Abstract
Introduction: Septic shock, a severe manifestation of infection-induced systemic immune response, poses a critical threat resulting in life-threatening multi-organ failure. Early diagnosis and intervention are imperative due to the potential for irreversible organ damage. However, specific and sensitive detection tools for the diagnosis of septic shock are still lacking. Methods: Gene expression files of early septic shock were obtained from the Gene Expression Omnibus (GEO) database. CIBERSORT analysis was used to evaluate immune cell infiltration. Genes related to immunity and disease progression were identified using weighted gene co-expression network analysis (WGCNA), followed by enrichment analysis. CytoHubba was then employed to identify hub genes, and their relationships with immune cells were explored through correlation analysis. Blood samples from healthy controls and patients with early septic shock were collected to validate the expression of hub genes, and an external dataset was used to validate their diagnostic efficacy. Results: Twelve immune cells showed significant infiltration differences in early septic shock compared to control, such as neutrophils, M0 macrophages, and natural killer cells. The identified immune and disease-related genes were mainly enriched in immune, cell signaling, and metabolism pathways. In addition, six hub genes were identified (PECAM1, F11R, ITGAL, ICAM3, HK3, and MCEMP1), all significantly associated with M0 macrophages and exhibiting an area under curve of over 0.7. These genes exhibited abnormal expression in patients with early septic shock. External datasets and real-time qPCR validation supported the robustness of these findings. Conclusion: Six immune-related hub genes may be potential biomarkers for early septic shock.
Plain Language Summary
Septic shock is a dangerous condition that happens when an infection spreads through the body and triggers a strong immune response, leading to the failure of multiple organs. Recognizing and treating septic shock quickly is crucial to prevent lasting damage to the body’s organs. However, doctors currently do not have highly effective tools to diagnose septic shock early. In this study, we looked at genetic information from patients with early septic shock. We used a large public database to find patterns in gene activity that could help identify the condition. By analyzing the genes, we could tell which types of immune cells were involved. We discovered that certain immune cells, like neutrophils, M0 macrophages (a kind of white blood cell that helps fight infections), and natural killer cells, were more active in patients with septic shock. We also found genes that are active during immune responses and disease progression. These genes were mostly involved in the body’s defense system, cell communication, and energy use. Among these genes, six stood out as being closely connected to M0 macrophages. These six genes could potentially serve as early warning signs for doctors to detect septic shock, as they were good at distinguishing between patients with and without the condition. In summary, the study identified six genes that might be useful for spotting septic shock early on. These findings could lead to better diagnostic tools, helping doctors to treat patients before their condition becomes critical.
Introduction
Sepsis is a disease with high morbidity and mortality, characterized by a dysregulated response to infection and resulting in multi-organ dysfunction [1]. Septic shock is a severe complication of sepsis that manifests as exacerbating the immune response and significantly heightens the risk of death due to circulatory, cellular, and metabolic abnormalities [2, 3]. Early intervention and treatment of septic shock are important to minimize the pathophysiologic response and improve patient survival [4]. However, specific and sensitive detection tools for the diagnosis of septic shock are still lacking.
The pathophysiologic process of septic shock is closely related to the dysregulation of the immune system [5]. The intricate interactions of the immune system involve a wide variety of immune cells, and the inflammatory response triggered by infection may lead to dynamic changes in immune cells [6]. Over-activation of some immune cells and the release of large amounts of cytokines contribute to vasodilatation and microcirculatory disorders, triggering a systemic immune response that may ultimately lead to organ failure and shock, posing a life-threatening risk to patients [7, 8]. The distinct pattern of immune cell infiltration observed in the early stages of septic shock emphasizes the intimate connection between immunity and disease progression [9]. Therefore, it is important to understand the immune cell infiltration status for early diagnosis and intervention in septic shock.
Weighted gene co-expression network analysis (WGCNA) can effectively integrate gene expression and clinical characterization data to evaluate functional pathways and identify molecular biomarkers [10]. WGCNA has been proven to be valuable in studying human sepsis, autoimmune diseases, and various cancers [11, 12]. In addition, CIBERSORT could quantify the cellular composition of immune cells and has been successfully used to estimate the level of immune cell infiltration in various diseases [13]. In this study, we aimed to explore the relationship between early septic shock and immune infiltration using public microarray datasets and identify hub genes associated with early septic shock and immune infiltration through WGCNA and protein-protein interaction (PPI) networks, providing valuable insights into its pathogenesis. We hope our study has the potential to contribute to the development of novel molecular markers for the early diagnosis and intervention in septic shock.
Materials and Methods
Data Collection and Processing
In this study, the analyzed data were obtained from the Gene Expression Omnibus (GEO) database and processed using R4.0.5 (https://www.r-project.org/). Dataset GSE57065 was downloaded from the GEO database, including 82 blood samples from 28 patients with septic shock (28 ICU patients enrolled at the onset of septic shock, with blood samples collected at 30 min [H0], 24 h [H24], and 48 h [H48] after the onset of septic shock) and 25 blood samples from healthy controls (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000540949). Gene expression matrix files were obtained using the platform annotation file GPL570 and “AnnoProbe” package in R (https://github.com/ableno/AnnoProbe). Additionally, dataset GSE115736 was downloaded to obtain gene expression data from different types of leukocytes derived from adult donor peripheral blood mononuclear cells, including CD4+ T cells, CD8+ T cells, eosinophils, memory B cells, monocytes, myeloid DCs, myeloid DC CD123, naive B cells, neutrophils, NK cells, nonclassical monocytes, and plasmacytoid DCs (ratio = 5:5:2:3:5:3:2:5:3:5:2:2). Dataset GSE154918 included blood samples from healthy controls, patients with uncomplicated infections, sepsis patients, septic shock patients, sepsis follow-up patients, and septic shock follow-up patients. We selected 40 blood samples from controls and 19 blood samples from septic shock patients in dataset GSE154918.
Immune Infiltration Analysis
CIBERSORT was used to calculate the score of infiltrating immune cells in patients with septic shock. The ESTIMATE software was then used to evaluate three scores: the stromal score, the immune score, and the ESTIMATE score. The results were visualized using the R package, ggplot2, pheatmap, ggpubr, and ggcorrplot.
WGCNA
The SAPS II score is important for prognostic prediction in ICU patients and could often be used to assess disease severity [14]. A scale-free gene co-expression network was constructed using the R package “WGCNA” to analyze genes with the top 25% coefficients of variation in 82 disease samples. The “hclust” was performed to cluster sample data according to time, SAPS II, ESTIMATE, immune, and stromal scores. The “pickSoftThreshold” was performed to select a soft threshold (β = 9) to construct a scale-free topology. Then, the adjacency matrix was calculated and transformed into the topological overlap matrix (TOM) and the corresponding dissimilarity matrix (1-TOM), and the genes were clustered using the average chained hierarchical clustering method. According to the criteria of the hybrid dynamic shear tree method, the minimum number of genes was set to 30, and the height of the cut tree graph was set to 0.3. Then, the signature genes of each module were calculated sequentially, and the modules were analyzed by clustering.
Hub Module Screening
The “Pearson” method was used to analyze the correlation between modules and immune scores and disease progression to screen the hub modules. Immune score or disease-related genes were selected from hub modules with gene significance >0.2 and module membership >0.8, respectively, and intersections were taken to obtain core genes.
Enrichment Analysis
To further explore the biological functions of core genes, we performed gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of core genes using the Genecodis 4.0 database (https://genecodis.genyo.es/) with p value <0.05.
Identification of Hub Genes
PPI networks were constructed using the STRING database (https://string-db.org/) for core genes. The MCC, DMNC, Degree, Closeness, and EPC algorithms in CytoHubba, a plugin of Cytoscape, were used to obtain hub genes (the top 10 genes for each algorithm). The expression levels of hub genes in the disease samples and controls were analyzed using rank sum tests and presented in box plots. Additionally, the receiver operating characteristic (ROC) curve was performed to characterize the diagnostic value of the hub genes for 28 samples of septic shock (H0) and 25 controls using pROC. The diagnostic value of the hub genes was verified in dataset GSE154918 using pROC. The correlations between the hub genes and the differential immune cells were calculated using the “Pearson” method. Moreover, the dataset GSE115736 was used to show the expression of hub genes in different immune cell types.
Patient Samples and Protocols
We obtained 13 blood samples from healthy controls (n = 13) and 33 blood samples from patients with septic shock (n = 11), collected at 30 min (0 h), 24 h, and 48 h after the onset of septic shock. This study was approved by the Ethics Committee of the Second Affiliated Hospital of Shandong First Medical University (2022-094). Written informed consent was obtained from all patients. Detailed patient characteristics are displayed in online supplementary Table S2. Total RNA was extracted from these blood samples using the Blood RNA Extraction Kit (Magen, China) and reversed to cDNA using the FastKing cDNA First Strand Synthesis Kit (TIANGEN, China). Subsequently, real-time qPCR was performed using SuperReal PreMix Plus (SYBR Green, TIANGEN) to validate the expression levels of hub genes. GAPDH and ACTB served as internal reference genes. The primers were synthesized at Biomed (Beijing, China), and the specific sequences are presented in online supplementary Table S3. Data were relatively quantified using the 2−ΔΔct method. Group differences were assessed using a one-way ANOVA.
Drug Prediction
The DGIdb database (version: 3.0.3, http://www.dgidb.org/) was utilized to predict potential drugs for the treatment of septic shock using hub genes as drug targets.
Results
Immune Cell Analysis
The immune cell status of early septic shock samples and controls was evaluated using CIBERSORT. The percentage composition of immune cells in each sample is shown in Figure 1a, revealing the highest abundance of neutrophils and monocytes. Furthermore, significant differences in the levels of 12 immune cells were observed between early septic shock samples and controls. For instance, the levels of neutrophils and M0 macrophages were increased and the levels of CD8+ T cells and NK cells resting were decreased in early septic shock samples compared to controls (Fig. 1b). Notably, there was a significant decrease in CD4 naïve T cells and CD4+ T cells resting and a significant increase in activated CD4+ T cells (Fig. 1b). Additionally, ESTIMATE and immune scores were significantly decreased in the early septic shock samples (H0) compared to controls, while stromal scores were significantly increased (Fig. 1c). Taken together, the infiltration levels of immune cells were significantly changed during the development of early septic shock.
Identification of Hub Modules Associated with Immunity and Disease Progression
Genes in early septic shock samples were clustered using WGCNA according to time, SAPS II, and immune scores (Fig. 2a). β = 9 was chosen as the soft threshold to construct the scale-free network (Fig. 2b). Finally, 14 modules were identified through dynamic tree-cutting and module merging (Fig. 2c, d). Among them, the black module had the highest positive correlation with time to septic shock (p = 2e-05) and the blue module had the highest negative correlation with time to septic shock (p = 0.04), while both modules had a significant correlation with the immune score (Fig. 2e). Therefore, the black and blue modules were selected as hub modules for subsequent analysis.
Identification of Core Genes and Enrichment Analysis
In the black module, we identified 104 genes related to immune score and 103 genes related to onset time (Fig. 3a). Similarly, in the blue module, we identified 150 genes related to immune score and 63 genes related to onset time (Fig. 3b). The above genes were united to obtain 166 genes related to onset time and 254 genes related to immune cells, respectively. Finally, 164 core genes were identified through overlapping both sets and may be crucial in the development of early septic shock (Fig. 3c). GO analysis revealed that core genes were mainly distributed in the cytoplasm, Golgi apparatus, and lysosome. They mainly exhibited metabolism-related molecular functions, such as transferase activity and mannose binding. Additionally, they were involved in biological processes such as immunity, inflammation, cellular signaling, and metabolism, including cell migration and IMP salvage regulation (Fig. 3d). Similarly, KEGG analysis showed that core genes were mainly enriched in hypoxia-inducible factor (HIF)-1, Fc epsilon RI, and metabolic pathways (Fig. 3e).
Identification of Hub Genes and Expression Profile Analysis
A PPI network of 164 core genes was constructed using the STRING database (Fig. 4a). Subsequently, 6 hub genes were identified using Cytoscape based on the PPI network, including platelet endothelial cell adhesion molecule-1 (PECAM1), F11 receptor (F11R), integrin α-L (ITGAL), intercellular adhesion molecule 3 (ICAM3), hexokinase 3 (HK3), and MCEMP1 (Fig. 4b). The expression levels of hub genes exhibit significant differences between controls and early septic shock samples (Fig. 5a). There appeared to have a tendency for the expression of these genes to revert to the normal controls over time. These results suggested that hub genes were significantly dysregulated during the onset of septic shock but that it is a restorative process within 48 h of the patient’s admission to the ICU. Additionally, the expression levels of hub genes were validated using real-time qPCR (Fig. 5b; online suppl. Table S4 for raw data). The expression levels of PECAM1, ITGAL, and MCEMP1 showed significant differences between controls and early septic shock samples, particularly at 30 min (0 h) after the patient’s admission to the ICU. Although there were no significant differences in the expression of ICAM3, HK3, and F11R, these genes exhibited expression trends that were generally consistent with the bioinformatics analysis.
Hub Genes and Immune Cells
The correlation between the six hub genes and differential immune cells was calculated using the “Pearson” method (Fig. 6a). The results showed that each hub gene was associated with multiple immune cells. Notably, the hub genes were all strongly correlated with M0 macrophages, with HK3 and MCEMP1 significantly positively correlated with M0 macrophages, PECAM1, F11R, ITGAL, and ICAM3 significantly negatively correlated with M0 macrophages (Fig. 6b). Additionally, the expression of hub genes in different immune cell types was investigated using the dataset GSE115736 (Fig. 7). The results revealed higher expression of F11R in eosinophils, HK3 in nonclassical monocytes (followed by monocytes and eosinophils), ICAM3 in neutrophils, ITGAL in NK cells (followed by nonclassical monocytes and CD8+ T cells), MCEMP1 in neutrophils (followed by monocytes), and PECAM1 in nonclassical monocytes (followed by monocytes and neutrophils).
Diagnostic Efficacy of Hub Genes in Septic Shock
The ROC curve analysis was performed to evaluate the diagnostic value of the hub genes. The results revealed AUC values of 0.921, 0.999, 0.743, 0.841, 1, and 0.849 for F11R, HK3, ICAM3, ITGAL, MCEMP1, and PECAM1, respectively (Fig. 8a). Furthermore, the diagnostic value of the hub genes was verified in the dataset GSE154918. The AUC values for F11R, HK3, ICAM3, ITGAL, MCEMP1, and PECAM1 were 0.7, 1, 0.85, 0.722, 1, and 0.686, respectively (Fig. 8b). These results implied the superior ability of hub genes in the diagnosis of septic shock.
Drug Prediction
The results showed that three of six hub genes, ICAM3, ITGAL, and PECAM1, were found to be targeted by 20 drugs in the DGIdb database (Fig. 9). Among these, ITGAL was targeted by the most drugs, such as mycophenolate mofetil and staurosporine.
Discussion
Despite certain advancements in septic shock treatment, this disease remains a significant public health challenge with high morbidity and mortality rates [15]. This is mainly attributed to irreversible injury to vital organs and the development of complications [16]. The pathophysiological role of inflammation and immunity is considered crucial in the development of septic shock [17]. The field is under extensive research aimed at identifying immune-related biomarkers for early intervention [18, 19]. In the present study, we identified several immune-related genes that could potentially serve as biomarkers for early septic shock.
During septic shock, there is an intricate interplay between hyperactivated inflammation and immunosuppression [20]. Neutrophil overactivation in septic shock may cause persistent reactive oxygen species-induced injury and induce dysfunction in vascular cells [21]. T-cell dysfunction in patients with septic shock would involve a decrease in CD8+ T-cell, CD4+ T-cell, and NK-cell proliferation favoring immunosuppression [22]. Our study revealed significant differences in immune infiltration between early septic shock and controls. Notably, the infiltration levels of neutrophils and M0 macrophages were significantly increased in early septic shock samples, consistent with findings by Kong et al. [23]. Additionally, activated CD4+ T cells, CD8+ T cells, and resting NK cells were reduced. Importantly, our results showed a significant decrease in ESTIMATE score and immune score, along with a significant increase in stromal score in early septic shock samples (H0). These findings suggested that, in general, patients with early septic shock may be in a state of immunosuppression and immune dysregulation. These results not only provided support for previous studies but also indicated a complex interaction between pro- and anti-inflammatory responses, suggesting that early septic shock is closely related to the immune microenvironment.
Hub modules (black and blue) associated with immunity and disease progression were identified through WGCNA analysis. The GO and KEGG results indicated that these core genes in hub models were closely associated with various biological processes such as immunity, inflammation, cell signaling, and metabolism. Notably, the core genes were mainly enriched in HIF-1, Fc epsilon RI, and metabolic pathways. Among them, HIF-1 is involved in multiple processes of inflammation, and its regulation of multiple hypoxia-related genes may have potential prognostic value in septic shock [24]. Fc epsilon RI is associated with inflammation, and it triggers allergic degranulation in a mouse model of endotoxemia [25]. Furthermore, different metabolic profiles have been suggested to be a major contributor to immune imbalances associated with sepsis [26].
Six hub genes were further identified in hub modules, including PECAM1, F11R, ITGAL, ICAM3, HK3, and MCEMP1. These genes showed significant dysregulation at 30 min and 24 h of ICU admission, while expression tended to normal levels at 48 h. Correlation analysis revealed that each hub gene was significantly associated with multiple immune cells, while all hub genes were significantly associated with M0 macrophages, further emphasizing the importance of macrophage activation in early septic shock. PECAM1, a member of the immunoglobulin superfamily, is involved in leukocyte regulation in a mouse model of sepsis [27, 28]. In a mouse model of endotoxin shock, PECAM1-deficient mice show increased sensitivity to lipopolysaccharide than the wild type and exhibit excessive inflammatory responses [29]. F11R, also known as junctional adhesion molecule-A, is a transmembrane protein, and its deletion leads to increased infiltration and phagocytosis of neutrophils within the gut in the mouse model of sepsis [30]. HK3, a key enzyme that catalyzes glycolysis, is significantly upregulated in neonatal sepsis and has the potential to differentiate patients with sepsis and septic shock [31]. Proteomic analysis of monocytes derived from septic shock patients indicates an upregulation of glycolysis enzymes and an impaired inflammatory phenotype, suggesting potential disturbances in both immune function and metabolic processes [32]. Increased MCEMP1 is detected in human and mouse models of septic shock [33]. MCEMP1 knockdown enhances T-lymphocyte and NK-cell activity while reducing the release of inflammatory factors in septic mice [34]. ICAM3 mediates crosstalk between neutrophils and NK cells [35]. These findings were consistent with our results, suggesting that the hub genes may play an important role in the immune microenvironment of early septic shock. Although ICAM3 and ITGAL/CD11a play essential functions in the immune response [36, 37], their role in immune function in septic shock needs further investigation.
In addition, the changes in the expression of hub genes flanked the strong immune dysregulation in the early stage of septic shock, and the expression levels tended to normalize at 48 h, which may be related to treatment effects and self-regulatory mechanisms. ROC analysis revealed that hub genes have good specificity and sensitivity in the diagnosis of early septic shock. Drug prediction results showed that three hub genes were potentially linked to related drugs. Among them, mycophenolate mofetil, as an immunosuppressant, was able to alleviate inflammation in a mouse model of lipopolysaccharide-induced acute lung injury [38]. Resveratrol and staurosporine-containing PH liposomes also demonstrated ameliorative effects on septic shock models, respectively [39, 40]. Taken together, these results suggested that immune-related hub genes have potential as biomarkers for early septic shock.
However, our study has some limitations. First, data analysis relied on a single data source with a limited sample size. Second, the data analysis relied on public database analysis and lacked clinical and functional validation. Future studies should focus on in vivo and in vitro experiments to validate the relationship between hub genes and the immune microenvironment in early septic shock, as well as to evaluate the potential of these hub genes as biomarkers for early septic shock in larger cohorts.
Conclusion
In conclusion, we identified six hub genes (PECAM1, F11R, ITGAL, ICAM3, HK3, and MCEMP1) that may be potential biomarkers for early septic shock and play critical roles in regulating the immune microenvironment. These findings provide a basis for future improvements in therapeutic strategies for septic shock.
Statement of Ethics
This study was approved by the Ethics Committee of the Second Affiliated Hospital of Shandong First Medical University (2022-094). Written informed consent was obtained from all patients.
Conflict of Interest Statement
The authors declare no conflicts of interest in this work.
Funding Sources
This work was supported by the National Natural Science Foundation of China (No. 81960342).
Author Contributions
B.L.: data curation and writing original draft. Y.F.: methodology. Y.F. and B.L.: investigation. X.Z. and Y.F.: visualization. B.L., Y.F., and X.Z.: conceptualization and validation. F.G. and W.S.: resources, investigation, and methodology. H.L., J.H., and Z.T.: project administration and resources. X.Z., H.L., and J.H.: software. Z.T.: writing review and editing. All authors read and approved the final manuscript.
Additional Information
Juntao Hu and Zhanhong Tang contributed equally and share the co-corresponding authorship.Edited by: H.-U. Simon, Bern.
Data Availability Statement
Data for this study were obtained from the public database GEO (https://www.ncbi.nlm.nih.gov/geo/), specifically from datasets GSE57065, GSE154918, and GSE115736. All remaining data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.