The bacterial MrpORP is a novel Mrp/NBP35 protein involved in iron-sulfur biogenesis

Despite recent advances in understanding the biogenesis of iron-sulfur (Fe-S) proteins, most studies focused on aerobic bacteria as model organisms. Accordingly, multiple players have been proposed to participate in the Fe-S delivery step to apo-target proteins, but critical gaps exist in the knowledge of Fe-S proteins biogenesis in anaerobic organisms. Mrp/NBP35 ATP-binding proteins are a subclass of the soluble P-loop containing nucleoside triphosphate hydrolase superfamily (P-loop NTPase) known to bind and transfer Fe-S clusters in vitro. Here, we report investigations of a novel atypical two-domain Mrp/NBP35 ATP-binding protein named MrpORP associating a P-loop NTPase domain with a dinitrogenase iron-molybdenum cofactor biosynthesis domain (Di-Nase). Characterization of full length MrpORP, as well as of its two domains, showed that both domains bind Fe-S clusters. We provide in vitro evidence that the P-loop NTPase domain of the MrpORP can efficiently transfer its Fe-S cluster to apo-target proteins of the ORange Protein (ORP) complex, suggesting that this novel protein is involved in the maturation of these Fe-S proteins. Last, we showed for the first time, by fluorescence microscopy imaging a polar localization of a Mrp/NBP35 protein.


Results
The Mrp ORP proteins are two-domain proteins. Querying the INTERPRO portail indicated that Mrp ORP belongs to the Mrp/NBP35 ATP-binding proteins (IPR019591), a very large protein family ubiquitous in the domains of Life (Supplementary Table S1). This protein family encompasses the prokaryotic Mrp and ApbC, and the eukaryotic Nbp35 and Cfd1 proteins involved in Fe-S cluster biogenesis 12,16,24 . Most of the 17,511 members of the Mrp/NBP35 ATP-binding protein family contained a single conserved functional domain, the P-loop containing nucleoside triphosphate hydrolase domain (P-loop NTPase, IPR027417). However, in a few cases this domain is found associated with other domains (Supplementary Table S2). In Mrp ORP , the P-loop NTPase domain is associated with a dinitrogenase iron-molybdenum cofactor biosynthesis domain (Di-Nase, IPR003731), which is usualy found in proteins involved in the biosynthesis of the iron-molybdenum cofactor (FeMo-co), such as NifB and NafY 25 . The association between P-loop NTPase and Di-Nase domains was observed in 99 members of the Mrp/Nbp35 ATP-binding family (Supplementary Table S3). They corresponded mainly to proteins from anaerobic organisms, such as Thermodesulfobacteria, Clostridia and Desulfovibrio (Supplementary Table S3).
The phylogenetic analysis of the Mrp/NBP35 ATP-binding protein family showed that sequences from the three domains of Life are mixed on the tree (Fig. 1), indicating that horizontal gene transfers among domains occurred during the diversification of this protein family. According to this phylogeny, the Mrp ORP protein is more related to the eukaryotic Nbp35 and Cfd1 than to the bacterial ApbC and Mrp (Fig. 1, purple triangles). The 99 sequences harboring an association between the P-loop NTPase and Di-Nase domains do not form a monophyletic group (Fig. 1, pink triangles). In fact, they belong to different parts of the tree, indicating that the association between these two domains occurred several times independently. The phylogenetic analysis of the 110 sequences displaying the highest sequence similarity with the P-loop NTPase domain of Mrp ORP disclosed its closest relatives that are mainly from Deltaproteobacteria and revaled again phylogenetic relationships at odd with the current systematics confirming that the evolution of the Mrp/NBP35 ATP-binding family has been heavily impacted by horizontal gene transfers ( Supplementary Fig. S1). Again, sequences harboring both the P-loop NTPase and Di-Nase domains appeared intermixed with sequences containing only the P-loop NTPase domain, confirming that such an association occurred mainy times during evolution.
Genes coding for Mrp ORP proteins from the sulfate reducer deltaproteobacteria DvH and DdG20 were both located in the orp genes cluster ( Supplementary Fig. S2) 19,22 . The Mrp ORP proteins from DvH and DdG20 have a molecular mass of 50 kDa and 43 kDa, respectively, with a P-loop NTPase domain of 30 kDa and a Di-Nase domain of 13 kDa for Mrp ORP with DvH exhibiting a suplementary linker between the two domains (Fig. 2). The sequence alignment of the two Mrp ORP with E. coli Mrp, S. enterica ApbC, S. cerevisiae Nbp35 and Cfd1, showed that the typical deviant Mrp Walker A (GKGGhGK[ST]), Walker B motifs, and CXXC motifs are conserved in Mrp ORP (Fig. 2) 12,16,26 . We found also that in the Mrp ORP , four cysteine residues were present in the N-terminal part of the P-loop NTPase domain of Mrp ORP proteins (Fig. 2, blue crosses) with only one of them conserved in eukaryotic Nbp35 (Fig. 2, black asterisk). These cysteine residues might form a non-canonical motif: CX 3 CX 20 CXC in Mrp ORP of DvH and CXCX 5 CX 4 C in Mrp ORP of DdG20.
Yet, the association between a P-loop NTPase and a Di-Nase domains raises the question about the role of Mrp ORP proteins. We then investigated the biochemical properties of this new type of Mrp-like protein.
The conserved CXXC motif from the P-loop NTPase domain of Mrp ORP binds a Fe-S cluster. We, first, investigated the presence of a Fe-S cluster bound to Mrp ORP . The UV-visible spectrum of the aerobically isolated DdG20 Mrp ORP purified from E. coli, showed no absorbance corresponding to a Fe-S signature (Fig. 3 clusters signature but with a A 400 /A 280 ratio of 0.18 lower than the full-length protein (Fig. 3, Inset). The two conserved cysteine residues, Cys215 and Cys218, of the Mrp ORP CXXC motif were then replaced by alanine residues by site directed mutagenesis (Fig. 2, red asterisks). The UV-visible absorption spectrum of the corresponding reconstituted variant protein Mrp ORP C215A/C218A exhibited a drastic decrease in the absorbance at 400 nm with an A 400 /A 280 ratio of 0.08 compared to 0.3 for the wild-type protein (Fig. 3, dashed line). From these results, we conclude that the CXXC motif of the P-loop NTPase domain of Mrp ORP is involved in the binding of a Fe-S cluster.
The Di-Nase domain of Mrp ORP binds a 3Fe-4S cluster. We noticed that the reconstituted Mrp ORP C215A/C218A exhibited a weak absorbance around 400 nm compared to the apo-protein spectrum and that the P-loop NTPase domain had a ratio A 400 /A 280 lower than the full-length Mrp ORP (Fig. 3). To explore the hypothesis that the Di-Nase domain could bind a Fe-S cluster, the strep-tagged-Di-Nase domain of Mrp ORP (Mrp ORP _CT) was anaerobically produced and isolated from DvH. The as-isolated domain has a brown color and exhibited an UV-visible spectrum with a broad absorption band at 420 nm, with a shoulder at 325 nm ( Fig. 4A, solid line), and a A 420 /A 277 ratio equal to 0.57. These features disappeared upon reduction with sodium dithionite (Fig. 4A, dashed line). The Mrp ORP _CT contained 3.2 ± 0.1 Fe per polypeptide chain, and an extinction coefficient of 15200 M −1 cm −1 at 420 nm (with an extinction coefficient per iron of around 4750 M −1 cm −1 ), a value within the expected range for Fe-S cluster-containing proteins. As previous studies described Mrp/Npb35 proteins (Nbp35, Ind1 and ApbC) as dimeric proteins 9,12,16,26 , the quaternary structure of Mrp ORP _CT was analyzed by gel filtration (Supplementary Fig. S3). The Mrp ORP _CT domain eluted mainly as a dimer ( Supplementary Fig. S3).
The Mrp ORP _CT exhibits an EPR signal in the as-isolated form, in the perpendicular-mode, with g values at 2.012, 2.009 and 1.96, that decreases in intensity by increasing the temperature (data not shown), while the dithionite-reduced form is EPR silent (Fig. 4B) 1+ , this center presents three high-spin ferric ions, which are spin coupled, with a S total = 1/2 state, and upon reduction to the [3Fe-4S]° oxidation state, by one electron, yields an integer spin (S total = 2) species, which is not observed in the perpendicular measurement mode (Fig. 4B). The g-values are not similar to the ones found in typical [3Fe-4S] cluster containing proteins, but this protein does not present in its primary sequence the usual binding motif for this type of metal cluster. Nevertheless, the g-values determined for Mrp ORP _CT are close to the ones reported for ThiI that has been shown to bind a [3Fe-4S] cluster 27 .
Therefore, the spectroscopic data together with the presence of only 3 conserved cysteine residues in the primary sequence of this domain (Fig. 2), support the hypothesis of a cuboidal [3Fe-4S] 1+ cluster (S total = 1/2) being present in Mrp ORP _CT, that can be reduced to [3Fe-4S]° (S total = 2) 28 .
The cyclic voltammogram of Mrp ORP _CT presents a reversible signal at −445 ± 10 mV (signal I in Fig. 4C) and another at −645 ± 10 mV (signal II in Fig. 4C) of which it is only observed the cathodic counterpart. The first  Mrp ORP transfers its Fe-S cluster to apo-aconitase. We then analyzed whether holo-Mrp ORP was able to transfer, in vitro, Fe-S clusters to apo-protein, such as Aconitase B (AcnB). AcnB is known to be active only when its [4Fe-4S] cluster is inserted into the protein 29 . AcnB was purified aerobically from recombinant E. coli and Fe-S cluster was completely removed to obtain an inactive apo-AcnB. Reconstituted Mrp ORP was then incubated with Apo-AcnB in an anaerobic chamber in the presence of DTT and the enzymatic activity of AcnB was determined at periodic time intervals. The AcnB activity increased with time in the presence of a fixed concentration of holo-Mrp ORP (Fig. 5A). No significant AcnB activities were observed when apo-Mrp ORP was used instead of holo-Mrp ORP (data not shown) or when 15 μM of Fe 2+ and S 2− were added instead of holo-Mrp ORP (Fig. 5C).
Using a fixed concentration of apo-AcnB with various concentration of holo-Mrp ORP, we determined the amount of holo-Mrp ORP necessary to activate apo-AcnB (Fig. 5B, full square). Approximately 4 μM of holo-Mrp ORP were required to activate 2 μM of AcnB, i.e. in a ratio of 2 molecules of holo-Mrp ORP for 1 of AcnB (Fig. 5B, full square). As the presence of about 4 moles of iron and sulfide atoms are necessary to aconitase to be active 29 , a homodimer of holo-Mrp ORP sharing 4 iron and 4 sulfide atoms are transferred to apo-aconitase. The holo-Mrp ORP is then able to transfer its Fe-S cluster to apo-aconitase.
Additionally, we found that the reconstituted P-loop NTPase domain of Mrp ORP transferred its Fe-S cluster to apo-AcnB resulting in an AcnB activity comparable to the full-length protein (87.4%) (Fig. 5C). When the same experiment was performed with the holo-Di-Nase domain of Mrp ORP an aconitase activity of 44% of the full-length Mrp ORP activity was observed (Fig. 5C).
Altogether, these results show that, in vitro, the Fe-S cluster that is transferred from Mrp ORP to AcnB is preferentially the one bound to the P-loop NTPase domain. Mrp ORP is able to transfer its Fe-S clusters to its physiological ORP partners. We previously showed that Mrp ORP interacts in vivo with the ORP complex that contains Fe-S binding proteins, especially the Orp3 and Orp4 proteins that each exhibits two [4Fe-4S] ferredoxin-like motifs 19 . We thus tested whether Mrp ORP was able to transfer its Fe-S clusters to its physiological partner proteins. For this purpose, from DvH cell extract, the His-tagged Orp3, that we systematically co-purified with its partners Orp4 and Orp8, was treated to remove metal centers (Fig. 6, dashed line). In vitro Fe-S transfer assays were performed using the reconstituted holo-form of a Strep-tagged Mrp ORP and the His-tagged apo-Orp3-Orp4-Orp8 complex. The reconstituted holo-Mrp ORP and apo-Orp3-Orp4-Orp8 were incubated for 90 min under anaerobic conditions and the proteins were separated using a Ni-NTA column. After separation, the UV-visible spectrum of the eluted fraction exhibited strong absorption bands at 420 nm and 325 nm with an A 400 /A 280 ratio of 0.42, a spectrum similar to the spectrum of anaerobically purified holo-Orp3-Orp4-Orp8 proteins exhibiting A 400 /A 280 ratio of 0.53 (Fig. 6, solid line and inset). These results demonstrate that, in vitro, Mrp ORP can efficiently transfer Fe-S cluster to its physiological partner, the ORP complex.

Mrp ORP exhibits a polar localization in DvH.
To further characterize Mrp ORP, we then assessed its cellular localization in DvH using fluorescence microscopy imaging (Fig. 7). To achieve this goal, the full length Mrp ORP was fused to the green fluorescent protein (GFP). In order to express the fusion mrp ORP -gfp gene from the native promoter, the fusion was introduced into the DvH chromosome at the orp locus, replacing the endogenous wild-type mrp ORP gene. Because GFP does not fluoresce in the absence of oxygen, cells were first grown under anaerobic conditions, and the pictures were acquired less than 10 min after contact with air. From our previous study, we showed that this time of air incubation doesn't affect the localization of a FtsZ-GFP fusion in DvH 30 . The profile of the fluorescence signal observed during the initiation step of the cell cycle of DvH for Mrp ORP appeared to be localized at one pole for 78% of the cells (Fig. 7A). Western blot analysis of the Mrp ORP -GFP using anti-GFP antibodies revealed only one band corresponding to the fusion protein, Mrp ORP -GFP, suggesting that the integrity of the fusion proteins was conserved ( Supplementary Fig. S4). The P-loop NTPase domain was mostly located at one (58% of cells) or two poles (37% of cells) (Fig. 7B) whereas the fluorescence of the GFP-Di-Nase fusion was diffused in the cytoplasm (Fig. 7C). No growth defect was observed whatever the recombinant strain used. These results revealed a polar spatial localization of Mrp ORP linked to the Mrp/NBP35 domain.

Discussion
P-loop NTPases are one of the largest class of proteins with subgroup members involved in a wide variety of essential cellular functions 8 . The Mrp/NBP35 ATP-binding protein subclass comprises proteins present in all three kingdoms of life and mainly involved in Fe-S cluster biogenesis [7][8][9][10][11]15 .
In this study, we characterised Mrp ORP , a novel type of Mrp/NBP35 ATP-binding protein. Mrp ORP is distinct form the other members of this family by the fact that it associates a P-loop containing nucleoside triphosphate hydrolase domain (P-loop NTPase) with a dinitrogenase iron-molybdenum cofactor biosynthesis domain (Di-Nase). The phylogenetic analysis of the Mrp/NBP35 ATP-binding protein family showed that the association between both domains occurred several times independently.
Characterization of the reconstituted wild-type and mutant Mrp ORP proteins, with the results described from biochemical analyses of other members of the Mrp/NBP35 ATP-binding protein family, indicated that the conserved CXXC motif of the P-loop NTPase domain coordinates a [4Fe-4S] cluster between two Mrp ORP molecules. Interestingly, our data suggest that, in spite of the lack of classical Fe-S binding motif, the Di-Nase domain does have a 3Fe-4S cluster that can exist in two redox states [3Fe-4S] 1+ and [3Fe-4S] 0 , with a reduction potential of is put aside by the fact that similar data were obtained from two different preparations and all the manipulation were performed inside the anaerobic box in a one-day purification procedure. However additional work is needed for definitive identification of this cluster. From our preliminary results, the cluster in the Di-Nase domain of Mrp ORP seems to be clearly different from the heterometalic sulfide cluster (S 2 MoS 2 CuS 2 MoS 2 ) noncovalently bond to the polypeptide chain of Orp8, another Di-Nase one-domain protein belonging to the ORP complex 31,32 . It is also different from the IssA protein from Pyrococcus furiosus belonging to the same family and that binds thioferrate through a cationic sequence in the C-terminal tail not found in Orp9 33 . A multiple alignment of the Di-Nase domain of proteins associating this domain with the P-loop NTPase domain as in Mrp ORP, shows two cysteine_histidine rich conserved motifs: the CXHFGHCE motif located at the beginning of the Di-Nase domain and the CDH sequence located at the end of the domain (Supplementary Fig. S5). As cysteine and histidine residues can coordinate a Fe-S cluster, our hypothesis is that these conserved residues are involved in the binding of the [3Fe-4S] cluster in the Di-Nase domains of Mrp ORP . Altogether, these results allow us to propose that Mrp ORP is a novel member of the Mrp/NBP35 ATP-binding family which can bind at least two Fe-S clusters, one interdomain [4Fe-4S] cluster in the P-loop NTPase domain and one [3Fe-4S] cluster in the Di-Nase domain.
We further showed that the Fe-S cluster of Mrp ORP can be transferred to apo-proteins, such as the aconitase and the ferredoxin-like proteins (Orp3 and Orp4) of the ORP complex previously shown to interact with Mrp ORP in vivo 19 . The genomic clustering of mrp ORP with the ORP encoding genes, further, is in total agreement with the idea that Mrp ORP might be dedicated to the maturation of the Fe-S containing metalloproteins belonging to the ORP complex. To date, the only known target identified for bacterial Mrp is the TcuB protein, a protein necessary for tricarballylate catabolism 10 . We propose here the ORP complex as a novel target of Mrp/NBP35 ATP-binding proteins.
We then investigated the cellular localization of Mrp ORP that was mainly observed at one pole of the cell. Such polar localization for a Fe-S carrier has never been described before and questions the localization of others prokaryotic Mrp/NBP35 ATP-binding protein because we showed that the polar localization is probably linked to the P-loop NTPase domain. Interestingly, this localization is found to be in adequation with the putative apo-target localization, as Orp3 and Orp4 exhibit a C-terminal amphipatic helix shown in MinD to be responsible for polar binding (Fig. S6) 34 . Polar localization was previously observed for protein involved in several biological processes, such as cell division, chemotaxis, signal transduction, cellular differentiation, virulence and bacterial respiration [35][36][37][38][39][40] but never reported for a Fe-S protein maturation factor. We demonstrated that the Fe-S cluster present in the Di-Nase domain is not efficienly transferable to apo-aconitase. Then, the role of this domain is still unclear. The presence in the Di-Nase domain of a high content of conserved proline residues (TPPPHXPGXXP), that have been shown to be involved in protein-protein/domain interaction, might be responsible for the specificity of interaction of Mrp ORP with dedicated apo-partners to which the Fe-S is transferred (in magenta in Supplementary Fig. S5) 41 . Alternatively, the Fe-S cluster present in the Di-Nase domain might have a structural role in Mrp ORP . Hence, the presence of these unusual [3Fe-4S] clusters has been observed in enzymes, such as nitrate reductase, [NiFe] hydrogenase and ThiI 27,42,43 . Although, their role in those proteins has not been established, those centers have been considered to be involved in electron transfer and recently in sulfur transfer as in Thil 27,43 .
It has been shown that Nbp35 proteins possess an extra stable 4Fe-4S cluster absent in others Mrp/NBP35 ATP-binding proteins characterized to date 13,24 . The role of this 4Fe-4S located in the N-terminal extension of Nbp35 is still unclear. Curiously, Mrp ORP contains also four cysteine residues included in a non-canonical motif in the N-terminal part of the P-loop NTPase with only one cysteine residue conserved in Nbp35 (Fig. 2). This feature added to the phylogenic position of Mrp ORP , suggest that bacterial Mrp ORP are closer to eukaryotic Nbp35 than bacterial Mrp and ApbC.
Fe-S cluster biogenesis in anaerobic bacteria is poorly documented, while these organisms rely heavily on Fe-S cluster enzymes 44 . This study starts to fill this gap by describing that Mrp ORP is likely an Fe-S cluster biogenesis factor in DvH and DdG20. Genome scanning analysis revealed that DvH possesses a minimal ISC system constituted by a cysteine desulfurase and a scaffold protein (NifU type) 45 . In addition, we detected homologues of the E. coli SufB and SufD proteins that might constitute a minimal SUF system, reminiscent of what is observed in archaea. To date, Mrp/NBP35 ATP-binding proteins in Fe-S biogenesis have been proposed to act as Fe-S scaffold and carriers 12,16,24 . Interrestingly, we noticed redundancy of Mrp proteins in DvH and other SRM. In DvH, two other Mrp/NBP35 ATP-binding proteins are detected, DVU1847 and DVU2330, and are composed solely of the P-loop NTPase domain containing the conserved deviant Walker box and the CXXC motif. Dvu2330 belonged to an operon encoding proteins involved in the biogenesis of Fe-S hydrogenases and Dvu1847 is included in an operon encoding a L-isoaspartate O-methyltransferase. The outstanding redundancy of Mrp/ NBP35 ATP-binding proteins in DvH and other SRM raises the question of the role of each of these proteins in these anaerobic microorganisms. Our phylogenomic study shows clearly that the three Mrp/Nbp35 ATP-binding proteins from DvH are closer to the eukaryotic Mrp/NBP35 than the bacterial and archaeal proteins.
Future studies will determine whether the assembly of Fe-S cluster on Mrp ORP is dependent of the general Fe-S biogenesis of DvH (ISC or SUF) or if Mrp ORP acts in parallel of these systems.

Methods
Bacterial Strains, Plasmids and Growth Conditions. Strains and plasmids used in this study are listed in Table S4. Escherichia coli DH5α and TG1 strains were grown in Luria-Bertani (LB) medium at 37 °C with the appropriate antibiotic when required (0.27 mM for ampicillin, 0.15 mM for chloramphenicol). Cultures of DvH were grown in medium C 46 at 33 °C in an anaerobic atmosphere supplemented with 0.17 mM of kanamycin or 0.15 mM of thiamphenicol when required. Anaerobic work was performed using an anaerobic chamber (COY Laboratory Products or MBraun) filled with a 10% H 2 -90% N 2 mixed-gas atmosphere. Before placement inside the anaerobic chamber, solutions were made anoxic by flushing with N 2 for removal of O 2 . Solutions, glass and plastic materials were equilibrated for at least 12 hours inside the anaerobic chamber before use.

Construction of Plasmids Used for Protein Production in E. coli. Standard protocols were used
for cloning, ligation and transformations. Custom oligonucleotides used are listed in Table S5. All restriction endonucleases and DNA modifications enzymes were purchased from New England Biolabs. Plasmids DNA were purified using the High Pure Isolation Plasmid Kit (Roche Diagnostics). PCR products were purified using MiniElute kits (Qiagen). For construction of pJF119-2109His, pJF119-3202His and pJF119-Nter3202His the appropriated primers described in Table S5 were used to amplify the dvu2109 and dde3202 genes from genomic DNA. The obtained PCR products and the pJF119 plasmid were digested with EcoRI and BamHI restriction enzymes and ligated into the multiple cloning site of the plasmid to obtain pJF119-2109His, pJF119-3202His and pJF119-Nter3202His, respectively. For all constructs, successful ligations were confirmed via DNA sequencing and subsequently transformed into TG1 E. coli cells.

Construction of Plasmids Used for Protein Production in DvH.
For construction of pBMC-6C3::3202strep, pBMC6C3::Cter2109strep and pBMC6C3::2103His the appropriated primers described in Table S5 were used to amplify the dde3202, dvu2109 and dvu2103 genes from genomic DNA. The obtained PCR products and the pBMC6C3 plasmid were digested with NdeI and SacI restriction enzymes and ligated into the multiple cloning site of the plasmid to obtain pBMC6C3::3202strep, pBMC6C3::Cter2109strep and pBMC-6C3::2103His, respectively. For all constructs, successful ligations were confirmed via DNA sequencing and subsequently electroporated into DvH cells.
Site Directed Mutagenesis. Simultaneous mutations of cys215 and cys218 residues from Mrp ORP were generated by oligonucleotide-directed mutagenesis using pBMC6C3::3202His as the PCR template and the Q5 site directed mutagenesis kit from Biolabs. The primers 3202cysmutF and 3202cysmutR were designed using the online NEB primer design software NEBaseChanger TM . ploop-ntpase-gfp and di-nase-gfp were constructed and inserted into the mrp ORP locus. With this construction, although the wild-type copy of mrp ORP is still present, it does not present the σ 54 binding site unlike the Mrp ORP -GFP allowing the expression of the fusion of interest in physiological conditions. The mrp ORP amplicon obtained by using the primer pair NterDVU2109_XhoI/CterDVU2109_NdeI and the plasmid pNot19Cm-Mob-XS-gfp 30 were cut with XhoI and NdeI. A gel extraction of the plasmid pNot19Cm-Mob-XS-gfp was done to insert mrp ORP into this plasmid using the XhoI and NdeI sites to obtain the plasmid pNot19Cm-Mob-XS-mrp ORP -gfp. To obtain, pNot19Cm-Mob-XS ploop-ntpase-gfp and pNot19Cm-Mob-XS-di-nase-gfp, the amplicons ploop-ntpase-gfp and di-nase-gfp were amplified by PCR using pNot19Cm-Mob-XS mrp orp -gfp as template and the primer pairs Nter2109-XhoI/CterGFP-SpeI and domCter2109-dir-XhoI/CterGFP-SpeI, respectively. The amplicons were cut with XhoI and SpeI and inserted into the plasmid pNot19Cm-Mob-XS. The 3 plasmids were then transferred into E. coli WM3064 and subsequently transferred by conjugation to DvH cells. Cells carrying the chromosomal recombination with the target fusion were selected for their resistance to thiamphenicol and checked by PCR using primers pair: DVU2108_UP/CterGFP-SpeI. A western blotting on DvH cells was also down by using an anti-GFP antibody, as described by Fievet et al., 2015 to control the production of Mrp ORP -GFP fusion 30  Protein purity was analyzed in a 12.5% Tris-Tricine SDS-PAGE. The fraction containing the protein was concentrated with centrifugal filter units (cut-off of 5 kDa) and freezed in liquid nitrogen until further use. Aconitase (AcnB) was purified as described for AcnA 47 . After purification of recombinant proteins, the eluted fractions were buffer exchanged with the buffer specified using a Hitrap Desalting column (GE Healthcare). Fractions that contained protein of interest at >95% purity, by SDS-PAGE analysis, were pooled and concentrated over a 10 kDa molecular mass cutoff membrane. Finally, the proteins were stored in liquid nitrogen. Protein concentration was determined using a Pierce TM 660 nm Protein Assay (Thermo) Pierce colorimetric assay. Bovine serum albumin (2 mg/mL, Sigma) was used as a standard.

[Fe-S] Cluster Reconstitution.
[Fe-S] cluster reconstitution was performed anaerobically in an anaerobic chamber (COY) at 18 °C as follows. Protein was reduced anaerobically with 5 mM DTT for at least 1 hour prior to Fe 2+ and S 2− addition. After pre-reduction, FeCl 3 was added to five-fold excess and incubated for approximately 2 minutes before an addition of a 5-fold excess of Li 2 S. The solution was incubated for 4 hours before excess salts and unbound iron were removed using a Hitrap Desalting column (GE Healthcare). For enzymatic reconstitutions, 5 mM L-cysteine and IscS (20 μM) were added in place of Li 2 S.
Quaternary structure determination. The quaternary structure of Mrp ORP_ CT was determined using a Superdex 200 10/300 GL size exclusion column (GE-Healthcare). The mobile phase used was 100 mM Tris-HCl, pH 7.5 and 500 mM NaCl and protein was injected on the colum at a flow-rate.of 1 mL/min. The standard used to create a standard curve were β-amylase (200 kDa) albumin (66 kDa) and carbonic anhydrase (29 kDa in the flow-through, while Orp3-His-tagged was eluted with buffer C containing 100 mM of imidazole. The eluted fraction containing Orp3 with co-eluted Orp4 and Orp8 was analyzed and an UV-visible spectrum was recorded. Aconitase Activity. Aconitase 52 . The corresponding sequences were aligned with MAFFT using the accurate option L-INS-I option, which allowed accurate multiple alignment construction. The resulting alignment was trimmed with BMGE (BLOSUM30 option) and used to infer a maximum likelihood tree with IQ-TREE v1.5.3 53 using the LG + I + G4 evolutionary model as suggested by the model selection tool implemented in IQ-TREE (BIC criterion). The robustness of the resulting tree was assessed with the non-parametric bootstrap procedure implemented in IQ-TREE (100 replicates of the original dataset) 54 .
Microscopy Experiments. The 3 fusion GFP strains were grown until the middle of the exponential growth phase (OD 600nm of approximately 0.4 to 0.5) in medium C. Cells were concentrated 2 times by centrifugation. The buffer used for this concentration contained 10 mM Tris-HCl (pH 7.6), 8 mM MgSO 4 , 1 mM KH 2 PO 4 . In order to stain DNA, this buffer was supplemented with 5 ng/μL of 4' ,6-diamidino-2-phenylindole (DAPI). After 20 min of incubation in the dark, the cells were washed three times in TPM buffer. The DNA was stained under anaerobic conditions to limit the exposure of the cells to air. The pictures were acquired after 10 min of air exposure, which was required for oxygen GFP maturation. The cells were placed between a coverslip and an agar pad of 2% agarose. Pictures were acquired with a Nikon TiE-PFS inverted epifluorescence microscope, 100x NA1.3 oil PhC objective (Nikon), and Hamamatsu Orca-R2 camera. For fluorescent images, a Nikon intenselight C-HGFI fluorescence lamp was used. Specific filters were used for each wavelength (Semrok HQ DAPI/CFP/GFP/YFP/ TxRed). Image processing was controlled by the NIS-Element software (Nikon). Electrochemical Measurements. All the electrochemical experiments were conducted inside an anaerobic chamber (MBraun) at room temperature, with 100 mM Tris-HCl 8.1, 500 mM NaCl, 2.5 mM desthiobiotin, 3 mM DTT used as electrolyte properly flushed with argon before entering in the chamber. A three-electrode configuration system, containing a reference electrode, Ag/AgCl (+205 mV vs SHE), a secondary electrode, platinum wire, and a work electrode, PGE, were used. To measure the cyclic voltammograms a µautolab (ECO Chemie, Utrecht, The Netherlands) was used being the data collect and analyzed on GPES software package (ECO chemie). Cyclic voltammetric measurements were performed on a potential window from +0.1 to −0.9 V (vs SHE), and the scan rate dependence investigated between 0.005 and 0.1 V s −1 .