Institutional Review Board approval

This work complied with all relevant ethical regulations and we obtained informed consent from all donors. The study protocols were approved by Maternidade Climério de Oliveira Ethics Committee, Federal University of Bahia (protocol number 037/2011, Ethics Committee approval number 034/11). All animal studies were conducted in Assessment and Accreditation of Laboratory Animal Care accredited Biosafety Level 2 and 3 facilities at the NIAID/NIH using a protocol (LPD-99E) approved by the NIAID Animal Care and Use Committee.

Human study design

BACH1 expression was tested in 30 healthy control individuals who had TB excluded through clinical and radiological investigation and who were IGRA-negative tested using the QuantiFERON Gold In Tube assay (3rd generation) (Qiagen), and in 30 individuals with culture-confirmed pulmonary TB AFB screening in sputum smears (by microscopy). Supporting sputum cultures (Lowenstein–Jensen solid cultures) were performed in all patients (Supplementary Table 1). Individuals with no symptoms and normal chest X-ray but with positive IGRA tests were clustered in the TBI group. Venous blood was collected in sodium heparin tubes for isolation of PBMC. Cells were cryopreserved at the biorepository of the Laboratory of Inflammation and Biomarkers, Fundação Oswaldo Cruz, Salvador, Brazil. For the immunological assays performed, selected samples from adult (>18-yr-old) HIV-negative individuals with confirmed PTB or controls were matched by age and sex (±5 yr). Sample size was determined on the basis of calculations of study power of 80% (alpha error, 5%) to detect differences in BACH1 expression >2% (arbitrary cut-off) between TB and healthy controls. Cryopreserved PBMCs were thawed and monocytes were column purified using CD14 beads. RNA extraction was performed using the Qiagen Easy RNA extraction kit. BACH1 mRNA levels were assessed by real-time PCR and gene fold-increase relative to β-actin (ACTB) was calculated.

BACH1 primers:

1.

Forward sequence: CACAATTCTTCCATAGACCCTC

2.

Reverse sequence: TCTGCCACTTCTCGCTCC

ACTB primers:

1.

Forward sequence: CACCATTGGCAATGAGCGGTTC

2.

Reverse sequence: AGGTCTTTGCGGATGTCCACGT

In a separate study, we analysed blood RNA-sequencing data from several timepoints (before diagnosis for progressors) from the Adolescent Cohort Study (ACS, GSE79362), a longitudinal cohort profiling individuals who progressed to active TB (n = 37 progressors) or from 106 controls with positive IGRA and/or TST test, who remained TB-free34 (Supplementary Table 2). Normalized data from GEO was applied to the multicohort analysis framework to calculate effect sizes reflecting BACH1 expression differences across clinical groups (controls and TB progressors). Hedge’s g was used to measure the effect size.

Animal experiments

Thy1.1 C57BL/6J, B6.SJL (CD45.1/1) and B6.SJL/C57BL/6 (CD45.1/2) mice (9–12-week-old, male) were obtained through a NIAID supply contract with Taconic Farms (Germantown, New York). Thy1.1 C57BL/6J mice were used as WT C57BL/6J controls. Bach1−/− mice (C57BL/6J background) were generously provided by Dr Kazuhiko Igarashi (Tohoku University Graduate School of Medicine). B6.Sst1S mice were donated by Dr Igor Kramnik (Boston University, USA) and crossed with Bach1−/− mice. Animal genotyping was performed by Transnetyx. Animals were housed under specific pathogen-free (SPF) conditions with ad libitum access to food and water, 20–26 °C, 30–70% humidity and a 12 h/12 h light/dark cycle.

Lung samples from Mtb-infected rhesus macaques (ID#: DG3X, DG5F, DGKA, DG9R, DGOI, DF4H, DGRI, DGNK, DG3P and DG4N) initially described in ref. 35 were used. Formalin-fixed paraffin-embedded tissue sections from Mtb-infected animals were stained for BACH1.

Generation of competitive mixed BM chimaeric mice preparation

B6.SJL/C57BL/6 (CD45.1/2) mice were lethally irradiated (900 rad) and reconstituted with a total of 2 × 106 donor BM cells from B6.SJL CD45.1/1 and C57BL/6 CD45.2/2 mice mixed in equal parts. Trimethoprim-sulfamethoxazole water was given to irradiated mice for 4 weeks. Mice were maintained on regular water for 8 additional weeks to ensure complete immune reconstitution before infection.

Bacterial strains and culture conditions

Mtb H37Rv strain was grown in 7H9 broth (Sigma-Aldrich) supplemented with 0.05% Tween 80 (Thermo Fisher) and 10% oleic acid-albumin-dextrose-catalase (OADC; BD Biosciences) at 37 °C. H37Rv expressing the red fluorescent protein (H37Rv-RFP) was a kind gift from Dr Joel Ernst (University of California San Francisco, USA) and grown in 7H9 broth (BD Biosciences) supplemented with 0.05% Tween 80 (Thermo Fisher), 10% OADC and 30 μg ml−1 kanamycin (Sigma-Aldrich) at 37 °C. Bacteria in mid-log phase (optical density (OD) of 0.6–1.0) were centrifuged at 5,000 r.p.m. for 10 min, resuspended in fresh 7H9 media and frozen at −80 °C in aliquots of ~108 bacilli per ml.

Primary cell cultures

Murine BMDMs were generated by dispersing cells (3–5 × 106 cells) from both femurs and tibiae and seeding them in Petri dishes (100 × 15 mm size) containing 10 ml of BMDM-differentiation media (DMEM/F12 containing 2 mM l-glutamine (Gibco), 10% fetal bovine serum (FBS), 2% HEPES (Life Technologies), 1 mM sodium pyruvate (Gibco), 25 μg ml−1 gentamicin (Gibco) and 20% L929-conditioned media) at 37 °C with 5% CO2. After 4 d of incubation, 10 ml BMDM-differentiation media without gentamicin was added. On day 6, BMDMs were detached by washing cells with cold 1× PBS.

In vivo Mtb infection

Animals were infected with H37Rv strain by aerosol at low-dose infection (~100–250 c.f.u. per mouse) or by intrapharyngeal (IPH) inoculation at high dose (~1,000–2,000 c.f.u. per mouse). For the IPH infection, mice were anaesthetized with isoflurane, and in a vertical body position their tongues pulled aside for pharynx exposure. Mtb inoculum (30 μl) was administered intrapharyngeally into the airway. Infection dose was confirmed by plating lung homogenates on Middlebrook 7H11 agar plates supplemented with 0.5% (v/v) glycerol and 10% (v/v) OADC-enrichment media. High-dose Mtb infection was used as a model of Mtb-induced pulmonary necrosis in immunocompetent C57BL/6 mice11,13.

Preparation of single-cell suspension from lungs

Lung lobes isolated from mice were washed with sterile 1× PBS, dissected into small pieces, digested in RPMI medium containing Liberase TL (0.33 mg ml−1; Sigma-Aldrich) and DNase I (0.1 mg ml−1; Sigma-Aldrich) at 37 °C for 45 min under agitation (200 r.p.m.) and added with FBS to block enzymatic digestion. Lung tissue was dispersed by passage through a 70-μm-pore-size cell strainer. Red blood cells were lysed with ACK buffer (Gibco) at room temperature for 3 min. Lung cells were washed with 1× PBS supplemented with 10% FBS, centrifuged at 1,500 r.p.m. for 5 min and the cell pellet resuspended in RPMI medium supplemented with 10% FBS. Cell numbers were counted using ViaStain acridine orange propidium iodide staining on a Cellometer Auto 2000 cell counter (Nexcelom).

Flow cytometry

Intravenous injection of 2 μg anti-CD45 (clone 30-F11; Invitrogen) 3 min before euthanasia was performed to distinguish cells within the lung vasculature and parenchyma13. Cocktails of conjugated or unconjugated antibodies diluted in 1× PBS containing 2% FBS and 10% Brilliant Stain buffer (BD Biosciences) were added to isolated cells and incubated for 30 min at 4 °C. Antibodies used were directed against CD11b (1:300, clone M1/70), CD11c (1:200, clone HL3), Ly6G (1:100, clone 1A8), CD24 (1:200, clone M1/69), CD19 (1:100, clone 1D3), B220 (1:100, clone RA3-6B2), CD4 (1:100, clone GK1.5 or RM4-5), CD45 (1:100, clone 30-F11), CD45.2 (1:100, clone 104), Siglec-F (1:200, clone E50-2440), TCR-β chain (1:100, clone H57-597) and TCR-γδ (1:100, clone GL3), all purchased from BD Biosciences; F4/80 (1:50, clone BM8) was purchased from Thermo Fisher; CD8-α (1:100, clone 53-6.7), CD11c (1:200, clone N418), CD44 (1:300, clone IM7), CD64 (1:100, clone X54-5/7.1), CD69 (1:100, clone H1.2F3), CD88 (1:100, clone 20/70), IA/IE (1:300, MHCII, clone M5/114), NK1.1 (1:100, clone PK136), CD45.1 (1:100, clone A20) and Ly6C (1:200, clone HK1.4) were purchased from BioLegend; monoclonal rabbit unconjugated Gpx4 (1:100, clone EPNCIR144) was purchased from Abcam. Unconjugated monoclonal rabbit antibody was detected with donkey F(ab’)2 anti-rabbit IgG H&L pre-adsorbed (1:700, Abcam) and rabbit IgG monoclonal antibody (1:100, Abcam) was used as primary isotype control. Ultraviolet Fixable Live/Dead cell stain dye was purchased from Molecular Probes (Invitrogen) and the staining was performed following manufacturer instructions. Samples were acquired on a FACSymphony A5 SORP flow cytometer (BD Biosciences) or an LSR II Fortessa flow cytometer and results were analysed using FlowJo v.10 software (Three Star).

Determination of glutathione, total antioxidant status and lipid peroxidation levels in lungs

Lungs were homogenized in 1× PBS and centrifuged at 13,000 g at 4 °C for 10 min to remove tissue matrix and cell debris. Supernatants were sterilized by 0.22-μm filtration and stored at −80 °C. Glutathione levels, TAS and lipid peroxidation levels were measured using the Glutathione assay kit, the antioxidant assay kit and the TBARS assay kit (all from Cayman), respectively, following manufacturer instructions. Glutathione, TAS and lipid peroxidation levels were normalized on the basis of total protein content determined using Pierce protein assay (Thermo Fisher) according to manufacturer instructions.

Multiplex cytokine arrays

Levels of cytokines and chemokines in lung tissue homogenates were assessed using a MILLIPLEX MAP mouse cytokine/chemokine magnetic bead panel kit (Millipore Sigma) according to manufacturer instructions and measured using a MAGPIX instrument (R&D Systems). Total protein concentration was measured using Pierce Protein assay (Thermo Fisher) and values were used to normalize cytokine/chemokine levels on the basis of total protein content.

Histopathology, immunohistochemistry and necrotic tissue detection

For histological analysis, lungs were fixed with 10% formaldehyde, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E) or Ziehl-Neelsen (ZN). Samples were examined under light microscopy and images scanned using an Aperio VERSA scanner (Leica Microsystems).

Fixed tissue was paraffin embedded and 10-μm-thick sections were prepared. Slides were deparaffinized and treated with AR6 buffer (Akoya Biosciences) for 20 min at 100 °C. Slides were next placed in AR9 buffer (Akoya Biosciences) at 100 °C and allowed to cool to room temperature for ~45–50 min. Samples were permeabilized using 0.2% TritonX 100 (Millipore Sigma) for 10 min and blocked using fetal calf serum (FCS) and/or isotype-matched non-specific immunoglobulin. Primary antibody against BACH1 (Proteintech) was employed at a dilution of 1:250. Tissues were washed, stained with ImmPRESS HRP horse anti-rabbit staining kit (Vector Laboratories) and counterstained with haematoxylin. Slides were mounted and imaged using an Aperio VERSA scanner (Leica Microsystems). For human lung tissue, epitope retrieval was performed using citrate buffer (pH 6.0) in bath water at 98 °C for 45 min. Following endogenous peroxidase blocking (peroxidase blocking solution, Dako), tissue sections were incubated with BACH1 primary antibody (1:500; Proteintech) for 18 h at 4 °C. After washing with 1× PBS, all sections were then incubated with Advance HRP Link buffer for 20 min and then subjected to an additional round of PBS washing followed by incubation with Advance HRP enzyme (Dako) for 20 min. Chemical reactions were developed with 3,3-diaminobenzidine (Dako) and sections were counterstained with Harris hematoxylin. One representative field per sample (×200) was selected by a single experienced pathologist. All immunomarkers were analysed within these same selected fields.

Necrotic tissue detection in the lungs was assessed by inoculating mice with a solution of SytoxGreen dye (Thermo Fisher) at 50 μM intravenously 10 min before mouse euthanasia. Lungs were then collected, washed with 1× PBS and immediately fixed with 10% formaldehyde at 4 °C for 48 h. SytoxGreen fluorescence in whole lung tissue was examined in a motorized stereo microscope (Leica M205 FA) and images captured with a CFC345 cooled monochrome camera (Leica) using LAS X (Leica Microsystems) software. Images were processed using Imaris 8.4.1 (Bitplane) software, and QuPath v.0.4.0 and Image J v.1.53t for visualization and quantification.

Bacterial load determination

Mtb burden in the lung and spleen homogenates was assessed by serial dilution and plating onto 7H11 agar Petri dishes supplemented with 0.5% (v/v) glycerol and 10% (v/v) OADC-enrichment media. Bacterial colonies were enumerated after 21 d incubation at 37 °C.

Lipid peroxidation staining

Lipid peroxidation was measured in lung single-cell suspension as well as in BMDM cultures by using click-iT lipid peroxidation imaging kit (Life Technologies) according to manufacturer instructions. Briefly, cells were incubated with the linoleamide alkyne (LAA) reagent (alkyne-modified linoleic acid) at 37 °C for 1 h and then washed with 1× PBS by centrifugation for 5 min at 450 g. Cells were then stained extracellularly with specific antibodies at 4 °C for 30 min to determine their phenotype, followed by incubation with Live/Dead detection reagent as described above. Fixation of these cells was performed by adding cytofix/cytoperm buffer (BD Bioscience) for 1 h at 4 °C, washing and then resuspension in 1× PBS. LAA fluorescence was analysed by flow cytometry.

RNA-seq, scRNA-seq and data analyses

Single-cell suspensions generated from lungs of Mtb-infected and uninfected mice were used for RNA-seq or scRNA-seq studies. For RNA-seq, total RNA was isolated from the suspensions using the RNeasy Plus mini kit (QIAGEN). For all samples, low-quality bases were removed and adapters were trimmed using Trimmomatic v.0.32. After the quality check, sequences were aligned to the Mus musculus genome (GRCm39) using STAR (v.2.7.0). After mapping, the output was converted to count tables with the tximport package in R (4.2.1). The count gene expression matrix was examined using the DESeq2 package in R (4.2.1) to identify DEGs. The changes in gene expression levels were considered significant when statistical test values (false discovery rate (FDR) adjusted P value) were lower than 0.05 and the fold change/difference higher than ±1.4. Candidate DEGs were visualized in volcano plots and Venn diagrams using the VennDiagram package in R (4.2.1). The obtained DEGs were scanned with the REACTOME pathway database using the compare Cluster package in R (4.2.1).

For scRNA-seq, equal numbers of live cells from each sample of the Bach1−/− and C57BL/6 WT groups were pooled, stained with propidium iodide at a dilution of 1:100 and live cells were sorted for each group. From this sorted population, 10,000 cells of each group were loaded per lane on a 10X Genomics Next GEM chip, and single-cell GEMs were generated using a 10× Chromium Controller. Subsequent steps to generate complementary DNA libraries were performed following the 10× Genomics protocol. Libraries were pooled and sequenced on an Illumina NextSeq 2000 system. The sequenced data were processed using Cell Ranger (v.6.1.2) to demultiplex the libraries. The count tables were then further processed and analysed using Seurat (v.4.0) in R (v.4.1.0). Cells were then filtered for less than 8% mitochondrial contamination and data were normalized, scaled and merged. FindVariableFeatures and RunHarmony functions were used to integrate the data. Principal components analysis was performed to find neighbours and clusters, and UMAP reduction was performed with 30 dimensions. FindAllMarkers with a filter of log fold change ≥0.25 and percent of cells expressing the marker ≥0.25 was used to identify gene markers that distinguish the cell clusters, and the clusters were manually assigned cell types on the basis of identified canonical markers (Extended Data Fig. 4). FindMarkers was used to identify DEGs between groups. Genes with a log fold change ≥0.5, percent of cells expressing the marker ≥0.2 and adjusted P ≤ 0.05 were considered significant and displayed using a volcano plot. AddModuleScore was used to calculate the average expression of a set of genes in the ferroptosis pathway, and results were visualized using feature and violin plots.

In vitro macrophage infection and cell death measurement

Mtb was grown in complete 7H9 broth media at 37 °C for 7 d and bacterial concentration determined by spectrophotometry at 600 nm. Mtb cultures were centrifuged at 5,000 r.p.m. for 10 min, resuspended in OptiMEM and sonicated for 30 s to reduce bacterial clumping. BMDMs were exposed to either H37Rv or H37Rv-RFP at an MOI of 10 for 3–4 h, washed three times with 1× PBS and then cultured in fresh OptiMEM media. L929-conditioned media were added to the cultures to a final concentration of 2.5% on days 1 and 3.

Necrotic cell death was evaluated by staining adherent cells with Fixable Viability Dye eFluor780 (eBioscience) according to manufacturer protocol. BMDMs were stained with Live/Dead staining solution (1:500 diluted in 1× PBS) at room temperature for 10 min in the dark and then incubated with anti-CD11b antibody (eBioscience) for an additional 20 min. Macrophages were washed with 1× PBS following centrifugation at 450 g for 5 min and fixed with cytofix/cytoperm buffer (eBioscence) for 1 h at 4 °C. Fixed cells were then detached, washed, resuspended in 1× PBS with 1% BSA (MP Biomedicals) and analysed by flow cytometry.

Mitochondrial superoxide assay

BMDM cultures were washed with Hankʼs balanced salt solution with calcium and magnesium (HBSS/Ca/Mg; Gibco) following centrifugation at 450 g for 5 min to remove residual culture media. Cells were stained with 5 μM MitoSOX dye (diluted in HBSS/Ca/Mg; Life Technologies) at 37 °C for 10 min according to manufacturer protocol. Extracellular staining for CD11b as cell fixation was performed as described above. MitoSOX fluorescence intensity was measured using a flow cytometer.

Quantification and statistical analysesStatistics

The results shown in figure legends are presented as mean ± s.e.m. The sample size (n) and numbers of independent experiments for the in vivo experiments are described in the graphics or provided in the figure legends. For in vitro experiments, the number of experimental replicates is described in the figure legend. Cytokine/chemokine levels were normalized to total protein concentration per experiment using GraphPad Prism 9.0 software, and clustered and visualized as a heat map using the R package pheatmap. A publicly available RNA-seq data obtained from https://ogarra.shinyapps.io/tbtranscriptome/ as referred to in ref. 33 was re-analysed to determine mRNA levels of BACH1 in TB patients as well as in different experimental settings in vivo. Statistical analyses were performed with GraphPad Prism 9.0 using either unpaired two-tailed t-test for comparison between two groups or one-way analysis of variance (ANOVA) for multiple comparisons. The median values with interquartile ranges were used as measures of central tendency and dispersion, respectively, for parameters whose values exhibited a non-Gaussian distribution. The Mann–Whitney test (for two groups), Kruskal–Wallis test with Dunn’s multiple comparisons or linear-trend post-hoc tests (for more than two groups) were used to compare continuous variables. Statistical differences were considered significant when P < 0.05.

Figure visualization

Figures were generated in Adobe Illustrator 2023 (v.27.7) and R (4.2.1), incorporating images from BioRender.com.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.