Ethics statement

Human samples and clinical data in this study were obtained from studies approved by the Samsung Medical Center Institutional Review Board (clinicaltrials.gov identifier: NCT00970801) and the Institutional Research and Ethics Committee at Chungnam National University Hospital (approval numbers: CNUH 2019-04-046, CNUH 2020-07-082, CNUH 2020-09-015; Daejeon, Korea). All patients provided informed consent, and the protocol adhered to the guidelines of the Declaration of Helsinki.

Patients and clinical samples

ATG7 and inflammatory cytokines/chemokines (IL1B, IL6, CXCL2, and CCL2) expression in PBMCs was evaluated in NTM patients (Mabc, n = 31; Mmass, n = 22) and HCs (n = 39) from the previously studied cohort²⁵. Thirteen surgical lung resection patients were selected, and their NTM infection was confirmed through microbiological evaluation. These patients were diagnosed at Chungnam National University Hospital (Daejeon, Korea) between January 2018 and December 2022. Sputum samples were collected for acid-fast bacilli smears and cultures. TB/NTM genotyping was performed using real-time PCR with the PANA qPCR™ assay (Panagene, Korea). Lung tissues were obtained from FFPE tissue blocks.

Data analysis of the human PBMC samples

nCounter assay results for human PBMCs from NTM cohort were obtained from a previous study (GSE290289)²⁵. Raw counts were normalized using positive controls and housekeeping genes with nSolver 4.0 software. The normalized counts for each gene were then compared between the NTM-PD patients and HCs using t-tests. For the heatmap, the normalized counts were converted to Z-scores, which were calculated based on the mean and standard deviation of each gene. ATG7 mRNA levels were measured by qRT-PCR and normalized to GAPDH expression using the 2^−ΔΔCt (delta-delta Ct) method. The primer sequences are listed in Supplementary Table 3.

Mice

The mice used in individual experiments were age- (6–8 weeks old) and sex-matched. To generate transgenic in innate immune cell-specific Atg7-deficient mice (Atg7 ^fl/flLysM-Cre^+/+) using the Cre/loxP recombination system, Atg7-floxed mice (Atg7 ^fl/fl, C57BL/6 background)⁶² were crossed with B6.129P2-Lyz2 ^tm1(cre)Ifo/J mice (004781, The Jackson Laboratory) expressing Cre recombinase under the control of a Lyz2 promoter. F1 progeny from Atg7 ^fl/flLysM-Cre^+/+ × Atg7 ^fl/fl crosses were genotyped for the presence (LysM–Cre⁺) or absence (LysM–Cre⁻) of the LysM–Cre allele. Cre-mediated recombination was confirmed by PCR using genomic DNA from mouse tails. The genotyping primers were as follows: IMR3066 (Mutant reverse, 5′-CCCAGAAATGCCAGATTACG-3′); IMR3067 (Common, 5′-CTTGGGCTGCCAGAATTTCTC-3′); IMR3068 (WT reverse, 5′-TTACAGTCGGCCAGGCTGAC-3′). The absence of expression of ATG7 in BMDMs was confirmed by immunoblot analysis using an ATG7 antibody (8558, 1:1000, Cell Signaling). Mice were housed at 21–22 °C, with 30–70% humidity on a 12 h light/dark cycle with ad libitum access to food and water. All procedures involving animals were reviewed and approved by the Institutional Animal Care and Use Committee, Chungnam National University College of Medicine (202109A-CNU-180; Daejeon, South Korea). These procedures were conducted in compliance with the guidelines of the Korean Food and Drug Administration.

Cells

The isolation of PBMCs from heparinized venous blood was conducted using the density gradient centrifugation method using Ficoll-Hypaque (Lympoprep; 1114545, axis-sheid). Plastic adherent monocytes derived from PBMCs were cultured in complete RPMI 1640 (10-040-CV, Corning) supplemented with recombinant human macrophage-colony stimulating factor (M-CSF; M6518, Sigma Aldrich) for 7 days to differentiate MDMs⁶³ at placed at 3 × 10⁶ cells per 96-well plate. Primary BMDMs isolated from Atg7 cWT and Atg7 cKO mice were cultured in Dulbecco’s modified Eagle’s medium (DMEM; 10-013-CVRC, Corning) supplemented with 10% fetal bovine serum (FBS; 16000-044, Gibco), 1% Antibiotic-Antimycotic (15240062, Gibco). BMDMs were differentiated for 5 days in the presence of M-CSF (416-ML-010, R&D Systems). HEK293T (ATCC, CRL-3216) cells were cultured in DMEM and maintained at 37 °C in a humidified incubator with 5% CO₂.

Mycobacterial strains and culture

The smooth and rough ATCC19977 strain of Mabc, red fluorescent protein (RFP)-labeled Mabc-R, and GFP-labeled Mabc-S, Mint (ATCC 13950), Mav (ATCC 25291), and Mmass (KMRC-00136-13018) were used for this study. Mabc CIP 104536 T R and S morphotype carrying a pMV262-mWasabi or RFP plasmid enabling the expression of mWasabi or RFP were employed. Mabc CIP 104,536 S- and R-type strains were generously provided by Laurent Kremer (Université de Montpellier, Montpellier, France). Each Mycobacterium strain was cultured on 7H10 (Difco, 262710) plates supplemented with 10% oleic albumin dextrose catalase (OADC; 212240, BD Bioscience) for 4 days or 2–4 weeks to confirm colony formation and the absence of contamination. A single colony was selected using a loop and inoculated into 25 ml of 7H9 liquid medium without added surfactant but supplemented with 10% OADC. This pre-culture was incubated at 37 °C for 3 days or 2–4 weeks before initiating the main culture. Once the absorbance reached an OD_600nm of 0.5 or higher, 10 ml of the seed culture was transferred to 1 l of the same main culture medium composition and incubated at 37 °C with rotary shaking. The culture was terminated when the absorbance approached an OD_600nm of 0.6. The bacterial culture was then centrifuged to pellet the cells, and the medium components were removed by resuspending and centrifuging the cells three times in PBS buffer. Following centrifugation, an equal volume of glass beads was added to the pelleted bacterial cells, and twice their volume of PBS was used to homogenize the mixture using a Mixer Mill 400 (Retsch, Haan, Germany) at 30 Hz for 4 min, resulting in the separation of the mycobacteria into single-cell units. The homogenized mycobacteria were then aliquoted into cryogenic vials for storage. Upon thawing, the strains were diluted and quantitatively cultured on agar plates to verify the colony-forming units per milliliter (CFU/mL).

Antibodies and reagents

Anti-IL-1β (1:1000, CST#12242), anti-Caspase-1 (1:1000, CST#2225), anti-Caspase-3 (1:1000, CST#9662), anti-cleaved Caspase-3 (1:1000, CST#9661), anti-GSDMD (1:1000, CST#39754), anti-ATG7 (1:1000, CST#8558), anti-ACTB (1:4000, CST#4970), anti-mouse IgG (1:4000, CST#7076) and anti-rabbit IgG (1:4000, CST#7074) were purchased from Cell Signaling Technology for western blotting. Anti-DFNA5/GSDME [EPR19859]-N-terminal (1:1000, ab2159) were purchased from Abcam. Anti-LC3A/B (1:400, PM036) was purchased from MBL International. Anti-LAMP1 (1:400, sc-19992) was purchased from Santa Cruz Biotechnology. Alexa Fluor 405-conjugated anti-rat IgG (1:400, A48261), Alexa Fluor 488-conjugated anti-rabbit IgG (1:400, A17041), Alexa Fluor 594-conjugated anti-rabbit IgG (1:400, A21207), Pam₃CSK₄ (tlrl-pms), and Dihydroethidium (1:200, D11347) were purchased from Invitrogen. MitoTracker Deep Red FM (M22426) was purchased from Thermo Fisher Scientific. RIPA buffer (CBR002) was purchased from LPS solution and 1× blocking solution (OP105-500) was from Biofact. Protease inhibitor (4693116001) and phosphatase inhibitor (4906837001) cocktails were purchased from Roche. MPO (1:200, AF3667) was purchased from R&D System. Anti-Histone H3 [Citruline Arg17, Citruline Arg2, Citrulline Arg8] (1:200, NB100-57135) was purchased from Novus Biologicals. InVivoMAb mouse IgG2a isotype control (BE0085), InVivoMAb anti-mouse Ly6G (BE0075-1) were purchased BioXCell. Alexa Fluor™ 700 rat anti-mouse CD45 (1:250, 56-0451-82), PE-Cyanine7 rat anti-Ly6G (25-9668-82) were purchased from Invitrogen. Zombie Violet™ Fixable Viability Kit (423114), APC anti-mouse CD64 (FcγRI) (1:250, 161006), and PE anti-mouse MERTK (Mer) (151506) were purchased from BioLegend. APC-Cy™7 rat anti-CD11b (1:250, 557657) and Annexin V-PI staining kit (556547) were purchased from BD Biosciences.

RNA extraction and qRT-PCR

Total RNA from cells and tissues was isolated using TRIzol reagent (15596018, Invitrogen) according to the manufacturer’s instructions. cDNA synthesis was performed using Reverse Transcriptase Premix (EBT 1515C, ELPIS-BIOTECH). qRT-PCR was performed using SYBR Green Master Mix (208056, Qiagen) and primers for the indicated genes in a Rotor-Gene Q 2plex system (Qiagen). Data were analyzed using the 2^–ΔΔCt threshold cycle method; Gapdh was used for normalization. Primer sequences (mouse, human) are shown in Supplementary Table 3.

Intracellular CFU assay

For the assessment of intracellular bacterial viability, Mabc, Mav, Mmass, or Mint-infected cells were lysed in distilled water to release the intracellular bacteria. Serial dilutions of the lysates were plated on 7H10 agar plates and incubated at 37 °C. After 3–4 days (Mabc and Mmass) or 21 days (Mav, Mint) of incubation, the colonies were counted to measure the bacterial loads.

Mouse in vivo infection model

Atg7 cWT and Atg7 cKO (8–10 weeks old) mice were anesthetized and challenged intranasally with Mabc-S (1 × 10⁶ or 1 × 10⁷ CFU/mouse), Mabc-R (2 × 10⁶ CFU/mouse), Mav (1 × 10⁶ or 1 × 10⁷ CFU/mouse), Mint (1 × 10⁷ CFU/mouse), or Mmass (1 × 10⁷ CFU/mouse). At the indicated times points post-infection, the mice were euthanized and the lungs were harvested to measure the bacterial burden. Lung tissues were homogenized using a tissue homogenizer (OMNI TH; OMNI Inc.) in PBS with detergent (PBST), and serial dilutions of the homogenates were plated on 7H10 agar plates. After 3–4 days (Mabc and Mmass), or 21 days (Mav and Mint) of incubation the colonies formed in the plates were counted.

Lentiviral shRNA production and transduction

shRNA plasmids (ATG7, sc-41447-SH, Santa Cruz Biotechnology) were used for silencing of human ATG7. For lentivirus production, ATG7 shRNA plasmid, pRSV-Rev (12253, Addgene), pMDLg/pRRE (12251, Addgene), and pMD2.G (12259, Addgene) were transfected into HEK293T cells (crl-3216, ATCC) via the Lipofectamine 3000 (L300015, Invitrogen) for 72 h. After transfection, the supernatants were collected and filtered through the filter with 0.45 μm pore (Millipore) and then centrifuged. The collected supernatants were stored at −80 °C. For lentiviral infection, human primary monocytes and MDMs cultured in 96-well plates were infected with a lentiviral vector (MOI of 10) for 36 h, followed by subsequent NTM infection.

Neutrophil depletion

Atg7 cWT and Atg7 cKO mice were infected intranasally with Mabc-S or Mav (1 × 10⁶ CFU/mouse). For in vivo neutrophil depletion, mice were treated with 0.5 mg of anti-Ly6G mAb (neutrophil-depleting, clone 1A8; BE0075-1, BioXcell) or anti-IgG2a Ab (isotype control, clone 2A3; BE0085, BioXcell) via intraperitoneal injection every 1 or 2 days throughout the indicated depletion periods.

Immunohistochemical staining and analysis

For histological analysis in NTM patients, whole FFPE tissues were sectioned onto coated slides, deparaffinized with xylene, and hydrated in a series of alcohol solutions. The sections underwent antigen retrieval by heating in a pressure cooker (containing 10 mmol/L sodium citrate [pH 6.0]) for 3 min. Endogenous peroxidase blocking was achieved using 0.03% hydrogen peroxide for 10 min. The sections were then incubated at room temperature for 1 h with the following primary antibodies: ATG7 (1:500, ab54272, Abcam). Liver hepatocytes served as positive controls, and tonsil tissues as negative controls. For automated staining quantification, a two-step computational algorithm (based on color deconvolution) was validated using QuantCenter software (available on 3DHistech image analysis platform, Sysmex, Budapest, Hungary). The DensitoQuant module was trained for the recognition of immunohistochemical staining.

Lung tissues dissected from Atg7 cWT and Atg7 cKO mice were fixed with 10% formaldehyde solution before paraffin embedding. The sectioned tissues (thickness of 4μm) were stained with Hematoxylin and Eosin (H&E) and then scanned using the PANNORAMIC 300 Flash DX device (3DHISTECH Ltd, Budapest, Hungary). To quantify the inflamed area from stained lung tissues, red-stained areas with high thresholds in three representative lungs were measured as a percentage of the total area using FIJI software. For immunofluorescence, lung paraffin blocks were sectioned and stained with specific targets: anti-MPO (1:200, R&D systems), anti-Histone H3 (1:200, Novus biologicals), Dihydroethidium (1:200, Invitrogen), anti-mouse Ly6G (1:200, BioXcell), and PI (1:200, P3566, Thermo Fisher Scientific). The sections were then washed three times in PBS (CBP007B, LPS solution), and then incubated in fluorescent antibodies: Alexa Fluor 488-conjugated Goat anti-rabbit IgG (1:400), Alexa Fluor 594-conjugated Donkey anti-rabbit IgG (1:400). Coverslips were then mounted onto the slides c using Prolong™ Gold Antifade Mountant with DNA Stain DAPI (P36931, Invitrogen). Immunofluorescence images were acquired with Zeiss LSM 900. 3D reconstruction image acquisition by Aivia (Leica microsystems).

Enzyme-linked immunosorbent assay (ELISA)

Cell supernatants and tissue homogenates were collected and stored at −80 °C. Supernatants were used to measure mouse IL-1β (88-7013-22, Invitrogen), mouse IL-6 (555240, BD Bioscience), mouse TNF (560478, BD Bioscience), human IL-1β (557953, BD Bioscience), and human IL-6 (555220, BD Bioscience) secretion according to the manufacturer’s instructions.

Western blotting

Cell lysates from Atg7 cWT and Atg7 cKO mice BMDMs and lung homogenates were prepared in RIPA buffer (CBR002, LPS solution) containing protease inhibitor (4693116001, Roche) and phosphatase inhibitor (04906845001, Roche). Samples were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to nitrocellulose membrane. Membranes were blocked in 1x blocking solution (OP105-500, Biofact) for 30 min at room temperature. The membranes were probed overnight at 4 °C with primary antibodies as described: Anti-pro-IL-1β, anti-Caspase-1, anti-Caspase-3, anti-cleaved Caspase-3, anti-GSDMD, anti-ATG7, anti-ACTB, and anti-DFNA5/GSDME. The membranes were washed using TBS-T buffer and further incubated for 1 h at room temperature with appropriate secondary antibodies: anti-mouse IgG (Cell Signaling Technology, #7076, 1:4000) and anti-rabbit IgG (Cell Signaling Technology, #7074, 1:4000). The immune-reactive proteins were detected by using Immobilon western chemiluminescent HRP substrate (WBKLS0500, Merck), and were visualized by iBright^TM Image system (CL750, Invitrogen). The intensity value of target protein bands was normalized by using ImageJ software (National Institutes of Health, USA, 1.52a) for densitometry analysis.

Tissue processing and spatial transcriptomic generation

Following the resection of lungs from Mav-infected Atg7 cWT and Atg7 cKO at 21 dpi, tissues underwent fixation in 10% formalin and subsequent embedding in paraffin. RNA extraction was performed on FFPE samples utilizing the RNeasy FFPE Kit (#73504, Qiagen), followed by DV 200 evaluation using an Agilent TapeStation. FFPE tissue samples were then prepared according to the Visium Spatial Gene Expression for FFPE protocol (#CG000409, 10xGenomics Pleasanton, CA, USA). Libraries were constructed utilizing the Visium Spatial Gene Expression for FFPE Kit, Mouse Transcriptome v.1 (PN-1000521, 10× Genomics) along with the Dual Index Kit TS Set A (PN-1000251, 10xGenomics). Subsequent sequencing was conducted on an Illumina NovaSeq6000 S1 Rgt Kit v1.5 200 cycles, employing paired-end 200 bp FlowCells. For the data analysis pipeline, reads were aligned to the mouse reference genome (mm10-2020) using Space Ranger (10× Genomics). Matrix and H&E image analysis were performed using Seurat v.4.0 (R 4.1) for further analysis.

Spatial transcriptomic data pre-processing

The aggregated HDF5matrix was imported into R and split by sample. Feature-barcode matrices for each sample were imported into the R package “Seurat” (Version 5.0.1) for normalization, quality control, batch effect correction, dimensionality reduction, and Louvain clustering. Spots expressing less than 300 features were excluded from downstream analysis.

scRNA sequencing and data analysis

Following the resection of lungs from Mav-infected Atg7 cWT and Atg7 cKO at 21 dpi, tissues were dissociated using Liberase TL (#05401020001, Roche). Cell viability and number were assessed using the LUNA-FX7™ Automated Cell Counter (Logos Biosystems) with acridine orange/PI stain. Single-cell suspensions were generated using the Chromium Next GEM Single Cell 3’ Kit v3.1 (16 rxns, PN: 1000268) in conjunction with the Chromium Next GEM Chip G Single Cell Kit (48 rxns, PN: 1000120) and the Dual Index Kit TT Set A (96 rxns, PN: 1000215), following the manufacturer’s protocol (10xGenomics). Libraries were sequenced on an Illumina NovaSeq 6000 platform using the NovaSeq 5000/6000 S2 Reagent Kit (200 cycles) v1.5 (PN: 20028315). The sequencing was performed in paired-end mode with Read1: 28 cycles, Read2: 90 cycles, i7 Index: 10 cycles, and i5 Index: 10 cycles. Reads were aligned to the mouse reference genome (mm10-2020-A) using Cell Ranger v8.0.1 (10× Genomics) and gene expression matrices were generated for downstream analysis. Low-quality cells expressing fewer than 200 genes were removed from the dataset. After quality control, 25,040 cells were retained. Seurat’s standard preprocessing workflow was applied, including normalization, data scaling, Harmony integration for batch effect correction, dimensionality reduction, and clustering based on the first 30 principal components. Thirty-seven cell clusters were annotated into 22 cell types based on differentially expressed genes and canonical cell type markers, with references from the PanglaoDB, including AT1 (Hopx, Akap5), AT2 (Sftpc, Stfpa1), Ciliated (Scgb1a1, Sec14l3), Capillary (Cd93, Hpgd), Vein (Vwf), Lymph (Fbln2, Ccl21a, Mmrn1), Fibroblast (Col14a1, Col13a1), Pericyte (Acta2, Gucy1a1), Myeloids (Cd68, Itgam, Fcgr3), T cells (Cd3d, Cd4, Cd8a), NK cells (Nkg7, Gzma, Ccl5), B/Plasma cells (Cd19, Igkc, Ighm, Mzb1), and five innate immune cell types: Interstitial macrophage (Cd68, Fcgr1, Stab1, Aoah, Sdc3), alveolar macrophage (Marco, Siglecf, Itgax, Olr1), neutrophil (Csf3r, Ly6g, Il1r2, S100a8, Trem1), monocyte (Ccr2, Ifitm3, Tet2, Sirpb1c, Klra2, Gpr141), and dendritic cell (Zbtb46, Slamf7, Traf1).

Cell type population heatmap with deconvolution

We conducted deconvolution of cell types within spatial transcriptomic spots using the python package “Cell2location”, with the scRNA sequencing data from GSE151974 serving as the reference²⁸. Subsequently, we integrated the results using Seurat object metadata into the R package “Seurat” for downstream analysis. Mean average expression in each spot was calculated based on unsupervised clusters, and the results were visualized using the Heatmap function in the R package “pheatmap” (Ver 1.0.12).

Differential expression analysis to pathway enrichment analysis (GO Biological Process)

Differentially expressed genes were identified using the “FindMarkers”, “MAST” and “FindAllMarkers” function in the “Seurat”. Pathway enrichment analysis was performed using the R package “ClusterProfiler” (Ver 4.10.0) with gene identifiers mapped using the R package “org.Mm.eg.db” package (Ver 3.18.0). Statistical significance for enrichment was assessed using the hypergeometric test with Benjamini-Hochberg correction for multiple comparisons, as implemented in the “ClusterProfiler” package, with a p-value threshold of <0.05. The selected GO terms represent biological processes. As enrichment testing is inherently one-sided, the test directionality was not explicitly stated.

Statistical approach for comparing gene set scores

The average expression of selected gene set signatures from signaling pathways in the databases HALLMARK (HALLMARK_INFLAMMATORY_RESPONSE, HALLMARK_IL6_JAK_STAT3_SIGNALING, HALLMARK_IL2_STAT5_SIGNALING, HALLMARK_TNFA_SIGNALING_VIA_NFKB, HALLMARK_REACTIVE_OXIGEN_SPECIES_PATHWAY), REACTOME (R-MMU-5620971, R-MMU-5218859), KEGG (mmu04215), and GO_BP (GO:0097527) was calculated for each spot and cell in the Spatial Transcriptomes and scRNA sequencing datasets using the “AddModuleScore” function. Data visualization was performed using “DimPlot”, “Featureplot”, and “SpatialFeaturePlot” function in the “Seurat”.

Transmission electron microscopy

For transmission electron microscopy analysis, lung tissues or BMDMs from Atg7 cWT and Atg7 cKO were sequentially fixed with 2.5% glutaraldehyde and 1% osmium tetroxide on ice for 2 h and washed with PBS. The cells were then dehydrated in ethanol and propylene oxide series, embedded in Epon 812 mixture, and polymerized in an oven at 60 °C for 24 h. The 70 nm sections acquired from polymerized blocks were collected on 150 mesh copper grids, counterstained with 4% uranyl acetate for 10 min and lead citrate for 10 min, and examined with a KBSI Bio-HVEM system (JEM-1400Plus at 120 kV and JEM-1000BEF at 1000 kV, JEOL). Data were collected using a OneView camera with DM software (Image size: 4 × 4k, Gatan Inc., Pleasanton, CA, USA). ImageJ software was used for the quantification of damaged mitochondria.

Flow cytometric analysis

To analyze apoptotic cells, Mabc-R-infected BMDMs were harvested and stained with FITC-conjugated Annexin V and PI as provided by the Annexin V-PI staining kit (556547, BD Biosciences). For neutrophil infiltration staining, Mabc-R or Mav-infected Atg7 cWT and Atg7 cKO lung samples were harvested using a mouse lung dissociation kit (130-095-927, Miltenyi Biotech). Briefly, the lungs of the mice were transferred to a C tube containing the 1× S buffer with the mix solution (enzyme D and A in S buffer). The GentleMACS programs were run with a 30 min incubation on the MACSmix™ Tube Rotator at 37 °C. The cell suspension was subsequently passed through a 70-µm strainer, and the cell pellets were resuspended in Zombie-violet Live-Dead stain (1:1000) and incubated for 15 min in the dark. Cells were then washed with PBS containing 2% FBS (v/v). Cells were then stained for 30 min at 4 °C with following specific antibodies: Alexa Fluor™ 700 rat anti-mouse CD45 (1:250), APC anti-mouse CD64 (FcγRI) (1:250), PE anti-mouse MERTK (Mer) (1:250), PE-Cyanine7 Rat Anti-Ly6G (1:250), and APC-Cy™7 Rat Anti-CD11b (1:250). All stains were carried out in PBS containing 2% FBS (v/v). After washing twice with PBS containing 2% FBS, stained cells were resuspended in 2% paraformaldehyde and assayed immediately on the BD FACSDiscover™ S8 Cell Sorter (NFEC-2025-02-303724, BD Biosciences). Data were collected and analyzed using BD Flow Jo (ver10.8.1, BD Biosciences), BD FACSChorus™ software (BD Biosciences).

Immunofluorescence and confocal microscopy analysis

Cells were cultured on coverslips and infected with NTM. Subsequently, the cells were washed three times with PBS, fixed with 4% paraformaldehyde for 15 min, and permeabilized with 0.25% Triton X-100 (Sigma-Aldrich) for 10 min. After a 2 h incubation with primary antibodies at room temperature, cells were washed with PBS to remove excess primary antibodies. The samples were incubated with secondary antibodies for 2 h and nuclei were stained with DAPI for 5 min at room temperature. After mounting, fluorescence images were captured using confocal laser microscope (TCS SP8, Leica).

To quantify the colocalization of mycobacteria with LC3/LAMP1 and mitotracker with LC3/LAMP1, images were captured using confocal laser-scanning microscopy. Each experiment was performed on triplicate coverslips, and the results are expressed as the mean and standard deviation. Pearson’s correlation coefficient was measured by ImageJ software.

Measurement of mtROS

To quantify mtROS, BMDMs and lung tissues were incubated with 2.5 μM MitoSOX^TM Red Mitochondrial Superoxide Indicator (M36007, Invitrogen). After 20 min, the cells were washed and measured by confocal microscope (LSM900, Zeiss). MitoSOX intensity was analyzed using ImageJ software.

RNA sequencing analysis

Paired-end sequencing reads were generated on the Illumina sequencing NovaSeq platform. Trimmomatic v0.38 was used to remove adapter sequences and trim bases with poor base quality. The cleaned reads were aligned to the Mus musculus (mm10) using HISAT v2.1.0, based on the HISAT and Bowtie2 implementations. Aligned data (SAM file format) were sorted and indexed using SAMtools v 1.9. After alignment, the transcripts were assembled and quantified using StringTie v2.1.3b. Gene-level and Transcript-level quantification was calculated as raw read count, fragments per kilobase of transcript per million mapped reads (FPKM) and transcripts per million. Differential expressed gene (DEG) analysis and RLE normalization were performed by DESeq2 v 1.38.3. Statistical significance of differential expression gene was determined using DESeq2 nbinom WaldTest. Fold change and P-value were extracted from the result of WaldTest. All P-values are adjusted by Benjamini–Hochberg algorithm to control false discovery rate. The significant gene list was filtered by |fold change| ≥ 2 & raw P-value < 0.05. Gene-enrichment and functional annotation analysis for significant gene list were performed based on gProfiler (https://biit.cs.ut.ee/gprofiler/orth), and KEGG pathway database (https://www.genome.jp/kegg/) respectively. The heatmap data was produced utilizing the heatmap (version 1.0.12). All data analysis and visualization of differentially expressed genes was conducted using R 4.2.2 www.r-project.org.

Optical bioluminescence imaging

For optical fluorescence imaging, Atg7 cWT and Atg7 cKO were intranasally infected with ERFP-Mabc-R (2 × 10⁶ CFU/mouse) for 7 days. To check bacterial colonization in the lung, ex vivo fluorescence imaging was performed using an optical imaging system (Night OWL LB983, Berthold Technologies, USA). At 7 days after bacterial infection, mice were sacrificed and lungs were excised for ex vivo imaging. ERFP-Mabc-R were detected with excitation/emission wavelength at 580/620 nm with an exposure of 7 s. Data were analyzed with IndiGo software (Berthold Technologies).

Statistics and reproducibility

For statistical analysis, data obtained from at least three independent experiments [mean ± standard deviation (SD) or standard error of the mean (SEM)] were analyzed using a two-tailed Student’s t-test, Mann–Whitney U-test, and one-way analysis of variance (ANOVA) with Tukey’s or Dunnett’s multiple comparison tests. The sample size was determined based on biological/technical replicates with sufficient reproducibility. For analysis of the human PBMC sample, 3 groups (NTM patients with Mabc and NTM patients with Mmass as experimental groups and healthy group as the control group) were examined by Kruskal-Wallis test with post hoc using Dunn’s multiple comparison test. For the Visium spatial transcriptomics and scRNA sequencing data, each group (Atg7 cKO as the experimental group and Atg7 cWT as the control group) consisted of one mouse. The unit of analysis was “spot” for Visium and “cell” for scRNA sequencing. Sample sizes (n) for all plots are indicated in the corresponding figure legends. Two-group comparisons of module scores between Atg7 cKO and Atg7 cWT were performed using two-sided, unpaired Welch’s t-tests (R packages ggpubr v0.6.0 and ggplot2 v3.4.4). For comparisons among more than two groups, normality was assessed using the Shapiro-Wilk test (stats package, v4.3.2), and homogeneity of variances was evaluated using Levene’s test (car package, v3.1-2). If both assumptions were met (P  >  0.05), a one-way ANOVA (stats package, v4.3.2) was performed, followed by Tukey’s Honestly significant difference (HSD) post hoc test for pairwise comparisons. Effect sizes (Cohen’s d) and their 95% confidence intervals were calculated using the effsize package (v0.8.0) and are provided in the Source Data file. Statistical significance was defined as P  <  0.05. Exact p-values are reported in the figures or figure legends; p-values below the numerical precision limit of R (double-precision floating point) are reported as being less than the corresponding threshold.

Reporting summary

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