Proteolytic cleavage orchestrates cofactor insertion and protein assembly in [NiFe]-hydrogenase biosynthesis

Metalloenzymes catalyze complex and essential processes, such as photosynthesis, respiration, and nitrogen fixation. For example, bacteria and archaea use [NiFe]-hydrogenases to catalyze the uptake and release of molecular hydrogen (H2). [NiFe]-hydrogenases are redox enzymes composed of a large subunit that harbors a NiFe(CN)2CO metallo-center and a small subunit with three iron–sulfur clusters. The large subunit is synthesized with a C-terminal extension, cleaved off by a specific endopeptidase during maturation. The exact role of the C-terminal extension has remained elusive; however, cleavage takes place exclusively after assembly of the [NiFe]-cofactor and before large and small subunits form the catalytically active heterodimer. To unravel the functional role of the C-terminal extension, we used an enzymatic in vitro maturation assay that allows synthesizing functional [NiFe]-hydrogenase-2 of Escherichia coli from purified components. The maturation process included formation and insertion of the NiFe(CN)2CO cofactor into the large subunit, endoproteolytic cleavage of the C-terminal extension, and dimerization with the small subunit. Biochemical and spectroscopic analysis indicated that the C-terminal extension of the large subunit is essential for recognition by the maturation machinery. Only upon completion of cofactor insertion was removal of the C-terminal extension observed. Our results indicate that endoproteolytic cleavage is a central checkpoint in the maturation process. Here, cleavage temporally orchestrates cofactor insertion and protein assembly and ensures that only cofactor-containing protein can continue along the assembly line toward functional [NiFe]-hydrogenase.

More than one-third of all proteins in nature have been shown to bind metal cofactors (1). Metalloenzymes catalyze complex and essential processes, such as photosynthesis, respiration, and nitrogen fixation (2). Industrial interest stems from the excellent catalytic performance and specificity of certain metalloenzymes (e.g. in biotechnological and pharmaceutical applications) (3). Furthermore, metalloproteins receive growing attention due to the discovery of numerous diseases related to cofactor biosynthesis (4). Coordination of synthesis and assembly of multisubunit metalloproteins must be tightly controlled to ensure that only correctly folded and catalytically active enzymes are delivered to their subcellular destination. The physiological and molecular orchestration of events (e.g. coordination of protein and cofactor synthesis or assembly of cofactors into apoprotein) is not fully understood. Our comprehension of metal cofactor biosynthesis has been facilitated by the study of [NiFe]-hydrogenases, which served as excellent model systems (5,6). Due to their capability to catalyze the reversible reduction of protons to molecular hydrogen (H 2 ), hydrogenases have been discussed for potential bioenergy applications and industrial production of H 2 (7).
Typically, [NiFe]-hydrogenases are heterodimers of a large and a small subunit. The "large subunit" binds the catalytic NiFe(CN) 2 CO cofactor (8,9), and the "small subunit" contains three iron-sulfur (FeS) clusters involved in electron transfer. The nickel ion of the bimetallic active site is coordinated via four conserved cysteinyl thiolates, two of which additionally ligate an iron ion. The iron site carries one carbonyl (CO) 2 and two cyanide (CN) ligands. These unique ligands are proposed to maintain the low-spin Fe(II) state important for H 2 activation (10,11). Fig. 1 presents our current working model for hydrogenase maturation and suggests the following sequence of events: (i) synthesis of CO and CN, ligation to iron, and incorporation of the fully coordinated iron into the large subunit pro-protein by the coordinated activity of HypCDEF (12); (ii) nickel insertion by concerted activity of HypA, HypB, and SlyD (6, 13) (these accessory proteins are not required at high Ni 2ϩ concentrations) (5,14); (iii) nickel acting as recognition site for a specific endopeptidase (15); cleavage of the C-terminal extension of the pro-protein that harbors the [NiFe]-cofactor, inducing conformational changes; and (iv) complex formation with the small subunit to yield active [NiFe]-hydrogenase (16,17).
Escherichia coli synthesizes three distinct membrane-bound [NiFe]-hydrogenases: hydrogenase-1 (Hya), hydrogenase-2 (Hyb), and hydrogenase-3 (Hyc). The large subunit of each hydrogenases is synthesized as a pro-protein with a C-terminal extension, cleaved off by a specific endopeptidase. Besides the proteins mentioned above (HypABCDEF), orthologous accessory proteins have been described. For example, HybG is involved in delivery of the completed Fe(CN) 2 CO moiety to the large subunit of hydrogenase-2, whereas its homologue HypC delivers the cofactor to hydrogenase-3. Table 1 describes the proposed function of [NiFe]-hydrogenase maturation proteins from E. coli used in this study. Most enzymes and substrates involved in biosynthesis and assembly of the bimetallic cofactor have been identified. However, details about the coordination of protein synthesis and cofactor assembly are largely unknown.
To identify the sequence of events in active site biosynthesis and investigate the individual role of each accessory protein, an approach is required that allows reconstructing the maturation pathway. Here, we developed an enzymatic in vitro assay to study the synthesis of active [NiFe]-hydrogenase 2 from E. coli. The starting point is the unprocessed, large subunit pro-protein (HybC) and five individual [NiFe]-hydrogenase maturation proteins in purified form. The experiments focus on the question of when and why the C-terminal extension of HybC is cleaved off. Making use of biochemical analysis, circular dichroism, and infrared spectroscopy, we found that proteolytic cleavage coordinates cofactor insertion and protein assembly during the maturation of [NiFe]-hydrogenases.

The in vitro system to study [NiFe]-hydrogenase maturation
The anoxic reaction mixture was composed of purified HybC, the core maturation complex HybG-HypDE (GDE), purified accessory proteins His HypE and His HypF in stoichio-metric amounts, and an activation mix including ATP, carbamoyl phosphate, NiCl 2 , FeSO 4 , and sodium dithionite. The reaction was started by the addition of endopeptidase HybD in catalytic amounts. After 30 min of incubation time at room temperature, aliquots of the reaction mixture were analyzed by SDS-PAGE for cleavage of HybC (ϳ63.5 kDa) into the processed form, HybC Ϫ15AA (ϳ62 kDa) ( Fig. 2A)   Involved in acquisition and insertion of the nickel into large subunit HybC (53). HypB 31.5 GTPase activity; binds nickel and forms with SlyD the HypB-SlyD complex that activates nickel release and insertion into the large subunit (54,55). HybG (HypC) 8.7 Small iron-and CO 2 -binding protein; proposed to deliver CO 2 for reduction to CO by HypD (27). Forms with HypD the core complex of hydrogenase-2 maturation machinery (HybG-HypD), that is involved in delivering the completed Fe(CN) 2 CO moiety to the large subunit HybC (14). HypD 41.4 The only FeS cluster-containing Hyp protein; acts as a scaffold for assembly of the Fe(CN) 2 CO moiety of the cofactor prior to its transfer to the large subunit (28 due to protein degradations that appear as a smear on SDS-PAGE rather than separate bands (data not shown). Infrared spectroscopy showed the CO/CN vibrational fingerprint of the cofactor for in vitro (complete) and in vivo matured hydrogenase-2 (Fig. 2B). The presence of two CO bands suggested copopulation of two redox species (18,19). We tentatively assign these peaks to the "activated" Ni-C and fully reduced Ni-R state (1965 and 1947 cm Ϫ1 , respectively). Both Ni-C and Ni-R are reported to be involved in the catalytic cycle of [NiFe]-hydrogenase (7). In the absence of maturation proteins (GDE) or activation mix (AM), no formation of cofactor was observed ( Fig. 2B). However, upon closer inspection, traces of putative CO/CN bands at lower wave numbers were detected.

Analysis of hydrogenase-2 activity matured in vitro
Although HybC potentially harbors a [NiFe]-center, the large subunit alone does not exhibit hydrogenase activity. The presence of hydrogenase-2 small subunit HybO is required for enzyme activity (16). After maturation of HybC as described above, we therefore added a fraction of enriched HybO to HybC and analyzed for hydrogen conversion activity by ATR FTIR spectroscopy. Here, in vitro (complete) matured hydrogenase-2 was reduced by deuterium gas (D 2 ), and the release of deuterium ions into the H 2 O bulk was followed in real time (Fig. 3A). Recombination of D ϩ and OH Ϫ ions gave rise to intense bands at 2505 and 1445 cm Ϫ1 that can be assigned to semi-heavy water (O-D stretch and HDO bending, respectively). These bands are a highly specific marker for hydrogen uptake activity. See the legend to Fig. 3 for further details on the observed difference spectrum, which also comprises spectral features of film swelling. Fig. 3B shows comparable activity of hydrogenase-2 matured in vivo. In the absence of the GDE complex or AM, hydrogenase-2 was limited to residual D 2 oxidation activity. The D 2 -minus-N 2 difference spectrum of in vitro maturated hydrogenase-2 shows cofactor reduction upon D 2 uptake (Fig.  3B, inset). At least three reduced species in the fingerprint region of the cofactor were populated (CO bands 1951, 1936, and 1927 cm Ϫ1 ), and a similar red shift was observed in the CN region (2075 and 2063 cm Ϫ1 ). Future work will identify the involved redox species in greater detail. Additionally, the samples were analyzed for H 2 uptake activity by following hydrogenase-dependent reduction of benzyl viologen photometrically (Fig. 3C). The specific activity of purified in vivo maturated hydrogenase-2 was 1.88 Ϯ 0.11 mol of H 2 oxidized/min/mg. This value represents 100% activation and serves as an estimate of the maximum hydrogenase-2 activity that can be restored in vitro. Under the same reaction conditions, the highest activity obtained was in the presence of all components of the in vitro assay (Fig. 3C, complete). Without the addition of an equimolar ratio of His HypE and His HypF to the reaction mix, only ϳ70 and ϳ30% of enzyme activity could be restored, respectively. Only background activity that varied between an undetectable level and 6.8% of maximal activity was observed when GDE, HybD, HybO, or AM was omitted. The background activity is most probably due to nonspecific assembly of cofactor in trace amounts (compare Fig. 2B). HybO alone shows no hydrogenase activity. Furthermore, incubation performed under aerobic conditions completely abolished H 2 uptake activity.

The C-terminal peptide on HybC is essential for interaction with the hydrogenase maturation machinery
HybC is known to form a stable complex with the accessory protein HybG (20). HybG and HypD are the core complex of the hydrogenase-2 maturation machinery. With our in vitro system, we addressed the question of whether the 15-amino acid C-terminal extension on HybC is required for interaction with the HybG protein. Complex formation of Strep-tagged HybG with HybC, HybC Ϫ15AA , or HybC ϩ5AA was investigated. HybC Ϫ15AA is a genetic construct equal to truncated HybC (i.e. without the 15 amino acids at the C terminus). HybC ϩ5AA equals HybC but includes an artificial extension of five amino acids (Gly-Leu-Cys-Gly-Arg) at the C terminus (Fig. 4A).
The plasmid p-hybG was expressed together with either p-hybC or p-hybC Ϫ15AA plasmid in the strain FTD-G ( Table 2). The respective cell extracts were subjected to chromatography on a Strep-Tactin affinity column. The result shows that HybG interacted exclusively with HybC but not with the truncated variant HybC Ϫ15AA (Fig. 4B). Furthermore, we addressed the question of whether the C-terminal extension on HybC is important for recruitment of other proteins involved in maturation. To probe this, the plasmid pT-hybG Strep -hypDEF was expressed together with either HybC Ϫ15AA , HybC, or HybC ϩ5AA . Analysis of the isolated protein complexes after Strep-Tactin affinity chromatography revealed that only HybC interacted with the core complex HybG-HypD, whereas both HybC Ϫ15AA and HybC ϩ5AA did not interact (Fig. 4C, lanes 1-3). To determine whether HypE and/or HypF are required for HybG and HypD to form a complex with HybC, the experiment was repeated in a hypF or hypE deletion background. The plasmid pT-hybG-hypDE was expressed into hypF deletion strain FTD-F, whereas plasmid pT-hybG-hypDF was expressed in the hypE deletion strain FTD-E. The lack of HypF or HypE (Fig. 4C, lanes 4 -6 (⌬hypF) and lanes 7-9 (⌬hypE)) appeared to have no negative impact on the ability of HybC to interact with the HybG-HypD core complex (Fig. 4C, lanes 5 and 8). Our study also demonstrates that extension of the C terminus by five amino acids (HybC ϩ5AA ) prevents interaction with HybG-HybD (Fig. 4C, lanes 1, 4, and 7). Altogether, these results indicated that the C-terminal extension on HybC is important for interaction with HybG-HypD of the maturation machinery. HybG and HypD form the central maturation complex, with neither HypE nor HypF being essential for complex formation with HybC.  (60). B, matured hydrogenase-2 as isolated (in vivo, black) or as synthesized in vitro (in vitro complete, red) did not differ significantly in D 2 uptake activity. Residual activity was observed in the absence of the HybG-HypDE complex (in vitro w/o GDE, blue) or activation mixture (in vitro w/o AM, green). Note the D 2 -N 2 difference spectrum of in vitro synthesized hydrogenase-2 (inset). Here, the population of at least three reduced species is shown (positive CO frequencies at 1951, 1936, and 1927 cm Ϫ1 ). C, kinetic analysis of H 2 uptake activity. The complete assay was composed of purified HybC, GDE, HypE, HypF, the endopeptidase HybD, HybO, and AM. Reactions were conducted in the absence of one or more substrates at a time, as indicated. The samples were analyzed for activity by following H 2 -dependent reduction of benzyl viologen photometrically under the same reaction conditions. The specific activity of purified hydrogenase-2 (in vivo) was 1.88 mol of H 2 oxidized/min/mg. This activity represents 100% activation and provides an estimate of the maximum activity expected for the in vitro matured hydrogenase. The maximal attainable activity obtained under this condition was in the presence of all components of the in vitro assay (complete). Without the addition of an equimolar ratio of His HypE and His HypF to the reaction mix, ϳ70 or 30% of enzyme activity could be restored, respectively. Only background activity was observed when GDE, HybD, AM, or HybO was omitted or when the incubation was performed under aerobic conditions. HybO alone shows no hydrogenase activity. Four independent experiments and the S.D. value is shown.

Cleavage of the C-terminal peptide on HybC after cofactor assembly is essential for the formation of active hydrogenase
We investigated at which stage during maturation the C-terminal extension of HybC is cleaved off. The individual compositions of protein complexes of the in vitro assays were analyzed at different maturation steps by SDS-PAGE (Fig. 5A). Lane 1 shows the complete reaction mix but lacking HybD endopeptidase. The reaction was started by the addition of HybD, and immediately afterward, an aliquot was analyzed (lane 2, t 0 ). No HybC Ϫ15AA was detected. After t ϭ 10, 20, and 30 min of incubation at room temperature, HybC Ϫ15AA was observed in increasing amounts (lanes [3][4][5]. No additional processing of HybC after 20 min was observed. Both unprocessed Strep HybC and processed Strep HybC Ϫ15AA were isolated from the reaction mix by Strep-Tactin affinity chromatography (lane 6). These results demonstrate that the processed Strep HybC Ϫ15AA , but not the unprocessed HybC protein, forms a heterodimer with the small subunit HybO. Removal of the C-terminal extension appears to be a prerequisite for formation of catalytically active [NiFe]hydrogenase.
To determine whether the C-terminal extension of HybC is required for synthesis of active hydrogenase also in vivo, plasmids encoding either N-terminal His-tagged HybC or equally tagged HybC Ϫ15AA were introduced into a hybC deletion strain of E. coli (21). After anaerobic growth, recombinant HybC and HybC Ϫ15AA proteins were isolated by histidine affinity chromatography (Co 2ϩ -charged resin). The HybC preparation exhibited hydrogenase activity (Fig. 3, B and C; in vivo) and showed the (CN) 2 CO signature of a [NiFe]-center ( Fig. 2B; in vivo), but  (lanes 1-3). The experiment was repeated with extracts, in which either HypF (GDE ⌬hypF; lanes 4 -6) or HypE (GDF ⌬hypE; lanes 7-9) was lacking. Again, only HybC formed a complex with HypG Strep . HypF and HypE were found to be not essential for the interaction of HybG Strep and HybC (lanes 5 and 8).
the HybC Ϫ15AA preparation was inactive and devoid of cofactor (data not shown). These results support the aforementioned findings that truncated HybC (HybC Ϫ15AA ) is not able to recruit the hydrogenase maturation machinery and thus hinders maturation into active enzyme also in vivo.

Conformational changes of HybC upon removal of the C-terminal peptide
To detect possible conformational changes associated with cleavage of pro-protein, we purified and analyzed HybC, HybC Ϫ15AA , and HybC ϩ5AA by CD and native gradient gel electrophoresis. Far-UV CD spectroscopy (150 -250 nm) is sensitive to changes in secondary structure of proteins, whereas near-UV CD (250 -350 nm) monitors alterations in the environment of aromatic amino acid side chains (i.e. changes in the tertiary structure of proteins) (22)(23)(24). Notably, the C-terminal extension does not contain any aromatic amino acids. Spectra of HybC (red) and HybC ϩ5AA (black) were found to be almost identical (Fig. 6). Thus, the five-amino acid extension of the C terminus of HybC does not lead to structural changes in the protein. In contrast, the truncated HybC Ϫ15AA protein showed significant spectral differences in far-and near-UV CD spectroscopy (Fig. 6, A and B). Changes of far CD spectra are probably due to differences in the content of regular secondary structural features, such as the ␣-helix and ␤-sheet. Near-CD spectra indicate changes in the environment of aromatic amino acid side chains (e.g. changes in tertiary structure of the protein after removal of the C-terminal extension). Native PAGE (Fig.  6C, left) showed clear changes in migration behavior of HybC after removal of the C-terminal extension. The five-amino acid extension of HybC did not lead to changes in migration behavior, because both HybC ϩ5AA and HybC migrated faster and  showed broad, diffuse protein bands. In contrast, upon removal of C-terminal extension, the HybC Ϫ15AA migrated slower and showed a sharper protein band. SDS-PAGE was used to resolve and confirm the small differences in mass between HybC variants (Fig. 6C). In sum, these results indicate that conformational changes in secondary and tertiary structure of HybC take place upon removal of C-terminal extension.

Discussion
Maturation of [NiFe]-hydrogenases in vitro provides the possibility to prove and refine our understanding of [NiFe]-cofac-tor biosynthesis. The presented approach exclusively relies on purified components and allows studying the order of events individually. Our maturation assay included HypE and HypF, both of which have been shown to synthesize the CN ligands from carbamoyl phosphate, as supplied with the activation mixture (25,26). Although HypE and HypF are absolutely required for synthesis of the cyano-ligands, partial cleavage of HybC took place in their absence. This is due to the presence of HypE that was supplied to the reaction mix in the form of the GDE complex. The reason for cleavage in the absence of HypF is the presence of Fe(CN) 2 CO on the GDE complex that allows for partial assembly of [NiFe]-cofactor (14). Only the complete assay, including HypE and HypF, allows for multiple maturation turnover, as supported by formation of [NiFe]-cofactor and maturation of hydrogenase-2 in vitro at activity levels near those of the in vivo matured enzyme. Lenz and co-workers (61) recently identified the accessory protein HypX that synthesizes the CO ligand of O 2 -tolerant [NiFe]-hydrogenases from formyl-tetrahydrofolate. However, the anaerobic route of biosynthesis remains ambiguous because E. coli lacks HypX or orthologue enzymes (27).
An interesting aspect of performing in vitro reconstitution with purified proteins is whether a particular protein has a direct participation in the maturation process or not and which minimal combination of components is sufficient to achieve complete maturation. This capability provides a rapid means of "quality control" with regard to the reproducibility of the purification procedures for the individual enzymes by confirming the activity of the purified proteins individually. The functionality of the purified GDE complex, HypE, HypF, HybD, HybC, and HybO was confirmed by activity measurements of maturation assays complemented with the corresponding missing protein, as shown in Fig. 3C. FTIR analysis performed in previous and current studies indicates that anaerobically purified GDE complex carries two cyano-ligands, one CO ligand and bound CO 2 (14). Whereas HypD alone shows infrared bands assigned to one CO and two CN ligands (28), purified HybG exhibits a band characteristic of bound CO 2 (27). Metal analysis determined by inductively coupled plasma mass spectrometry (ICP-MS) showed that the GDE complex contained ϳ4.4 iron ions/mol of ternary complex, whereas HybG contained substoichiometric amounts of iron (ϳ0.3 mol of iron/mol of HybG). The amount of nickel, copper, cobalt, and zinc was below the detection limit (14). Employing the GDE complex that does not harbor the Fe(CN) 2 CO moiety could not restore hydrogenase activity. Therefore, the presence of the Fe(CN) 2 CO moiety seems to be necessary for hydrogenase-2 maturation in vitro. No hydrogenase activity could be measured when the GDE complex was replaced by HybG, HypD, or HypE individually or in combination after separate purification (14). This finding indicates that the physical ternary complex carrying the Fe(CN) 2 CO moiety is essential for successful activation of hydrogenase-2. Anerobically purified HypD protein preparations typically contain five equivalents of iron and four sulfides per protein molecule (28). Four of these iron ions are associated with a low-potential [4Fe-4S] cluster (29). Electron paramagnetic resonance (EPR) and Mossbauer spectroscopy on HypD revealed a diamagnetic [4Fe-4S] 2ϩ cluster (30). The fifth iron ion was proposed to coordinate the (CN) 2 CO ligands (28). HypD is the only Hyp [4Fe-4S] cluster containing protein and is therefore likely to be capable of performing redox chemistry and delivering electrons for CO and CN ligand generation (29). During the course of this study, HypD was found to catalyze the CO-dependent reduction of the electron acceptor methyl viologen with a maximal rate of 24 milliunits/mg of protein. However, the reverse reaction, CO 2 -dependent oxidation of reduced methyl viologen, performed in the presence of ATP could not be observed (data not shown). Analysis of anaerobically purified HypF revealed that the protein contained only substoichiometric amounts of iron and zinc (ϳ0.1 mol of iron and 0.4 mol of zinc per HypF monomer). The UV-visible spectrum did not indicate the presence of any cofactor. Similar results were obtained for the aerobically purified HypF (26). The carbamoyl phosphate phosphatase activity of HypF was routinely confirmed by the release of inorganic phosphate directly in non-denaturing polyacrylamide gels (31). The functionality of purified HypE and HypF was also routinely verified by observation of the cyanation on HypE from carbamoyl phosphate in an ATP-dependent condensation reaction using ATR FTIR (25). The addition of only HypF or HypE to the reconstitution reaction had no effect on the level of hydrogenase activity, but the addition of both (complete reaction) increased the hydrogenase activity about 2-fold. This suggests that an equimolar ratio of HypE and HypF is needed for full reconstitution of hydrogenase activity. Anerobically purified HybC precursor isolated from any hydrogenase deletion mutant of hypF, hypE, hypD, or hybD was unprocessed, was not associated with the small subunit HybO, and lacked enzyme activity. These HybC preparations were colorless and lacked cofactor or any associated metal, as indicated by UV-visible spectrum and ICP-MS analysis. In a Hyp-competent genetic background, the HybC precursor can be matured in vivo and in vitro and delivers active hydrogenase-2. In contrast, the isolated HybC ϩ5AA could not be matured, was brown, and had substoichiometric amounts of an oxygen-sensitive iron-sulfur cluster (0.2 mol of iron/mol of HybC ϩ5AA ). This cluster is coordinated by the conserved cysteinyl residues that normally ligate the NiFe(CN) 2 CO cofactor, as was shown by UV-visible spectroscopy and EPR (32).
Isolation of the small subunit HybO in its active form was more challenging. The pC-hybO plasmid that encodes for N-terminal His-tagged protein ( His HybO) was transformed in E. coli strain FTD-O, which lacks the genes encoding HybO and the large subunit of hydrogenase-2 ( Table 2). Overproduction and purification of His HybO was not possible due to inclusion body formation. A low level of soluble protein without isopropyl 1-thio-␤-D-galactopyranoside induction (ϳ0.70 mg of protein per 5 g wet weight of anaerobically grown cells) could be isolated by histidine affinity chromatography, but His HybO was labile due to the imidazole-containing elution buffer and could not restore hydrogenase activity in vitro. We therefore separated the soluble fraction containing His HybO by ultracentrifugation and subsequent ion-exchange chromatography. The His HybO fraction was brown and most likely present as holoprotein with three iron-sulfur clusters. It should be noted that the HybO protein concentration in hydrogenase deletion mutants was usually at a low level, particularly if the large subunit is absent (17). This can be due to rapid turnover of HybO. These results suggest that direct physical contact between the large and small subunit is required to ensure stability of the small subunit against degradation.
The results provide evidence that the C-terminal extension of the large subunit HybC is removed only after cofactor assembly and before complex formation with the small subunit HybO has occurred. Proteolytic cleavage is thus a checkpoint that guarantees formation of catalytically active [NiFe]-hydrogenase exclusively. The underlying mechanism was shown to be a larger conformational change associated with removal of the 15 C-terminal amino acids. We suggest that proteolytic cleavage of HybC triggers structural changes and induces formation of the [NiFe]-center from the Fe(CN) 2 CO moiety and nickel (33). It has been hypothesized that changes in conformation precede the internalization of the metal center within the large subunit (5). Crystal structures show that the [NiFe]-cofactor lies close to the interface with the small subunit (34).
The data presented in this study suggest that the role of the C-terminal extension is maintaining a "correct" conformation of the HybC protein that facilitates interaction with accessory proteins HybG-HypDEF. We could demonstrate that maturation proteins, particularly HybG, interact specifically with HybC but are incapable of interacting with truncated HybC Ϫ15AA . Most likely, the different conformation adopted by HybC Ϫ15AA prevents the interaction. During synthesis of hydrogenase-3, HypC (a homologue of HybG) forms a stable complex with the pro-protein of the large subunit of hydrogenase-3. This complex was found to accumulate in mutants unable to synthesize the NiFe(CN) 2 CO cofactor (35). Therefore, it is likely that complex formation between the unprocessed large subunit and the hydrogenase maturation machinery makes HybC accessible for the incoming cofactor (5). Because we observe formation of the HybC-HybO heterodimer exclusively with processed HybC, it seems that the C-terminal extension on HybC prevents complex formation with HybO. Upon removal of the C-terminal peptide, far-CD spectroscopy indicates rather insignificant conformational changes in secondary structure. In contrast, global changes in tertiary structure of the protein are suggested by the observed difference in the near-UV spectra between HybC and HybC Ϫ15AA . This difference is most likely due to the tertiary folding of the polypeptide chain, which can place the aromatic amino acid side chains in chiral environments, thus giving rise to circular dichroism (22,24). This is in good agreement with the proposed refolding of HybC upon cleavage of the C-terminal peptide (5). The fully folded and cofactor-containing HybC large subunit can then engage with the small subunit to form catalytically active heterodimer. This would ensure that in wildtype cells, only a large subunit with a complete [NiFe]-cofactor proceeds along the maturation path. A similar function was suggested for the C-terminal extension of hydrogenase-3 of E. coli and the membrane-bound [NiFe]-hydrogenase of Ralstonia eutropha (36,37).
Biosynthesis and insertion of [NiFe]-cofactor involves a multistep pathway catalyzed by different compositions of accessory proteins (20,38). Precursors of [NiFe]-hydrogenase are unsta-ble in aqueous solutions and highly sensitive to oxygen (39,40); thus, it seems likely that accessory and pro-protein form a supramolecular complex during maturation. A stable "supercomplex" consisting of HybG, HypD, HypE, and HybC was isolated in this work. Furthermore, multiprotein complexes containing HypA, HypB, SlyD, and pro-protein of hydrogenase-3 were observed earlier (49).
The C-terminal extension of the large subunit varies greatly both in sequence and length among different hydrogenases and organisms. Intriguingly, there are examples of hydrogenases whose large subunit lacks a C-terminal extension including H 2 -sensory and energy-converting [NiFe]-hydrogenases (41,42). During the course of this study, the large subunit of [NiFe]hydrogenase from Carboxydothermus hydrogenoformans (CooH) or Thermoanaerobacter tengcongensis (EchE), in which the C-terminal extension is missing (43,44), could not be synthesized in an active form in E. coli. In reciprocal heterologous complementation studies, anaerobically purified heterologous HybG-HypDE complex, where HypD was replaced by HypD either from C. hydrogenoformans or from T. tengcongensis, was not capable of assembling and maturing active hydrogenase-2 in E. coli (data not shown). How insertion of the [NiFe]-cofactor and subunit assembly are coordinated in these enzymes remains to be investigated. However, it is possible that the hydrogenase-specific maturation machinery in these organisms encodes for a peptide that functionally replaces the C-terminal extension. In support of this is the finding that the extension must not be covalently linked to the large subunit but can be part of a separate polypeptide (45).
In conclusion, this study highlights the role of the C-terminal extension on the hydrogenase catalytic subunit. The 15-amino acid peptide promotes the complete sequence of maturation events by directing and binding maturation proteins and ensuring that only the cofactor-containing large subunit can continue on the assembly line toward active [NiFe]-hydrogenase.

Bacterial strains and growth conditions
All E. coli strains and plasmids used are listed in Table 2. Strains were generally grown anaerobically in LB or in modified TB medium. The antibiotics kanamycin, chloramphenicol, and ampicillin were added to the medium at final concentrations of 50, 15, and 100 g/ml respectively. For overproduction of proteins, E. coli strains, BL21(DE) (46), FTD147 (21), or the indicated strains were transformed with the appropriate plasmid and grown anaerobically in modified TB medium at 37°C until the optical density at 600 nm (A 600 ) reached 0.3. Gene expression was induced by the addition of 0.2 g/ml anhydrotetracycline or 150 M isopropyl 1-thio-␤-D-galactopyranoside followed by incubation at 30°C for 3-5 h or after the culture reached an optical density at 600 nm between 0.7 and 0.9 (exponential phase cultures). Cells were harvested by centrifugation at 10,000 ϫ g for 15 min at 4°C. Large-scale protein purification was performed with 20 liters of anaerobic TB medium (3-4 g of cells/liter culture).

Strain and plasmid construction
Strains were constructed by the introduction of mutations from E. coli donor strains into recipient strains of MC4100 derivatives by P1kc-mediated phage transduction. Plasmid p-hybC served as DNA template for the introduction of mutation into the hybC gene using the QuikChange site-directed mutagenesis strategy of Stratagene. The oligonucleotide primers used to introduce a stop codon at position Val-553 of HybC were 5Ј-GCCTGTGCGGTACACTAAGTGGATGCTGACG-GC-3Ј and 5Ј-GCCGTCAGCATCCACTTAGTGTACCGCA-CAGGC-3Ј. To generate the pT-hybG-hypDF and pT-hybG-hypDE plasmids, the pT-hybG-hypDEF plasmid (14) was digested with BamHI or HindIII, which released either the complete hypE gene or the complete hypF gene, respectively. The p-hybG-hypDEF plasmid was used as a template for amplification of hypD, hypE, hypF, and hybG genes via PCR. The resulting PCR fragments were digested with NdeI and HindIII and ligated into NdeI-/HindIII-digested pET28A to generate the plasmid pT-hybG-hypDEF.

Preparation of crude extracts and affinity purification of protein
All steps were carried out under a nitrogen/hydrogen mixture (95:5) in an anaerobic chamber (Coy Laboratories) and at 4 ºC unless stated otherwise. Cells containing Strep-tagged protein were resuspended at a ratio of 1:3 (w/v) in buffer W (100 mM Tris/HCl and 150 mM NaCl, pH 8.0) including 2 mM sodium dithionite, 5 mg/liter DNase, and 0.2 mM PMSF. Cells were disrupted by sonication (40 watts for 16 min with 0.5-s pulses). Unbroken cells and debris were removed by centrifugation for 30 min at 10,000 ϫ g at 4°C to deliver the crude extract. Three-milliliter crude extracts were applied to a 1-ml Strep-Tactin-Sepharose column (IBA Technologies) using gravity flow. Unbound proteins were removed from the column by washing with 5 column volumes of buffer W. Specifically bound proteins were eluted with buffer W including 5 mM desthiobiotin. Desthiobiotin was subsequently removed by passage through a Hi-Prep 26/10 desalting column (GE Healthcare) equilibrated with buffer W without reducing agents and connected to an ÄKTA apparatus (GE Healthcare). Proteins were concentrated by centrifugation at 5000 ϫ g using centrifugal filters (Amicon Ultra, 50 K, Millipore) and stored under liquid nitrogen. His-tagged protein variants were purified from cells transformed with the appropriate plasmid derivative (Table 2). Wet cell paste was resuspended at a ratio of 1:4 (w/v) in buffer A (50 mM Tris/HCl, pH 8.0, 300 mM NaCl) including 5 mg/liter DNase and 0.2 mM PMSF. Crude extracts derived from ϳ10 g of wet cell paste were loaded onto a 1.5 ϫ 10-cm column of Co 2ϩcharged resin (Talon resin, Clontech). The column, which had been previously equilibrated with buffer A, was washed with 10 column volumes of the same buffer, followed by 5 column volumes of buffer A supplemented with 10 mM imidazole and then 3 column volumes of buffer A supplemented with 20 mM imidazole to remove nonspecifically bound proteins. His-tagged proteins were subsequently eluted with buffer A containing 300 mM imidazole. Imidazole was removed by a desalting column, and proteins were concentrated as described above.

Enrichment of an active small subunit HybO required for analysis of hydrogenase activity
The pC-hybO plasmid that encodes for N-terminal Histagged protein ( His HybO) was transformed in E. coli strain FTD-O (Table 2) and grown anaerobically in LB medium at 25°C without induction until the optical density at 600 nm reached 0.7. The cells were disrupted by three passages through a French pressure cell. His HybO purified by histidine affinity chromatography could not restore hydrogenase activity in vitro. We therefore separated soluble fractions containing His HybO by ultracentrifugation at 150,000 ϫ g for 2 h. Subsequently, the supernatant was loaded onto a Q-Sepharose HiLoad column, equilibrated with 100 mM Tris buffer, pH 8. The protein was eluted in a stepwise NaCl gradient. Enriched active holoprotein His HybO was recovered in the fractions eluted with 0.2 M NaCl.

The in vitro system for maturation of [NiFe]-hydrogenase-2 from E. coli
The 100-l complete reaction mixture in 50 mM MOPS KOH, pH 7.0, contained 4 M purified Strep-tagged HybC (39), 2 M purified GDE complex (14), 2 M His HypE, and 2 M His HypF (14). An activation mix with ATP, carbamoyl phosphate, NiCl 2 , FeSO 4 , and sodium-dithionite was added at final concentrations of 2.5 mM, 50 M, 100 M, 50 M, and 2 mM, respectively. The reaction mixtures were incubated at room temperature under nitrogen (95%) hydrogen (5%) in an anaerobic chamber. The reaction was started by the addition of a 1 M concentration of the endopeptidase His HybD. After a 30-min incubation, 20-l aliquots of the reaction mixture were analyzed for cleavage of HybC by SDS-PAGE using the ImageQuant software (Molecular Dynamics). Aliquots of reaction mix (80 l) were supplemented with 0.5 mg of the small subunit HybO-enriched fraction. After 15 min, the samples were analyzed for activity photometrically.

Isolation of in vitro matured hydrogenase-2
To isolate preparative amounts of active hydrogenase-2 that was matured in vitro, the reaction mix was scaled up to 5 ml. Matured Strep HybC was isolated from the reaction mix by Strep-Tactin affinity chromatography. Strep HybC was supplemented with the His HybO-enriched fraction. Active Strep HybC-His HybO heterocomplex was isolated by histidine affinity chromatography (Co 2ϩ -charged resin). Protein samples were concentrated and analyzed by ATR FTIR as reported earlier (47).

Isolation of in vivo matured hydrogenase-2
Hydrogenase-2 was isolated from FTD-147 strain, in which a fully active enzyme was restored by introducing genes coding for HybC together with maturation proteins HybG-HypDEF on an expression plasmids pASK-hybC and pT-hybG-hypDEF. The procedure for isolating an active enzyme was according to Ref. 48.

Determination of hydrogenase activity
H 2 uptake activity was determined by following the reduction of benzyl viologen at 578 nm using an Uvicon 900 dual-wavelength spectrophotometer. The 0.8-ml assays contained 4 mM benzyl viologen in 50 mM MOPS buffer, pH 7.0. Cuvettes were allowed to equilibrate with a 100% H 2 headspace. One unit of activity corresponds to oxidation of 1 mol of H 2 /min.

Non-heme iron and acid-labile sulfide determination
Iron and acid-labile sulfide of the GDE complex, HypD, and HybG were determined due to established protocols (51). Metal content of the GDE complex, HypD, HybG, HypE, HypF, and HybC variants was determined by ICP-MS, as described previously (32). For ICP-MS analysis, 0.1-0.2 mg of purified protein samples were analyzed for iron, nickel, zinc, cobalt, and copper.

Infrared spectroscopic characterization of HybC variants
All FTIR spectroscopy was conducted on a rapid-scan Tensor 27 spectrometer (Bruker Optik, Ettlingen, Germany) equipped with a three-reflection ZnSe/silicon crystal ATR cell (Smiths Detection, Wiesbaden Germany) as described earlier (27,28,47). The spectrometer was situated in an anaerobic gas chamber (Coy Laboratories, Grass Lake, MI) in a water-free atmosphere of typically 99% N 2 and 1% H 2 .

Further spectroscopic characterization of HybC
Protein concentrations were determined spectroscopically using an extinction coefficient of EM (280 nm) ϭ 97,310 M Ϫ1 cm Ϫ1 . Spectra were measured at a protein concentration of 1 mg/ml in 25 mM Tris-HCl buffer, pH 7.5, containing 150 mM NaCl using a Jasco J815 spectropolarimeter in the far-UV region (light path 0.1 mm, data accumulation 16ϫ) and near-UV region (light path 1 cm, data accumulation 32ϫ or 64ϫ), respectively. For all measurements, it was ensured that the proteins were stable under the given conditions.

Quality control and metal content
No hydrogen turnover activity is associated with any of the purified maturases: GDE complex, HypE, HypF, HybD, or HybO (14,16). Thus, catalytic competence and cofactor integrity of these enzymes had to be probed alternatively. This included the use of the maturation assay established in this study, UV-visible spectroscopy, FTIR spectroscopy, ICP-MS, and different biochemical activity assays. For example, the phosphatase activity of HypF was routinely confirmed by measuring the release of inorganic phosphate from carbamoyl phosphate (31). The specific interplay of HypE and HypF was probed by observing cyanation of HypE from inorganic phosphate in an ATP-dependent condensation reaction via surfaceenhanced IR absorption spectroscopy, as established earlier (25). Furthermore, the integrity of the Fe(CN) 2 CO cofactor as associated with HypD and the GDE complex was routinely probed by ATR FTIR (14,28). The presence of an iron ion and