Profiles of immune status and related pathways in sepsis: evidence based on GEO and bioinformatics

Sepsis, characterized as life-threatening sequential organ failure, is caused by a dysregulated host immune response to a pathogen. Conventional practice for sepsis is to control the inflammation source and administer highgrade antibiotics. However, the mortality rate of sepsis varies from 25–30% and can reach 50% if a septic shock occurs. In our current study, we used bioinformatics technology to detect immune status profiles in sepsis at the genomic level. We downloaded and analyzed gene expression profiles of GSE28750 from the Gene Expression Omnibus (GEO) database to determine differential gene expression and immune status between sepsis and normal samples. Next, we used the CIBERSORT method to quantify the proportions of immune cells in the sepsis samples. Then we explored the differentially expressed genes (DEGs) related to sepsis. Furthermore, gene ontology (GO) function and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were used to present potential signaling pathways in sepsis. We found that in the sepsis samples, the CD8 T cell fraction was consistently lower, based on the CIBERSORT method, whereas the neutrophil fraction was significantly higher in the sepsis samples. The GO function and KEGG pathway enrichment analysis identified 1573 DEGs that were significantly associated with neutrophil activation, neutrophil degranulation, neutrophil activation involved in the immune response, neutrophil-mediated immunity, and T cell activation in the biological processes group. In our study, we provided a first glance of associations between immune status and sepsis. Furthermore, our data regarding the reciprocal interaction between immune cells (neutrophils and CD8 T cells) could improve our understanding of immune status profiles in sepsis. However, additional investigations should be performed to verify their clinical value.


Introduction
Sepsis, a complex life-threatening organ dysfunction that ranks as the 10 th leading cause of death, is a perplexing imbalance between a pathogen and the body's immune response (Porte et al., 2019;Verdonk et al., 2017). It was reported that the rapidly increasing incidence of severe bloodstream infections with multidrug-resistant (MDR) pathogens have caused higher health care burdens for governments worldwide (Dalhoff et al., 2018). Sepsis not only causes primary infectious injury but also secondary damage to the infected tissues. Studies have been performed to examine the possible systemic effects of intensive sepsis that leads to the most severe consequence of septic shock, which causes significant morbidity and mortality (Muller-Redetzky, 2017;Osborn, 2017;Singer et al., 2016).
Currently, there are few molecular-based immunotherapies in existence for septic patients (Schrijver et al., 2019). In clinical practice for sepsis, the first step is to control the inflammation source and administer high-grade antibiotics (Liu et al., 2017;Sterling et al., 2015). Furthermore, vital organ support and even resuscitation may also be required for severe consequences (Busani et al., 2017). In the past few years, clinical trials from some large institutions have proven to be disappointing because of the complex heterogeneity of study populations and immunological phenotypes (Peters Van Ton et al., 2018). To date, researchers have explored immunosuppressive avenues for the treatment of sepsis, which leads to striking morbidity and mortality caused by sepsisinduced immunoparalysis (Bruse et al., 2019;Zijlstra et al., 2019). However, the current therapeutic focus has shifted from immunosuppressive strategies to enhancing the host's immune response (Esposito et al., 2017;Hagel et al., 2019).
It is well known that the initial immune response to infection is mounted by host cellular and humoral mediators, while neutrophils, as early responding immune cells, are recruited to the site of infection to exert their functions (Liu and Sun, 2019). However, recent studies showed that neutrophils may in fact be a double-edged sword in sepsis that could induce pyroptosis to fulfill their role in the active immune response. Therefore, it is crucial that we should pay close attention to the regulation of neutrophils when dealing with sepsis clinically.
Thus, in our current study, we used bioinformatics technology to detect immune status profiles in sepsis at the genome level. The Gene Expression Omnibus (GEO) database offers a pioneering medium of the genomic events in large cohorts worldwide, which serves as a public repository for archiving high-throughput microarray experimental data. We downloaded and analyzed the gene expression profiles of GSE28750 from the GEO database to determine the differential gene expression and immune status between sepsis and normal samples.

Data resources
The differentially expressed genes (DEGs) and the immune status during sepsis were investigated, relative to normal samples, after downloading and analyzing GSE28750 (Sutherland et al., 2011) profiles from the GEO database (Clough and Barrett, 2016) (http://www.ncbi.nlm.nih.gov/geo/) that essentially serves as a public repository, wherein high-throughput microarray experimental data is archived. The platform of GSE28750 was GPL 570 (Affymetrix Human Genome U113 Plus 2.0 Array).

Estimation of immune cell fractions
The well-designed CIBERSORT method (Newman et al., 2019) (http://cibersort.stanford.edu/), validated on gene expression profiles measured using microarrays, helped quantify the immune cell proportions in sepsis samples. CIBERSORT comprises 547 genes and specifically facilitates highly sensitive discrimination of 22 human hematopoietic cell phenotypes, including B cells, T cells, natural killer cells, macrophages, dendritic cells, and myeloid subsets. CIBERSORT established a P-value via the Monte Carlo method for deconvolution of each sample, offering a measure of confidence in our results, wherein the fractions of immune cell populations inferred at a threshold of <0.05 were considered accurate (Newman et al., 2015), and only patients conforming to this were then considered eligible for further investigation. The immune cell proportions were individually projected for each gene expression series, so for each sample, the sum of all estimates equaled 1.

Identification of DEGs
The downloaded original files were cataloged into sepsis and normal groups. The Bioconductor package 'affy' (http:// www.bioconductor.org/) standardized and transformed raw data into expression values (Gautier et al., 2004). The DEGs between early-detection sepsis and normal tissue samples were identified via applying a significance analysis of the empirical Bayes method within the Limma package (Ritchie et al., 2015). Adj. P-value < 0.01 and logFC > 1 were the designated cut-off criteria to select significant DEGs.
Functional enrichment analysis R language clusterProfiler package enrichment analysis facilitated the analysis of potential biological processes (BP), cellular components (CC) and molecular functions (MF) related to DEGs (Ashburner et al., 2000;Pickett and Edwardson, 2006;Yu et al., 2012). A KEGG pathway enrichment analysis presented potential signaling pathways. KEGG, as a comprehensive resource to ascertain functional and metabolic pathways, comprises exhaustive database compilations with detailed information on genomes, biological pathways, diseases, chemical substances, and drugs (Kanehisa and Goto, 2000;Ogata et al., 1999). A P-value of <0.05 was deemed statistically significant.

Estimation of immune cell fractions
The CIBERSORT fractions presented in Fig. 1B revealed CD8 + T cells were consistently lower in sepsis, compared with normal samples, whereas the neutrophil fraction was considerably higher in sepsis samples.

Identification of DEGs
Subsequent to pre-processing, a total of 1573 DEGs were identified in sepsis, relative to control samples. Fig. 2 presents a volcano plot of sepsis DEGs from each dataset.
GO function and KEGG pathway enrichment analysis R language clusterProfiler, used to apply GO function and KEGG pathway enrichment analysis, offered a detailed insight into DEGs, and the GO results were further categorized functionally to incorporate MF, BP, and CC. For MF, these DEGs were enriched for MHC class II protein binding complex, MHC protein binding complex, cytokine binding, protein tyrosine kinase binding, and protein kinase regulator activity. Moreover, these genes were significantly enriched in specific and tertiary granules, cytoplasmic vesicle lumen, vesicle lumen, and secretory granule lumen in the CC category. In the BP group, these DEGs were significantly associated with neutrophil activation, neutrophil degranulation, neutrophil activation involved in immune response, neutrophil-mediated immunity, and T cell activation ( Fig. 3 and Tab. 1). The results of the KEGG pathway analysis showed that DEGs were mainly enriched in pathways in the hematopoietic cell lineage, Th1 and Th2 cell differentiation, Th17 cell differentiation, inflammatory bowel disease (IBD), programmed death (PD) ligand 1 expression and the PD-1 checkpoint pathway in cancer, human T-cell leukemia virus 1 infection, the T cell receptor signaling pathway, primary immunodeficiency, Epstein-Barr virus infection and leishmaniasis (Fig. 4 and Tab. 2).

Discussion
Sepsis, characterized as life-threatening sequential organ failure, is caused by a dysregulated host immune response to a pathogen (Pei et al., 2018). It is vital that a balanced host immune response is maintained to eliminate systemic inflammatory responses and restore sequential organ functions. However, the underlying evolutionary mechanisms of host sepsis-induced inflammation, immunosuppression, and organ failure remain unknown (Drigo et al., 2018). Some immune modulators, such as Thymosin alpha 1 (Tα1), have been employed to great biological effect for septic patients with systemic inflammatory response syndrome (Pei et al., 2018;Pica et al., 2018). Although Tα1 seems to serve as an important alternative therapy supporting treatment for sepsis in these previous studies, sepsis manifests diversely, including systemic inflammatory response syndrome, and so identical treatment is not appropriate for all septic patients. Nevertheless, it is understood that there are powerful links between activation of first-line immune cells and the immunopathogenesis of sepsis (Kumar, 2018). Experiments investigating dysregulated activation of immune cells during sepsis progression could provide promising targets for immunomodulatory therapy.
In order to seek potential targets for immunomodulatory therapy, we have provided a first glance of associations between immune status and sepsis. In our current study, we first downloaded and analyzed the gene expression profiles of GSE28750 from the GEO database to investigate the   differential gene expression and immune status between sepsis and normal samples. Next, we used the CIBERSORT method to quantify the proportions of immune cells in the sepsis samples and detected highly sensitive and specific discrimination of 22 human hematopoietic cell phenotypes, including B cells, T cells, natural killer cells, macrophages, dendritic cells, and myeloid subsets. Significance analysis by the empirical Bayes methods within the Limma package was then applied to identify DEGs between early detection of sepsis samples and control samples based on the original CEL files.

B cells naive B cells memory Plasma cells T cells CD8 T cells CD4 naive T cells CD4 memory resting T cells CD4 memory activated T cells follicular helper T cells regulatory (Tregs) T cells gamma delta NK cells resting
We identified a total of 1573 DEGs in sepsis samples compared with normal tissue samples, and the fractions of CD8 + T cells were consistently lower as determined by CIBERSORT, whereas the fractions of neutrophils were significantly higher in the  sepsis samples. Furthermore, GO function and KEGG pathway enrichment analysis found that these 1573 DEGs were significantly associated with neutrophil activation, neutrophil degranulation, neutrophil activation involved in the immune response, neutrophil-mediated immunity, and T cell activation in the BP group.   During the first stage of the body's innate response to infection, neutrophils which serve as early responders play a key role in adaptive immune response progress, which includes anti-microbial CD4 + and CD8 + T-cell responses. It is well established that a reciprocal relationship exists between neutrophils and T cells, with neutrophils suppressing T cell activation. Research has revealed that neutrophils, by releasing reactive oxygen species, myeloperoxidase, and arginase to exert their effects, can suppress human T cell activation in vitro (El-Hag et al., 1986). A similar phenomenon of neutrophil-mediated T cell inhibition can be observed in both tumor patients and normal pregnancy. Recent research has found that with the increasing proportions of neutrophils, T cell function was remarkably reduced because of increasing arginase-1 levels in glioma patients (Kropf et al., 2007). Likewise, during normal pregnancy, it was found that the higher levels of arginase-1 expressed by neutrophils in the placenta and maternal blood were associated with T cell hyporesponsiveness. In our study, we identified that CD8 + T cell fractions were consistently lower in sepsis samples, while the neutrophil fraction was significantly higher in the sepsis samples.
In conclusion, we suggest a comprehensive estimate of associations between inflammatory response and sepsis. The fractions of both CD8 + T cells and neutrophils could improve our understanding of the heterogeneity of sepsis that promotes the immune status profiles in sepsis. More experiments are required to detect the reciprocal relationship between neutrophils and CD8 + T cells to elucidate the mechanism of action and identify prospective insights during sepsis progression.