2.1 Alveolar Epithelial Deficiency in Atg5 Increased Susceptibility to P. aeruginosa Infection

By analyzing our lung bulk RNA-seq dataset, comprising three wild-type (WT) mice and four P. aeruginosa-infected mice at 24 h postinfection, we uncovered a pronounced enrichment of autophagy-related pathways upon P. aeruginosa infection, as determined by single-sample gene set enrichment analysis (ssGSEA; Figure 1A). To delve into the enrichment of autophagy formation signature across distinct cell types, we employed our single nucleus RNA sequencing (SnRNA-seq) dataset derived from two WT mice and two P. aeruginosa-infected mice at 24 h postinfection (Figures 1B and S1). AUCell analysis revealed an escalated autophagy formation signature enrichment in P. aeruginosa-infected lungs compared to control lungs, with type II alveolar epithelial cells (AEC2) exhibiting the highest enrichment of this signature (Figure 1C–E). Through protein–protein interaction (PPI) analysis with the bulk RNA-seq dataset (Figure S2), we found that ATG5, a pivotal protein involved in the autophagy formation, plays a key role in the interaction network within P. aeruginosa-infected lungs. Therefore, we examined the role of ATG5 in AEC2 during P. aeruginosa infection, utilizing a mouse pneumonia model. PAO1 (1 × 10⁷ CFU/mouse) was intranasally delivered into each of AEC2-specific Atg5 conditional knockout (Atg5^ΔAEC2) mice and their littermate WT counterparts. Upon PAO1 infection, deficiency of Atg5 in AEC2 reduced the survival rate, resulting in a mortality rate of 45% for Atg5^ΔAEC2 mice, while WT mice exhibited a mortality rate of 13% after 72 h (Figure 1F). Moreover, Atg5^ΔAEC2 mice exhibited impaired bacterial clearance in both lungs and bronchoalveolar lavage fluid (BALF) compared to WT mice (Figure 1G,H). Notably, there was a marked increase in bacterial burdens in the blood, spleen, and liver of Atg5^ΔAEC2 mice, indicating enhanced dissemination of P. aeruginosa (Figure 1I–K). Collectively, these findings underscore the indispensable role of ATG5 in epithelial cells for resistance to P. aeruginosa infection.

2.2 Alveolar Epithelial Deficiency in Atg5 Exacerbated the Inflammatory Response and Activation of the AKT/PI3K/NF-κB Axis in Lungs During P. aeruginosa Infection

To gain deeper insight into the lethality associated with P. aeruginosa infection in Atg5^ΔAEC2 mice, lung samples were stained with hematoxylin and eosin staining (H&E) and scored in a blinded manner. The assessment revealed more severe lung injuries and heightened infiltration of inflammatory cells in Atg5^ΔAEC2 mice in comparison to their WT counterparts post P. aeruginosa infection (Figure 2A,B). Confirmation of increased total cells in BALF, notably neutrophils, further emphasized the heightened inflammatory state in Atg5^ΔAEC2 mice following exposure to PAO1 (Figure 2C,D). Given the pivotal role of cytokines in acute lung injury, we measured the levels of proinflammatory factors TNF-α and IL-1β, which were found to be significantly elevated in BALF from Atg5^ΔAEC2 mice compared to WT mice postinfection (Figure 2E). Western blotting validation underscored that the absence of Atg5 and autophagy in AEC2 aggravated the production of TNF-α and IL-1β in the lungs during P. aeruginosa infection (Figures 2F,G and S3). Furthermore, immunoblotting revealed heightened activation of AKT, PI3K, and NF-κB in the lungs of Atg5^ΔAEC2 mice compared to WT mice after PAO1 infection (Figures 2H and S3). Unexpectedly, while overall STAT3, another key proinflammatory pathway, increased, there was no evident elevation in phosphorylated-STAT3 during P. aeruginosa infection in Atg5^ΔAEC2 mice (Figures 2H and S3). Similarly, our airway epithelial cell delivery experiments revealed that lungs treated with Atg5-knockdown epithelial cells exhibited more severe lung injury and heightened infiltration of inflammatory cells following P. aeruginosa infection compared to those treated with control siRNA-transfected epithelial cells. Notably, administration of the AKT inhibitor AKTi ½ or the NF-κB inhibitor JSH-23 significantly alleviated the exacerbated lung injury and inflammatory cell infiltration induced by epithelial Atg5 knockdown (Figure S3). These data suggest that epithelial ATG5 protects against lung injury upon infection, potentially by controlling the inflammatory response and regulating the AKT/PI3K/NF-κB axis.

2.3 Loss of Atg5 Intensified the Dysregulation of Tight Junctions in Alveolar Epithelial Cells During P. aeruginosa Infection

The lung epithelial barrier serves as a crucial defense against bacterial infiltration, preventing bacterial entry into the circulatory system and thereby mitigating bacterial dissemination [8]. To further understand bacterial dissemination in Atg5^ΔAEC2 mice, we conducted an assessment of the tight junction protein ZO-1 through immunofluorescent staining (Figure S4A). Remarkably, ZO-1 expression exhibited a marked decrease in lung samples of Atg5^ΔAEC2 mice compared to their WT counterparts in response to P. aeruginosa infection, a trend that was consistently confirmed by western blotting (Figure S4B,C). Correspondingly, the targeted knockdown of Atg5 using siRNA resulted in a more pronounced reduction in ZO-1 levels in the murine AEC2 (MLE-12 cell line) following stimulation with P. aeruginosa (Figure S4D–F). These data underscore the critical role of ATG5 in the maintenance of epithelial tight junctions during P. aeruginosa infection

2.4 Alveolar Epithelial Atg5 Deficiency Intensified Inflammasome Activation and Pyroptosis in Lungs Upon P. aeruginosa Infection

Excessive inflammasome activation is known to contribute to exacerbated pathological damage [9]. We probed inflammasome expression levels through western blotting, revealing a substantial increase in NLRP3 and NLRC4 expression in Atg5^ΔAEC2 mice compared to WT mice upon P. aeruginosa infection (Figure 3A–F). Additionally, the levels of procaspase-11 and cleaved caspase-11 were elevated in mice lacking alveolar epithelial Atg5 compared to control mice following infection (Figure 3A–F). These results underscore that Atg5 deficiency intensifies the activation of pulmonary NLRP3, NLRC4, and caspase-11 inflammasomes induced by P. aeruginosa. Given the established link between inflammasome activation and a form of inflammatory cell death known as pyroptosis [10], we investigated the expression of pyroptosis-related proteins. Our analysis revealed an increase in procaspase-1 and cleaved caspase-1 in lung tissues of Atg5^ΔAEC2 mice after P. aeruginosa infection (Figure 3A–F). The activation of caspase-1 results in the cleavage of gasdermin D (GSDMD), releasing the N-terminal GSDMD (NT-GSDMD) and inducing pyroptotic cell death [10]. Notably, the concurrent enhancement of NT-GSDMD levels was observed alongside increased caspase-1 cleavage in Atg5 conditional knockout mice during infection (Figure 3A–F). Immunofluorescence further illustrated a substantial increase in cleaved caspase-1 and NT-GSDMD specks in Atg5^ΔAEC2 mice compared to WT mice (Figure 3G). Moreover, mice with AEC2 Atg5 deletion exhibited an augmentation in cleaved IL-1β compared to WT mice in response to P. aeruginosa challenge (Figure 2F). In summary, these findings establish the protective role of epithelial ATG5 against P. aeruginosa-induced injury by finely modulating inflammasome activation and pyroptosis in the lungs.

2.5 Loss of Atg5 Intensified Activation of Noncanonical Caspase-11 Inflammasome and GSDMD in Alveolar Epithelial Cells Upon P. aeruginosa Infection

To further explore the mechanisms of Atg5 in regulating inflammasomes and pyroptosis in alveolar epithelial cells during P. aeruginosa infection, we employed siRNA to interfere with Atg5 expression in MLE-12 cells. Unexpectedly, Atg5 loss did not augment the activation of NLRP3 and NLRC4 in AEC2 following P. aeruginosa exposure (Figure 4A–G). Moreover, there was no increase observed in the levels of cleaved caspase-1 and cleaved IL-1β in Atg5-silenced epithelial cells compared to control cells in response to P. aeruginosa infection (Figures 4A–G and S5). In contrast, an increase was noted in cleaved caspase-11 in Atg5 siRNA-transfected cells compared to control siRNA-transfected cells following exposure to P. aeruginosa (Figure 4A–G). While western blot showed unchanged procaspase-11 levels, immunofluorescence demonstrated an increase in total caspase-11 expression, possibly due to its ability to encompass both procaspase-11 and its cleaved form (Figure 4H). Caspase-11 orchestrates GSDMD activation through noncanonical inflammasome signaling. Consistently, this intensified activation of caspase-11 coincided with an increase in NT-GSDMD levels (Figure 4A–G). Immunofluorescence further illustrated augmented expression of NT-GSDMD in cells transfected with Atg5 siRNA compared to those transfected with control siRNA upon infection (Figure 4H). Taken together, these data indicate that ATG5 primarily implicates in the regulation of noncanonical caspase-11 inflammasome-mediated GSDMD activation in alveolar epithelial cells during P. aeruginosa infection.

2.6 Loss of Atg5 Impedes Mitophagy and Aggravated Mitochondrial Damage in Epithelial Cells During P. aeruginosa Infection

Mitochondria represent a major source of reactive oxygen species (ROS), which can cause cellular damage if not effectively managed [11]. To counteract this, mitophagy—a specialized form of autophagy—plays a crucial role in mitigating mitochondrial ROS (mtROS) by facilitating the removal of damaged mitochondria [12, 13]. To investigate the role of mitophagy in modulating inflammatory responses during P. aeruginosa infection, we employed two complementary approaches: pretreatment with the mitochondrial division/mitophagy inhibitor Mdivi-1 and knockdown of Atg5 in mouse lung tissues via siRNA transfection. Both interventions significantly reduced lung injury and inflammatory cell infiltration compared to controls following P. aeruginosa infection (Figure S6), highlighting the critical role of mitophagy in maintaining tissue homeostasis under infection-induced stress. Moreover, in Atg5-silenced MLE-12 cells, LC3B puncta, a marker associated with autophagosome formation, displayed decreased accumulation around mitochondria in response to P. aeruginosa, indicating a compromised mitophagic process compared to control cells (Figure 5A). Surprisingly, the knockdown of Atg5 in epithelial cells led to a decrease in the expression of PINK1, Parkin, and PHB2 upon P. aeruginosa treatment, implying that the loss of Atg5 not only disrupts mitophagy by impeding autophagosomal formation but also inhibits both ubiquitin-dependent and receptor-mediated mitophagy pathways during infection (Figure 5B,C). Consistent with this, silencing Atg5 in MLE-12 cells resulted in aggregated mitochondrial damage, as evidenced by a substantial decrease in mitochondrial membrane potential (Figure 5D) and a significant increase in the production of both total ROS and mtROS during PAO1 infection (Figure 5E). Additionally, considering that mtDNA is released under various pathological conditions, such as oxidative stress and mitochondrial dysfunction [14], we measured cytosolic mtDNA, which was significantly elevated in Atg5-knockdown cells compared to control cells postinfection (Figure 5F). These findings collectively underscore the essential role of ATG5 in mitophagy and the maintenance of mitochondrial homeostasis in alveolar epithelial cells during P. aeruginosa infection.

2.7 Loss of Atg5 in Epithelial Cells Facilitated Extracellular mtDNA Release During P. aeruginosa Infection, Subsequently Triggering cGAS–STING–NLRP3 Axis Activation in Macrophages

Recent work highlighted the role of GSDMD in promoting the release of mtDNA from the cytosol into the extracellular space [15]. We speculated that silencing Atg5 would lead to the accumulation of mtDNA in the extracellular space. Quantitative PCR analysis revealed a significant increase in mtDNA levels in the culture medium supernatant of MLE-12 cells with Atg5 knockdown following P. aeruginosa infection, compared to control cells (Figure 6B). The escape of mtDNA into the cytosol triggers the activation of the cGAS–STING–NLRP3 axis [16]. The extracellular release of mtDNA may subsequently activate immune cells, offering a plausible explanation for the observed discrepancy where alveolar epithelial Atg5 deficiency intensified NLRP3 inflammasome activation in lungs, but loss of Atg5 in alveolar epithelial cells did not. To further investigate this phenomenon, we exposed RAW264.7 murine macrophages to the 0.22 um filtered medium supernatant of P. aeruginosa-infected MLE-12 cells. Western blotting and immunofluorescence results demonstrated a significant increase in the expression of cGAS, STING, and NLRP3 in macrophages treated with supernatant from Atg5-silenced MLE-12 cells compared to control cells (Figure 6C–E). Consistent with these findings, our airway epithelial cell delivery experiments showed that lungs receiving Atg5-knockdown epithelial cells exhibited increased expression of cGAS, STING, and NLRP3 in macrophages following P. aeruginosa infection, compared to lungs treated with control siRNA-transfected epithelial cells (Figure S7). To validate these findings, we employed the mitochondria-targeted peptide SS-31, known for neutralizing ROS and inhibiting mtDNA release [17]. Indeed, SS-31 markedly decreased mtDNA levels in both the cytoplasm and culture medium supernatant of Atg5-knockdown epithelial cells during P. aeruginosa infection (Figure 6A,B). Furthermore, the heightened expression of cGAS, STING, and NLRP3 in macrophages treated with supernatant from P. aeruginosa-infected Atg5-silenced MLE-12 cells was significantly mitigated by SS-31 treatment of the epithelial cells (Figure 6C–E). In addition, we used RU.521, a selective inhibitor targeting the catalytic site of cGAS [18]. This inhibitor significantly suppressed downstream STING and NLRP3 expression in macrophages treated with supernatant from Atg5-silenced MLE-12 cells infected with P. aeruginosa (Figure 6F–H). This observation suggests that inhibiting cGAS activity, which impairs cGAMP production, indirectly affects STING expression. This suggests a possible feedback mechanism wherein the downstream signaling components of the cGAS-STING pathway are modulated by upstream inhibition. Altogether, these results underscore the pivotal role of ATG5 in controlling extracellular mtDNA accumulation from alveolar epithelial cells and subsequent cGAS–STING–NLRP3 signaling activation in macrophages during P. aeruginosa infection.

4.1 Mice

Six to eight weeks old C57BL/6 mice were purchased from the Jackson Laboratory. The Atg5^ΔAEC2 mice were generated on a C57BL/6J background by crossing Atg5^f/f mice with Sftpc^CreER mice to produce Sftpc^CreER;Atg5^f/f mice [39]. To induce Atg5 knockout specifically in AT2 cells, tamoxifen was administered intraperitoneally at a dose of 50 mg/kg once daily for 5 consecutive days. The mice were bred and housed at the University of North Dakota, with all animal procedures approved by the University of North Dakota Institutional Animal Care and Use Committee (2210-0045) and conducted in compliance with animal care and institutional protocols.

4.2 Cell Culture, Transfection, and Treatment

Murine epithelial cells (MLE-12) and macrophage cells (RAW264.7) were acquired from the American Type Culture Collection. RAW264.7 cells were cultured in high glucose Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and 100 U/mL penicillin–streptomycin antibiotics, maintained in a 37°C incubator with 5% CO₂. MLE-12 cells were cultured in Dulbecco's modified Eagle medium/nutrient mixture F-12 medium. Transfection of MLE-12 cells with small interfering RNAs was carried out using RNAiMAX (Invitrogen), following the manufacturer's instructions. The Atg5 siRNA was procured from Santa Cruz Biotechnology.

4.3 Bacteria Preparation and Infection Experiments

P. aeruginosa strain PAO1 was graciously provided by Dr. S. Lory (Harvard Medical School, Boston, MA, USA). The bacteria were cultivated in lysogeny broth (LB) medium at a constant temperature of 37°C with shaking at 220 rpm until reaching an OD₆₀₀ of 0.6–0.8. For in vivo experiments, mice were anesthetized with ketamine (80 mg/kg) and xylazine (10 mg/kg), then intranasally instilled with 1 × 10⁷ clonal-forming units (CFUs) of PAO1 in 30 µL phosphate-buffered saline (PBS). The health status of mice was closely monitored, and euthanasia was performed upon observing moribund conditions to generate survival curves using Kaplan–Meier methods and to facilitate additional assessments.

For airway epithelial cell delivery experiments, we utilized a previously established intratracheal delivery model in mice, which achieves a cell-retention efficiency exceeding 10% [40]. C57BL/6J mice were administered 20 µL of 2% polidocanol (PDOC; Sigma-Aldrich, St. Louis, MO, USA) in PBS intratracheally to prepare the lungs. After 24 h, MLE-12 cells transfected with Atg5 siRNA or siNC (1 × 10⁷ cells per mouse) were delivered via intratracheal injection. The next day, mice were infected with P. aeruginosa (PAO1) and lung tissues were harvested 24 h postinfection for analysis. For inhibitor experiments, mice were pretreated with the AKT inhibitor AKTi ½ (20 mg/kg) or the NF-κB inhibitor JSH-23 (40 mg/kg) via intraperitoneal injection for 2 consecutive days prior to PAO1 infection. Lung tissues were collected 24 h after infection for further analysis.

For Atg5 knockdown in vivo experiments, 20 µg of Atg5 siRNA or siNC was mixed with 7.5 µL of Entranster-in vivo transfection reagent (Engreen Biosystem) and 7.5 µL of saline to a total volume of 30 µL. The mixture was administered intranasally to C57BL/6J mice. After 24 h, mice were infected with PAO1 (1 × 10⁷ CFU). Lung tissues were harvested 24 h postinfection for analysis. For mitophagy inhibition experiments, mice were pretreated with the mitochondrial division/mitophagy inhibitor Mdivi-1 (50 mg/kg) via gavage for 5 consecutive days. Following pretreatment, mice were infected with PAO1 (1 × 10⁷ CFU). Lung tissues were collected 24 h postinfection for further analysis.

In the epithelial infection experiments, MLE-12 cells were infected with PAO1 at an MOI of 20 for 2 h after cell fusion reached 60%–80%. To assess the impact of SS-31, MLE-12 cells were treated with 200 nM SS-31 along with PAO1. Following the infection, supernatant was collected, subjected to centrifugation, and bacteria were removed using a 0.22 µm filter. The resulting supernatant was then utilized to culture RAW264.7 cells for 24 h. For RU.521, RAW264.7 cells were pretreated with 20 uM RU.521 for 2 h, followed by culturing with the supernatant of PAO1-treated MLE-12 cell medium for 24 h.

4.4 Bacterial Burden Assay

Lung, spleen, and liver homogenates, BALF, and blood collected from mice were appropriately diluted in PBS to various concentrations. Subsequently, the diluted samples were evenly spread on LB agar dishes. These dishes were incubated overnight in a 37°C environment, after which bacterial colonies were counted for analysis.

4.5 Histological Analysis

Lung tissues were fixed in 4% paraformaldehyde (PFA), followed by a 48-h dehydration in 75% alcohol at 4°C. Subsequently, the tissues were embedded in paraffin using a routine histological procedure. H&E staining, following the standard staining protocol, was then performed for histological examination. The lung injury was scored in a blinded manner according to previous method [41].

4.6 Mitochondrial Membrane Potential Assay

The assessment of mitochondrial membrane potential was conducted employing the JC-1 Mitochondrial Membrane Potential Assay Kit (Cayman Chemical), following the manufacturer's guidelines. In brief, MLE-12 cells, infected with PAO1 for 2 h as described earlier, were treated with 1 µg/mL of JC-1 and incubated for 30 min. Fluorescence intensity was quantified using a fluorescence plate reader at 560-nm excitation with a 595-nm emission filter and at 485-nm excitation with a 535-nm emission filter.

4.7 Measurement of Reactive Oxygen Species

Intracellular ROS were assessed using H₂DCF-DA staining solution (Invitrogen), following the manufacturer's stipulated guidelines. mtROS levels were quantified employing MitoSOX Red Mitochondrial Superoxide Indicators (Invitrogen). MLE-12 cells, subjected to the specified treatments, were incubated with 0.5 µM MitoSOX reagent for 30 min. Following two washes, fluorescence emissions were recorded using confocal laser scanning microscopy at 595 nm.

4.8 Immunoblotting

Lung tissues or whole-cell lysates were prepared in RIPA buffer (Thermo Fisher), enriched with a protease inhibitor cocktail and a phosphatase inhibitor cocktail (Invitrogen). Protein concentrations were quantified using the BCA Protein Assay Kit (Thermo Fisher). Subsequently, 20 µg of protein were electrophoretically separated on SDS–PAGE gels and transferred onto PVDF membranes. The membranes were then blocked in 5% skim milk in 1 × TBST for 1 h, followed by an incubation with primary antibodies (1:1000 dilution; ATG5, TNF-α, IL-1β, PI3K, p-PI3K, AKT, p-AKT, p-P65, CASP1, CASP11, NLRP3, GSDMD, PINK1, Parkin, and PHB2 from Cell Signaling Technology; NLRC4 from ABclonal; P65 from Invitrogen; STAT3 and p-STAT3 from Santa Cruz Biotechnology; ZO-1 from Abcam; cGAS and STING from Proteintech) at 4°C overnight. Subsequent incubation involved secondary antibodies conjugated to HRP (ABclonal, 1:10,000 dilution). Signal detection was achieved using an enhanced chemiluminescence detection kit (Cytiva).

4.9 Immunofluorescence and Confocal Microscopy

Lung sections (5 µm thickness) and cells cultured on coverslips underwent fixation in 4% PFA, permeabilization in 0.1% Triton X-100, and blocking in PBS supplemented with 10% goat serum. Primary antibodies (1:200 dilution; ZO-1 from Abcam; LC3B, NLRP3, PINK1, PHB2, cleaved GSDMD, and cleaved CASP1 from Cell Signaling Technology; cGAS and STING from Proteintech) were incubated in the blocking buffer at 4°C overnight. Subsequently, Goat anti-Mouse Alexa Fluor 488 Antibody, Goat anti-Mouse Alexa Fluor 594 Antibody, Goat anti-Rabbit Alexa Fluor 488 Antibody, and Goat anti-Rabbit Alexa Fluor 555 Antibody (Invitrogen; 1:1000 dilution) were employed to bind primary antibodies at room temperature for 1 h. Nuclei were counterstained with DAPI, and images were acquired using an Olympus FV3000 confocal microscope. For mitochondrial labeling, cells were incubated with MitoTracker Red (Invitrogen, M7512).

4.10 Enzyme-Linked Immunosorbent Assay

Paired antibodies (capture and detection) and standard recombinant mouse IL-1β and TNF-α were used to determine mouse cytokine concentrations in BALF according to manufacturer's instructions.

4.11 Measurement of mtDNA

MLE-12 cells underwent PBS washing and harvesting, with 10% reserved for whole-cell DNA extraction. The remaining cells were resuspended in prechilled mitochondrial extraction buffer. Subsequently, these cells were subjected to 20 passes through a 25-G syringe on ice and two-step centrifugation to facilitate separation of mitochondria and cytosolic fractions [42]. The purification of DNA from whole-cell extracts, cytosolic fractions, or culture media was carried out using the DNA Extraction Kit (TIANGEN) per manufacturer's instructions. mtDNA was quantified by qPCR using primers specific for the mitochondrial D-loop region (Forward: 5′-AATCTACCATCCTCCGTGAAACC-3′, Reverse: 5′-TCAGTTTAGCTACCCCCAAGTTTAA-3′), cytochrome c oxidase (Forward: 5′-GCCCCAGATATAGCATTCCC-3′, Reverse: 5′-GTTCATCCTGTTCCTGCTCC-3′) or a specific region of mtDNA that is not inserted into nuclear DNA (Forward: 5′-CTAGAAACCCCGAAACCAAA-3′, Reverse: 5′-CCAGCTATCACCAAGCTCGT-3′). Nuclear DNA encoding telomerase reverse transcriptase (Forward: 5′-CTAGCTCATGTGTCAAGACCCTCTT-3′, Reverse: 5′-CTAGCTCATGTGTCAAGACCCTCTT-3′), 18S ribosomal RNA (Forward: 5′-TAGAGGGACAAGTGGCGTTC-3′, Reverse: 5′-CGCTGAGCCAGTCAGTGT-3′), and β2 microglobulin (Forward: 5′-ATGGGAAGCCGAACATACTG-3′, Reverse: 5′-CAGTCTCAGTGGGGGTGAAT-3′) was used for normalization.

4.12 Transcriptome Analysis

The transcriptome library was prepared using the TruSeq RNA Sample Preparation Kit (Illumina, San Diego, CA, USA) with 5 µg of total RNA as input. PolyA selection was performed using oligo(dT) beads to isolate mRNA, which was subsequently fragmented. Complementary DNA (cDNA) synthesis was conducted using the SuperScript Double-Stranded cDNA Synthesis Kit with random hexamer primers (Invitrogen, CA). The resulting cDNA underwent end-repair, phosphorylation, and addition of an “A” base in accordance with Illumina's library preparation protocol. Library size selection was performed on a 2% low range ultra agarose gel, followed by amplification using Phusion DNA Polymerase with 15 PCR cycles. Paired-end sequencing was carried out on the Illumina HiSeq X Ten platform by Majorbio (Shanghai, China). Raw paired-end reads were trimmed and quality-checked using SeqPrep and Sickle with default parameters. Clean reads were then aligned to the reference genome in orientation mode using TopHat (version 2.0.0). ssGSEA was performed via GSVA package (1.42.0; method = “ssgsea,” kcdf = “Poisson”) [43]. PPI analysis was performed utilizing the STRING database. Autophagy-related signatures are available in Table S1.

SnRNA-seq was performed by SingulOmics Corporation (https://singulomics.com/). After quality control (nFeature_RNA > 400 and nFeature_RNA < 4000 and percent.mt < 5) and normalization with SCTransform(), integration was performed by FindIntegrationAnchors() and IntegrateData() function in Seurat (v4.1.1) [44, 45]. Principal components analysis and clustering were conducted with 30 PCs and a resolution of 0.4 in FindClusters(). Single-cell differential expression gene (DEGs) analyses were performed via MAST R package (1.20.0) [46]. Likelihood ratio tests between the full and reduced model formulas were used to identify DEGs (Benjamini and Hochberg method was performed for multiple testing corrections, FDR < 0.05). Autophagy formation gene signature is available in Table S1. Calculation and visualization of AUC scores were conducted via AUCell package (1.16.0) in R (4.1.0) [47, 48].

4.13 Statistical Analysis

All data were presented as mean ± SD from at least three independent experiments. Statistical analyses were done using Graphpad Prism (8.0.2). Multiple group comparisons were analyzed using one-way ANOVA or two-way ANOVA. Two group comparisons were analyzed by nonpaired t-test. Statistical significance was defined as a p value < 0.05.