Non-Coding RNA Networks as Potential Novel Biomarker and Therapeutic Target for Sepsis and Sepsis-Related Multi-Organ Failure

According to “Sepsis-3” consensus, sepsis is a life-threatening clinical syndrome caused by a dysregulated inflammatory host response to infection. A rapid identification of sepsis is mandatory, as the extent of the organ damage triggered by both the pathogen itself and the host’s immune response could abruptly evolve to multiple organ failure and ultimately lead to the death of the patient. The most commonly used therapeutic strategy is to provide hemodynamic and global support to the patient and to rapidly initiate broad-spectrum empiric antibiotic therapy. To date, there is no gold standard diagnostic test that can ascertain the diagnosis of sepsis. Therefore, once sepsis is suspected, the presence of organ dysfunction can be assessed using the Sepsis-related Organ Failure Assessment (SOFA) score, although the diagnosis continues to depend primarily on clinical judgment. Clinicians can now rely on several serum biomarkers for the diagnosis of sepsis (e.g., procalcitonin), and promising new biomarkers have been evaluated, e.g., presepsin and adrenomedullin, although their clinical relevance in the hospital setting is still under discussion. Non-codingRNA, including long non-codingRNAs (lncRNAs), circularRNAs (circRNAs) and microRNAs (miRNAs), take part in a complex chain of events playing a pivotal role in several important regulatory processes in humans. In this narrative review we summarize and then analyze the function of circRNAs-miRNA-mRNA networks as putative novel biomarkers and therapeutic targets for sepsis, focusing only on data collected in clinical settings in humans.


Introduction
Sepsis is a life-threatening clinical syndrome caused by a dysregulated host inflammatory response to infection, often associated with multiple organ dysfunction syndrome and death [1]. It is recognized as a leading cause of death worldwide, as highlighted by the Global Burden of Disease, which estimated 48.9 million incident cases in 2017, accounting for 19.7% of all global deaths [2]. The economic and health burden of sepsis worldwide is alarming; mortality in sepsis patients has been estimated to be ≥10%, rising above 40% when evolving to septic shock [1]; in 2011 the total sepsis-related costs for US hospitals accounted for more than US $20 billion [1].
The definition of sepsis has undergone several revisions over the years because of the highly variable clinical spectrum: the 2001 American College of Chest Physicians/Society of Critical Care Medicine (ACCP/SCCM) Consensus Conference Committee criteria for the host systemic inflammatory response syndrome (SIRS) are currently outdated because The objective of this review is to summarize the current findings of circRNA-miRNA-lncRNA networks in the context of sepsis as a biomarker and therapeutic target with a focus on their clinical use in a hospital setting and their effectiveness in providing reliable data for the improvement of clinical practice in the adequacy of early diagnosis and treatment of sepsis and septic shock.

Methodology of Literature Search
A comprehensive literature search was carried out in the MEDLINE database (search terms: "sepsis" + "noncodingRNA", "miRNA", "circRNA", "mRNA", "lncRNA", "RNA", "network", "biomarker", "therapy", "prognosis", "organ failure"). The search has been restricted to papers published in English without time limit. The authors sought literature by examining reference lists in original articles and reviews. We have included in this review only systematic reviews, metanalyses, randomized trials and randomized controlled trials, selecting studies in which the main objective of the study was the identification and function of circRNA, miRNA, or mRNA networks in the context of novel biomarkers with remarkable prognostic value and/or therapeutic target for sepsis and sepsis-related organ failure.
Each author involved independently evaluated the results of the literature research extracting the most pertinent knowledge, while others verified the accuracy and completeness of the extracted data. Each author made a judgement as to whether the search results were different or confounding, trying to provide as complete an overview of the field as possible to date.

Role of Biomarkers in Sepsis
According to the Biomarkers Definitions Working Group, a biological marker or biomarker is "a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacologic responses to a therapeutic intervention" [26].
A biomarker finds application across four domains or functional classes: • As a diagnostic tool, i.e., a biomarker able to confirm a disease; • As a tool able to stage or to stratify disease severity; • As a prognostic tool; • An effective tool for prediction and monitoring of clinical response of an intervention [26].
Biomarkers of sepsis hold the promise of closing the gap in obtaining mycobacterial cultures by providing clinicians with clinically useful data. There is a strong demand for new and accurate sepsis biomarkers, especially in the era of personalized medicine in which physicians must increasingly tailor clinical and therapeutic management to each patient. In the following part of our manuscript, without aiming to address this topic comprehensively, the role of some of the major biomarkers currently used in sepsis will be analyzed by discussing the merits and demerits of their use during the management and treatment of such a serious disease that has its cornerstone in the timeliness of identification and early and appropriate intervention.

C-Reactive Protein (CRP)
C-Reactive Protein (CRP) is a plasma protein belonging to the group of the so-called acute phase reactants which may increase rapidly during inflammatory conditions or secondary to non-specific acute inflammatory stimuli [26]. The acute-phase proteins are produced in the liver during inflammatory states under the control of cytokines: CRP is mainly synthesized through Interleukin-6 (IL-6) and Interleukin-1β (IL-1β) stimulation via the transcription factors STAT3 and NF-κB [27]. CRP, as a component of the innate immune system, during infection may recognize various pathogens associated molecular patterns (PAMPs) such as phospholipid fragments released from damaged cells consequently activating the complement system and finally inducing the death of the targeted cells [28]. CRP is released into the bloodstream after 4-6 h after an inflammatory stimulus and a plasma peak is reached in 36 to 50 h [14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29]. Several conditions besides infection can result in the elevation of the CRP-serum level [30]; a meta-analysis of Simon  Liu et al. [32], in a systematic review and meta-analysis including 45 studies and 5654 patients, showed an acceptable level of sensitivity of 75% (95% CI: 69-79%) but a weak level of specificity of 67% (95% CI: 58-74%) for the ability of CRP to differentiate patients with sepsis vs. non-infectious inflammatory state/disorders. Tan et al. [33], comparing the ability of CRP and PCT to serve as biomarkers for sepsis diagnosis show similar sensitivity (CRP: 80%, 95% CI: 63-90%, procalcitonin: 80%, 95% CI: 69-87%) but significantly lower specificity for CRP (61%; 95% CI: 50-72%) than procalcitonin (77%; 95% CI: 60-88%) [33]. A possible explanation for the lower diagnostic accuracy of CRP as a sepsis biomarker (low specificity and moderate sensibility) could account for the slow-release kinetics as a consequence of the inflammatory stimulus and its increase also due to other pathological conditions besides infections (e.g., trauma, burns, surgery or various immune-inflammatory conditions [34,35].
Finally, the limits showed that CRP remains a widely used diagnostic and therapeutic biomarker in sepsis to date, mainly because a decrease in its values correlates with the success of antimicrobial treatment [36].

Procalcitonin (PCT)
Procalcitonin is the precursor of calcitonin, released by the C-cells of parathyroid glands. Assicot et al. [37] in 1993 for the first time described the association between PCT serum levels and severe bacterial infection. Compared to CRP, the PCT has a better kinetic profile, increasing within 3-6 h after the onset of infection reaching its serum peak after 6-8 h [29]. Several studies investigated the diagnostic performance of PCT. A meta-analysis of Uzzan et al. [35], including studies from 1996 to 2004, showed a higher accuracy of PCT levels than CRP levels for the diagnosis of sepsis (Global diagnostic accuracy odds ratios: CRP 5. 43 [35]. However, the authors included a restricted cohort study based only on surgery or trauma patients, and thus the conclusion cannot be extended to patients other than surgical conditions [35].
To date, there are no established cut-off values of serum PCT concentrations that are able to discriminate sepsis versus septic shock [29].

Presepsin
Presepsin, the N-terminal fragment of 13 kDa of the sCD14 (the soluble form of the receptor of lipopolysaccharide-lipopolysaccharide binding protein), is an emerging biomarker and early indicator of bacterial infections [40]. Presepsin, as part of the Toll-like receptor group, takes part of the innate immune system, binding several pathogen-associated molecular patterns (PAMPs) such as lipopolysaccharide (LPS) of Gram-or peptidogligans [40]. In recent years, sCD14 has become one of the most widely sepsis biomarkers studied: the level of sCD14 increased significantly in patients with sepsis and septic shock compared with healthy people, and the change was significantly related to the severity and prognosis of the disease [29,[41][42][43][44]. The diagnostic power of presepsin in detecting sepsis showed with a pooled sensitivity of 77-86% and a specificity of 73-78% [41][42][43][44]. Nevertheless, presepsin still needs wider investigation and further validation and comparison with standard sepsis biomarkers prior to being recommended for the hospital-setting.

Non-CodingRNA
Non-codingRNA, including long non-codingRNAs (lncRNAs), circularRNAs (cir-cRNAs) and microRNAs (miRNAs), take part in a complex chain of events playing a pivotal role in several important regulatory processes in humans. Non-codingRNAs are traditionally classified based on the length of nucleotides (nt) in small non-codingRNAs (sncRNAs) and long non-codingRNAs. The ncRNAs with 200 nt or lower are referred to as sncRNAs including microRNAs (miRNAs), small interfering RNAs (siRNAs) and piwi-interactingRNAs (piRNAs) whereas those with 200 nt or higher are referred as lncRNAs, such as promoter-associated transcripts (PATs), enhancerRNAs (eRNAs) and circular-RNAs (circRNAs).
Numerous studies have demonstrated that exosomes, phospholipid bilayer vesicles that originate from the membrane vesicles of the endosomes and are secreted by almost all cells, contain a variety of functional molecules that are crucial mediators of intercellular communication including lncRNAs, circRNAs and miRNAs. After exosomes are released into the tissue fluid, they arrive at the target cells and begin to deliver the different molecules contained (such as circRNAs), thus initiating functional responses and inducing subsequent phenotypic changes [45][46][47]. Recent studies have shown altered expressions of several non-coding RNAs such as lncRNAs, circRNAs and miRNAs during sepsis. Interestingly, noncoding RNAs have also been found to participate in the pathogenesis of multiple organ system failure through different mechanisms. In the following section of the review, the role of these three classes of noncoding RNAs in the pathophysiology of sepsis and sepsis-related multi-organ failure (MOF) will be examined.

CircRNAs
CircRNAs are an endogenous non-codingRNA which have as their main characteristic a closed loop structure with a covalent bond linking the 3 to 5 ends [48,49].
CircRNAs are highly conserved molecules expressed in a broad range of human cells, both in physiologic and pathologic conditions [50]. Memczak et al. detected 1950 circRNAs in HEK293 cells, 1903 circRNAs in the brain and fetal cells of mouse and through 724 circRNAs from different developing stages of Caenorhabditis Elegans (a nematode worm about 1 mm in length) [48].
Enuka et al. [51] estimate that circRNAs have a half-life of 18.8-23.7 h, roughly 2.5-fold higher that their linear form [52]. Also, circRNAs are insensitive to the common degradation pathways (i.e., RNAase or RNAexonucleases) [53]. Mature circRNAs are usually found in cytoplasms, whereas immature circRNAs that are still susceptible to intronic splicing remain in the nucleus [54].

CircRNAs Biogenesis
CircRNAs are primarily generated by the transcription of exonic and/or intronic sequences of a protein-coding gene in a complex reaction catalyzed by the RNA polymerase II (RNApol II). Both circRNAs and mRNAs, in their linear forms, are produced by the same precursor or pre-mRNAs undergoing a totally different splicing mechanism [55]. CircRNAs are produced by spliceosome machinery through a back-splicing process, resulting in a covalently closed loop structure between the 3 upstream splice site and the 5 downstream splice site [56][57][58][59][60][61].

CircRNAs Functions
CircRNAs act by separating miRNAs from their target mRNAs, thereby influencing miRNA-mediated gene suppression or expression. The interaction between circRNA-miRNA appears to be critical to the optimal functionality of our organism [62,63]. CircR- NAs may act as a dynamic scaffold influencing protein interactions having the potential to regulate protein function by binding, storing, sequencing and isolating proteins to specific subcellular locations [64]. Finally, nuclear circRNAs may act as regulators of the transcription by promoting the extended activity of RNA polymerase II [46,65] or through other alternative pathways which are still being studied.
Considering all of these features, the tissue development specific expression patterns and the putative crucial regulatory functions in various diseases, circRNAs thus have all of the qualities to be used as novel biomarkers in several pathologic conditions [19,21,66,67].

Long Non-CodingRNAs (lncRNAs)
The lncRNAs (200 nt or more) belong to the large class of non-codingRNAs that perform housekeeping functions in numerous biological processes through the regulation of gene expression at the post-transcriptional and transcriptional level [68][69][70]. Changes in the expression levels of lncRNAs affect the malignancy phenotype in various types of cancer such as colorectal, lung, liver, breast, ovarian cancers and leukemia [71,72], and are linked to development state [73] and may be observed during various phases of T-cell differentiation [74].

lncRNAs Biogenesis
To date, the lncRNAs biogenesis is not completely clarified. Transcriptome-wide studies show that lncRNAs biogenesis owns a peculiar expression pattern that is cell type-specific and stage-specific [75,76], which is also regulated by cell type and stagespecific stimuli [77] Briefly, lncRNAs are widely interspersed in the genome. Enhancers, promoters, and intragenic regions are the main DNA elements from which lncRNAs are transcribed [78].

lncRNAs Functions
The molecular function of lncRNAs may be summarized as four archetypes: (1) signal: lncRNAs can serve as molecular signal able to activate/silence specific genome sequences by interacting with chromatin and recruiting the chromatin modifying systems [76]; (2) Decoy: lncRNAs play a central role in the regulation of genome transcription, mainly by inhibiting it but occasionally also by activating it [76,77]; (3) Guide: lncRNAs can modify the chromatin structure guiding specific proteins to specific targets that ultimately cause gene silencing [76,77]; (4) Scaffold: lncRNAs can serve as a scaffold for assembling two or more proteins by inducing changes in chromatin as a consequence. They may therefore play a role in the activation or transcriptional silencing of specific genome sequences [76,77].

MicroRNAs (miRNAs)
miRNAs are endogenous non-coding transcripts of 19-22 nucleotides that modulate the translation of target mRNAs at a post-transcriptional level [79].

miRNAs Biogenesis
The RNA polymerase II transcribes a primary miRNA called pri-miRNA (~500-3000 nt); starting with this precursor a premature miRNA (pre-miRNA) is formed [80,81]. Through exportin 5, the pre-miRNA is exported in the cytoplasm and an miRNA duplex is processed. The "miRNA duplex" is bound into the RISC (RNA-induced silencing complex) with the final release of the mature miRNA [80]. As previously mentioned, circRNAs sequester mature miRNAs, acting as an miRNA sponge, through the interaction between the RNA binding proteins, diminishing miRNA functions. The most notably and predominant function of miRNAs is the binding of complementary sequences of the target linear RNAs that results in the inhibition of mRNA translation [82].

Non-Coding RNA and Sepsis
During sepsis the immune system is widely activated; the level and extent of the immune response triggered by the pathogen is different from subject to subject and depends largely on his/her state of immunocompetence [83,84].
Studies performed mostly in experimental animal models investigating the role of non-codingRNAs in the modulation of inflammation have identified some interesting networks that have been shown to be actively involved. Accumulating evidence shows that lncRNAs and miRNAs are involved in the sepsis inflammatory response, but the role played by different non-coding RNA networks in different biological contexts is extremely complex, variable, and involves the intervention of several mediators and effectors. An example will make clear the complexity of the field of research we are discussing: lncRNA taurine upregulated gene 1 (lncRNA TUG1) seems to participate in several pathophysiological processes. The overexpression of TUG1 has demonstrated the ability to alleviate the inflammatory response (including apoptosis and autophagy) in acute lung injury [85]. These relevant clinical effects seem to be mediated via targeting miR-34b-5p [86] and miR-27a-3p [87]. On the other hand, TUG1 silencing reduces the inflammation and apoptosis of renal tubular cells in an ischemia-reperfusion model via targeting of miR-449B-5p [88], and regulates the expression of various genes such as matrix metalloproteinase [89], protects against myocardial ischemia upregulating miR 142-3p [90] and downregulating miR-29a-3p [91] through the modulation of the miR-532-5p/Sox8 axis [92] and the miR-145-5p-Binp3 axis [93,94]. These are just a few of the many and complex interactions highlighted to date.
Despite the mentioned challenge of accurately defining the roles played by specific non-coding RNAs, some research has helped us identify some molecules of likely clinical interest. LncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) has been linked to sepsis. MALAT1 downregulation inhibits the LPS-induced inflammatory response by preventing the release of IL-6 and TNF-α and the NF-κB signaling pathway by upregulating miR-150-5p [95]. The MALAT1/miR-214/Toll-like receptor (TLR)5 signaling pathway dysregulation enhances the risk of post-burn sepsis by promoting greater production of proinflammatory cytokines [96]. LncRNA nuclear-enriched abundant transcript 1 (NEAT1) seems to also be involved in sepsis progression. In rats, NEAT1 knockdown could significantly improve the sepsis-induced myocardial injury preventing cardiac insufficiency and consensually increasing the ejection fraction (p < 0.05) [97]. Furthermore, another study indicates that NEAT1 inhibits the LPS-induced progression of sepsis in RAW264.7 cells by modulating the miR-31-5p/POU2F1 axis [98], suggesting that NEAT1, which also positively correlates with Th1 and Th17 levels [99], will be a potential target for clinical treatment of sepsis-induced organ damage.
Several miRNAs have been found to influence the course of sepsis, providing another substantial amount of evidence [100]. Over a hundred of different miRNAs have been investigated [101], identifying more or less prominent roles in the pathophysiology of sepsis-related injury by influencing the level of inflammasome induced.
What emerges is therefore an extremely interconnected network in which it is difficult to disentangle, especially if the objective is, as in our case, to try to determine whether there are the premises to attribute the role of biomarker of sepsis to one or more of the non-coding RNAs analyzed so far. It is a picture in which for each different pathological condition studied, different networks of non-coding RNA play different roles, making it impossible to assign a unique label to a specific lncRNA or miRNA.

Assessment of the Clinical and Prognostic Value of Non-Coding RNAs as Biomarkers in Sepsis
The role actually played in sepsis by the many non-coding RNAs that have been studied so far is only partially clarified. Many studies using cell and animal models have tried to demonstrate that non-coding RNAs may be used to control the multi-organ damage due to the septic process, but considerable challenges must be overcome in order to successfully translate these approaches into clinical practice. The concrete risk is to discuss notions that remain theoretical, not having the clinical relevance to propose themselves as concrete novelties in the future day-to-day clinical practice. Attempting to provide indications that are as translatable as possible into clinical practice, the following will be listed only the evidence available to date in which it was assessed the usefulness of miRNAs in sepsis in human clinical settings.

circRNAs
Using the research criteria previously described, only three studies to date have evaluated circ-RNAs by addressing the issue of their reliability as biomarker and prognostic value in clinical settings of sepsis in humans.
Wei et al. [102] explored the clinical values of circular RNA protein kinase C iota (circ-PRKCI) and its target microRNA-545 (miR-545) in sepsis patients. Plasma samples of 121 sepsis patients and 60 healthy controls were collected. Decreased circ-PRKCI expression and increased miR-545 expression were observed in sepsis patients compared to healthy controls, both of which had close correlations with sepsis risk. Decreased circ-PRKCI and increased miR-545 expressions were associated to 28-day mortality risk in sepsis patients, which were slightly lower than the predictive values of APACHE II score and SOFA score for predicting 28-day mortality risk.
Tian et al. [103] studied changes in circRNA expression in exosomes by circRNA microarray analysis in sepsis patients. ROC analysis showed that hsa_circRNA_104484 and hsa_circRNA_104670 have the potential to be used as novel diagnostic biomarkers and molecular therapeutic targets for sepsis.
Hong et al. [104] evaluated circFADS2 expression, a circRNA with putative protective roles in LPS-induced inflammation. Their results suggest that CircFADS2 is upregulated in sepsis to suppress LPS-induced lung cell apoptosis.

lncRNAs
Using the research criteria previously described, twelve studies addressed the issue of reliability as a biomarker and prognostic tool of lncRNAs in clinical settings of sepsis in humans. Table 1 shows the main findings as well as the diagnostic power of the different lncRNAs evaluated.

miRNAs
The research that has evaluated miRNAs with this aim in clinical settings of sepsis are of significant number, attracting the interest of researchers. We have tried to summarize in Table 2 the main elements of all available studies and in Figure 1 the main regulatory mechanisms of the pro-inflammatory state and interaction with host defense mechanisms that may substantiate their role as biomarkers in sepsis.     The research that has evaluated miRNAs with this aim in clinical settings of sepsis are of significant number, attracting the interest of researchers. We have tried to summarize in Table 2 the main elements of all available studies and in Figure 1 the main regulatory mechanisms of the pro-inflammatory state and interaction with host defense mechanisms that may substantiate their role as biomarkers in sepsis.

Conclusions
When facing sepsis or septic shock, time management is crucial to ensure the maximum chance of survival for the patient. Since the combined sensitivity and specificity of actual biomarkers (e.g., CPR, PCT, IL-6, etc.) do not allow for the rapid ascertainment of the diagnosis [14,15], they cannot always discriminate early enough between infectious and non-infectious patients and rapidly evolving patients from the more stable ones, whilst sepsis-related adverse outcomes rise with every hour of delay of proper intervention; new early biomarkers are urgently needed.
Non-codingRNA, including lncRNAs, circRNAs and miRNAs, cooperate in a comprehensive network deeply involved in gene function modulation, playing a pivotal role in several important regulatory mechanisms in humans, including the pathophysiology of sepsis and the sepsis-associated organ dysfunction.
The numerous studies carried out in the field of non-coding RNAs, their biogenesis, and their biological and clinical significance in recent years have identified the finding of the selective enrichment of exosomes with various networks of non-codingRNAs, different in health and disease according to options and mechanisms almost unknown today. This issue is certainly one of those most deserving of attention and investigation in the future.
Among the many studies conducted in clinical settings to ascertain the effective role of specific non-coding RNAs in supporting the early diagnosis of sepsis and in guiding therapeutic management, some have provided data that are worthy of highlighting and may provide a concrete starting point for future investigations. Some lncRNa, such as