Shared function and moonlighting proteins in molybdenum cofactor biosynthesis

Introduction

The transition element molybdenum is the only 4d series element of the periodic table with a biological role identified in bacteria, archaea and eukaryotes (Hille et al., 2014). The biologically active form is molybdate, which enters the cell by active transport systems (Self et al., 2001). As part of the catalytic active center in enzymes, molybdate is coordinated to a dithiolene group of a tricyclic pyranopterin monophosphate named molybdopterin (MPT), thereby forming the molybdenum cofactor (Moco) (Johnson and Rajagopalan, 1982). Only in nitrogen fixing bacteria can molybdate be inserted into a FeS cluster based iron-molybdenum cofactor (FeMoco), which forms the active site of nitrogenase (Ribbe et al., 2014).

Moco is the essential component of a large group of redox enzymes, which catalyze diverse reactions (Hille, 1996). More than 60 different Moco-containing enzymes were identified in bacteria, archaea, plants and animals to date (Hille et al., 2014). The physiological roles of molybdoenzymes are fundamental to most organisms since they are often involved in essential reactions (Zhang and Gladyshev, 2008; Madsen, 2011; Carpenter et al., 2012). Oxo-transfer in addition to hydroxylation reactions can be catalyzed by molybdoenzymes using a large diversity of organic (e.g. dimethylsulfoxide (DMSO), dimethylsulfide (DMS), formate, xanthine, ethylbenzene) and inorganic substrates (e.g. nitrate, selenate, chlorate, sulfite, arsenite). Some of the better-known Moco-containing enzymes include xanthine dehydrogenase in humans, assimilatory nitrate reductase in plants, and formate dehydrogenase in bacteria and archaea.

The Moco-containing enzymes are categorized on the basis of the structures of their molybdenum centers, dividing them into three families: the xanthine oxidase (XO) family, the sulfite oxidase (SO) family and the dimethyl sulfoxide (DMSO) reductase family (Hille, 1996) (Figure 1). The XO family is characterized by an MPT-Mo^VIO/S(O⁻) core in the oxidized state, with one equivalent of the pterin cofactor (designated MPT) coordinated to the metal. Enzymes of the SO family coordinate a single equivalent of the pterin cofactor with an MPT-Mo^VIO/O(Cys) core in its oxidized state, with the cysteine ligand being provided by the enzyme (Hille, 1996). The DMSO reductase family is exclusively present in prokaryotes and has a more diverse ligandation environment than the cofactor of the other families. All enzymes of this family have two equivalents of molybdopterin guanine dinucleotide (MGD), a variant of MPT containing an additional GMP group attached to the phosphate group, as the organic compound of their Moco (Hille, 1996). The molybdenum coordination sphere is usually completed by a Mo=O or Mo=S group with a sixth ligand in the MPT₂-Mo^VIO/S(X) core. The sixth ligand, X, can be a serine, a cysteine, a selenocysteine, an aspartate or a hydroxide and/or water molecule. The Y-ligand shown in Figure 1 can be either a sulfido or an oxo-ligand.

Figure 1:

General scheme for the biosynthesis of Moco.

A scheme of the biosynthetic pathway of Moco biosynthesis. The upper part shows the three conserved steps of Moco biosynthesis present in all organisms, the formation of cPMP, MPT and Mo-Moco. In bacteria, Mo-MPT is further modified by the addition of the nucleotides GMP or CMP, forming either the bis-MGD or the MCD forms of the cofactor. Enzymes of the DMSO reductase family are present exclusively in prokaryotes and bind the bis-MGD cofactor. Here, the molybdenum atom contains a ligand Y, which is either an oxo or a sulfido group, and a ligand X which can be a cysteine, a selenocysteine, a serine, an aspartate or a hydroxo-ligand. Moco can be further modified by the replacement of one oxo-ligand by a sulfido-ligand, forming the sulfurated Moco present in enzymes of the XO family in bacteria and eukaryotes. Enzymes of the SO family are present in bacteria and eukaryotes and bind the Mo-MPT cofactor with a proteinogenic cysteine ligand.

The chemical nature of Moco has been determined by Rajagopalan and coworkers in 1982 (Johnson and Rajagopalan, 1982). In all organisms including humans, Moco is synthesized by a conserved biosynthetic pathway that can be divided into three steps, according to the stable biosynthetic intermediates which can be isolated (Figure 1) (Rajagopalan, 1996): the synthesis of cyclic pyranopterin monophosphate (cPMP) (Wuebbens and Rajagopalan, 1993), conversion of cPMP into MPT by introduction of two sulfur atoms (Pitterle et al., 1993), and insertion of molybdate to form Mo-MPT (Joshi et al., 1996). In bacteria, a fourth step may be present that includes the further modification of Mo-MPT by the addition of nucleotide monophosphates to the phosphate group of MPT, forming the bis-molybdopterin guanine dinucleotide (bis-MGD) or molybdopterin cytosine dinucleotide (MCD) cofactors (Neumann et al., 2011; Reschke et al., 2013).

Moco biosynthesis has been intensively and best studied in the bacterium Escherichia coli. Here, a total of nine proteins were identified with a specific and essential function in Moco biosynthesis (Shanmugam et al., 1992; Rajagopalan, 1996). These proteins are encoded by genes that are organized in operon structures in E. coli: moaABCDE, mobAB, mocA, moeAB and mogA. However for two genes, namely moaB and mobB, an involvement in Moco biosynthesis has not been proven to date. Additionally, two more proteins are involved in Moco biosynthesis, for which also a role in other cellular pathways has been assigned: IscS and TusA (Zhang et al., 2010; Dahl et al., 2013; Leimkühler et al., 2017) (Table 1). IscS is the house keeping l-cysteine desulfurase mobilizing the sulfur from l-cysteine for FeS cluster biosynthesis, thionucleosides in tRNA, biotin, thiamin, and lipoic acid (Hidese et al., 2011). TusA, in contrast, has been originally identified as essential protein for the biosynthesis of 5-methylaminomethyl-2-thiouridine (mnm⁵s²U34) modified nucleosides in tRNAs for Glu, Gln and Lys (Ikeuchi et al., 2006). Thus, the biosynthesis of Moco shows a link to the assembly of FeS clusters and the synthesis of thio-modified tRNAs by sharing protein components for sulfur mobilization (Leimkühler et al., 2017).

Mutations in the human Moco biosynthetic genes lead to death in early childhood (Schwarz, 2016). The affected patients show severe neurological abnormalities, such as attenuated growth of the brain, seizures, and dislocated ocular lenses (Johnson and Duran, 2001). Moco is present in five enzymes in humans: sulfite oxidase (SO), xanthine oxidoreductase (XOR), aldehyde oxidase (AOX1) and two mitochondrial amidoxime reducing components, mARC1 and mARC2 (Hille et al., 2011; Mendel and Kruse, 2012). SO is the only one of the five molybdoenzymes that is essential for humans (Duran et al., 1978, 1996).

For Moco biosynthesis in humans, six proteins were identified with an essential role for all molybdoenzymes in humans: MOCS1A, MOCS1B, MOCS2A, MOCS2B, MOCS3 and gephyrin (Reiss and Johnson, 2003; Mendel and Kruse, 2012) (Table 1). Genes and the encoded proteins were named in humans as MOCS (molybdenum cofactor synthesis) (Reiss et al., 1998b; Reiss, 2000). Additionally, at least two further proteins are involved in Moco biosynthesis: the l-cysteine desulfurase NFS1 (Marelja et al., 2013) and the human Moco sulfurase HMCS (Peretz et al., 2007). In comparison to the components involved in Moco biosynthesis identified in bacteria, homologous proteins in humans are present, however, these consist either of two fused domains or alternatively fusions at the gene level are present, which by alternative splicing produce the active protein subunits. These include bicistronic transcripts of the MOCS1 and MOCS2 loci, which only after alternative splicing result in proteins for MOCS1A, MOCS2A, MOCS2B and a MOCS1A-MOCS1B fusion protein (Mendel and Leimkühler, 2015). Recently, MOCS2B was shown to interact not only with MOCS2A to form an active MPT synthase, but also with the Ada Two A containing (ATAC) acetyltransferase complex and with the double-stranded RNA-dependent protein kinase R (PKR) resulting in altered protein phosphorylation response (Suganuma et al., 2016). MOCS2B, thus, has a second role apart from Moco biosynthesis in humans, classifying it as a moonlighting protein. Further, MOCS3 is a fusion protein of a N-terminal MoeB-domain with homologies to ubiquitin-activating enzymes and a C-terminal rhodanese-like domain (RLD) with a catalytic sulfur-transfer activity. MOCS3 was shown to be a shared-function protein involved in both Moco biosynthesis and thiolation of 5methoxycarbonylmethyl2thiouridine (mcm⁵s²U) modified nucleosides in cytosolic tRNAs for Gln, Glu and Lys (Chowdhury et al., 2012). Further, the protein gephyrin is composed of two protein domains in humans, with a N-terminal MogA-like domain (G-domain) and a C-terminal MoeA-like domain (E-domain), which are connected by a long linker region (Tyagarajan and Fritschy, 2014). For gephyrin a moonlighting role apart from Moco biosynthesis has been described in neuronal cells, in which gephyrin is involved in glycin receptor clustering. HMCS, in addition, is a two-domain protein exclusively present in eukaryotes, which is involved in Moco sulfuration of XOR and AOX1 by providing the sulfur from its N-terminal l-cysteine desulfurase domain and transferring it to the Moco bound to the binding C-terminal domain (Peretz et al., 2007). Last but not least, NFS1 is a l-cysteine desulfurase with a main cellular localization in mitochondria where its primary role is dedicated to FeS cluster assembly (Biederbick et al., 2006). However, NFS1 was also identified to be localized in the cytosol where it interacts with MOCS3 and thereby provides the sulfur both for Moco biosynthesis and tRNA thiolation (Marelja et al., 2013) (Table 1).

In this review, Moco biosynthesis will be summarized and the shared function and moonlighting roles of proteins involved in Moco biosynthesis will be highlighted and their role in each pathway explained in detail.

The conversion of 5′GTP to cPMP

The first link between the Moco biosynthesis and FeS cluster assembly is realized in the first step of Moco biosynthesis, in which the FeS cluster containing protein MoaA is involved. In this step, the conversion of 5′GTP into cPMP proceeds through a complex rearrangement reaction, where the C8 of guanine is inserted between the C2′ and C3′ atoms of the ribose (Wuebbens and Rajagopalan, 1995; Hover et al., 2013; Mehta et al., 2013) (Figure 2). The synthesis of cPMP is catalyzed by the two enzymes MoaA and MoaC in bacteria and the reaction has been studied in detail using the bacterial enzymes from Staphylococcus aureus (Hover et al., 2013; Mehta et al., 2013). MoaA harbors two [4Fe4S] clusters (Hänzelmann and Schindelin, 2004, 2006), and belongs to the radical S-adenosyl methionine (SAM) superfamily (Sofia et al., 2001; Hänzelmann et al., 2004). Radical SAM enzymes catalyze a reaction in which SAM is reductively cleaved by an electron provided from the [4Fe4S] cluster bound at the N-terminus of the enzymes, to generate a 5′-deoxyadenosyl radical (5′-dA•) (Frey et al., 2008). This organic radical species generated on MoaA abstracts a H-atom from the substrate 5′GTP, bound to the C-terminal [4Fe4S] cluster of MoaA. In this reaction, a (8S)-3′, 8-cyclo-7,8-dihydroguanosine 5′triphosphate (3′,8-cH₂GTP) intermediate is formed (Figure 2). This intermediate is then converted by MoaC to cPMP, a reaction which involves pyrophosphate cleavage in addition to the formation of the cyclic phosphate group (Hover et al., 2013, 2015). The crystal structures of MoaC from E. coli and Thermus thermophilus showed that the protein forms a hexamer assembled as a trimer of dimers (Wuebbens et al., 2000; Kanaujia et al., 2010).

Figure 2:

Synthesis of cPMP from 5′GTP.

The conversion of 5′GTP to cPMP is catalyzed by MoaA (PDB: 1TV8) and MoaC (PDB: 1EKR) in bacteria and MOCS1A and MOCS1B in eukaryotes. All carbon atoms of the 5′GTP are found within cPMP. The C8 atom from the 5′GTP is inserted between the C2′ and C3′ atoms of the ribose, forming the 3′,8-cH₂GTP intermediate. This reaction is catalyzed by the MoaA or MOCS1A proteins, which are S-adenosylmethionine(SAM)-dependent enzymes. In the reaction catalyzed by MoaC or MOCS1B, the pyrophosphate group is cleaved, forming cPMP. In eukaryotes this step is localized in mitochondria. The [4Fe4S] clusters present on MoaA or MOCS1A are shown schematically.

In humans, the MOCS1 locus encodes two proteins, MOCS1A and MOCS1B, which are homologous to the bacterial MoaA and MoaC proteins, respectively (Reiss et al., 1998a). An unusual bicistronic transcript of the MOCS1 locus was identified, with open reading frames for both MOCS1A and MOCS1B in a single transcript, separated by a stop codon (Reiss et al., 1998b; Gross-Hard and Reiss, 2002). Splice variants of the MOCS1 locus were identified that either express the MOCS1A protein separately or a MOCS1A-MOCS1B fusion protein. The splice variant designated as type I expresses MOCS1A from exons 1-9, while type II and type III variants contain a truncation of exon 9 and express the resulting MOCS1A-MOCS1B fusion protein from exons 1-8 and exon 10 (Gray and Nicholls, 2000; Gross-Hard and Reiss, 2002). No evidence was found for the expression of a separated MOCS1B protein from the bicistronic MOCS1A-MOCS1B splice-type I cDNA, indicating that MOCS1B is only expressed as a fusion with MOCS1A, while MOCS1A, in contrast, is expressed as a separate protein (Hänzelmann et al., 2002). Both transcripts were shown to contain a mitochondrial targeting sequence for the translocation to mitochondria.

The presence of two FeS clusters on MOCS1A are directly linking the first step of Moco biosynthesis to FeS cluster biosynthesis in mitochondria. In eukaryotes, the mitochondria constitute the main compartment for the biosynthesis of FeS clusters. In humans, the main proteins required for FeS cluster biosynthesis are NFS1, ISD11, ISCU and frataxin, which form the quaternary core complex (Tsai and Barondeau, 2010; Bridwell-Rabb et al., 2011; Schmucker et al., 2011). ISD11 thereby is exclusively present in eukaryotes and was described as a stabilizing factor for NFS1, being essential for the activity in FeS cluster formation in mitochondria (Lim et al., 2013). After the synthesis of the FeS cluster on ISCU, the cluster is transferred by the help of carrier proteins to mitochondrial acceptor proteins like MOCS1A (Uhrigshardt et al., 2010).

After its synthesis in mitochondria, cPMP is transported to the cytosol (Teschner et al., 2010), where all further modification reactions occur. In the past, cPMP has been proposed to pass through the mitochondrial membrane without specific transport proteins, due to its hydrophobic nature (Wuebbens and Rajagopalan, 1993). However, a transporter involved in exporting cPMP to the cytosol was proposed in plants to be ATM3 (a homolog of human ABCB7) (Teschner et al., 2010). Surprisingly, ABCB7 in humans is also suggested to transport an essential molecule for FeS cluster biosynthesis to the cytosol (Hausmann et al., 2008). It has been proposed for yeast and mammals, that the mitochondrial transporter Atm1 (ABCB7 in mammals) exports either fully formed FeS clusters or a special form of sulfur that is required for cytosolic FeS cluster (CIA) biosynthesis (Lill et al., 2006; Hausmann et al., 2008). This would present essential functions of the mitochondria in both cytosolic FeS cluster and Moco biosynthesis. However, it still remains unclear how cPMP is specifically transported to the cytosol in humans and which specific compound essential for FeS cluster biosynthesis ABCB7 transports to the cytosol.

Synthesis of MPT from cPMP

In the next step of Moco biosynthesis, cPMP is converted to MPT in a reaction catalyzed by MPT synthase (Pitterle and Rajagopalan, 1989, 1991, 1993; Pitterle et al., 1990, 1993). In this reaction two sulfur atoms are inserted to the C1′ and C2′ positions of cPMP (Daniels et al., 2008). MPT synthase forms an (αβ)₂ heterotetramer composed of two MoaD and two MoaE subunits in bacteria (Rudolph et al., 2001). The sulfur atoms required for this reaction are present at the C-terminus of MoaD in form of a thiocarboxylate group (Gutzke et al., 2001; Leimkühler et al., 2003). The X-ray structures of the E. coli and S. aureus MPT synthases showed that the central dimer is formed by two MoaE subunits containing one MoaD at each end (Rudolph et al., 2001; Daniels et al., 2008). The MoaE subunit was shown to bind the cPMP molecule (Wuebbens and Rajagopalan, 2003). Each of the formed MoaE-MoaD dimers thereby act independently during the reaction. Studies of the reaction mechanism showed that the first sulfur is added at the C2′ position of cPMP by the thiocarboxylate group of one MoaD molecule, resulting in a hemisulfurated cyclic thiopyranopterin intermediate as product (Figure 3). In the next reaction, the cyclic phosphate group of the intermediate is hydrolyzed (Daniels et al., 2008). After transfer of the first thiocarboxylate sulfur to cPMP by one MoaD subunit, a new thiocarboxylated MoaD molecule is bound to MoaE (Wuebbens and Rajagopalan, 2003; Daniels et al., 2008). The second MoaD-COSH then attacks the C1′ atom of the hemisulfurated intermediate and MPT is formed (Figure 3).

Figure 3:

The biosynthesis of MPT from cPMP.

In the MPT synthase reaction, MoaD or MOCS2A bind in their thiocarboxylated forms to MoaE or MOCS2B, respectively, forming active MPT synthases (PDB: 2QIE for E. coli MoaD-MoaE) that convert cPMP to MPT in a two-step process with a hemisulfurated MPT intermediate. After transfer of the first thiocarboxylate sulfur to cPMP by one small MPT synthase subunit, a new thiocarboxylated MoaD/MOCS2A molecule is bound to MoaE or MOCS2B. The regeneration of the small MPT synthase subunit is catalyzed by MoeB (PDB: 1JW9) in bacteria or MOCS3 in eukaryotes. MOCS2B was additionally identified to have a moonlighting function in eukaryotes (summarized in the gray box). The model shows the interaction of MOCS2B with MBIP of the ATAC complex. Interaction of PKR supresses its autophosphorylation and the downstream phosphorylation of JNK and elF2a, thereby regulating both transcription and translation initiation.

In humans, the MOCS2 locus encodes the two subunits of MPT synthase, with MOCS2A being the MoaD homolog and MOCS2B being the MoaE homolog. In analogy to the MOCS1 locus, MOCS2A and MOCS2B are transcribed leading to a bicistronic transcript, however, with overlapping reading frames (Stallmeyer et al., 1999a). Since the C-terminus of MOCS2A is indispensable for the MPT synthase reaction, MOCS2A and MOCS2B are translated independently from the bicistronic transcript (Stallmeyer et al., 1999a). The expression of the overlapping MOCS2A and MOCS2B genes thereby was shown to be realized by two alternative MOCS2 splice forms, named I and III. The alternative splicing results in two different first exons, which lead to the alternative transcripts (Hahnewald et al., 2006). Studies on the human MPT synthase reaction showed that the mechanism of MPT formation is identical to the one studied for the E. coli proteins (Figure 3). For MPT synthase, active hybrid proteins with the E. coli large or small subunits of MPT synthase can be formed (Leimkühler et al., 2003).

A novel moonlighting role for MOCS2B

Recent studies revealed that human MOCS2B has a moonlighting role in the regulation of transcription and translation (Suganuma et al., 2016). The study showed that apart from its role in Moco biosynthesis, MOCS2B interacts with the ATAC histone acetyltransferase complex, a protein complex that consists of 13 subunits and serves as transcriptional co-activator (Figure 3). The human ATAC complex contains the mitogen activated protein kinase (MAPK) and the upstream kinase-binding inhibitory protein (MBIP). Phylogenetic analyses showed that in flies MBIP was evolutionary derived from an ancestral MOCS2B protein. Furthermore, in Drosophila it was shown that the MOCS2B homolog dMoaE forms a fusion protein with MBIP, revealing a direct coupling of Moco biosynthesis and transcriptional regulation in flies (Suganuma et al., 2012). In contrast, in Metazoa, the MBIP protein may have been derived from an ancestral MOCS2B protein, but has lost its catalytic activity in Moco biosynthesis. In humans, however, MOCS2B was shown to interact with the MBIP protein of the ATAC complex, and by this interaction the phosphorylation PKR was decreased, the double-stranded RNA-dependent protein kinase R (Suganuma et al., 2016). The decreased PKR phosphorylation further suppressed the phosphorylation of c-Jun N-terminal kinase (JNK) and of the eukaryotic translation initiation factor 2a (elF2a). The inhibited activity of PRK in total resulted in an enhanced concentration of the ternary complex required for translation initiation, thereby enhancing the translation initiation of iron responsive mRNA. This negative regulation of PRK through MOCS2B in the ATAC complex was revealed to be crucial for the translation of ferritin, the cytosolic iron storage protein involved in cellular iron homeostasis. Thus, by the interaction of MBIP and MOCS2B transcription and translation initiation of proteins for iron homeostasis are controlled. Moco biosynthesis and iron homeostasis therefore seem to be linked by a non-catalytic moonlighting function of MOCS2B. When MOCS2B is available in the cell and not in complex with MOCS2A, iron responsive translation pathways are initiated by the interaction of MOCS2B with the MBIP protein of the ATAC acteyltransferase complex. This link might be advantageous to the cell, since Moco biosynthesis requires FeS clusters for the activity of MOCS1A. By the enhanced translation of ferritin, MOCS2B might control the cellular iron level of the cell and thereby ensures that enough FeS clusters are assembled to form an active MOCS1A protein. This novel moonlighting role of MOCS2B reveals a direct link of the iron metabolism and Moco biosynthesis in the cell.

Sulfur mobilization involves sharing of protein functions in prokaryotes

After MPT synthesis, the regeneration of sulfur at the C-terminal glycine of MoaD is catalyzed by MoeB in bacteria (Leimkühler et al., 2001; Schindelin, 2005). MoaD thereby dissociates from MoaE and forms a heterodimeric complex with MoeB. The X-ray structure of the (MoaD-MoeB)₂ complex showed that the C-terminus of MoaD is activated by MoeB under ATP consumption, leading to the formation of an acyl-adenylate group at the terminal glycine of MoaD (Lake et al., 2001; Leimkühler et al., 2001; Schmitz et al., 2007) (Figure 4). In a second step, sulfur is directly transferred from a sulfur transferase to MoaD-AMP in the (MoaD-MoeB)₂ complex, releasing (MoeB)₂, AMP and MoaD thiocarboxylate (MoaD-COSH). After the dissociation of MoaD-COSH from the complex, MoaD reassociates with MoaE to form the active MPT synthase heterotetramer (Tong et al., 2005; Schmitz et al., 2007). In the sulfur transfer reactions to MoaD, the proteins IscS and TusA were shown to be involved in E. coli, forming a sulfur-relay system in which a persulfide sulfur is formed on a conserved cysteine residue of these proteins as intermediate (Leimkühler and Rajagopalan, 2001; Zhang et al., 2010) (Figure 4). It is believed that the persulfide sulfur of TusA-SSH acts as sulfur donor for the (MoaD-MoeB)₂ complex by attacking the MoaD-AMP bond, releasing AMP and creating a transient MoaD perthiocarboxylate intermediate, which further is reductively cleaved (Dahl et al., 2013; Iobbi-Nivol and Leimkühler, 2013).

Figure 4:

Sulfur-transfer network for the formation of active MPT synthase.

For the formation of the thiocarboxylate group on MoaD in baceria, MoaD assembles with MoeB (PDB: 1JW9). In this complex, adenylated MoaD is formed by attachment of an AMP moiety by the MoeB protein. MoaD-AMP is then sulfurated by the persulfide group formed on the TusA protein. The sulfur for TusA-SSH formation is derived from the l-cysteine desulfurase IscS (PDB: 3LVJ). IscS (PDB: 3LVM) was shown to additionally interact with IscU for FeS cluster formation or with ThiI for thiamine or s⁴U formation in tRNA. After the sulfur transfer reaction, MoaD-COSH dissociates from the MoeB-dimer to reassociate with MoaE. In humans, the sulfur is transferred from cytosolic NFS1 via a cysteine on MOCS3-RLD further onto MOCS2A. NFS1 was shown to be localized in the cytosol for Moco biosynthesis and additionally in mitochondria, where its primary role is assigned to the assembly of FeS clusters. For FeS cluster assembly, a quaternary complex is formed, consisting of NFS1, ISD11, ISCU and frataxin. The vertical dashed line implies that the sulfur transfer networks in bacteria and humans use different protein components.

In E. coli, the sulfur transferase TusA for Moco biosynthesis thereby is a shared function protein, since it was originally shown to be the sulfur delivery protein for (c)mnm⁵s²U34 thio-modifications in nucleosides of tRNAs for Glu, Gln and Lys (Figure 4). For the formation of (c)mnm⁵s²U, a sulfur-relay system was identified involving the initial sulfur mobilization by the l-cysteine desulfurase IscS and the proteins TusA, TusBCD, TusE and MnmA (Ikeuchi et al., 2006). TusA thereby directly interacts with IscS, stimulates its desulfurase activity and directs the sulfur flow to 2thiouridine formation. After the formation of a persulfide on Cys19 of TusA, TusA transfers the sulfur onto Cys 278 of TusD, which forms a complex with TusB and TusC. Further, TusE interacts with TusBCD, accepts the sulfur from TusD, and transfers the sulfur to Cys199 of MnmA. MnmA binds tRNA and ATP and activates the bound tRNA by forming an activated acyladenylated intermediate on U34 (Numata et al., 2006b). Subsequently, a nucleophilic attack by the persulfide sulfur of Cys199 of MnmA generates a tRNA thiocarbonyl group and releases AMP (Kambampati and Lauhon, 2003; Numata et al., 2006a,b).

For both pathways, the l-cysteine desulfurase IscS catalyzes the formation of the persulfide group on TusA (Shi et al., 2010). E. coli IscS was shown to be the primary sulfur donor for numerous sulfur-containing molecules with important biological roles, including FeS clusters, thiamin, Moco and thionucleosides in tRNA (Leimkühler et al., 2017) (Figure 4). To perform its important role in multiple biological pathways, E. coli IscS was shown to interact with a number of different proteins (Shi et al., 2010), namely (i) IscU in complex with CyaY, Fdx and IscX for FeS cluster formation, (ii) TusA for either the (c)mnm⁵s²U34 modification of tRNA or the biosynthesis of Moco, or (iii) ThiI for either the synthesis of thiamine or the s⁴U8 modification of tRNA (Hidese et al., 2011). Different binding sites for some of these molecules were mapped on E. coli IscS (Shi et al., 2010). Overall, all sulfur-containing molecules in E. coli depend on the availability of IscS. Therefore, the sulfur transfer pathways to sulfur containing biomolecules like Moco biosynthesis, FeS cluster assembly and thiolation of tRNAs are tightly connected by sharing the main sulfur-providing protein in the cell. How the sulfur transfer to each sulfur-containing molecule is specifically channeled is not completely understood to date.

Eukaryotic MOCS3 is a two-domain protein with multiple interaction partners

MOCS3 is the homolog of MoeB in humans (Matthies et al., 2004). Sequence alignments of MOCS3 revealed that it is a two-domain protein with an N-terminal MoeB-like domain, and an additional C-terminal domain showing homologies to rhodaneses (MOCS3-RLD) (Bordo and Bork, 2002; Matthies et al., 2004). MOCS3-RLD was shown to act as direct sulfur donor for the formation of the thiocarboxylate group on MOCS2A (Matthies et al., 2004) (Figure 4). A persulfide group was identified on the highly conserved cysteine residue of the six amino acid active site loop of MOCS3-RLD (Matthies et al., 2005). In total, MOCS3 can be regarded as a multi-functional and multi-domain protein combining the adenylation of MOCS2A with the subsequent sulfur-transfer reaction. It was shown that the N-terminal MoeB-like domain of MOCS3 activates the C-terminus of MOCS2A by addition of AMP, in a similar manner as reported for the E. coli proteins MoeB and MoaD. In the second reaction, the MOCS2A acyl-adenylate is converted to a thiocarboxylate by sulfur transfer from a persulfide present at MOCS3-RLD, and in turn, thiocarboxylated MOCS2A is generated and released (Matthies et al., 2005; Mendel and Schwarz, 2011) (Figure 4). In analogy to the E. coli system, it was shown that the sulfur for MOCS3 originates from the l-cysteine desulfurase NFS1 in the cytosol in humans cells, transferring sulfur from l-cysteine in a sulfur relay system via a persulfide group formed on the l-cysteine desulfurase via MOCS3-RLD further onto MOCS2A (Krepinsky and Leimkühler, 2007; Marelja et al., 2008; 2013).

Crucially, MOCS3 is not only involved in Moco biosynthesis but also in the formation of thio-modified mcm⁵s²U34 nucleosides in tRNA for Gln, Glu and Lys in the cytosol of human cells (Chowdhury et al., 2012). To perform this shared function role in tRNA thiolation, MOCS3 interacts with the ubiquitin-related modifier protein1 (URM1) (Figure 4). In analogy to the MOCS2A protein, URM1 contains a conserved C-terminal double glycine-motif on which a thiocarboxylate group is formed for sulfur transfer for the mcm⁵s²U34 modification in tRNA (Xu et al., 2006; Pedrioli et al., 2008; Schmitz et al., 2008; Leidel et al., 2009). MOCS3 thereby activates URM1 in the presence of ATP by formation of activated URM1-AMP (Figure 4). URM1 is further transferred to the persulfide group formed on Cys412 of MOCS3-RLD via a disulfide bond. After cleavage of the disulfide bond, thiocarboxylated URM1 is released. The sulfur of URM1-COSH is further transferred to mcm⁵U34 group of the tRNA^Glu,Gln,Lys, aided by the CTU1 and CTU2 proteins under ATP consumption (Chowdhury et al., 2012).

In total, Moco biosynthesis and tRNA thiolation are directly connected in humans by sharing the common MOCS3 enzyme (Figure 4). Further, the sulfur delivery pathway to MOCS3 involves the l-cysteine desulfurase NFS1 in the cytosol (Marelja et al., 2013). In eukaryotes, however, NFS1 was described to be mainly localized in the mitochondria where it is involved in the assembly of FeS clusters (Biederbick et al., 2006) (Figure 4). A role of NFS1 for the CIA pathway of FeS cluster biosynthesis in the cytosol has not been described so far. Thus, the role of the cytosolic form of NFS1 seems to be exclusively restricted to Moco biosynthesis and thio-modifications of certain tRNAs.

Insertion of molybdate into MPT

After synthesis of the dithiolene moiety in MPT, molybdate can be inserted to form Mo-MPT. In this step, the gene products of moeA and mogA are involved in E. coli (Figure 5). The crystal structure of MogA showed the form of a homotrimer (Liu et al., 2000), while the crystal structure of MoeA showed a dimeric structure of two subunits, each containing four subdomains, with one of the subdomains being structurally related to MogA (Schrag et al., 2001; Xiang et al., 2001). It was revealed that both the E. coli MoeA and MogA proteins are essential for the incorporation of molybdenum to MPT in a concerted manner (Nichols and Rajagopalan, 2002). The reaction was shown to be catalyzed in a two-step process in which MogA first activates MPT in an ATP-dependent reaction and MoeA in the second step mediates molybdate ligation to the formed MPT-AMP (Figure 5). The molybdate insertion step thereby is coupled to a deadenylation step catalyzed by MoeA, finally resulting in Mo-MPT formation (Nichols and Rajagopalan, 2005). While in bacteria molybdate insertion is catalyzed by the two proteins MogA and MoeA, during evolution to higher organisms these two proteins were fused to a single two-domain protein [Cnx1 in plants (Stallmeyer et al., 1995) and gephyrin in mammals (Stallmeyer et al., 1999b)]. The two domains of these proteins were named G-domain (homologous to MogA) and E-domain (homologous to MoeA). The MPT-AMP intermediate (Figure 5) was first identified in the crystal structure of the plant Arabidopsis thaliana Cnx1 G-domain (Kuper et al., 2004). It was shown that the Cnx1 G-domain catalyzes this adenylation of MPT in a reaction dependent on Mg²⁺ and ATP (Llamas et al., 2004, 2006). The Cnx1 E-domain was shown to be required for the subsequent Mg²⁺ dependent molybdenum-insertion. Obviously the coupling of both functions into one protein was beneficial for a protected product-substrate channeling of the fragile intermediate MPT-AMP (Belaidi and Schwarz, 2013).

Figure 5:

Insertion of molybdate into MPT.

The bacterial MogA (PDB: 1DI6) and MoeA (PDB: 1G8I) proteins catalyze the specific incorporation of molybdenum into MPT in a multistep reaction with an adenylated MPT intermediate (MPT-AMP). While MogA forms the MPT-adenylate intermediate, MoeA mediates molybdenum ligation to MPT at low concentrations of MoO₄²⁻ in a Mg²⁺ dependent manner. The same reaction is catalyzed by gephyrin in humans (PDB: 4PD0 for gephyrin-E and 1JLJ for gephyrin-G). For gephyrin, a moonlighting function was described in neuronal cells (gray box), being involved in neuroreceptor clustering (here the interaction with the GABA_A receptor is shown). Gephyrin is believed to form higher oligomers at synaptic sites by homotrimerization of its G-domain and homodimerization of its E-domain. Gephyrin also presents a link to the to the cytoskeleton by direct interaction with profilin and Mena/VASP.

In humans, molybdate insertion is catalyzed by the two-domain protein gephyrin and follows the same mechanism as described for plants (Stallmeyer et al., 1999b; Belaidi and Schwarz, 2013) (Figure 5). Recent crystallographic analyzes of the gephyrin-E domain in the apo state and in complex with ADP, AMP and molybdate provided insights into the catalytic step of deadenylation (Kasaragod and Schindelin, 2016). The crystal structures revealed an adenine-binding region on the E-domain that recognizes the AMP moiety of the AMP-MPT substrate. Formation of the pyrophosphate linkage appeared to be coupled to significant conformational changes with in the protein, especially of an assigned nucleotide recognition loop. The metal was shown to be bound at another conserved site, in close proximity to where the dithiolene group of MPT was predicted to be localized (Kasaragod and Schindelin, 2016).

The moonlighting role of gephyrin in neuroreceptor clustering

Apart from Moco biosynthesis, gephyrin has an essential moonlighting function in the central nervous system, where it clusters inhibitory glycine and γ-amino butyric acid (GABA) type A receptors at postsynaptic synapses (Schwarz, 2005) (Figure 5). In addition to its interaction with the glycine receptor, gephyrin also interacts with the intracellular loop of the GABA_A receptor. Gephyrin thereby is a central player at inhibitory synapses, as it is the structural receptor scaffold and acts at a platform for facilitating protein–protein interactions, bringing receptors, cytoskeletal elements and signaling proteins into close proximity (Maric et al., 2014). Gephyrin thus regulates clustering and diffusion of these receptors of fast inhibitory transmission. The gephyrin protein network in particular also tethers the GABA_A receptors to the cytoskeleton by direct interaction with profilin and other proteins like Mena/VASP (Maric et al., 2014) (Figure 5). Gephyrin in total is comprized of three domains, the N-terminal G-domain, the C-terminal E-domain and a central domain referred to as linker. The linker connects the G- and E-domains and harbors numerous sites for post-translational modifications and interaction with proteins that regulate synapse formation and function (Tyagarajan and Fritschy, 2014). Gephyrin is believed to form higher oligomers at synaptic sites by homotrimerization of its G-domain and homotomerization of its E-domain. Further, gephyrin possesses a non-conserved surface-exposed loop in the E-domain, which regulates its postsynaptic clustering (Lardi-Studler et al., 2007). In addition to post-translational modifications, gephyrin is the subject of extensive alternative splicing. These modifications are regulating the role of gephyrin in Moco biosynthesis and as a postsynaptic scaffoldig molecule (Fritschy et al., 2008; Herweg and Schwarz, 2012). While the role of gephyrin in Moco biosynthesis is essential to humans for the activity of SO, it was shown that Moco biosynthesis in the central nervous system only takes place in astrocytes (Tyagarajan and Fritschy, 2014). Thus, it has been suggested that gephyrins main function in neurons is restricted to the regulation of inhibitory neurotransmission.

The attachment of nucleotides to Mo-MPT in bacteria

The majority of molybdoenzymes in bacteria like E. coli contain the bis-MGD cofactor (Hille et al., 2014). These molybdoenzymes are grouped into the DMSO reductase family and contain two dithiolene moieties from two MPT backbones ligated to the molybdenum atom. In bis-MGD, Moco is further modified by the addition of a GMP group to the phosphate of each MPT molecule. The synthesis of bis-MGD is catalyzed by MobA in a two-step reaction under GTP consumption (Reschke et al., 2013). In the first step, a bis-Mo-MPT intermediate is formed on MobA (Figure 6). This reaction depends on Mo-MPT formation by MoeA (Temple and Rajagopalan, 2000), however, no further molecules are required for the reaction. In the second step, GTP is hydrolyzed and GMP is added to the C4′ phosphate of each MPT molecule in bis-Mo-MPT, forming the bis-MGD cofactor (Palmer et al., 1996; Lake et al., 2000). The crystal structure of MobA showed that the protein is a monomer with an overall two-domain architecture, in which the N-terminal domain of the molecule adopts a nucleotide (GTP)-binding Rossmann fold, and a possible MPT binding site that is localized to the C-terminal half of the protein (Lake et al., 2000; Stevenson et al., 2000). After bis-MGD formation, the cofactor can be inserted into target molybdoenzymes by help of specific Moco-binding molecular chaperones.

Figure 6:

Formation of dinucleotide and sulfurated forms of Moco.

Mature Moco can be further modified in prokaryotes by the addition of GMP or CMP to the C4′ phosphate of MPT via a pyrophosphate bond. The bis-MGD cofactor is formed by the MobA protein (PDB: 1E5K), which specifically binds GTP, in a two-step reaction via the formation of a bis-Mo-MPT intermediate. Bis-MGD is present in enzymes like the E. coli TorA or NarGHI protein, which coordinate the molybdenum atom by an amino acid ligand X, which is either a serine or an aspartate. Bis-MGD can be further modified by sulfuration. In the reaction for bis-MGD sulfuration for enzymes like formate dehydrogenase FdhF in E. coli, the proteins FdhD (PDB: 2PW9) and IscS (PDB: 3LVJ) are involved. FdhD likely binds bis-MGD and IscS transfers the sulfur to the Mo atom by exchanging an oxo-group and adding the sulfido ligand. Afterwards, sulfurated bis-MGD is inserted into FdhF, which further cordinates bis-MGD by a selenocysteine ligand. Enzymes like the nitrate reductase NapAB contain a cysteine ligand instead. MCD is formed by the MocA protein, which acts specifically on CTP. For PaoABC, MCD is further modified in E. coli by exchange of the equatorial oxygen to a sulfido ligand, forming sulfurated MCD. This step is carried out by the PaoD protein in conjunction with IscS. After sulfuration, MCD is inserted by the aid of PaoD into PaoABC. Further, formed Mo-MPT can also directly be inserted into enzymes of the SO family, which is MsrP in E. coli or SO and mARC in humans. For enzymes of the XO family in humans, like XOR or AOX, the two-domain protein HMCS is involved, containing a N-terminal l-cysteine desulfurase domain (binding PLP) and a C-terminal MOCS-domain involved in Mo-MPT binding. The vertical dashed line implies that Moco sulfuration in bacteria and humans use different protein components.

For enzymes of the xanthine oxidase family in bacteria, Mo-MPT can be modified by the addition of a CMP nucleotide to form the molybdopterin cytosine dinucleotide cofactor (MCD) (Neumann et al., 2009; Iobbi-Nivol and Leimkühler, 2013) (Figure 6). This reaction is catalyzed by the MocA protein in E. coli, which shares a high level of amino acid sequence identity with MobA (Neumann et al., 2011). Both MobA and MocA were shown to be specific for the respective purine or pyrimide nucleotide. The reaction catalyzed by MocA thereby resembles the second part of the reaction catalyzed by MobA, acting as a MPT-CTP transferase by covalently linking CMP to Mo-MPT and releasing pyrophosphate (Neumann et al., 2009). However, in contrast to MobA, MCD is the end product and a bis-cofactor is not formed. Instead, the MCD cofactor for all enzymes of the XO family is further modified by the insertion of an equatorial sulfido ligand to the molybdenum atom that is essential for enzyme activity.

Moco sulfuration in bacteria

After the synthesis of the bis-MGD cofactor or the MCD cofactor in E. coli, these cofactors can be further modified by the addition of a terminal sulfido group as ligand to the molybdenum atom (Figure 6). While a MPT-Mo^VIOS(O^-) coordination is characteristic for enzymes of the XO family in pro-and eukaryotes, the presence of a sulfido group at the bis-MGD cofactor was recently revealed to be present in bacterial enzymes of the DMSO reductase family, like the E. coli formate dehydrogenase FdhF or the periplasmic nitrate reductase NapAB from Cupriavidus necator (Raaijmakers and Romao, 2006; Coelho et al., 2011; Thome et al., 2012; Hille et al., 2014). The reaction of Moco sulfuration is catalyzed by Moco bound to molecular chaperones in conjunction with a l-cysteine desulfurase that provides the sulfur atom for the reaction. For E. coli FdhF, the reaction of bis-MGD sulfuration was described to involve the chaperone FdhD and the l-cysteine desulfurase IscS (Thome et al., 2012). During this reaction, a persulfide sulfur is transferred from the l-cysteine desulfurase via a cysteine on the molecular chaperone to the bound bis-MGD cofactor (Figure 6). After this reaction, FdhD is expected to insert the modified bis-MGD into the target enzyme FdhF.

For Moco sulfuration of enzymes of the XO family, the reaction generally is catalyzed in a similar manner, involving a Moco-binding chaperone and the l-cysteine desulfurase IscS. Here, chaperones like PaoD for the periplasmic aldehyde oxidoreductase PaoABC in E. coli bind Moco (in case of PaoD the MCD cofactor is bound) and thereby protect it from oxidation (Otrelo-Cardoso et al., 2014). By a suggested interaction with the l-cysteine desulfurase IscS, the equatorial oxygen ligand of Mo-MPT is replaced by a sulfido ligand (Neumann et al., 2007) (Figure 6). After the Moco sulfuration reaction, the mature cofactor is inserted into the apo-target molybdoenzyme in a specific reaction, involving the molecular chaperone that recognizes its specific partner protein (Neumann et al., 2007). Since Moco is deeply buried in these prefolded proteins, these chaperones may additionally act in the final folding after Moco insertion. In total, the reaction of Moco sulfuration in bacteria reveals an involvement of the l-cysteine desulfurase IscS, thereby adding further interaction partners to the complex sulfur transfer network of IscS.

Moco sulfuration in eukaryotes

As a final step of Moco maturation among eukaryotes, the members of the XO family require the addition of a terminal inorganic sulfur ligand, which is catalyzed by a specific two-domain Moco sulfurase, e.g. ABA3 in plants or HMCS in humans (Figure 6). The best studied eukaryotic Moco sulfurase thereby is the ABA3 protein from A. thaliana. ABA3 is a homodimeric two-domain protein (Bittner et al., 2001) with its N-terminal domain sharing structural and functional homologies to the l-cysteine desulfurases SufS. The C-terminal domain was shown to bind Moco and belongs to the Moco sulfurase C-terminal (MOSC) domain superfamily of proteins (Anantharaman and Aravind, 2002) (Figure 6). This domain, however, shares no amino acid sequence homologies to the Moco-binding chaperones identified in bacteria. In a pyridoxal phosphate-dependent manner, the N-terminal domain of ABA3 decomposes l-cysteine to yield alanine and sulfane sulfur (Heidenreich et al., 2005), the latter being bound as a persulfide to a highly conserved cysteine residue of ABA3. After binding of Mo-MPT to the C-terminal domain of ABA3, Moco is sulfurated on this domain via an intramolecular persulfide relay from the N-terminal domain (Lehrke et al., 2012; Wollers et al., 2008). After Moco sulfuration, the matured cofactor is inserted into the target proteins AOX or XOR (Figure 6). This shows that for ABA3 and its orthologs, the l-cysteine desulfurase has been fused to a MOSC domain to evolve into a new activator protein highly specific and exclusive for AOX and XO. In bacteria, this reaction is catalyzed by two proteins, with the l-cysteine desulfurase being shared with other sulfur transfer pathways.

Outlook

In general, the biosynthesis of Moco is conserved in all organisms containing molybdoenzymes. However, elucidation of the details of the pathway for Moco biosynthesis in prokaryotes and eukaryotes has revealed that differences exist among the kingdoms of life. For example, in higher eukaryotes, individual Moco biosynthesis proteins perform moonlighting functions as a consequence of gene sharing. Moonlighting roles often are the consequence of alternative splicing events, thus, these dual protein functions seem to be restricted to eukaryotes. Further, protein roles with an involvement in Moco biosynthesis are shared among different pathways, in particular several pathways are connected by sharing the same sulfur transfer protein. Here, a link between the biosynthesis and maturation of molybdoenzymes, the assembly and distribution of FeS clusters and the formation of thio-modified tRNAs has been identified in the last years: (i) The synthesis of the first intermediate in Moco biosynthesis requires an FeS-cluster containing protein, (ii) the initial sulfur transferase for the dithiolene group in Moco is common also for the synthesis of FeS clusters, thiamin and thiolated tRNAs, (iii) the modification of the Mo first coordination sphere with a sulfur atom additionally involves a sulfur transferase, and (iv) iron homeostasis is controlled on the transcriptional and translational level by proteins involved in Moco biosynthesis in eukaryotes. Overall, by sharing protein components between different cellular pathways, these pathways are tightly connected to each other and provide a concerted level of regulation, so that e.g. Moco is only synthesized when FeS clusters are present being required for its synthesis.