Sex as a biological variable. Our study examined both male and female mice, with similar findings observed in both sexes. However, in certain experiments as noted, only male mice were used to minimize variability in phenotype.

Animals. All mice were in the C57BL/6J background, and littermate controls were used for experiments when feasible. C57BL/6 mice were purchased from CLEA Japan and bred and housed under specific pathogen-free conditions. Mice deficient for Aldh1a1 were generated by CRISPR/Cas9-mediated gene targeting in C57BL/6 zygotes using CRISPR RNA (5′-AGTTCTTAACCCTGCAACTG-3′) as targeting guide, and STOP sequences were inserted by electroporation. Mice were group-housed and fed standard chow at ambient temperature of 25°C with 50% humidity on average and a 12-hour light-dark cycle in individually ventilated cages.

Reagents. Naphthalene and disulfiram were purchased from FUJIFILM Wako Pure Chemical Corporation. DEPs (SRM2975; National Institute of Standards and Technology) and PM (SRM2786; National Institute of Standards and Technology) were purchased from Sigma-Aldrich. Alda-1 was purchased from MedChemExpress.

Alda-1 in vivo administration. Alda-1 (in 50% DMSO/50% PEG, at 0.8 mg/kg/h) or vehicle control (50% DMSO/50% PEG) was administered using an ALZET osmotic pump (DURECT Corp.) as described previously (69). Osmotic pumps were implanted 2 or 3 days before naphthalene or DEP exposure to minimize the impact of surgical procedure on the outcome of the experiments.

Naphthalene exposure. Eight- to nine-week-old mice (C57BL/6, Aldh1a1–/–, Aldh1a1+/+, and Aldh1a1+/– littermates or age-matched pairs) received intraperitoneal injection of naphthalene (200 mg/kg). For trachea tissue live imaging, naphthalene was administered 4 h prior to imaging. For histology, immunofluorescence, electron microscopy, flow cytometry, mucociliary transport assay, RNA-seq, and pneumonia experiments, mice received the second naphthalene injection 2 weeks after the first injection. Two weeks after the second injection, mice were sacrificed or used for pneumonia experiments as described below.

For Alda-1 treatment, WT mice were divided into 4 groups: vehicle control without injury, Alda-1 without injury, vehicle control with naphthalene exposure, and Alda-1 with naphthalene exposure. Four days after intraperitoneal injection of naphthalene (200 mg/kg), mice were sacrificed and trachea and lung tissues were isolated for mucociliary transport assay and western blotting, respectively.

DEP exposure. DEP powder was suspended in PBS at 2 mg/mL (w/v), vortexed for 2 minutes, sonicated for 10 minutes in a cooled water bath, aliquoted, and stored at –80°C until use. Upon thawing for each experiment, aliquots were sonicated for 5 minutes immediately before use. Mice (C57BL/6, Aldh1a1–/– versus WT or Aldh1a1+/– littermates or age-matched pairs, aged 10–12 weeks) were lightly anesthetized with isoflurane and exposed to either 100 μg DEPs or PM in 50 μL of PBS or 50 μL PBS without DEPs or PM. For trachea tissue live imaging, DEPs or PM were administered twice at 2 and 16 h prior to imaging. For MDA measurement in BALF, DEPs were administrated 3 times per day for 2 consecutive days (6 times total). Alternatively, DEPs were administered 6 times every second day, and mice were sacrificed 24 h after the last exposure for quantitative PCR or 3 days after the last exposure for histology, immunofluorescence, flow cytometry, and mucociliary transport assay. Pneumonia experiments were performed 1 day after the sixth exposure as described below.

MDA measurement. After DEP or PBS administration, the trachea was exposed under deep anesthesia. BALF was collected in 1 mL of PBS using a 22G catheter and centrifuged at 1,000g at 4°C for 5 minutes, and the supernatant was collected for assay. The concentration of free MDA was quantified utilizing an aromatic hydrazine-based method, employing the Colorimetric Lipid Peroxidation (MDA) Assay Kit (Abcam).

Histology and immunofluorescence staining of lung tissue sections. Mice were sacrificed using CO2 inhalation. Isolated lung tissues were fixed with 4% paraformaldehyde (PFA) overnight at 4°C and embedded in paraffin. Tissue sections (6 μm thickness) were prepared using a microtome (SLEE medical) and mounted onto adhesive glass slides (Matsunami Glass). Rehydration was performed with xylene, followed by a standard ethanol dilution series. For histology, sections were stained with H&E solutions (FUJIFILM Wako). For immunostaining, antigen retrieval was performed at 98°C for 45 minutes using an ImmunoSaver device (FUJIFILM Wako). Sections were permeabilized with 0.1% Triton X-100 for 10 minutes and then blocked with Blocking One Histo buffer (Nacalai Tesque) at room temperature for 1 hour. After blocking, sections were incubated with primary antibodies diluted in Can Get Signal immunoreaction enhancer solution (Toyobo) overnight at 4°C: ALDH1A1 (Cell Signaling Technology; D4R9V), 4-HNE (JaICA; HNEJ-2), CC10 (Abcam; EPR19846), CYP2F2 (Santa Cruz Biotechnology; F-9), and S. pneumoniae (Abcam; ab20429). After incubation with fluorescently conjugated secondary antibodies, nuclear staining with DAPI (100 nM in PBS, 5 minutes) was performed after the final antibody application. After DAPI staining, slides were washed 4 times in PBS and sealed in mounting medium (Prolong Diamond Antifade Mountant; Thermo Fisher Scientific) for microscopy observation.

ROS, lipid peroxide, and acrolein labeling, and plasma membrane staining of isolated trachea tissues for live imaging. Mice were euthanized by CO2 inhalation and tracheas were dissected. For ROS and lipid peroxide labeling, isolated trachea tissues were incubated with LipiRADICAL Green (2.5 μM; Funakoshi), CellROX Deep Red (10 μM; Thermo Fisher Scientific), and CellTracker Red CMTPX (1:1,000 dilution; Thermo Fisher Scientific) in DMEM/F12 without phenol red (Thermo Fisher Scientific) for 30 minutes at 37°C. For acrolein labeling, trachea tissues were incubated with AcroleinRED (10 μM; Funakoshi) and CellMask Plasma Membrane Deep Red (1:1,000 dilution; Thermo Fisher Scientific) for 20 minutes at 37°C. For time-lapse recording of cilia movement, trachea tissues were incubated with CellMask Plasma Membrane Orange (Thermo Fisher Scientific) at 1:1,000 dilution for 30 minutes at room temperature. After labeling, tissues were rinsed with DMEM/F12 twice and mounted using medical adhesive (Daiichi Sankyo) onto glass slides with 0.4 mm high ridges, covered with DMEM/F12, and sealed with cover glasses for microscopy observation.

Airway epithelial injury model in vitro. C57BL/6 male mice (6–8 weeks) were euthanized by CO2 inhalation, and ALI cultures of mouse airway epithelial cells were prepared from trachea and main bronchi without expansion, as described previously (70, 71). Briefly, after enzymatic digestion and fibroblast deprivation (70, 71), collected nonadherent cells were resuspended in MTEC proliferation medium: DMEM/F12 supplemented with 5% (v/v) FBS, Insulin-Transferrin-Selenium (Thermo Fisher Scientific), 1.5 mM l-glutamine, 0.1 μg/mL cholera toxin, 0.025 μg/mL murine EGF (PeproTech Inc.), 0.03 mg/mL bovine pituitary extract (Thermo Fisher Scientific), Y-27632 (Cayman Chemical), 0.05 μM RA (Sigma), and antibiotics (Penicillin-Streptomycin-Amphotericin B mixture; Lonza). The cells were then seeded onto a 6.5 mm Transwell 0.4 μm pore polyester membrane insert (Corning) at 8 × 104 cells/cm2 and incubated at 37°C with 5% CO2. At 100% confluence, differentiation was induced by removing apical media and replacing the basal media with differentiation medium: DMEM/F12 supplemented with 0.1% (w/v) Bovine Albumin Fraction V (Thermo Fisher Scientific), Insulin-Transferrin-Selenium (Thermo Fisher Scientific), 1.5 mM l-glutamine, 0.025 μg/mL cholera toxin, 0.005 μg/mL murine EGF (PeproTech Inc.), 0.03 mg/mL Bovine Pituitary Extract (Thermo Fisher Scientific), 0.1 μM RA (Sigma), and antibiotics (Penicillin-Streptomycin-Amphotericin B mixture; Lonza). For naphthalene-induced injury, naphthalene was applied in the basal differentiation medium at 10 μM for 10 days from differentiation day 7. After naphthalene exposure, cells were incubated in normal differentiation medium for 4 days, in the presence or absence of pan-ALDH inhibitor disulfiram (2 μM), depending on the experimental design. Ethanol and DMSO were used as carrier controls for naphthalene and disulfiram, respectively.

After exposure, cells on the membrane insert were fixed with 4% (w/v) PFA in PBS for 15 minutes at room temperature. After fixation, membranes were removed from the inserts and placed on a glass slide for immunostaining. After washing with PBS containing 0.05% (v/v) Triton X-100 (PBS-T), cells on the membrane were blocked with 5% (w/v) BSA in PBS-T for 1 hour; incubated with a combination of primary antibodies against acrolein (Abcam; 10A10), ALDH1A1 (Cell Signaling Technology; D4R9V), ZO-1 (Invitrogen; ZO1-1A12), ODF2 (Abcam; ab43840), or TUBA (Santa Cruz Biotechnology; 6-11B-1) in PBS-T containing 2% (w/v) BSA overnight at 4°C; and incubated with fluorescently conjugated secondary antibodies in PBS-T containing 2 % (w/v) BSA for 1 hour. After nuclear staining with DAPI in PBS-T for 5 minutes, membranes were washed with PBS-T and sealed with a drop of mounting medium (Prolong Diamond Antifade Mountant; Thermo Fisher Scientific) for microscopy observation.

Microscopy imaging and analyses. H&E images and some of the immunofluorescence images were captured using a BZ-X800 microscope (Keyence Corp.). Confocal imaging was performed using a STELLARIS 5 WLL confocal microscope (Leica Microsystems) equipped with LAS X software.

For trachea tissue live imaging, Z-stack images were captured using the confocal microscope, and a confocal image with maximal fluorescence intensity was selected for each analysis. CellROX positive and LipiRADICAL Green positive cells per 200 × 200 (40,000) μm2 were counted using ImageJ (NIH). For quantitative analyses of acrolein adducts in ALI culture (in vitro), Z-stack images were captured using the confocal microscope every 1 μm for a total of 34 sections, and merged gray scale images were analyzed using ImageJ software. For cilia height measurement, ALI culture samples from Aldh1a1+/+ and Aldh1a1–/– with or without naphthalene exposure were immunolabeled for TUBA and ZO-1, and fluorescence images were captured using the confocal microscope, as described above. The 3D reconstitutions of Z-stack images were used for cilia height measurement. For each sample, 2 images (96.88 × 96.88 μm2, each containing 20–30 ciliated cells) were captured, and cilia height was determined as the distance from ZO-1 to the top of the TUBA signal in each ciliated cell. Approximately 50 ciliated cells per sample were analyzed. Image analysis was performed using ImageJ.

For quantifying the percentage of ciliated epithelial surfaces in lung tissue sections, immunofluorescence images were captured using the BZ-X800 microscope from the top to the bottom of 6 μm sections every 0.6 μm and merged, and the total epithelial length and ciliated surface length of tissue samples were manually measured using the line tool of ImageJ.

For detailed morphological observation of cilia, super-resolution microscopy was performed using a Nikon AX R Confocal Microscope System with a Spatial Array Confocal detector equipped with ×100 objective lens (PLAN APO λD ×100 /1.45 oil) and NIS-Elements image acquisition software. Imaging of fixed samples was performed using Galvano mode. For recording cilia movement, time-lapse imaging was performed using resonant mode at 29.3 fps.

Scanning electron microscopy. Tissue samples were fixed by perfusion with 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), sliced into 2 mm pieces, and immersed in the same fixation buffer. After washing, the specimens ware postfixed with 1% osmium tetroxide in 0.1 M phosphate buffer (pH 7.4) containing 1% potassium ferrocyanide and conductive stained with 1% tannic acid solution and 1% osmium tetroxide solution. The specimens were dehydrated in a graded series of ethanol, substituted with 100% ethanol, dried by the critical-point drying method, and coated with osmium tetroxide by the vacuum deposition method. Electron micrographs were captured with an S-4800 field emission scanning electron microscope (Hitachi High-Technologies Corp.).

Mucociliary transport assay. Mucociliary transport was analyzed using fluorescent beads (Fluoresbrite, 0.5 μm; Polysciences) as described previously (45). Briefly, after removing unnecessary surrounding tissues, an isolated trachea was opened from the dorsal side. With luminal surface facing upward, the tissue was placed in a rectangular space surrounded by 0.4 mm2 high vinyl ridges on a glass slide. The tissue was secured using medical adhesive (Daiichi Sankyo) as described above. Immediately after applying a drop of prewarmed fluorescent beads (1:500 dilution in DMEM/F12; 37°C) onto the luminal surface, a cover glass was placed on top for microscopy observation. Time-lapse images were recorded at 136 ms intervals using the STELLARIS 5 WLL confocal microscope with ×20 objective lens. Beads were tracked and analyzed using the TrackMate plug-in for Fiji (72–74). Traveling linearity and directional uniformity were calculated by analyzing 10 individual beads per record. Two recordings were performed for each trachea. In each record, 300–600 beads were detected within the area (258.33 × 64.20 μm), and 10 beads with recording of more than 6 continuous frames (0.816 seconds) were randomly selected for calculations. For individual beads, displacement α was defined as the direct distance from initial position (frame t = 0, coordinate: xt = 0, yt = 0) to the endpoint (frame t = 6, coordinate: xt = 6, yt = 6), and total track distance β was defined as the total traveling distance within the same time frame (from t = 0 to t = 6). Traveling linearity was calculated by dividing α by β.

(Equation 1)

Traveling linearity = α/β

For calculating directional uniformity, the individual trajectory vector ϕ was determined as displacement from frame 0 to frame 6, and the average trajectory vector Φ was defined as the average of 10 individual trajectory vectors within the same record. Directional uniformity was calculated by dividing the length of Φ (displacement as a group) by the average size of 10 individual ϕ vectors (individual displacement).

(Equation 2)

Directional uniformity = Φ/ ϕaverage

In vivo imaging. Mice were anesthetized using an anesthetic combination (medetomidine, midazolam, and butorphanol), and fluorescent carboxylate–modified microspheres with a diameter of 0.2 μm (1:100 dilution; Invitrogen; F8807) were administered intranasally. At 1, 6, and 24 hours after administration, in vivo imaging was performed using IVIS Lumina III version 4.7 (PerkinsElmer). Settings for imaging were as follows: lamp level, low; excitation, 680 nm; emission, 790 nm; epifluorescence; binning, medium; field of view, A; F-stop = 2; and acquisition time = 1 s. Total flux (photons/s) was measured within equally sized rectangular regions of interest using Living Image software (PerkinElmer).

Flow cytometry. After removal of blood cells by cardiac perfusion with 5 mL of ice-cold PBS via the right ventricle and instillation of dissociation buffer (HBSS containing 0.2 U/mL Liberase and 20 μg/mL DNase I, both from Roche) via a trachea catheter, lung lobes were isolated. Minced lung tissues were digested in dissociation buffer at 37°C for 30 minutes. The reaction was stopped by adding an excess volume of buffer containing 10% (v/v) FBS, and the mixtures were filtered through a 70 μm filter to remove undigested parts and debris. After blocking with anti-mouse CD16/CD32 (BioLegend: 93), leukocytes were labeled using combinations of fluorophore-conjugated antibodies: anti-CD45 (BioLegend; 30-F11), anti-CD11b (BioLegend; M1/70), anti-Ly6G (BioLegend; 1A8), Ly6C (BioLegend; HK1.4), anti-CD11c (BioLegend; N418), anti-SiglecF (BD Biosciences; E50-2440), anti-CD4 (BioLegend; GK1.5), and anti-CD8a (BioLegend; 53-6.7).

For labeling epithelial populations, digested tissues were cleaned using a Debris Removal kit (Miltenyi) and then blocked with anti-mouse CD16/CD32. Cell surface markers were labeled using anti-CD45, anti-CD31(BioLegend; 390), and anti-CD326 (BioLegend; G8.8). After labeling dead cells using Zombie Aqua (BioLegend), cells were fixed and permeabilized using the Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences). Intracellular airway epithelial marker proteins were labeled using anti-CYP2F2 (Santa Cruz Biotechnology; F-9) and anti-TUBA (Cell Signaling Technology; D20G3). After staining, cell suspensions were filtered through a 70 μm strainer. Flow cytometry was performed on an LSRFortessa system (BD Biosciences).

Real-time RT-PCR. Total RNA extraction was performed using an RNeasy Mini Kit (Qiagen) for trachea tissues and Trizol Reagent (Thermo Fisher Scientific) for the other tissues. cDNA was prepared using PrimeScript RT Master Mix (Takara Bio Inc.) containing random hexamer and oligo-dT primer. Alternatively, SMART MMLV Reverse Transcriptase (Clontech) with oligo-dT primer was used. Real-time PCR was performed using KAPA SYBR Fast (KAPA Biosystems) and the CFX Connect Real-Time System (Bio-Rad). The primers used for the mouse genes are as follows: Gapdh, 5′-GTTGTCTCCTGCGACTTCAAC-3′ and 5′-CCAGGGTTTCTTACTCCTTGG-3′; Rpl13a, 5′-GGCTGAAGCCTACCAGAAAGT-3′ and 5′-TCTTTTCTGCCTGTTTCCGTA-3′; β-actin, 5′-TGTTACCAACTGGGACGACA-3′ and 5′-GGGGTGTTGAAGGTCTCAAA-3′; Aldh1a1, 5′-GGCTTAATCCAACAGATTCATTCACCT-3′ and 5′-ACACCTGGGGAACAGAGCA-3′; Aldh1a2, 5′-CACAGGAGAGCAAGTGTGTGA-3′ and 5′-TAGTTGCAAGAGTTGCCCTGT-3′; Aldh1a3, 5′-AAACCCACGGTCTTCTCAGAT-3′ and 5′-CTTTGTCCAGGTTTTTGGTGA-3′; Aldh1a7, 5′-AGCTTAATCTGGCAGAATCAGAGTCT-3′ and 5′-TCAGAGGAATAACCCCGAGGAAT-3′; Aldh2, 5′-TTTATGAACAGTGGCCAGACC -3′ and 5′-TCGTTGATGATCCTCCCATAG-3′; Aldh3a1, 5′-GATCCTAACTCCAAGGTGATGC-3′ and 5′-ACCCGTTTGATGAGCTTATTGT-3′; Aldh3a2, 5′-GATCCTAACTCCAAGGTGATGC-3′ and 5′-ACCCGTTTGATGAGCTTATTGT-3′; Aldh3b1, 5′-GAAGCATTTCAAGCGACTCC-3′ and 5′-CAGGCTTCTCACAGTCACCA-3′; Aldh3b2, 5′-GCAACGATGGCTTCCTCTAC-3′ and 5′-AGCCTATGGCCCAGCTTATC-3′; Aldh3b3, 5′-AGCGCTTTATGCCTATTCCA-3′ and 5′-ACGGAGGCCATTAAGCTTCT-3′; Aldh4a1, 5′-TGGAAGCACACCTCCTCTCT-3′ and 5′-AAGGGCGACAACTGGTACTG-3′; Aldh5a1, 5′-TTACTGGCTCAACAGCAACG-3′ and 5′-TGTTTGAGCAAACGCAAGTC-3′; Aldh6a1, 5′-ATCCTCGTAGGGGAGGCTAA-3′ and 5′-TTAATTCTTCGCCCATCCAG-3′; Aldh7a1, 5′-GGAAGGAATAGGCGAGGTTC-3′ and 5′-AGTGATGATTCCCACCAAGC-3′; Aldh8a1, 5′-GCAAAGCACATTTGGAGAAAG-3′ and 5′-AGCGGGACTCATCCTTAATGT-3′; Aldh9a1, 5′-GGCCAGTTTCTGTGTCATCAT-3′ and 5′-CCCTTCACAGCATTCTCCATA-3′; Aldh16a1, 5′-CTTCTCCTTTCCGCACAGTC-3′ and 5′-CCATGAGCATTGATCCACAC-3′; Aldh18a1, 5′-ATGGTTACCGCTTTGGACTG-3′ and 5′-CTTCCATGCTCGGAGAAGTC-3′; Txnrd1, 5′-CAGTTCGTCCCAACGAAAAT-3′ and 5′-GCACATTGGTCTGCTCTTCA-3′; Hmox1, 5′-TGCTCGAATGAACACTCTGG-3′ and 5′-TCTCTGCAGGGGCAGTATCT-3′; Foxj1, 5′-CAGACCCCACCTGGCAGA-3′ and 5′-TGAAGGCCCCACTGAGCA-3′; Muc5ac: 5′-AGTTGCCAGTGTCTACAGCC-3′ and 5′-CTGGAAGTCATCAGCCTGCA-3′; Scgb1a1, 5′-ACAATCACTGTGGTCATGCTGT-3′ and 5′-AGGGTATCCACCAGTCTCTTCA-3′; Trp63: 5′-GTCAGCCACCTGGACGTATT-3′ and 5′-CTCATTGAACTCACGGCTCA-3′; Krt13: 5′-AACAAGGCTGGAACAGGAGA-3′ and 5′-CACATCCTGCAGTCCTCTCA-3′; Cxcl1, 5′-GCTGGGATTCACCTCAAGAA-3′ and 5′-TCTCCGTTACTTGGGGACAC-3′; Ccl2, 5′-AGGTCCCTGTCATGCTTCTG-3′ and 5′-TCTGGACCCATTCCTTCTTG-3′; Csf2, 5′-GGCCTTGGAAGCATGTAGAG-3′ and 5′-CCGTAGACCCTGCTCGAATA-3′; Il1b, 5′-TGTGGCAGCTACCTGTGTCT-3′ and 5′-TGTTCATCTCGGAGCCTGTA-3′; Il6, 5′-AAGCCAGAGTCCTTCAGAGAGATA-3′ and 5′-CAGGGGTGGTTATTGCATCT-3′; Il17a, 5′-TCCAGAAGGCCCTCAGACTA-3′ and 5′-AGCATCTTCTCGACCCTGAA-3′.

RNA-seq. Lung tissues were isolated from 2 mice per group (Aldh1a1+/+ and Aldh1a1–/–) and digested as described above for RNA extraction and purification. RNA-seq sample preparation was performed using an RNeasy Mini Kit (Qiagen). Sequencing libraries were constructed through library preparation following the recommended protocol for the TruSeq stranded mRNA Library Prep kit (Illumina). Fragment size of the libraries was confirmed with a LabChip DNA High Sensitivity Reagent Kit (PerkinElmer). Libraries were sequenced on a NovaSeq 600 (Illumina) in the 101-base single-read mode. Among the known RA-responsive 532 genes (75, 76), 153 genes with more than 100 reads were selected, as shown in Supplemental Table 1, and plotted against all genes (log2 [fold change] [x axis] against average fragments per kilobase million [y axis]). Pathway analysis was performed using integrated differential expression and pathway analysis (iDEP2.0, GAGE) using the Gene Ontology database for biological processes and TF.Target.RegNetwork for RAR target genes, respectively.

Open-source data exploration. RNA-seq data for human ALDH family genes were downloaded via the ENCODE Expression Atlas on July 10, 2023 (77, 78). Mouse and human lung single-cell data were sourced from the LungMAP Consortium (U01HL122642) (LungMAP IDs: LMEX0000004396 for human and LMEX0000004397 for mouse) using the human or mouse ShinyCell browser and downloaded from www.lungmap.net (LungMAP Data Coordinating Center; 1U01HL122638) on September 2, 2023. ALDH1A1 expression levels in ciliated cells from healthy donors and patients were analyzed using a publicly available dataset (https://cellxgene.cziscience.com/collections/6f6d381a-7701-4781-935c-db10d30de293) using BBrowser X (BioTuring Inc., study 846a6f259f9d4d85b07789b03eb4e4aa) (79).

Bacterial pneumonia. For the bacterial pneumonia model, S. pneumoniae strain TIGR4 was used (80). Male C57BL/6 mice aged 10–13 weeks, pre-exposed to either naphthalene or DEPs, were used for bacterial pneumonia experiments. Mice were anesthetized by intraperitoneal injection of anesthetic mixtures (medetomidine, midazolam, and butorphanol) before infection and intranasally instilled with 1–2 × 108 CFU of TIGR4 in 20 μL PBS. Bacterial culture, instillation, and CFU determination were performed as described previously (80).

Western blot. For ALDH1A1 detection, lung lysates were prepared using SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 5% 2-mercaptoethanol, and 0.001% bromophenol blue), separated by electrophoresis on 4%–15% polyacrylamide gel, and transferred onto a PVDF membrane. The membrane was incubated with antibodies against ALDH1A1 or β-actin in EveryBlot (Bio-Rad), followed by incubation with antibodies against rabbit IgG conjugated with HRP. For TUBA detection, lung lysates were prepared using RIPA buffer, and protein concentrations were determined by the bicinchoninic acid method for loading normalization. Lysates were heat denatured in SDS sample buffer, separated on 10% polyacrylamide, and transferred onto PVDF membrane. The membranes were incubated with anti-TUBA (Cell Signaling Technology; D20G3, 1:1,000) followed by HRP-conjugated anti-rabbit IgG, or with HRP-conjugated anti–β-actin (13E5, 1:1,000) in Can Get Signal reagent. The peroxidase activity was detected by ImmunoStar Zeta (FUJIFILM Wako).

Statistics. All experiments were conducted at least twice, and statistical analyses (1-way ANOVA followed by post hoc tests or 2-tailedStudent’s t test) were performed using GraphPad Prism 9. Data points and mean values are presented unless otherwise noted. Mouse survival curves were compared with the log-rank test. P < 0.05 was considered statistically significant. Samples sizes are detailed in figure legends.

Study approval. All procedures were approved by the IACUC of The University of Osaka.

Data availability. Sequencing data are available from the National Center for Biotechnology Information Gene Expression Omnibus (GSE267105, GSE287365, and GSE296445). All data are available in the main text or the supplemental materials.