Analysis of the HypC-hycE complex, a key intermediate in the assembly of the metal center of the Escherichia coli hydrogenase 3.

The formation of a complex between the specific chaperone-type protein HypC and the precursor form of the large subunit HycE in the maturation pathway of hydrogenase 3 from Escherichia coli has been studied by targeted replacement of amino acids in both proteins. HypC and its homologs contain the motif MC(L/I/V)(G/A)(L/I/V)P at the amino terminus, from which the methionine residue is post-translationally removed. The exchange of the cysteine residue led to complete loss of the ability to interact with the precursor form of HycE, but replacement of the proline residue had no effect. Site-directed replacement of the conserved cysteine residues in HycE involved in nickel binding was also performed. Exchange of Cys(241) resulted in the inability of the HycE variant to interact with HypC and to incorporate nickel. The variants of HycE in which Cys(244) and Cys(531) were replaced by alanine residues were unable to incorporate nickel, although the mutated proteins could interact with HypC. Intriguingly, the precursor of HycE in which the Cys(534) residue was exchanged could form the complex with HypC, could incorporate nickel, and was C-terminally processed, but it delivered an inactive enzyme. Our findings are in favor of a model in which binding of HypC masks Cys(241); Cys(244) and Cys(531) bind the iron and nickel moieties, respectively; and C534 closes the bridge between the two metals after C-terminal processing has taken place.

Despite the growing knowledge about the structures and catalytic or other functional roles for metal centers in many metalloproteins, information as to how metal centers are biosynthetically assembled is scarce. Recent studies on several systems (1)(2)(3)(4)(5) revealed the involvement of a surprisingly complex cascade of reactions in the maturation process in which numerous auxiliary proteins are implicated. One class of them was found to participate in metal center assembly by stabilizing a partially unfolded apoprotein conformation that is competent for the specific binding of a metal ion or a metal cofactor (1)(2)(3)(4)(5). This mechanism is reminiscent of the role postulated for molecular chaperones, which facilitate the correct folding of nascent polypeptide chains in vivo but are not components of the final structures. Accordingly, these particular auxiliary proteins were designated specific chaperone-type proteins (1)(2)(3)(4)(5), their mode of action and the basis of their specificity being not yet fully understood.
One type of enzymes whose metal center formation has been studied in some detail are hydrogenases. They play a central role in the metabolism of many microorganisms by catalyzing the production or consumption of molecular hydrogen according to the reaction, H 2 7 2H ϩ ϩ 2e Ϫ . These enzymes can be divided into two major families: iron-only hydrogenases ([Fe]hydrogenases) and nickel-iron hydrogenases ([NiFe]-hydrogenases) (6,7). They are composed of a small electron transfer subunit (28)(29)(30)(31)(32)(33)(34)(35) and a large catalytic subunit (45-65 kDa). In [NiFe]-hydrogenases, the large subunit harbors the metal center ligated to the protein by one motif in the amino-terminal portion of the polypeptide (RXCX 2 CX 3 H) and a second one at the carboxyl-terminal region (DPCX 2 CX 2 (H/R)) (6,8). A major and so far highly conserved characteristic of [NiFe]-hydrogenases is their complex biogenesis, which involves at least seven accessory gene products (9). Although this pathway is far from being understood, substantial progress has been made regarding the maturation of the catalytic subunit. A current model suggests that maturation is initiated by the binding of iron together with the diatomic ligands followed by nickel insertion. Finally, the metal-containing precursor is converted to the mature form by proteolytic removal of a C-terminal extension, mediated by a specific protease (10 -13). The three-dimensional structure of one of these proteases, namely the hydrogenase 2-specific protease of Escherichia coli, HybD (14), has been resolved recently by x-ray crystallography. Another accessory protein, HypC, is a specific chaperone-type protein required for the maturation of the catalytic subunit of the hydrogenase 3 (5). Such genes coding for auxiliary proteins have been found in all organisms synthesizing [NiFe]-hydrogenases.
Here we report on the interaction between the precursor form (pre-HycE) of the large subunit of the hydrogenase 3 from E. coli and the chaperone-type protein HypC. We identified cysteines both in HypC and pre-HycE as being essential for the interaction. A model is presented for the role of the cysteine residues of pre-HycE in the metal incorporation process.
Site-directed Mutagenesis-Site-directed mutagenesis was performed by recombinant polymerase chain reaction either on the pACE1 plasmid (18) to yield hycE mutants or on the pJA1021 plasmid (17) to yield hypC mutants. The Expand High Fidelity polymerase chain reaction system (Roche Molecular Biochemicals) was used. This approach is based on the polymerase chain reaction amplification of an entire plasmid DNA by mutagenic primers divergently oriented but overlapping in their 5Ј-ends (19 -21). The resulting linear DNA molecules are then transformed into the recA Ϫ DH5␣ strain and replicating circles are subsequently recovered by recombination between the homologous ends of the linear template. The cysteine residues at positions 241, 244, 531, and 534 of HycE were replaced by alanine via this procedure using the following primer pairs: EC241A-EC241Arev, EC244A-EC244Arev, EC531A-EC531Arev, and EC534A-EC534Arev (Table I). The Cys 2 and the Pro 6 residues of HypC were replaced using the following primer pairs: HypC2rev-HypC2A, HypC2rev-HypC2S, HypC2rev-HypC2R, and HypCP6rev-HypCP6 (Table I). All mutations were confirmed by DNA sequencing. The resulting mutated plasmids were used to transform E. coli HD705 in the case of the pACE1 derivatives or DHP-C in the case of the pJA1021 plasmid series.
Growth Conditions and Preparation of Extracts-Strains were cultivated anaerobically in TGYEP medium containing 30 mM sodium formate as described previously (5) and supplemented with a 1 M concentration each of sodium molybdate and sodium selenite and 5 M nickel chloride. When required, chloramphenicol was added at a final concentration of 30 g/ml. Following the addition of a 1% inoculum, cultures were grown at 37°C until they reached an optical density of 1.0 at 600 nm (A 600 ). Cells were harvested by centrifugation and washed in 50 mM Tris-HCl, pH 7.4. Cytoplasmic fractions obtained by ultracentrifugation at 100,000 ϫ g (named S100 extracts), and crude extracts were prepared as described previously (5).
Enzyme Assays-Total hydrogenase activity of crude extracts was assayed by H 2 -dependent reduction of benzyl viologen as described by Ballantine and Boxer (22) and expressed in mol of H 2 /min/mg of total protein.
Polyacrylamide Gel Electrophoresis and Western Blotting-Proteins separated on a 10% SDS polyacrylamide gel by using the discontinuous buffer system of Laemmli (23) were transferred onto nitrocellulose membranes and blotted with anti-HycE antibody (1:1000) or anti-HypC antibody (1:500). For antibody detection, the protein A-horseradish peroxidase conjugate (Bio-Rad, Munich, Germany) and the Chemiluminescent Lumilight kit from Roche Molecular Biochemicals were used. Electrophoresis on nondenaturing gels was conducted as described previously (5), employing 40 g of total protein from S100 extracts.
Nickel Incorporation Experiments-Strains were cultivated at 37°C in 250 ml of TGYEP medium supplemented with 30 mM formate, 1 M sodium molybdate, 1 M sodium selenite, and 150 nM 63 Ni (specific radioactivity, 190 Ci/mol). At an A 600 of 1.0 -1.2, the cells were harvested by centrifugation, washed, and resuspended in 50 mM Tris-HCl, pH 7.4, containing a 30 g/ml concentration each of phenylmethylsulfonyl fluoride and DNase I. S100 extracts were then prepared and applied onto a 10% nondenaturing gel. After electrophoresis, the gel was dried and exposed for autoradiography for 1 week.
Covalent Modification of Protein Thiols and Nondenaturing Polyacrylamide Gel Electrophoresis (ND-PAGE) 1 -40 g of total protein from S100 extracts were incubated for 15 min with or without 15 mM dithiothreitol followed by the addition of one-fourth volume of 0.5 M iodoacetamide or potassium iodoacetate in 1 M Tris-HCl, pH 8.7. After 5 min or 30 min at room temperature, the samples were mixed with the sample buffer and analyzed by electrophoresis on a 10% nondenaturing polyacrylamide gel.

Identification of Key Residues in
HypC for the Maturation of the Hydrogenase 3-Among the auxiliary proteins required for hydrogenase maturation is a family of small proteins from 80 to 110 residues in size with an acidic character (pI ϳ4 -5). They are products of hypC, hybG, or hupF genes. It was shown that HypC from E. coli is able to form a stable complex with the immature form of the large subunit of the hydrogenase 3, HycE (5). A comparison of the amino acid-derived sequence of HypC with those of homologs from other organisms is given in Fig. 1. Noteworthy is the presence of a conserved motif at the aminoterminal part of the protein, namely MC(L/I/V)(G/A)(L/I/V)P. Amino-terminal sequencing of the purified HypC protein indicated that the methionine residue is post-translationally removed (data not shown), yielding an N terminus with a reactive thiolate. To examine its possible role in the maturation process, this cysteine residue (Cys 2 ) and Pro 6 from HypC were chosen as candidates for site-directed mutagenesis and were replaced individually by either neutral (alanine and serine) or charged (arginine) amino acids and by alanine and threonine, respectively. The resulting pJA1021 derivatives carrying the mutated genes were used to transform strain DHP-C, which lacks the chromosomal hypC gene.
The transformants were grown under fermentative conditions, and the cells were analyzed for H 2 -dependent benzyl viologen reduction (Table II) and for processing of the large subunit HycE (Fig. 2). Transformants carrying plasmids in which Cys 2 of HypC was replaced were devoid of hydrogenase 3 activity, supporting the idea that the maturation of the enzyme may be blocked. In contrast, replacement of Pro 6 had no effect. To ascertain that the cellular level of the mutant proteins was not changed, the transformants were analyzed by immunoblotting. It was found that under fermentative conditions they produced the same amount of HypC as present in the strain transformed with the plasmid carrying the wild-type hypC gene (data not shown). Fig. 2 displays the immunoblotting analysis of the maturation state of HycE in the transformants carrying the different hypC alleles. HycE was present in the parental strain MC4100 in the mature form (Fig. 2, lane 1) and in the DHP-C strain in the precursor form (Fig. 2, lane 3). DHP-C transformants with the pJA1021 plasmid (Fig. 2, lane 4) or that coding for the Pro 6 variant (data not shown) exhibited a partial restoration of the maturation of HycE, confirming that these substitutions did not abolish the HypC function. Laser densitometry analysis indicated that the processing level was approximately 30%, which agrees with previous findings (17) that complementation of a DHP-C strain by the respective gene in trans restores maturation of hydrogenase 3 only to a limited extent. The lack of full complementation has been in- 1 The abbreviations used are: ND, nondenaturing; PAGE, polyacrylamide gel electrophoresis

Interaction between HypC and Pre-HycE
terpreted previously by the assumption that maturation follows a channeled pathway that requires the stoichiometric amount and activity of the components involved (17, 24 -26). Any effect of overexpression of the hypC gene in trans on the maturation process can be ruled out (Table II). In agreement with the inability of the Cys 2 mutant proteins to support the generation of hydrogenase activity, HycE is not processed in these strains (Fig. 2, lanes 5-7).
To gain further insight into the impairment of the HypC function, the ability of the HypC variants to form a complex with the precursor form of HycE during the maturation process was examined. A ND-PAGE system was employed to study in vivo complex formation between HypC and pre-HycE (5). S100 extracts from strains transformed with the different plasmids were separated by ND-PAGE, and Western blot experiments were performed using antisera directed against HypC. The results show that the Cys 2 mutants were no longer able to form a complex with the immature form of HycE (Fig. 3). They migrate in the position of free HypC with the interesting exception of the C2S variant. There is no obvious explanation for its retarded migration, but it could be the consequence of some modification at the hydroxyl or at the amino group of the amino-terminal serine. Again, the variant in which Pro 6 was replaced showed the same pattern as the wild-type protein (data not shown).
Site   (37), Archaeoglobus fulgidus (38), Methanobacterium thermoautotrophicum (39), and Helicobacter pylori J99 (40). The Clust-alW algorithm was used with a PAM250 weight table, and the consensus sequence is indicated above the sequence alignment. replacement in HypC, a plausible assumption was that one of the conserved cysteine residues in HycE may be also involved. Hence, each of the cysteine residues present in the two nickelbinding motifs of HycE were substituted by an alanine residue via site-directed mutagenesis, which led to genes coding for the C241A, C244A, C531A, and C534A mutant HycE proteins. Under fermentative conditions, none of the mutants exhibited hydrogenase 3 activity (Table II). These observations were corroborated by the results of the immunoblotting analysis of HycE. As already noted previously for other auxiliary proteins involved in hydrogenase maturation (17,18,24,26), the deletion of the chromosomal hycE gene cannot be fully complemented by expression of hycE in trans (Fig. 4, lane 3), possibly because of some imbalance in stoichiometry between pre-HycE and the components of the maturation machinery. Despite a low processing level, the expression of hycE in trans yielded hydrogenase 3 activity nearly identical to that of the parental strain (Table II). With the striking exception of the C534A mutant, none of the other mutant proteins were subjected to HycI-mediated proteolysis (Fig. 4). Intriguingly, therefore, replacement of Cys 534 by alanine allows C-terminal processing but prevents development of hydrogenase activity (Table II).
We then investigated whether the mutated HycE polypeptides, notably the C534A variant, could incorporate nickel. To this end, cultures were grown in presence of 63 Ni, S100 extracts were separated by ND-PAGE, and 63 Ni-labeled proteins were identified by autoradiography (Fig. 5) and Western blot analysis (Fig. 6). In the case of the parental strain MC4100 (Fig. 5,  lane 1, band 1), a major 63 Ni signal comigrated with material immunoreacting with antiserum directed against HycE. It corresponds to the mature form of HycE as displayed by ND-PAGE (Fig. 6A, lane 1). In addition, a slower migrating 63 Nilabeled protein band also present in the HD705 extract (which is devoid of the hycE gene) can be seen (Fig. 5, lane 2). It may represent the large subunit of either hydrogenase 1 or 2. In the case of the HD709 extract, two major pre-HycE conformers produced in absence of the HycI protease and identified by Western blot analysis (Fig. 6A, lane 3) contained 63 Ni (Fig. 5,  lane 3, bands 2 and 3), one of them being associated with HypC (band 2) (see Fig. 6). Finally, from the four HycE variants, only that carrying the C534A replacement gave a 63 Ni-labeled protein band that immunoreacted with anti-HycE antiserum and was equivalent to the signal obtained with the transformant carrying the plasmid with the wild-type hycE gene (Fig. 5,  lanes 5 and 9, band 1). These results support the notion that nickel incorporation in the C534A variant is not accompanied by the generation of hydrogenase 3 activity.
In order to check if any of the cysteine substitutions in HycE impaired the ability for complex formation with HypC during the maturation process, each of the transformants was analyzed with the procedure described previously (5). As expected, the C534A protein that underwent HycI-mediated processing was able to specifically interact with HypC (Fig. 6B, lane 7). The same result was obtained for the transformants possessing the HycE variants C244A and C531A. On the other hand, the C241A variant was unable to interact with HypC (Fig. 6B). It was previously pointed out (5) that only a certain conformer of pre-HycE is competent to bind HypC (Fig. 6A, lane 3, band 2), whereas the majority of the immature polypeptide cannot participate in this interaction (Fig. 6A, lane 3, band 3), at least under the experimental conditions employed. Indeed, the absence of the characteristic fastest migrating immunoreactive form of HycE in the C241A mutant (Fig. 6A, lane 4, band 2) is paralleled by the absence of a HypC-pre-HycE complex (Fig.  6B, lane 4).

Effect of Reducing Agents and Alkylating Agents on the Stability of the HypC-Pre-HycE Complex-The results described thus far had indicated that cysteine residues both in HypC and
HycE are essential for the complex formation between the two proteins. This prompted the analysis of whether the formation of a disulfide bridge could be responsible for the protein-protein interaction. For this purpose, S100 extracts of strain HD709, which contain large amount of the complex (Fig. 6B, lane 3), were analyzed under both reducing and nonreducing conditions by SDS-PAGE followed by immunoblotting analysis using antiserum directed against HycE. There was no alteration of the migration of the HycE polypeptide under nonreducing conditions (Fig. 7, lanes 2 and 5). As a control, extracts from strain DHP-C (⌬hypC), which contain the precursor form of HycE, were analyzed under identical conditions, and an identical pattern was observed. These observations provide circumstantial evidence that HypC does not interact with pre-HycE via the formation of a disulfide bridge that under nonreducing conditions should give rise to a slower migrating species.
To further characterize the interaction, HD709 extracts were treated with different alkylating or reducing reagents and then separated by ND-PAGE. Although some dissociation occurred as indicated by the appearance of free HypC, the presence of 10 or 25 mM dithiothreitol was not sufficient to fully resolve the complex (data not shown). In the absence of reducing agents, the complex is extremely stable, since no dissociation is detectable after 2 h of incubation.
The amenability of the cysteine residues within the complex to alkylation was then determined by their reaction with iodoacetic acid or iodoacetamide. Thiols potentially masked in disulfide bonds under these native conditions were reduced by preincubation with dithiothreitol, and the electrophoretic behavior of HypC associated or not with pre-HycE was examined by ND-PAGE. The reaction of such S100 extracts from strain HD709 treated with each of the alkylating agents either under reducing or nonreducing conditions resulted in the complete dissociation of the complex after only 5 min of incubation (Fig.   FIG. 4.

Immunoblotting analysis of HycE precursor and mature forms in transformants expressing HycE mutant forms.
Crude extracts (30 g of total proteins) were subjected to SDS-PAGE as indicated in Fig. 2. Lane 1, MC4100;  8). Any accessible thiolate group not involved in disulfide bond formation should react and subsequently carry one negative charge after modification with iodoacetic acid. The different electrophoretic mobility of the liberated HypC revealed that alkylation of the protein had occurred. DISCUSSION A key role in the maturation process of [NiFe]-hydrogenases has been demonstrated recently for the chaperone-type protein HypC and its interaction with the HycE apoprotein of the E. coli hydrogenase 3 (5). This complex has been identified in each of the mutants with a lesion in hyp genes, supporting the idea that its formation may constitute an early step in the maturation process (5).
In the present work, it was shown that cysteine residues both from HypC and the precursor form from the large subunit are crucial for the interaction. Fig. 9 displays the function of the thiolates in liganding the [NiFe] cluster in the mature subunit (Fig. 9A) as delineated from the three-dimensional structure of the Desulfovibrio gigas enzyme (8) and compares it with the role postulated for these thiolates during the maturation process on the basis of the results reported (Fig. 9B). The proposed model implies that (i) Cys 241 is masked in pre-HycE by the interaction with HypC, (ii) Cys 244 coordinates the iron, (iii) Cys 531 coordinates the nickel, and (iv) Cys 534 provides a free thiolate attacking the nickel and iron atoms after proteolytic removal of the carboxyl terminus and thereby closes the bridge. Below we will discuss the evidence available that supports this model.
Masking of Cys 241 by HypC-When each of the four cysteine residues playing a role in the ligation of the [NiFe] cluster of the mature subunit was replaced by an alanine, the C241A variant was the only one that was no longer able to interact with HypC. It is assumed that the effect is specific, since an identical replacement of Cys 244 only 3 residues away allowed the interaction to take place. In addition, exchange of the well conserved Arg 239 residue by leucine also in close vicinity of Cys 241 did not prevent complex formation. 2 It is assumed that Cys 241 of pre-HycE directly or indirectly interacts with Cys 2 from the HypC protein. ture of the interaction is unknown at present and requires the purification of substantial amounts of the maturation intermediate. It could consist in the interaction of each thiol as hydrogen donor with a strong acceptor group from the other partner. Alternatives would be that the thiols of Cys 2 of HypC and Cys 241 of HycE could be sandwiched by some metal ion, like iron or zinc. However, the in vitro addition of chelating agents as EDTA or 2,2Јdipyridyl has no influence on the complex stability (data not shown). Interestingly, once the complex is formed it cannot be titrated by a surplus of a HypC variant in which the N-terminal cysteine was replaced by an alanine or serine residue. 2 If the interaction were noncovalent and assuming that the specificity of recognition is not only determined by the two cysteine residues, this should have been the case. Disulfide bonding as a possibility can be ruled out because the complex is readily dissociated by iodoacetic acid or iodoacetamide and because no migration difference of the complex under SDS-PAGE was observed under oxidative compared with reducing conditions. Coordination of the Iron by Cys 244 and of Nickel by Cys 531 -There are substantial number of arguments that the two metals are incorporated separately into the precursor of HycE. The most convincing ones are that the precursor formed in a strain grown in absence of nickel can be matured in vitro by the addition of nickel alone (11). This provides support for the contention that iron has already been incorporated and that nickel is inserted independently as a late step. Furthermore, the maturation endopeptidase scans the precursor for the presence of nickel by binding to the metal, which excludes the existence of the bridged cluster (14,27). Evidence that Cys 244 carries the iron in the precursor can only be deduced from the structure of the mature cluster (8). The direct interaction between Cys 541 and nickel, however, has been proven in the case of [NiFeSe]-hydrogenase from Desulfovibrio baculatus in which the cysteine is naturally substituted by a selenocysteine whose selenium atom has been shown to directly interact with nickel (28,29).
Cys 534 of Pre-HycE Contains a Free Thiol-An intriguing finding was also that Cys 534 of HycE is not essential for maturation. The C534A variant forms a complex with HypC, accepts nickel, and is proteolytically processed but does not yield an active enzyme. Incorporation of nickel and processing, although not interaction with HypC, was recently demonstrated for an analogous mutant in the large subunit of the NADreducing hydrogenase of Alcaligenes eutrophus (26). Cys 534 , which provides the thiolate for bridging the iron and the nickel in the mature cluster, does not interact with the metals during the maturation process.
Working Model-Based on these results, a working model on the sequence of events in the assembly of the [NiFe] center is proposed (Fig. 9). First, the thiolate of Cys 241 is masked by interaction with HypC. Since nickel can be incorporated into this complex, Cys 241 cannot be involved in the initial ligation of the metals of the center. Second, iron binding takes place at Cys 244 , and nickel is coordinated by Cys 531 .
The key intermediate for the assembly of metal center, therefore, is the complex between pre-HycE carrying the iron at Cys 244 and the chaperone-type protein HypC. It is assumed that the C terminus of HycE sticks out of the complex so that nickel can be added to Cys 531 . By an unknown reaction, the linkage between pre-HycE and HypC is then cleaved, rendering pre-HycE a substrate for the endopeptidase. Only nickel-containing pre-HycE from which HypC has been released can be cleaved by the endopeptidase. 2 This cleavage triggers a conformational switch feeding the C terminus toward the two metals and closing the bridge between them by reaction with the free thiol of Cys 534 . For a proof of this model, a number of questions have to be answered. The most relevant ones are the disclosure of the chemical nature of the interaction of pre-HycE with HypC and the determination of the reaction that leads to the release of HypC.