The circular RNA hsa_circ_0003091 regulates sepsis-induced lung injury by sponging the miR-149/Smad2 axis

Differential expression of circRNAs in sepsis patients

First, we collected clinical blood samples from sepsis patients (n=3) and healthy controls (n=2) and performed transcriptomic profiling to identify deregulated circRNAs. A total of 44 circRNAs were differentially expressed (35 upregulated and 9 downregulated) between sepsis patients and healthy subjects (Figure 1A, 1B).

Figure 1. Differential expression of circRNAs in sepsis patients. (A) The scatter plot shows the circRNAs expression profiles between sepsis patients and healthy subjects. (B) Dysregulated circRNAs between sepsis patients and healthy subjects. (C, D) GO and KEGG enrichment analyses of dysregulated circRNAs between the two groups. (E, F) Validation of expression changes in top 10 circRNAs. *p < 0.05; ***p < 0.001.

Gene ontology (GO) enrichment and kyoto encyclopedia of genes and genomes (KEGG) pathway analysis of circRNA genes

Further, we found that upregulated circRNAs were associated with metabolic processes and catalytic activity in sepsis patients, while downregulated circRNAs were related to cellular processes (Figure 1C). The KEGG analysis revealed that cytokine-cytokine receptor interactions were associated with upregulated circRNAs, while fructose and mannose metabolism was related to downregulated circRNAs (Figure 1D).

Validation of circRNAs expression by qRT-PCR

Consistent with the sequencing analysis, the qRT-PCR results revealed that, compared to healthy controls, 7 circRNAs were upregulated and 3 were downregulated in both peripheral blood and bronchoalveolar lavage fluid from sepsis patients (Figure 1E, 1F). Particularly, the expression of has_circ_0003091 was ranked top 1 and was selected for subsequent analyses.

Expression of has_circ_0003091 in LPS-induced ALI mice

To explore the effects of has_circ_0003091 (mmu_circ_0015268) in endothelial cell dysfunction and pulmonary vascular injury, its expression was determined in the lung and lung EC tissues from LPS-induced ALI mice. The histological analysis showed that sepsis led to pulmonary injury (Figure 2A, 2B). Additionally, the levels of hepatic injury (ALT activity) and renal injury (creatinine concentration) markers were dramatically increased in sepsis mice compared to healthy controls (Figure 2C, 2D). These data indicated the successful establishment of the sepsis model. Compared to the lung tissues of healthy mice, ALI mice presented enhanced expression of has_circ_0003091 (Figure 2E). Meanwhile, we also observed significantly elevated expression of has_circ_0003091 in LPS-treated primary murine pulmonary microvascular endothelial cells (PMVECs) compared to untreated cells (Figure 2F). Altogether, these results showed that LPS induction significantly enhanced the expression of has_circ_0003091 in the lung and lung ECs tissues of ALI mice.

Figure 2. Expression of mmu_circ_0015268 in LPS-stimulated lung tissues and PMVECs. ALI model was evaluated by histological analysis (A, B), hepatic-injury marker (ALT activity) (C) and renal injury marker (creatinine concentration) (D). (E) Expression of mmu_circ_0015268 in whole lungs after LPS treatment. (F) Expression of mmu_circ_0015268 in primary lung microvascular endothelial cells (PMLECs) stimulated with LPS. **P < 0.01.

Upregulation of has_circ_0003091 contributes to lung inflammation, endothelial dysfunction, and vascular injury

Increased expression of has_circ_0003091 was associated with increased production of inflammatory cytokines in lung tissues, including IL-1β, IL-6, TNF-α (Figure 3A), and BAF (Figure 3B). The H&E and TUNEL stainings showed a higher pathological injury (Figure 3C, 3E) and apoptotic ratio (Figure 3D, 3F) of lung tissues in LPS-induced mice. Moreover, these changes were accelerated by has_circ_0003091 overexpression and alleviated by has_circ_0003091 suppression (Figure 3A–3F). Consistently, the protein levels of endothelial-specific adhesion marker E-selectin, ICAM1 and VCAM1, and apoptotic markers were highly induced by LPS in lung tissues (Figure 4A and Supplementary Figure 1). Further, significantly reduced levels of endothelial AJs proteins p120-catenin, β-catenin, and E-cadherin were observed in LPS-induced mice (Figure 4A), suggesting enhanced pulmonary vascular injury.

Figure 3. mmu_circ_0015268 contributed to lung inflammation and vascular injury. Adult C57BL/6 mice were randomly assigned into 6 groups: sham, Model, Model + empty vector (EV), Model + mmu_circ_0015268 overexpressing vector (OV), Model + negative control (NC), Model + mmu_circ_0015268 shRNA (shRNA). ELISA was used to detect the contents of TNF-α, IL-6, and IL-1β in whole lung tissues (A) and BAL fluid (B). (C) Paraffin-embedded lung tissue samples were stained for H&E (magnification 200 X). (D) TUNEL assay of lung tissue sections for evaluation of apoptosis (magnification 200 X). (E, F) Quantification of lung injury scores and apoptosis. *P < 0.05; **P < 0.01 vs. sham; #P < 0.05, vs. model.

Figure 4. Upregulation of mmu_circ_0015268 promoted endothelial dysfunction in vivo and in vitro. (A) Representative blot results showing the levels of BAX, BCL2, Caspase3, cleaved PARP1, E-selectin, ICAM1, VCAM1, P120-catenin, β-catenin, and E-cadherin in lung tissues. (B) PMLECs were transfected with mmu_circ_0015268 overexpressing vector (OV) or mmu_circ_0015268 shRNA (shRNA), followed by LPS stimulation. Representative blot results showing the levels of BAX, BCL2, Caspase3, cleaved PARP1, E-selectin, intercellular cell adhesion molecule-1 (ICAM1), vascular cell adhesion protein 1 (VCAM1), P120-catenin, β-catenin, and E-cadherin in PMLECs. (C, D) Immunofluorescence intensities of ICAM1 and VCAM1 in vivo and in vitro for different groups.

To verify the role of mmu_circ_0015268 on endothelial dysfunction, we transfected PMVECs with mmu_circ_0015268 overexpressing plasmids or mmu_circ_0015268 shRNA. Increased levels of ICAM1, VCAM1, and E-selectin were observed after has_circ_0003091 overexpression, suggesting aggravation of LPS-caused endothelial activation. Consistently, compared to LPS-treated control PMVECs, the LPS-treated PMVECs with has_circ_0003091 overexpression presented significantly reduced levels of AJs proteins p120-catenin, β-catenin, and E-cadherin, which was tightly associated with increased endothelial activation. On the other hand, mmu_circ_0015268 knockdown alleviated these damages (Figure 4B and Supplementary Figure 2). The quantitative evaluations of each protein are presented in the Supplementary Figures. Moreover, we applied immunofluorescence to detect the expression of ICAM1 and VCAM1 in LPS-treated animals and primarily cultured cells. Consistently, the intensities of ICAM1 and VCAM1 were enhanced in the model group, which was further enhanced by mmu_circ_0015268 overexpression and suppressed by mmu_circ_0015268 knockdown (Figure 4C, 4D). Collectively, these results indicated that mmu_circ_0015268 can promote lung injury responses in mice.

mmu_circ_0015268 negatively regulates the expression of miR-149

After bioinformatics analyses, miR-149 was identified as a potential target of has_circ_0003091 (Figure 5A). Compared to the mutant has_circ_0003091 cells, we observed significantly reduced luciferase activity in the miR-149 and WT has_circ_0003091 co-transfected cells (Figure 5B), suggesting the direct interaction between has_circ_0003091 and miR-149. Then, we examined the expression of miR-149 in lungs and PMVECs under ALI. The expression of miR-149 was significantly decreased in pathogenic lung tissues and LPS-stimulated PMVECs (Figure 5C). Compared to controls, has_circ_0003091 overexpression significantly impaired miR-149 expression in the mice’s lung tissues and isolated ECs (Figure 5C). Additionally, a negative correlation between miR-149 and has_circ_0003091 expression was observed in PMVECs (Figure 5D), consistent with the above results. Overall, these results suggested that has_circ_0003091 negatively regulates the expression of miR-149.

Figure 5. mmu_circ_0015268 negatively regulated miR-149 expression. (A) Putative binding site between has_circ_0003091 and miR-149. (B) Representative bar graphs showing the luciferase activity in the cells after indicated treatments. (C) Expression of mmu_circ_0015268 in lungs and isolated ECs. (D) The PMLECs were transfected with mmu_circ_0015268 overexpressing vector (OV) or mmu_circ_0015268 shRNA (shRNA), followed by LPS stimulation. The expression of mmu_circ_0015268 was measured in different groups. **p < 0.01; ***p < 0.001 vs. sham; ^#p < 0.05 vs. model.

Smad2 is a direct downstream target of miR-149

The putative binding site between miR-149 and Smad2 was found using the TargetScan database and indicated that has_circ_0003091 exerted its regulatory function by sponging miR-149 and subsequently affecting the Smad2 expression in sepsis-induced ALI (Figure 6A). Additionally, after miR-149 mimic transfection, only Smad2-WT cells exhibited significantly reduced luciferase activity. No luciferase activity changes were observed in the Smad2-MUT cells, suggesting direct Smad2 targeting by miR-140 (Figure 6B). In the animal model, the levels of Smad2 in the lung tissues and isolated ECs were elevated (Figure 6C). Moreover, mmu_circ_0015268 could positively regulate Smad2 expression (Figure 6D). By transfecting PMVECs with miR-149 mimics or inhibitors, the mRNA and protein levels of Smad2 were significantly decreased or elevated, respectively (Figure 6E, 6F). These findings revealed that Smad2 was a downstream target of miR-149.

Figure 6. Smad2 is a direct downstream target of miR-149. (A) Putative binding site between Smad2 and miR-149. (B) Representative bar graphs showing the luciferase activity in the cells after indicated treatments. (C) Expression of Smad2 in lungs and isolated ECs. (D) The PMLECs were transfected with mmu_circ_0015268 overexpressing vector (OV) or mmu_circ_0015268 shRNA (shRNA), followed by LPS stimulation. Expression of Smad2 in different groups was measured. (E, F) The mRNA and protein levels of Smad2 in differentially treated PMVECs. **p < 0.01; ***p < 0.001 vs. sham; ^#p < 0.05 vs. model.

mmu_circ_0015268 inhibits LPS-induced lung injury via the miR-149/Smad2 axis

The expression of Smad2 was significantly enhanced by has_circ_0003091 overexpression and impaired by miR-149 overexpression in LPS-stimulated PMVECs. Moreover, the elevated Smad2 expression was suppressed by miR-149 mimics (Figure 7A), suggesting that has_circ_0003091 exerted its function by regulating the miR-149/Smad2 axis. Furthermore, the Western blot analysis showed that mmu_circ_0015268-induced apoptotic markers and endothelial activation were effectively restored by miR-149 overexpression, including BAX, BCL2, Caspase3, cleaved PARP1, E-selectin, ICAM1, VCAM1, P120-catenin, β-catenin, and E-cadherin in PMVECs (Figure 7B and Supplementary Figure 3). These results suggested that mmu_circ_0015268 promoted LPS-stimulated apoptosis and inflammation via the miR-149/Smad2 axis in PMVECs.

Figure 7. mmu_circ_0015268 inhibited LPS-induced lung injury via the miR-149/Smad2 axis. The PMLECs were transfected with mmu_circ_0015268-expressing vector and miR-149 mimics alone or combined, followed by LPS stimulation. (A) Expression of Smad2 in different groups. (B) Western blot of BAX, BCL2, Caspase3, cleaved PARP1, E-selectin, intercellular cell adhesion molecule-1 (ICAM1), vascular cell adhesion protein 1 (VCAM1), P120-catenin, β-catenin, and E-cadherin in different groups. **p < 0.01; ***p < 0.001.

Clinical samples

Blood samples were collected from three sepsis patients and two healthy subjects from the China-Japan Friendship Hospital following the Helsinki Declaration. Table 1 shows the demographic characteristics of septic patients. All samples were frozen in liquid nitrogen and stably stored at − 80° C until RNA extraction and sequencing at the Beijing Liuhe Huada gene technology company. The protocols and procedures of this study were strictly reviewed and approved by the Ethics Committee of the China-Japan Friendship Hospital. All subjects enrolled signed the informed consent.

RNA sequencing

The PAXgene Blood RNA Kit (QIAGEN) was used to extract total RNA from clinical blood samples. A Nanodrop spectrophotometer (Thermo Fisher) was used to determine RNA concentration. Total RNA extraction and library construction were performed by the Annoroad Gene Technology Co., Ltd. (Beijing, China). The R software was used for subsequent data processing. The expression profile of circRNAs was screened based on the fold-change (FC) (FC > 2.0) and p-value (p < 0.05) thresholds. The genes related to the circRNAs screened were analyzed using GO and KEGG enrichment.

Construction of sepsis-induced ALI model

First, C57BL/6 mice (10 weeks, 25-30 g) were randomly divided into 6 groups: Sham, Model, Model + empty vector (EV), Model + mmu_circ_0015268 overexpressing vector (OV), Model + negative control (NC), Model + mmu_circ_0015268 shRNA (shRNA). All mice were housed under a 12:12 h light/dark photocycle. LPS (0.5 mg/kg body weight) (Sigma) was used to induce sepsis via intravenous (i.v.) injection. An equal volume of sterile normal saline was used as vehicle control. Adenoviral vectors containing mmu_circ_0015268 cDNA or shRNA were packed into adenovirus particles, and i.v. injected one week before ALI induction. The animal study protocols were approved by the Institutional Animal Care and Use Committee of our hospital.

Isolation of lung endothelial cells

After digestion with a mouse lung dissociation kit (Miltenyi Biotec), CD45+ immune cells were removed using the obtained CD45 single-cells conjugated microbeads. Then, CD31 conjugated microbeads were used to select endothelial cells. Finally, a suspension with CD45-negative and CD31-positive was selected.

Cell culture

PMVECs were cultured using DMEM and transfected with mmu_circ_0015268 or mmu_circ_0015268 shRNA adenoviral vectors. After 24 h of transfection, cells were treated with LPS (100 ng/mL). Gene analysis was carried out 6 h after transfection.

Enzyme-linked immunosorbent assay (ELISA)

Tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and interleukin-1β (IL-1β) were quantified using ELISA kits (R&D Systems). After sample addition and 2 h of incubation with detection antibody, plates were incubated with a substrate solution and the optical absorbance at 450 nm was measured using a plate reader (Molecular Devices Corp, Menlo Park, CA).

qRT-PCR

First, Total RNA was isolated from lung tissues or cells and reversely transcribed into cDNA. The expression of screened circRNAs was validated using qRT-PCR. Primers used in this study are shown in Table 2.

Western blot analysis

Lysis Buffer and BCA Protein Assay Kit from Thermo Fisher were employed to extract total protein and determine their concentrations, respectively. After separation using 10% SDS-PAGE, protein bands were transferred to the PVDF membrane. After 1 h of blocking with 5% non-fat milk, samples were incubated overnight with primary antibodies against BAX (1:1000, Abcam, Boston, MA, USA), BCL2 (1:1000, Abcam, Boston, MA, USA), Caspase3 (1:1000, Abcam, Boston, MA, USA), cleaved PARP1 (1:1000, Abcam, Boston, MA, USA), E-selectin (1:1000, Abcam, Boston, MA, USA), ICAM1 (1:1000, Abcam, Boston, MA, USA), VCAM1 (1:1000, Abcam, Boston, MA, USA), P120-catenin (1:1000, Abcam, Boston, MA, USA), β-catenin (1:1000, Abcam, Boston, MA, USA), and E-cadherin (1:1000, Abcam, Boston, MA, USA) at 4° C. Finally, samples were incubated for 1 h with HRP-conjugated secondary antibodies and the protein bands were analyzed.

Hematoxylin and eosin (H&E) staining

Changes in lung tissue morphology were determined using H&E staining. Briefly, lung tissue sections were stained with H&E following a routine protocol, and results were observed and recorded at 200 x magnification.

TUNEL assay

The TUNEL assay (Promega, Madison, WI, USA) was used to determine cell apoptosis. Apoptotic cells were measured under a fluorescence microscope (200 x).

Immunofluorescence

Briefly, tissue sections or PMVECs were washed with PBS, fixed with 4% polyformaldehyde, and penetrated with 0.5% Triton X-100. After 30 min of blocking with 5% BSA, sections were incubated with primary antibodies against ICAM1 and VCAM1 (Abcam) at 4° C overnight, followed by 30 min of incubation at 37° C with Alexa Fluor 488 and Alexa Fluor 555-labeled IgG. Then, DAPI staining was conducted for cell nuclei detection, and a light microscope was used to visualize the results.

Dual-luciferase reporter assay

Lipofectamine 2000 was used to co-transfect HEK-293T cells with the luciferase report vector containing the 3’ UTR of WT-has_circ_0003091 or MUT- has_circ_0003091 and miR-149 agomir or agomir NC. After 24 h, the Luciferase Assay Reporter System (Promega) was used to evaluate the luciferase activity.

Statistical analyses

The data were analyzed using GraphPad Prism 8.0 and are presented as means ± standard deviations (SDs). The data were compared using Student's t-test or ANOVA. A p < 0.05 was defined as a significant difference between indicated groups.