Introduction
Dengue was listed by the World Health Organization (WHO) as one of the top ten global health threats announced at the beginning of 2019 [
1]. In the past few decades, Dengue has become the fastest-growing mosquito borne disease in the world [
2‐
4]
, seriously endangering human health [
5,
6]. Vaccine research and development continue to make progress [
7‐
12], but the antibody-dependent enhancement (ADE) limits the effectiveness of vaccines [
13‐
17]. Asymptomatic infections increase incidence of dengue [
16,
18] and effective treatments have not been identified. Therefore, it’s urgent to explore the pathogenic mechanism of dengue fever and screen out molecular markers for better diagnosis and treatments.
Autophagy, a catabolic process that degrades damaged or abnormal intracellular components to recover nutrients, is essential for maintaining cell and body homeostasis [
19,
20], and benefits to proliferation and infection of the Dengue virus (DENV) [
21‐
24]. In DENV-ADE infection, cross-reactive antibodies mediate infection by inducing autophagy related proteins, and then suppress the innate immunity mediated by the mitochondria antiviral protein (MAVS) [
25]. Immune response affects directly or indirectly host response to DENV in varying degrees, including symptomatic infection, asymptomatic infection [
26,
27], dengue shock syndrome (DSS) and dengue hemorrhagic fever (DHF) [
28‐
30]. Therefore, it is essential to explore autophagy and immune response during DENV infection.
Transcriptomics researches are beneficial to assist researchers in better understanding disease causes [
31] and locating biomarkers [
32‐
34]. Our precious transcriptomics researches contributed to understand viral evolution and its impact on pathogenicity and vaccine development of DENV [
35‐
38]. However, studies [
39‐
41] published focused on multi-gene researches, and single analytical method (differentially expressed genes (DEGs) analyses), and did not link genomics with immune landscapes. In this study, we used a combination of DEG analyses, weighted co-expression network analysis (WGCNA) and Receiver Operator Characteristic Curve (ROC) to identify, validate and test biomarkers with diagnostic value of stages and severity in independent datasets, and applied the CIBERSORT website to analyze immune landscape differences among three stages and between DHF and Dengue Fever (DF) and explore correlations between genes and immune cells.
Discussion
This study combined DEGs analysis, WGCNA and ROC to identify, valid and test potential biomarkers associated with the staging and severity of Dengue, and used GO enrichment analysis, KEGG analysis and GSEA analysis to explore potential reasons resulting in DHF. The CIBERSORT website was also applied to explore immune differences during Dengue infection. Our research was the first to show that CD38 and ZNF595 had clinical significance in the stage and severity of Dengue and that they could be used as biomarkers to distinguish clinical stages and severity for Dengue patients. It was worth noting that Plasma cells, activated memory CD4+ T cells, and Monocytes also showed distinguishing value.
CD38 which was identified, verified and tested in independent datasets could distinguish three clinical stages of Dengue fever and was significantly associated with plasma cells, but it could not use to predict severity which was similar to precious results [
52] and distinguish different serotypes. Our results showed expression levels of
CD38 and fractions of plasma cells were similar in four serotypes. We identified, verified and tested
ZNF595 in independent datasets that could predict DHF. Mechanistically, EA-DEGs which inhibited viral replication were down-regulated in DHF, and a related-autophagy gene (
CCL2) expressed differently and increased significantly in DHF, all of which suggested DENV regulated autophagy to lead to DHF.
In addition, we analyzed systematically immune differences among three stages and between DHF and DF in the acute stage, and correlation between immune cells and genes. Immune analysis results showed fractions of monocytes, activated mast cells, M1 macrophages and neutrophils referred to innate immune response increased obviously in the EA stage compared with the C stage; significant increments can be observed in fractions of plasma cells and activated memory CD4+ T cells comparing the LA stage with the C stage; increasing fractions of plasma cells, CD8+ T cells, activated memory CD4+ T cells, follicular helper T cells, regulatory T cells (Tregs) and gamma delta T cells can be discovered in the LA stage compared with the EA stage. Above results were suitable for any serotype, suggesting different serotypes did not have obvious immune differences. Increasing immune response eliminated DENV which avoided Dengue symptoms [
27] and immune response kinetics dependent on initial lymphocyte numbers, were observed to correspond with illness severity [
53]. Our studies showed neutrophil and humoral immune response are activated in DHF in the LA stage, but the whole immune system was damaged in DHF compared with DF, which was a possible reason leading to DHF.
Interestingly, KEGG enrichment showed DEGs (between the EA stage and the C stage, and between the EA stage and the LA stage) were enriched in Coronavirus disease —COVID-19, Influenza A, Hepatitis C and Measles pathways, implying Coronavirus [
54], Alphainfluenzavirus influenzae [
55], Hepatitis C virus [
56] and Measles morbillivirus also up regulated these genes during infection and had shared pathogenic mechanism.
Gene expression profiles from public databases (GEO) have been applied to explore Dengue biomarkers by researchers. Several studies have used multiple genes to distinguish Dengue patients from healthy samples [
39,
40], DHF from DF [
41] and different stages [
57]. However, multi-gene lists can limit the sensitivity and specificity of DEGs as disease biomarkers [
58,
59].
Compared with previous studies [
39‐
41], our study had several advantages. First of all, this study was based on single gene analyses which allowed us to identify a stable and robust biomarker and these biomarkers were identified and verified on independent datasets which increased the accuracy of our study. Secondly, we combined three methods, including DEG, WGCNA and ROC to screen, verify and test biomarkers for Dengue diagnoses which increase the accuracy of our results, differing from precious studies only depending on the DEG analysis. Thirdly, we analyzed 22 types of immune cells including less attention immune cells based on RNA-sequence which contributed to increase understand of the whole immune response during DENV infection and explore correlations between genes and immune cells. A previous study showed that peripheral lymphocyte subset alteration can be independent predictors for clinical characteristics and treatment efficacy of COVID-19 [
60]. Interestingly, we found that the fraction of Plasma cells, activated memory CD4+ T cells and Monocytes in Dengue patients also had clinical characteristics and can distinguish these clinical stages for Dengue patients. Finally, not just for single serotype, our study included four serotypes which was involved in separate and combined analysis.
There is still a shortcoming of this study, because the study only uses public datasets to analyze, verify and test, without clinical verification and test. Our research has provided help in distinguishing the stage and severity of Dengue infection and understanding pathogenic mechanism of different serotypes, and will help analyze mechanisms of DHF and benefit to clinical treatment in the future.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.