Elucidation of the Dual Role of Mycobacterial MoeZR in Molybdenum Cofactor Biosynthesis and Cysteine Biosynthesis

The pathway of molybdenum cofactor biosynthesis has been studied in detail by using proteins from Mycobacterium species, which contain several homologs associated with the first steps of Moco biosynthesis. While all Mycobacteria species contain a MoeZR, only some strains have acquired an additional homolog, MoeBR, by horizontal gene transfer. The role of MoeBR and MoeZR was studied in detail for the interaction with the two MoaD-homologs involved in Moco biosynthesis, MoaD1 and MoaD2, in addition to the CysO protein involved in cysteine biosynthesis. We show that both proteins have a role in Moco biosynthesis, while only MoeZR, but not MoeBR, has an additional role in cysteine biosynthesis. MoeZR and MoeBR were able to complement an E. coli moeB mutant strain, but only in conjunction with the Mycobacterial MoaD1 or MoaD2 proteins. Both proteins were able to sulfurate MoaD1 and MoaD2 in vivo, while only MoeZR additionally transferred the sulfur to CysO. Our in vivo studies show that Mycobacteria have acquired several homologs to maintain Moco biosynthesis. MoeZR has a dual role in Moco- and cysteine biosynthesis and is involved in the sulfuration of MoaD and CysO, whereas MoeBR only has a role in Moco biosynthesis, which is not an essential function for Mycobacteria.


Introduction
Among the metabolic pathways requiring sulfur transfer are those leading to the formation of FeS clusters, biotin, thiamin, lipoic acid, molybdopterin (MPT), and sulfur-containing bases in RNA [1]. In addition, a new pathway for cysteine biosynthesis involving sulfur transfer has been elucidated in Mycobacterium tuberculosis recently [2,3]. This pathway involves the ubiquitin-fold protein CysO that contains a C-terminal thiocarboxylate group at the last glycine of a GG-motif as sulfide donor [4]. Similar proteins involved in sulfur transfer are the ThiS protein involved in thiamin biosynthesis [5] and the MoaD protein involved in molybdopterin biosynthesis [6]. The M. tuberculosis protein MoeZR has been shown to transfer sulfur onto CysO for the formation of the thiocarboxylate group. The cysteine synthase M (CysM) catalyzes the addition of O-phosphoserine to the carboxy-terminus of the protein-bound CysO-thiocarboxylate to generate a cysteine-CysO adduct [7]. The protease Mec + hydrolyzes the CysO-cysteine adduct to release cysteine and regenerate CysO [2]. The CysO activating enzyme MoeZ belongs to a superfamily of proteins consisting of related proteins that are members of pathways involved in the transfer of sulfur-containing moieties to metabolites [8]. Members of this family are referred to as MoeB, MoeBR, MoeZR, MoeZ and MOCS3 [9]. MoeB, the molybdopterin synthase activating enzyme in molybdenum cofactor (Moco) biosynthesis, is the best characterized protein from this family [10]. E. coli MoeB was shown to activate the small subunit of MPT synthase MoaD to form an acyl-adenylate intermediate at its C-terminal glycine [11]. Subsequently, the MoaD acyl-adenylate is converted to a thiocarboxylate either by a L-cysteine desulfurase using L-cysteine as sulfur source [12] or by a rhodanese-like protein containing a protein-bound sulfane sulfur [13]. In all eukaryotic and several bacterial MoeB homologs a rhodanese-like domain is fused to the C-terminus of MoeB (referred to as MoeBR and MoeZR) [9]. This rhodanese-like domain was shown to be directly involved in the generation of the thiocarboxylate group of the MoaD homolog in humans [13]. While all sequenced eukaryotic MoeB-homologs so far contain the C-terminal rhodanese-like domain, the bacterial homologs are much more divergent. Here, either a MoeB-homolog is present, a MoeBhomolog with a C-terminal rhodanese-like domain (MoeBR), a MoeZ-homolog, a MoeZ-homolog with a C-terminal rhodaneselike domain (MoeZR) or a combination of several of these proteins [9]. MoeBR and MoeZR proteins share a high amino acid sequence identity. The main difference between both proteins are two conserved CXXC motifs near the C-terminus of the MoeBdomain. While these two motifs are present in all MoeB and MoeBR proteins, MoeZ and MoeZR homologs are mostly missing the second motif completely, while the first motif comprises the consensus sequence NYRD [9].
Recently, the role of genes involved in Moco biosynthesis were described for M. tuberculosis [14]. While E. coli contains single copies of the genes for Moco biosynthesis, a distinguishing feature of members of the M. tuberculosis complex is their possession of multiple homologs associated with the steps of the Moco biosynthesis pathway for the conversion of GTP to cyclic pyranopterin monophosphate (cPMP), the formation of molybdopterin (MPT) by insertion of two sulfur atoms into cPMP and the final insertion of molybdate to the dithiolene sulfurs of MPT, thus forming Moco [14,15]. According to our previous nomenclature we refer to the protein for the moeB2 gene as MoeBR and for the protein to the moeB1 gene as MoeZR. A gene region containing moaA1-moaB1-moaC1-moaD1 and moeB2 was acquired by horizontal gene transfer in M. tuberculosis H37Rv and M. bovis BCG [14]. In addition, a moaX gene was identified located downstream of this cluster which comprises a fusion of moaD and moaE and was shown to form active components of MPT synthase, since the moaX gene was able to complement both, a M. smegmatis moaD2 and moaE2 mutant [14].
Previously, the effects of several M. tuberculosis mutants in genes for Moco biosynthesis were described [14]: a M. tuberculosis moeB1 mutant is defective in arresting phagosome maturation [16]; moaC1 and moaX mutants showed a reduced ability to parasitize macrophages [17]; and a moaC1 mutant was attenuated for growth in primate lungs [18]. In addition, the gene expression profile of M. tuberculosis in mice specifically identified moaB2 as part of a gene cluster which was highly expressed in vivo [19]. Most recently, independent transposon mutants in moaC1 and moaD1 in the M. tuberculosis W-Beijing strain, GC1237, were identified by highcontent, phenotypic cell-based screening as defective in the ability to arrest phagosome maturation [20]. An important role for nitrate reductase under anaerobic growth conditions has been ascribed for M. tuberculosis [14].
In this study we investigated the role of MoeBR and MoeZR in Moco biosynthesis. Mycobacterial MoeBR and MoeZR were purified and characterized in their activities. Both MoeBR and MoeZR showed thiosulfate:sulfurtransferase activity. After coexpression of MoeZR and MoeBR with either MoaD1 or MoaD2 in E. coli, sulfurated MoaD-homologs for Moco biosynthesis were obtained. The sulfur-transfer was verified by mass spectrometry. Direct interaction between MoeZR, MoeBR and MoaD1, MoaD2 or CysO were analyzed by copurification studies. Our results show that both MoeBR and MoeZR have a role in Moco biosynthesis. While MoeBR preferentially interacts with both MoaD proteins, MoeZR has a dual role in the cell and is able to transfer the sulfur to either of the two MoaDs or CysO, a protein involved in cysteine biosynthesis.

Test of functional complementation of E. coli moeB 2 cells with MoeBR and MoeZR
To analyze the role of the duplicated E. coli MoeB homologs from M. tuberculosis MoeBR and MoeZR in Moco biosynthesis, functional complementation studies of an E. coli moeB 2 strain were performed. To investigate the functional properties of both proteins, M. tuberculosis MoeBR and MoeZR were cloned into expression vectors (obtained from J. Kuper, EMBL Hamburg), resulting into fusion proteins with an N-terminal MBP-tag. For functional complementation studies, the corresponding vectors were expressed in the E. coli moeB mutant strain (plus 20 mM IPTG). Complementation of the strains by the Mycobacterial proteins would result in the production of active nitrate reductase, a Moco-containing enzyme, the activity of which is dependent on the ability of cells to synthesize Moco. The activity of nitrate reductase can be quantified in crude cell extracts ( Table 1). The values in Table 1 show that MoeBR but not MoeZR alone was able to complement the E. coli moeB 2 strain at least to some extent (6-fold increase in comparison to the background), thus, it is speculated that MoeZR is unable to interact with the E. coli MoaD protein. To circumvent this, a coexpression with the corresponding Mycobacterium homolog would be necessary. For this purpose, we cloned the genes moaD1, moaD2, moaX and additionally cysO from M. bovis total DNA, which is commercially available. The amino acid sequences are identical with the ones of the M. tuberculosis proteins, so the M. bovis sequence should be suitable for protein expression and interaction studies of the purified proteins with M. tuberculosis MoeBR and MoeZR. MoaX is a fusion protein of MoaD and MoaE and was shown before to harbor MPT synthase activity [14]. The genes were cloned into P15A origin based expression vectors which allowed simultaneous replication of two vectors after transformation in the E. coli moeB 2 cells. The results in Table 1 show that in combination with MoaD1 and MoaD2, MoeBR and MoeZR were both able to functionally complement the E. coli moeB 2 strain. The determined nitrate reductase activities for MoeZR were 13-14 fold higher in conjunction with MoaD1 and MoaD2 in comparison to the background, which implies that MoeZR is well able to activate the Mycobacterial MoaD homologs but not MoaD from E. coli. The nitrate reductase activities for the complemented cells with MoeBR in addition to MoaD1 and MoaD2 were 1.9-2.3 increased in comparison to the single complementation with Nitrate reductase activity was determined in crude cell extracts by a spectroscopic assay using benzyl viologen as described by Jones and Garland [27]. One unit of nitrate reductase activity is described as the production of 1 mmol of nitrite per min per O.D. 600. Assays were performed in triplicate. Significant differences at a level of P,0.01 (Student's t test) are indicated as **. All values represent means 6 standard deviation (SD) for n = 9. MoeBR, showing that MoeBR better interacts with its own partners. The nitrate reductase activities of MoeZR and MoeBR in conjunction with the MoaD homologs were comparable (Table 1). No functional complementation was obtained by coexpressing MoaX.

Purification of MoeBR and MoeZR
To further characterize MoeBR and MoeZR, both proteins were purified after heterologous expression in E. coli. For purification of MoeZR and MoeBR, both genes cloned in the pMalC2x vector were used. Expression of the proteins in BL21(DE3) cells resulted in N-terminal MBP-tagged recombinant proteins. MoeZR and MoeBR were purified by amylose affinity chromatography. One major band was visible on SDS-polyacrylamide gels with a size corresponding closely to the calculated molecular masses of the MBP-MoeZR fusion protein of 94 kDa and MBP-MoeBR with a mass of 93 kDa ( Figure 1). However, some degradation products are visible on the SDS-polyacrylamide gel. The band at 45 kDa was determined to be MBP, as verified by MALDI peptide mapping. MoeZR and MoeBR were purified with a yield of 4.4 mg/L or 7.8 mg/L E. coli cells. To determine whether structural differences between MoeBR and MoeZR exist, CD spectra of both proteins were recorded. As shown in Fig. 2 the far-ultraviolet CD spectra revealed mainly no differences in the average composition of secondary-structural elements of MoeZR and MoeBR.

Analysis of the sulfurtransferase activity of MoeZR and MoeBR
For MoeB-like proteins with a C-terminal rhodanese-like domain, it was shown before that they contain a conserved cysteine residue, which is part of a highly conserved six-amino acid active-site loop, that is essential for thiosulfate sulfurtransferase activity by formation of a persulfide group during catalysis. After purification of MoeBR and MoeZR, it was of interest to compare the activities of the C-terminal rhodanese-like domain of the purified proteins. In vitro, thiosulfate sulfurtransferase activity was measured by the method described by Sörbo [21]. The data in Table 2 show that MoeBR and MoeZR express thiosulfate sulfurtransferase activity. The k cat and K M values for MoeZR and MoeBR with thiosulfate are comparable as revealed by the k cat / K M values, however, the k cat /K M with cyanide is only half for MoeBR in comparison to MoeZR. In total, the k cat values are 112-196 times higher in comparison to human MOCS3 [9] and comparable with yeast Uba4 [22], homologous proteins from eukaryotes.

Copurification of MoeBR and MoeZR with MoaD1, MoaD2 and CysO
To additionally confirm that MoeZR interacts with MoaD1 and MoaD2, it should be possible to copurify the protein complexes after coexpression in E. coli. Plasmids expressing His 6 -MoaD1, His 6 -MoaD2, and His 6 -CysO were cotransformed with MBP-MoeZR and MBP-MoeBR and subjected to Ni-NTA affinity chromatography after coexpression. The results in Figure 3 show, that with His 6 -MoaD1, His 6 -MoaD2 and His 6 -CysO, MBP-MoeZ was copurified. In contrast, MBP-MoeBR was only copurified in the mixtures containing His 6 -MoaD1 and His 6 -MoaD2. Thus, MoeBR does not interact with CysO. The results were additionally confirmed by immunoblot analyses using an MBP-antibody ( Figure 3) and in addition by MALDI peptide mapping (data not shown). Both analyses confirmed clearly, that MoeBR does not interact with CysO.

Electrospray Mass Spectrometry of MoaD1, MoaD2 and CysO after coexpression with MoeBR and MoeZR
To directly prove the existence of a thiocarboxylate group at the C-terminal glycine of either His 6 -MoaD1, His 6 -MoaD2 or His 6 -CysO after coexpression with MBP-MoeBR or MBP-MoeZR, respectively, the purified proteins were subjected to electrospray ionization (ESI) mass spectrometry, which allows the detection of the oxygen versus sulfur exchange due to its characteristic mass shift of 16 Da. As shown in Fig. 4, His 6 -MoaD1 and His 6 -MoaD2 contained a C-terminal thiocarboxylate group after coexpression with MBP-MoeBR and MBP-MoeZR. In contrast, on His 6 -CysO a thiocarboxylate group was only determined when MoeZR was present for coexpression. The presence of the modification at the C-terminal glycine was verified by MS/MS based on differences after carboxyterminal fragmentation analyses (data not shown). This clearly shows that both MoeBR and MoeZR are able to activate MoaD1 and MoaD2, however, only MoeZR is able to form a stable complex with CysO which leads to the formation of the C-terminal thiocarboxylate group on CysO.

Discussion
Our studies involve the characterization of Mycobacterial MoeZR and MoeBR. A distinguishing feature of members of the M. tuberculosis species is their possession of multiple homologs associated with the first step of Moco biosynthesis [14,23]. Two homologs of the E. coli MoeB protein were identified in M. tuberculosis H37Rv and M. bovis BCG Pasteur ( Table 3). The corresponding genes were annotated as moeB1 and moeB2 [14]. According to our previous nomenclature we refer to the protein for the moeB2 gene as MoeBR and for the protein to the moeB1 gene as MoeZR, since both proteins contain a C-terminal rhodanese-like domain, which is present in some bacterial and eukaryotic homologs including human MOCS3, but not in E. coli MoeB [9]. For M. tuberculosis MoeZR it was described that it plays a role in a novel cysteine biosynthesis pathway [2]. This cysteine biosynthesis pathway involves a small sulfur carrier protein, CysO, which carries a C-terminal thiocarboxylate group and a cysteine synthase, CysM [4]. CysM reacts with O-phosphoserine to form an a-aminoacrylate intermediate [7]. The sulfur is then provided by CysO and cleavage of cysteine results in the release of carboxylated CysO [3]. For the regeneration of CysO thiocarboxylate, it has been suggested that MoeZR first activates CysO by acyl-adenylation of the C-terminus and then the thiocarboxylate group is formed by involvement of a sulfane sulfur bound to the Cterminal rhodanese-like domain of MoeZR [2]. The sulfur source so far remains unknown. These studies were performed in vitro by using crude cell extracts, in vivo evidence for the involvement of MoeZR in CysO activation was not shown to date. It was only reported, that the expression of MoeZR is upregulated under the same conditions that produce upregulation of CysO and CysM expression, mainly conditions of oxidative stress [3]. Our studies performed in E. coli give direct evidence, that CysO specifically interacts with MoeZR and that a thiocarboxylate group is formed on CysO only in the presence of MoeZR. However, our studies also show that MoeZR has a dual role in Mycobacteria and is also    CysO, however, it is able to interact with both MoaD1 and MoaD2. Since the Mycobacterial MoeZR and MoeBR proteins showed a higher ability to complement the E. coli MoeB function when their own MoaD congeners were present. The main difference between the E. coli and the Mycobacterial MoeB-like proteins is that both Mycobacterial congeners contain a C-terminal rhodanese-like domain. Thus, a fusion of the rhodanese-like protein to the MoeB-domain was sustained during evolution in Mycobacteria. In contrast in E. coli, IscS was shown to be the primary sulfur donor for Moco biosynthesis, transferring the sulfur to MoaD [12]. Recently, a separate rhodanese-like protein was identified in E. coli, named YnjE, which was shown to act as a mediator between the sulfurtransfer from IscS to the MoaD/ MoeB complex in E. coli, making the interaction more specific [24]. Thus, in E. coli the sulfur transfer pathway is realized by several protein partners, and not by fusion proteins. The primary sulfur donor for Moco biosynthesis in Mycobacteria remains to be elucidated. However, IscS homologs were identified [25] (Table 3), which might act as the primary sulfur donor to the rhodanese-like domain, as shown for the human MOCS3 protein [26].
Purification and characterization of MoeBR and MoeZR showed, that both proteins had the same secondary structure. MoeBR and MoeZR were purified in an active form which showed comparable thiosulfate:cyanide sulfurtransferase activities, however, thiosulfate is most likely not the physiological sulfur source, since the K m values were unphysiological. The k cat values with thiosulfate were comparable to the ones identified for other MoeB-like proteins containing a C-terminal rhodanese-like domain, like the human yeast Uba4 [22].
We also planned to purify and characterize MoaX. MoaX is a fusion protein of MoaD and MoaE [14]. For the accurate function of MoaD, the C-terminal glycine has to be accessible for the adenylation and sulfurtransfer reaction by the MoeB-homolog and for the subsequent sulfurtransfer to cPMP in conjunction with MoaE. Williams et al. [14] showed that MoaX was able to complement both, an M. smegmatis moaD2 mutant and a moaE2 mutant, assuming that an active MPT synthase was produced. The assumption was that MoaX is cleaved in the cell and a separate and active MoaD protein is generated. Our studies show that MoaX was not able to complement the E. coli moeB mutant in conjunction with MoeBR or MoeZR, and in addition single E. coli moaD and moaE mutant strains were also not complemented by moaX (data not shown). Thus, in E. coli the system for cleavage of MoaX seems to be missing. Unfortunately, we did not succeed to express and purify MoaX for further characterization, since the majority of the protein was expressed in inclusion bodies (data not shown). So far, the role of MoaX for Mycobacteria remains unclear. Especially since only M. tuberculosis and M. bovis species have acquired a MoaX homolog. It also remains possible that MoaX has a role in the cell apart from Moco biosynthesis, however, this is only a speculation.
It is not clear yet, why different Mycobacteria strains contain several homologs of genes for Moco biosynthesis. The moaA1-moaB1-moaC1-moaD1 cluster and moeB2 (coding for MoeBR) were acquired by horizontal gene transfer in M. tuberculosis and M. bovis [14]. Also, the moaA3-moaB3-moaC3-moaX gene cluster was acquired by horizontal gene transfer in these strains. Williams et al. [14] showed that the moaA1-moaD1 cluster is dispensable    (Fig. 6). The role of MoeBR seems to be restricted to Moco biosynthesis and most likely it preferentially interacts with MoaD1 in the cell, since both genes have been aquired simultaneously by horizontal gene transfer. However, both MoeBR and MoeZR retained their ability to interact with both MoaD1 and MoaD2 (Fig. 6). How both processes, the biosynthesis of Moco and cysteine are regulated in the cell, has to be determined in future studies. Since L-cysteine is the likely sulfur donor for Moco biosynthesis and an L-cysteine desulfurase acts as sulfur donor to MoeZR, MoeZR might be the link for both biosynthetic pathways and could be the main switch for feedback regulation.

Bacterial Strains, Plasmids, Media, and Growth Conditions
The bacterial strains and plasmids used in this work are listed in Table 4. E. coli BL21(DE3) cells were used for expression of the Mycobacterial proteins. Bacterial cultures were generally grown in LB medium under aerobic conditions at 30uC. When required, 150 mg/mL ampicillin or 50 mg/mL chloramphenicol was added to the medium.

Protein Expression and purification
The clones for the expression of MoeBR and MoeZR were obtained from the EMBL in Hamburg. For coexpression of M. tuberculosis MoeZR or MoeBR, and M. bovis MoaD1, MoaD2, MoaE2, or CysO, the corresponding plasmids (Table 4)    Functional complementation of an E. coli moeB mutant strain For functional complementation, the E. coli moeB 2 strain was transformed with the corresponding expression plasmids for E. coli MoeB (pMW15eB) and MoaD (pMW15aD), control, and the Mycobacterium tuberculosis MoeBR (pMal_MtMoeBR), MoeZR (pSM2) alone or for coexpression with M. bovis His 6 -MoaD1 (pCO10), His 6 -MoaD2 (pCO13), His 6 -CysO (pCO16), His 6 -MoaX (pSM1). For quantitative determination of nitrate reductase activity, the transformed E. coli cells were grown aerobically at 30uC in 5 ml of LB medium containing 15 mM nitrate. Protein expression was induced with 20 mM IPTG. Nitrate reductase activity in crude cell extracts was determined by a spectroscopic assay using benzyl viologen as described by Jones and Garland [27]. Nitrate reductase activity is related to the amount of cells measured at an O.D. at 600 nm. One unit of nitrate reductase activity is described as the production of 1 mmol of nitrite per min per O.D. 600.

Thiosulfate:cyanide sulfurtransferase activity
Thiosulfate:cyanide sulfurtransferase activities of MoeZR and MoeBR were measured by the classic colorimetric method after Sörbo [21], which is based on the absorbance of the complex formed between ferric ion and thiocyanate at 460 nm. Reaction mixtures in 100 mM Tris-acetate (pH 8.6), contained varying concentrations of sodium thiosulfate (0.5-60 mM) and varying concentrations of potassium cyanide (0.1-40 mM) in a volume of 0.5 mL. Reactions were started by the addition of the enzyme in a range of 0.5-1 mM, depending on the MoeB-like variant used. After an incubation time of 0.5-1 min at 25uC, formaldehyde (15%, 250 mL) was added to quench the reaction. Color was developed by the addition of 750 mL of ferric nitrate reagent [100 g of Fe(NO 3 ) 3 69H 2 O and 200 mL of 65% HNO 3 per 1500 mL]. After a further incubation for 10 min, thiocyanate (complexed with iron) was quantified at 460 nm using e = 4200 M 21 6cm 21 .

Electrospray Ionization MS
Aliquots (1-3 mL) of MoaD1, MoaD2, CysO purified from extracts coexpressed with MoeBR or MoeZR dissolved in 5 mM NH 4 OAc buffer were diluted 1:1 with methanol followed by addition of 1-10% formic acid (final protein concentration 1-10 pmol/mL) and applied to gold-coated nanospray glass capillaries, which were placed orthogonally in front of the entrance hole of a Q-TOFmicro instrument (Micromass, Manchester, U.K.). A voltage of approximately 1000 V was applied to the capillary, and ions were separated by the time-of-flight analyzer of the mass spectrometer. Protein spectra were deconvoluted using the MaxEnt1 software package (Micromass, Manchester, UK).

CD spectroscopy
Far-UV CD spectra of 0.1 mg/mL enzyme samples were recorded in 100 mM Tris, 200 mM NaCl, pH 7.4 using a Jasco J-715 CD-spectrophotometer. The scanning mode was set step-wise, each nm a data pitch was recorded, the response time was 4 seconds and each measurement was repeated 3 times.