NarJ Chaperone Binds on Two Distinct Sites of the Aponitrate Reductase of Escherichia coli to Coordinate Molybdenum Cofactor Insertion and Assembly*

Understanding when and how metal cofactor insertion occurs into a multisubunit metalloenzyme is of fundamental importance. Molybdenum cofactor insertion is a tightly controlled process that involves specific interactions between the proteins that promote cofactor delivery, enzyme-specific chaperones, and the apoenzyme. In the assembly pathway of the multisubunit molybdoenzyme, membrane-bound nitrate reductase A from Escherichia coli, a NarJ-assisted molybdenum cofactor (Moco) insertion step, must precede membrane anchoring of the apoenzyme. Here, we have shown that the NarJ chaperone interacts at two distinct binding sites of the apoenzyme, one interfering with its membrane anchoring and another one being involved in molybdenum cofactor insertion. The presence of the two NarJ-binding sites within NarG is required to ensure productive formation of active nitrate reductase. Our findings supported the view that enzyme-specific chaperones play a central role in the biogenesis of multisubunit molybdoenzymes by coordinating subunits assembly and molybdenum cofactor insertion.


Understanding when and how metal cofactor insertion occurs into a multisubunit metalloenzyme is of fundamental importance.
Molybdenum cofactor insertion is a tightly controlled process that involves specific interactions between the proteins that promote cofactor delivery, enzyme-specific chaperones, and the apoenzyme. In the assembly pathway of the multisubunit molybdoenzyme, membrane-bound nitrate reductase A from Escherichia coli, a NarJassisted molybdenum cofactor (Moco) insertion step, must precede membrane anchoring of the apoenzyme. Here, we have shown that the NarJ chaperone interacts at two distinct binding sites of the apoenzyme, one interfering with its membrane anchoring and another one being involved in molybdenum cofactor insertion. The presence of the two NarJ-binding sites within NarG is required to ensure productive formation of active nitrate reductase. Our findings supported the view that enzyme-specific chaperones play a central role in the biogenesis of multisubunit molybdoenzymes by coordinating subunits assembly and molybdenum cofactor insertion.
Molybdoenzymes are involved in numerous metabolic reactions in the carbon, nitrogen, and sulfur cycles and crucial for all forms of life (1). With the exception of nitrogenase, the active site of molybdoenzymes contains a molybdenum cofactor (Moco) 3 that has an ubiquitous basic structure composed of a molybdenum atom coordinated to one or two molecules of a tricyclic pyranopterin (2,3). The past few years have seen spectacular advances in our understanding of the molecular mechanisms of Moco biosynthesis, a highly conserved biosynthetic pathway (4 -8). In contrast, information concerning biogenesis of molybdoenzymes is scarce. Molybdoenzyme biogenesis, the process that ensures productive formation of active molybdoenzymes, generally involves both metal cofactor insertion and multisubunit assembly. In prokaryotes, the Moco insertion process is a cytoplasmic post-translational event (9) often assisted by enzyme-specific chaperones (10 -13).
Dissimilatory nitrate reductase A from Escherichia coli (NarGHI) is one of the best studied multisubunit molybdoenzymes (14) and can be considered as a model system for studying the biogenesis process in prokaryotic enzymes. NarGHI is a heterotrimeric enzyme comprising a Moco and an iron-sulfur-containing catalytic subunit (NarG, 139 kDa), an iron-sulfur-containing subunit (NarH, 58 kDa) and a quinol-oxidizing membrane-bound heme b subunit (NarI, 26 kDa) (14,15). NarGH is located in the cytoplasm, anchored to the cytoplasmic membrane by NarI. When liberated from the membrane, the NarGH complex retains its activity using artificial electron donors such as benzyl viologen. Finally, the enzyme-specific chaperone NarJ plays an essential role for nitrate reductase A activity, facilitating Moco insertion into NarG (11).
As observed for other known molybdoenzymes (16 -19), the crystal structure of the NarGHI complex (20,21) reveals that Moco is an extended molecule deeply buried into the enzyme complex at the NarG-H subunit interface. This observation suggests that the subunit assembly and the Moco incorporation must be tightly coordinated. Such coordination has been shown using the membrane-bound nitrate reductase from Thermus thermophilus. Indeed, the NarJ-assisted Moco insertion step can only occur once the apoenzyme complex is attached to the cytoplasmic membrane (22).
In this work, we have provided new insights about molybdoenzyme biogenesis in E. coli. We have revealed a novel function for the enzymespecific chaperone NarJ that goes beyond its reported implication in the Moco incorporation process. Our results demonstrated that NarJ interacts at two distinct binding sites on the NarG subunit within the apoenzyme complex, one interfering with membrane anchoring and another being involved in Moco insertion. Overall, the enzyme-specific NarJ protein coordinates assembly and molybdenum cofactor acquisition of the heterotrimeric enzyme during the biogenesis process.

MATERIALS AND METHODS
Bacterial Strains and Growth Conditions-The E. coli strains and plasmids used in this work are described in Table 1. BL21(DE3) was employed for protein overexpression using pET22b-type expression plasmids (Novagen). BTH101 was used as recipient strain for two-hybrid assays. In-frame deletion of narJ was performed according to the Hamilton procedure (23). The mob,narJ double mutant strain was obtained by P1 transduction as described before (15). Bacterial cultures were grown aerobically in L-broth medium at 37°C. For biochemical studies, strains were grown anaerobically in L-broth medium supplemented with glucose (0.3%) and nitrate (0.2%). When required, the appropriate antibiotics and 0.2 mM isopropyl-1-thio-b-D-galactopyranoside were added.
Plasmids-Deletions in narG gene were performed by PCR using pT18-NarG, pT25-NarG, or pNarGHJI plasmids as templates. Plasmids marked with the (⌬1-40) designation produce a truncated NarG protein for the first 40 amino acids. Conversely, pT18-NarG-(1-40) plasmid expresses the first 40 amino acids of NarG fused to the N terminus of the T18 domain of adenylate cyclase.
Bacterial Two-hybrid Assay-Protein interactions were detected using a bacterial two-hybrid approach as described previously (24,25). Protein interactions were estimated by measurement of ␤-galactosidase activity in cells grown to mid-log phase at 30°C in L-broth medium supplemented with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside.
Cell Fractionation-Cells were cultured anaerobically as indicated above, harvested at late exponential phase, and washed in 50 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 buffer (buffer A). The cells were used immediately or stored as pellets at Ϫ20°C until required. Bacterial cells were broken by passage through a French press. The lysate was centrifuged for 10 min at 20,000 ϫ g to remove any unbroken cells. The crude extract obtained (CE) was diluted 5-fold with buffer A and centrifuged at 250,000 ϫ g for 90 min. The cell supernatant and the membrane fraction were again centrifuged for 90 min at 250,000 ϫ g, yielding the soluble (S) and membrane (M) fractions.
Immunological Quantitation-The amount of nitrate reductase antigen present in cell fractions was quantified by performing rocket immunoelectrophoresis analysis using a polyclonal antiserum specific for nitrate reductase A as described previously (26).
Activation of Nitrate Reductase in Crude Cell Extracts-The MobAdependent activation of nitrate reductase was based on that described previously (10). The specified quantities of purified MobA-His 6 protein were added to 100 l of crude extract; the volume of the reaction medium was adjusted to 150 l with buffer A. The NarJ-dependent activation experiment was performed using the same procedure by adding specified quantities of purified NarJ-His 6 (11). In this work, in vitro activation of nitrate reductase precursors present in crude extracts was only considered to avoid a time-dependent inactivation of the sample occurring during cell fractionation. The assay mixtures were incubated under strictly anaerobic conditions in a glove box (95% N 2 , 5% H 2 ) at 37°C for 90 min and placed on ice to stop the reaction. Aliquots of the reaction mixture were assayed for nitrate reductase activity.
Enzyme Assays-Nitrate reductase activity in cell fractions was measured at 37°C by following the oxidation of reduced benzyl viologen spectrophotometrically at 600 nm coupled to the reduction of potassium nitrate (27). Nitrate reductase specific activity is expressed in mol of nitrite produced min Ϫ1 mg Ϫ1 of nitrate reductase as detected by rocket immunoelectrophoresis.
Overexpression and Purification of His 6 -tagged Proteins-MobA-His 6 was purified as described (28). Overexpression of NarJ-His 6 was performed as described previously (11). Soluble fractions were prepared as described above using 20 mM Tris-HCl, pH 7.6, 0.5 M NaCl, 5 mM imidazole buffer (binding buffer) and applied to a 4-ml Ni 2ϩ affinity column (Qiagen) equilibrated with the binding buffer. After washing the column, the NarJ-His 6 protein was eluted using binding buffer supplemented with 200 mM imidazole. The pooled fractions were immedi-FIGURE 1. NarJ influences apoNarGH localization. Nitrate reductase distribution after cell fractionation of various strains was evaluated by rocket immunoelectrophoresis and expressed as a percentage of the sum of immunoprecipitated nitrate reductase present in the soluble and membrane fractions. The black bars correspond to the membrane fraction, and the gray bars correspond to the soluble fraction. Holonitrate reductase produced by MC4100 strain (WT) was used as a positive control, whereas localization of apoenzyme produced either from narJ, mob,narJ strains or from mob strains was investigated. The mob strain was further transformed with a plasmid allowing expression of the NarJ protein (mob/pNarJ). Results are mean values from at least three independent experiments. It is important to notice that the amount of nitrate reductase present in each mutant strain (narJ, mob,narJ, or mob) as detected by rocket immunoelectrophoresis on crude extracts is similar to that detected in the wild-type strain (data not shown). ately dialyzed against 20 mM Tris-HCl, pH 7.6, 0.1 M NaCl buffer, frozen in liquid nitrogen, and stored at Ϫ80°C until used. The whole procedure was carried out at 4°C. Interaction Study by Biosensor Experiments-The surface plasmon resonance (BIAcore apparatus) was used to analyze the interaction between NarJ and NarG variants. All experiments were carried out at 25°C. NarJ-His 6 was purified as indicated above and immobilized on a sensor chip CM5 (BIAcore) through amine coupling as described previously (11). Crude extracts (70 ml at 50 mg of total protein/ml) containing NarG variants were prepared in HBS buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3.4 mM EDTA, 0.005% surfactant P20) and injected over the test and control (no protein immobilized) surfaces at a flow rate of 10 ml/min. The sensor surface was regenerated with a 1-min injection of 1 mM NaOH.

NarJ Maintains the Aponitrate Reductase in a Soluble State before
Moco Insertion-Enzyme-specific chaperones interact with the apoenzyme early during the maturation process. The apoenzyme acquires thereby a conformation suitable for subsequent incorporation of the metal cofactor (29). In the same way, NarJ attachment to the aponitrate reductase is a prerequisite for Moco incorporation (11,30). In the case of the nitrate reductase from T. thermophilus, it has been demonstrated that NarJ-assisted Moco incorporation occurs within the membranebound apoenzyme complex. Further, NarJ is required for membrane attachment of the thermophilic apoenzyme complex (22). Thus, we investigated whether the cellular distribution of the apoenzyme from E. coli is influenced by its interaction with NarJ ( Fig. 1). Initially, the absence of NarJ did not appear to affect the cellular distribution of the apoenzyme; the apoenzyme produced by a narJ strain was mainly associated with the membrane fraction as in the wild-type parent (Fig. 1). However, in a mob mutant in which Moco biosynthesis is arrested, chromosomal or plasmidic expression of NarJ increased significantly the amount of soluble enzyme. Although only 15% of soluble apoenzyme was found in the mob,narJ strain, nearly 40 and 80% were present in the mob and in mob/pNarJ strains, respectively. Conversely, in a Moco-sufficient strain (WT), overexpression of NarJ did not affect the localization of the holonitrate reductase. Thus, before Moco insertion, the NarJ chaperone maintains the apoenzyme in a soluble state.

NarJ-assisted Moco Insertion
Step Precedes Membrane Anchoring of the Apoenzyme-We next asked whether the NarJ-driven modification of apoenzyme distribution influences the NarJ-assisted Moco incorporation process. To address this point, we followed Moco insertion using in vitro activation assays on crude extracts from strains displaying either mostly soluble or membrane-bound aponitrate reductases (Fig. 2). Initially, in vitro MobA-dependent activation assays were performed using mob or mob/pNarJ strains expressing 40 and 80% of soluble apoenzyme, respectively. Interestingly, the highest activity value was obtained with the strain mob/pNarJ that expressed a mainly soluble apoenzyme ( Fig.  2A). In a different set of experiments, in vitro NarJ-dependent activation assays were performed using a nar strain transformed with a plasmid expressing either the soluble NarGH complex or the membrane-associated NarGHI complex (Fig. 2B). Here again, the highest level of activation was observed with the nar/pNarGH extract displaying 80% of soluble apoenzyme in contrast to the nar/pNarGHI displaying 20% of soluble enzyme. In both approaches, the higher level of soluble apoenzyme in the crude extracts resulted in a higher level of activation. Further, as seen in Fig. 2, the highest activity values were obtained with extracts in which the apoenzyme had interacted in vivo with NarJ (mob strain) as compared with the cases in which NarJ has been added in vitro (nar derivatives). This is in full agreement with the reported chaperone function of NarJ i.e. apoenzyme stabilization (31).
Since maintaining the apoenzyme in a soluble state improved the yield of activation, we next asked whether, once solubilized, the membrane-bound apoenzyme can be activated in vitro. Solubilization by Triton X-100 did not improve the in vitro activation level of the mob crude extract in which 60% of the apoenzyme is membrane-bound (data not shown). As a control, Triton X-100 addition to a mob/pNarJ extract in which the apoenzyme is mainly soluble did not inhibit the level of activation. These results indicated that, despite solubilization, the membrane-bound apoenzyme can no longer incorporate Moco.
Taken together, these results clearly indicated that (i) the apoenzyme maintained in a soluble state by NarJ in a mob strain is in a competent conformation for Moco insertion and (ii) once bound to the NarI subunit, the apoNarGH complex has acquired a definitive conformation that is no longer compatible with a NarJ-assisted Moco insertion proc-FIGURE 2. The level of in vitro activation of the aponitrate reductase is directly dependent on its cellular distribution. A, increased solubility of the apoenzyme by NarJ in a mob strain results in higher level of activation. MobA-dependent activation experiments were performed using crude cell extracts issued from mob or mob/pNarJ strains. BV:Nitrate, benzyl viologen:nitrate. B, premature membrane anchoring of the apoenzyme reduces the level of NarJ-dependent activation. Experiments were performed using crude cell extracts issued from a nar strain expressing the NarGH (nar/ pNarGH) or the NarGHI (nar/pNarGHI) subunits. Benzyl viologen:nitrate reductase activities are given as mol of nitrite formed min Ϫ1 mg Ϫ1 of nitrate reductase present in the crude extracts as detected by rocket immunoelectrophoresis and were measured as described under "Materials and Methods." The curves are exponentials fitted to experimental data.
ess. Moco incorporation is apparently a cytoplasmic event that must take place at a particular stage of nitrate reductase biogenesis before membrane attachment of the apoenzyme.
Evidence for Two NarJ-binding Sites on the NarG Precursor-We showed that NarJ maintains the apoNarGH complex in a soluble state. One explanation could be that NarJ interferes with membrane anchoring of the apoenzyme. Previous studies from this laboratory indicated that NarJ specifically interacts with the catalytic subunit NarG within the apoenzyme complex (11). The x-ray structure of the NarGHI complex reveals that both NarG and NarH subunits associate with NarI, the membrane-anchor subunit (Fig. 3) (20). A major interaction with NarI centers on the N-terminal tail of NarG. Such an extended domain of NarG would constitute a good candidate for NarJ binding in terms of membrane-anchoring interference. To verify this hypothesis, the region extending between Met-1 and Gln-40 (Fig. 3) was fused to the N terminus of the T18 domain of the adenylate cyclase and tested for NarJ interaction in the two-hybrid system. Remarkably, the fusion strongly interacts with NarJ (Fig. 4A). As reported before, the NarJ-NarG protein interaction can also be followed using BIAcore (11). NarJ protein was thus immobilized and tested for its interaction with the NarG-(1-40)-T18 fusion protein. As compared with the sensorgram obtained with a nar extract (nar/pT18-Zip), NarJ appears to specifically interact with the N-terminal tail of NarG (Fig. 4B). As a positive control, an extract containing the full-length NarG fused to the T18 domain showed specific interaction with NarJ.
In a complementary set of experiments, we assessed the ability of the truncated NarG variant to interact with NarJ. Unexpectedly, deletion of the first 40 residues of NarG does not hamper its interaction with NarJ, as seen using a bacterial two-hybrid assay (Fig. 4A). Moreover, the truncated protein still interacts with NarH (Fig. 4A) or with the Moco biosynthetic proteins that promote Moco delivery (data not shown), suggesting a structural integrity of the protein. BIAcore analysis confirms that the NarG variant specifically interacts with NarJ (Fig. 4C). Taken together, these data indicated the presence of at least two NarJ-binding sites within NarG.
Further, the ability of the truncated NarG protein to yield an active and membrane-bound complex was assessed. Although it completely lacked any physiological quinol:nitrate oxidoreductase activity (data not shown), the truncated version of NarG exhibited benzyl viologen:nitrate  . Evidence for two NarJ-binding sites within NarG. A, interactions between NarG variants and NarJ or NarH were probed in a wild-type BTH101 strain using a twohybrid approach. The indicated values represent ␤-galactosidase activities expressed in Miller units. Negative controls using the Zip fusion proteins have been performed and have systematically given 50 Miller units. The data are the average of at least three independent experiments that vary by no more than 10% from the mean. B and C, qualitative analysis of the interaction between NarJ-His 6 and NarG variants using a BIAcore approach. Sensorgrams of interaction between immobilized NarJ-His 6 and the NarG variants are expressed in resonance units. The control flow cell (no protein immobilized) was subtracted from the experimental flow cell, allowing direct visualization of the specific binding. Crude extracts issued from the nar strain transformed with the following plasmids were injected over both control and test flow cells. In B, T18 derivative plasmids were employed: pT18-Zip, pT18-NarG, and pT18-NarG- . In C, T25 derivative plasmids were employed: pT25-Zip, pT25-NarG, and pT25-NarG-(⌬1-40).
oxidoreductase activity (Table 2), indicating the presence of Moco in the enzyme complex and maintenance of its structural integrity (32). The finding that the enzyme is active with an artificial electron donor (benzyl viologen), but not with the physiological electron donor (quinol), suggested a disturbed membrane anchoring to NarI resulting in the loss of electron transfer throughout the NarGHI complex. Further, localization analysis showed that membrane anchoring is affected (55% as compared with 97% for the wild-type active enzyme) ( Table 2). Such an observation is not surprising considering that the N-terminal tail of NarG is involved in NarI subunit interaction together with NarH (20). Overall, these results demonstrated that the NarJ-binding site located within the truncated protein is essential for Moco incorporation.
NarJ Controls the Quality of the Enzyme Addressed to the Membrane via Binding to the N-terminal Tail of NarG-According to our hypothesis, NarJ binding to the N-terminal tail of NarG interferes with membrane anchoring of the apoenzyme complex. Deletion of this fragment should thus enhance anchoring of the apoenzyme prior to the Moco insertion step. This should be visualized by a low specific enzymatic activity of the N-terminally truncated NarG variant present in the membrane fractions. Indeed, the benzyl viologen:nitrate oxidoreductase specific activity (expressed in mol of nitrite produced, min Ϫ1 mg Ϫ1 nitrate reductase) is an intrinsic parameter of the enzyme that assesses the presence of Moco and enzyme integrity. In a wild-type context (nar/ pNarGHJI strain), no significant differences in terms of specific activity were observed between the enzyme present in the soluble or the membrane fractions (70 -100 units) (Table 2). Interestingly, the specific activity of the truncated enzyme located into the membrane fraction was much lower (5-fold less) than the one measured for the soluble counterpart. This indicated a low proportion of Moco-containing and active enzymes in the membrane fractions. These data suggested that an uncontrolled membrane anchoring of both the apo-and holoenzyme complexes occurs in the absence of the N-terminal tail of NarG.
In a different but complementary set of experiments, we determined whether the N-terminal tail of NarG provided in trans in a wild-type strain could titrate in vivo NarJ from the nitrate reductase biogenesis process ( Table 2). A plasmid construction allowing expression of a fusion protein between the first 40 amino acids of NarG and the T18 domain of the adenylate cyclase was employed to transform the wildtype strain. Interestingly, the specific enzymatic activity measured in the crude extract from the WT/pT18-NarG-(1-40) strain is much lower than that prepared from the wild-type strain transformed with the vector control (20 versus 94 units) ( Table 2). Thus, the N-terminal tail of NarG provided in trans negatively influences the level of maturation of the nitrate reductase. Cell fractionation was performed using both strains and showed that the enzyme distribution was not affected by the presence of the NarG fragment. Further, in the presence of the N-terminal tail provided in trans, the specific activity of the membranebound enzyme was reduced as compared with the one measured for the soluble counterpart (22 versus 75 units) ( Table 2). Overall, these data indicated that deletion of the N-terminal fragment of NarG or providing it in trans results in a drastic reduction of the quality of the nitrate reductase complex attached to the cytoplasmic membrane as evaluated by the enzyme specific activity.

DISCUSSION
Molybdenum cofactor acquisition has become an important area of study given the wide range of functions carried out by molybdoenzymes in all domains of life. Moco is systematically found deeply buried into the enzyme complexes, suggesting that the subunit assembly and the cofactor incorporation must be coordinated. The Moco insertion process involves several accessory proteins (30), among which the enzyme-specific chaper-  ones facilitate the process by the initial formation of a complex with the apoenzyme (10,11,13). The work presented here has answered two fundamental questions on molybdoenzyme biogenesis. (i) Considering the multisubunit character of numerous molybdoenzymes, at which particular stage of assembly does the Moco insertion event occur? (ii) How could such an event be controlled? Here, using the heterotrimeric nitrate reductase A from E. coli as a model, we demonstrated that the enzyme-specific chaperone NarJ binds two distinct sites on the apoenzyme to coordinate both molybdenum cofactor insertion and multisubunit assembly.
Our results support a model depicted in Fig. 5 for how sequential interaction between NarJ and the apoenzyme promotes assembly and Moco incorporation within a multisubunit molybdoenzyme. At first, both the NarG and the NarH subunits associate, forming the apoNarGH complex. At this stage, we showed that the complex may interact with the membrane-anchor NarI subunit, leading to the formation of an inactive NarGHI complex that can no longer incorporate the cofactor. Such an assertion is supported by examination of the crystal structure of the membrane-bound apoenzyme, produced in a mob strain, which shows a closed conformation identical to the native enzyme (20,33). Alternatively, the apoNarGH complex may enter the maturation pathway by the initial interaction with the NarJ protein. At this stage, NarJ determines the fate of the apoenzyme complex by shifting the equilibrium toward the Moco insertion pathway. In addition to its protective role on the apoenzyme, we showed (i) that NarJ binding to the N-terminal tail of NarG prevents premature membrane anchoring of the apoNarGH complex and (ii) that NarJ binding to another site of NarG allows the interaction with the Moco delivery machinery constituted by several proteins involved in the final stages of Moco biosynthesis (30). Accordingly, we showed that NarJ titration using the N-terminal tail of NarG not only reduces the level of Moco-containing enzyme but also redirects some of the apoNarGH complex toward membrane anchoring prior to Moco insertion. Once NarG has folded consecutively to cofactor insertion, the NarJ chaperone and the Moco biosynthetic proteins are dislodged by an unknown mechanism, and the active holoNarGH complex is then free to interact with NarI via the N-terminal tail of NarG and NarH. To summarize, the fate of the nitrate reductase apoprotein complex in the cytoplasm can be determined by the equilibrium of three factors: the folding kinetic of the apoenzyme, the rate of membrane anchoring, and the efficiency of the Moco incorporation process. Our results were consistent with NarJ controlling each of these factors, an action that goes beyond its restricted chaperone function.
Likely, another NarJ-binding site exists in the N-terminally truncated version of NarG, in which NarJ facilitates Moco insertion. The exact location of this second binding site remains unclear. Considering that the association and dissociation criteria for NarJ binding onto either site of NarG are the same, i.e. binding on the apoNarGH precursor and release upon Moco incorporation, one can envision that both NarJbinding sites are structurally connected. Such coupling may enhance the co-association and dissociation steps and minimize the possibility that NarJ remains at the N-terminal tail of NarG, preventing membrane anchoring of the holoNarGH complex. Alternatively, NarJ can be released from the N-terminal tail of NarG with an intrinsic time constant, providing sufficient time for Moco incorporation to be completed before membrane anchoring.
Recently, studies on the membrane-bound nitrate reductase (NarCGHI) from T. thermophilus revealed that multisubunit assembly and Moco incorporation are coupled (22). However, a different biogenesis pathway has been delineated as compared with the E. coli enzyme. Indeed, the NarJassisted Moco incorporation step strictly requires prior membrane anchoring of the apoNarGH complex to the NarCI anchor subunits (22). Further, sequence analysis of the NarG subunit from T. thermophilus revealed the absence of the N-terminal tail targeted by NarJ in the E. coli NarG protein.
Such an observation could explain the different biogenesis scheme depicted for the thermophilic enzyme complex.
The underlying concept of this work is that coordination of multisubunit assembly and metal cofactor insertion is fundamental for productive synthesis of a multisubunit metalloprotein. Our results bring interesting comparisons with other systems such as the [NiFe] hydrogenases (34). The fully mature and active hydrogenase is a heterodimer made up of a large subunit that carries the [NiFe] center and an electron-transfer small subunit (35). A key intermediate during maturation is a complex between the precursor form of the large subunit and a chaperone protein (36,37). After maturation of the large subunit, the final step of hydrogenase maturation is the formation of the heterodimer (38). In the case of periplasmic hydrogenases, it clearly appears that the existence of the heterodimer is a prerequisite for translocation as only the small subunit harbors the twin arginine-containing signal peptide that directs the complex to the Tat translocon (39,40).