Media and growth conditions

Media used were Luria–Bertani (LB) rich medium or M9 minimal medium supplemented with glycerol (0.4%) and MgSO₄ (1 mM). l-Arabinose (Ara) (0.2%), casamino acids (0.05%), thiamine (50 μg ml⁻¹) and mevalonate ( MV), 1 mM were added when required. Solid media contained 1.5% agar. Antibiotics were used at the following concentrations: chloramphenicol (Cam), 30 μg ml⁻¹; kanamycin (Kan), 30 μg ml⁻¹; spectinomycin (Spc), 100 μg ml⁻¹; and ampicillin (Amp), 100 μg ml⁻¹. Note that Kan was used at 80 μg ml⁻¹ for transduction experiments involving the ΔiscS mutants as recipients. For growth at pH 7.0 or 4.5, cultures grown in LB rich medium were diluted 1:100 into 200 μl of LB or LB–HCl (pH 4.5) in 96-well plates. Growth was monitored using a TECAN spectrophotometer by recording OD₆₀₀ every 10 min over 16 h at 37°C.

Strains and plasmids

All strains used in this study are E. coli K-12 MG1655 derivatives. The mnmA mutant, FBE584, was obtained by in-frame deletion of mnmA and replaced with a kanamycin cassette (26). FBE597 was obtained from the E. coli KEIO Knockout Collection (27) by P1 transducing the mnmE::kan allele into MG1655 background. The mnmE::cam mutant was constructed by deleting and replacing the mnmE gene with a chloramphenicol resistance encoding cassette giving rise to the FBE707 strain (26). The conditional RExBADmnmA mutant, FBE583, was obtained by amplifying a fragment carrying the aadA7I (spectinomycin resistance gene) and araC genes by polymerase chain reaction (PCR) from TG1 spec RExBAD (FBE319) using the ML39 primers (28). The linear fragment was then inserted upstream of the mnmA gene in E. coli MG1655 carrying the λ red expression plasmid pKD46 (26). All mutants were P1-transduced into different E. coli K-12 backgrounds (Supplementary Table S1). The CF8310 strain, a MG1655 strain derivative carrying the T7 RNA polymerase encoding gene on a lambda prophage (DE3), was used as recipient for P1 transduction of ΔmnmA::kan, giving rise to the FBE598 strain. Strain FBE605 (ΔiscUA ΔsufABCDSE MVA⁺, kanR piscR::lacZ Δlac) is a MG1655 strain derivative, which can synthesize isopentenyl diphosphate from the introduced eukaryotic MVA-dependent pathway (29). Strain FBE605 was derived from BR404 (30) as follows. First, the CamR cassette was removed from the suf locus, yielding to BR412 that was used as a recipient for P1 transduction of the ΔiscUA::cat mutation, to generate FBE033. The Cat cassette was removed from the iscUA locus by using the pCP20 plasmid. PCR analysis was used to check the absence of the iscUA genes and the suf operon. DNA sequence analysis of the isc region was carried out and MVA-dependent viability was controlled. For all constructions, transductants were verified by PCR, using primers hybridizing upstream and downstream of the deleted gene. When necessary, antibiotic resistance cassettes were eliminated using plasmid pCP20 as described (31). Strains used in this study are listed in Supplementary Table S1.

Plasmid construction

To construct the pBADmnmA⁺ plasmid, the mnmA gene was first amplified from E. coli MG1655 chromosomal DNA using the ML40 primers, digested by EcoRI and XhoI and cloned into the EcoRI and SalI sites of pBAD24. The mnmA gene was also subcloned into the pET28a vector by PCR amplification from pBADmnmA⁺ to introduce the BamHI restriction site and the TEV nucleotide sequence in 5′ and the HindIII restriction site in 3′ using the ML52 primers. The PCR fragment was purified, digested by BamHI and HindIII and cloned into the BamHI and HindIII sites of pET28a. mnmA variants containing mutations were obtained by site-directed mutagenesis using the ML49–50-51 primers (NEB, E0554S). The sequence of the expression plasmids was confirmed by sequencing. Primers and plasmids used in this study are listed in Supplementary Tables S2 and S3, respectively.

Synthetic lethality test

Cells were cultivated in M9 minimal medium supplemented with 0.05% casamino acids and 0.4% glycerol overnight at 37°C. Saturated cultures were washed the next day twice with PBS and then diluted 1:50 into LB rich medium with or without 0.2% l-arabinose. Cultures were incubated at 37°C for 8 h. The optical density at 600 nm was measured over time.

Hydrogen peroxide sensitivity test

Overnight cultures were diluted using serial dilution in sterile PBS and 5 μl was directly spotted onto LB plates containing 1 mM H₂O₂. The plates were incubated overnight at 37°C before growth was recorded.

Overexpression of E. coli MnmA wild-type and the D99A-C102A and D99A-C102A-C199A variants

The mnmA gene was sub-cloned into the pET15b plasmid using a ligation independent cloning strategy by Eurofins to produce a 6His-MnmA protein construct whose 6His tag can be cleaved by the H3C protease. The mnmA_D99A-C102A and mnmA_{D99A-C102A-C199A} genes were sub-cloned into the pET28a plasmid. The plasmids were transformed into E. coli BL21(DE3) Star Codon Plus competent cells. One colony was used to inoculate 100–200 ml of LB medium supplemented with ampicillin or kanamycin (50 μg ml⁻¹) for MnmA wild-type and variants, respectively. 60 ml (20 ml) of this culture grown overnight at 37°C was used to inoculate 6 L (2 L) of LB medium supplemented with the same antibiotic. Cultures were grown at 37°C to an OD₆₀₀ 0.6–0.8, and overexpression was induced with 0.5 mM Isopropyl β-D-1-thiogalactopyranoside. After 4 h incubation at 37°C, cells were collected by centrifugation and stored at −20°C.

Purification of E. coli wild-type MnmA and variants

Cells were resuspended in 50 mM Tris-HCl (pH 7.5), 500 mM NaCl, 10% glycerol, containing RNase A (2 μg ml⁻¹), benzonase (1.6 U ml⁻¹, Sigma Aldrich), lysozyme (0.3 mg ml⁻¹), 1 mM PMSF PhenylMethylSulfonyl Fluoride, 1 mM β-mercaptoethanol and disrupted by sonication. Cells debris were removed by centrifugation at 25 000 rpm for 1 h at 4°C. The supernatant was then loaded on an immobilized metal affinity Ni-NTA column (HisTrap 5 ml, GE Healthcare) equilibrated in 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 1 mM PMSF and eluted with a linear gradient of 0–1 M imidazole. The proteins were collected, dialyzed twice against 1 L of 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 1 mM β-mercaptoethanol in the presence of the PreScission Protease (25 μg per mg wild-type MnmA) or TEV protease (43 μg/mg MnmA variants). Wild-type MnmA was further purified at 1 ml min⁻¹ onto a gel filtration column (Hiload 26/60 Superdex 200, GE Healthcare) equilibrated with 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, using an ÅKTA system. The as-purified proteins were concentrated to 24 mg ml⁻¹ (wild-type) with an Amicon Ultra filter device (30 kDa cutoff, Millipore) or 4–5 mg.ml⁻¹ (variant) with an Amicon Ultra filter device (10 kDa cutoff, Millipore), frozen in liquid nitrogen and stored at −80°C.

The GST–3C-protease (PreScission, a gift from S. Mouilleron) was expressed using pGEX-2T recombinant plasmids. After induction at 25°C with 0.1 mM IPTG for 20 h, the protein was purified using glutathione–Sepharose chromatography.

[Fe–S] cluster reconstitution and purification of holo-MnmA wild-type and variants

The reconstitution of the [4Fe–4S] cluster and purification of holo-MnmA were performed in a glove box containing <0.5 ppm O₂. After incubation of 100 μM as-purified MnmA with 10 mM dithiothreitol for 10 min, a 5.5-fold molar excess of ferrous ammonium sulfate and l-cysteine, as well as 2 μM E. coli cysteine desulfurase CsdA, were added, and incubation was extended overnight at room temperature. After centrifugation for 20 min at 12 300 g, holo-MnmA was loaded onto a Superdex 200 increase 10/300 GL column (GE Sciences) equilibrated in 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 5 mM dithiothreitol (DTT). The peak containing the holo-MnmA was then concentrated to 10–19 mg/ml on a Vivaspin concentrator (10 kDa cutoff). The same protocol was used for analyzing the capacity of the variants to host [Fe–S] clusters.

Quantification methods

The Bradford method was used to quantify the protein (32). The Fish and Beinert methods were routinely used after cluster reconstitution to quantify iron and sulfide, respectively (33,34).

SEC-MALS

Size exclusion chromatography coupled with multi-angle light scattering (SEC-MALS) experiments were performed using an HPLC-MALS system (Shimadzu) equipped with light scattering detector (mini DAWN TREOS, Wyatt Technology), refractive index detector (Optilab T-rEX, Wyatt Technology) and UV detector (SPD-20A, Shimadzu). Holo-MnmA (100 μl at 2 mg ml⁻¹) was injected on a Superdex 200 10/300 GL increase column (GE Healthcare) equilibrated in 50 mM Tris-HCl (pH 7.5), 200 mM NaCl buffer, in the presence or absence of DTT at a flow rate of 0.5 ml.min⁻¹. Molar masses of proteins were calculated using the ASTRA 6.1 software (Wyatt Technology) using a refractive index increment (dn/dc) value of 0.183 ml g⁻¹.

CD analysis

Circular dichroism (CD) spectra were recorded on a Chirascan-plus CD Spectrometer (Applied Photophysics). The far ultraviolet spectra (195−260 nm) were measured at 20°C in quartz cells of 0.5 mm optical path length. The final concentration of MnmA proteins was 2 μM in 25 mM Tris-HCl (pH 7.5), 100 mM NaF. Spectra were acquired at a resolution of 1 nm, with time per points set at 1 s and a bandwidth of 1 nm. All spectra were corrected from the contribution of the buffer and are an average of ten accumulations.

Preparation of bulk tRNA and in vitro transcribed Ec-tRNA^Glu

Bulk tRNA from various E. coli strains was purified as reported (35). The E. coli tRNA^Glu_UUC (Ec-tRNA^Glu) was synthesized in vitro by T7-RNA polymerase transcription as described in (36). Before use, the tRNA transcript was refolded by heating at 65°C for 15 min then 45°C for 15 min, and finally cooling at 4°C for 30 min.

In vitro enzyme assay

Holo-MnmA (1 or 10 μM) and Ec-tRNA^Glu_UUC (15 μM) were incubated at 37°C in 100 μl of 50 mM Tris-HCl (pH 7.5), 200 mM NaCl in the presence of 0.25 mM ATP, 2.5 mM MgCl₂ and 1 mM Na₂S under anaerobic conditions for 60 min. The reaction was stopped by adding 1 μl of 3 M formic acid, followed by the addition of 3 μl of 1 M Tris (pH 8.5) to adjust the pH of the solution to 6.5.

tRNA digestion and analysis of modified nucleosides

tRNA (15 μM) was digested overnight in 100 μl of 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 0.1 mM ZnSO₄ at 37°C by nuclease P1 (two units, Sigma) followed by the addition of alkaline phosphatase for 2 h at 37°C (1 unit, Sigma). HPLC-tandem mass spectrometry analyses were performed with an ExionLC chromatographic system coupled with a QTRAP6500+ mass spectrometer (AB SCIEX INSTRUMENTS) equipped with a Turbo Spray IonDrive source used in the positive ionization mode. HPLC separation was carried out with a 2 × 150 mm, 2.7 μm Poroshell HPH-C18 column (Agilent, France) at 0.4 ml min⁻¹ and at 35°C. A linear gradient of 0–15% acetonitrile in 0.1% formic acid over 7 min was used as the mobile phase. Mass spectrometry detection was carried out in the multiple reactions monitoring mode to obtain high sensitivity and specificity. The transitions used to quantify s²U (or s⁴U that has a similar fragmentation pattern) were 261→129 and 261→112, corresponding to the loss of ribose. Under the HPLC conditions used, s²U is eluted faster (4.3 min) than s⁴U (4.6 min). Quantification was performed by external calibration. The 5-methylaminomethyl-2-thiouridine (mnm⁵s²U) and 5-carboxymethylaminomethyl-2-thiouridine (cmnm⁵s²U) standards were synthesized via nucleophilic substitution of 5-pivaloyloxymethyl-2-thiouridine with methylamine or tetrabutylammonium salt of glycine, respectively, according to the previously described procedures (37). The structure and homogeneity of both nucleosides were confirmed by ¹H and ¹³C NMR, mass spectrometry and reversed-phase HPLC analysis.

Characterization of the [4Fe–4S] cluster by UV-Visible and EPR spectroscopies

UV-visible absorption spectra were recorded in quartz cuvettes (1 cm optic path) under anaerobic conditions in a glove box on a Cary 100 UV-visible spectrophotometer equipped with optical fibers.

The EPR continuous wave measurements were performed on a spectrometer Bruker ELEXSYS-E500 operating at 9.38 GHZ, equipped with SHQE cavity cooled by a helium flow cryostat ESR 900 Oxford Instruments. The EPR spectra of frozen solution of 220 μM E. coli MmnA with reconstituted cluster, without treatment and reduced with 5.5 mM dithionite in 50 mM Tris-HCl (pH 7.5), 200 mM NaCl was recorded at 10, 15 and 20 K, under nonsaturating conditions and using the following parameters: a microwave power of 2–10 mW, a modulation amplitude of 0.4 mT, a modulation frequency of 100 kHz and an accumulation of 4 scans. For the quantification of unpaired spins, a Cu-EDTA standard sample (200 μM) was used. The simulation of the EPR spectrum was performed with the Easyspin software (http://www.easyspin.org/).

Escherichia coli MnmA binds a [4Fe–4S] cluster

Escherichia coli MnmA with an N-terminal histidine tag was purified under aerobic conditions using Ni-affinity chromatography. The faint brownish color and the weak absorbance at 410 nm suggested the presence of an [Fe–S] cluster (Supplementary Figure S3A). The tag was then removed using the H3C protease and the protein was further purified by size-exclusion chromatography, which led to a mixture of monomeric and dimeric species, as shown by SEC-MALS analysis (Supplementary Figure S3B and S3C). Because the monomer/dimer ratio increases with DTT concentration (Supplementary Figure S3B), the dimer is formed by disulfide bonds between two monomers. As-purified MnmA retained a faint brownish color and its UV-visible spectrum exhibited a band at around 410 nm (Supplementary Figure S3D). These unexpected features, which disappeared upon further aerobic purification, strongly suggested the presence of an [Fe–S] cluster within the as-purified protein, which would be destroyed by air during purification, and prompted us to evaluate the potential of MnmA to assemble a well-defined cluster. For that purpose, in vitro reconstitution of the cluster was carried out by anaerobically treating the protein with ferrous iron and l-cysteine in the presence of a cysteine desulfurase, as a source of sulfur. The protein was then purified by size-exclusion chromatography, also under anaerobic conditions, leading to a homogeneous brownish protein, indicative of the presence of an [Fe–S] cluster in the protein that was subsequently called holo-MnmA. SEC-MALS analysis (Supplementary Figure S3E) indicated that holo-MnmA is almost exclusively in the monomeric form in solution, the molar mass of the protein being 40.0 ± 2.6 kDa, close to the theoretical molar mass of the monomer (41 kDa).

Quantification of the protein-bound iron and sulfur content gave 3.02 ± 0.2 Fe and 3.9 ± 0.3 S per MnmA monomer, consistent with the presence of one [4Fe–4S] cluster per monomer. Accordingly, the UV-visible spectrum of purified holo-MnmA displays a broad absorption band at around 410 nm that is specifically characteristic for the presence of a [4Fe–4S] cluster (Figure 2A). Anaerobic addition of dithionite to holo-MnmA led to a rapid decrease in the intensity of the 410 nm absorption band, suggesting a fast reduction of the cluster (Figure 2A). The frozen solutions of holo-MnmA, before and after reduction, were analyzed by continuous wave EPR spectroscopy at different temperatures (Figure 2B and Supplementary Figure S3F). No EPR signal was observed for holo-MnmA solution, suggesting an S = 0 [4Fe–4S]²⁺ state, and excluding S = 1/2 paramagnetic clusters such as [4Fe–4S]⁺ or [3Fe-4S]⁺. Upon reduction, the EPR spectrum of the protein recorded at 15 K exhibited a signal with a rhombic g tensor (g_x = 1.88, g_y = 1.92, g_z = 2.045) characteristic of the S = 1/2 [4Fe–4S]⁺ state (Figure 2B). Temperature variations allowed to discriminate between a [2Fe-2S]⁺ cluster and a [4Fe–4S]⁺ cluster because of their different spin relaxation behaviors (38). At 40 K, the EPR lines of the reduced holo-MnmA sample broadened beyond recognition (Supplementary Figure S3F), in agreement with the presence of a [4Fe–4S]⁺ cluster, which usually cannot be detected above 35 K, in contrast to [2Fe–2S]⁺ clusters. Quantification with Cu-EDTA standard solution indicated the presence of around 0.5 spin per holo-MnmA monomer.

Figure 2.

Spectroscopic and enzymatic characterizations of E. coli MnmA. (A) UV-visible spectra of 40 μM apo-MnmA (dotted line), 40 μM holo-MnmA (thick line) and reduced holo-MnmA after 20 min of incubation with 1 mM dithionite (dashed line). The spectra were recorded with 40 μM protein in 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, 5 mM DTT in 1 cm (apo-MnmA and reduced holo-MnmA) or 1 mm (holo-MnmA) pathlength cuvettes and normalized. (B) X-band EPR spectrum (10 mW microwave power; modulation amplitude of 0.4 mT) of 220 μM holo-MnmA reduced with 5.5 mM dithionite at 15 K. The experimental (black solid line) and simulated (dashed line) spectra are superimposed. The cluster was simulated with the following values of the g-tensor: g_x = 1.880, g_y = 1.920, g_z = 2.045 and Gaussian distribution deviations σ(g_x) = 0.04, σ(g_y) = 0.02, σ(g_z) = 0.04. (C) In vitro tRNA sulfuration activity tests of MnmA under anaerobic conditions. After tRNA digestion, s²U was separated by HPLC-MS/MS and quantified using a synthetic s²U standard. The data shown are mean values based on three different experiments, with the standard error of the mean indicated as a bar. In vitro transcribed Ec-tRNA^Glu (15 μM) was incubated for 1 h at 37°C in 50 mM Tris (pH 7.5), 200 mM NaCl with apo or holo-MnmA (wild-type or mutant) in the presence or absence of 1 mM Na₂S, 2.5 mM MgCl₂, 0.25 mM ATP. Ec-tRNA^Glu alone 15 μM (1), apo-MnmA 1 μM (2) or 10 μM (3), 1 μM holo-MnmA and no Na₂S (4), no MgCl₂ (5) or no ATP (6), holo-MnmA 1 μM (7) or 10 μM (8), 1 μM holo-MnmA_D99A-C102A mutant (9) or 1 μM holo-MnmA_{D99A-C102A-C199A} mutant (10). (D) HPLC MS/MS detection of s²U and s⁴U (m/z 261). Samples were analyzed using the two transitions 261→129 and 261→112; the most intense one corresponds to the loss of the ribose moiety. Left: mixture of s²U and s⁴U synthetic standards (0.5 pmole injected); middle: Ec-tRNA^Glu alone after hydrolysis; right: Ec-tRNA^Glu (15 μM) after incubation with holo-MnmA (10 μM) in the presence of Na₂S, ATP, MgCl₂ and hydrolysis.

The [4Fe–4S] cluster is required for the tRNA sulfuration activity of E. coli MnmA

The sulfuration activity of E. coli MnmA was tested using in vitro transcribed Ec-tRNA^Glu as a substrate in the presence of ATP, Mg²⁺ and inorganic sulfide, a sulfur donor often used in enzyme sulfuration assays (39–41). The thiolated nucleosides were quantified after digestion of the tRNA products by nuclease P1 and alkaline phosphatase. The s²U product was first identified by its elution position after HPLC-coupled mass spectrometry (MS), then quantified by MS/MS using a synthetic s²U standard (Figure 2C). The fragment spectrum corresponding to the loss of sugar (-132) was monitored (Figure 2D). Holo-MnmA (1 μM) was able to catalyze sulfuration of tRNA (15 μM) in contrast to apo-MnmA (1 or 10 μM) (Figure 2C). Moreover, the sulfur atom that was inserted into the nucleoside is derived exclusively from the inorganic sulfide salt, and not from the [4Fe–4S] cluster, since no sulfuration could be observed in the absence of sulfide. These control experiments indicated that the cluster and inorganic sulfide were required for the tRNA sulfuration activity of MnmA. Moreover, the amount of product formed increased with enzyme concentration (1 μM versus 10 μM in Figure 2C). Mg-ATP was also required, in agreement with its role in activating the substrate via adenylation of the C2 oxygen atom (Figure 1A and Supplementary Figure S2) (8).

MnmA is necessary for E. coli to resist to stress

It was previously shown that mnmA mutation in E. coli leads to reduced growth rate (14). We reproduced this observation (Figure 3A) and also noticed that the mutant exhibited a small colony phenotype when plated on rich medium (Figure 3C). We constructed the ΔmnmA strain and checked by HPLC-MS that the bulk tRNA from this strain did not contain mnm⁵s²U nor cmnm⁵s²U (Table 1). By monitoring growth in liquid culture, we found that the ΔmnmA strain showed slower growth in the presence of a mild acid stress (i.e. growth medium at pH 4.5) (Figure 3B) and that it exhibited hypersensitivity to H₂O₂ (Figure 3C). Note that survival plate assays run in parallel showed that the viability of the mnmA strain was not impaired at pH 4.5.

Figure 3.

Phenotypes of ΔmnmA and derivative strains. (A) Growth in LB pH 7.0 medium. Strains studied are wild-type (wt) (FBE051), ΔmnmA (FBE584), ΔiscS (FBE653) and ΔiscS ΔmnmA (FBE703). Data are representative of three independent experiments (n = 3). (B) Growth in LB–HCl (pH 4.5). Strains studied are wild-type (wt) (FBE051), ΔmnmA (FBE584), ΔiscS (FBE653) and ΔiscS ΔmnmA (FBE703), (n=3). (C) Hypersensitivity of ΔmnmA to H₂O₂ (1 mM). Strains tested are wt, ΔmnmA, ΔiscS and ΔiscS ΔmnmA. Each spot represents a 10-fold serial dilution (n = 3). (D) The ΔmnmE and ΔmnmA mutations are lethal synthetic in E. coli. The RExBADmnmA ΔmnmE strain (FBE601) was grown in LB rich medium without (- ara) or with (+ ara) 0.2% l-arabinose, (n = 3).

Table 1.

Analysis of s²C, mnm⁵s²U and cmnm⁵s²U content (in pmoles) of bulk tRNA (15 μM) from various strains (mean of three measurements)

	s2C	cmnm5s2U	mnm5s2U
Wild-type	9.5 ± 1.7	107.6 ± 39.8	1080.7 ± 647.0
ΔmnmA	14.8 ± 2.2	0.0 ± 0.1	1.9 ± 1.0
ΔiscUA ΔsufABCDSE	0.0 ± 0.0	64.5 ± 20.9	1421.9 ± 601.7
ΔttcA	0.4 ± 0.1	68.0± 44.4	1756.1 ± 1003.5

U34-tRNA modification pathway in E. coli

Biochemical studies together with a genetic approach coupled with mass spectrometry (ribonucleome analysis) have shown that MnmA receives sulfur from the IscS cysteine desulfurase (6,22) and the TusABCDE sulfur relay pathway (7). Thus, the prediction is that the ΔiscS strain should exhibit defects similar to those of the ΔmnmA strain, as shown in S. enterica (16). The ΔiscS mutation was found to cause extreme sensitivity to pH 4.5, even to a larger extent than that caused by the ΔmnmA mutation (Figure 3B). Such a feature has never been described for IscS so far. Introducing the ΔmnmA mutation in a ΔiscS background did not worsen the defect (Figure 3B), in line with the idea that MnmA depends upon IscS for getting sulfur. Similarly, when challenging the strains with H₂O₂ (Figure 3C), the ΔiscS strain displayed hypersensitivity comparable to that of the ΔmnmA strain, and the combination of both deletions failed to enhance the defect.

Besides hypersensitivity to H₂O₂, we wished to set-up another MnmA-associated phenotypic test. For this, we took advantage of the synthetic lethality between mnmA and mnmE mutations. MnmE, along with MnmG, adds the aminomethyl and carboxymethylaminomethyl groups to the C5 position of U34 (42). We constructed a strain named RExBADmnmA, in which expression of the mnmA gene was under the control of an arabinose inducible promoter and tested whether it could accommodate a ΔmnmE mutation. We observed that the ΔmnmE mutation could be introduced by transduction only when arabinose was added to the medium (Figure 3D). This confirmed that a strain lacking both mnmA and mnmE is not viable, as previously shown by using a strain, in which the mnmA gene was deleted and the mnmE gene put under the control of an arabinose inducible promoter (42).

Then, we reasoned that if iscS was required for MnmA to function, then a strain carrying both ΔiscS and ΔmnmE should not be viable. Therefore, we asked whether we could introduce a ΔiscS mutation into a ΔmnmE mutant. The result was negative, suggesting that a ΔiscS mutation recapitulated the ΔmnmA mutation, in agreement with IscS providing the sulfur atom to MnmA.

Taken together, these genetic analyses showed that the mnmA gene is crucial for E. coli to grow under both balanced and stress conditions, and established its close functional interaction with iscS, in agreement with early studies in S. enterica (23).

The ligands of the [4Fe–4S] cluster are most likely Asp99, Cys102 and Cys199

The D99A, C102S and C199A mutants of E. coli MnmA were reported to lack in vitro U34-tRNA sulfuration activity using cysteine as the sulfur donor, and the IscS and Tus proteins as sulfur carriers (8). We produced the double D99A-C102A and the triple D99A-C102A-C199A mutants (Supplementary Figure S4). Both mutants were correctly folded, as shown by their circular dichroism spectrum, which did not differ from that of the wild-type protein (Supplementary Figure S4F). After cluster reconstitution, they both almost completely lost the absorption band at 410 nm (Supplementary Figure S4G) and were shown to contain very little Fe, namely 0.46 ± 0.03 and 0.25 ± 0.05 iron per MnmA monomer, respectively. Finally, they did not exhibit any catalytic activity (Figure 2C, columns 9 and 10) in standard in vitro assays with sulfide used as a sulfur donor. Therefore, the mutagenesis results point out Asp99, Cys102 and Cys109 as being involved in cluster binding and catalysis (Figure 1B) (8). But one cannot unambiguously conclude on the in vitro analysis of only the double and triple mutant variants that all three residues contribute to ligation. Yet, we did not analyze the single variants because, in many instances, a single mutation of a cluster ligand was shown to be not sufficient to disrupt cluster ligation (43,44). Therefore, in addition, we tested the functional importance of the Asp99, Cys102 and Cys109 residues by changing each of them to alanine and testing the activity of the resulting single mutants in vivo using the series of phenotypic tests described above (Supplementary Figure S5). When expressed in trans from plasmids, all mutated variants, namely D99A, C102A and C109A failed to (i) suppress growth rate defects of the ΔmnmA recipient in rich medium (Supplementary Figure S5A, top), (ii) suppress hypersensitivity of the ΔmnmA strain to acid (Supplementary Figure S5A, bottom) and oxidative stress (Supplementary Figure S5B) and (iii) allow growth in the absence of arabinose in the RExBADmnmA ΔmnmE containing background (Supplementary Figure S5C). Note that expression of all three alleles gave rise to large amounts of soluble protein so that the possibility of a destabilizing effect of the introduced mutations was ruled out (Supplementary Figure S5D). Hence, based upon their inability to complement different defects of the ΔmnmA mutant, we concluded that substitution of Asp99, Cys102 or Cys109 by alanine yielded a nonfunctional MnmA protein. Altogether, the simplest interpretation of the in vitro characterization of the mutants and the in vivo complementation assays is that Asp99, Cys102 and Cys109 are the ligands of the [4Fe–4S] cluster.

Biosynthesis of mnm⁵s²U and cmnm⁵s²U in tRNAs occurs in the absence of the ISC and SUF [Fe–S] cluster biogenesis systems

Since maturation of [Fe–S] proteins in E. coli relies on the two ISC and SUF machineries, we analyzed the s²U content in bulk tRNA from the ΔiscUA ΔsufABCDSE strain (29) after tRNA digestion and HPLC-MS analysis (Table 1) to check whether MnmA maturation also uses these pathways. The ΔiscUA ΔsufABCDSE strain is not viable but can be grown by introducing the eukaryotic [Fe–S]-independent mevalonate-dependent isoprenoid biosynthesis pathway (29) to supply isopentenyl diphosphate, which is essential and normally synthesized by the [Fe–S] dependent IspG/IspH proteins. First, we observed that s²C was absent in bulk tRNA from the mutated strain (Table 1), in full agreement with TtcA, the unique enzyme responsible for s²C formation, requiring a [4Fe–4S] cluster (39) assembled by the ISC system (24). Then, we compared the content in mnm⁵s²U and cmnm⁵s²U in bulk tRNA from a wild-type and the ΔiscUA ΔsufABCDSE strains. Unexpectedly, we found that the amounts of mnm⁵s²U and cmnm⁵s²U from bulk tRNA in the mutant strain were comparable to those in the wild-type strain. This suggests that [Fe–S] cluster biogenesis of the E. coli MnmA protein can occur in the absence of the ISC and SUF systems.