Most septicemia-related deaths arise from the death of a large number of immune effector, thus severely impairing the immune response and creating a sustained immunosuppressive state [16, 17]. In particular, the apoptosis of lymphocytes, such as, innate, immune and adaptive immune cells is associated with an elevated risk for secondary infections as well as a poor prognosis in sepsis patients [18]. Therefore, specific apoptotic and anti-apoptotic pathways may represent attractive targets for sepsis [19]. In general, mechanisms underlying the expression of genes in sepsis may involve circular RNAs (circRNAs) [20], a stable form of non-coding RNA [21]; often these act as sponges for miRNA [22]. These sponges carry out critical functions in gene transcription and gene expression [23] and are known to engage in a range of cellular events, including differentiation, proliferation, apoptosis, and autophagy [24, 25]. Furthermore, in sepsis patients, circRNAs are known to be more abundant in immune cells than in healthy subjects [26]. The identified biological functions, as well as other characteristics, suggest that transcriptional and post-transcriptional regulation might carry out essential functions in the pathophysiology of sepsis [27, 28].

AS as well as mRNA isoforms participate in critical functions in nearly every cell process and pathway. Therefore, it was not unexpected that both were determined to be essential in effective antiviral responses [29]. AS events have been documented to participate in making fine adjustments to both responses. As an example, the Toll-like receptor signaling pathways, which are associated with innate immunity, are known to be controlled by AS events as well as alternative polyadenylation [30]. AS events also carry out critical functions in the activation of lymphocytes. The activation of T- and B-cells can considerably alter gene expression and AS events [31]. The changes in AS in these cells are not well-understood. However, one recent study profiled the B-cell transcriptional response to stimuli that activate B-cells [31]. The results from that study revealed that AS events, in particular the use of exons, affect numerous genes, with pronounced enrichment occurring in genes associated with functions in signaling and receptors [32].

Our study showed that the Cassetteexon, A5SS, A3SS were the most variable splicing events in sepsis. Approximately 75% of RAS events are complex splicing events. The splicing patterns in sepsis may have higher levels of heterogeneity, thus indicating the complexity of splicing regulation in sepsis. Our PCA plot also demonstrated that the mRNA AS events themselves could represent a solid comprehensive signature for detecting sepsis. As reported earlier, the alternative splicing of sphingomyelin phosphodiesterase1(SMPD1) functions as a negative regulator of acidic sphingomyelinase (ASM) activity. The activity of acidic ASM or SMPD carries out essential functions associated with the development of immune responses and organ failure in sepsis [33, 34]. In addition, a reduction in intron 5 retention is known to be a risk factor for severe sepsis and septic shock [35]. It has been reported that the polymorphic variants associated with exon 8 that are located at the 3´-UTR of the HLA-G gene are connected with septic shock that can occur in patients that are critically ill [36]. These results suggest that alternative splicing involves regulation of the immune/inflammatory response in sepsis, which might have aa crucial function in the mechanism of sepsis.

In this study we identified various AS patterns in genes related to cell apoptosis, including SHISA5, IFI27, HTT, MCL1, NR3C1, DELE1, NUDT2, FHIT. All of these genes showed significantly different RNA expression levels in the different groups. Interestingly, 5pMXE and 3pMXE were the most common events. As described earlier, the single-pass ER transmembrane protein SHISA5 has been reported to serve as a novel negative regulator of the processes underlying constitutive autophagy [36]. This finding is supported by the fact that SHISA5 deficiency alone is known to be sufficient to induce autophagy [37]. SHISA5 has been demonstrated to be a substrate in autolysosomal proteolysis, which is removed eventually from the site of action during the process of autophagy [38]. In addition, IFI27 was previously reported to be robustly associated with Th2 cells and Th1 cells, as well as aDC [38]. Furthermore, IFI27 has been shown to correlate positively with viral load, but it correlates negatively with counts of CD4 cells [39]. It is possible that IFI27 might participate in the mechanisms underlying immunodeficiency, and the expression of IFI27 might be associated with sepsis exacerbation, therefore representing a potential therapeutic target. These results indicated that the 5pMXE alternative splicing pattern of SHISA5 and IFI27 might be important targets of the immune cell apoptosis process in sepsis. Furthermore, skipping exon 6 of the FAS gene, which encodes the transmembrane domain, produces a soluble protein [39]. On the other hand, when this exon is included, a membrane receptor is produced, which triggers signaling pathways associated with cell death [40]. These findings indicate that the distinct AS patterns involved in sepsis pathophysiology might display essential functions in immune cell apoptosis in sepsis.

RNA-binding proteins (RBPs) are known to be crucial effectors in the expression of many genes [41] and are involved in the regulation of nearly every aspect of RNA functionality, including transcription, splicing, modification, intracellular trafficking, translation, and decay [42, 43]. Importantly, gene expression differences between sepsis survivors and non-survivors have been detected previously; furthermore, multiple genes related to immune function were poorly expressed in non-survivors [44]. This study has demonstrated that RBP expression was substantially different among the different study groups and clearly separated the sepsis group from the infection and healthy groups. Interestingly, the number of RBPs decreased with increasing severity of sepsis. The differential expression of RBPs during the process of sepsis may lead to differences in AS, thus affecting various aspects of cell function. GO analysis of the DE RBPs in sepsis identified terms that were mostly enriched in the immune/inflammatory response; we identified significant differences in 8 down-regulated RBPs in sepsis, including DDX24, CBFA2T2, NOP, ILF3, DNMT1, FTO, PPRC1, NOLC1. These changes indicate that the distinct expression levels of RBPs related to immune/inflammatory genes play important roles in the molecular mechanisms underlying the pathophysiology of sepsis.

The human nucleolar protein 14 (NOP14) gene has been reported previously to be located on chromosome 4p16.317 and is a key gene in the process of sepsis. This gene might be associated with pre‑18S rRNA processing that occurs during sepsis, or it might be involved in the inflammation that takes place during sepsis, acting through the regulation of miRNA‑2909 expression [45]. EMG1 and NOP14 are known to be members of a family of repressed environmental stress response (ESR) genes, which also includes genes that encode ribosomal proteins (RPs) as well as additional proteins known to be associated with RNA metabolism and the synthesis of proteins [46]. Furthermore, the enhancer binding factor 3 (ILF3) functions as a stable heterodimeric complex to stabilize mRNAs and regulate gene expression [47]. As reported previously, miR-215-5p expression is protective in inflammation injury that occurs in sepsis caused by H9c2 by targeting ILF3 and LRRFIP1 [48]. DDX24 is a DEAD-box helicase whose role in cells is largely unknown. However, DDX24 is capable of binding ssRNA and dsRNA. Interestingly, it has been shown to negatively affect RLR-dependent innate immune activation through several mechanisms [49]. DDX24 has been shown to compete with RIG-I for VSV RNA binding through its ability to bind RNA, exerting a direct inhibitory effect on viral perception [50]. At the same time, the qRT-PCR results of this study showed that the expression of DDX24 in the sepsis group was significantly lower than that in the healthy group, so we speculated that it plays a protective role in the innate immune response in sepsis through negative regulation.DNA methyltransferases1(DNMT1) -mediated DNA methylation is involved in many human diseases by affecting many types of cellular processes, including cell growth, cell cycle progression, metastasis, apoptosis, development, and tumorigenesis [51]. Fubing Ma et al. report that DNMT1-mediated increased DNA methylation plays a key role in LPS-induced sepsis by regulating the SMAD2/DNMT1/miR-145 negative regulatory loop [52]. The first discovered RNA demethylase obesity-associated protein (FTO) [53], involved in cell proliferation, apoptosis, cell cycle, migration, invasion, drug resistance and other processes [54]. As an oncogene, FTO promotes IDH mutations through the FTO/MYC/CEBPA signaling pathway, which in turn leads to tumorigenesis [55]. The human fragile histidine triad (FHIT) gene is a tumor suppressor gene, and heterozygous deletion (LOH), homozygous deletion, and abnormal expression of the FHIT gene have been implicated in several types of human malignancy [56]. In addition, FHIT has been reported to increase mitochondrial calcium release and promote apoptosis [57].The mechanism of action of FTO, NUDT2 and FHIT in sepsis is not clear, and we speculate that FTO, NUDT2 and FHIT may promote the development of sepsis by participating in apoptosis, and the mechanism needs further study. Collectively, these results suggested that genes encoding RBPs, such as DDX24, NOP, ILF3, DNMT1, FTO, PPRC1, NOLC1,represent novel targets for the molecular mechanisms that regulate sepsis and may be involved in the immune response by regulating AS events in key genes that are associated with cellular apoptosis.

Our study has some limitations that should be considered, even though we detected significant differences in AS in sepsis compared with a healthy group. However, we are still unclear whether changes in AS are related to certain bacterial categories. The factors that exert impact on AS in sepsis needs further research. Similarly, we need to investigate the mechanisms and clinical outcomes of ILF3 regulation on crucial genes in sepsis.