2.1 Decreased Nrf2 expression in S. aureus-induced sepsis patients

To investigate the role of Nrf2 in S. aureus-induced sepsis (SA-sepsis), we first analyzed gene expression data from patients with septic shock caused by S. aureus infection.²⁶ The transcriptomes of peripheral whole blood from healthy volunteers (control, n = 22) and septic shock patients (patient, n = 102) were analyzed, and they showed separate clustering patterns (Figure S1A). Additionally, gene expression levels were found to be similar within each subject (Figure S1B). Nrf2 mRNA level was lower in septic shock patients than in controls (p = 0.012; Figure 1A).

To evaluate the potential clinical relevance of Nrf2 in SA-sepsis, we compared the levels of Nrf2 in serum samples from patients with sepsis and intensive care unit (ICU) patients without sepsis. We observed significantly lower Nrf2 levels in the SA-sepsis group (Figure 1B). The results revealed a negative correlation between serum Nrf2 levels and the Acute Physiology and Chronic Health Evaluation II (APACHE II) score (r = −0.42, p = 0.016; Figure 1C) as well as the Sequential Organ Failure Assessment (SOFA) score (r = −0.48, p = 0.005; Figure 1D). Additionally, the SA-sepsis group exhibited more severe organ damage compared with the control group. Meanwhile, serum level of procalcitonin (PCT)—a marker of bacterial infection²⁷—was much higher in the SA-sepsis group than in the control group (p = 0.002; Figure 1E and Table 1). The lower PaO₂/FiO₂ ratio in the SA-sepsis group (p = 0.002; Figure 1F) indicated that lung function was impaired in sepsis patients. Approximately 81.8% of patients in the SA-sepsis group had ALI, which was significantly higher than the proportion of control patients (p = 0.026; Table 1). Although there was no difference in 28-day survival between the 2 groups (p = 0.121; Figure 1G), the mortality rate of the SA-sepsis was higher than that of the control (58.3 vs. 31.8%; Table 1). Thus, the downregulation of Nrf2 in SA-sepsis leads to multiple organ dysfunction, with the expression level of Nrf2 being negatively correlated with sepsis severity.

The results revealed a downregulation of Nrf2 expression in patients with septic shock caused by S. aureus infection. Lower levels of Nrf2 were associated with more severe organ damage and showed a negative correlation with the severity scores of sepsis.

2.2 Differentially expressed genes and enrichment analysis

Whole-genome transcriptome analysis of blood samples from septic shock and control patients revealed 1023 differentially expressed genes (DEGs), including 607 upregulated and 416 downregulated genes (Figure S1C; p < 0.05). The top 50 DEGs are presented in a heatmap (Figure S2). Many of the top DEGs (Figure S2) are known to be involved in redox reactions (TXN and ATP5F1E), regulation of mitochondrial function (NDUFA1, ATP5F1E, ATP5MG, and ATP5MJ), and immune regulation (MCEMP1, IRAK3, FCER1G, LILRA5, IL1R2, and IL18RAP).

To identify molecular functions and signaling pathways that are significantly altered in sepsis, we performed gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analyses. KEGG pathways associated with all DEGs (Figure 2A) and upregulated genes (Figure 2B) suggested that OXPHOS (KO00190) plays an important role in sepsis. Consistent with these results, GO molecular function terms for genes upregulated in sepsis were enriched in OXPHOS-related functions such as electron transfer activity, cytochrome-c oxidase activity, and NAD⁺ nucleosidase activity (Figure 2C). These results indicate that altered OXPHOS and mitochondrial function play an important role in the development of sepsis.

Transcriptomic analysis revealed 1023 DEGs in patients with septic shock, which are potentially associated with oxidative-reduction reactions, mitochondrial function, and immune regulation (Figure S2). Enrichment analysis highlighted significant alterations in the OXPHOS pathway (Figures 2A and B), may suggesting the importance of disrupted mitochondrial function in the development of sepsis.

2.3 Occurrence of oxidative stress injury and mitochondrial dysfunction during SA-ALI in mice

To further examine the role of Nrf2 in SA-ALI, a SA-ALI model was established using WT C57BL/6 mice (Figure S3A). Hematoxylin and eosin (H&E) staining of lung tissue revealed that with increasing S. aureus concentration, edema, necrosis, and alveolar and interstitial inflammation were aggravated compared with the control group (Figure 3A). The alveolar structure was largely destroyed after being injected with the highest concentration of bacteria. Bronchoalveolar lavage fluid (BALF) cell counts revealed that inflammatory cells had infiltrated the lung tissue of SA-ALI mice (p < 0.05; Figure 3B); and real-time quantitative PCR (RT-qPCR) analysis showed that interleukin (IL)−6, IL-1α, and TNF-α were upregulated while IL-10 was downregulated in this group (Figure 3C). In line with the findings from clinical samples, Nrf2 mRNA expression was much lower in the SA-ALI group (p < 0.001; Figure 3D). Serum lactate dehydrogenase (LDH), malondialdehyde (MDA), and myeloperoxidase (MPO) activities were also significantly increased in the sepsis model (Figures 3E–G).

Mitochondrial function is thought to play a critical role in the development of ALI in sepsis.²⁸ PGC-1α, mitochondrial transcription factor A (Tfam) and nuclear respiratory factor 1 (NRF1) are implicated in mitochondrial biogenesis triggered by environmental stressors,²⁹ while mitofusin 1 (MFN1) and optic atrophy 1 (OPA1) are involved in mitochondrial fusion.³⁰ We found that mitochondrial DNA (mtDNA) copy number was decreased in the lung tissue of SA-ALI mice (Figure 3H). Additionally, it was accompanied by downregulation of PGC-1α, Tfam, NRF1, MFN1, and OPA1 (Figure 3I) mRNA, indicating that mitochondrial biogenesis and fusion were inhibited at the transcriptional level during S. aureus infection. Thus, sepsis caused by S. aureus leads to ALI and is accompanied by oxidative stress injury and mitochondrial dysfunction.

Using a SA-ALI mouse model, we observed increased lung damage, inflammation, and altered cytokine expression. Nrf2 mRNA expression was significantly lower in the SA-ALI group, and markers of oxidative stress and inflammation were elevated. Additionally, mitochondrial dysfunction markers and reduced mitochondrial biogenesis and fusion-related gene expression were detected, indicating the occurrence of oxidative stress injury and mitochondrial dysfunction during SA-ALI in mice.

2.4 Nrf2 deficiency aggravates SA-ALI with mitochondrial dysfunction

We examined whether loss of Nrf2 aggravates SA-ALI using Nrf2^−/− mice (Figures S3B and C). H&E staining of lung tissue indicated that inflammatory cell infiltration was more severe in Nrf2^−/− mice than in the WT (Figure 4A). In Nrf2^−/− mice with S. aureus infection, alveolar walls were thickened and contained numerous neutrophils; alveoli showed structural damage; and many inflammatory cells had infiltrated around the trachea (Figure 4A). The lung injury score showed that Nrf2^−/− mice had severe lung tissue damage (p = 0.0045; Figure 4B). The number of cells in BALF was increased in sepsis and Nrf2 deficiency increased inflammatory cell infiltration (Figure 4C). However, there was no significant difference in BALF protein concentration between Nrf2^−/− and WT mice (Figure 4D). SA-ALI mice also showed a marked increase in serum superoxide dismutase (SOD), which can convert superoxide anion radicals into two new form—oxygen and H₂O₂,³¹ but the antioxidant enzyme defense system was not activated in the absence of Nrf2 (Figure 4E). The change in serum LDH level was consistent with our previous results (Figure 3E), but Nrf2 knockout did not alter the level of LDH compared with the WT (Figure 4F). An examination of mtDNA copy number showed that loss of Nrf2 further inhibited mitochondrial function in sepsis (Figures 4G and H) as well as in healthy mice (Figure 4I).

In addition to Nrf2, another regulator of mitochondrial functions is PHB2. We evaluated PHB2 mRNA level and found that it was decreased when under the attack of S. aureus in vivo (Figure 4J). Consistent with our experimental results, PHB2 also showed a decreasing trend in septic shock patients (Figure S3D).

Collectively, the results showed that Nrf2 deficiency exacerbated lung tissue damage, inflammatory cell infiltration, and mitochondrial dysfunction in SA-ALI. Additionally, the mRNA level of PHB2, another regulator of mitochondrial functions, was decreased in both mice and septic shock patients.

2.5 Nrf2 inhibition exacerbates mitochondrial dysfunction in vitro

To explore the mechanism of SA-ALI in greater detail, we used A549 cells treated with lipoteichoic acid (LTA), a major component of the S. aureus cell wall. Western blot analysis showed that Nrf2 expression decreased with prolonged LTA (10 μg/mL) treatment, especially after 4 h, whereas PHB2 level was unchanged (Figure 5A). We also examined phospho-Nrf2 (p-Nrf2) expression with LTA time gradients and concentrations, and found that to be expressed more with prolonged LTA treatment and increased concentration (Figures 5A and B). However, the expression of total-Nrf2 (t-Nrf2) is opposite to that of p-Nrf2. The ratio of p-Nrf2/t-Nrf2 showed an increasing trend as the concentration of LTA and treatment time increased. When cells were treated with higher concentrations of LTA (20 μg/mL) for 4 h, Nrf2 and PHB2 levels were decreased, which also reduced voltage-dependent anion channel 1 (VDAC1) expression (Figure 5B). Cytochrome c has an antioxidant function in cells and is a marker of mitochondrial function.³² Treatment with a low concentration of LTA (2.5 μg/mL) resulted in the release of a large amount of cytochrome c—a reactive oxygen species scavenger and initiator of the apoptosis cascade—from mitochondria. However, intracellular cytochrome c was depleted at a high concentration (20 μg/mL) of LTA (Figure 5B).

To clarify the role of mitochondria in the effects exerted by Nrf2, the Nrf2 inhibitor ML385 was used in this study (Figure 5C). Neither ML385 (10 μM) nor LTA (20 μg/mL) affected cell vitality but cotreatment with both agents had an inhibitory effect, as determined with the cell vitality assay (Figure 5D). There was no significant difference in ATP production following treatment with ML385 and LTA, but there was an increasing trend overall (Figure 5E). MMP—which drives ATP synthesis and is a key parameter for evaluating mitochondrial function³³—was reduced by LTA and ML385 separately, as determined using the MMP probe JC-1 (Figure 5F). Moreover, the coapplication of LTA and ML385 synergistically enhanced this effect (Figure 5F). These results indicate that Nrf2 inhibition aggravates mitochondrial impairment in lung cells.

Our experiments demonstrated that as the treatment duration and concentration of LTA increased, the expression level of Nrf2 decreased in vitro. Additionally, the expression level of PHB2 decreased with an increase in the concentration of LTA treatment. Nrf2 inhibition, along with LTA treatment, led to reduced cell viability and MMP, indicating that Nrf2 inhibition exacerbates mitochondrial dysfunction in lung cells.

2.6 Nrf2 modulates PHB2 expression and intracellular distribution during SA-ALI

To examine the intracellular distribution of PHB2 in SA-ALI, we evaluated PHB2 protein levels in different cellular compartments. Our results show that the expression of total-PHB2 in lung tissues of WT mice decreased under S. aureus infection, and the absence of Nrf2 further reduced the expression of total-PHB2 (Figures 6A and B). Subsequently, we performed mitochondrial and cytoplasmic fractionation in lung tissue. Western blot analysis showed that mitochondrial PHB2 (mito-PHB2) level was decreased in Nrf2^−/− mice and was further reduced by SA-ALI (Figures 6C and D). In contrast, Nrf2 deficiency combined with SA-ALI led to an increase in cytoplasmic PHB2 (cyto-PHB2) level (Figures 6C and E). Immunofluorescence analysis revealed a decrease in PHB2 expression following LTA and ML385 treatment (Figure S4). Thus, during SA-ALI, both intracellular and mitochondrial levels of PHB2 were decreased, while the cytoplasmic level of PHB2 was significantly increased, and these effects was enhanced in the absence of Nrf2. Subsequently, we transfected Nrf2 overexpression (Nrf2-OE) plasmids into A549 cells in vitro. Nrf2-OE resulted in increased p-Nrf2 and PHB2 expression compared with the negative control (NC) group, which was further enhanced by LTA stimulation (Figure 6F). Mitochondrial and cytoplasmic fractionation of Nrf2-OE A549 cells showed elevated levels of both mito- and cyto-PHB2 compared with the NC group (Figure 6G). Moreover, the initially decreased expressions of mito- and cyto-PHB2 in the NC group were restored under LTA attack in the Nrf2-OE group (Figure 6G).

In short, the absence of Nrf2 in SA-ALI leads to a decrease in mito-PHB2 levels and an increase in cyto-PHB2. Conversely, overexpression of Nrf2 promotes the protection of PHB2 against LTA-induced damage. These findings indicate a positive regulatory role of Nrf2 in modulating PHB2.

2.7 Nrf2 binds to the PHB2 promoter and enhances gene transcription

To determine whether Nrf2 directly enhances PHB2 transcription, we performed a bioinformatic analysis and the dual-luciferase reporter assay. The 1023 DEGs identified between sepsis and nonsepsis patients (Figure S1C) were input into ChIP-X Enrichment Analysis Version 3 (ChEA3), a web-based transcription factor enrichment analysis tool that includes 6 primary reference gene set libraries from multiple sources.³⁴ Our research demonstrated the significance of Nrf2 as a crucial transcription factor in sepsis, and notably, Nrf2 exhibited enrichment when PHB2 was employed as input (Figure 7A). In the dual-luciferase reporter assay, the fluorescence intensity was higher in A549 cells cotransfected with PHB2 wt and Nrf2-OE luciferase plasmids compared with cells transfected with the former plasmid only (Figure 7B). These results suggest that Nrf2 directly induces PHB2 transcription by binding to its promoter.

To verify whether the direct regulation of PHB2 by Nrf2 indeed affects mitochondrial function, we introduced PHB2 siRNA to inhibit the expression of PHB2. We found that under LTA stimulation, Nrf2-OE in A549 cells significantly increased the mtDNA copy number. However, after using PHB2 siRNA, the replication process of mitochondrial genes was significantly decreased (Figures 7C–E). However, after using PHB2 siRNA, we observed that although the mtDNA copy number (NADH dehydrogenase subunit 1, ND-1; cytochrome c oxidase subunit I, COX-1; and complex IV) remained higher than the LTA stimulation group, it was significantly lower than the PHB2 siRNA negative control group (Figures 7C–E). This suggests that in addition to promoting mitochondrial biogenesis and improving mitochondrial function through the expression of downstream mitochondrial protective genes,¹⁷ Nrf2 can also enhance mitochondrial function by promoting the expression of PHB2. Additionally, we also measured the ATP concentration in A549 cells. Under the influence of LTA, the cellular ATP concentration significantly increased (Figure 7F). Nrf2-OE further elevated the ATP concentration, but under the interference of PHB2 siRNA, the ATP concentration decreased (Figure 7F). The observed results may be attributed to the following factors: under inflammatory attack, increased intracellular oxidative stress leads to an accumulation of free radicals, stimulating mitochondrial hyperfunction and enhancing OXPHOS efficiency. However, a decrease in ATP concentration was observed when PHB2 siRNA was used in Nrf2-OE cells (Figure 7F), indicating the loss of the protective role of PHB2 in mitochondrial function and cell survival. This finding indirectly suggests that the Nrf2/PHB2 pathway is involved in maintaining mitochondrial functional homeostasis.

In summary, Nrf2 binds to the PHB2 promoter and directly induces its transcription, as shown by bioinformatics analysis and dual-luciferase reporter assays. The direct regulation of Nrf2/PHB2 pathway affects mitochondrial function, as evidenced by the introduction of PHB2 siRNA to inhibit PHB2 expression in Nrf2-OE cell.

4.1 Patients and ethics

Serum samples were collected from patients diagnosed with SA-sepsis (n = 11) and nonsepsis control patients (n = 22) at the ICU of Wuxi Hospital of Traditional Chinese Medicine and Affiliated Hospital of Jiangnan University during the period from March 1, 2019 to August 31, 2020. All sepsis patients were enrolled consecutively. Whole blood and clinical data were collected from patients after diagnosis as sepsis within 24 h of their admission to the ICU. Sepsis was defined according to Sepsis-3.⁴⁹ We obtained the APACHE II score^{54, 55} and the SOFA score^{56, 57} of each patient—which revealed the severity of organ dysfunction during sepsis—by examining clinical data and laboratory indices (Table S1). The exclusion criteria were age <18 years; history of immune deficiency diseases or malignancy; inability or refusal to provide written, informed consent; being in a terminal state of sepsis; or having undergone/undergoing chemotherapy or hormone therapy within 4 weeks of ICU admission. Medical and nursing records were reviewed to collect patient demographics, ventilator parameters, and laboratory findings (Table 1).

All patients or their families expressed willingness to participate in the study through provision of signed, informed consent. Experiments involving human subjects were approved by the Ethics Committee of the Affiliated Hospital of Jiangnan University, Wuxi, China (ethics no. LS2020038).

Peripheral venous blood samples were collected from patients using a vacuum blood collection tube without anticoagulant. After they had been allowed to stand for 30 min, the samples were briefly centrifuged and the supernatant was aspirated, followed by centrifugation (6800 g, 10 min). The serum was stored at −80°C. Serum Nrf2 level was determined using a commercial ELISA kit (Abcam; Cat. No: ab277397).

4.2 RNA-sequencing dataset

The GSE95233 dataset from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/; derived from the GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array) included 22 control patients and 51 patients with septic shock; the latter underwent two blood draws, yielding 102 samples in total. We compared the 102 samples from septic shock patients (from GSM2500371 to GSM2500472) to control samples (from GSM2500349 to 2500370) and used the “limma” package of R software (R Foundation for Statistical Computing, Vienna, Austria) to identify DEGs between sepsis patients and controls.

4.3 GO and KEGG pathway enrichment analyses

We used the “ClusterProfiler” package of R software⁵⁸ to perform GO and KEGG pathway enrichment analyses to determine the molecular functions and signaling pathways of the identified DEGs.

4.4 Reagents and antibodies

LTA was purchased from Sigma–Aldrich (Cat. No: L3140). ML385 (Cat. No: HY-100523), a Nrf2 inhibitor, was from MedChemExpress. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (Beyotime; Cat. No: C1018) or Hoechst33342 (Yeasen; Cat. No: 40732ES03). Antibodies used in this study are listed in Table S2.

4.5 Cell culture and treatment

A549 lung epithelial cells were purchased from the American Type Culture Collection (ATCC) and cultured in Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin solution (all from Gibco; Cat. No: 16140071 and 11965092) at 37°C with 5% CO₂. A reserve solution of ML385 with a concentration of 10 mM was dissolved in DMSO. 5 mg of LTA was dissolved in 500 μL of ultrapure water. In the LTA time gradient experiment, cells were treated with a culture medium containing 10 μg/mL of LTA and incubated for 0.5, 2, and 4 h, respectively. In the LTA concentration gradient experiment, cells were treated with a culture medium containing 2.5, 5, 10, and 20 μg/mL of LTA, respectively, and incubated for 4 h. In the MMP and western blotting experiments, cells were pre-treated with ML385 at a concentration of 10 μM for 0.5 h and then exposed to 20 μg/mL of LTA for 4 h.⁵⁹

4.6 Bacterial strains and growth conditions

Bacterial cell recovery and preservation were performed as previously described.⁶⁰ Briefly, lyophilized powder of S. aureus strain ATCC29523 was cultured overnight (37°C, 220 rpm) to mid-log phase (optical density = 600 nm) in brain heart infusion-chloramphenicol broth. The cell suspension was evenly spread on nonselective Luria Broth medium for single colony counting. The bacterial solution was mixed with 40% glycerol in a 1:1 ratio and stored at −80°C.

4.7 Animals and ethics

Animal experiments were approved by the Animal Ethics Committee of Jiangsu Institute of Schistosomiasis Control (ethics no. IACUC-JIPD-2019043). WT C57BL/6 mice (8 weeks old, weighing 18−25 g) were obtained from Jiangnan University (Wuxi, China). Male Nrf2^−/− mice (8 weeks old, weighing 18−25 g) were obtained from the Model Animal Research Center (Nanjing, China).⁶¹ Genotyping (Figure S4B) and RT-qPCR (Figure S4C) were performed to confirm Nrf2 deficiency. The SA-ALI model was established by intraperitoneal injection of S. aureus into 8-week-old mice.^{7, 62} Specific amounts of SA were suspended in a suitable solution and then injected into the abdominal cavity of mice using a syringe, avoiding blood vessels, bladder, and any viscera.

WT and Nrf2^−/− mice were divided into control (100 μL of phosphate-buffered saline [PBS] was injected intraperitoneally), SA1 (2 × 10⁷ CFU S. aureus dissolved in 100 μL PBS), SA2 (1 × 10⁸ CFU S. aureus dissolved in 100 μL PBS), and SA3 (3 × 10⁸ CFU S. aureus dissolved in 100 μL PBS) groups according to the concentration of S. aureus that was injected.^{63, 64} After intraperitoneal injection of PBS/SA suspension in mice for 6 h,⁶⁵ inhale anesthesia with a dose of 2.0% isoflurane for 30 min. After 6 h of intraperitoneal injection of PBS/SA in mice, the mice were anesthetized and dissected to obtain whole blood, serum, BALF, fresh lung tissue, and paraformaldehyde-fixed lung tissue. Immediately after anesthesia, collect 0.8–1 mL of whole blood from the posterior vena cava. After a 30-min stasis, follow the steps in section 4.11 to obtain the serum. Simultaneously, open the airway and chest of the mice and use the method described in section 4.8 to obtain BALF. Within 20 min after anesthesia, collect fresh lung tissue from the mice. Store a portion at −80°C and fix another portion in 4% paraformaldehyde at room temperature. The paraformaldehyde-fixed lung tissue will be paraffin-embedded within 48 h. The portion of the fresh lung tissue is used for western blot and qRT-PCR. Bacterial cultures of BALF and peripheral whole blood confirmed that the sepsis model was successfully established (Figure S4A).

4.8 BALF

After tracheotomy, the alveoli of mice were lavaged using sterile pre-chilled PBS. The BALF was centrifuged and protein concentration in the supernatant was determined using a bicinchoninic acid assay protein quantification kit (Vazyme; Cat. No: E112-01). The pellet was used for cell counting; the total number of cells in BALF was determined by flow cytometry (BD C6; BD Biosciences).

4.9 RT-qPCR

RNA was extracted from lung tissue or cells using TRIzon reagent (CWBIO; Cat. No: CW0580) according to the manufacturer's instructions. The PrimeScript RT kit (Takara; Cat. No: RR047A) and SYBR Mix (Yeasen; Cat. No: 11198ES08) were used for reverse transcription and qPCR, respectively. Primer sequences and PCR conditions are listed in Tables S3 and S4. Expression levels of each gene were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and expressed as a fold of control. Observe the dissociation curve for the presence of only a single amplification peak to verify reaction specificity. This experiment was performed using LightCycler 480 Real Time PCR System (Roche). The 2^−∆∆Ct method was used to quantify the reaction mtDNA transcriptional activity was measured by examining the expression of ND-1, COX-1, and complex IV.

4.10 Histopathology and lung injury score

Mouse lung specimens were fixed in 4% paraformaldehyde for 48 h and then embedded in paraffin (ASP200S and EG1150H; Leica). Paraffin blocks were cut into 4 μm thick sections on a microtome (RM2245; Leica) that were imaged under a pathology scanner (Pannoramic MIDI; 3DHISTECH) after H&E staining. The lung injury score was assigned according to American Thoracic Society criteria based on examination of five randomly selected fields by two pathologists who were not involved in the study. The average score of the five random regions is taken as the lung injury score of this mouse, which is represented by this slice.

4.11 Serum biochemical indicators

After collecting whole blood from mice, the samples were placed in 1.5 mL EP tubes and allowed to stand for 30 min to facilitate the separation of plasma and serum. The samples were centrifuged for 10 min at 425 g and then for 20 min at 106 g. The supernatant was carefully aspirated to obtain 300 μL of serum at least, which was immediately stored at −80°C. The concentration of serum LDH (Cat. No: A020-2-2), MDA (Cat. No: A003-1-2), SOD (Cat. No: A001-3-2), and MPO (Cat. No: A044-1-1) was measured using commercial kits from Nanjing Jiancheng Bioengineering Institute (Nanjing, China).

4.12 Western blotting

Total protein from lung tissues and cells was extracted and western blotting was performed as previously described.⁶⁰ Briefly, cells and tissue extracts were resolved by 10 or 12% SDS-PAGE, transferred to nitrocellulose membrane, and incubated using various primary antibodies (Table S2) diluted 1:1000 at 4°C overnight, then incubated for 1 h at room temperature with the corresponding HRP-conjugated secondary antibody. Protein bands were visualized and imaged using ECL Chemiluminescence Assay Kit (Vazyme; Cat. No: E412-01) and Gel Image Analysis System (Universal Hood III; Bio-Rad). After importing the image into ImageJ software (National Institutes of Health, Bethesda, MD, USA), the gray values of bands were calculated after removing the background signal. VDAC1 was used as the loading control for mitochondrial proteins and GAPDH or β-actin was used as the loading control for whole-cell and cytoplasmic proteins.

4.13 Mitochondrial isolation

Fresh mouse lung tissue (0.1 g) or cells (5 × 10⁶) were collected and homogenized on ice and lysed to obtain intact and purified mitochondria using the Mitochondria Isolation kit (Bioss; Cat. No:C5032) according to the manufacturer's protocol.

4.14 Measurement of cellular ATP concentration and cell viability

Cell viability was assessed using Cell Counting Kit-8 (Vazyme; Cat. No: A311-01) and cellular ATP concentration was measured using the Enhanced ATP Assay kit (Beyotime; Cat. No: S0027), both according to the manufacturer's instructions.

4.15 MMP assay

The MMP assay was performed using a kit (Solarbio Science Amp Technology; Cat. No: M8650) according to the manufacturer's instructions. Fluorescence was visualized by confocal laser scanning microscopy at 400 × magnification (Model LSM880; Carl Zeiss).

4.16 Plasmid construction and transfection

The Nrf2-OE plasmid pcDNA3.1(+)-Nrf2-3 × FLAG-P2A-EGFP (forward primer [CMV-F], CGCAAATGGGCGGTAGGCGTG; reverse primer [EGFP-SEQR], GACACGCTGAACTTGTGGC) and the control vector H2713 pcDNA3.1(+)−3 × FLAG-P2A-EGFP were designed and constructed by OBiO Technology. A549 cells were seeded at a density of 5 × 10⁵ cells per well in six-well plates and transfected with 2 μg DNA per well mixed with 5 μL Liposomal Transfection Reagent (Yeasen; Cat. No: 40802ES02) according to the manufacturer's protocol.

4.17 Immunofluorescence analysis

Immunocytochemistry was performed as previously described.⁶⁶ Fluorescence was visualized by confocal laser scanning microscopy at 400 × magnification and analyzed using ZEN2 (blue edition) software (Carl Zeiss).

4.18 Dual-luciferase reporter assay

The PHB2 luciferase plasmid pGL4.10-promoter PHB2 wt (forward primer [RVprimer3], CTAGCAAAATAGGCTGTCCC; reverse primer [Luc2-N-Re], CGTCTTCGAGTGGGTAGAATG) and control vector H352 pGL4.10 were constructed by OBiO Technology.

The PHB2 luciferase plasmid and Nrf2-OE plasmid were cotransfected into cells and the dual-luciferase reporter assay was performed using a kit (Vazyme; Cat. No: DL101-01) according to the manufacturer's protocol. Fluorescence was detected using a microplate reader (SynergyH4; Berthold Technologies).

4.19 PHB2 RNA silencing

To enhance gene silencing efficiency, a mixture of three different PHB2 siRNAs was used: 5′-GCCACATCACAGAATCGTA-3′; 5′-GAACCCTGGCTACATCAAA-3′; 5′-CCAAGGACTTCAGCCTCAT-3′. The siRNAs were commercially synthesized (RiBobio). For negative controls, scrambled RNA oligonucleotides were used. For each well, 50 nM of each of the three oligos were transfected using Hieff Trans™ Liposomal Transfection Reagent (Yeasen, Cat. No: 40802ES02) according to the manufacturer's instructions. After incubation for 15 min, the medium was replaced with DMEM containing 10% fetal bovine serum. Transfected cells were used for the subsequent experiments 48 h after transfection.

4.20 Statistical analysis

Data are presented as means ± standard deviation. The difference between the two groups was evaluated for significance with an unpaired t-test. Continuous variables in clinical data were tested for normality. The independent samples t-test and Wilcoxon rank-sum test were applied to normal and non-normal data, respectively. The chi-squared test was used to evaluate categorical variables in clinical data. Correlations were determined by linear regression analysis. Cumulative survival rates were determined with the log-rank test. A p value < 0.05 was taken to reflect a significant difference in all tests. Prism v9.0 (GraphPad), Cytoscape v 3.8.2 (https://cytoscape.org/release_notes_3_8_2.html), SPSS v20 (IBM), R v.4.1.1, and Bioconductor v3.13 (https://bioconductor.org/news/bioc_3_13_release/) were used for data analysis and image processing.