Background
Sepsis is one of the leading causes of death in intensive care units (ICUs) worldwide. It represents a difficult challenge to the intensivist because this condition lacks good diagnostic tools and adequate evidence-based treatments. The incidence of sepsis has increased considerably in recent years, and the morbidity and mortality rates associated with the syndrome are extremely high, especially for septic shock, in which mortality ranges from 40% to 60% in different populations [
1‐
4].
The molecular basis underlying sepsis mechanisms are complex and still only partially understood [
5]. Studies have demonstrated that the pathophysiology of sepsis is related to a dysregulated inflammatory response and oxidative stress [
6]. Recently, the presence of circulating extracellular vesicles (EV) shed from different cell types during sepsis and their relationship with sepsis pathophysiology was established [
7‐
9]. Previous studies demonstrated that a special type of EV, named
exosomes and mostly originated from platelets, induce superoxide production and apoptosis in vascular cells during sepsis by inflammatory and redox mechanisms [
10,
11]. Also, our group established a correlation between exosomes and organ dysfunction by demonstrating exosome-induced myocardial depression in isolated heart and papillary muscle preparations. This dysfunction was mediated by a mechanism associated with myocardial nitric oxide production [
12].
Recent studies demonstrated that exosomes may play a role in cell-to-cell communication in normal physiology and several diseases, including cancer, through carriage of functional messenger RNA (mRNA) and microRNAs (miRNAs, miRs) [
13‐
15]. Even though the presence of miRNAs in plasma during sepsis was previously demonstrated [
16,
17], the genetic content of exosomes during sepsis was evaluated in very few studies [
18‐
20].
In the present pilot study, we isolated exosomes from plasma of patients with septic shock and analyzed their entire miRNA content. We also assessed mRNAs, specifically looking for those related to inflammatory response and oxidative stress, because previous studies by our group and others demonstrated that these mechanisms mediate the effect of exosomes during sepsis [
10,
11]. In addition, we correlated the differential expression of miRNA to outcomes and used pathway enrichment analysis to elucidate their most important possible biological mechanisms. miRNA profiling revealed that the genetic material contained in exosomes from patients with sepsis is related to modulation of immune system and cell cycle regulation. The data presented in this article may help to generate new insights into sepsis pathophysiology and ultimately provide research windows to development of novel therapies.
Methods
A detailed description of the methods is provided in Additional file
1.
Patients and control subjects
In this prospective observational study, we recruited 24 patients in the initial 24 h of a first episode of septic shock who were admitted to the ICUs of Hospital Sirio-Libanes, Hospital São Paulo/Federal University of São Paulo, and Hospital das Clinicas/University of São Paulo, all in São Paulo, Brazil, from March 2012 to August 2013. Septic shock was defined as the need for vasopressor support after adequate fluid resuscitation [
21]. We excluded patients with severe anemia (hemoglobin < 7.0 g/dl) or active bleeding, those with known platelet disorders or conditions that cause thrombocytopenia other than sepsis (e.g., heparin-induced thrombocytopenia or chemotherapy), those with full heparin use (low molecular weight heparin or unfractionated) or use of any other medications that interfere with platelet function, those with a life expectancy no longer than 24 h, and patients with active cancer. The included patients had 30-ml blood samples collected from a central venous or arterial catheter up to 24 h after ICU admission (D0) and from a central venous or arterial catheter or by peripheral puncture 7 days after treatment (D7) for those who survived or were not discharged from the hospital. Twelve healthy volunteers without comorbidities and paired in a 2:1 ratio for sex and age were used as control subjects. The study was approved by the institutional ethics and review board of Hospital Sirio-Libanes (Comitê de Ética em Pesquisa, protocol number HSL 2010/61), which served as the coordinating institution, and also from the other institutions (Comitê de Análise para Projetos de Pesquisa, Hospital das Clinicas, Universidade de São Paulo, and Comitê de Ética em Pesquisa, Universidade Federal de São Paulo). Informed consent was obtained from the patients or their representatives.
Isolation and characterization of exosomes
Plasma was filtered through 0.2-μm membranes; exosomes were isolated by ultracentrifugation as previously described [
12]; and protein extracts were assessed by Western blotting using antibody against exosomal flotillin-1 (ab41927; Abcam, Cambridge, UK). For validation, the number of particles and the particle size were measured using a nanoparticle-tracking analysis device (NanoSight LM10; Malvern Panalytical, Malvern, UK) coupled to a charge-coupled device camera and a laser emitting a 60-mW beam at 405 nm. Video acquisitions were performed in five recordings of 60 seconds each. At least 1000 particles were tracked in each sample. Nano flow cytometry was carried out using human antibodies against cell surface antigens CD9 (exosome marker) and CD41 (platelet marker) using a CytoFLEX flow cytometer (Beckman Coulter Life Sciences, Indianapolis, IN, USA). We used violet side scatter and fluorescent polystyrene beads (Megamix-Plus FSC and SSC; BioCytex, Marseille, France) with known sizes (100, 160, 200, 240, 300, 500, and 900 nm) to identify vesicles smaller than 1 μm. The analysis was performed using CytExpert 2.1 software (Beckman Coulter Life Sciences). Plasmatic concentrations of interleukin (IL)-1β, IL-6, IL-8, IL-10, IL-13, tumor necrosis factor-α, and transforming growth factor-β were measured in seven patients with sepsis by enzyme-linked immunosorbent assay.
miRNA profiling and inflammatory and oxidative stress gene expression analysis
RNA and miRNA were extracted using the miRNeasy Mini Kit (Qiagen, Valencia, CA, USA) and were reverse-transcribed into complementary DNA (cDNA). Using TaqMan® Array Microfluidic Cards technology (Applied Biosystems/Thermo Fisher Scientific, Foster City, CA, USA) and qRT-PCR, we simultaneously evaluated 754 mature miRNAs (TaqMan MicroRNA Array v3.0; Life Technologies/Thermo Fisher Scientific, Grand Island, NY, USA) and 90 genes related to immune response (TaqMan Low Density Array Immune Profiling; Life Technologies/Thermo Fisher Scientific). The oxidative stress study consisted of 84 genes associated with oxidative stress and antioxidant defense using a commercial PCR array (PAHS065-A; Qiagen). Raw data were analyzed using ExpressionSuite v1.0.3 software (Life Technologies/Thermo Fisher Scientific), and all cycle threshold (Ct) values were evaluated using StatMiner version 5 (Integromics, Granada, Spain). Considering the cDNA preamplification step, Ct values greater than or equal to 32 for miRNAs and oxidative stress and greater than or equal to 35 for the immune study were excluded according to the manufacturers’ instructions.
For the miRNA analysis, the geNorm method was used to identify the best reference controls. A median of miR-17, miR-20a, and miR-106a expression was then used for data normalization. The expression levels of immune and stress oxidative genes were normalized to 18S ribosomal RNA and β-actin, respectively [
22]. The relative gene expression data were calculated using the comparative cycle threshold (2
−ΔΔCt) method [
23].
Target prediction, network pathway analysis, and cluster analysis
Ingenuity Pathway Analysis (IPA) software (
www.ingenuity.com; Qiagen Bioinformatics, Redwood City, CA, USA) and the MicroRNA Target Filter software tool (Qiagen Bioinformatics), which relies on three algorithms (TargetScan, TarBase, and miRecords), were used to identify putative targets of the miRNAs differentially expressed in septic exosomes versus healthy controls and in the comparison of patients with sepsis according to survival. IPA Network maintains a graphical database of networks of interacting genes (Ingenuity Knowledge Base; Qiagen Bioinformatics). Only experimentally validated targets were selected, indicating that only networks already demonstrated in previous studies to be associated with the specific miRNA expression are shown. A list containing the differentially miRNAs was uploaded into the IPA software and analyzed on the basis of content as of October 2016. The significance of the association between each list and the canonical pathways was measured by Fisher’s exact test. As a result, a
p value was obtained, and values below 0.05 were considered significant. To build the figures, molecules were represented as nodes, and the biological relationship between two nodes was represented as a line. These connections were supported by at least one reference from the literature, from a textbook, or from canonical information stored in the database [
24].
Using the normalized 2−ΔΔCt values, we constructed heat maps for comparison of patients with sepsis with healthy control subjects and patients with sepsis according to outcome, where the colors represent a scale of the smaller (blue) to larger (red) expression ratios for each miRNA as compared with the control samples (value = 1/white color). A further analysis by hierarchical cluster based on Euclidean correlation was then done. Each row represents a miRNA, and each column represents a patient sample. The color scale shown in the cluster grouping illustrates the normalized 2−ΔΔCt values for the miRNAs for all samples, where red and blue represent the expression above or below the average, respectively.
Statistical methods
The statistical tests were applied only to targets expressed in at least 50% of the samples from each group. The 2
−ΔΔCt values were tested for normal distribution using the Shapiro-Wilk test. For both paired and nonpaired comparisons, the Wilcoxon test was used. Benjamini-Hochberg adjustment was used to correct
p values (
p < 0.05) for false discovery rates [
24]. All tests were performed using R software version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).
Discussion
Our study comprehensively assessed the profile of miRNAs and mRNAs of exosomes from patients with septic shock in two stages of treatment. We demonstrated a differential regulation of miRNAs during sepsis, and the analysis of their targets indicates that they may potentially influence the immune system. Interestingly, the same pathways are still modified after 7 days of ICU admission, when most patients were already in a clinical recovery phase, as suggested by their reduced Sequential Organ Failure Assessment score. Our results suggest a subclinical and prolonged inflammatory response in sepsis, as previously reported in severe community-acquired pneumonia [
26].
Earlier data demonstrated the role of miRNAs in the inflammatory/immune response in sepsis [
27] and trauma. A study in which researchers analyzed the miRNA profiles of ten patients with polytrauma identified some miRNAs that were also differentially expressed in our patients (hsa-let-7b-5p, hsa-miR-151a-3p, hsa-miR-16-5p, hsa-miR-186-5p, hsa-miR-221-3p, hsa-miR-25-3p, hsa-miR-28-5p, hsa-miR-340-3p, and hsa-miR-618) [
28]. This could indicate a similar response of the body to a serious infectious or inflammatory injury.
Very few previous studies demonstrated the presence of miRNAs in exosomes during sepsis. In a mouse model, miRNA-125b was identified in endothelial progenitor cell-derived exosomes, and sepsis induced a downregulation of this miRNA [
19]. In our study, miRNA-125b showed greater expression in patients with sepsis with persistent expression after 7 days. In another study, some miRNAs from endothelial progenitor cell-derived exosomes (miR-15a, miR-27a, and miR-34a) had differential expression in plasma of patients with sepsis [
18]. In our study, the expression of exosomal miR-27a was six times higher among sepsis survivors. These miRNAs modulate the inflammatory response and the cell cycle [
18], a finding similar to that derived from our analysis.
Whereas the comparison between sepsis and controls identified miRNAs predominantly related to inflammation, the evaluation regarding outcomes identified miRNAs mostly related to cell cycle modulation. Cell cycle modulation was described in the context of sepsis and acute kidney injury [
29‐
31] and correlated with miRNA regulation [
29,
32]. In this scenario, miRNA regulation of target genes may lead the reentry into the proliferative cell cycle and differentiation, such as in fibrotic repair responses following tissue injury [
33], or, more likely, miRNAs may regulate cell cycle arrest to prevent differentiation of damaged cells under hostile conditions [
34,
35]. Whether it is conceivable that the miRNAs described here may affect both mechanisms, it is not possible without functional studies or on the basis of our results to identify which pathways are more influenced by these miRNAs.
The evaluation of mRNA related to oxidative stress demonstrated a significant and persistent increase in MPO mRNA. One major pathway regulated by MPO is the degradation of melatonin [
36], which is reduced during sepsis, with MPO playing a role in this process [
37,
38]. Another mRNA identified is FOXM1, a transcription factor related to cell cycle progression and activated by oxidative stress [
39]. Overexpression of FOXM1 in septic mice protects against lung injury and increases survival [
40,
41]. In our study, the increase in FOXM1 mRNA could also indicate a mechanism related to cell cycle regulation.
There is a paucity of information on how exosomal miRNAs interact with their targets to produce their effects. Most authors agree that exosomes are efficient carriers of proteins and genetic material to neighboring or more distant cells. However, the mechanisms of exosome–recipient cell interaction are unclear. The interaction usually starts with exosome internalization by various proposed pathways, including clathrin-mediated endocytosis, phagocytosis, or macropinocytosis [
42]. Once in the recipient cell, exosomal miRNA and mRNA are usually functional and can interact with their targets to synthesize new proteins or modulate gene expression [
13]. This has been demonstrated mostly in laboratory cell lines but not in sepsis, so any assumption regarding the interplay between exosomal genetic content and target cells during sepsis would be speculative.
Our study has limitations worth mentioning. First, this is an exploratory study with a small number of patients included, so our findings should be considered hypothesis-generating. Also, several comparisons were made, and as such, our analysis was corrected with a multiple correction-specific test (Benjamini-Hochberg) to minimize false-positive results. We included only patients in the first episode of sepsis to avoid those with chronic inflammation or immunosuppression that could influence the results. This criterion may limit the generalizability of our findings. Because we did not quantify the number of exosomes in the plasma of patients and healthy volunteers, the effect of sepsis on the exosomal counts was not estimated, and we could not normalize exosomal miRNA and mRNA levels to the amount of plasma exosomes. Plasmatic concentrations of cytokines were below the detection limit of the method in our patients, probably owing to the long-term storage of these samples, so we could not correlate the results of inflammatory cytokines with the exosomal genetic content. Also, we assessed only mRNA related to inflammatory response and oxidative stress, because these two important mechanisms were previously demonstrated to mediate the effects of exosomes in sepsis [
11,
12]. With this approach, it was not possible to evaluate other pathophysiological targets of exosomal mRNA. In order to compare the effects of infectious and noninfectious injury, we did not include a control group composed of noninfected ICU patients. In addition, we did not evaluate the function of these miRNAs in sepsis, so the pathways that they influence were not demonstrated in this study.