The complex multiprotein systems for the assembly of protein-bound iron–sulfur (Fe–S) clusters are well defined in Gram-negative model organisms. However, little is known about Fe–S cluster biogenesis in other bacterial species. The ISC (iron–sulfur cluster) operon of Mycobacterium tuberculosis lacks several genes known to be essential for the function of this system in other organisms. However, the cysteine desulfurase IscSMtb (Rv number Rv3025c; Mtb denotes M. tuberculosis) is conserved in this important pathogen. The present study demonstrates that deleting iscSMtb renders the cells microaerophilic and hypersensitive to oxidative stress. Moreover, the ∆iscSMtb mutant shows impaired Fe–S cluster-dependent enzyme activity, clearly indicating that IscSMtb is associated with Fe–S cluster assembly. An extensive interaction network of IscSMtb with Fe–S proteins was identified, suggesting a novel mechanism of sulfur transfer by direct interaction with apoproteins. Interestingly, the highly homologous IscS of Escherichia coli failed to complement the ∆iscSMtb mutant and showed a less diverse protein-interaction profile. To identify a structural basis for these observations we determined the crystal structure of IscSMtb, which mirrors adaptations made in response to an ISC operon devoid of IscU-like Fe–S cluster scaffold proteins. We conclude that in M. tuberculosis IscS has been redesigned during evolution to compensate for the deletion of large parts of the ISC operon.
Iron–sulfur (Fe–S) proteins and sulfur-containing cofactors are ubiquitous in all kingdoms of life and participate in a plethora of fundamental processes of biosynthetic pathways [1–4]. Fe–S proteins are involved in redox systems, central metabolism, respiration, gene regulation, RNA modification, and DNA repair and replication. In humans, insufficiency in Fe–S cluster biogenesis leads to some forms of sideroblastic anaemia as well as Friedreich's ataxia, a degenerative disease of the nervous system .
To understand the mechanism by which these fragile clusters are built, maintained and repaired has been of major interest to the research community over the past two decades [2,3]. Owing to the multitude of Fe–S proteins (>150 predicted proteins in Escherichia coli), their biosynthetic machineries have a major effect on most essential metabolic pathways. Knowledge of the individual differences in Fe–S cluster biogenesis of different bacterial species may provide tools to better understand evolutionary adaptations made in response to changing habitats and strategies to combat bacterial pathogens.
To date, our knowledge of bacterial Fe–S cluster assembly mechanisms is based primarily on investigations made in Gram-negative bacteria. E. coli uses two multiprotein systems, ISC (iron–sulfur cluster) and SUF (sulfur metabolism), to assemble and deliver Fe–S clusters. The SUF system is active only under conditions of oxidative stress or iron limitation, whereas the ISC system is believed to be the housekeeping system that serves most Fe–S proteins and enzymes using sulfur as their cofactor [5,6]. Both systems function as complex machineries in which up to eight individual proteins encoded by the iscRSUA-hscBA-fdx operon (Figure 1) or the sufABCDSE operon participate in a multistep process which delivers [2Fe–2S] and [4Fe–4S] clusters to effector apoproteins. The central components of both systems are cysteine desulfurases (IscS and SufS). These PLP (pyridoxal 5′-phosphate)-dependent enzymes convert L-cysteine into L-alanine leading to the release of a sulfur atom that forms a persulfide on a cysteine residue of the enzyme [7,8]. In the case of E. coli IscS, sulfur is transferred to the apo-form of the Fe–S scaffold protein IscU by protein–protein interaction leading to the formation of a [2Fe–2S] cluster in the presence of iron. Once two [2Fe–2S] clusters have been generated on an IscU dimer, an [4Fe–4S] cluster can be formed and transferred to Fe–S effector proteins or the A-type carrier protein IscA. This protein is believed to have a Fe–S cluster-delivery role. Release of IscU-bound Fe–S clusters involves the ATP-hydrolysing activity of the co-chaperones HscA and HscB . The products of the stress-induced SUF operon are similar to those of the ISC-operon. They include a scaffold protein (SufB), an A-type carrier (SufA) and probably an ATPase (SufC) triggering the release of Fe–S clusters from SufB [2,3].
Comparison of ISC operons from different bacterial species
The considerable amount of knowledge available concerning Fe–S cluster biogenesis in Gram-negative bacteria is primarily based on observations made in ISC-knockout mutants. The phenotypes of these mutant strains clearly demonstrate the importance of this operon under Fe–S cluster-damaging conditions. E. coli strains with defects in Fe–S cluster biogenesis are hypersensitive to H2O2 since Fe–S clusters are perturbed by stress conditions such as oxidative stress [8,9]. Furthermore, the deletion of iscS in E. coli leads to distinct alterations of growth rate under changing oxygen tensions and impaired activity of Fe–S cluster-dependent enzymes such as Acn (aconitase) . These mutants were also used to show that IscS has been highly conserved during evolution, which facilitated heterologous complementation studies using IscS copies of different organisms [9–12]. It was thus shown that the slow-growth phenotype of an E. coli ∆iscS strain could be rescued by overexpressing the yeast IscS homologue Nfs1 (NiFS-like 1). Furthermore, IscSEc (where EC denotes E. coli.) complements the deleted Nfs1 gene in yeast when expressed with a presequence that shuttles the protein to the mitochondrion [10,11].
The formation of Fe–S clusters and the role of cysteine desulfurases in Gram-positive bacteria are less well understood. We performed a genomic analysis revealing that most low-GC-content Gram-positive bacteria, such as Clostridium difficile, display a reduced ISC operon where genes such as iscA and the chaperones hscA and hscB are missing (Figure 1). In the actinomycete Mycobacterium tuberculosis, a Gram-positive bacterium with a high GC content, the reduction of the ISC operon is even more dramatic and only the gene coding for the cysteine desulfurase IscS (IscSMtb, where Mtb denotes M. tuberculosis; Rv number Rv3025c) is retained (Figure 1). Since M. tuberculosis contains a SUF operon comparable with those found in many other bacteria, it is not known whether IscSMtb is also involved in Fe–S cluster generation in vivo . However, a conserved region around the active-site cysteine residue clearly indicates that IscSMtb is a group I (IscS-like) cysteine desulfurase . Furthermore, it was shown previously that affinity-purified IscSMtb is capable of reconstituting the [4Fe–4S] cluster of WhiB3 in vitro [15,16]. WhiB3 belongs to a family of regulatory Fe–S proteins present only in actinomycetes and their bacteriophages [17–20].
The survival and spread of M. tuberculosis relies on the successful adaptation to extremely variable environments ranging from high oxygen tensions in the lung to hypoxic conditions in the granuloma. In addition, this intracellular pathogen encounters changing redox conditions inside and outside the macrophage. These rapidly changing environments require fine-tuned Fe–S cluster biogenesis and repair. Indirect evidence of a role for IscSMtb in the oxidative stress response is given by microarray data showing that iscS is up-regulated after exposure to nitric oxide and H2O2 . To better understand the role of IscS in mycobacterial Fe–S cluster biogenesis we created a ∆iscSMtb mutant which showed a phenotype strongly associated with impaired Fe–S cluster assembly. Since an IscU homologue is missing in the mycobacterial ISC operon, we tested for direct interaction of IscSMtb with mycobacterial Fe–S proteins. Intriguingly, we were able to identify an extensive IscS–Fe–S protein-interaction network. Repeating these assays with the well-described IscSEc instead of IscSMtb led to markedly different results. Finally, we provide structural data explaining the failure of IscSMtb to interact with IscU and the evolutionary deletion of this gene from the mycobacterial ISC operon.
We conclude that the mycobacterial IscS has a role in Fe–S cluster assembly and in combating oxidative stress, although the respective ISC operon does not imply such a role. In addition we show that the mycobacterial protein differs from IscS homologues found in eukaryotes and Gram-negative bacteria.
MATERIALS AND METHODS
Bacterial strains, culture conditions and cloning strategies
M. tuberculosis H37Rv was grown in 7H9 broth (Difco) supplemented with 0.2% glycerol, Middlebrook ADC (albumin/dextrose/catalase) enrichment and 0.05% Tween 80 or on solid Middlebrook 7H10 medium (Difco) supplemented with OADC (oleic acid/ADC). For selection, hygromycin, streptomycin and kanamycin were used at a concentrations of 50 μg/ml, 25 μg/ml and 20 μg/ml respectively and sucrose was used at a concentration of 5%. Microaerophilic growth was performed using the GasPak™ pouch system from Becton Dickinson (EZ Campy system) or by overlaying M. tuberculosis colonies on 7H10 agar with 10 ml of top agar (7H9 medium with 0.75% agar). H2O2 exposure was performed in 7H9 broth with 5 mM H2O2 for 3 h. E. coli BL21(DE3) cells were used for expression of His6-tagged IscSMtb in LB medium containing 50 μg/ml ampicillin. All cloning steps were performed as described previously using the In-Fusion PCR Cloning kit (Clontech) . Site-directed mutagenesis was performed using the QuikChange II system (Agilent) according to the manufacturer’s recommendations. For cytotoxicity assays, THP-1 macrophages were grown in RPMI 1640 medium supplemented with 10% (v/v) FBS and activated with 200 nM PMA for 48 h. Cells were washed and infected with mycobacteria at an MOI (multiplicity of infection) of 5. MRC-5 fibroblasts were grown in MEM (minimal essential medium) supplemented with 10% (v/v) FBS and seeded into 96-well plates at 20000 cells per well. Cells were allowed to adhere and myocobacteria were added at an MOI of 10. After 24 h, extracellular bacteria were removed by washing with PBS and, after an additional 48 h of incubation in growth medium at 37°C under 5% CO2, 10% PrestoBlue® (Invitrogen) was added and the reaction was incubated for 1 h after measurement of fluorescence at 560/590 nm.
Allelic exchange and unmarked deletion of iscS
Details of the primers and plasmids used are shown in Supplementary Tables S1 and S2 (http://www.biochemj.org/bj/459/bj4590467add.htm). M. tuberculosis H37Rv cells were washed three times in 10% (v/v) glycerol and transformed with pGA44:iscS followed by selection on 7H10 agar containing streptomycin. These clones were transformed with pJG:Rv3026/trmU, and positive clones were selected on 7H10 agar containing hygromyin and kanamycin. Clones containing both plasmids were plated on 7H10 agar with streptomycin and sucrose. The deletion of trmU (tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase) was performed using pJG:Rv3023/iscS.
DNA extraction and Southern blot analysis
Mycobacteria were treated overnight at 37°C with SET buffer [25% sucrose, 50 mM sodium EDTA and 50 mM Tris/HCl (pH 8)] containing 3 mg/ml lysozyme. After the addition of RNase (subsequent incubation at 37°C for 30 min) and proteinase K (subsequent incubation 55°C for 30 min) DNA was extracted with phenol/chloroform and ethanol-precipitated. For Southern blot analysis, 4 μg of DNA was digested with AclI and AvrII and separated by agarose gel electrophoresis. After denaturation the DNA was blotted on to a Hybond N+ membrane. For probe labelling and detection the ECL direct nucleic acid labelling and detection kit (GE Healthcare) was used following the manufacturer’s recommendations. A PCR product spanning Rv3026/trmU without iscS was used as the probe.
RNA extraction and PCR
For gene expression analysis, RNA was extracted with TRIzol® (Invitrogen) and treated with DNase I (Roche) before generation of the cDNA template. cDNA was synthesized using the RevertAid First Strand cDNA Synthesis kit (Fermentas) using random hexamer primers. cDNA corresponding to 10 ng of input RNA was used in each RT (reverse transcription)–PCR supplemented with specific primer pairs (200 nM each) listed in Supplementary Table S2 and SYBR Green Master mix (Applied Biosystems). qPCR (quantitative real-time PCR) was performed with the 7900HT Fast Real-Time PCR system (Applied Biosystems) with the following parameters: 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. Melt curve analysis was used to confirm specific amplification of each primer pair.
Y2H (yeast two-hybrid) assays
The Matchmaker Gal4 Two-hybrid System 3 (Clontech) was used to screen for proteins that interact with IscSMtb and IscSEc. The respective plasmids based on pGBKT7 and pGADT7 (Supplementary Table S1) were co-transformed into yeast strain AH109 according to the recommendations of the manufacturer. Transformants were plated on SD (synthetic defined) minimal medium (−Leu/−Trp) and subcultured on high stringency plates [SD/−Ade (adenine)/−His/−Leu/−Trp/X-α-galactosidase].
Purification of His6-tagged IscSMtb
Mid-exponential-phase culture (2 litres) was induced with 0.5 mM IPTG and incubated for 12 h at 16°C. E. coli BL21(DE3) cells were lysed in lysis buffer [50 mM Tris/HCl (pH 8), 500 mM NaCl, 5 mM imidazole, 10% (v/v) glycerol and 1% (w/v) Tween 20] using a French press. After clearance by centrifugation, the lysates were incubated with 1 g of PrepEase resin (USB) for 1 h at 4°C followed by separation on a PolyPrep chromatography column (Bio-Rad Laboratories). The resin was washed with two column volumes of buffer containing 10 mM imidazole and eluted with 250 mM imidazole. After dialysis against 25 mM Tris/HCl (pH 7.5) and 100 mM NaCl the protein was further purified by gel filtration on a HiLoad 16/60 Superdex 200 column (GE Healthcare).
Experiments were performed with S-tagged WhiB3, Acn or ShdB expressed from pETduet-1 as prey and His-tagged IscSMtb expressed from pQE80 as bait. Empty pQE80 was used as a negative control. E. coli BL21(DE3) cultures (500 ml) were grown and induced as described above. Prey and bait cultures were mixed, and lysis and resin binding were performed as described above [lysis buffer: 25 mM Tris/HCl (pH 8), 200 mM NaCl, 5 mM imidazole and 2% (v/v) glycerol]. After elution with lysis buffer containing 250 mM imidazole, proteins were separated by SDS/PAGE (4–12% gel) and transferred on to a nitrocellulose membrane. S-tagged proteins were detected using an S-tag monoclonal antibody at a dilution of 1:5000 (Merck).
Enzyme assays with L-cysteine as substrate contained 200 mM Tris/HCl (pH 7.5), 50 mM NaCl, 0.25 mM PLP and 0.5–100 mM L-cysteine dissolved in water and 5 mM DTT. Before enzyme assays for determining the activity of IscS with L-selenocysteine were performed, L-selenocystine had to be reduced to obtain L-selenocysteine. For this purpose L-selenocystine was first solved in a water/methanol (1:2, v/v). Subsequently, the pH was increased by the addition of two equivalents of a 1 M NaOH solution while cooling on ice. NaBH4 was added in portions under stirring until the yellow mixture became colourless. The pH was adjusted to pH 5.5 to obtain the free amino acid instead of the sodium salt. The L-selenocysteine assay mixtures contained 200 mM Tris/HCl (pH 7.5), 50 mM NaCl, 0.25 mM PLP and 3.75–30 mM L-selenocysteine. The enzyme assays were started by the addition of purified protein (100 μg in 1 ml) and were performed at 37°C. Several 50 μl samples were collected over a period of 20 min, and reactions were terminated by mixing each sample with 30 μl of 5% perchloric acid and 38% ethanol. After neutralization with 20 μl of 20 mM Tris/HCl (pH 8) buffer containing 23 mM K2CO3, the precipitated salts were removed by centrifugation (16200 g for 10 min at 22°C). Subsequently, the product, L-alanine, was quantified by ultra-high-pressure liquid chromatography as its o-phthaldehyde derivative. Assays were linear over time and proportional to the protein concentration applied.
For Acn activity, exponentially growing cultures were lysed by beadbeating with 0.1 mm zirconia beads in ice-cold assay buffer. After centrifugation (16200 g for 10 min at 4°C), Acn activity was measured in the supernatant in a coupled enzymatic reaction using an Acn activity assay kit (Sigma–Aldrich). SdhB (succinate dehydrogenase) activity was measured according to the method of Munujos et al. . Whole-cell lysates (10 μl; 200 ng/μl total protein) were added to 90 μl of reaction buffer in a 96-well plate and colour development was monitored at 500 nm in a Tecan M200 microplate reader at 30°C.
Protein crystallization and X-ray data collection
Approximately 1000 buffer conditions were screened for protein crystals with 10 mg/ml IscSTB in 25 mM Tris/HCl (pH 7.5) and 100 mM NaCl. The best condition was refined using the hanging-drop vapour-diffusion method in drops of 2–4 μl at 18°C. Diffraction-quality crystals were grown in a crystallization solution consisting of 0.2 M ammonium sulfate, 0.1 M Bis-Tris (pH 5.5), 10% (w/v) PEG 3350 and 5% (v/v) glycerol. All crystals were cryoprotected in reservoir mother liquor supplemented with 25% (v/v) glycerol and directly flash-cooled in liquid nitrogen. Diffraction data were collected on beamlines PXII and PXIII of the Swiss Light Source (Villigen, Switzerland). All data were indexed, scaled and integrated using XDS .
Structure determination and model refinement
IscSMtb was crystallized in the P1 space group with unit cell dimensions of approximately a=66 Å (1 Å=0.1 nm), b=78 Å, c=93 Å, α=94°, β=104° and γ=99°, with four molecules in the asymmetric unit. Phase determinations were carried out by molecular replacement using Phaser , part of the CCP4 Suite . IscSMtb was modelled on the published structure of IscS from E. coli  with PDB code 3LVM used as search model. The initial molecular replacement models were manually adjusted in COOT, part of the CCP4 Suite , and refined with REFMAC5 . Structure figures were prepared with PyMOL (http://www.pymol.org). All crystallographic statistics are listed in Supplementary Table S3 (http://www.biochemj.org/bj/459/bj4590467add.htm). Co-ordinates and structure factors for IscSMtb have been deposited in the PDB under accession code 4ISY.
An FDR (false discovery rate) of 1%, a minimum peptide Mascot score of 25 and a minimum of two matching peptides per protein were used as the identification criteria for Mtb proteins. A normalized spectrum count was used to compare the relative abundance of proteins. A detailed description of the LC–MS/MS methods used in the present study can be found in the Supplementary Online Data (http://www.biochemj.org/bj/459/bj4590467add.htm).
iscSMtb mutant is microaerophilic, hypersensitive to H2O2 stress and displays impaired activity of Fe–S cluster-dependent enzymes
The iscS gene of M. tuberculosis was predicted to be essential for cell viability by high-density mutagenesis . To create an unmarked deletion we first complemented M. tuberculosis H37Rv with pGA44:iscS, a plasmid carrying a tetracycline-repressible pristinamycin-dependent promoter. We then deleted iscS using pJG:Rv3026/trmU which contains the genes flanking iscS and the sacB gene for counterselection. After selection on medium containing sucrose, successful allelic exchange was confirmed by Southern blotting (Figure 2A). Compared with the wild-type strain, the growth rate of the complemented ∆iscS strain was slightly lower and the addition of tetracycline led to a pronounced deceleration of growth (Figure 2B). However, even in the presence of tetracycline, there was observable growth allowing the conclusion that IscS is not fully essential. To confirm this we performed a second deletion using pJG:Rv3026/trmU on a wild-type background. After counterselection, small colonies were visible after 10 weeks of incubation at 37°C and the successful deletion of iscS was confirmed by PCR (results not shown). Complementation of this mutant was possible with both plasmids expressing IscS from the strong HSP60 (heat-shock protein 60) promoter (pMV261B:iscSMtb) or from the putative endogenous promoter upstream of iscS (pMD31:iscSMtb). Incubating the mutant strain for 3 months revealed a distinct smooth and ruffled colony morphology which could be converted into the typical rough and spread-out colony morphology of wild-type M. tuberculosis upon complementation with iscS (Figure 2C). Interestingly, it was reported previously that a whiB3-knockout mutant showed similar colony morphology, and a defect in cord formation appears to be a feature of both mutants . These data show that iscS is not essential on rich medium, but its deletion leads to a slow-growth phenotype and altered morphology. When subcultured several times in liquid medium the growth rate could be increased and finally reached H37Rv wild-type levels. However, when these bacteria were plated on solid medium, the growth rate was again heavily reduced and the colonies were smooth, indicating a stable phenotype.
Unmarked deletion of
iscS in M. tuberculosis
The generation of the ∆iscSMtb mutant allowed us to identify characteristics associated with a putative role of IscS in Fe–S cluster assembly. Incubation of wild-type, knockout and complemented knockout mutant M. tuberculosis cells with 5 mM H2O2 for 3 h showed increased sensitivity of the mutant strain towards oxidative stress (Figure 3A). Further proof of a role for IscS in protecting M. tuberculosis from oxidative stress was obtained on overexpression of the protein from the strong pristinamycin-dependent promoter as this rendered the cells resistant to H2O2 (Figures 3A and 3B).
Growth characteristics of ∆
iscS under various conditions and cytotoxicity assays
To test whether the deletion of iscSMtb leads to altered activity of Fe–S cluster-dependent enzymes we quantified Acn activity in mycobacterial whole-cell lysates (Figure 3C). The [4Fe–4S] cluster of this TCA (tricarboxylic acid) cycle enzyme is essential for the conversion of citrate into isocitrate. Compared with the wild-type strain, the ∆iscSMtb mutant displayed approximately 50% less Acn activity, which is comparable with the activity seen in the ∆iscS mutant of E. coli . Acn activity could be restored through complementation (Figure 3C). Equal total amounts of Acn were identified by LC–MS/MS in the examined whole-cell lysates, ruling out reduced enzymatic activity due to different amounts of enzyme (Supplementary Figures S1A and S1B at http://www.biochemj.org/bj/459/bj4590467add.htm). We also tested for activity of the Fe–S protein SdhB in the respective strains. This enzyme also showed impaired activity in the mutant (Supplementary Figure S1C).
The biogenesis of Fe–S clusters is sensitive to oxygen since iron and sulfide can be modified by oxygen or ROS (reactive oxygen species), eventually leading to cluster decay. When grown under microaerophilic conditions, the growth rate of the mutant could be enhanced and was almost comparable with that of the H37Rv wild-type strain grown under the same conditions, indicating that IscS is needed for cluster assembly or repair under normal oxygen conditions (Figure 3D).
Putative targets of IscSMtb are Fe–S proteins with important roles in the virulence of M. tuberculosis. Furthermore, phagocytosis shifts the bacteria into redox-active cell compartments, which are dedicated to the elimination of microbes. We thus tested the ∆iscS strain in THP-1 macrophage and MRC-5 lung fibroblast virulence assays. At an MOI between 5 and 10, wild-type M. tuberculosis is cytotoxic for activated macrophages as well as fibroblasts and this can be quantified using cell viability assays [28,29]. After 72 h of infection, the ∆iscS strain was significantly less cytotoxic for macrophages and fibroblasts than the H37Rv wild-type or the complemented mutant strain, indicating an attenuated phenotype in vitro (Figure 3E).
It was shown recently that the deletion of trmU has an effect on the intracellular redox state of E. coli . To rule out the possibility that the phenotype we observed for the ∆iscS strain was not due to impaired TrmU activity as a consequence of insufficient sulfur transfer, we deleted trmU in M. tuberculosis (Supplementary Figure S2A at http://www.biochemj.org/bj/459/bj4590467add.htm) and performed Acn- and H2O2-sensitivity assays. The ∆trmU strain performed like wild-type M. tuberculosis in both assays indicating that the impaired Fe–S protein activity we observed in the ∆iscS strain is independent from TrmU (Supplementary Figures S2B and S2C).
iscS of E. coli does not rescue the small colony morphology or Acn activity of the ∆ iscSMtb mutant
The high conservation of iscS genes allowed for successful heterologous complementation experiments in iscS mutants of several organisms [10–12]. In order to identify distinct features of the mycobacterial protein we also tried to complement the ∆iscSMtb strain by overexpressing the well-described IscS protein of E. coli from a mycobacterial promoter. However, we were not able to rescue the small colony morphology of the knockout strain (Figure 4A). This failure for complementation was not due to low transcription or expression levels of IscSEc since we were able to detect and quantify the protein by LC–MS/MS in the mycobacterial cytosol (Figure 4B). We also tested for Acn activity in this IscSEc complemented mutant. Expression of IscSEc did not restore the full activity of the enzyme (Figure 3C).
Complementation of ∆
iscS with iscS from M. tuberculosis or E. coli
IscSMtb uses both
L-cysteine and L-selenocysteine as substrates
To rule out differences in enzyme kinetics as a possible explanation for the failure of IscSEc in complementing ∆iscSMtb, we assayed for cysteine desulfurase activity of recombinant IscSMtb. So far IscSMtb cysteine desulfurase activity was only shown indirectly by a shift of its UV–visible spectrum after the addition of L-cysteine . We quantified the product L-alanine after the addition of different concentrations of L-cysteine to the affinity-purified IscSMtb. Assuming Michaelis–Menten kinetics, we determined a Km value of 6.1 mM (Figure 4C) and, at a concentration of 12 mM L-cysteine, a specific activity of 0.31 μmol/min per mg, which is similar to previously published data on IscSEc . Several cysteine desulfurases, including IscSEc, were shown to provide selenium for selenoproteins such as FdhF (formate dehydrogenase H) and selenium-containing tRNAs using L-selenocysteine as a substrate . We were interested in similar activity for IscSMtb and performed enzymatic assays with this substrate. L-selenocysteine was generated from reduced L-selenocystin as described in the Materials and methods section. We were able to identify a high specific activity for this substrate at a concentration of 7.5 mM L-selenocysteine (Figure 4D). At higher concentrations, enzyme activity was decreased most probably due to acidification of the reaction buffer by the substrate, which has a pH of 5.5 (results not shown). However, we could clearly show that IscSMtb is capable of using both L-cysteine and L-selenocysteine as substrates. These data indicate that, despite similar enzymatic activities, IscSMtb is functionally distinct from its E. coli counterpart upon expression in the mycobacterial host.
IscSMtb, but not IscSEc, interacts with several mycobacterial Fe–S and sulfur-accepting proteins
An intracellular excess of free iron and sulfur is toxic for most bacteria and eukaryotes. Thus, in the IscS proteins examined so far, sulfur is liberated from cysteine and transferred to the scaffold protein IscU . Since an IscU homologue is absent from the mycobacterial ISC operon, we were interested in the direct interaction of IscSMtb with putative sulfur-accepting Fe–S proteins. IscSMtb is capable of reconstituting the [4Fe–4S] cluster of apo-WhiB3, a mycobacterial transcriptional regulator, in vitro . Using Y2H assays we were able to show an interaction of both proteins (Figure 5A). Furthermore, interactions were identified for other known Fe–S protein targets of the ISC system, such as the enzymes Acn (Rv number Rv1475c), SdhB (Rv number Rv3319), FdhF (Rv number Rv2900c) and the SirA (sulfite reductase; Rv number Rv2391). In addition, IscSMtb interacts with the non-Fe–S protein and mycobacterial ISC operon member TrmU (Figure 5A). Y2H assays were repeated with switched vectors in which IscSMtb was fused to the binding domain of the reporter system (pGBKT7-IscSMtb) and mycobacterial Fe–S proteins to the activating domain of the reporter plasmid pGADT7. These experiments gave identical results (not shown; see Supplementary Table S1 for the plasmid constructs).
IscSMtb interacts with several mycobacterial Fe–S proteins
Protein interactions identified in the Y2H experiments were confirmed for three mycobacterial Fe–S proteins (WhiB3, Acn and SdhB) by in vitro pull-down experiments. S-tagged Fe–S proteins overexpressed in E. coli were bound to His-tagged IscSMtb and captured on nickel resin. Protein complexes could be identified after immunoblotting and detection with an anti-S-tag antibody (Figure 5B).
To identify reasons for the failure of IscSEc to rescue the small colony phenotype of ∆iscSMtb we repeated the Y2H experiments using IscSEc. First, we confirmed published experiments on the positive interaction of IscSEc with IscUEc indicating that our Y2H assay can be used to test for putative protein interactions of IscSEc with Fe–S proteins  (Table 1). When IscSEc was tested for interaction with mycobacterial Fe–S proteins, a positive result could only be obtained for IscSEc and WhiB3. In contrast with IscSMtb, IscSEc failed to interact with SdhB, SirA and Acn. Thus the negative complementation phenotype that we observed for the ∆iscSMtb::iscSEc strain could be due to failure of IscSEc in interacting with sulfur-accepting proteins of M. tuberculosis.
|Vector .||pGBKT7:IscUEc .||pGBKT7:whiB3 .||pGBKT7:sdhB .||pGBKT7:Acn .||pGBKT7:sirA .||pGBKT7:IscSMtb .||pGBKT7:53 .|
|Vector .||pGBKT7:IscUEc .||pGBKT7:whiB3 .||pGBKT7:sdhB .||pGBKT7:Acn .||pGBKT7:sirA .||pGBKT7:IscSMtb .||pGBKT7:53 .|
We also tested for interaction of IscSMtb with IscU of E. coli. Although IscSMtb interacts with several mycobacterial Fe–S proteins, an interaction with this central sulfur-accepting protein of Gram-negative bacteria was not detectable (Table 1). The mycobacterial SUF operon contains an IscU/NifU-like gene product (Rv1465) with 24% homology with the respective E. coli protein. This protein also failed to interact with IscSMtb (Table 1). The reason for these negative results could be a C-terminal α-helix absent from IscSMtb, a part of the protein that is mandatory for the IscS–IscU interaction in E. coli (see the structural data below) [34,35]. Of note, not only IscSMtb, but also the Fe–S acceptor protein WhiB3, failed to interact with the putative IscU/Rv1465 protein (Table 1).
Overall protein fold and active site are conserved in IscSMtb and IscSEc
To better define the distinct characteristics of the IscSMtb protein we crystallized the protein and solved the structure at 2.6 Å resolution. The crystallographic data collection statistics are given in Supplementary Table S3. As expected, IscSMtb forms a homodimer with the typical large and small domains seen in most PLP-dependent enzymes (Supplementary Figure S3 http://www.biochemj.org/bj/459/bj4590467add.htm) .
IscSMtb and IscSEc share similar secondary structure and topology except for a C-terminal α-helix absent from IscSMtb as shown by superimposition of the structures in Figure 6(A). Residues of the C-terminal α-helix were shown previously to be essential for the IscS–IscU interaction in E. coli [34,35].
Structural alignment of IscSMtb (blue) and IscSEc (green) monomers
In its active-site pocket, IscS catalyses the desulfuration of cysteine yielding alanine. This involves the formation of a Schiff base between the substrate cysteine residue and the cofactor PLP followed by nucleophilic attack of the bound cysteine residue by Cys329 in IscSMtb and Cys328 in IscSEc. The highly homologous pockets of both proteins are rich in charged and/or polar amino acid side chains. The PLP group is anchored via an internal Schiff base with Lys205 in M. tuberculosis and Lys206 in E. coli (Figure 6B), and further stabilized by hydrogen bonds with identical amino acids in both enzymes indicating similar substrate-binding and catalysis mechanisms. Key to the versatile sulfur transfer activity of IscS proteins is the dual role of the active-site cysteine, which participates in catalysis of sulfur from the substrate cysteine and in transfer of a subsequently bound persulfide to effector proteins. The active-site cysteine residue is organized in a large and flexible loop, which does not display electron density in either the IscSEc structure (Cys328–Lys333) or our IscSMtb structure (Cys329–Ala334). However, the loop flanking amino acids, which are believed to ensure the necessary flexibility of the IscS loop, are conserved in both proteins, again indicating similar sulfur transfer mechanisms.
The surface amino acid composition of IscSMtb and IscSEc are highly distinct
Since IscSEc partially failed to interact with mycobacterial Fe–S proteins, we tested for differences in the properties of surface amino acids of the two IscS proteins using the Chimera software . Interestingly, the electrostatic surface potential is dramatically different in large areas of the proteins (Figure 7), which is mirrored by net charge calculations of −34 e for IscSMtb and −12 e for IscSEc. This is most pronounced in the vicinity of the IscSMtb active site, where large patches of the protein display a negative charge which is not the case for IscSEc. Intriguingly, the polarity of the surface amino acids is much more conserved in the two proteins with predominantly hydrophilic side chains as expected for cytosolic proteins. IscSMtb shows only a slightly higher content of hydrophobic areas as shown in Supplementary Figure S4 (http://www.biochemj.org/bj/459/bj4590467add.htm).
The electrostatic potential of the IscSMtb and the IscSEc dimer
The N-terminal part of IscSMtb is essential for protein–protein interactions
To map interaction domains of IscSMtb with sulfur-accepting proteins we performed deletion-mapping experiments of the C-terminus and probed these shorter versions of the protein in Y2H assays. It was found that large parts of the C-terminus could be deleted without affecting protein–protein interactions. However, putative regions between amino acids 68 and 143 are essential for interaction (Supplementary Figure S5A (http://www.biochemj.org/bj/459/bj4590467add.htm). This N-terminal part of IscSMtb is rich in arginine and tryptophan residues and a unique WXRRX5RRRX16W motif can be found in IscS proteins of actinomycetes, but not in proteins of other Gram-positive or Gram-negative bacteria (Supplementary Figure S5A). The arginine residues of the two IscS monomers form a large surface-exposed hydrophilic patch, whereas two tryptophan residues of each monomer come together in a cleft formed by the dimer interface (Supplementary Figures S5B and S5C). The structural composition of this part of the mycobacterial protein is clearly distinct from its E. coli counterpart where interaction-sensitive arginine residues are primarily found towards the C-terminus . In an attempt to identify the surface amino acids that are essential for proper IscSMtb function and protein–protein interaction, we performed structure-guided alanine-scanning mutagenesis on IscSMtb, choosing polar and surface-exposed amino acids in the interaction-sensitive IscSMtb regions. As shown in Supplementary Figure S6(A) (http://www.biochemj.org/bj/459/bj4590467add.htm), several single, double and triple point mutations were generated and tested for complementation in the slow-growing ∆iscSMtb strain. Both tryptophan residue mutations led to only partial complementation. Alanine exchange of the first two arginine residues in the motif had no effect on complementation, whereas triple mutations of Arg90–Arg92 led to complete abrogation of IscS complementation on the ∆iscS background. However, when expression levels of these mutated proteins were quantified by LC–MS/MS, significantly smaller amounts of protein could be detected compared with the wild-type protein, indicating that the mutated amino acids are essential for protein stability (Supplementary Figure S6B). To rule out promoter mutations in the mycobacterial expression plasmid that might explain the low levels of expressed protein, we individually cloned the mutated IscSMtb genes into an E. coli expression vector and purified the proteins from E. coli lysates. The resulting expression levels of the mutated proteins were equally low in comparison with the wild-type protein (results not shown). Again, this indicates that the affected amino acids are important for structural integrity of IscSMtb making it difficult to draw conclusions on their individual role in protein–protein interactions.
In the present study we generated genetic, structural and functional data to characterize the role of the mycobacterial cysteine desulfurase IscS in Fe–S cluster biogenesis. The largely depleted and unusual ISC operon found in mycobacteria and other Gram-positive bacteria initially led to the assumption that these organisms rely exclusively on the SUF system for Fe–S cluster assembly and repair [13,39]. However, we provide evidence that the deletion of iscS in M. tuberculosis leads to a phenotype which is associated with dysfunctional Fe–S cluster assembly. Since the typical scaffold proteins for cluster assembly and transfer are absent from the actinobacterial ISC operon, two scenarios for IscSMtb-mediated Fe–S cluster assembly are conceivable: (i) IscSMtb makes use of so far unidentified mycobacterial scaffold proteins or scaffold proteins of the SUF system; or (ii) a direct interaction of IscSMtb and apo-acceptor proteins is involved in sulfur transfer mediated by IscSMtb.
The latter theory of scaffold-protein-independent cluster assembly is supported by data on the mutational analysis of the ISC operon of Gram-negative bacteria. The complementation of an E. coli strain lacking the complete ISC operon with IscSEc alone partially restored the activity of some Fe–S-dependent enzymes, clearly indicating that IscS is capable of contributing to Fe–S cluster formation even in the absence of the scaffold protein IscU . It is well known that IscS has a general Fe–S cluster biogenesis-independent role in sulfur trafficking. A downstream gene product of both the C. difficile ISC operon and the M. tuberculosis ISC operon is the 2-thiouridylase TrmU (Figure 1), a non-Fe–S enzyme that has been shown to receive sulfur from IscS for the formation of 2-thiouridine in E. coli tRNA . Another example of sulfur donation by IscS is thiamine biosynthesis involving the non-Fe–S protein ThiI . These data clearly indicate that IscS can be a direct sulfur donor to several effector proteins. It is thus conceivable that IscS provides sulfur not only to non-Fe–S proteins by direct protein interaction, but also to the apo-forms of Fe–S proteins. Our protein–protein interaction studies, together with the data on impaired Acn and SdhB activity in the mutant strain, support this theory which represents a novel and so far undescribed mechanism of direct sulfur transfer to Fe–S proteins.
This raises the question of why M. tuberculosis performs IscU-independent cluster assembly. It was recently shown that the ISC-system in E. coli is inactivated by ROS, whereas the SUF system is induced and active only under oxidizing stress mediated by ROS . The labile element within the ISC operon was identified as IscU. Clusters built on this protein were easily oxidized and destroyed, whereas IscS itself retained full activity when exposed to ROS [3,43]. Intracellular mycobacteria encounter a hostile environment in which macrophages constantly produce ROS and reactive nitrogen species that have potent antimicrobial activity. M. tuberculosis has evolved several strategies to withstand this stress and to persist within human phagocytes . A ROS-labile IscU protein may be unfavourable for survival in this intracellular niche, and IscS-driven cluster assembly independent of IscU could be a requisite to promote intracellular survival. Our data showing that high levels of IscS protect from H2O2 killing support this theory.
We were surprised by our finding that IscSMtb converted L-selenocysteine into L-alanine in a highly efficient and specific way. In fact, the activity of the enzyme was higher in the presence of L-selenocysteine than in the presence of L-cysteine. The M. tuberculosis genome lacks genes involved in the biogenesis of selenoproteins including selenocysteine-specific tRNAs and selenophosphate synthase, which was shown to be an acceptor of IscS-derived selenium in other organisms . However, in the presence of selenium, bacteria synthesize L-selenocysteine through the cysteine biosynthesis pathway leading to its unspecific misincorporation into proteins . This effect is even more pronounced in bacteria lacking specific selenium-incorporation pathways, as is the case for M. tuberculosis . Owing to the high reduction potential, free L-selenocysteine is considered toxic to the cell, thus IscSMtb could be involved in the detoxification of this highly reactive amino acid.
Another unsolved problem in mycobacterial Fe–S assembly is the source of iron. In E. coli, there is some evidence that the bacterial frataxin CyaY may function as an iron donor or at least as an iron-sensing protein during Fe–S cluster assembly on IscU. In addition, there are in vitro data implying that IscA may provide iron for the E. coli ISC system . However, no CyaY or IscA homologue can be identified in the mycobacterial genome. Further studies are needed to answer these important questions.
In the present study, we have addressed the strong genomic variability of ISC operons found in different organisms. This variability implies distinct Fe–S cluster-assembly mechanisms and structural differences of the respective ISC proteins. In fact, the IscSMtb structure we solved mirrors adaptations made in response to an operon devoid of scaffold proteins since the protein lacks a C-terminal helix that is involved in the formation of the IscS–IscU complex in E. coli . Interestingly, it was shown previously that E. coli IscU binds to an IscS region rich in amino acids with negatively charged residues . Consistent with charge-dependent protein–protein interactions, it is the IscU mutation K103E (positive to negative) that abrogates the IscS–IscU interaction. We were able to show in the present study that the surface of IscSMtb predominantly contains negatively charged residues and the electrostatic potential, especially in the vicinity of the active site, is highly divergent in IscSMtb and IscSEc. This overall negative net charge in IscSMtb may be a requirement for sulfur transfer to Fe–S proteins, which often carry positively charged Fe–S active sites [47,48].
In conclusion, the results of the present study provide insight into a novel mechanism by which M. tuberculosis copes with environmental stress through Fe–S cluster biogenesis and/or repair by the versatile enzyme IscS. It is evident that IscS plays a decisive role in sulfur trafficking and sulfur metabolic pathways which are essential for survival and virulence of M. tuberculosis . Its unique structural features not present in IscS proteins of many other bacteria or the human IscS-like protein may provide opportunities for therapeutic interventions against this important pathogen.
formate dehydrogenase H
multiplicity of infection
reactive oxygen species
tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase
Jan Rybniker envisaged the project. Jan Rybniker, Jeffrey Chen and Gaëlle Kolly performed the gene knockouts. Jan Rybniker, Pia Hartmann and Edeltraud van Gumpel performed Y2H assays. Jan Marienhagen performed the IscS enzymatic assays. Jan Rybniker performed enzymatic assays with M. tuberculosis whole-cell lysates. Jan Rybniker and Florence Pojer established the protocols in protein purification and crystallization and determined the crystal structures of IscS. All authors contributed to the development of the paper.
We thank Diego Chiappe, Romain Hamelin and Dr Marc Moniatte (Proteomics Core Facility, EPFL, Lausanne, Switzerland) for mass spectrometric analysis of IscS protein samples and Professor Matteo dal Peraro, Matteo Thomas Degiacomi and Veronica Monticone (Laboratory for Biomolecular Modeling, EPFL, Lausanne, Switzerland) for their help with the electrostatic surface potential maps.
This work was supported by the BMBF (Bundesministerium für Bildung und Forschung) [grant number 01KI1017 to (J.R. and P.H.)] and the Swiss National Science Foundation [grant number 31003A_140778]. J.M.C. is a recipient of postdoctoral fellowships from the Canadian Thoracic Society and Canadian Institutes of Health Research.
Co-ordinates and structure factors for Mycobacterium tuberculosis IscS have been deposited in the PDB under accession code 4ISY.