and given that in vivo, both cysteines of CcmH are necessary for cytochrome c maturation during anaerobic growth (14).

DISCUSSION
The maturation of c-type cytochromes in Gram-negative bacteria occurs in the periplasm; it is in this oxidizing compartment that the newly synthesized polypeptide chain is translocated and covalently linked to the heme group. Although it is still not known how this process is carried out, it is clear that attachment of the heme to the (presumably unfolded) apocyt requires a remarkably specific reduction of both cysteines of its heme binding motif. In System I, CcmH and CcmG are the two proteins of the cytochrome c maturation (Ccm) machinery that were suggested to be implicated in this high fidelity process.
The crystal structure of the periplasmic domain of Pa-CcmH* in the oxidized state, illustrated above, is the first available three-dimensional structure of a CcmH protein and reveals that this enzyme adopts a three-helix bundle fold that was never observed in any other thiol-oxidoreductase; as discussed below, this unique structural fold mirrors the functional properties of Pa-CcmH*. Although the geometry of the disulfide bond is the same as described for TRX and TRX-like proteins (39), the topology of the 2 active site Cys residues (Cys 25 and Cys 28 ) is reversed (Fig. 1). Indeed, in Pa-CcmH*, the N-terminal Cys 25 is buried, whereas the C-terminal Cys 28 is solvent- exposed; this suggested to us that, at variance with TRX proteins, the Cys 28 is the one presumed to perform the nucleophilic attack on the apocyt disulfide bond. Although this hypothesis is based on the crystal structure of oxidized Pa-CcmH*, it would imply a mechanism different from that described in TRX and TRX-like proteins where the reactive cysteine is the N-terminal one, as substantiated by a model previously proposed for the reduction of apocyt by CcmH (14).
This peculiarity of the active site cysteines of Pa-CcmH* is also supported by their pK a values that we found to be quite similar (8.4 Ϯ 0.1 for Cys 25 and 8.6 Ϯ 0.1 for Cys 28 ). In TRX proteins characterized by aspecific thiol-oxidoreductase activity, the pK a value of the cysteine performing the initial nucleophilic attack on its target substrate is significantly lower (i.e. ϳ7); as recently discussed for ResA, a TRX-like protein involved in System II (12), the higher pK a values of the Pa-CcmH* active site thiols may ensure that this component of the Ccm apparatus is unreactive toward nonspecific substrates.
Although a detailed description of the catalytic mechanism of Pa-CcmH* would profit from the determination of the structure of the reduced state, we propose that Ser 36 , which in Pa-CcmH* is close to the active site (Fig. 1, B and C) and strictly conserved in all CcmH homologues (Fig. 2), may play a role in either one or both of two crucial events, i.e. (i) recognition of the conserved His residue in the heme binding motif of apocyt and/or (ii) activation of Cys 28 . Indeed, on the basis of its pK a ϭ 8.6 Ϯ 0.1, Cys 28 is expected to be protonated in the reduced state, and thus its activation may be mediated by the polarization of the water molecule (Fig. 1C), which is H-bonded to the strictly conserved Ser 36 . The lack of strong activating acidic residues close to the active site of Pa-CcmH* may be necessary to avoid Cys 28 performing a nucleophilic attack easily on aspecific disulfide-containing substrates. Interestingly, a careful examination of the crystal structures of all available oxidized and reduced TRXs shows that a topologically equivalent water molecule is always present and, indeed, it has been proposed to mediate the activation of the reactive N-terminal Cys residue in these oxidoreductases (41).
In TRX and TRX-like proteins, resolution of the mixed disulfide complex is subsequently carried out via a nucleophilic attack by the C-terminal Cys residue on the intermolecular disulfide (40). In the case of Pa-CcmH*, however, the higher pK a of the N-terminal Cys 25 and the absence of activating acidic residues in the active site region support a different mechanism. Rupture of the mixed disulfide and release of reduced Pa-apocyt should involve a second redox partner of the Ccm maturase complex, such as CcmG (14,15); indeed, previous in vitro experiments with the R. capsulatus homologue Ccl2 supported an interaction with the CcmG homologue HelX (13).
How can Pa-CcmH* cope with the requirement to recognize and interact with two different redox partners such as Pa-apocyt and Pa-CcmG? Although it cannot be excluded that reduced Pa-CcmH* may adopt a different conformation, the crystal structure reported here highlights the presence of a small pocket on the surface of Pa-CcmH*, close to the active site disulfide bond; it is tempting to speculate that this cavity, surrounded by conserved hydrophobic and polar residues, could represent the recognition site for the heme binding motif of apocyt. This possibility is made more likely since a small cavity with similar features has been recently identified in the structure of reduced ResA and implicated in apocyt recognition (11). On the other hand, it is plausible that Pa-CcmG recognition may be mediated by some conserved basic/polar residues located near the active site of Pa-CcmH* (i.e. Arg 24 and Gln 29 ); indeed, the unusually acidic nature of the redox-active center of E. coli CcmG has been hypothesized to be important in the interaction with its CcmH partner (17,18).
Finally, we want to highlight that our kinetic data on Pa-CcmH* provide the first quantitative in vitro evidence for the recognition and binding of a Ccm component to its apocyt substrate; this result was obtained by mixing with a nonapeptide (dansyl-KGCVACHAI), which is identical to the heme binding motif of Pa cyt c 551 , the physiological partner. The measured off-rate constant (in the 0.2-1 s Ϫ1 range) combines an adequate affinity (low M) with the need to release reduced apocyt to other component(s) of the System I maturase complex.
On the basis of the structural and functional data on Pa-CcmH* presented here, it is conceivable to envisage an assembly line of holo cytochrome c in which reduced CcmH (a non-TRX-like thiol-oxidoreductase) specifically recognizes, binds, and reduces oxidized apocyt via the formation of a mixed disulfide complex, subsequently resolved by CcmG, a TRX-like thiol-oxidoreductase. This view supports the general picture whereby redox cascades are mediated by the interaction between proteins alternating a TRX-like and a non-TRX-like fold (19,47). Further characterization of the multienzymatic redox pathway involved in the maturation of c-type cytochromes will require structural and biochemical investigations of the recognition processes within the ternary complex involving CcmH, apocyt, and CcmG.