Biochemical and Structural Analysis of the Molybdenum Cofactor Biosynthesis protein MobA

Molybdopterin guanine dinucleotide (MGD) is the form of the molybdenum cofactor that is required for the activity of most bacterial molybdoenzymes. MGD is synthesized from molybdopterin (MPT) and GTP in a reaction catalyzed by the MobA protein. Here we report that wild type MobA can be co-purified along with bound MPT and MGD demonstrating a tight binding of both its substrate and product. In order to study structure-function relationships, we have constructed a number of site-specific mutations of the most highly conserved amino acid residues of the MobA protein family. Variant MobA proteins were characterized for their ability to support the synthesis of active molybdenum enzymes, to bind MPT and MGD, to interact with the molybdenum cofactor biosynthesis proteins MobB and MoeA, and by X-ray structural analysis. Our results suggest an essential role for glycine 15 of MobA, either for GTP-binding and/or catalysis, and an involvement of glycine 82 in the stabilization of the product-bound form of the enzyme. Surprisingly, the individual and double substitution of asparagines 180 and 182 to aspartate did not affect MPT binding, catalysis, and product stabilization. wild and MGD in the ratio of Based on comparisons and the high resolution X-ray structure, we have constructed a number of site-directed variants of the E. MobA protein. Our biochemical and structural analyses indicate an essential role for 15 in the conversion of to and of glycine 82 in stabilizing the product-bound form of Inactive variant forms of do not lose the ability to with and the distinct from


INTRODUCTION
The transition metal molybdenum is an essential element for most living organisms. It is required for the activities of a range of enzymes that catalyze two electron redox reactions (1). Molybdenum in these enzymes is always found co-ordinated to an organic cofactor, which in its most simple form is molybdopterin (MPT). The biosynthesis of MPT is by an evolutionarily conserved multi-step pathway.
The molecular structures of most of the proteins involved in MPT synthesis have now been determined, and the biochemistry of its assembly and insertion into molybdoenzymes is an area of intense study (2)(3)(4)(5)(6)(7).
The biosynthesis of the molybdenum cofactor (Moco) can be divided into three stages. The first step is the rearrangement of a guanine nucleotide to form precursor Z (8). The second stage is the introduction of two sulfur atoms into the pyran ring of precursor Z to give MPT (9)(10)(11). The third step is the chelation of molybdenum and the formation of active Moco. Whereas in eukaryotes this step is catalyzed by two domain proteins (Cnx1 in plants, gephyrin in mammals), in bacteria the separate MogA and MoeA proteins have been implicated in this latter step (5,6,12,13). Unique to prokaryotes, a further modification of the basic MPT structure by attachment of a mononucleotide to the terminal phosphate of MPT occurs (14). This modification is essential for the activity of most, but not all, bacterial Mococontaining enzymes. In Escherichia coli, most of the molybdoenzymes require the GMP-modified form of molybdopterin, molybdopterin guanine dinucleotide (MGD) for their activity, although enzymes from other bacteria that utilize molybdopterin cytosine dinucleotide have also been reported (14,15). In E.
coli, the GMP attachment step is catalyzed by the cellular protein MobA (16). Mutants in mobA fail to synthesize MGD, and accumulate elevated quantities of MPT (14). The E. coli mobA gene is cotranscribed with a further gene, mobB, encoding a nucleotide-binding protein, that is not absolutely required for MGD synthesis (17, 18).
Recent work has suggested that in vivo, molybdenum cofactor biosynthesis most probably occurs on protein complexes rather than by the separate action of the biosynthetic enzymes (19). In particular it seems that Mo insertion and nucleotide attachment to form MGD in E. coli are intimately linked. Using a two-hybrid approach, it has been shown that the MobA  pH8.0, 300mM NaCl, 20mM imidazole, 10% (v/v) glycerol) and lysed by sonication. The crude cell extract which was prepared after centrifugation (26) was loaded onto a 1ml Ni-nitrilotriacetic acid column that had been previously pre-equilibrated with washing buffer (50mM Na phosphate, pH6.0, 300mM NaCl, 20mM imidazole, 10% (v/v) glycerol). After loading, the column was washed with eight column volumes of washing buffer and the bound protein eluted by application of three column volumes of washing buffer supplemented with a further 480mM imidazole. Eluted protein was applied immediately onto a 10ml PD10 desalting column that had been previously equilibrated with desalting buffer (100mM Tris-HCl, pH7.2). Protein was washed through the column with the same buffer and Crystallization and data collection. Crystals of the MobA variants were obtained and cryoprotected as described previously for the wild type (20). X-ray data were recorded at a temperature of 100 K at the Synchrotron Radiation Source in Daresbury (UK), either on station PX9.5 using a 165 mm MAR-Research charge-coupled device (CCD) detector, or on station PX9.6 using an Area Detector Systems Corporation Quantum 4 CCD detector. The resultant data were scaled and merged using the HKL by guest on March 24, 2020 http://www.jbc.org/ Downloaded from package (32) and all subsequent downstream processing and statistical analysis was effected using programs from the CCP4 suite (33). X-ray data collection parameters are summarized in Table I. Given that all crystals were isomorphous with the wild-type crystals (20) (i.e. space group P2 1 2 1 2 with approximate cell parameters of a = 76.6 Å, b = 41.8 Å, c = 54.5 Å), the wild-type coordinates (PDB accession code 1E5K) could be used as a starting model in each case.
Prior to refinement, in order to reduce model bias, the substituted residues were truncated to alanine, the temperature factors were reset to the overall value estimated from the Wilson plot in the program TRUNCATE (34), and all the coordinates were subjected to small random shifts (not exceeding 0.3 Å).
After rigid body refinement to convergence the structures were subjected to restrained refinement with REFMAC5 (35) and the solvent was automatically modelled and refined with ARP (36). Interactive model building with the program O (37) was used to correct errors in the models and to introduce the side chains of the substituted residues where appropriate. After the final refinement cycle the resultant structures were evaluated using PROCHECK (38). Model parameters are summarized in Table I.

Site-directed mutagenesis of the E. coli mobA gene
Based on protein sequence alignments (Fig 1), a number of amino acid residues are completely conserved across the MobA family of proteins. Many of these highly conserved residues fall in the putative substrate binding pocket of the E. coli MobA protein that has been previously identified (20, 21). In order to probe the functions of these residues, we have constructed a number of site-directed mutations in the mobA gene ( Fig. 1). Conserved residues R19, K25, D101, R156 were initially substituted by alanine, the conserved glycine residues (G15, G22, G78 and G82), were substituted by leucine, and the two invariant asparagines (N180 and N182) were replaced singly and in combination by aspartate.
To test the effect of the mutations on the ability of the cell to synthesize active molybdenum cofactor in vivo, we measured the specific nitrate reductase activity of the mobAB mutant with plasmids encoding the histidine-tagged wild type or variant proteins. As shown in Fig. 2, only two mutations led to a complete loss of MobA function. Substitution of glycine 15 for leucine (G15L), or aspartate 101 for alanine (D101A) totally abolished MobA activity. However, further analysis of the D101A mutation indicated that most of the protein was present in the cells in the form of inclusion bodies, which could account for the lack of activity. Therefore we made a further substitution of aspartate 101 for asparagine (D101N).
The mob mutant expressing D101N showed reduced, but still detectable, nitrate reductase activity. Of the other point mutations, the mob mutant expressing the K25A-substituted protein repeatedly showed lower nitrate reductase activity. The other mutations gave essentially similar levels of active nitrate reductase as the wild type protein.

Co-purification of molybdenum cofactor forms with his-tagged MobA
We next wanted to study individual activities of the purified proteins in order to understand which of the functional properties of MobA were affected by the mutations. As a consequence of the reaction catalyzed by MobA it should interact with its substrates (MPT, GTP) as well with the reaction product (MGD). We have shown previously that the G-domain of the MPT-binding protein Cnx1 (39), as purified from E. coli, contains bound MPT (13). We therefore reasoned that the purified wild type MobA protein may also contain one or more forms of bound molybdenum cofactor. Consistent with this, we were able to demonstrate that the histidine-tagged wild type protein could be co-purified with up to 50pmoles of molybdenum cofactor per mg of protein (1.1mmol Moco/mol MobA). When the analysis was repeated under conditions of mild oxidation (that prevents cleavage of the phosphodiester bond), it was apparent that most of the bound molybdenum cofactor was in the form of the reaction product, MGD (Fig. 3B). We typically observed a ratio of MPT:MGD of 1:3 for the wild type protein.
We next expressed and purified the variant MobA proteins. All of the his-tagged variant proteins (apart from the D101A variant) expressed to a similar level and were isolated to the same degree of purity as the his-tagged wild type MobA (Fig. 3A). Total Moco analysis revealed that each of the amino acid substituted proteins was also capable of binding and co-purifying the cofactor (data not shown). The MobA led to a notable increase in the ratio of bound MPT relative to MGD. Taken together with the observation that both the K25A and D101N replacements resulted in a marked decrease in the level of nitrate reductase activity (Fig. 2), this indicates that the altered MPT:MGD ratio is of functional significance.
G15L was the only modified form of MobA that showed a complete absence of nitrate reductase activity, consistent with an inability to synthesize MGD. However, the G15L variant of MobA, as purified, contained a relatively high amount of bound MPT. This may be due to the fact that there are higher levels of MPT in the cell when MobA is inactive (14) which probably accounts for the relatively high amount of MPT bound. This supports the idea that the defect in the G15L variant is not at the level of MPT binding, but must instead be at the level of binding of the other substrate (GTP) or at the catalytic step. The leucine side chain probably partially occludes the guanine binding pocket or alternatively disrupts the local structure sufficiently to prevent GTP binding. In order to ascertain whether this MobA variant was capable of interacting with the reaction product, we expressed the plasmid-encoded his-tagged G15L MobA protein in a strain carrying the wild type chromosomal copy of mobA. As shown in Table II, under these circumstances, the G15L protein could also interact with MGD. It is notable that this variant only binds half the wild type level of MGD. This is presumably due to a lower affinity of the variant MobA protein for the product resulting from occlusion or disruption of the guanine binding pocket.
An unexpected result was obtained with the MobA G82L variant. Despite repeated attempts, we were only able to detect MPT in the purified G82L protein (Fig. 3C). This observation is particularly surprising given the fact that when expressed in the mob mutant strain TP1000, the G82L variant of MobA was able to support a nitrate reductase activity level that was almost 60% of wild type, indicating that it was capable of synthesizing MGD. These observations imply that the binding of the reaction product, MGD, is destabilized as a result of the substitution.

Structural analysis of amino acid substituted MobA proteins.
All of the variant MobA proteins that were analyzed for Moco binding were also subjected to crystallization trials. Of these, five yielded crystals that were suitable for X-ray data collection. These were the R19A, G22L, D101N, N180D and N182D substituted proteins, and their structures were subsequently determined to resolutions ranging from 1.65 to 2.00 Å. In all cases, significant structural changes were restricted to the vicinity of the site of substitution (Fig. 5). The largest changes were observed for the G22L variant, where the preceding five residues, which form the central part of the consensus loop, adopted a different conformation. A salt bridge, present in the wild type MobA structure between D101 and K25 was absent in the D101N variant, and both of these side chains had slightly different conformations. By contrast the R19A-substituted structure was essentially identical to that of the wild type protein. Furthermore, no significant changes were apparent for either the N180D or N182D variants, although the substituted side chains adopted slightly different configurations relative to the wild type. It is possible that those variants that did not yield crystals had significantly different structures to the wild type protein. However, this seems unlikely since, with the exception of G15L, all were active and even the latter was able to bind both MPT and MGD. It is more likely that subtle changes in surface by guest on March 24, 2020 http://www.jbc.org/ Downloaded from charge and/or side chain conformations were sufficient to prevent crystal lattice formation under the wild type crystallization conditions.

DISCUSSION
In this study we have substituted ten of the most highly conserved amino acid residues in the MobA protein family. Almost all of these conserved residues cluster around the proposed substrate binding pocket of MobA (20, 21). Residues G15, R19 and G22 are found in the consensus loop, most of which is poorly ordered in the MobA structure, that is proposed to wrap around the substrates and partially sequester them from solvent exposure during catalysis. This may be a mechanism whereby the enzyme prevents futile hydrolysis of GTP by water once it has been bound in a catalytically favourable conformation. R19 has been proposed to play a crucial role in the catalytic mechanism of MobA by stabilizing the transition state. Our results do not support this contention. Substitution of R19 for alanine gave a MobA protein that was functionally active and that could bind MPT and MGD in the same ratio as inability to effect efficient catalysis. A substitution of K25 for alanine also gave a significant reduction of MobA activity and a two fold decrease in the level of bound Mo-cofactor, that was mainly due to a reduction of the bound MGD (Fig 3B).
Residues N180 and N182 have been postulated to form part of the MPT binding site, and to contact O4 and N5 of MPT (21). We constructed single aspartate substitutions at each of these residues as well as a doubly-substituted protein. All three variant proteins were active and bound the same quantities of MPT and MGD as wild type MobA. The high resolution structures of the singly-substituted proteins were essentially identical to wild type. Our experimental data suggest that these highly conserved residues are not strictly required for MPT binding, so that the position of MPT as extrapolated by Lake et al. Bacillus subtilis (43). Conserved residues that were subjected to mutagenesis in this study are indicated by asterisks under the sequence.     a The figures in brackets indicate the values for outer resolution shell. b R merge = ∑(| I j -< I j > |)/∑< I j >, where I j is the intensity of an observation of reflection j and < I j > is the average intensity for reflection j. c R iso = ∑(| F PH -F P |)/∑| F P |, the mean fractional isomorphous change between the wild type structure factors (F P ) and the variant structure factors (F PH ). d The R-factors R cryst and R free are calculated as follows: R = ∑(| F obs -F calc |)/∑| F obs | x 100, where F obs and F calc are the observed and calculated structure factor amplitudes, respectively. e Diffraction-component precision index (44) -an estimate of the overall coordinate errors calculated in REFMAC (35). f As calculated using PROCHECK (38). g After least-squares superposition based on all main chain atoms.