Background
Sepsis is an inflammatory response caused by infection with pathogenic microorganisms. The body shock caused by it is called septic shock. Sepsis is a reaction to systemic infections [
1,
2]. Septic shock, associated with critical hypotension, is common acute diseases in the ICU (intensive care unit) [
2,
3]. It is estimated that about 8 million people worldwide die from sepsis (usually septic shock) every year, and abnormalities in the circulatory system, cells, and metabolism can significantly increase mortality [
1,
4].
Most of septic shock is caused by microbial infections (bacteria, viruses, fungi, etc.) [
5]. In early microbial infections, humoral reactions are activated, in which immune cells (macrophages, neutrophils, etc.) recognize and destroy invading organisms [
6]. Reduced blood vessel volume, cardiac dysfunction and peripheral vasodilation are major causes of septic shock [
6,
7]. In view of this, active fluid resuscitation and anti-infective symptomatic treatment are performed in these patients [
8‐
10]. However, 28-day and hospital mortality in patients remain very high [
8]. Moreover, the probability of re-admission after discharge from hospital is higher than that of ordinary ICU patients, and a considerable proportion of patients have cognitive dysfunction after treatment [
11‐
13].
Diagnosis and prognostic detection of diseases at the molecular level are now the general trend of development, which is also widely used by researchers in sepsis [
14,
15]. Mohammed et al. used high-throughput sequencing technology to identify potential biomarkers and signaling pathways related to septic shock [
16]. In addition, some researchers use TSD (transcriptomic signature distance) and meta-analysis to analyze the transcriptome data of septic shock patients [
17,
18]. Machine learning is a branch of computer science and statistics that play an important role in the detection, diagnosis and treatment of diseases [
19,
20]. Machine learning has also been used to study septic shock [
21,
22]. However, most of these studies use machine learning to predict the progression of septic shock. Machine learning is rarely used to identify potential diagnostic and prognostic biomarkers of septic shock. Therefore, in order to identify potential diagnostic gene biomarkers of septic shock, machine learning method was performed, followed by prognostic analysis in this study. Our study could be valuable in understanding the pathological mechanism of septic shock and exploring novel diagnostic gene biomarker for the diagnostic and therapy of the disease.
Discussion
Based on the machine learning method, 15 DEmRNAs, such as HMGN3, CACNA2D3, DUSP3, MGST1, CLEC5A, KIF1B, RAB13, ARHGEF18 and FCER1A, were determined as the superlative diagnostic gene biomarkers. The final survival analysis showed that only FCER1A and ARHGEF18 had obvious prognostic value.
HMGN3 (high mobility group nucleosomal binding domain 3) plays an important regulatory role in pancreatic cells [
33]. In patients with sepsis, high blood sugar is a risk factor for poor prognosis. During sepsis, the rapid changes in microvascular circulation in skeletal muscle have a serious hindrance to the delivery of insulin [
34]. HMGN3 can reduce the level of glucagon in the plasma [
35] to maintain stable blood sugar level in the body. In this study, HMGN3 was down-regulated in patients, which laid the foundation for further verification of the role in sepsis.
MGST1 (microsomal glutathione s-transferase 1), an important redox and detoxification enzyme, play a crucial role in cell defense and hematopoiesis [
36,
37]. CLEC5A (c-type lectin domain containing 5A) is a Syk (spleen tyrosine kinase) coupled c-type lectin, mainly expressed in myeloid cells, such as macrophages and neutrophils [
38], participates in host defense, inflammation, platelet activation and development [
39]. KIF1B (kinesin family member 1B) gene belongs to the kinesin superfamily, which is responsible for encoding proteins that transport mitochondria and synaptic vesicle precursors within the cell [
40]. In addition, KIF1B is found to be a tumor suppressor gene [
41,
42], which has a potential role in mitochondrial morphological changes. KIF1B and mitochondrial metalloproteinase YME1L1 (YME1 like 1 ATPase) coordinately regulate mitochondrial fission to induce mitochondrial apoptosis [
43]. In the early stage of sepsis, released NO (nitric oxide) can directly block mitochondrial respiration and cause body shock when accumulated to a certain degree [
6]. The potential role of KIF1B in mitochondria suggested that it may play a role in septic shock. RAB13 (RAB13, member RAS oncogene family) is present in all macrophage-related cells [
44]. In our study, MGST1, CLEC5A, KIF1B and RAB13 were all up-regulated in patients. This showed that MGST1, CLEC5A, KIF1B and RAB13 could play a crucial role in septic shock.
KLRF1 (killer cell lectin like receptor F1) is an activating homodimeric C-type lectin-like receptor, which plays an important role in regulating the activity of natural killer cells and monocytes [
45]. Recently, UPP1 (uridine phosphorylase 1) is reported to play an important role in immune and inflammatory biological process of disease [
46‐
48]. Previous studies have found that the expression of UPP1 is increased in the brain of sepsis rats [
49]. HDAC4 (histone deacetylase 4) plays an important regulatory role in sepsis and may be an effective target for sepsis treatment [
50,
51]. The expression level of NARF (nuclear prelamin A recognition factor) in multiple sclerosis (a chronic neuroinflammatory disease) was increased [
52]. So far, we have not found any studies on ECRP (ribonuclease A family member 2C, pseudogene) and LHFPL2 (LHFPL tetraspan subfamily member 2) in inflammatory or immune diseases. This article may first report that ECRP and LHFPL2 play a role in the progression of septic shock. In our study, KLRF1 (down-regulated), UPP1 (up-regulated), HDAC4 (up-regulated), NARF (up-regulated), ECRP (up-regulated) and LHFPL2 (up-regulated) were all abnormally expressed and could be considered as potential diagnostic biomarkers. These results suggested that KLRF1, UPP1, HDAC4, NARF, ECRP and LHFPL2 play a key role in septic shock. It provides a potential direction for further research on septic shock.
The protein encoded by ARHGEF18 (Rho/Rac guanine nucleotide exchange factor 18) plays an important role in activating eosinophils and other white blood cells [
53]. Sepsis is a high-risk disease caused by host reaction disorder and endangering the safety of life [
54]. Eosinophils are components of white blood cells of the immune defense system, and play a role in evolution of inflammation and disease [
55,
56]. FCER1A (Fc fragment of IgE receptor Ia) is an IgE receptor (immunoglobulin receptor), which is the initiating factor of allergic reactions and plays a role in allergic inflammation [
57,
58]. The interaction between FCER1B and other immunoglobulin-related inflammatory genes will increase the risk of asthma [
59]. In this study, ARHGEF18 and FCER1A were related to survival. In the enriched GO function, ARHGEF18 is mainly involved in regulating cell death and apoptosis. FCER1A is mainly involved in regulating immune regulation and metabolic processes. This further showed that ARHGEF18 and FCER1A may be related to the survival of septic shock patients.
The MAPK (mitogen-activated protein kinase) signaling pathway play a crucial part in the regulation of diseases, such as anti-inflammatory, analgesic, protective injury, etc. [
60]. MAPK contains three sub-pathways p38MAPK (p38 mitogen-activated protein kinase), ERK-1/2 (extracellular signal-regulated kinase), and JNK (c-Jun-terminal kinase) [
61,
62]. Among them, the p38MAPK and JNK signaling pathways play a role in hamowanie wzrostu, inflammation and pro-apoptotic signaling [
60]. MAPK pathway can be activated by extracellular signals, such as cytokines involved in inflammatory response, growth factors that regulate growth and metabolism, bacterial complexes [
60]. Inhibiting the activation of the MAPK pathway can reduce lung injury caused by septic shock [
63]. In the KEGG enrichment, CACNA2D3 and DUSP3 were taken part in the MAPK signaling pathway. CACNA2D3 (calcium voltage-gated channel auxiliary subunit alpha2delta3) plays an important role in canceration [
64‐
66]. CACNA2D3 is expressed in low levels in endometrial cancer tissues and cells [
64]. Overexpression of CACNA2D3 in vitro significantly inhibits tumor cell proliferation and migration [
64]. CACNA2D3, as a new tumor suppressor gene, can significantly inhibit lymph node metastasis of esophageal squamous cell carcinoma in clinical studies [
67]. Lymph nodes are immune sites for lymphocytes, which lays the foundation for studying the role of CACNA2D3 in septic shock. DUSP3 (dual specificity phosphatase 3), also called VHR (vaccinia-H1 related phosphatase), is a founding member of the bispecific protein phosphatase group [
68]. DUSP3 plays a role in Staphylococcus aureus infection [
69], DUSP3, a positive regulator of innate immune response [
70], is the main protein tyrosine phosphatase in macrophages mediating cellular processes (including immune responses) [
71]. This further illustrates that MAPK signaling pathway may play an irreplaceable role in septic shock by regulating related genes such as CACNA2D3 and DUSP3.
However, this study has certain limitations. Firstly, the sample size of the RT-PCR experiment is small, which may lead to a certain degree of error. More blood samples from septic shock patients are further needed to verify the expression of the identified mRNAs. Secondly, the molecular mechanism of DEmRNAs during septic shock has not been studied. More experiments are needed to further research the underlying mechanism of the disease.
Conclusions
In this study, in order to identify potential diagnostic gene biomarkers of septic shock, machine learning method was performed, followed by prognostic analysis. 15 superlative diagnostic gene biomarkers (KLRF1, UPP1, RAB13, KIF1B, CLEC5A, NARF, DUSP3, FCER1A, CACNA2D3, HMGN3, ECRP, HDAC4, LHFPL2, MGST1 and ARHGEF18) for septic shock were identified by machine learning analysis. It is worth mentioning that ARHGEF18 and FCER1A were related to survival. CACNA2D3 and DUSP3 participated in MAPK signaling pathway to regulate septic shock. Identified diagnostic gene biomarkers may be helpful in the diagnosis and therapy of patients with septic shock. This study can provide a basis for the research of septic shock.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.