Mycobacterium tuberculosis SufR responds to nitric oxide via its 4Fe–4S cluster and regulates Fe–S cluster biogenesis for persistence in mice

Abstract

The persistence of Mycobacterium tuberculosis (Mtb) is a major problem in managing tuberculosis (TB). Host-generated nitric oxide (NO) is perceived as one of the signals by Mtb to reprogram metabolism and respiration for persistence. However, the mechanisms involved in NO sensing and reorganizing Mtb's physiology are not fully understood. Since NO damages iron-sulfur (Fe–S) clusters of essential enzymes, the mechanism(s) involved in regulating Fe–S cluster biogenesis could help Mtb persist in host tissues. Here, we show that a transcription factor SufR (Rv1460) senses NO via its 4Fe–4S cluster and promotes persistence of Mtb by mobilizing the Fe–S cluster biogenesis system; suf operon (Rv1460-Rv1466). Analysis of anaerobically purified SufR by UV–visible spectroscopy, circular dichroism, and iron-sulfide estimation confirms the presence of a 4Fe–4S cluster. Atmospheric O₂ and H₂O₂ gradually degrade the 4Fe–4S cluster of SufR. Furthermore, electron paramagnetic resonance (EPR) analysis demonstrates that NO directly targets SufR 4Fe–4S cluster by forming a protein-bound dinitrosyl-iron-dithiol complex. DNase I footprinting, gel-shift, and in vitro transcription assays confirm that SufR directly regulates the expression of the suf operon in response to NO. Consistent with this, RNA-sequencing of MtbΔsufR demonstrates deregulation of the suf operon under NO stress. Strikingly, NO inflicted irreversible damage upon Fe–S clusters to exhaust respiratory and redox buffering capacity of MtbΔsufR. Lastly, MtbΔsufR failed to recover from a NO-induced non-growing state and displayed persistence defect inside immune-activated macrophages and murine lungs in a NO-dependent manner. Data suggest that SufR is a sensor of NO that supports persistence by reprogramming Fe–S cluster metabolism and bioenergetics.

Graphical abstract

Highlights

• •Mycobacterium tuberculosis (Mtb) upregulates the suf operon for Fe–S cluster biogenesis in response to nitric oxide (NO).
• •A transcription factor SufR senses NO via its 4Fe–4S cluster and directly regulates the expression of suf operon.
• •SufR-regulated Fe–S cluster biogenesis confers respiratory and redox features that promote recovery of Mtb from NO stress.
• •SufR activity is required to support the NO-dependent persistence of Mtb in macrophages and mice.

1. Introduction

About 90% people infected with Mycobacterium tuberculosis (Mtb) remain asymptomatic for tuberculosis (TB). This indicates that the host immunity effectively suppresses the bacterial replication without eradicating the pathogen. The ability of the host to produce nitric oxide (NO) by an inducible nitric oxide synthase (iNOS) is known to modulate immunity [1] and microbial physiology [2], thereby controlling diverse infections [3,4] including TB [5]. Upon infection with Mtb, lesional macrophages in humans and macaques express functional iNOS [6,7], and exhaled breath of TB patient contains NO [8]. Importantly, iNOS activity seems to control TB in humans [9,10]. Mechanistically, NO inhibits respiration and arrests growth of Mtb [2]. The bacterial mechanisms responsible for sensing NO and mobilizing adaptation programs are poorly understood.

Previous models suggested that Mtb exploits a three-component system, DosR/S/T, to induce transcriptional changes, growth arrest, and a switch from aerobic to anaerobic respiration under NO stress [11,12]. However, the NO-mediated activation of DosR regulon is transient [13] and other gases, such as carbon monoxide (CO) and oxygen (O₂)-limitation, similarly activate the DosR pathway [14,15]. Therefore, the mechanism that specifically coordinates changes in metabolism, growth, and respiration of Mtb in response to NO is not fully understood. In this context, a recent study demonstrated active degradation of several iron-sulfur (Fe–S) cluster proteins coordinating respiration, central metabolism, and amino acid biosynthesis in NO-treated Mtb [13]. Importantly, a seven-gene operon, the Suf system (Rv1460-Rv1466) that is likely involved in Fe–S cluster biogenesis/repair [16], showed prolonged, elevated expression in response to NO [13]. Interestingly, excluding Rv1460 (sufR), other genes of the suf operon are essential [16,17]. Since the reaction of NO with Fe–S clusters generates a lethal dinitrosyl-iron dithiol complex (DNIC) [18], active degradation and calibrated regeneration of Fe–S clusters via the suf operon represents a potential adaptive defense against NO. Despite these reports, how Mtb senses NO and regulates Fe–S cluster biogenesis remains uncharacterized. Filling this knowledge gap is crucial for understanding the molecular underpinning of Mtb persistence.

Recently, the first protein of the Suf system (SufR; Rv1460) has been shown to coordinate a 2Fe–2S and function as a putative regulator of the suf operon [19]. However, the C-terminus of SufR contains five cysteine residues (C₁₇₅–C₁₇₉–C₁₉₂–C₂₁₈–C₂₂₀), which suggests serving as a ligand for a 4Fe–4S cluster [20,21]. The presence of a 2Fe–2S cluster was demonstrated in vitro that invariably results in poor Fe–S cluster incorporation [19,22]. Also, without a functional assay (e.g., DNA binding), the authenticity of the 2Fe–2S form of SufR cannot be validated. Lastly, phenotypic characterization of the sufR mutant remained ambiguous as one study showed the requirement of SufR for growth under standard culture conditions but not under stress (e.g., iron-limitation) [19]. In contrast, other studies reported the opposite findings [22,23]. Several discrepancies thus exist in the previously reported characterization of SufR, and consequently, in our understanding of its physiological function. In this study, we performed biochemical, biophysical, and genetic characterization of SufR. We show that SufR contains a NO-sensitive 4Fe–4S and regulates Fe–S biogenesis, redox balance, and bioenergetics of Mtb to promote persistence in response to NO in vivo.

2. Material and methods

2.1. Culture conditions

The Mtb H37Rv, MtbΔsufR, and sufR-Comp were grown in Middlebrook 7H9 broth (Becton, Dickinson and Company (BD), USA) medium supplemented with 0.2% glycerol, 0.5% BSA, 0.2% dextrose, and 0.085% NaCl (ADS) with 0.05% Tween 80 as described previously [23]. For culturing on solid medium, Mtb strains were cultured on 7H10/7H11 agar medium (Becton, Dickinson and Company (BD), USA) supplemented with 1x OADC (Becton, Dickinson and Company (BD), USA) and 0.2% glycerol. E. coli cultures were grown in LB medium (HIMEDIA, India). Whenever required, antibiotics were added to the culture medium (for E. coli, 100 μg/mL kanamycin (Amresco, USA) and 150 μg/mL hygromycin (Sigma-Aldrich, India); for Mtb strains, 50 μg/mL hygromycin).

2.2. Generation of the sufR-complemented strain

Primer pair: 5’-ATCGAAGCTTGTCCGTCCCTGCCGATCTCAC-3’ and 5’- ATGCGGTACCAACGCTCTTGCTGGCCTCTG-3’ having restriction site HindIII and KpnI were used to amplify sufR from wild type Mtb. PCR was performed and the product contained the wild-type sufR gene, encoded by Rv1460 with the promoter region (567 bp upstream), was cloned into pCV125 (integrated plasmid). Sequence was verified and the constructed plasmids were transformed into the MtbΔsufR strain. Expression of complemented strain was confirmed by qRT-PCR.

2.3. RNA sequencing experiments

The Mtb strains were grown to an O.D.₆₀₀ of 0.4 and exposed to 0.5 mM diethylenetriamine-nitric oxide (DETA/NO [Sigma-Aldrich, India]) for 4 h at 37 °C. The experiment was carried out with three independent biological replicates. Total RNA extraction was conducted using the FastRNA® Pro Blue Kit (MP Biomedicals, USA) in accordance with the manufacturer's instruction and further purified using RNeasy spin columns (Qiagen, USA) as described [23]. Following purification, the RNA was quantified and assessed for purity by a 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). RNA samples with an RIN (RNA Integrity Number) value > 8 were processed further for sequencing. Ribosomal RNA (16s and 23s rRNA) was removed by hybridization with magnetic beads-coupled oligonucleotide (MICROBExpress Kit, Life Technologies, USA) and concentration of enriched mRNA was quantified by Qubit RNA HS Assay Kit (Life Technologies, USA). RNA-seq was performed as described [23]. In brief, libraries were prepared using NEB Next Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs, USA), according to manufacturer's instructions. The library size distribution and quality were assessed using a high sensitivity DNA Chip (Agilent Technologies, USA) and sequenced in HiSeq 2500 platform (Illumina, USA) sequencer using 1X50 bp single-end reads with 1% PhiX spike-in control.

2.4. Differential gene expression and statistical analysis for RNA-Seq

Raw reads were obtained for Mtb H37Rv strain as fastq files. The reference genome sequence (.fna) and annotation (.gff) files for the same strain (accession number: NC_000962.3) were downloaded from the ncbi ftp website ("ftp.ncbi.nlm.nih.gov"). The annotation file was customized with the addition of annotations for non-coding RNAs [24]. The format of the annotation file (.gff) was changed to.bed format using an in-house python script. The raw read quality was checked using the Fast QC software (version v0.11.5; http://www.bioinformatics.babraham.ac.uk/projects/fastqc). BWA (version 0.7.12-r1039) [25] was used to index the reference genome. Reads with raw read quality >= 20 were aligned using BWA aln -q option. SAMTOOLS (version 0.1.19-96b5f2294a) [26] was used to filter out the multiply mapped reads. BEDTOOLS (version 2.25.0) [27] was used to calculate the reads count per gene using the annotation file (.bed). The normalization and differential gene expression analysis for the conditions were carried out using edgeR as mentioned previously [28]. Genes with at least 10 reads were selected for each comparative analysis. DGE analysis was done in RStudio (1.1.447) with R version 3.4.4 (http://www.rstudio.com/).

2.5. Aconitase assay

The activity of aconitase (Acn) was measured by monitoring the disappearance of cis-aconitate at 240 nm in a UV spectrophotometer (Thermo Scientific Biomat 3S, USA) as described [29]. One unit (U) of aconitase activity is defined as 1 μmol cis-aconitate formed or converted per minute. Reaction mixtures (1 mL) for Acn contained 25 mM Tris-HCl (pH 8.0), 100 mM NaCl, and 50 μg Mtb cell lysates. Reactions were initiated by adding 0.15 mM cis-aconitate and monitored by following the disappearance of cis-aconitate at 240 nm after every 15 s for total 30 min. Absorbance at 240 nm was plotted against time. Acn activity was calculated from linear portion of the curve in initial 5 min when reaction follows 0^th order of reaction. An extinction coefficient of 3,500 M^-1cm^-1 was used to calculate the rates.

2.6. Western blot

Whole cells lysate (50 μg) was separated on 12% SDS-PAGE and then transferred onto a PVDF membrane (GE Healthcare, Piscataway, NJ, USA). Membrane were blocked in 5% (w/v) nonfat dry milk and incubated for 3 h at room temperature with primary antibody (Acn and Cbs 1:10000 dilution). After washing with 1X TBST, membranes were incubated in goat anti-rabbit IgG HRP-conjugated secondary antibody (1:10000 dilution) for 1 h. The autoradiography signals were visualized using ECL advance Western blotting detection kit (BioRad, USA).

2.7. OCR and ECAR measurements

The Mtb strains adhered to the bottom of a XF cell culture microplate (Agilent technologies, USA), at 2X10⁶ bacilli per well by using Cell-Tak (a cell adhesive). Oxygen Consumption Rate (OCR) and Extracellular Acidification Rate (ECAR) were measured using Agilent XF Extracellular Flux Analyser. Assays were carried out in unbuffered 7H9 media (pH 7.35) with glucose 2 mg/mL as carbon source. Basal OCR and ECAR were measured for initial 21 min before the automatic addition of freshly prepared DETA-NO (0 mM, 0.25 mM, 0.5 mM and 1 mM) in 7H9 unbuffered media, through port A of cartridge plate. Three measurements were taken after 1 h of incubation. Carbonyl cyanide 3-chlorophenylhydrazone (CCCP) (Sigma-Aldrich, India) was added at 10 μM concentration to achieve maximum rate of respiration. Raw data of OCR and ECAR was CFU normalized for 2X10⁶ CFU/well. Spare respiratory capacity, was calculated from % OCR value, by subtracting third basal reading (normalized as 100%) from first point after CCCP addition.

2.8. CellRox Deep Red staining and flow cytometry

Logarithmically growing Mtb strains were treated with DETA-NO (0 mM, 0.25 mM, 0.5 mM and 1 mM) and then incubated at 37 °C with shaking for 2 h. 200 μL cells were treated with CellROX® Deep Red reagent (Thermo Fisher, USA) at a final concentration of 5 μM for 30 min at 37 °C and analyzed on BD FACSVerse flow cytometer with 640/665 nm excitation and emission respectively. We collected 5000-10,000 events for each sample wherever possible.

2.9. Animal experiments

For the chronic model of infection, 5- to 6-week-old female BALB/c, C57BL/6 and iNOS^-/- mice (n = 6 per group) were infected by aerosol with approximately 100 bacilli per mouse with the Mtb strains using a Madison chamber aerosol generation. At indicated times post infection, mice were euthanized, and the lungs were harvested for bacillary load, tissue histopathology analysis, and pathological scoring as described [23]. The remaining tissue samples from each mouse were homogenized and bacillary load was quantified by plating serial dilutions of tissue homogenates onto Middlebrook 7H11-OADC agar plates supplemented with lyophilized BBL MGIT PANTA antibiotic mixture (polymyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin, as supplied by BD; USA). Colonies were observed and counted after 4 weeks of incubation at 37 °C.

2.10. Construction of Mtb SufT knockdown strain

For construction of Mtb SufT knock down strain (SufT-KD), we followed CRISPR interference (CRISPRi) technology as described previously [30]. Anhydrotetracycline (ATc, Cayman, USA) 200 ng/mL was added for the induction of sufT specific–guide RNA (sgRNA) and dCas9 every 48 h. SufT-KD culture was divided equally when A₆₀₀ reached 0.1–0.2 and cultured in the presence or absence of ATc. The dCas9 was expressed by pRH2502 from a TetR-regulated uvtetO promoter and sgRNA in pRH2521 under the control of a TetR-regulated smyc promoter (Pmyc1tetO). To deplete sufT, gene specific sgRNAs were designed for two regions between 98-127 bp and 171–191 bp of sufT and cloned in pRH2521. Depletion of sufT was verified by qRT-PCR and based on the significant repression of sufT, we chose sgRNA targeting sufT region between 98 and 127 bp for further study. For qRT-PCR analysis, total RNA was extracted from SufT-KD and control strains after 24 h of ATc treatment. After DNase treatment, total 600 ng of RNA was used for cDNA synthesis by using Random hexamer oligonucleotide primer (iScript Select cDNA Synthesis Kit, BioRad, USA). Gene specific primers (Table S2) and iQ SYBER Green Supermix (BioRad, USA) were used for RT-PCR (StepOne Plus, Thermo, USA). Gene expression was normalized to Mtb 16S rRNA expression level.

2.11. NO exposure and recovery

Logarithmically grown Mtb strains (OD₆₀₀ of 0.8) were diluted to an OD₆₀₀ of 0.1 and exposed to six doses of 100 μM DETA-NO, once every 6 h. Recovery from exposure to DETA-NO was monitored by recording the OD₆₀₀ after each addition of DETA-NO, as well as every 24 h after addition. Two biological replicates were used for each strain. For single dose experiments, different doses of DETA-NO (0.5 mM, 1.25 mM, and 2.5 mM) were added. After 4 and 24 h cells were harvested, washed with 1X PBS and plated on OADC-7H11 plates. Colonies were counted after 3–4 weeks of incubation at 37 °C.

2.12. Cell line experiments

RAW264.7 murine macrophage cell line was activated by treatment with IFNγ (100U/mL, Invitrogen, USA) for 12 h before infection and LPS (100 ng/mL, Sigma-Aldrich, India) for 2 h before infection. Activated RAW264.7 macrophages were infected with wt Mtb, MtbΔsufR, and sufR-Comp strains at multiplicity of infection (MOI) 2 for 2 h, followed by washing thoroughly to remove extracellular bacteria with warm DMEM medium and suspended in the same containing 10% FBS. For CFU determination, macrophages were lysed using 0.06% SDS-7H9 medium diluted in PBS/Tween and plated on OADC-7H11 at indicated time points. Colonies were counted after 3–4 weeks of incubation at 37 °C.

2.13. Purification of SufR under anaerobic conditions

The entire ORF of Mtb sufR (Rv1460) was PCR-amplified using gene-specific oligonucleotides (pET28asufRF and pET28asufRR; Table S2), digested with NdeI-HindIII, and ligated into similarly digested His-tag-based expression vector, pET28a (TAKARA BIO, Clontech Laboratories, CA, USA) to generate pET28a:SufR. A N-terminal histidine-tagged SufR was overexpressed in E. coli BL21 λDE3 by 0.6 mM IPTG [Isopropyl β-d-1-thiogalactopyranoside; MP Biomedicals, USA (60 min, 30 °C)]. To facilitate Fe–S cluster formation, cultures were incubated on ice for 18 min prior to induction and were supplemented with 300 μM ferric ammonium citrate and 75 μM l-methionine (Amresco, USA) and purified as described [31]. Purification was performed under strict anaerobic conditions inside an anaerobic glove box (PLas-Labs, Lansing, MI, USA) maintaining ≈2.0 ppm O₂ by volume, and buffers and solutions were appropriately deoxygenated. The sufR gene on pET28a-SufR was mutated using oligonucleotide-based site-directed mutagenesis approach to create individual cysteine to alanine substitutions. After the PCR, DpnI was added into the reaction mixture to digest the wild-type plasmid that was used as the template. The reaction mixture containing the mutated sufR gene was used to transform E. coli BL21 λDE3. Sequences of primers used to create mutations are shown in (Table S2). Resulting clones were verified by sequencing, and the mutant Cys variants of the wt SufR were purified as described earlier. In order to generate apo-SufR, the holo-SufR was incubated with EDTA (Ethylenediaminetetraacetic acid, Sigma-Aldrich, India) and potassium ferricyanide in a molar ratio of protein: EDTA: ferricyanide in 1:50:20 at 25 °C and incubated for 20–30 min till the extensive loss of color. The solution was passed through PD10 desalting column and stored at −80 °C.

2.14. UV–Vis, CD analysis, and gel filtration of SufR

The UV–visible absorption spectroscopy was carried out in a Thermo scientific spectrophotometer (Thermo scientific, USA) at 25 °C. Absorption spectra of SufR WT (native/holo) and mutants were recorded immediately as the elution fractions were collected during the purification. In order to study the effect of air oxidation, on Fe–S cluster stability, freshly purified holo-SufR was transferred to an anaerobic quartz cuvette, exposed to air by opening the cap and mixing by pipetting for 2 min. The cuvette was then sealed and monitored by UV–visible spectroscopy over time (Thermo scientific, USA). To study the effect of NO, DTH and H₂O₂ on [4Fe–4S] cluster of SufR, the absorption spectra of freshly purified protein were recorded at indicated concentration and different time intervals in anaerobic quartz cuvette. CD measurements were conducted in a Jasco J-715 spectropolarimeter (Jasco, USA). Far-UV spectra were measured from 190 to 250 nm range and near-UV spectra from 300 nm to 650 nm range. Protein concentration used for the far-UV CD measurements was 10–20 μM and for the near-UV measurements was 140–150 μM. Cells of 1.0 cm path length were used for the measurements of the far- and near-UV spectra, respectively. Three repeat scans were obtained for each sample. The averaged baseline spectrum was subtracted from the averaged sample spectrum. The protein was dissolved in 5 mM phosphate buffer pH 7.4 and 150 mM NaCl. Results are expressed as molar ellipticity [θ] (deg cm² dmol^-1), calculated from the following formula [θ]λ = θ/[c]∗l∗10∗n, where θ is the measured ellipticity in degrees at wavelength λ, c is the protein concentration in mg/mL, l is the light path length in centimeters and n is the number of amino acids. CD Pro software was used to analyze the data.

To determine the molecular mass of apo- and holo-SufR protein, analytical size-exclusion chromatography experiments were performed with Superdex 200 increase, 10/300 GL analytical column (GE Healthcare Life Sciences, USA). The column was pre-equilibrated and eluted with the running buffer (10 mM phosphate buffer, 10% glycerol and 100 mM NaCl at pH 7.4) at a constant flow rate of 0.5 mL/min. Molecular mass of the proteins was determined by using gel filtration molecular mass standard (Carbonic Anhydrase, Albumin, Alcohol Dehydrogenase, β-Amylase, Apoferritin and Thyroglobulin). Ve/Vo was plotted as a function of log₁₀Mr of the standard protein where Ve is the elution volume of the protein, V_o is the void volume of the column and Mr is the molecular weight of the particular protein. Blue dextran was used to determine the void volume (Vo). Running buffers were purged with nitrogen gas and were degassed thoroughly to remove any dissolved atmospheric oxygen. The experiments were conducted under anaerobic conditions.

2.15. EPR spectroscopy of SufR

For EPR spectroscopy, holo SufR was treated inside the anaerobic glove box with aliquots of freshly prepared proline NONOate (Cayman Chemicals, Ann Arbor, MI, USA). Aliquots of SufR were placed in an anaerobic cuvette and titrated by injection with aliquots of a 2.5 mM stock solution of proline NONOate (Cayman chemicals, USA). Samples were then transferred to EPR tubes and immediately frozen in liquid nitrogen. EPR characteristics of NO-treated SufR were analyzed by subjecting the samples to continuous-wave spectrometers at liquid nitrogen temperature as described previously [32]. NO-treated SufR was analyzed by continuous-wave EPR on a perpendicular mode X-band EPR spectrometer operating at 100-kHz modulation frequency and equipped with liquid nitrogen cryostat and a dual mode X-band cavity (JES200 ESR spectrometer, JEOL, USA). Field calibration was done by using a standard NMR G meter. The background signal from the buffer was subtracted from the spectra.

2.16. Chemical analysis of iron and sulfide

The total iron content and acid-labile sulfide content of holo-SufR, was measured using a previously described procedure [33]. For iron estimation, freshly purified holo-SufR (0.1 mL) was heated at 95 °C degree for 30 min after treatment with 22% HNO₃ (0.1 mL). Samples were cooled to ambient temperature followed by addition of 0.6 mL ammonium acetate (7.5% w/v), 0.1 mL freshly prepared ascorbic acid (12.5% w/v) and 0.1 mL ferene (10 mM). The concentration of iron present in the protein was determined by measuring the absorbance of the product, iron-ferene complex at 593 nm, which was compared with a standard curve prepared from dilutions of freshly prepared Fe(III) solution in the range of 0–200 μM. To measure acid-labile sulphide content, freshly prepared Na₂S.9H₂O solution was used to prepared standard solution in the range of 52–260 μM of S²⁻. Protein sample/standard (200 μL) was mixed with 0.6 mL of zinc acetate (1% w/v) followed by addition of 50 μL of NaOH (12% w/v). After incubation of 15 min, 0.15 mL of N, N-dimethyl-p-phenylenediamine dihydrochloride (0.1% w/v dissolved in 5 M HCl) and 0.15 mL of freshly prepared 10 mM FeCl₃ (dissolved in 1 M HCl) was added. The reaction mixture was further incubated for 30 min at room temperature and the absorbance of the product, methylene blue, was measured at 670 nm. For iron and sulfide estimation three independent preparations of holo-SufR were analyzed to ensure the reproducibility and three dilution of each sample were considered.

2.17. Electrophoretic mobility shift assay (EMSA)

For EMSA, promoter fragment of sufR (∼170 bp) and blaC (∼ 100 bp) were PCR amplified from the Mtb genome. The 5’ end was labeled by [Υ-32_P]-ATP using T4 polynucleotide kinase (MBI Fermentas, USA) as per the manufacturer's instructions. The labeled oligonucleotides were passed through a 1 mL Tris-EDTA, pH 7.5 equilibrated Sephadex G-50 column and elute was vacuum dried. Blunt ended duplex DNA was prepared by annealing radiolabelled and complementary cold oligonucleotides in a 1:2 M ratio in 50 μL reaction system containing TE and 1X saline-sodium citrate buffer (3 M sodium chloride and 0.3 M sodium citrate). The reaction mixture was heated to 95 °C for 5 min and then cooled in a thermal cycler. Annealed mixtures were resolved by 10% native PAGE in 1X Tris-borate EDTA (TBE). Assembled DNA substrates were visualized on an X-ray film and substrates were subsequently purified by excising the respective desired bands. DNA substrates were eluted by incubating corresponding gel pieces in TE at 4 °C for 6 h. Binding reactions were performed in binding buffer (25 mM Tris-HCl, 1 mM DTT, 0.1 mg/mL BSA, 5 mM MgCl₂; pH 7.4) for 30 min at 4 °C and 6% polyacrylamide gel was used to resolve protein-DNA complexes in an anaerobic chamber by electrophoresis at constant voltage (50 V). 0.5 nM ³²P-labeled DNA substrates were incubated with indicated concentrations of various forms of SufR (apo, holo, H₂O₂-, and NO-treated). For competition with unlabeled DNA, fragments of sufR and blaC (∼100–200 bp upstream of translational start codon) were PCR amplified from the Mtb genome and used in various amounts to outcompete binding of holo-SufR to ³²P-labeled DNA fragments. Gels were exposed to auto radiographic film and visualized by a phosphor imager (Typhoon FLA-9000, GE Healthcare Life Sciences, USA). All the above reactions and separations were performed under anaerobic condition inside glove box (Plas-Labs, Lansing, MI, USA).

2.18. In vitro transcription assays

The DNA templates (170 bp) including the sufR promoter regions were PCR amplified using primers PsufR F1/PsufR R1 (Table S2). The amplicons (50 nM) were pre-incubated with different concentration of holo- and apo-SufR in the transcription buffer (50 mM Tris HCl, (pH 8.0), 10 mM magnesium acetate, 100 mM EDTA, 100 mM DTT, 50 mM KCl, 50 mg/mL BSA, and 5% glycerol) for 30 min at room temperature. Single-round transcription reactions were initiated with the addition of 100 nM Mtb RNAP-σA holo enzyme, 100 μM NTPs, 1 μCi [α-32P]-UTP, 50 μg mL⁻¹ heparin and incubated at 37 °C for 20 min. The reactions were terminated by addition of 2X formamide dye (95% formamide, 0.025% (w/v) bromophenol blue, 0.025% (w/v) xylene cyanol, 5 mM EDTA and 0.025% SDS and 8 M urea) and heated at 95 °C for 5 min followed by snap chilling in ice for 2 min. The transcripts were resolved on an 8% TBE-urea-PAGE gel. To study the effect of NO, transcription assays were performed as mention above after treating the holo-SufR with proline NONOate for 5 min followed by purification. All the treatments and the reactions were performed inside anaerobic glove box under anoxic condition.

2.19. DNase I footprinting

5 nM of ³²P-labeled DNA substrate was incubated with increasing concentration of purified holo-SufR in binding buffer (25 ​mM Tris- HCl (pH 7.5), 1 ​mM DTT, 100 ​μg/mL BSA, 5 ​mM MgCl2, 5 ​mM CaCl2), and samples were incubated for 30 ​min at 4 ​°C inside glove box under anaerobic condition. Reactions were initiated by the addition of DNase I to a final concentration of 0.05 units and incubated for 2 ​min at room temperature. The reactions were terminated by the addition of 150 ​μL of TE (pH 7.5) followed by incubation at 75 ​°C for 15 ​min to deactivate DNase I enzyme. The sample was further subjected to vacuum evaporation and the pellet thus obtained was re-suspended in loading dye (80% (v/v) formamide, 0.1% (v/v) BPB, and 0.1% (v/v) xylene cyanol) and analyzed on a denaturing 15% polyacrylamide gel containing 7 ​M urea. The gel was dried, and the bands were visualized with a Typhoon FLA-9000 phosphor imager (GE Healthcare Life Sciences, USA). Maxam and Gilbert A ​+ ​G ladder was prepared as described previously [34].

2.20. Ethics

Animal experimentation: This study was carried out in strict accordance with the guidelines provided by the Committee for the Purpose of Control and Supervision on Experiments on Animals (CPCSEA), Government of India. The protocol of animal experiment was approved by the Institute's Animal Ethical Committee (IAEC), Indian Institute of Science (IISc), Bangalore, India (Approval number: CAF/Ethics/544/2017). All efforts were made to minimize the suffering.

2.21. Statistical analysis

All data were graphed and analyzed with Prism v8.0 (GraphPad) unless otherwise stated. Statistical analyses were performed using Student's t-test (two-tailed). Where comparison of multiple groups was made either one-way or two-way ANOVA with Bonferroni multiple comparisons was performed. Differences with a p value of <0.05 were considered significant. Statistical significance for RNA-seq was calculated using the QL F-test followed by Benjamini-hochberg method of multiple testing correction.

2.22. Miscellaneous

The molar extinction coefficient per [4Fe–4S]²⁺ cluster (ε413 = 16200 M^-1 cm^-1) was determined from in vitro reconstituted protein. Protein concentration of SufR throughout the study was calculated using A₂₈₀ reading. Anti-SufR polyclonal antibody was used to monitor the expression and purification of wt SufR and mutant proteins by Western blot. Samples (5 mg of protein per slot) were resolved by 12% reducing SDS-PAGE and transferred to nitrocellulose membrane. Blots were probed with primary and secondary (horseradish peroxidase-conjugated anti-rabbit IgG) antibodies and processed.

2.23. Data availability

The RNA-sequencing data presented in the manuscript are deposited into the NCBI Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc&equals;GSE154169 with the accession number GSE154169.

3. Results and discussion

3.1. Mtb SufR contains a 4Fe–4S cluster

To understand the role of Mtb SufR in sensing NO and regulating Fe–S cluster homeostasis, we first examined the biochemical and biophysical characteristics of the SufR Fe–S cluster. We anaerobically purified histidine-tagged SufR from E. coli cultured in growth conditions optimized for the maximum incorporation of Fe–S clusters in vivo [31]. As isolated, native SufR (holo-SufR) displayed a characteristic straw brown color, with an absorption maximum indicative of a 4Fe–4S cluster at 413 nm (molar absorption coefficient [ε] at 413 = 16200 M^-1 cm^-1) (Fig. 1A and Fig. S1 A-C) [32]. Treatment of holo-SufR with a one-electron donor, sodium dithionite (DTH), caused partial bleaching of brown color and loss of absorbance at 413 nm, consistent with the presence of a redox-responsive 4Fe–4S cluster [32] (Fig. 1A). Gel filtration performed under anaerobic conditions suggested a molecular mass of ∼ 54 kDa for native SufR, consistent with a dimer. Clusterless SufR (apo-SufR) also eluted as a dimer, indicating that the Fe–S cluster does not bridge SufR monomers (Fig. 1B). Moreover, the identity of cysteine residues that coordinates the 4Fe–4S cluster was experimentally validated. The alanine mutant of two putative cysteine ligands (C₁₇₅A_C₁₇₉A) was sufficient to disrupt the 4Fe–4S cluster as evident from the loss of absorbance at 413 nm (Fig. 1C).

Fig. 1.

Mtb SufR coordinates 4Fe–4S cluster that responds to NO and H₂O₂. (A) Anaerobically-purified SufR was subjected to UV–visible spectroscopy. A clear peak at 413 nm indicates presence of a 4Fe–4S cluster. The Fe–S cluster was reduced by addition of 5 mM DTH. (Inset) (i) unbound Ni²⁺-NTA, (ii) brown colored SufR bound to Ni²⁺-NTA beads, and (iii) anaerobically purified holo-SufR. (B) Superdex 200 HR gel filtration chromatography of SufR. (Inset: standard curve of the protein standards versus their elution parameter Ve/Vo where, Ve is the elution volume and Vo is void volume). (C) UV–visible spectroscopy of cysteine mutants of SufR (M1-C₁₇₅A, M2-C₁₇₅A_C₁₇₉A, and M3- C₁₇₅A_C₁₇₉A_C₁₉₂A). (D) A far-UV and (E) Near-UV CD-spectrum of holo- and apo-SufR. (F) UV–visible spectra were acquired before and after addition of proline NONOate (2.5 mM). A time dependent increase in absorbance at 350 nm indicates monomeric DNIC formation. (G) Holo-SufR exposed to various concentration of proline NONOate (SufR:NO). EPR data were acquired using a continuous-wave EPR spectrometer at a microwave frequency of 9.667 GHz and microwave power of 2 mW at liquid nitrogen temperature. The peak at 2.025 g is consistent with the formation of monomeric DNIC. (H) UV–visible spectra of holo- SufR exposed to either atmospheric O₂ for 2 min or (I) H₂O₂ (1 mM). (Inset) Rate of the 4Fe–4S cluster loss was determined by calculating the percent loss of absorbance at 413 nm upon exposure to air and H₂O₂. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

A recent study using aerobically purified SufR, demonstrated that SufR is a monomer and coordinates a 2Fe–2S cluster upon Fe–S cluster reconstitution in vitro [19]. In general, Fe–S clusters are inherently unstable under aerobic conditions [35] and stabilize only under anaerobic conditions [31]. Fe–S clusters can be assembled in vitro, but this process has poor yield [31]. Therefore, the technique of assembling the Fe–S cluster in vivo followed by anaerobic purification is more likely to provide native configuration of SufR Fe–S cluster. We clarified this by analyzing native SufR using circular dichroism (CD). The far-UV CD-spectrum showed two minima at 208 and 224 nm and indicated 70% α-helical content in the secondary structure (Fig. 1D). The near-UV CD-spectrum displayed two characteristic maxima at 330 and 420 nm and two minima near 350 nm and 550 nm, as described earlier for [4Fe–4S] cluster (Fig. 1E) [31,36] As expected, the near-UV CD-spectrum of apo-SufR did not show a [4Fe–4S] cluster (Fig. 1E). Lastly, the estimation of iron and sulfide ions in the native SufR revealed the association of 3.23–3.75 iron atoms per SufR monomer and similar amount of sulfide ions per SufR monomer (Fig. S1D). In sum, we demonstrate that Mtb SufR contains a 4Fe–4S cluster.

3.2. SufR responds to NO through its 4Fe–4S cluster

We exposed native SufR to the fast-releasing NO donor; proline NONOate (T_1/2–1.8 s, pH 7.4) and subjected it to UV–visible spectroscopy. Treatment with proline NONOate gradually reduced absorbance at 413 nm and formed a new chromophoric feature at ∼350 nm over time with a clear isosbestic point (Fig. 1F). These spectral features were consistent with a dinitrosyl-iron dithiol complex (DNIC), wherein the sulfide ligands of the 4Fe–4S clusters were displaced by NO to form [Fe-(NO)₂] [37]. The DNIC formation was confirmed by subjecting NO-treated SufR to continuous-wave electron paramagnetic resonance (cw-EPR). A strong EPR signal centered at g = 2.03 was observed in a dose-dependent manner, consistent with the formation of the monomeric DNIC complex on SufR (Fig. 1G) [32].

The suf operon showed induction under oxidizing stress (e.g., H₂O₂) [38]. Therefore, we determined whether the Fe–S cluster of SufR is responsive to atmospheric O₂ and H₂O₂. Exposure to atmospheric O₂ resulted in a slow decline in absorbance at 413 nm (Fig. 1H). A plot of ΔA_413nm against time revealed the loss of ∼75% of the 4Fe–4S cluster only at 20 h post-O₂ exposure. However, exposure of native SufR to 1 mM H₂O₂ resulted in the rapid loss of absorption at 413 nm (Fig. 1I). A plot of ΔA_413nm against time revealed the loss of ∼75% of the 4Fe–4S cluster in 20 min. Higher concentrations of H₂O₂ (10 mM and 100 mM) resulted in the loss of the cluster within 1–2 min (Figs. S2A–C). Taken together, these data demonstrate that the 4Fe–4S cluster of SufR is sensitive to NO and H₂O₂.

3.3. SufR binds upstream of suf operon promoter and represses expression

To investigate if SufR mediates Fe–S cluster biogenesis by regulating the suf operon expression, we first assessed the DNA-binding properties of SufR. We carried out electrophoretic mobility shift assays (EMSA) using a radioactively (³²P)-labeled 170 bp DNA fragment encompassing the promoter region of suf operon [19]. We used four different forms of SufR; holo-SufR, apo-SufR, NO-treated SufR, and H₂O₂-treated SufR. Holo-SufR bound to the suf promoter region, whereas apo-SufR, NO-treated SufR, and H₂O₂-treated SufR showed no DNA binding (Fig. 2A–C). DNA binding was outcompeted by 50-100-fold excess of unlabeled suf promoter DNA, but not by an unrelated promoter fragment (blaC) (Fig. 2D). These findings suggest that Mtb SufR is an NO- and H₂O₂-responsive, sequence-specific DNA-binding protein.

Fig. 2.

DNA binding activity of SufR. (A) Binding of SufR to ³²P-labeled suf promoter (0.5 nM). The concentrations of SufR were 0.125 μM, 0.25 μM, 0.5 μM, 1 μM, and 2 μM. Holo-SufR (0.5 μM) was pretreated with (B) proline NONOate (NO; 2.5-, 5-, and 10-fold molar excess) or (C) H₂O₂ (25, 50, 75, 100, 250, 500, 1000 and 5000 nM) and were incubated with ³²P-labeled suf promoter. (D) Holo-SufR (0.5 μM) was incubated with ³²P-labeled suf promoter and DNA binding was competed out using 12.5-, 25-, 50-, and 100-fold molar excess of unlabeled competitor promoter of suf operon and blaC. (E) Effect of SufR on transcription from suf promoter in vitro. Increasing concentrations of holo-SufR inhibit transcription. (F) Apo-SufR was completely ineffective in inhibiting transcription. (G) Holo-SufR (0.50 μM) pre-treated with proline NONOate (NO; 2.5- and 5-fold molar excess) stimulated transcription from suf promoter. The graph below indicates densitometry analysis of the suf transcript (n ​= ​2). One-way analysis of variance (ANOVA) with Bonferroni's post hoc was used to determine statistical significance. ‘ns’ non-significant ‘∗’ p<0.05 ‘∗∗’ p<0.01 ‘∗∗∗’ p<0.001.

We next performed in vitro transcription using Mtb RNA polymerase as described [39]. A single round of transcription of the suf promoter fragment (170 bp [-63 to ​+ ​107 bp]) produced a single transcript of 107 nucleotides, consistent with the leaderless transcription of the suf operon [40]. Addition of holo-SufR repressed transcription from the suf promoter in a dose-dependent manner, whereas apo-SufR did not (Fig. 2E–F). Importantly, the treatment of holo-SufR with proline NONOate (NO) reversed the repression of the suf promoter (Fig. 2G). These results indicate that holo-SufR binds to the suf promoter and represses expression, whereas NO-damaged or clusterless apo-SufR lacks DNA-binding and transcription repression. The data suggest that occupancy of Fe–S cluster on SufR serves as an indicator of Fe–S cluster-sufficient or -deficient conditions. During low Fe–S cluster demand, holo-SufR represses the Suf system to limit excessive assembly of Fe–S clusters. In contrast, loss or damage of the Fe–S cluster in SufR (apo-SufR) signals heightened demand for Fe–S clusters that results in the de-repression of the suf operon and mobilization of Fe–S cluster assembly. In support of this, we depleted an essential gene of the suf operon (Rv1466; sufT) involved in Fe–S cluster maturation [41] using CRISPR interference (CRISPRi-sufT). The depletion of SufT is expected to perturb Fe–S cluster biogenesis and increase the pool of apo-SufR, which could result in the de-repression of the suf operon in Mtb. Consistent with this, expression of the suf operon was induced in CRISPRi-sufT as compared to wild-type Mtb, indicating that Mtb de-represses the suf operon in response to abnormal Fe–S cluster biogenesis caused by SufT depletion (Fig. S3).

Lastly, we performed DNase I footprinting to identify the binding site of holo-SufR on the promoter region of the suf operon (Fig. 3A). A clear region of protection from DNase I digestion was evident with increasing molar ratios of SufR:DNA (Fig. 3A–B). The protected region contains a perfect inverted repeat (ACACT-N₅-TGTGA) separated by 5 bp (Fig. 3A–B). This inverted repeat forms a part of a larger inverted repeat (ATTTTGTCACACT-N₅-TGTGAAAAT). Consistent with the footprinting data, mutations in the inverted repeat completely abolished binding, thus confirming that the SufR binds to the palindrome ACACT-N₅-TGTGA (Fig. 3C–E). Altogether, using multiple techniques, we confirmed that SufR functions as a NO-sensitive DNA-binding transcription factor in Mtb.

Fig. 3.

DNase I footprint of holo-SufR bound to the suf promoter. Sequence upstream of the suf operon (KA10t/KA11t) were used for DNase I footprinting. Bold letter G represents transcription start site (TSS) for the suf operon. (A) ³²P-KA10t was annealed with cold KA11t. (B) ³²P-KA11t was annealed with cold KA10t to examine footprint in the reverse strand. 5 nM³²P-labeled dsDNA was incubated either in absence or presence of holo-SufR (125 ​nM, 250 ​nM, 500 ​nM, and 1 ​μM) and treated with 0.05 U of DNase I for 2 ​min. A ​+ ​G is Maxam and Gilbert sequencing ladder. DNase I protected/unprotected/hypersensitive nucleotides are marked with brown/black/blue and two inverted repeats are marked by arrows (blue and black). Mutations in the inverted repeat completely abolished binding of holo-SufR. (C–E) EMSA was performed with wild type (KA10t and KA11t) and oligonucleotides with mutations in the inverted repeats (KA10tm1, KA11tm1, KA10tm2 and KA11tm2) at increasing concentrations of NaCl at two different concentrations of holo-SufR (0.25 μM and 0.5 μM). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.4. NO regulates expression of the suf operon and Fe–S cluster pathways in Mtb

Having shown that SufR responds to NO via its 4Fe–4S cluster and directly regulates the transcription of suf operon promoter, we next asked if SufR coordinates Mtb's adaptation under NO stress. To examine this possibility, we first investigated the role of SufR in regulating transcriptome of Mtb in response to NO. All genes of the suf operon are essential except sufR [16]. Therefore, we utilized sufR-deficient (MtbΔsufR) strain of Mtb [23]. We generated the sufR-complemented strain (sufR-Comp) by integrating sufR (Rv1460) along with its native promoter (∼500 bp upstream of sufR) in the genome of MtbΔsufR. Using qRT-PCR, we confirmed the restoration of sufR expression to wild type (wt) Mtb levels in sufR-Comp (Fig. S4.). Since exposure to 0.5 mM of NO donor diethylenetriamine-nitric oxide (DETA-NO) arrested the growth of wt Mtb, MtbΔsufR, and sufR-Comp without affecting viability (Fig. S5), we performed RNA-sequencing (RNA-seq) on Mtb strains treated with 0.5 mM of DETA-NO for 4 h.

Our RNA-seq data recapitulated the previously published NO-responsive transcriptome of Mtb, with the induction of the DosR, IdeR, Zur, and CsoR regulons, and reduced expression of RNA and protein biosynthesis pathways (fold change≥2; FDR≤0.05; Fig. S6 and Table S1A) [13]. These pathways were similarly affected in MtbΔsufR and sufR-Comp in response to NO (Fig. S6 and Tables S1B–C). Since Fe–S cluster proteins are the most susceptible targets of NO [42], we found that the expression of ∼50% of genes encoding Fe–S cluster proteins was altered in wt Mtb, MtbΔsufR, and sufR-Comp under NO stress (Fig. 4 and Table S1D). The strong reactivity of NO towards heme iron of primary cytochrome oxidase is thought to arrest respiration [2]. However, we found that NO does not affect the expression of genes encoding primary terminal oxidase (cytochrome bc1-aa3 complex; qcrABC) in Mtb (Fig. 4). Nonetheless, downregulation of Fe–S cluster-containing NADH dehydrogenase complex I (nuo operon) and other genes encoding Fe–S cluster proteins involved in central metabolism (acn, prsA, and udgA) are consistent with the inhibition of primary respiration by NO. Moreover, the expression of genes associated with alternate form of respiration (e.g., NADH dehydrogenase type II [ndh], cytochrome BD oxidase [cydABCD], nitrate reductase [narH], sulfite reductase [sirA], hydrogenase [hycP]) [43], carbon catabolism (korAB, frdB, icl1, and pckA), and branched chain amino acid (BCAA) biosynthesis [44] were induced by NO. (Fig. 4).

Fig. 4.

SufR mainly regulates Suf system involved in Fe–S cluster biogenesis in response to NO. Total RNA from three biological replicates of untreated (UT) and DETA-NO treated (T) wt Mtb, MtbΔsufR, and sufR-Comp was isolated and subjected to RNA-seq. Heat maps indicate log₂ fold change of differentially expressed genes (DEGs) belonging to various functional categories (obtained from Mycobrowser, EPFL, Lausanne). Genes were considered differentially expressed on the basis of the false discovery rate (FDR) of ≤0.05 and absolute fold change of ≥1.5. BCCA: branched-chain amino acid.

The association between NO, respiration, and metabolism is further indicated by the upregulation of Fe–S cluster enzymes associated with the biosynthesis of respiratory cofactors such as isoprenoid (lytB2 and gcpE), molybdopterin (moaA2), thiamin (thiC), quinolinate (nadA-B), and sulfur metabolites (cysH) [[45], [46], [47], [48]] (Fig. 4; Table S1D). Also, NO induced the expression of transcription factors containing redox-sensitive Fe–S clusters (e.g., whiB1, whiB2, and whiB3) [32,49,50] (Fig. 4). We found that NO induced comparable expression changes in Mtb and MtbΔsufR (Fig. 4), indicating that SufR is not a global regulator of Mtb's transcriptome under NO stress.

Upregulation of the Fe–S cluster transcriptome indicates an increased demand for Fe–S cluster biogenesis. In agreement, expression of the suf operon (sufRBDCSUT; Rv1461-Rv1466) was stimulated 15- to 25-fold by NO in Mtb (Fig. 4). Also, NO induces the cysteine biosynthetic machinery that supplies sulfur for Fe–S cluster biogenesis (Fig. 4). However, genes coordinating biosynthesis of Fe-binding heme (hemZ) and encoding Fe-storage proteins (bacterioferretin; bfrA-B) were downregulated (Fig. 4). These observations suggest that Mtb prioritizes the assembly of Fe–S clusters under NO stress. In contrast to wt Mtb, the suf operon remains basally expressed in NO-treated MtbΔsufR (Fig. 4). A direct comparison of the expression data confirmed a uniformly reduced expression of the suf operon in MtbΔsufR as compared to wt Mtb under standard growing conditions and NO stress (Fig. 5A–D). Since SufR is a putative repressor of the suf operon [19], the operon's diminished expression in MtbΔsufR was unexpected. One possibility is that by replacing 345 bp fragment internal to the sufR (+308 to +653 bp) with the hygromycin resistance cassette (lox-hyg^rgfp-lox) [23], we might have interrupted the NO-inducibility of the downstream suf genes (sufBDCSUT) in MtbΔsufR (Fig. 5A). Consistent with this, the partial transcript of sufR originated from the undeleted region (+1 to +307 bp) that is present upstream to the deleted fragment (+308 to +653 bp) retained NO-inducibility in MtbΔsufR (Fig. 5C–D). The basal expression of the sufBDCSUT genes in MtbΔsufR is likely due to alternative transcription start site (TSS2) present upstream of sufB (Fig. 5A). As expected, sufR-Comp expressing a native copy of sufR restored the induction of sufR in response to NO, whereas rest of the operon remained basally expressed (Fig. 5B–D and Fig. S7). While the suf operon was not induced, the expression of pathways requiring Fe–S clusters remained induced in NO-treated MtbΔsufR and sufR-Comp. Thus, we anticipate that the heightened demand for Fe–S clusters under NO stress is unlikely to be satisfied in MtbΔsufR or in sufR-Comp. The restoration of NO-inducibility of sufR but not of the suf operon in sufR-Comp provides an opportunity to assess the contribution of SufR other than regulating Fe–S cluster homeostasis under NO stress. Altogether, the transcriptomic data indicate that SufR is mainly required to adjust the NO-responsive expression of the suf operon in Mtb.

Fig. 5.

Deregulation of the suf operon in MtbΔsufR and sufR-comp. (A) Genomic organization of the suf operon upon disruption of sufR. The internal fragment of sufR ORF from +308 bp to +653 bp was deleted (del) and replaced by hygromycin resistance cassette (loxP-hyg^r-gfp-loxP). The sufR sequence from +1 to +307 remained undeleted (UD) and resulted in the expression of an aberrant sufR transcript. The two transcription start sites TSS1 (coincides with the start codon ‘gtg’ of sufR) and TSS2 (upstream) of sufB were marked as per the published work [19]. (B) Exponentially grown cultures of wt Mtb, MtbΔsufR, and sufR-Comp were either left untreated (UT) or treated (T) with 500 μM DETA-NO for 4 h. Total RNA was isolated and subjected to RNA-seq and qRT-PCR analysis. Heat map depicting expression of the suf genes (log₂fold-change≥1; FDR≤0.05). Note that sufR expression detected in MtbΔsufR samples originated from the reads mapping onto the undeleted (UD) region of sufR. (C, D) qRT-PCR data showing the expression of the suf genes where MtbΔsufR and sufR-Comp were compared with untreated and DETA-NO-treated wt Mtb. Results are expressed as mean ± standard deviation (Mean ± SD).

3.5. NO irreversibly damages Fe–S clusters of aconitase in MtbΔsufR

Altered expression of the Fe–S pathways involved in metabolism and respiration by NO indicates that NO might modulate these processes in Mtb. To investigate this idea, we evaluated a 4Fe–4S-containing enzyme aconitase (Acn) activity, which functions as a critical gatekeeper of the TCA cycle, and shows sensitivity to NO due to a solvent-exposed Fe atom [51]. A gradual decrease in Acn activity over time was observed in Mtb exposed to 0.5 mM of DETA-NO, indicating Fe–S cluster damage. At 12 h post-exposure, DETA-NO triggered a 40% reduction in Acn activity without decreasing its abundance (Fig. 6A). Importantly, re-culturing DETA-NO-treated Mtb in a DETA-NO-free medium significantly restored Acn activity, indicating efficient mobilization of Fe–S cluster regeneration machinery (Fig. 6A).

Fig. 6.

SufR coordinates Fe–S homeostasis, respiratory reserves, and redox balance in response to NO. Aconitase (Acn) activity in response to 0.5 mM DETA-NO. (A)wt Mtb and (B)MtbΔsufR. At 12 h post-treatment, strains were cultured in a DETA-NO free 7H9 broth for 24 h and recovery (Rec) of Acn activity was quantified. Intracellular levels of Acn remained unaltered. A non-Fe-S cluster protein cystathionine-β-synthase (Cbs) of Mtb was used as the loading control. A change in oxygen consumption rate (OCR) of (C) wt Mtb and (D) MtbΔsufR was quantified upon addition of indicated concentrations of DETA-NO. The uncoupler CCCP was used to determine the spare respiratory capacity (SRC). The first and second vertical lines indicate point of addition of DETA-NO and CCCP, respectively. (E) SRC of wt Mtb and MtbΔsufR with increasing concentrations of DETA-NO. Percentage SRC was calculated by subtracting basal OCR (before adding DETA-NO) from CCCP-induced OCR considering basal OCR as 100%. Statistical significance for the OCR was obtained by comparing OCR upon treatment with different doses of DETA-NO with untreated (two-tailed, unpaired Student's t-test.). Comparisons whose P value is <0.05 were indicated with different symbols. Symbols: (): comparison to 0.25 mM; (): comparison to 0.5 mM; and (∗): comparison to 1.0 ​mM ∗//p<0.05; ∗∗//p<0.01; ∗∗∗//p<0.001. For SRC, one-way analysis of variance (ANOVA) with Bonferroni's post hoc was used to determine statistical significance. ‘ns’ non-significant ‘#’ p<0.05 ‘##’ p<0.01.

We next examined if defective induction of the suf operon impairs regeneration of NO-damaged Fe–S clusters in MtbΔsufR. The Acn activity in MtbΔsufR was similar to wt Mtb under aerobic growing conditions (Fig. 6B). Moreover, like wt Mtb, Acn activity decreased in MtbΔsufR under NO stress over time (Fig. 6B). However, in contrast to wt Mtb, reactivation of Acn upon removal of NO stress was absent in MtbΔsufR (Fig. 6B). Importantly, sufR-Comp that maintains sufR expression but lacks NO-inducibility of the suf operon also failed to reinstate Acn activity (Fig. S8A). Data suggest that NO-mediated induction of the suf operon rather than SufR alone is critical for the repair of NO-damaged Fe–S clusters in MtbΔsufR. A previous study indicated that Acn activity is dependent upon a stand-alone cysteine desulfurase (IscS) in Mtb under standard culture conditions [52]. These findings, along with our data, suggest that Mtb prefers IscS under aerobic conditions and Suf system under NO stress for biogenesis of Fe–S clusters. Similar roles were assigned for Isc and Suf systems in E. coli [53].

3.6. NO depletes spare respiratory capacity and perturbs redox homeostasis of MtbΔsufR

Fe–S cluster-containing enzymes are crucial for maintaining carbon catabolism, oxidative phosphorylation (OXPHOS), and redox balance [45,54]. Therefore, we exploited Seahorse XF Flux technology to analyze the influence of NO on oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), which are measurable readouts of OXPHOS and glycolysis, respectively [55]. To quantify the basal and maximum rates of OCR and ECAR, we cultured Mtb and MtbΔsufR in 7H9-glucose in an XF microchamber, then exposed it to DETA-NO, and finally to the uncoupler carbonyl cyanide m-chlorophenyl hydrazine (CCCP). Addition of CCCP stimulates respiration to the maximal capacity manageable by Mtb. The difference between basal and CCCP-induced OCR provides an estimate of the spare respiratory capacity (SRC) available for sustaining stress-mediated bioenergetics (e.g., nitrosative and oxidative conditions) [56]. Under normal growing conditions, Mtb displayed a basal OCR of 20 ± 0.14 pmol/min, which increased to 85 ± 11.7 pmol/min in response to uncoupling stress by CCCP (Fig. 6C). This indicates that Mtb normally functions at a submaximal OXPHOS (∼25%) capacity. Under similar conditions, MtbΔsufR utilizes ∼ 35% of its maximal respiratory capacity, which is more than wt Mtb (Fig. 6D). Similar to the uncoupler CCCP, NO also depolarizes the cytoplasmic membrane to arrest respiration and growth [57,58]. Consistent with this, and as seen with CCCP, pretreatment with DETA-NO also increased basal OCR of wt Mtb and MtbΔsufR (Fig. 6C, D). However, while 1 mM of DETA-NO was required to increase basal OCR of wt Mtb significantly, 0.25 mM was sufficient for MtbΔsufR (Fig. 6C, D). Data suggest that NO stimulated basal OCR, possibly by collapsing proton motive force (PMF), and that MtbΔsufR is more sensitive to membrane depolarization by NO.

We also found that DETA-NO pretreatment progressively reduced the ability of bacteria to increase OCR in response to CCCP (Fig. 6C, D). As a result, DETA-NO significantly decreased SRC in a dose-dependent manner in both Mtb and MtbΔsufR (Fig. 6E). However, SRC of MtbΔsufR was significantly lower than wt Mtb under normal culture conditions and upon exposure to 0.25 mM DETA-NO (Fig. 6E). These results indicate that Mtb mobilizes its reserved respiratory capacity to sustain bioenergetics in response to NO. Data also suggest that the inherently reduced SRC of MtbΔsufR due to diminished Fe–S cluster biogenesis increased its vulnerability towards bioenergetic exhaustion by NO.

Measurement of basal ECAR with and without CCCP treatment indicated that wt Mtb and MtbΔsufR operate at a suboptimal glycolytic capacity of 20% and 15%, respectively (Figs. S8B–C). Similar to OCR, DETA-NO pretreatment progressively reduced the ability of Mtb to increase ECAR in response to CCCP (Fig. S8B). However, unlike wt Mtb, MtbΔsufR significantly increased basal ECAR and retains CCCP-induced ECAR in response to 0.25 mM and 0.5 mM of DETA-NO (Fig. S8C). This suggests an increased reliance of MtbΔsufR on glycolysis to handle the bioenergetic needs under NO stress. Similar to MtbΔsufR, the NO-induced changes in OCR, SRC, and ECAR were recapitulated in sufR-Comp (Figs. S8D–F), indicating that the increased expression of the suf operon rather than sufR alone is critical for Mtb's response to NO.

Lastly, we asked whether NO perturbed redox homeostasis in Mtb. We used a genetic biosensor (Mrx1-roGFP2) to measure the redox potential of a physiologically relevant antioxidant, mycothiol (MSH), as a proxy for the cytoplasmic redox potential (E_MSH) of Mtb [59]. Ratiometric measurements of emission at 510 nm after excitation at 405 and 488 nm can easily quantify any changes in redox physiology [59]. In response to an oxidant or a reductant, the biosensor ratio showed a rapid increase or decrease, respectively [59]. Mtb and MtbΔsufR expressing Mrx1-roGFP2 were treated with 0.25 mM, 0.5 mM, 1 mM DETA-NO, and the biosensor ratio was measured. Exposure of Mtb to NO did not increase the Mrx1-roGFP2 ratio, indicating that wt Mtb robustly maintains cytoplasmic E_MSH in response to NO (Fig. 7A). In contrast, NO induces a slightly higher oxidative shift in E_MSH of MtbΔsufR than wt Mtb in a dose-dependent manner (Fig. 7A). Using a ROS-sensitive fluorescent dye Cell ROX, we confirmed that NO induces oxidative stress in MtbΔsufR but not in wt Mtb (Fig. 7B). Altogether, our data indicate that the suf operon's NO-mediated induction is required to regenerate Fe–S clusters, maintain respiratory reserves, and buffer redox imbalance.

Fig. 7.

DETA NO induces oxidative stress in MtbΔsufR. (A) wt Mtb and MtbΔsufR expressing Mrx1-roGFP2 were exposed to the indicated concentrations of DETA-NO and the biosensor ratio (405/488 nm) was measured using flow cytometry. Data shown are the result of three independent experiments performed in triplicate (mean ± SD). One-way analysis of variance (ANOVA) with Bonferroni's post hoc test was employed to determine statistical significance. (B) wt Mtb and MtbΔsufR treated with the indicated concentrations of DETA-NO for 2 ​h and stained with CellRox Deep Red reagent to measure endogenous ROS. Data shown are the result of three independent experiments performed in triplicate (mean ​± ​SD). Two-way analysis of variance (ANOVA) with Bonferroni's post hoc test was employed to determine statistical significance between different doses of DETA-NO. Symbols: (∗): comparison to Mtb, (#): comparison to untreated MtbΔsufR. ‘∗’ p<0.01 ‘∗∗’ p<0.01‘∗∗∗’ p<0.001.

3.7. SufR is required to recover from NO-induced growth arrest and persistence in vivo

Next, we investigated the biological consequence of compromised Fe–S homeostasis and bioenergetics by assessing the phenotype of MtbΔsufR under NO stress. First, we investigated the survival phenotype of MtbΔsufR under NO stress in vitro. A single dose of various concentrations of DETA-NO (0.5 mM, 1.25 mM, and 2.5 mM) did not influence the survival of MtbΔsufR (Fig. 8A and Fig. S9). Repeated exposure to low doses of NO is known to arrest Mtb's growth for an extended duration [60]. Administration of 0.1 mM of DETA-NO every 6 h for 36 h induces an extended period of growth arrest followed by recovery of wt Mtb at day 7 post-treatment (Fig. 8B). In contrast, MtbΔsufR resumed growth only at day 16 post-treatment with NO (Fig. 8B). The sufR-Comp strain showed a recovery defect largely similar to MtbΔsufR, reinforcing that the NO inducibility of the entire suf operon rather than only sufR is necessary for the timely resumption of growth. We confirmed this using CRISPRi-sufT strain, which expresses reduced levels of another Fe–S cluster assembly factor- SufT. Similar to MtbΔsufR, diminished levels of sufT delayed recovery of Mtb from NO-induced growth inhibition (Fig. 8C).

Fig. 8.

MtbΔsufR displays defect in recovering from NO-induced growth arrest in vitro and NO-dependent persistence in mice. (A)Mtb strains were exposed to DETA-NO for 24 h and survival was monitored by enumerating CFUs. (B)Mtb strains were either left untreated (UT; empty symbols) or treated (T; filled symbols) with repeated doses of 0.1 mM of DETA-NO for 36 h at an interval of 6 h and growth was monitored over time. (C)Mtb strains expressing CRISPRi vector without (Control) or with sufT-specific guide RNA (sgRNA, sufT KD) were left untreated (UT) or treated (T) with 0.1 mM of DETA-NO at an interval of 6 h for 36 h and growth was monitored over time. To deplete SufT, 200 ng/mL of anhydrotetracycline (ATc) was added for the induction of sufT specific–guide RNA and dCas9 every 48 h. (D) RAW264.7 macrophages activated with IFNγ (100U/mL.) and LPS (100 ng/mL) were infected with Mtb strains at a multiplicity of infection (moi) of 10 and survival was monitored by enumerating CFUs. Data are the results of two independent experiments performed in triplicates. Inbred (E) BALB/c mice (n = 6), (F) C57BL/6 (n = 5), and iNOS knockout (iNOS^-/-, n = 5) were given an aerosol challenge with Mtb strains and assessed for survival in lungs at indicated time points. Results are expressed as Mean ​± ​SD. Two-way analysis of variance (ANOVA) with Bonferroni's post hoc test was employed to determine statistical significance for the intramacrophage and pulmonic load between different strains. p value is <0.05 were indicated with symbols. Symbols: (∗): comparison to wt Mtb; (#): comparison to sufR-Comp. ∗/#p<0.05; ∗∗/##p<0.01; ∗∗∗p<0.001. In fig B and C open and close symbol are untreated and DETA-NO treated respectively.

Immunologically activated macrophages are known to induce nitrosative stress in Mtb [61]. In line with this, MtbΔsufR showed 10- and 15-fold reduced survival in immune-activated RAW 264.7 macrophages as compared to wt Mtb at day 2 and 4 post-infection, respectively (Fig. 8D). These observations prompted us to investigate the NO-dependent phenotype of MtbΔsufR in vivo. Previous studies have reported the requirement of SufR for persistence of Mtb in mice [22,62]. However, it remains to be addressed if SufR coordinates pathogen's persistence in response to NO. wtMtb and MtbΔsufR showed comparable growth in the lungs of BALB/c mice during the acute phase of infection (0–2 weeks) (Fig. 8E). However, MtbΔsufR was cleared progressively from the lungs, with a more than 1.5-log decline in bacterial burden by 8 weeks (Fig. 8E). The histopathological changes observed in animals' lungs at 8 weeks post-infection were proportionate to the bacterial burden (Fig. S10). The magnitude of pulmonary damage was highest in case of wt Mtb (2.75 ± 0.5), intermediate in sufR-Comp (2.0 ± 0.0), and lowest in MtbΔsufR (0.25 ± 0.50) (Fig. S10B). Lastly, to clarify NO's role in the persistence defect of MtbΔsufR, we infected a highly susceptible mouse strain lacking inducible nitric oxide (iNOS^-/-) [5]. We found that the persistence defect of MtbΔsufR was abolished in iNOS^-/- mouse, indicating the requirement of SufR for the persistence of Mtb in response to NO (Fig. 8F–G).

Surprisingly, the persistence defect of MtbΔsufR was somewhat rescued in animals and macrophages infected with sufR-Comp (Fig. 8D–E). Since suf operon's expression is also responsive to Fe-limitation encountered in vivo (41), it is possible that SufR induces the expression of suf operon under Fe-limitation from a promoter that is distinct from NO-responsive promoter in sufR-Comp. Consistent with this, overexpression of SufR led to its binding inside the ORF of Rv1461 (sufB) in Mtb [63]. Moreover, several other transcription factors (e.g., Rv0081, Rv0023, Rv1189, Rv3765c, Rv3849, Rv0260c, and glnR) bind and alter the expression of suf operon [63,64]. Some of these transcription factors, along with SufR, could regulate suf operon's expression to restore persistence of sufR-Comp in animals and macrophages. Alternatively, other enzymes involved in Fe–S cluster coordination in Mtb (e.g., IscS) (42) partially counterbalance the repressed Suf system's effect in sufR-Comp in vivo. Consistent with this, our unpublished data suggest that IscS and Suf systems compensate for the loss of each other in mediating Mtb's survival inside macrophages. Altogether, data show that SufR enables NO-dependent persistence of Mtb during infection.

4. Conclusions

Fe–S cluster production is tightly regulated to promote Fe–S formation when the necessity for the clusters is heightened (e.g., ROI/RNI/iron-limitation) and to limit unnecessary production when the demand is low (e.g., hypoxia) (43). Deregulation of Fe–S cluster biogenesis can lead to toxic accumulation of iron and polysulfides inside cells (43). Therefore, the calibrated expression of Fe–S cluster biogenesis is important. Here we show that Mtb SufR is required to regulate Fe–S cluster biogenesis in Mtb under NO stress. Our findings provide mechanistic insights into how Mtb exploits Fe–S cluster regulation and biogenesis under NO stress to favor the pathogen's persistence.

We found that NO more severely inhibits spare respiratory capacity of MtbΔsufR as compared to wt Mtb. Since spare respiratory capacity depends mainly on the recruitment of previously inactive respiratory complexes (55), the SUF system's sustained activation can provide a reserve of Fe–S cluster-containing respiratory complexes to maintain electron transfer in response to NO. Induction of the suf operon in response to conditions that damage Fe–S clusters (e.g., H₂O₂, NO, iron-starvation, antibiotics, phagosomal pH, and sputum) [22,23,38,39,[65], [66], [67], [68]] indicates that Mtb relies on Suf-dependent Fe–S cluster coordination to maintain persistence. The capacity of SufR to sense and respond to a range of cues, such as NO, H₂O₂, and iron limitation, possibly empowers Mtb to transduce different redox signals into transcriptional responses crucial for persistence in vivo.

One limitation of our study is the lack of complementation in various in vitro assays. Since the polar effects of sufR disruption interfered with NO-inducibility of the downstream suf genes, the phenotypic changes exhibited by MtbΔsufR were mainly due to basal expression of the suf operon. Consistent with this, the restoration of expression of sufR alone did not rescue the phenotype of MtbΔsufR in vitro. Surprisingly, while both MtbΔsufR and sufR-Comp showed defective recovery from NO-mediated growth arrest, the mutant strain recovered earlier than the complemented strain. In this context, we noticed that NO exposure upregulated the DOS dormancy regulon more in MtbΔsufR as compared to wt Mtb but below 2-fold (FDR< 0.05) cutoff (Fig. S11). Notably, the expression of the DOS regulon was restored to wt Mtb levels in sufR-Comp under NO stress (Fig. S11). Therefore, marginally better recovery of MtbΔsufR than sufR-comp could be a consequence of elevated DOS regulon in the mutant. Agreeing to this, an Mtb strain lacking DOS dormancy regulator (MtbΔdosR) showed defective recovery from NO-mediated growth cessation [60]. Interestingly, a previous study reported overexpression of the suf operon in MtbΔdosR under hypoxia [15], signifying a regulatory loop between SufR and DosR in Mtb. In addition to SufR and DosR/S/T system, the Fe–S cluster containing regulators such as WhiB3 and WhiB1 also respond to NO [32,50,69]. Further, using bacterial-one-hybrid system, another study reported binding of WhiB3 to the promoter region of sufR [64], suggesting that further experiments are needed to fully understand the mechanism underlying the regulation of the suf operon in Mtb.

Previous studies on the sufR mutant did not clarify the polar effects on the downstream suf genes. Pandey et al., reported a marginal induction of sufD, sufC, and sufT and a basal expression of sufB, sufS, and sufU in the sufR-deleted strain (ΔSufR_TB) [22]. ΔSufR_TB grew similar to wt Mtb under standard growth conditions but showed survival defect under redox stress and inside macrophages [22]. The mutant also displayed persistence defect in mice [22]. In contrast to our findings, Pandey et al., reported a significantly better survival of the sufR-complemented strain (ΔSufR_TB::pJEBsufRTB) than wt Mtb under diverse in vitro stress conditions and macrophages, and full rescue of the persistence defect in mice [22]. Another study generated three identical truncated mutants of sufR (ΔRv1460stop_1.19, ΔRv1460stop_5.19, and ΔRv1460stop_5.20) by introducing a premature stop codon at position 122 [19]. Surprisingly, ΔRv1460stop_1.19 and ΔRv1460stop_5.19 grew slowly than wt Mtb under standard growing conditions, whereas growth of ΔRv1460stop_5.20 was comparable to wt Mtb. Intriguingly, the activity of Fe–S cluster enzymes was not affected in ΔRv1460stop_5.20 but diminished in the reported sufR-complemented strain [19]. None of these studies examined the expression of full suf operon both in the sufR mutant and the complemented strain under normal and/or NO stress conditions to rule out polar effects. We believe that the reported discrepancies in the sufR complementation could be due to the use of non-native promoters (e.g., mycobacterium optimum promoter [MOP] [22] and hsp60 [19]) to restore SufR expression in previous studies rather than the NO-responsive native sufR promoter used in this study. Altogether, future work is required to explore the breadth of SufR-mediated gene regulation and the role of additional regulators in coordinating the expression of the suf operon.

Lastly, the suf operon was uniformly induced in clinical isolates of Mtb belonging to five globally circulating lineages during survival inside macrophages [70]. These results indicate that regulation of Fe–S cluster biogenesis is a part of core processes that remain conserved in diverse Mtb lineages evolved under selection pressure inside the human host. Altogether, we propose a new model of mycobacterial persistence in which SufR senses NO through its Fe–S cluster to coordinate Fe–S cluster biogenesis and regulate metabolism, respiration, and redox balance (Fig. 9).

Fig. 9.

A model for SufR-mediated regulation of Fe–S cluster biogenesis in Mtb. SufR appears to coordinate the expression of the suf operon in a negative feedback loop. Under Fe-sufficient conditions, the Fe–S cluster bound form of SufR (holo-SufR) binds to the promoter of the suf operon and represses Fe–S cluster biogenesis. Exposure to NO or H₂O₂ damages Fe–S clusters to generate DNIC or apo-forms of SufR, respectively, which abrogates SufR DNA-binding resulting in de-repression of the suf operon. The elevated expression of the SUF Fe–S assembly system regenerates Fe–S clusters of metabolic enzymes, respiratory complexes, and redox sensors to maintain Fe–S cluster homeostasis, spare respiratory capacity, and redox balance. The NO-sensing properties of SufR via its 4Fe–4S cluster and persistence defect exhibited by MtbΔsufR in mice in an iNOS-dependent manner suggest that SufR integrates host environmental signals to Fe–S cluster homeostasis in Mtb.

Author contributions

KA, AT, and AS participated in the design of the study. KA, AT, KS, NM, AJ, RKJ, RSR, and SNC, carried out the experiments. AR, VN, BG, GN, and ASN contributed to reagents and analyzed the data. KA and AS conceived the study, supervised the project, analyzed the data and drafted the manuscript. All authors read and approved the final manuscript.

Funding sources

The Mtb work was supported by the following Wellcome Trust/DBT India Alliance Grants, IA/S/16/2/502700 (AS), IA/E/16/1/503017 (KA), and in part by DBT grants BT/PR13522/COE/34/27/2015, BT/PR29098/Med/29/1324/2018, and BT/HRD/NBA/39/07/2018–19 (A.S.), DBT-IISc Partnership Program grant 22-0905-0006-05-987 436, and the Infosys Foundation. AS and KA are senior- and early-career fellows of Wellcome Trust/DBT India Alliance.

Notes

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We confirm that no competing financial interests exist.

Abbreviations

Mtb

Mycobacterium tuberculosis

NO

nitric oxide

iNOS

inducible nitric oxide synthase

DNIC

dinitrosyl-iron dithiol complex

DTH

sodium dithionite

CD

circular dichroism

EMSA

electrophoretic mobility shift assay

DETA-NO

diethylenetriamine-nitric oxide

OCR

Oxygen Consumption Rate

SRC

spare respiratory capacity

CCCP

carbonyl cyanide m-chlorophenyl hydrazine

ECAR

extracellular acidification rate

Appendix A. Supplementary data

The following are the Supplementary data to this article: