Study population and clinical characteristics
A total of 903 sepsis patients (mean age: 60.9±17.1 years; 31.7% female) and 1,150 healthy individuals (mean age: 48.1±15.2 years; 38.1% female) were enrolled in this study between February 2013 and June 2018. The Consolidated Standards of Reporting Trials (CONSORT) flowchart of this study is shown in Additional file 3. The clinical characteristics of the enrolled subjects (903 sepsis patients) are presented in Table 1. The 903 septic cases encompassed 181 patients with sepsis subtypes (20.0%), 368 patients with severe sepsis (40.8%), and 354 patients with septic shock (39.2%). The most common source of infection was respiratory tract infection (512/903, 56.7%), followed by Abdominal infection (112/903, 12.4%). The main pathogens identified in this study were Acinetobacterbaumannii (237/903, 26.2%), P. aeruginosa (109/903, 12.1%), Escherichia coli (97/903, 10.7%), Staphylococcus aureus (79/903, 8.7%), and Klebsiella pneumoniae (61/903, 6.8%). Gram-positive and gram-negative infections accounted for 12.1% (109/903) and 32.8% (296/903) of all the infections, respectively, and patients with polymicrobial infections comprised 12.1% (109/903) of all the patients.
Effect of ADAM17 genetic variations on susceptibility to sepsis development
A total of 903 sepsis cases were separated according to the severity of sepsis into three subgroups, namely the sepsis subtype, severe sepsis and septic shock subgroups, to evaluate the association of the ADAM17 -172A>G polymorphism with sepsis progression (Table 2). Our data showed that the frequencies of the AG/GG genotypes and G allele of rs12692386 in the sepsis subtype subgroup significantly differed from those in the severe sepsis subgroup (29.3% vs. 45.9%, P=0.0005 for genotype; 17.1% vs. 25.8%, P=0.0039 for allele) and septic shock subgroup (29.3% vs. 42.4%, P=0.0032 for genotype; 17.1% vs. 23.3%, P=0.0193 for allele), which suggests a functional role for the AG/GG genotype in promoting the progression of sepsis from sepsis subtype to severe sepsis and septic shock. Additionally, our data showed no significant differences in genotype/allele frequencies of ADAM17 -172A>G polymorphism between patients and healthy controls (all P>0.05), which suggests that this SNP might not influence susceptibility to sepsis (Table 3).
The 28-day ICU mortality was also analyzed to evaluate the effect of this ADAM17 SNP on the clinical outcome of sepsis patients. As presented in Table 4, the frequencies of non-survivors among the sepsis patients with the -172AG/GG genotypes and G allele were distinctly higher than those among patients with the AA genotype (50.7% vs. 39.3%, OR=0.630, 95% CI=0.444-0.894, P=0.0188) and A allele (29.3 % vs. 21.8 %, OR = 1.482, 95 % CI = 1.125-1.952, P= 0.0188). In addition, the Kaplan-Meier survival analysis indicated that the 28-day survival in septic patients with -172AG/GG genotypes was much worse than in the AA genotype carriers (log-rank test 4.696, P=0.030) (see Additional file 4).
Effect of ADAM17 -172A>G polymorphism on the expression of ADAM17 and proinflammatory cytokines
We divided 80 randomly selected sepsis cases into subgroups based on organ failure and the patients’ 28-day mortality, and statistically significant differences in ADAM17 expression were observed between the multiple organ dysfunction (MODS) and non-MODS subgroups (Fig. 1a). ADAM17 expression was significantly decreased in the sepsis patients who survived during the 28-day period compared with the patients who did not survive (Fig. 1a). To confirm the impact of ADAM17 rs12692386 polymorphisms on ADAM17 expression in vitro, PBMCs were isolated from 18 randomly selected healthy individuals and stimulated with LPS (800 ng/mL) for 6 hours in vitro. ADAM17 expression was significantly increased in the LPS-stimulated PBMCs compared with the control PBMCs (Fig. 1b). Additionally, we analyzed the secretion of cytokines in the cell culture medium and found significantly higher concentrations of TNF-α, IL-1β and IL-6 in the culture medium from LPS-stimulated PBMCs than in that from control PBMCs. Furthermore, PBMCs with the -172AG/GG genotypes exhibited significantly higher expression levels of ADAM17 than those carrying the AA genotype upon LPS stimulation (Fig. 1c). Moreover, we detected the impact of this ADAM17 genetic variation on the expression levels of cytokines. The concentrations of TNF-α (350.5 ± 37.6 pg/mL, P=0.0009) and IL-6 (958.6 ± 84.0 pg/ml, P=0.013) in LPS-stimulated PBMCs with the AG/GG genotypes were clearly higher than those with the AA genotype (149.9 ± 65.5 pg/ml and 625.0 ± 300.4 pg/ml, respectively) (Fig. 1c).
EGR1 expression is upregulated in sepsis and is associated with ADAM17
Because the -172A>G polymorphism is located in the promoter region of ADAM17 gene, we speculated that the A-to-G nucleotide exchange influences the binding affinity of transcription factors to this site and thereby contributes to the modulation of ADAM17 gene expression. A bioinformatics analysis predicted that three transcription factors, namely, EGR1, specificity protein 1 (SP1) and TFAP2A (AP2α), would bind to the -172A>G promoter region in an allele-specific manner (Fig. 2a). To investigate whether these transcription factors are involved in sepsis, we detected their expression in LPS-stimulated HUVECs and RAW264.7 cells and found that the expression of EGR1, but not SP1 and AP2α, was significantly increased in LPS-pretreated cells (Fig. 2b-d). Moreover, in agreement with the trend found for ADAM17 expression in these two cell lines, LPS stimulation of RAW 264.7 cells notably increased EGR1 expression to a significant level at 0.5 hours and to a peak level at 1 hour, and the peak time was delayed in HUVECs (Fig. 2e). Importantly, EGR1 mRNA expression was obviously higher in sepsis patients than in healthy controls (Fig. 2f). We further analyzed PBMCs isolated from healthy controls and found that the LPS-stimulated PBMCs showed significantly higher expression of EGR1 than the control PBMCs (Fig. 2g).
ADAM17 -172A>G polymorphism upregulates the ADAM17 gene expression via enhancing the binding affinity of its promoter region with the transcription factor EGR1
Because the -172A>G polymorphism is located in a potential EGR1-binding sequence in the ADAM17 promoter region, ChIP-qPCR and luciferase reporter assays were performed to determine the binding of EGR1 to the -172 position in the ADAM17 promoter region and to distinguish the differences in binding capacity between the rs12692386 A and G alleles to EGR1. ChIP-qPCR conducted with HUVECs indicated that EGR1 bound to the promoter region of the ADAM17 gene (Fig. 3a,b). For the luciferase reporter assays, a 6307-bp ADAM17 promoter- luciferase reporter and the corresponding reporter with a single point mutation (A-to-G) on the ADAM17 rs12692386 site were constructed, and these constructs were cotransfected into HEK293T cells with pCMV-EGR1 (Fig. 3c). As shown in Fig. 3d, EGR1-overexpressing HEK293T cells transfected with pGL3-G exhibited significantly higher luciferase activities than cells transfected with pGL3-A, which indicates that ADAM17 -172A>G contributes to the increased binding affinity of EGR1 to the ADAM17 promoter. We further investigated the impact of EGR1 on the regulation of ADAM17 expression in an LPS-induced cell model of sepsis. The construction and efficiency of infection of EGR1 overexpressed and silent adenovirus were presented (see Additional file 5). EGR1-overexpressing or EGR1-silenced virus successfully infected RAW264.7 cells and HUVECs (Fig. 3e). The presence of EGR1 significantly increased ADAM17 expression in LPS-induced cells, whereas these effects were completely abrogated by EGR1 knockdown (Fig. 3f). These findings confirm a positive role of EGR1 in the modulation of ADAM17 expression in sepsis.
Effects of the EGR1/ADAM17 signaling pathway on proinflammatory cytokine secretion and apoptosis in vitro
The concentrations of TNF-α, IL-1β and IL-6 in and the apoptosis rate of LPS-pretreated cells were then analyzed to confirm the impact of the EGR1/ADAM17 signaling pathway on sepsis. Our results showed that EGR1-silenced RAW264.7 cells displayed significantly lower expression levels of TNF-α (31.5 ± 3.1 pg/mL, P=0.029) and IL-6 (144.5 ± 33.1 pg/mL, P=0.029) compared with the control cells (47.5 ± 2.1 pg/mL and 417.3 ± 1.4 pg/mL, respectively) upon LPS stimulation. Decreased expression of TNF-α(17.3 ± 1.7 pg/mL, P=0.029) and IL-1β (4.4 ± 2.9 pg/mL, P=0.029) was also observed in LPS-stimulated EGR1-silenced HUVECs compared with those in LPS-stimulated cells (30.7 ± 4.9 pg/mL and 15.7 ± 2.9 pg/mL, respectively) (Fig. 4a). In addition, the silencing of EGR1 through EGR1-RNAi infection significantly decreased the apoptotic rate of RAW264.7 cells upon LPS stimulation (Fig. 4b).
EGR1 inhibition reduces the sepsis-induced inflammatory responses and improves the survival rate of endotoxemic mice
Given the important role of EGR1 in the upregulation of ADAM17 expression in LPS-induced cells, we investigated whether EGR1 silencing is involved in the attenuation of the inflammatory response in vivo. ADAM17 gene expression was decreased in the lung and liver tissues of As-ODN-pretreated endotoxemia mice compared with those of LPS-treated control mice, and this effect was accompanied by the downregulation of endothelial cell injury factors, including ICAM1 and VCAM1 (Fig. 5b, c). In addition, As-ODN administration afforded mice a significant survival benefit following sepsis for 72 hours (Fig. 5d). Plasma levels of IL-1β (68.4 ± 16.3 pg/mL, P=0.026), IL-6 (761.5 ± 775.0 pg/ml, P=0.015), and TNF-α (199.5 ± 159.7 pg/ml, P=0.0411) in As-ODN-pretreated endotoxemia mice were significantly decreased compared with those in the LPS-treated control mice (102.50 ± 28.68 pg/ml, 2914.0 ± 1540.0 pg/ml, 458.8 ± 147.5 pg/ml, respectively) at 6 hours post-LPS injection (Fig. 5e).