The Transcription Factor DDIT3 Regulates Macrophage Function by Inhibiting KLF10 to Attenuate ALI/ARDS Inflammation

Abstract

ALI and ARDS are among the most common diseases in intensive care units, and are characterized by rapid onset and high mortality rates. Dysregulation of macrophage homeostasis is closely associated with the inflammatory cascade in ALI/ARDS. Transcription factors play critical roles in maintaining macrophage immune function. However, the mechanisms by which transcription factors and macrophages regulate the inflammatory imbalance in ALI/ARDS remain largely undefined. Here, we illustrate the role of DDIT3 in regulating macrophage immune function. Our comprehensive bioinformatics analysis revealed that DDIT3 is a key transcriptional regulator of ARDS with superior diagnostic potential. DDIT3 is highly expressed in macrophages and promotes M1 macrophage activation in ARDS patients, and in vivo/vitro models of ALI. DDIT3 deficiency significantly reduces the proportion of M1 macrophages and notably decreases in the secretion of inflammatory cytokines. RNA-seq and CUT&Tag analyses identified KLF10 as a target gene of DDIT3. Moreover, we demonstrated that the inhibition of KLF10 activity reversed the anti-inflammatory effect of DDIT3 silencing by restoring M1 macrophage polarization and inflammatory cytokine secretion. Our findings confirm that DDIT3 is a key regulator of macrophage polarization and inflammatory mediator secretion, suggesting that DDIT3 is a potential therapeutic target for ALI/ARDS.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10753-026-02451-4.

Introduction

Acute lung injury (ALI) and Acute Respiratory Distress Syndrome (ARDS) are a life-threatening condition characterized by acute hypoxemic respiratory failure, pulmonary edema, and dysregulated inflammation. It accounts for approximately 10% of intensive care unit (ICU) admissions and has been associated with mortality rates of up to 40% in recent studies [1, 2]. Despite advances in supportive care, current therapeutic strategies remain largely ineffective, as no specific pharmacologic therapy has been shown to significantly improve patient prognosis. A hallmark of ARDS pathology is an uncontrolled inflammatory response that drives immune activation, leading to alveolar-capillary barrier disruption and lung injury [3]. Macrophages, the predominant inflammatory cells in the lung microenvironment, are pivotal in orchestrating this response. Activated macrophages, particularly those differentiated from recruited circulating monocytes, secrete a spectrum of proinflammatory cytokines, thereby triggering a “cytokine storm” [4]. Therefore, a comprehensive understanding of the alterations in macrophage functional phenotypes and cellular processes in ALI/ARDS pathology may lead to the development of new treatment strategies.

Transcription factors (TFs) are intracellular proteins that regulate gene expression by specifically recognizing and binding to defined DNA sequences, thereby controlling the transcriptional activity of target genes [5]. In immune cells such as macrophages, TFs orchestrate the expression of a broad spectrum of genes involved in inflammation, cell survival, differentiation, and metabolic reprogramming [6, 7]. During ALI/ARDS, the activation of proinflammatory transcription factors drives macrophage polarization toward a proinflammatory (M1) phenotype and sustains the production of cytokines and chemokines [8]. DDIT3 (DNA damage-induced transcript 3), also known as GADD153 (G1 phase arrest and DNA damage-inducible protein 153) or CHOP (C/EBP homologous protein), is a transcription factor induced by diverse stress conditions, including autophagy and endoplasmic reticulum stress. It is well-characterized for regulating genes involved in cell cycle arrest and/or apoptosis [9–11]. Previous studies have identified DDIT3 as a driver of dyserythroiesis and a potential therapeutic target to restore defective erythrocyte differentiation in myelodysplastic syndrome patients [12]. Under prolonged glutamine deficiency, DDIT3 activation balances cellular adenosine triphosphate production by upregulating glycolysis via transcriptional effects while inhibiting oxidative phosphorylation through direct interaction with mitochondrial respiratory chain components [13]. Additionally, recent findings revealed that DDIT3 contributes to metabolic dysregulation by mediating insulin resistance through adipose tissue macrophage polarization [14]. While DDIT3 has been implicated in metabolic and immune pathways, its role as a transcriptional regulator of macrophage-derived inflammatory cytokines in ALI/ARDS remains poorly understood.

KLF10 (Kruppel-like factor 10) is a zinc finger transcription factor that plays an important role in regulating immune and inflammatory responses [15–17]. Previous studies have shown that KLF10 can exert anti-inflammatory effects by modulating cytokine expression and suppressing pro-inflammatory gene transcription in macrophages [18]. However, its upstream regulation and precise role in ALI remain poorly understood. These observations prompted us to investigate whether DDIT3 might regulate KLF10 to influence macrophage activation during pulmonary inflammation.

In this study, we identified DDIT3 as a key transcriptional regulator with superior diagnostic efficacy for ARDS through integrated bioinformatics analyses. Functional characterization revealed that DDIT3 critically modulates inflammatory responses via macrophages in both murine and cellular models of acute lung injury. We demonstrated that DDIT3 is highly expressed in macrophages during ALI/ARDS and is indispensable for maintaining macrophage homeostasis by regulating macrophage activation and proinflammatory cytokine secretion through the downstream effector KLF10. Collectively, these findings establish the DDIT3-KLF10 axis as a promising therapeutic target for ALI/ARDS.

Materials and Methods

Datasets and Preprocessing

We downloaded data from the Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo) and ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) databases for six transcriptional profiles (GSE196399, GSE32707, GSE65682, GSE236713, GSE243066, and E-MTAB-5273) for ARDS patients and healthy individuals. The characteristics of the six datasets are shown in Supplementary Table 1. The datasets GSE89953 and GSE32707 were used as the training cohort, and the other 4 datasets were used as the validation cohort. The datasets were subjected to batch effect correction via the “sva” package in R [19]. The combined datasets included a total of 90 ARDS patients and 36 healthy individuals. Additionally, 564 genes were taken from the transcription factor database (http://bioinfo.life.hust.edu.cn/AnimalTFDB) and examined.

Differentially Expressed Genes (DEGs) Analysis

Differential gene analysis was performed on the training dataset via the “limma” package [20]. The screening criteria were |log₂ fold change|> 0.585 and pvalue < 0.05 for significantly differentially expressed genes. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysises and visualization of DEGs were performed via the Metascape (www.metascape.org/) online tool with q < 0.05.

Co-Expression Network Construction

To further screen for genes highly correlated with ARDS traits, we selected the top 75% of genes with the median absolute deviation and constructed a gene co-expression network via the R package “WGCNA“ [21]. The pink soft threshold function was subsequently used to select the appropriate soft threshold. When the power of β was equal to 15 (R² = 0.92), the adjacency matrix was converted into a topological overlap matrix (TOM), which was used to construct a clustering dendrogram of the samples. Afterward, the estimation of module membership (MM) and gene significance (GS) linked the modules to clinical features. Pivotal modules were designated those with the highest pearson MM and an absolute pvalue of 0.05. High module connectivity and clinical significance were indicated by MM > 0.8 and GS > 0.2, respectively. We overlaid the highest ARDS-related module genes, transcription factor-related genes, and DEGs by applying a Venn diagram.

Diagnostic Marker-based Prediction Model Development and Validation

To develop a classification prediction model with high accuracy and stability, we integrated 12 machine learning algorithms and 113 algorithm combinations [22]. The comprehensive algorithms include Lasso, Ridge, Enet, Stepglm, SVM, glmBoost, LDA, plsRglm, RandomForest, GBM, XGBoost, and Naive Bayes. To train the model, the training cohort was selected as the reference dataset and the other 4 datasets were used as the validation cohort. The performance of the model was then evaluated by calculating the area under the ROC curve (AUC) for each model and model gene, and then visualizing the results via a heatmap.

Cells Infiltration Estimation

Single-sample gene set enrichment analysis (ssGSEA) implemented in the R package GSVA was employed to quantify the relative infiltration of 28 immune cells in the training cohort [23]. Using the “corrplot” package, we created a correlation heatmap to visualize the correlations between different infiltrating immune cells.

RNA Extraction and Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)

10 mL of peripheral venous blood was collected from each of the six patients with ARDS and healthy individuals from the First Affiliated Hospital of Zhengzhou University. The Berlin consensus definition served to make the diagnosis of ARDS [24]. Detailed information on the clinical patient samples is shown in Supplementary Table 2. Blood samples from each participant were handled with Ficoll solution (Solarbio Life Sciences, Beijing, China) to isolate peripheral blood mononuclear cells (PBMCs). Total RNA was then extracted from PBMCs with an RNA extraction kit (TRIzon Reagent, Shanghai, China) and reverse transcribed into cDNA. The cDNA products were then diluted 10-fold for real-time fluorescence quantitative qRT-PCR analysis utilizing a qPCR kit (cat. no. R223-01, Vazyme). One replicate well was set for each sample, and the experiment was repeated three times. In the end, the relative expression levels of the detected mRNAs were normalized to GAPDH. Primers can be found in Supplementary Table 3. The relative expression of mRNAs was calculated using the 2-ΔΔCt method and normalized to GAPDH. Patients voluntarily participated in this study and signed the consent form, which was approved by the Ethics Department of the First Affiliated Hospital of Zhengzhou University (2024-KY-0156-001) and by the Declaration of Helsinki of the World Medical Association.

Cell Culture

RAW264.7 cells were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). The cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS, cat. no.164210-500, Pricella Biotechnology) and 1% Penicillin-Streptomycin Solution (P.S, cat. no. PB180120, Pricella Biotechnology) and placed in an incubator maintained at 37 °C under humidified atmospheric conditions consisting of 5% CO₂. RAW264.7 cells were plated in 24-well plates (3 × 10⁵ cells/well) for 24 h, then stimulated with or without 1 µg/ml lipopolysaccharide (LPS, cat. no. L4391, Sigma) for 24 h, after which the cells were collected for subsequent experiments [25].

Construction of Animal Models and Experimental Procedures

7–8 week old male C57BL/6J mice were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). The study was approved by the Ethics Committee for Animal Experiments of the First Affiliated Hospital of Zhengzhou University. 10 mice were randomly divided into control and experimental groups. The mice in the experimental group were injected with LPS (10 mg/kg) via an endotracheal drip, and the mice in the control group were treated with an equal amount of saline. The mice were sacrificed 24 h later.

Western Blotting Analysis

RAW264.7 cells were collected, protein was extracted via RIPA buffer (cat. no. 89900, Thermo Fisher Scientific) in the presence of a proteinase inhibitor cocktail (cat. no. P8340, Sigma), and protein concentrations were measured using BCA protein assay kit (cat. no. P1511-5, PPLYGEN). The proteins were loaded onto 12% SDS-PAGE and transferred onto PVDF membranes, which were blocked with 5% non-fat milk at room temperature for 1 h, and incubated with primary antibodies overnight at 4 °C. After the membrane was washed with PBST three times, it was incubated with the secondary antibody at room temperature for 1 h [26]. Finally, a chemiluminescence reagent imaging system was used to detect the bands, and all the bands were measured using ImageJ software. The primary antibodies used are shown in Supplementary Table 3.

Immunohistochemistry Analysis

Lung tissues were obtained from the mice and embedded in paraffin. After dewaxing and antigen thermal repair, 3% H₂O₂ solution was applied for 10 min to inactivate endogenous peroxidase, and 1% BSA was used for blocking. The primary antibodies against DDIT3 and F4/80 were incubated at 4 °C overnight, and the corresponding secondary antibodies were added at room temperature and incubated for 2 h. Subsequently, the samples were incubated with DAB and hematoxylin staining solution. Finally, the samples were observed under a microscope and analyzed using ImageJ software (Bethesda, USA). The primary antibodies used are presented in Supplementary Table 4.

Lentiviral Packaging and Cells Infection

pLKO.1-Ddit3-con, and pLKO.1-Ddit3-sh plasmids (Miaoling Bio) were co-transfected along with psPAX2 and pMD2G helper plasmids (Addgene) into 293 T cells using Neofect regent (Neofect Biotech). The supernatant containing the lentivirus was collected and filtered with 0.22 μm filters (Millipore) after 48 h, after which the RAW264.7 cells were infected immediately with polybrene (5 µg/mL) for 10 h. After 48 h of infection, the cells were selected with 5 µg/mL puromycin for stable transfection and named Luciferase-shRNA and Ddit3-shRNA, respectively.

Enzyme-linked Immunosorbent Assay (ELISA)

The expression levels of the cytokines, IL-1β, IL-6, and IL-18 were evaluated by ELISA following the manufacturer’s instructions (Elabscience). The evaluation was performed on 3 duplicates of each group to ensure consistency.

Staining and Flow Cytometric Analysis of M1 Macrophages

The cells were suspended in buffer I at a concentration of 0.1 × 10⁶ cells/25µL and stained with PE-conjugated anti-mouse CD86 at room temperature for 30 min in the dark. Cells were washed with buffer I. The cells were then resuspended in 200µL buffer I and stain with the viability marker DAPI on ice for 15 min in the dark. Cells were analyzed within 1 h of staining using Becton Dickinson LSR Fortessa. The primary antibodies used are presented in Supplementary Table 4.

RNA-seq and Bioinformatics Analyses

Before sequencing, 2 million cells were collected and washed twice with PBS. The RNA was extracted and used as the cDNA input, and a cDNA library was prepared with an Illumina TruSeq DNA library kit, and sequenced on an Illumina NovaSeq 6000 platform. The original data were filtered and compared with the reference genome, which can be downloaded from the Ensembl database (http://www.ensembl.org/index.html, mouse genome build mm10). HISAT2 version 2.2.1 [27] (http://daehwankimlab.github.io/hisat2/) was used to read the counts for each gene in each sample and gene expression was assessed via StringTie version 2.1.7 (https://ccb.jhu.edu/software/stringtie/) [28]. FPKM (Fragments Per Kilobase of exon model per Million mapped fragments) was calculated to evaluate the expression level. DEGs were defined by p.adjust < 0.05 and |fold change| >2 using the DESeq2 version 1.38.3 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html) [29]. GO and KEGG analyses of the DEGs were performed via the Metascape (www.Metascape.org/) online tool, with q < 0.05.

CUT&Tag

The cells were harvested, counted and centrifuged for 3 min at 600×g at room temperature. A total of 1 × 10⁵ cells were washed twice in 500 µL of wash buffer by gentle pipetting. Concanavalin A coated magnetic beads were prepared as the kit manual described, 10 µL of activated beads were added per sample and incubated at RT for 10 min. According to the kit instructions, cells were sequentially incubated with ConA Beads, primary antibody (anti-DDIT3 antibody, secondary antibody (Goat anti-Rabbit IgG)) and Hyperactive PG-TN5/PA-TN5 Transposon and then fragmented. The fragmented DNA was extracted from the samples and amplified by PCR. CUT&Tag libraries were constructed using HyperactiveTM In-Situ CUT&Tag Library Prep Kit (TD901, Vazyme) and sequenced on an Illumina NovaSeq platform, and 150-bp paired-end reads were generated. Trim Galore version 0.6.6 was used to remove adapter and low-quality reads [30]. Align paired-end reads using Bowtie2 version 2.3.5.1 [31] with options: “--end-to-end --very-sensitive”. Peak calling uses MACS2 version 2.1.1 [32] with threshold: q < 0.01, and the differential binding sites of Ddit3 were calculated using the DiffBind version 3.8.4 with the parameter log2foldchange > = 1 and FDR < 0.05 (https://www.cruk.cam.ac.uk/core-facilities/bioinformatics-core/software/diffbind.html). Scatterplots, correlation plots, and heatmaps are displayed using deepTools version 2.5.7 [33]. Annotation of peaks is performed using an R package ChIPseeker version 1.34.1 [34] (https://www.cruk.cam.ac.uk/core-facilities/bioinformatics-core/software/ ChIPseeker.html). The CUT&Tag peak/region < 1000 bp from the nearest transcriptional start site (TSS) was defined as the promoter. Hypergeometric Optimization of Motif Enrichment (HOMER) version 4.11 [35] (http://homer.ucsd.edu/homer/) was used to search for the binding motif.

Statistical Analysis

All data processing, statistical analysis, and plotting were conducted in R 4.3.0 software and GraphPad Prism 8.0.2. Statistical significance was determined with Student’s t-test for two groups. P < 0.05 was considered statistically significant.

Results

Identification of Transcription Factor-related Differential Genes (TFRGs) Between Healthy and ARDS Samples

To identify potential genes involved in ARDS, two ARDS datasets, GSE196399 and GSE243066, were merged to obtain the integrated training dataset after batch effect removal. The batch effect of the samples in the integrated GEO dataset was effectively mitigated after the batch removal procedure (Supplementary Fig. 1A-B). Significant differences between the two groups were shown by (Supplementary Fig. 1 C). GO enrichment analysis of the DEGs revealed significant associations with biological processes including mononuclear cell differentiation, regulation of inflammatory responses, DNA-binding transcription factor activity, and macrophage activation. In contrast, KEGG pathway enrichment analysis revealed strong enrichment of inflammation-related signaling pathways, such as the NF-κB signaling pathway, FoxO signaling pathway, and NOD-like receptor signaling pathway (Fig. 1A-C).

Fig. 1.

Identification of transcription factors-related genes via WGCNA.(A) Volcano map showing genes with significant difference between healthy and ARDS sample. (B) The top pathways revealed by gene ontology (GO) analysis of the differentially expressed genes between healthy and ARDS sample. (C) The top pathways revealed by gene Kyoto Encyclopedia of Genes and Genomes (KEGG) of the differentially expressed genes between healthy and ARDS sample. (D) Correlation analysis between module eigengenes and clinical traits. (E) The high correlation between GS and MM in the turquoise module. (F) The overlapping differentially expressed genes, WGCNA and transcription factor-related genes

Previous studies have shown that TFs play a central regulatory role in the initiation, maintenance, and extinction of the inflammatory response [36, 37]. To investigate the regulatory role of TFs in ARDS-associated inflammation, we performed WGCNA on the training dataset. Based on a dynamic tree-cutting algorithm and a module merging threshold of 0.25, nine distinct gene co-expression modules were identified for subsequent analysis. The resulting module clustering and merging process is illustrated in the dendrogram (Supplementary Fig. 1D). The associations between modules and clinical symptoms were then investigated using correlations between ME values and clinical features. Figure 1D shows that turquoise modules were positively correlated with ARDS (r = 0.85, p = 5e-35). Clinically significant modules were identified. Figure 1E shows that the turquoise module was highly correlated with ARDS in the scatter plot of MM versus GS (r = 0.79, p = 1e-200). We overlapped DEGs, transcription factor-related genes, and turquoise modules to identify 22 TFRGs for further analysis (Fig. 1F).

Construction and Validation of the Diagnostic Signatures Based on Integrative Machine Learning

To better identification of ARDS patients in clinical practice, we constructed diagnostic prediction models for ARDS by accepting the 22 TFRGs screened above in 113 combinations based on 10 machine-learning algorithms, and further calculated the average AUC value of each model. Interestingly, the best model was Lasso with the highest mean AUC (0.953) (Fig. 2A). In Lasso regression, the optimal λ was obtained when the partial likelihood bias was minimized (Fig. 2B-C). Finally, 18 transcription factor-related genes were identified, followed by ranking them according to their importance in the diagnostic prediction model (Fig. 2D). Based on gene ranking and a comprehensive literature review, DDIT3 was identified as being closely associated with inflammation-related diseases. The results of immune cell infiltration evaluated according to the ssGSEA algorithm showed that multiple immune cells differed significantly between ARDS patients and healthy persons (Supplementary Fig. 1E). Moreover, DDIT3 was positively correlated with M0 macrophages (Fig. 2E). The above results suggest that DDIT3 may be involved in the developmental process of ARDS by regulating macrophages.

Fig. 2.

Construction and validation of diagnostic models by transcription factor-associated differential genes (TFRGs).(A) The AUC value of 113 machine-learning algorithm combinations in the four testing cohorts. (B) Lasso coefficient computation. The vertical dashed line shows the ideal lambda value. (C) Ten-fold cross-validation for Lasso model parameter adjustment. Every curve represents a gene. (D) Ranking of 22 genes of relative importance in TFRG. (E) Correlation between DDIT3 and infiltrating immune cells

DDIT3 was Elevated in Vivo and in Vitro ARDS/ALI Models

To determine the role of DDIT3 in macrophage immune function during ALI/ARDS, we first examined Ddit3 expression in lung tissue of ALI mice. Using co-immunostaining for Ddit3 and F4/80 (a macrophage marker), we found that Ddit3 was predominantly localized to macrophages and was upregulated in lung tissues of ALI mice (Fig. 3A). Next, we examined DDIT3 expression in ARDS patients and healthy individuals PBMC using qRT-PCR. DDIT3 levels were also significantly elevated in ARDS compared with normal controls (Fig. 3B). To further characterize the role of Ddit3 in the pathogenesis of ALI, we found that Ddit3 expression levels were significantly elevated in LPS-stimulated RAW264.7 macrophages, bone marrow-derived macrophages (BMDM) (Fig. 3C-H). Taken together, these findings suggest that DDIT3 is highly expressed in macrophages and may promote lung inflammation, which prompted us to further explore the exact role of macrophage DDIT3 in ALI/ARDS.

Fig. 3.

DDIT3 is upregulated in the macrophages of in vivo and in vitro ALI models.(A) Immunofluorescence staining of the macrophage markers F4/80 (red) and Ddit3 (green) and DAPI (blue) in lung sections taken 24 h after ALI models. Scale bar, 20 μm. (B) Isolation of peripheral blood mononuclear cells (PBMCs) from peripheral blood samples from healthy and ARDS samples. Quantitative RT-PCR was used to analyze the expression levels of DDIT3 in PBMCs. GAPDH was used as internal reference. (C) qRT-PCR results showing Ddit3 mRNA expression levels in RAW264.7 cells stimulated with 1 µg/mL LPS for 24 h. β-actin was used as internal reference. (D) qRT-PCR results showing Ddit3 mRNA expression levels in BMDM cells stimulated with 100ng/mL LPS for 24 h. β-actin was used as internal reference. (E) Representative western blots showing Ddit3 protein levels in RAW264.7 cells stimulated with 1 µg/mL LPS for 24 h. (F) Quantitative analysis of DDIT3 protein levels from 3 independent experiments. (G) Representative western blots showing DDIT3 protein levels in BMDM cells stimulated with 100ng/mL LPS for 24 h. (H) Quantitative analysis of Ddit3 protein levels from 3 independent experiments. Statistical analysis is from 3 independent experiments, and the bar plot represents mean ± SD of triplicate samples. **p < 0.01, ***p < 0.001

Knockdown of Ddit3 Attenuates LPS-induced Lung Injury Inflammatory Response in RAW264.7

To further elucidate the mechanism of Ddit3 in LPS-induced ALI, we investigated whether Ddit3 knockdown could modulate the inflammatory response. First, Ddit3 knockdown was performed in RAW264.7 cells, and both qRT-PCR and Western blot analyses confirmed a knockdown efficiency exceeding 50% (Fig. 4A-C). Next, we utilized the Ddit3 knockdown RAW264.7 to assess the secretion of pro-inflammatory cytokines in ALI. We found that the supernatants of LPS-treated RAW264.7 cells showed increased secretion of IL-1β, IL-6, and IL-18 compared to the control group, whereas it was significantly reduced in the LPS + shDdit3 group. (Fig. 4D). Meanwhile, similar results were observed for the mRNA expression levels of pro-inflammatory cytokines in RAW264.7 cells (Supplementary Fig. 2). It is well known that M1 macrophages secrete pro-inflammatory factors, whereas M2 macrophages secrete anti-inflammatory factors [38]. To investigate the effects of Ddit3 on M1 macrophage polarization, we performed a flow cytometry assay and found that the percentage of M1 macrophages was significantly increased in LPS-treated RAW264.7 cells compared with the control group, whereas the number of M1 macrophages was decreased in the LPS + shDdit3 group (Fig. 4E-F). These results suggest that Ddit3 knockdown attenuates the LPS-induced lung injury inflammatory response in RAW264.7 cells by reducing the proportion of pro-inflammatory M1 macrophages.

Fig. 4.

Knockdown of Ddit3 attenuates LPS-induced lung injury inflammatory response in RAW264.7. (A) qRT-PCR results showing Ddit3 mRNA expression levels in Luciferase-shRNA or Ddit3-shRNA stransduced RAW264.7 cells. (B) Representative western blots showing Ddit3 protein levels in NC-shRNA or Ddit3-shRNA stransduced RAW264.7 cells. (C) Quantitative analysis of Ddit3 protein levels from 3 independent experiments. (D) Determination of the secretion of IL-1β, IL-6, and IL-18 in the supernatant of RAW264.7 cells by ELISA. (E) Representative flow cytometry M1 macrophage numbers were assessed by Cd86 staining of RAW264.7 cells. (F) Quantification of the proportion of 3 independent M1 macrophages. Statistical analysis is from 3 independent experiments, and the bar plot represents mean ± SD of triplicate samples. **p < 0.01, ***p < 0.001

Ddit3 Inhibits LPS-induced Inflammatory Response in ALI Through Gene Regulation

We hypothesized that Ddit3 acts as a transcription factor that can attenuate lung inflammation by modulating the transcriptional response, and that knockdown of Ddit3 leads to inhibition of macrophage pro-inflammatory gene expression. To test this, we performed RNA-seq of RAW264.7 and examined the effect of Ddit3 knockdown on LPS-stimulated macrophage gene expression. PCA of RNA-seq data showed that there were significant transcriptional differences between the Luci, LPS + Luci and LPS + Ddit3 groups (Fig. 5A). The results of differential analysis showed that the gene expression pattern was significantly changed after LPS + Ddit3 treatment, compared with LPS + Luci. A total of 823 differentially expressed genes were identified, including 456 up-regulated genes and 367 down-regulated genes (Fig. 5B). In addition, our transcriptional data also showed that IL-1β, IL-6, IL-18 mRNA levels were significantly reduced in the LPS + Ddit3 group compared to the LPS + Luci group (Fig. 5C). This is consistent with ELISA results. The results of GO analysis indicated that the up-regulated genes were enriched in M2 macrophage-like cells, such as myeloid cell differentiation, wound healing, epithelial cell migration, and regulation of macrophage activation (Fig. 5D). In contrast, down-regulated genes were mainly associated with chromosome segregation, positive regulation of cell cycle, regulation of innate immune response, cell activation involved in immune response, regulation of DNA binding TF activity. These RNA-seq results suggest that Ddit3 knockdown broadly regulates the transcriptional response of macrophages and may induce genetic programs that negatively regulate pro-inflammatory responses.

Fig. 5.

Ddit3 inhibits LPS-induced inflammatory response in ALI through gene regulation. (A) Principal component analysis of samples representing 3 biological replicates from RAW264.7 cells stimulated with Luci or LPS + Luci or LPS + shDdit3. (B) Volcano map showing genes with significant difference between LPS + Luci and LPS + shDdit3 group. (C) The FPKM value of IL-1β, IL-6, and IL-18. (D) GO analysis of DEGs showing biological processes enriched in DEGs upregulated (red) or downregulated (blue)between LPS + Luci and LPS + shDdit3 group in RAW264.7 stimulated with LPS. (E) KEGG analysis of DEGs showing biological processes enriched in DEGs upregulated (red) or downregulated (blue) between LPS + Luci and LPS + shDdit3 group in RAW264.7 stimulated with LPS. *p < 0.05, **p < 0.01, ***p < 0.001

Ddit3 as a Transcription Factor to Regulate Klf10 Expression

To investigate the role of Ddit3 as a transcription factor in macrophage gene regulation. We performed CUT&Tag sequencing to explore its genome-wide distribution. Principal component analysis and Spearman correlation analysis showed high reproducibility between the three sets of replicates, suggesting different gene expression patterns in the three groups (Supplementary Fig. 3A-B). CUT&Tag signaling heatmaps and profiles showed that the peak Ddit3 signal in the 3kb region around the TSS in the LPS + Luci group was dramatically elevated compared to the Luci group. While the DDIT3 peak signal was significantly reduced in the LPS + shDdit3 group (Fig. 6A-B). The genome annotation results showed that the CUT&Tag peaks in the Ddit3 knockdown group were mainly enriched in the distal DNA regions (34.57%) of the transcription start site (TSS) (Fig. 6C). In addition, compared with the LPS + Luci group, the Luci and LPS + shDdit3 groups significantly reduced chromatin accessibility to DNA regions associated with IL-1β, IL-6, and IL-18 genes (Fig. 6D). Both RNA-seq and CUT&Tag data suggest that Ddit3 regulates gene expression (i.e., 823 genes induced by DDIT3) and chromatin accessibility (i.e., 8507 DNA regions opened by Ddit3). Therefore, we want to know whether knockdown of Ddit3-regulated genes or enhancer regions was involved in suppressing proinflammatory responses in macrophages. We performed a joint analysis of genes associated with Ddit3-binding DNA and genes up-regulated after Ddit3 in silenced macrophages, and ultimately identified the shared genes Lipc, Tcim, Klf10 and Ptges (Fig. 6E-F). The results of RNA-seq and GSE65682 in the Luci/Healthy and LPS + Luci/ARDS groups showed that only Klf10 mRNA expression level was reduced in LPS + Luci/ARDS compared with the Luci/Healthy group (Supplementary Fig. 3C-D). These findings suggest that Ddit3, functioning as a transcription factor, may regulate the expression of the Klf10 and thereby modulate the release of inflammatory cytokines.

Fig. 6.

Ddit3 acts as a transcription factor to regulate Klf10 expression. (A) Heat maps of CUT&Tag signals around TSS for Luci (left), LPS + Luci (middle) and LPS + shDdit3 (right). (B) Representative peak plots of CUT&Tag signals around the TSS for Luci (blue), LPS + Luci (red) and LPS + shDdit3 (green). (C) Pie plot of the distribution of differential CUT&Tag peaks related to gene features after knockdown of Ddit3 treatment. (D) Gene locus of CUT&Tag at IL-1β, IL-6, and IL-18 loci with Luci or LPS + Luci or LPS + shDdit3. (E) Veen plot and heat map of CUT&Tag and RNA-seq showed Ddit3-enriched differential gene expression values. (F) Gene locus of CUT&Tag at Lipc, Tcim, Klf10 and Ptges loci with LPS + Luci or LPS + shDdit3

Ddit3 Knockdown Attenuates LPS-induced Inflammation in ALI by Upregulating Klf10

To investigate the relationship between Ddit3, Klf10 and inflammatory response. We examined the mRNA expression of KLF10 in PBMC from healthy individuals and ARDS patients using qRT-PCR. The results showed that KLF10 mRNA expression was significantly decreased in ARDS patients compared with healthy individuals (Fig. 7A). Meanwhile, in the Ddit3 knockdown RAW264.7 macrophages treated with LPS, the Klf10 expression level was significantly elevated (Fig. 7B-D). In conclusion, our results confirm that Ddit3 may regulate Klf10 expression in macrophages. Furthermore, we hypothesized that Klf10 is required to mediate the pro-inflammatory effects of Ddit3. To test this, we treated Ddit3-silenced macrophages with the KLF10 inhibitor (iKLF10) KLF10-IN-1 (#48 − 15) and LPS [39]. The results revealed that the inhibitory effect of knockdown of Ddit3 on the secretion of pro-inflammatory IL-1β, IL-6 and IL-18 factors was significantly reduced after iKLF10 treatment (Fig. 7E). Meanwhile, similar results were observed for the mRNA expression levels of pro-inflammatory cytokines in RAW264.7 cells (Supplementary Fig. 4). Flow cytometry assay showed that the number of M1 macrophages was significantly increased after the addition of iKLF10 inhibitor (Fig. 7F-G). This suggests that Klf10 is critical for Ddit3-mediated macrophage inflammatory factor release and M1 macrophage ratio.

Fig. 7.

Ddit3 knockdown attenuates LPS-induced inflammation in ALI by upregulating Klf10. (A) Isolation of peripheral blood mononuclear cells (PBMCs) from peripheral blood samples from healthy and ARDS samples. Quantitative RT-PCR was used to analyze the expression levels of KLF10 in PBMCs. (B) qRT-PCR results showing Klf10 mRNA expression levels in Luci or LPS + Luci or LPS + shDdit3. (C) Representative western blots showing Klf10 protein levels in NC-shRNA or Ddit3-shRNA. (D) Quantitative analysis of Klf10 protein levels from 3 independent experiments. (E) Determination of the secretion of IL-1β, IL-6, and IL-18 in the supernatant of RAW264.7 cells by ELISA. (F) Representative flow cytometry M1 macrophage numbers were assessed by Cd86 staining of RAW264.7 cells. (G) Quantification of the proportion of 3 independent M1 macrophages. *p < 0.05, **p < 0.01, ***p < 0.001

Discussion

DDIT3 has been implicated in regulating critical pathological processes, including immune-inflammatory responses, airway remodeling, apoptosis, and tissue repair [40–43]. Consistently, DDIT3 has been established as a key regulator in the pathogenesis of multiple respiratory diseases, such as chronic obstructive pulmonary disease, pulmonary fibrosis, asthma, and lung cancer. However, its molecular mechanisms in ALI/ARDS remain largely undefined. Here, we integrated machine learning-based bioinformatics analyses of multiple ARDS transcriptome datasets and identified DDIT3 as a potential early diagnostic biomarker for ARDS. Using clinical samples from ARDS patients, in vitro macrophage cell models, and in vivo mouse ALI models, we systematically investigated the role of DDIT3 in ARDS pathogenesis. Interestingly, our results reveal that DDIT3 exacerbates ALI/ARDS inflammatory responses by suppressing KLF10-mediated macrophage polarization toward an anti-inflammatory phenotype. These findings establish DDIT3 as a novel therapeutic target for modulating macrophage-driven inflammation in ALI/ARDS.

ARDS is a highly heterogeneous clinical syndrome with a high incidence in critical illness. Due to the highly variable underlying mechanisms of ARDS and the lack of specific biomarkers, diagnosis is clinical [44–46]. According to the LUNG-SAFE study, ARDS was not identified in more than half of patients when criteria for ARDS were met [47]. The poor reliability of some of the criteria in the Berlin definition may lead to poor understanding by clinicians. The problem of missed diagnosis and delayed diagnosis of ARDS due to many reasons remains unresolved. To address this challenge, multiple ARDS bulk RNA-seq datasets were used to identify TFRGs based on the WGCNA algorithm. Using these TFRGs, we developed a reliable diagnostic model for ARDS based on a variety of machine learning algorithms and their combinations that can autonomously select key genes. Through peripheral blood, cell and animal experiments, it was confirmed that DDIT3 blood markers have great potential for early identification of ARDS in clinical practice. However, the precise mechanisms by which DDIT3 contributes to the pathogenesis of ARDS require further investigation.

Excessive inflammation is thought to be a core part of the development of ALI/ARDS [48]. Among them, macrophages play an important role. Macrophages, which act as control switches of the immune system, are involved in the regulation of pro-inflammatory and anti-inflammatory responses [49]. Recent studies have shown that Chop deficiency regulates signal transducer 6 phosphorylation and peroxisome proliferator-activated receptor γ expression, promotes M2 macrophage production to protect mice from bleomycin-induced pulmonary fibrosis [41]. Interestingly, Wang et al. reported that Chop regulates signal transduction and activator of transcription factor 6 phosphorylation, thereby enhancing the expression of the mouse transcription factor EC, followed by transcription of IL-4 receptor α expression to facilitate M2 programming in macrophages [42]. These seemingly contradictory results suggest that DDIT3 is involved in the regulation of immune responses but has different mechanisms of action in different disease models. In our study, DDIT3 expression was found to be elevated in ARDS patients, mice, and macrophages, consistent with previous studies [50]. Furthermore, by evaluating macrophages knocked down for Ddit3 in vitro, we found that Ddit3 reduced the proportion of M1 macrophages, thereby inhibiting macrophage secretion of inflammatory cytokines. Our RNA-seq analysis further supported these findings, revealing that Ddit3 knockdown significantly altered macrophage transcriptional profiles, downregulating key pro-inflammatory genes (e.g., IL-1β, IL-6, IL-18) and upregulating genes associated with M2-like functions such as myeloid differentiation and wound healing, suggesting that Ddit3 modulates macrophage polarization through transcriptional reprogramming. However, the specific mechanism by which Ddit3 regulates macrophages needs to be further elucidated experimentally.

It is well known that transcription factors, as core regulators of gene expression, play an important role in the occurrence and development of respiratory diseases [5]. Several previous studies have shown that transcription factors regulate protein expression at the transcriptional level by binding to promoters or enhancers, thereby regulating macrophage polarization and cytokine release [51, 52]. Studies have shown that DDIT3 binds directly to the T-bet promoter to inhibit effector T cell activity in tumors [53]. To investigate the mechanism by which DDIT3 regulates macrophage polarization and cytokine release. Our CUT&Tag sequencing results showed that Ddit3 was mainly enriched in the enhancer region and the chromatin openness of pro-inflammatory cytokines was reduced in the ALI cell model with Ddit3 knockdown. This shows that Ddit3 is indeed able to regulate chromatin accessibility and gene transcription, which is consistent with the results of previous studies [54]. How does Ddit3 specifically regulate macrophage activity and transcription of cytokine genes? When KLF10 is deficient, it has been reported to promote exacerbation of inflammation such as heart and liver disease [55, 56]. Here, we show that Klf10 expression is strongly upregulated in the Ddit3 knockdown ALI model, and that the DNA-binding motif of Klf10 is highly enriched in open chromatin regions in the Ddit3 knockdown model, and our results suggest that Klf10 is a negative regulator of the pro-inflammatory response. To support this, KLF10 inhibitor and KLF10 deficiency led to loss of anti-inflammatory effects of Ddit3 knockdown and alterations in macrophages. Therefore, Ddit3 knockdown has the potential to activate Klf10 and lead to a decrease in the expression of pro-inflammatory cytokines. The specific mechanism of KLF10 regulating macrophages needs to be further elucidated in the future.

Although our findings advance understanding of macrophage function under specific experimental conditions, certain limitations should be considered when interpreting the results. The M1/M2 dichotomy, while offering a useful conceptual reference, does not fully encompass the plasticity and spectrum of macrophage activation observed in vivo [57–59]. Our use of RAW264.7 cells provided a controllable and reproducible system; nevertheless, primary macrophages or in vivo validation will be essential to delineate the physiological relevance and complexity of these responses.

In conclusion, our experimental results clearly correlate macrophages with the development of ALI/ARDS and demonstrate that deletion of DDIT3 is involved in LPS-induced ALI by upregulating transcription of KLF10(Fig. 8). Targeting the DDIT3-KLF10 signaling axis may provide a new strategy for preventing or ameliorating LPS-induced ALI.

Fig. 8.

Molecular mechanisms underlying that transcription factor DDIT3 attenuates ALI/ARDS via inhibition of KLF10 to macrophage function

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ALI

Acute lung injury

ARDS

Acute respiratory distress syndrome

ICU

Intensive care unit

TFs

Transcription factors

DDIT3

DNA damage-induced transcript 3

GADD153

G1 phase arrest and DNA damage-inducible protein 153

CHOP

C/EBP homologous protein

KLF10

Kruppel-like factor 10

GEO

Gene Expression Omnibus

DEGs

Differentially Expressed Genes

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

WGCNA

Weighted Gene Co-expression Network Analysis

TOM

Topological overlap matrix

MM

Module membership

GS

Gene significance

AUC

Area under the ROC curve

ssGSEA

Single-sample gene set enrichment analysis

PBMCs

Peripheral blood mononuclear cells

qRT-PCR

Quantitative Real-Time Polymerase Chain Reaction

LPS

Lipopolysaccharide

ELISA

Enzyme-linked immunosorbent assay

TFRGs

Transcription factor-related differential genes

PCA

Principal component analysis

BMDM

Bone marrow-derived macrophages

Author Contributions

**Mengfei Sun: ** Writing – original draft, Visualization, Validation, Software, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. **Qianqian Yang: ** Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation. **Yingshuai Tan: ** Methodology, Data curation, Validation. **Chunling Hu: ** Methodology, Data curation, Validation. **Shilong Zhao: ** Methodology, Data curation. **Xiaoxiao Lu: ** Methodology, Data curation. **Jing Gao: ** Supervision, Project administration. **Lihua Xing: ** Writing – review & editing, Supervision, Conceptualization.

Funding

This study was supported by the Province-Ministry Co-built Key Project of Henan Province, No. SBGJ202302042; the project of Education Department of Henan Province, No.24B320030; the project of Education Department of Henan Province, No.24A320023. the Joint construction project of Health Commission of Henan Province, No. LHGJ20210303; the project of Education Department of Henan Province, No.24A320047.

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics Approval and Consent to Participate

Patients voluntarily participated in this study and signed the consent form, which was approved by the Ethics Department of the First Affiliated Hospital of Zhengzhou University (2024-KY-0156-001) and by the Declaration of Helsinki of the World Medical Association. All animal experiments were approved by the Ethics Department of the First Affiliated Hospital of Zhengzhou University (2024031502).

Consent for Publication

All authors agree with the manuscript content.

Competing Interests

The authors declare no competing interests.