Journal list menu
Mechanistic characterization of sulfur transfer from cysteine desulfurase SufS to the iron–sulfur scaffold SufU in Bacillus subtilis
1 Introduction
Iron–sulfur clusters (Fe/S clusters) belong to the prevalent and most versatile group of cofactors in nature. Due to their structural variability and diversity of redox potentials they play important roles in many biochemical pathways such as respiration, iron storage, redox control or sulfur donation [1]. The biogenesis of Fe/S clusters in bacteria is dependent on complex machineries, such as the NIF [2], ISC [3] and SUF system [4]. These systems are widely distributed amongst bacterial species and higher organisms [5, 6]. While Gram-negative bacteria like Escherichia coli possess the two Fe/S cluster biogenesis systems ISC and SUF, most Gram-positive bacteria such as Bacillus spp., Enterococcus faecalis or Mycobacterium tuberculosis only feature a SUF system for Fe/S cluster biosynthesis [5, 7-9] (Fig. 1 ). In Bacillus subtilis, the single components of this system including the major scaffold protein SufU were shown to be essential [7].
During the assembly process, all Fe/S cluster biogenesis systems share common basic principles: The cluster is built upon a scaffold protein of the U- or A-type [10, 11], on which different cluster types can be assembled and which can transfer the clusters subsequently to an array of target proteins [10-16]. In addition, SufB of the SUF system possibly represents a further type of scaffold proteins [17, 18]. A further common feature is the presence of PLP-dependent cysteine desulfurases (NifS, IscS, SufSE in Gram-negatives or SufS in Gram-positives) employed for the incorporation of sulfur derived from cysteine [13, 19, 20]. These enzymes convert cysteine to alanine and transfer the sulfur to the corresponding scaffold proteins.
The interaction of cysteine desulfurase and scaffold protein for the ISC system has been part of previous studies [21, 22]. The general suggested mechanism is that the sulfide is bound as a persulfide on an active site cysteine of the cysteine desulfurase, from which it is transferred to the conserved cysteines of the scaffold protein. It has been shown that this process involves the formation of a heterodisulfide complex between scaffold and cysteine desulfurase [21, 23]. For the E. coli IscS/IscU interaction, it was shown that all three cysteine residues of IscU could be loaded by IscS, while a disulfide bond is formed between Cys328 of IscS and Cys63 of IscU [23]. In Azotobacter vinelandii it was found that Cys37 of IscU is important for the formation of the heterodisulfide complex with IscS [21].
In this study, we investigated the mechanism of sulfur transfer between SufS and SufU in B. subtilis as a model system for Fe/S cluster biogenesis in Gram-positive bacteria. By kinetic and inhibitory studies of SufS activity, we analyzed the influence of SufU and its single cysteine to alanine mutation variants on cysteine desulfuration. This led to a model proposing the single steps of sulfur transfer between the cysteine desulfurase and the major scaffold protein in the conserved Gram-positive SUF system.
2 Materials and methods
Materials and methods are described in the Supplementary data.
3 Results
3.1 Kinetics of sulfur transfer
As we have reported previously, the sulfide release activity of SufS increases up to 40-fold in the presence of apo-SufU [7]. In this study, we performed kinetic analyses of SufS activity under variation of both cysteine and apo-SufU substrates. The reactions were analyzed by the detection of the released amount of sulfide. Concentrations of cysteine were varied between 0 and 2 mM under constant SufU concentration, sufficient for maximum SufS activation (Fig. 2 A), and varying SufU concentrations (0–10 μM) were also tested under non-limiting concentrations of cysteine (Fig. 2B). We found that the kinetics of SufS follow a Michaelis–Menten-like behavior under both cysteine and SufU variation. The determined kinetic parameters are listed in Table 1 .
| Cysteine dependent | SufU dependent | |
|---|---|---|
 |
49.71 ± 15.81 | 2.63 ± 0.48 |
| v max/mU/mg | 153.54 ± 11.59 | 93.38 ± 9.68 |
| k cat/s−1 | 0.255 ± 0.007 | 0.155 ± 0.016 |
| Hill parameter | 0.61 ± 0.12 | 1.37 ± 0.22 |
To determine the reaction mechanism, we analyzed the SufS activity with varied SufU and cysteine concentrations. We used 1, 2.5 and 5 μM SufU and varied cysteine from 10 to 50 μM, or alternatively from 10 to 1000 μM (Fig. 2C and Supplementary data Fig. S1). The parallel curves (within experimental error) of the double reciprocal plots indicate a Ping-Pong bi–bi reaction mechanism. The Ping-Pong type is supported by the fact that we could not observe cysteine bound to the PLP cofactor of SufS either directly after purification or after incubation with cysteine (Supplementary data Fig. S2), indicating a fast and stable loading of the catalytic SufS cysteine with the persulfide without requirement of SufU.
Kinetic data that were reported in parallel to this work by Selbach et al. [24] are in agreement with our observations.
3.2 SufS is able to reconstitute an Fe/S cluster on SufU
Since it is possible to build an Fe/S cluster under anaerobic conditions on SufU by using ferric ammonium citrate and cysteine in the presence of SufS (see Supplementary data Fig. S3) [7], we analyzed the amount of labile sulfide bound to the reconstituted holo-SufU after enzymatic reconstitution to determine the number of sulfide transfer cycles per SufU monomer until complete loading. After purification by size exclusion chromatography, an amount of 1.65 ± 0.05 labile sulfide per SufU monomer was detected, indicating an average of two sulfide transfers until complete Fe/S cluster formation in vitro. Exact determination of the cluster type was not possible, since EPR and Mössbauer spectroscopy were not possible [7].
3.3 SufU cysteine to alanine variants do not activate SufS, and the SufU_C41A variant inhibits SufS in presence of native SufU
To investigate the molecular interaction between SufU and SufS, we have analyzed the role of the three highly conserved SufU cysteine residues responsible for Fe/S cluster binding by generating the SufU single amino exchange derivatives C41A, C66A and C128A. Ten micromolars of each SufU variant were tested for its effect on sulfide release together with 0.5 μM SufS. As shown in Fig. 3 A, none of the C to A variants was able to activate SufS beyond the basal level of activity. These results raised the question if the variants still had the capability to interact with SufS. To analyze this, we investigated how the activation of SufS was affected when both native SufU and one of the SufU exchange variants were present during the reaction.
In Fig. 3B, the specific activities of SufS in presence of 0.5 μM SufS, 10 μM SufU and 10 μM of each variant are shown. Apparently, only SufU_C41A had an impact on the activity of SufS in the presence of native SufU, while SufU_C66A and SufU_C128A did not show any significant effect on SufS activity when equimolar amounts of SufU were present. Under the same conditions, SufU_C41A shows a strong inhibition resulting in a total sulfide production of 0.61 ± 0.05 μM. In comparison, a similar quantity of 0.51 ± 0.02 μM sulfide is produced when SufS is incubated with cysteine and SufU_C41A in the absence of native SufU (Fig. 2A, column 4), or without any addition of SufU resulting in 1.3 ± 0.2 μM released sulfide. The slightly increased activity of SufS alone is most likely a result of DTT-enhanced persulfide reduction under these reaction conditions (see Supplementary data S.1.4), indicating that SufU_C41A is able to shield the SufS persulfide from the reductive environment.
To determine the inhibitory efficiency of SufU_C41A in more detail, we analyzed the inhibition by measuring the sulfide release with 0.5 μM SufS in presence of 5, 10 or 20 μM native SufU and varying SufU_C41A concentration (0–10 μM SufU_C41A in each assay). The results are shown in Fig. 4 .
In all three cases, the sulfide release by SufS decreases with increasing SufU_C41A concentrations. The curves in Fig. 4 show that maximum of inhibition is reached in all cases at equimolar ratios of SufU_C41A and SufS even in presence of 40-fold excess of native SufU, indicating a strong competitive effect of SufU_C41A. Using the quenching of SufS-dependent specific activity as a direct output for the inhibitory activity of SufU_C41A, an apparent K i of 41 nM was determined when assuming a 1:1 binding mode for enzyme and inhibitor (Supplementary data Fig. S4). This low inhibition constant suggests a tight binding mode of inhibition.
3.4 SufU and SufU_C41A form stable complexes with SufS in presence of cysteine
Since SufU_C41A has a high affinity towards SufS, we tried to detect possible interaction complexes by analytical size exclusion chromatography by using the purified His-tag constructs of the proteins (Table 2 ). In the control without SufU variants, the apparent protein peak with a mass of ∼113 kDa indicated the presence of a SufS dimer. No complexes of higher stoichiometry were found when incubating SufS-His6 with SufU-His6 or SufU_C41A-His6. Only when additionally incubated with cysteine, a peak shift to ∼154 kDa was observed in presence of SufU_C41A-His6, which indicates the formation of a (SufS)2(SufU_C41A)2 heterotetramer. This stoichiometry is in agreement with a recently published crystallographic model for the interaction of IscU and IscS from E. coli [22]. Further, the observation of this complex only for the inhibiting derivative SufU_C41A-His6 is coherent with the results published by Selbach et al. [24], who found that His-tagged variants of native SufU and SufS do not form stable complexes when incubated after separate production and purification [24]. The presence of DTT in excess during gel filtration did not influence the observed retention times of the complex, suggesting an interaction that is stable without formation of disulfide bonds between the interaction partners.
| Protein mixture | Determined complex size (kDa) |
|---|---|
| SufS | 113 |
| SufS + SufU | 113 |
| SufS + SufU_C41A | 113 |
| SufS + SufU + cysteine | 113 |
| SufS + SufU_C41A + cysteine | 154 |
In a second interaction approach, we tested the possibility of complex co-purification by using tag-free SufS, which upon overexpression in E. coli was subjected with the total cell crude extract to a Ni2+-NTA that was pre-loaded with the His-tagged SufU variants. After washing, the elution fractions were analyzed for complex formation by SDS–PAGE. Equal amounts of SufS were detected together with SufU_C41A and also native SufU (Fig. 5 A), indicating an equimolar complex formation. Again, the interaction took only place in presence of cysteine in the loading buffer. This approach demonstrates that a stable interaction of SufS and native SufU can be established in vitro dependent on the nature of the recombinant protein construct, which in case of N-terminally His-tagged SufS seems to disturb protein–protein contact with SufU. In the homologous IscS–IscU system of E. coli, the N-terminus of IscS, as seen in the crystal structure, is located closely to the interaction surface built with IscU [22, 25] (Fig. 5B), suggesting similar structural arrangements in the putative (SufS)2–(SufU)2 complex. Further, the SDS–PAGE in vitro analysis with tag-free SufS demonstrates that the SufU variants C66A and C128A are indeed unable to interact with SufS (Fig. 5A), providing a rational for the observed SufS activities dependent on these variants in presence or absence of native SufU.
4 Discussion
In this study, we investigated the reaction mechanism of the cysteine desulfurase SufS with its substrates cysteine and the scaffold protein SufU. The investigation revealed new molecular insights into the sulfur transfer reaction as one initial step during Fe/S cluster biogenesis as it is established in the SUF system of Gram-positive bacteria.
We find that all three cysteine residues of SufU are essential for the activation of SufS. If cysteine residue 41, 66 or 128 is replaced by alanine, the SufS activity remains at the basal level. This suggests that the variants may either not interact anymore with SufS to initiate further catalytic turnover or, on the other hand, that they might inhibit SufS activity by a tight and arresting interaction.
When put in competition with native SufU, the single cysteine exchange variants show different behavior. The variants C66A and C128A do not show any competitive effect, thus resulting in full activation of SufS. In contrast, the C41A variant inhibits SufS even if native SufU is present in excess. SufU_C41A was found to decrease sulfide production below the basal level observed with SufS alone, and tight binding inhibition of this variant was supported by the determination of a K i value of 41 nM. The gel filtration and co-purification experiments show that only native SufU and SufU_C41A are able to form a stable interaction with SufS in presence of the SufS substrate cysteine. In contrast to E. coli IscS/IscU this interaction seems not to depend on disulfide bond formation between these protein partners [23]. Interestingly, the N-terminal His-tag of SufS effectively destabilizes the interaction with native SufU [24]. The crystal structure of the highly homologous IscS shows, that the N-terminus is in close proximity to the active center of SufS (Fig. 5B). However, SufU is still able to activate the tagged SufS protein, which indicates that the N-terminal region, especially the His-tag is also flexible. Therefore we conclude that the His-tag extension prevents stable non-covalent interactions but still allows catalytic activation.
A similar situation was observed with the IscU-D39A variant of A. vinelandii [26]. The authors found that this variant could be purified in a stable stoichiometric complex with IscS. In contrast to our SufU_C41A variant, the IscU_D39A variant is loaded with a [2Fe–2S] cluster, which indicates a remaining functionality of the scaffold and most likely excludes a global structural change caused by this mutation. Due to the persisting interaction properties of SufU_C41A with SufS, global structural rearrangements are also unlikely in case of this variant.
Based on the presented data, we propose that all three cysteines of B. subtilis SufU are essential for the interaction with SufS, and that especially cysteine residue 41 is responsible for the initiation of the sulfide transfer. Accordingly, the following reaction mechanism is proposed (Fig. 6 ): (1) At first, SufS is loaded with a persulfide by PLP-dependent conversion of cysteine to alanine. (2) Non-covalent interaction with SufU is based on the presence of cysteine residues 66 and 128. (3) The persulfide is transferred from SufS to cysteine 41 of SufU, which is located in the center of a 6 amino acid long flexible loop (see Fig. 7 A). Cysteine 41 likely comes into contact with the persulfide-loaded active site cysteine of SufS, thereby accepting the persulfide sulfur atom. (4) After intermolecular sulfur transfer, converted SufU dissociates from SufS, while the sulfur might be intramolecularly transferred from cysteine 41 to either cysteine 66 or 128. Optionally, the altered conformation of SufU-Cys41-SH might allow sulfide transfer directly onto cysteine 66 or 128 in the next cycle. (5) After the dissociation, the new cycle initiates with the binding of a further cysteine on SufS.
Our earlier studies showed that holo-SufU is not able to activate SufS [7]. Holo-SufU contains a [4Fe–4S] cluster likely bound in a dimeric state (Supplementary data Fig. S3). Since we found in total two labile sulfides per SufU monomer after complete cluster reconstitution dependent on SufS activity as sulfide donor, this suggests that each SufU monomer is loaded two times by SufS until the Fe/S cluster is formed.
The transfer-mechanism is supported by the kinetic analysis, which suggests an ordered non-sequential bi–bi reaction mechanism [24], supporting the sulfide transfer at the beginning of the reaction. The persulfide on SufS seems to be effectively shielded from the environment, since large excess of DTT does not significantly increase sulfide production in absence of SufU. This is in contrast to NifS or IscS, which have partially higher basal activities of sulfide release and can be used to load non-scaffold Fe/S proteins [13, 27].
The structural flexibility of SufU cysteine 41, which seems to have an important role for the persulfide transfer, is given by the different minimized energy NMR structures of SufU (PDB: 2AZH). In these structures, cysteine 41 coordinates a zinc ion not only with its sulfide donor function (Fig. 7A), but alternatively also with the oxygen of its peptide carbonyl bond, thereby replacing the sulfide group for possible intramolecular sulfide transfer (Fig. 7B). The overlay of 10 energy minimized SufU NMR structures (Fig. 7C) shows that cysteine 41 can be found in different positions during zinc coordination, suggesting a possible pathway for both inter- and intramolecular persulfide transfer by an instant switch between the S- and O-coordination modes of cysteine 41.
In contrast to earlier studies [21, 23], we found that all conserved cysteine residues in the scaffold protein of the Gram-positive SUF system are important for the cysteine desulfurase-dependent sulfide transfer reaction. While the earlier studies investigated the cysteines on which sulfide can be transferred using mass spectrometry to identify persulfides or disulfide bonds between IscS and IscU, we used the strong catalytic activation of SufS by SufU, to analyze the relevance of the single cysteine residues conserved in SufU. In conclusion, this is the first study that provides a tentative model for the overall reaction mechanism of a cysteine desulfurase in conjunction with its activating scaffold partner from the SUF assembly system found in Gram-positive bacteria.
Acknowledgments
This work was supported by the DFG and the excellence initiative LOEWE of the state Hessen. We thank Alan Tanović and the group of Professor Roland Lill (Philipps-Universität-Marburg) for helpful discussions.
Appendix A A
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2011.01.005.
