HypF, a Carbamoyl Phosphate-converting Enzyme Involved in [NiFe] Hydrogenase Maturation*

HypF has been characterized as an auxiliary protein whose function is required for the synthesis of active [NiFe] hydrogenases in Escherichia coli and other bacteria. To approach the functional analysis, in particular the involvement in CO/CN ligand synthesis, HypF was purified from an overproducing strain to apparent homogeneity. The purified protein behaves as a monomer on size exclusion chromatography, and it is devoid of nickel or other cofactors. As indicated by the existence of a sequence motif also present in several O-carbamoyltransferases, HypF interacts with carbamoyl phosphate as a substrate and releases inorganic phosphate. In addition, HypF also possesses ATP cleavage activity that gives rise to AMP and pyrophosphate as products and that is dependent on the presence of carbamoyl phosphate. This and the fact that HypF catalyzes a carbamoyl phosphate-dependent pyrophosphate ATP exchange reaction suggest that the protein catalyzes activation of carbamoyl phosphate. Extensive mutagenesis of the putative functional motifs deduced from the derived amino acid sequence showed a full correlation of the resulting variants between their activity in hydrogenase maturation and the in vitro reactivity with carbamoyl phosphate. The results are discussed in terms of the involvement of HypF in the conversion of carbamoyl phosphate to the CN ligand.

Hydrogen metabolism in enterobacteria involves the activity of the products of three functional classes of genes that code for structural proteins, regulatory proteins, or for proteins involved in metal center biosynthesis and enzyme maturation. In Escherichia coli, the structural genes are organized in four operons, responsible for the formation of hydrogenase 1 (hya operon), hydrogenase 2 (hyb operon), hydrogenase 3 (hyc operon), and the putative hydrogenase 4 (hyf operon). Each operon contains the genes for the large and small hydrogenase subunit, for redox carriers, membrane anchor proteins plus components required for the maturation of the hydrogenase encoded by that specific operon, like the endopeptidase involved in proteolytic processing of the large subunit (for review see Refs. 1 and 2). Because the hydrogenases in E. coli serve different physiological functions, the expression of these oper-ons is differentially regulated. Hydrogenase 3, for example, which is a component of the formate hydrogen lyase system, is synthesized under fermentative conditions. Its formation requires the activity of the transcriptional activator FhlA and formate as inducer (3).
Considerable efforts have been directed to understand the incorporation of the [NiFe] metal center into the large hydrogenase subunit. A scenario is emerging indicating that iron and nickel insertion proceeds separately, whereby the incorporation of iron together with its CO and CN ligands precedes that of nickel (1,4,5). The function of the HypA and HypB proteins has been related to nickel insertion because hypA and hypB mutations can be phenotypically complemented by inclusion of high nickel concentrations in the medium (6,7) and because the HypB protein binds nickel (8). HypC, which is a small acidic protein, forms a complex with the precursor of the large hydrogenase subunit and, accordingly, has been postulated to play the function of a specific chaperone, maintaining a folding state to render the large subunit amenable for metal acquisition (9,10). The most detailed information for all maturation proteins is available for the endopeptidase which cleaves the C-terminal extension from the precursor of the large subunit once nickel has been inserted (5,11,12). It is thought that the endopeptidase controls the fidelity of nickel insertion (13) and also, by cleavage, induces the conformational shift required to thread the finished metal center into the interior of the large subunit (1,5,14).
An intriguing issue of [NiFe] center synthesis concerns the source and the pathway of biosynthesis of the CO and CN ligands. Circumstantial evidence indicates that the HypD, HypF, and HypE proteins are involved in this process. HypD is an Fe/S protein with unusual spectroscopy properties, 1 and HypF and HypE display sequence signatures that implicate functions in organic synthesis. HypF, for example, has been reported to carry a sequence motif characteristic of acyl phosphatases (15), and HypE shares motifs that are also found in aminoimidazole ribonucleotide synthetase and thiamin-phosphate kinase (16,17).
An important discovery in this connection was the finding that maturation of [NiFe] hydrogenases requires the availability of carbamoyl phosphate (CP). 2 A mutant of E. coli lacking CP synthetase (carAB genes) activity was unable to synthesize active hydrogenases 1-3. The deficiency was shown to be caused by a blockade of the maturation process (18). Citrulline in the medium was able to complement the lesion of the carAB mutant indicating that the carbamoyl moiety was crucial. It was postulated that CP is the precursor of either CO or CN or of both of them (18). In the present communication we show that the HypF protein recognizes CP as a substrate. A detailed mutational analysis of conserved sequence traits was performed, and it is shown that abolition of HypF function in the maturation process is fully correlated with its ability to interact with CP as a substrate. It is further shown that HypF cleaves ATP into AMP and pyrophosphate in the presence of CP.
EXPERIMENTAL PROCEDURES E. coli Strains, Plasmids, and Growth Conditions-Strain MC4100 (19) from E. coli was used as wild type, and strains JM109 (20) and DH5␣ (21) served as hosts for transformations. DHP-F (22) and DHP-F2 are derivatives of MC4100 with different deletions in the hypF gene; DHP-F lacks the segment of the gene reaching from amino acids 289 to 629, and DHP-F2 lacks the region from amino acids 59 to 629. DHP-F thus forms a shortened version of HypF that is detectable by anti-HypF antibodies in immunoblotting experiments of crude extracts, whereas no immunologically reacting material can be found in extracts from DHP-F2.
The plasmids employed in this work are listed in Table I, together with their genotype and source or derivation. Maintenance of plasmids in the transformants was selected for by inclusion of the appropriate antibiotics into the medium, ampicillin at 150 g/ml and chloramphenicol at 30 g/ml. If not indicated otherwise, bacteria were grown aerobically in LB medium (23) or anaerobically in TGYEP medium (24). Aerobic cultures were grown in Erlenmeyer flasks under rigorous rotatory shaking, and anaerobic growth took place in bottles filled with the medium to the top. Growth temperature was 37°C throughout.
Plasmid Construction-Plasmid pUCF18 was constructed for overexpression of the hypF gene. For this purpose, a 2.6-kb EcoRI-BamHI fragment from plasmid pTF5 (22) was inserted into plasmid pUC18 linearized by digestion with the same endonucleases. pUCF18, therefore, carries a hypF gene whose expression is under the control of the lac promoter.
Site-directed mutagenesis of the hypF gene was performed via inverse PCR using primers that carried the desired mutation (25). Alternatively, mutations were introduced with the aid of overlapping primers according to Ansaldi et al. (26). In each instance, plasmid pAF1 (22) was used as template, and the Expand High Fidelity PCR System from Roche Biochemicals (Penzberg, Germany) was employed. The amplified segment was phosphorylated, religated, and transferred into host DH5␣ via electroporation. Amplified fragments generated with the use of overlapping primers (26) were purified by passage over a QIAquick Spin column (Qiagen GmbH, Hilden, Germany) and directly used to transform strain DH5␣. The authenticity of the mutant hypF gene versions was verified by DNA sequencing using an ABI PRISM TM 310 sequencer (PE Applied Biosystems, Weiterstadt, Germany).
Purification of the HypF Protein-Cultures of strain JM109 transformed with plasmid pUCF18 were grown in 300 ml of LB medium in 2-liter Erlenmeyer flasks. After reaching an A 600 of 1.0, expression of hypF was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside at 1 mM final concentration. The cells were harvested after 3 h by centrifugation, resuspended in 10 mM Tris/Cl, pH 7.4, and sedimented again. The washed cells were suspended in 10 mM Tris/Cl, pH 7.4, 1 mM DTT (1% of the cultivation volume), and the suspension was brought to 20 g/ml phenylmethylsulfonyl fluoride and DNase I each.
The cells were broken by passage through a French press cell at 118 MPa, and the homogenate was clarified first by centrifugation at 10,000 ϫ g for 30 min (S10 supernatant) and subsequently for 2 h at 100,000 ϫ g (S100). The S100 fraction was adjusted to a protein concentration of 15 mg/ml and brought to an ammonium sulfate saturation of 35% by the addition of solid (NH 4 ) 2 SO 4 followed by stirring at 0°C for 30 min. The precipitate formed was collected by centrifugation (30 min at 15,000 ϫ g), dissolved in a minimum of buffer (10 mM Tris/Cl, pH 7.4, 1 mM DTT), and dialyzed against the same buffer.
The dialysate was fractionated by anion exchange chromatography on a Mono-Q HR 5/5 (Amersham Biosciences) column (1 ml) equilibrated with 10 mM Tris/Cl, pH 7.4, 1 mM DTT. Elution was with a gradient from 0 to 1 M sodium chloride at a flow rate of 60 ml/h. Fractions (1.5 ml) were monitored for presence of HypF by SDS-PAGE (27) and staining with Coomassie Blue. HypF-containing fractions were combined and subjected to ammonium sulfate precipitation at 50% saturation, and the precipitate formed was collected by centrifugation.
The sediment was dissolved in a small volume of Tris/Cl (10 mM, pH 7.4), 1 mM DTT containing ammonium sulfate at 20% saturation and applied to a 1-ml phenyl-Superose HR 5/5 column (Amersham Biosciences) that had been equilibrated with the same buffer. Bound proteins were eluted with a decreasing linear gradient from 20 to 0% ammonium sulfate at a flow rate of 30 ml per min. The elution of the HypF protein was monitored by SDS-PAGE. Fractions containing HypF were combined, and the protein was concentrated by ultrafiltration with the aid of the MICROSEP TM -Microconcentration system (Pall Gelman Laboratory, Dreieich, Germany).
The final purification step consisted of a gel filtration over a 300-ml Superdex XK 26/60 (Amersham Biosciences) column. It was developed with 25 mM Tris/Cl, pH 7.4, 100 mM sodium chloride, 1 mM DTT at a flow rate of 0.5 ml per min. 1-ml fractions were collected and assayed for HypF content by SDS-PAGE. Fractions containing HypF of apparent homogeneity were dialyzed against 25 mM Tris/Cl, pH 7.4, 1 mM DTT, 100 mM NaCl, 50% glycerol (v/v) and stored at Ϫ20°C.
Electrophoretic Techniques-SDS-PAGE was performed according to Laemmli (27) and non-denaturing PAGE as described by Drapal and Böck (9). Immunological detection of HypF in polyacrylamide gels after electrophoretic transfer onto nitrocellulose membranes was achieved with anti-HypF antibodies in a 1:1000 dilution and the use of the Lumi-Light Western blotting Substrate (Roche Diagnostics). Antibodies directed against HypF were generated by Eurogentec (Seraing, Belgium).
Determination of Carbamoyl-phosphate Phosphatase Activity-CP phosphatase activity was followed via the liberation of inorganic phos- Analysis of CP phosphatase activity in non-denaturing polyacrylamide gels was carried out as described by Mizuno et al. (29).
Dependence of ATP Hydrolysis by HypF Protein on the Presence of Carbamoyl Phosphate-ATP cleavage by HypF was measured in 1-ml reaction volumes containing 50 mM Tris/Cl, pH 7.4, 100 mM KCl, 1 mM MgCl 2 , 0.1 mM DTT, 25 nM HypF, 50 g of bovine serum albumin per ml, and [␥-32 P]ATP (specific radioactivity of 16.1 Ci/mol) and carbamoyl phosphate at the indicated concentrations. The reaction mixtures were incubated at 37°C, and 100-l samples were taken at the given intervals, mixed with 300 l of perchloric acid-washed charcoal suspension for 1 min, and centrifuged. The radioactivity of 200-l supernatant samples was determined in the scintillation counter. For control, samples without HypF protein or CP were analyzed. They were devoid of ATP cleavage activity.
[ 32 P]PP i -ATP Exchange Assay-PP i -ATP exchange activity catalyzed by HypF was determined in 1-ml assays containing 50 mM Tris/Cl, pH 7.4, 100 mM KCl, 1 mM MgCl 2 , 0.1 mM DTT, 0.1 mM ATP, 0.1 mM [ 32 P]PP i (specific radioactivity 25 Ci/mol), carbamoyl phosphate at the indicated concentrations, and 62.5 nM HypF protein. After 0, 1, 2, 4, 8, and 16 min of incubation at 37°C, 100-l samples were mixed with 25 l of HClO 4 (30%), and the mixture was combined with 30 l of acid-washed charcoal (12% suspension in water). After 5 min, 100 l of 1% cold sodium pyrophosphate solution in 14% HClO 4 was added, followed by incubation for 10 min with mixing and collection of the charcoal on Whatman GF/C glass fiber filters. The filters were washed twice with 5 ml of ice-cold 1% sodium pyrophosphate in 14% HClO 4 , once with 5 ml of ice-cold 1% sodium pyrophosphate in 1.4% HClO 4 , once with 5 ml of ice-cold distilled water, and dried. Their radioactivity was determined in a scintillation counter.
Analysis of Sequence Data-Homology searches were conducted with the aid of the NCBI data base of the National Institutes of Health and the ERGO data base of Integrated Genomics Inc. (Chicago, IL). DNA sequence analyses were performed with the GENEWORKS 2.5 program (Intelligenetics Inc., Geel, Belgium) or the Gene Tool Lite 1.D (Bio Tools Inc., Edmonton, Canada). Protein sequence alignments were calculated with either GENEWORKS 2.5 (or with MEGALIGN (DNAstar Inc., Madison, WI). The computer-based alignments were checked and amended "by hand" if appropriate.
Enzymes and Special Chemicals-Enzymes for restriction and modification of DNA were purchased from one of the following companies:

RESULTS
Sequence Characteristics of the HypF Protein-A schematic representation of sequence motifs strongly conserved in HypF homologs from bacteria and archaea is given in Fig. 1. The most intriguing characteristic is the existence of two perfect apparent "zinc finger" motifs from amino acid positions 109 -184 (30). They have been implicated in the binding of some bivalent cation, but experimental studies on such a function are lacking thus far. On the N-terminal side of these motifs a putative acyl phosphatase motif ranging from position 13-23 was discovered (15), but biochemical results demonstrating such an activity and its role in hydrogenase maturation are not available yet. Finally, in the C-terminal one-third of the protein there is a conserved sequence segment containing three histidine residues in a characteristic pattern. A data base search delivered best hits for enzymes with an O-carbamoylation activity in the synthesis of antibiotics or nodulation factors (see Fig. 1, top). Proteins possessing a similar motif, however, are also present in organisms like Pyrococcus horikoshii (31) or Sulfolobus tokodaii (32), which are not known to synthesize antibiotics or nodulation factors. The existence of this motif had prompted us to investigate whether CP is required for maturation of hydrogenases. Indeed, a mutant of E. coli devoid of CP synthetase activity was unable to develop active hydrogenases (18).
Purification and Properties of the HypF Protein from E. coli-The hypF gene on plasmid pUCF18 was overexpressed in strain JM109, and purified HypF was isolated from crude extracts by ammonium sulfate precipitation, Mono-Q HR anion exchange chromatography, hydrophobic interaction chromatography on a phenyl-Superose HR 5/5 residue, and gel filtration over a Superose XK 26/60 column, as detailed under "Experimental Procedures." The path of purification is displayed by Fig. 2A which presents a Coomassie-stained SDS-polyacrylamide gel in which the pooled fractions of each purification step were separated. Characteristically, from 142 mg of protein of a crude 30,000 ϫ g supernatant 9 mg of apparently purified HypF protein were obtained.
The UV-visible absorption spectrum of the purified protein, taken between 200 and 500 nm, did not indicate the presence of any cofactor absorbing in this range (data not shown). Size exclusion chromatography on a calibrated Superdex 200 column showed that HypF eluted at a position characteristic of a molecular mass of about 80 kDa that corresponds with the size retrieved from the migration in SDS gels and that delineated from the gene sequence, namely 81.9 kDa. Therefore, HypF as purified appears to be a monomeric protein. Analysis of purified HypF by atomic absorption spectroscopy and inductively coupled plasma atomic emission spectroscopy revealed that the protein (dialyzed against buffer lacking chelators) does not contain nickel, cobalt, copper, manganese, or molybdenum. However, the protein contained 1 iron atom per 7.8 HypF and 1 zinc atom per 2.5 HypF molecules. The substoichiometric ratio of the two metals may indicate that they are bound non-specifically. However, this needs further experimental analysis.
HypF Displays Carbamoyl-phosphate Phosphatase Activity-HypF shares a sequence signature motif with O-carbamoyltransferases, and it was tested to determine whether the protein purified can interact with CP. The liberation of inorganic phosphate was taken as a measure. Table II shows that HypF cleaves carbamoyl phosphate rather specifically; other acylphosphates or phosphoesters are only marginally hydrolyzed. The kinetics of CP cleavage by HypF were determined by assaying the hydrolysis rate at different substrate concentrations (Fig. 3). The liberation of phosphate followed Michaelis-Menten-type saturation kinetics. In several experiments K m values ranging from 260 to 330 M were obtained.
Next it was tested whether the CP phosphatase activity exhibited by purified HypF was also present in a freshly prepared crude extract. Cells of a mutant of E. coli (DHP-F) carrying a deletion in the chromosomal hypF gene were transformed either with plasmid pAF1 carrying hypF or with the vector. Extracts were prepared, and samples from S30 extracts were loaded on non-denaturing polyacrylamide gels, and the gels were analyzed for CP phosphatase activity (Fig. 2B). Enzyme activity developed specifically in the extract of the transformant carrying hypF on a plasmid, and it migrated in a position where the major band of the purified HypF was detected. Therefore, CP hydrolysis in the crude extract is catalyzed by the hypF gene product, and the purified HypF protein exhibits the same activity.
Carbamoyl Phosphate-dependent Liberation of AMP from ATP-The primary structure of HypF also contains a sequence motif (GXGXXGA) that resembles the "glycine-rich loop" motif of a family of ATP-binding proteins (GXGXXG(R/K)) (33). Its presence prompted the investigation whether HypF possesses ATP cleavage activity. [␣-32 P]ATP was therefore incubated with HypF protein, and the samples were applied to polyethyleneimine thin layer plates and separated with a solvent of 0.5 M KH 2 PO 4 (pH 3.4). Hydrolysis of ATP with the liberation of AMP could be observed but only when CP was included in the reaction mixture. In its absence, no significant ATP hydrolysis took place (data not shown).
To characterize the CP-dependent ATP hydrolysis activity of the HypF protein, initial reaction velocities were followed at different ATP concentrations keeping CP constant at 100 M and also at varying CP concentrations in the presence of 100 M ATP. Fig. 4 shows that the reaction follows Michaelis-Menten kinetics with both substrates. The kinetic constants of HypF in the CP phosphatase and the ATP hydrolysis reaction are summarized in Table III.
The apparent affinity of HypF for CP in the phosphatase reaction is far below that measured for it as substrate in the ATP hydrolysis reaction. By taking into account the rates of the two reactions, ATP cleavage in the presence of CP is kinetically favored.
The    5-7) and a crude extract of a ⌬hypF strain overexpressing hypF from a plasmid (⌬hypF/pHypF (lanes 3 and 4) were separated. In the control lanes 1 and 2, the identically treated extract of the ⌬hypF strain carrying the empty vector was applied (⌬hypF/pACYC184). The gel of in B was stained for CP phosphatase activity. Lanes 1 and 3 contain 60 g of protein, and lanes 2 and 4 contain 120 g of protein.
Lanes 5-7 contain 2, 4, and 6 g of protein, respectively. assessment, radioactively labeled pyrophosphate was incubated in the presence or absence of CP with unlabeled ATP, and samples were taken and assayed for the generation of radioactive ATP. The results obtained showed that such a PP i -ATP exchange indeed takes place when CP is present in the reaction mixture (Fig. 4B). The results also prove that HypF cleaves ATP into AMP and pyrophosphate and not by the sequential removal of two phosphate moieties, like in the selenophosphate synthetase reaction (36). The entry into a plateau might either be because of the attainment of the equilibrium in the PP-ATP exchange reaction or be caused by CP substrate limitation. The retardation of the reaction in the presence of 100 M CP is unspecific because CP at this concentration and higher inhibits HypF activity. 3 Mutational Analysis of HypF-To gain further insight into the function of HypF in the process of hydrogenase maturation and, in particular, to correlate the in vitro activities of the protein with the formation of the CO and CN ligands of the metal center, an extensive mutagenesis of the hypF gene has been carried out. Residues from the acylphosphatase, the zinc finger, and the O-carbamoyltransferase motifs were replaced either singly or in combination (see Fig. 1). The mutant gene 3 A. Paschos, A. Bauer, and A. Böck, unpublished data.

TABLE IV CP phosphatase activity of wild type and mutant HypF variants
Transformants were grown in TGYEP medium, and crude extracts of cells were centrifuged at 100,000 g, X and the resulting supernatant was subjected to ammonium sulfate precipitation at 30 % saturation. 200 g of protein of the dissolved and dialyzed sediment were taken for measurement of CP phosphatase activity.  products were analyzed for their in vivo stability when the mutated genes were expressed from a plasmid in strain DHP-F2 (⌬hypF) (Fig. 5A). With some remarkable exceptions, most of the mutant alleles yielded products at approximately the same level. The exceptions were HypF-R23K, HypF-C162A, and HypF-H475A/H476A. Intriguingly, exchange of a single cysteine from the C-terminal "zinc finger motif" (see Fig. 1, ZF2) destabilized the protein completely, whereas the replacement of the distal (neighboring) one in addition yielded a stable product (Fig. 5A, lanes 10 and 11). Surprisingly, the identical replacements at the N-terminal zinc finger motif delivered a stable product when one of the cysteines was exchanged but not in case of the double replacements (Fig. 5A, lanes 8 and 9).
The consequences of the mutations carried by the different HypF variants on their role in hydrogenase maturation was assessed by following the proteolytic processing of the precursor of HycE, which is the large subunit of hydrogenase 3 (Fig.  5B). The immunoblot indicates that exchange of the histidine residues in positions 475, 476, and 479 against alanine delivers gene products that are still functional in hydrogenase maturation. All the other variants were inactive. This was corroborated by measuring hydrogenase activity in non-denaturing polyacrylamide gels after separation of 100 g of protein of crude extracts from the mutants. Upon substrate staining, bands reflecting hydrogenase 2 and hydrogenase 1 activity were detected in case of the mutant variants HypF-H475A, HypF-H476A, and HypF-H479A.
Finally, it was important to correlate the proficiency of the HypF variants in the hydrogenase maturation process with their in vitro capacity to interact with CP. The transformants harboring the mutant alleles on a plasmid were grown anaerobically to an A 600 of 1.0. The cells were harvested, and crude extracts were prepared that were partially purified by centrifugation at 100,000 ϫ g and by ammonium sulfate precipitation. 200 g of protein of the 30% sediment fraction was analyzed for CP phosphatase activity (Table IV). Variants carrying the H475A, H476A, or H479A exchanges displayed full activity, whereas variant R23Q had detectable CP phosphatase activity but at a very low level. All the other variants that produced stable HypF protein were devoid of activity. DISCUSSION The synthesis of the CO and CN ligands of [NiFe] hydrogenases and their attachment to the iron poses a number of intriguingly novel features of bioinorganic chemistry. As some of the most toxic compounds in biology, it is predictable that they are not synthesized in the free state but rather bound to some adaptor, thus preventing their interaction with susceptible and essential cellular components. It is still unclear whether they are attached to the iron when it has been inserted into the large hydrogenase subunit or whether they are preformed, e.g. at some scaffold protein and transferred as the complete entity into the apoprotein. Moreover, it is still unclear how the precise stoichiometry of 2 CN and 1 CO per iron atom is achieved.
We have reported recently (18) that CP is required for the synthesis of active [NiFe] hydrogenases. The experimental evidence is as follows: (i) a mutant of E. coli devoid of CP synthetase activity was unable to synthesize all three hydrogenases based on the inability to mature and process the large subunits, and (ii) supplementation with citrulline as a source of CP restored this capacity. It has also been pointed out that reactions in metallo-organic chemistry have been described that convert a carbamoyl into a carbonyl or cyanyl moiety (18).
The results reported here convincingly show that the hydrogenase maturation protein HypF is interacting with CP as a substrate and that this interaction is involved in hydrogenase maturation. Purified HypF dephosphorylates CP specifically, and this activity can be discovered in electrophoretically separated extracts from an E. coli strain overexpressing hypF. Because the HypF protein has been purified under aerobic conditions, this also means that the CP phosphatase activity is oxygen-stable. A causal connection between CP and hydrogenase maturation by HypF is then provided by the mutational analysis. There is a clear and quantitative parallelism between maturation activity of HypF and the in vitro CP phosphatase activity. Somewhat surprising is the discovery that mutations in all three major signature motifs, the acylphosphatase, the zinc fingers, and the O-carbamoyltransferase motifs, can lead to the blockade of CP phosphatase activity. It indicates an integrated cooperativity between these domains in the cleavage reaction.
The delineation of the route how HypF could convert the carbamoyl moiety into one or both of the CO/CN ligands is not possible yet. It needs detailed analysis of the product formed in the CP-dependent ATP cleavage reaction and, especially, also the analysis of the activity of the HypE protein, since it has been shown recently that HypF and HypE form a complex (37). 3 We strongly emphasize in this context that the reactions assayed are those of the sole HypF protein in the absence of HypE or putative additional substrates. Although they present information on the existence of substrate-binding sites and types of reactions catalyzed, they do not preclude the possibility that the products measured are those of side reactions when reaction partners are missing, and the flux is blocked at the state of the same intermediate.
Bearing all this caveats in mind, speculations can be attempted on the fate of carbamoyl phosphate when acted upon by HypF and HypE. First, from the two reactions catalyzed by HypF, the CP phosphatase activity is kinetically somewhat less favored in comparison to the CP-dependent ATP cleavage. There are several explanations for this inefficiency. First, CP hydrolysis may represent a side reaction that is followed in the absence of other substrates, reflecting the transfer of the carbamoyl residue to a water molecule. Second, and alternatively, it could represent the carbamoyl transfer to some acceptor group of the HypF protein that is paralleled by the liberation of inorganic phosphate. The carbamoylated amino acid residue could then lose its acyl group because the conversion into the adduct is blocked, possibly due to the absence of HypE.
On the other hand, the incorporation of labeled pyrophosphate into ATP that is catalyzed by HypF in the presence of CP supports the view that CP is adenylated. Although exchange of the phosphoryl group of CP by an adenyl residue cannot be excluded (energetically this would not make sense), our favorite hypothesis is that adenylation takes place at the hydroxyl of the tautomeric form of the carbamoyl moiety yielding the iminoform of the carbamoyl adenylate. Removal of a hydrogen from the imino group would directly lead to the cyano moiety. It resembles the well known dehydration reaction with the formation of a phosphorylated intermediate. Interestingly, HypE shares structural similarity with PurM which catalyzes such a reaction in the purine biosynthetic pathway (16). Further work is aimed at this possibility.