Macrophage-derived microRNA-21 drives overwhelming glycolytic and inflammatory response during sepsis via repression of the PGE₂/IL-10 axis.

Abstract

Myeloid cells are critical for systemic inflammation, microbial control, and organ damage during sepsis. MicroRNAs are small non-coding RNAs that can dictate the outcome of sepsis. The role of myeloid-based expression of microRNA-21 (miR-21) in sepsis is inconclusive. Here we show that sepsis enhanced miR-21 expression in both peritoneal macrophages and neutrophils from septic C57BL/6J mice, and the deletion of miR-21 locus in myeloid cells (miR-21^Δmyel mice) enhanced animal survival, decreased bacterial growth, decreased systemic inflammation, and decreased organ damage. Resistance to sepsis was associated with a reduction of aerobic glycolysis and increased levels of the anti-inflammatory mediators prostaglandin E₂ (PGE₂) and IL-10 in miR-21^Δmyel in vivo and in vitro. Using blocking antibodies and pharmacological tools, we discovered that increased survival and decreased systemic inflammation in septic miR-21^Δmyel mice is dependent on PGE₂/IL-10-mediated inhibition of glycolysis. Together, these findings demonstrate that expression of miR-21 in myeloid cells orchestrates the balance between anti-inflammatory mediators and metabolic reprogramming that drives cytokine storm during sepsis.

Introduction

Sepsis is life-threatening organ dysfunction caused by a dysregulated host response to infection (1, 2). More than 48 million cases of sepsis and 11 million sepsis-related deaths were reported worldwide in 2017 (3). Interactions between the infecting microorganisms and the host innate immune system culminate in the complex pathogenesis events associated with sepsis (4, 5). Previous studies have demonstrated that the macrophage-mediated inflammatory response is involved in both exaggerated systemic inflammation and immune suppression in sepsis (6, 7). During sepsis, macrophages mediate the inflammatory response, depend on the metabolic switch, and are heavily involved in the inflammation and immune suppression during sepsis (7, 8).

Both pathogen and host-derived molecular patterns leads to the activation of a myriad of pro-inflammatory programs that are triggered by nuclear factor-kappa B (NF-kB), Janus kinases (JAKs), signal transducers and activator of transcription (STATs), and hypoxia-induced factor 1α (HIF-1α) (9). Activation of these pathways induces pro-inflammatory programs, which are crucial to controlling of bacterial infection (10–12). Furthermore, HIF-1α mediated expression of glycolytic enzymes and transports leads to aerobic glycolysis in macrophages (13, 14). Aerobic glycolysis is a critical metabolic pathway required for macrophage pro-inflammatory activity (15, 16). The blockage of glycolysis in macrophages reduces systemic inflammation and leads to improvement of sepsis outcomes in experimental models (7, 17). Exaggerated aerobic glycolysis and production of inflammatory mediators are critical determinants of organ dysfunction during sepsis (18).

Recent studies have shown that miRNA expression is associated with severe illness and suggest they may serve as key mediators of sepsis (19). microRNA (miRNAs) are small non-coding RNA molecules (18–23 nucleotides) that regulate gene expression by inducing transcript destabilization or translational repression (20, 21). miRNAs have an essential role in many biological processes, such as development, differentiation, cell survival, and inflammatory response (22). miRNAs are known to regulate phagocyte cytokine and chemokine responses, antimicrobial effector function, pathogen recognition, and tissue repair (22). The miRNA miR-21 is highly expressed in immune cells including as T/B lymphocytes, monocytes, macrophages, and dendritic cells (23). In vitro studies have shown that inflammatory stimuli such as LPS and cytokines such as IL-6, TFG-β, and TNF-α induce miR-21 expression (24, 25). Our laboratory has shown that miR-21 deficiency promotes the homeostatic generation of antiinflammatory macrophages by controlling the expression of STAT3 (26). miR-21 can also prevent phagocytosis by targeting GTPases (27), limit TLR signaling by inhibiting MyD88 and IRAK, and suppress NFκB signaling (23). Therefore, miR-21 has a complex role in macrophages regulating pro- and anti-inflammatory programs.

Experimental and clinical studies have shown that miR-21 is upregulated in acute and sustained in late sepsis (28). However, the role of miR-21 in LPS-induced endotoxemia or sepsis is inconclusive. McClure et al. have shown that treatment of BALB/c mice with a miR-21 inhibitor enhances animal survival and decreases bacterial loads after cecal ligation and puncture (CLP) (29). Whole-body miR-21^−/− mice are also protected from septic shock by reducing inflammasome activation and IL-1β processing and secretion (30). Delivery of a miR-21 antagomir is cardioprotective in septic mice by increasing the numbers of myeloid suppressor cells (28). On the other hand, Barnett et al. did not identify any differences in animal survival when global miR-21 deficient mice were CLP-induced sepsis (31). Although interesting, these studies did not reveal the cellular and molecular targets by which miR-21^−/− improves animal survival, and the role of miR-21 in aberrant and damaging metabolism during sepsis remains completely unresolved. Therefore, to understand the complex roles miR-21 may be playing in sepsis, it is necessary to both dissect the function of miR-21 in individual cell subsets as well as identify the molecular targets and pathways by which this miRNA acts. Here we found that miR-21 expression in macrophages and neutrophils inhibits the PGE₂/IL-10 axis and drives aberrant glycolysis and organ injury and lethality during sepsis. Our data show that miR-21 inhibition of anti-inflammatory responses favors metabolic reprogramming and mediates macrophage-dependent lung injury and animal mortality.

Materials and Methods

Mice

Female and male C57BL/6 and LysM^cre mice (8 weeks old; weight, 18 to 23 g) were obtained from The Jackson Laboratory. miR-21^fl/fl mouse strain was generated and donated by Dr. Ivan Mircea (Indiana University School of Medicine) trough Ozgene Ptv Ltd (Perth, Australia). The miR-21^fl/fl and LysM^cre mice were initially crossed at Indiana University School of Medicine to generate miR21^∆myel mice and then maintained at Vanderbilt University Medical Center. miR-21 deletion in macrophages was confirmed in each mouse before the experiments (Supplementary Figure 1). Mice were maintained according to National Institutes of Health (NIH) guidelines for the use of experimental animals with the approval of the Indiana University School of Medicine and Vanderbilt University Medical Center Committee for the Use and Care of Animals.

Polymicrobial sepsis and endotoxemia induction

Sepsis was induced by CLP as previously described (32). A 21 G needle was used to cause moderate sepsis. The survival rate was evaluated 6 h after CLP and twice per day every day during 7 days after CLP induction. Peritoneal lavage, blood, and lung were harvested 6 h after surgery to measure cytokines, markers of tissue damage, or bacterial burden in septic mice.

In another experimental setting, miR-21^fl/fl and miR-21^Δmyel mice were challenged i.p. with either a moderate (5mg/kg) or lethal dose (10 mg/kg) of LPS from E. coli serotype 0111: B4 (Sigma-Aldrich) or 0.9% saline control (33). Anti-IL10R treatment or isotype control (10 mg/kg, clone 1B1.3A BioXcel) was injected 1 hour before LPS injection (32). Mice survival was assessed over 7 days. Cytokine levels were measured in serum and/or peritoneal lavage fluid 12 hours after challenge with LPS.

Bacterial determination

Bacterial counts were determined, as previously described (32). In brief, peritoneal exudate and blood were collected from septic mice 6 h after CLP surgery, and serial dilutions were plated on Muller-Hinton agar dishes (Difco Laboratories, Detroit, Michigan). Colonies were counted after incubation overnight at 37°C. The results are expressed as log CFU (colony-forming unit) per ml of blood or peritoneal exudate.

Flow cytometry

Peritoneal cells were resuspended in FACS Buffer (PBS 1x, 2mM EDTA, 0,5% BSA). Cell suspension was blocked with anti-CD16/CD32 (clone 2.4G2; BD Biosciences Pharmingen) for 10 min at 4°C. BD Cytofix/Cytoperm™ was used to fixation and permeabilization of the cells according to the manufactory instructions (Biosciences Pharmingen). The cells were stained with mouse anti-CD11b FITC (1:200, BD biosciences) for 30 min at 4°C. To detect the proteins involved in glycolysis, we used the following antibodies (HK1 clone C35C4, GLUT1 clone D3J3A and HIF-1α D1S7W - Cell Signaling - all at 1:500), followed by incubation with anti-Rabbit secondary Ab (PE anti-IgG H+L 1:1000, Invitrogen). The cells were acquired by LSRII (BD Biosciences) at the VUMC Flow Cytometry Shared Resource core (FCSR). Data were analyzed with FlowJo Version 10.

Determination of neutrophil migration

Mice were submitted to CLP and sacrificed 6 h after surgery. Cells in the peritoneal cavity were harvested with cold PBS/EDTA. Cells were pelleted by centrifugation and resuspended in ACK lysis buffer for 5 min and resuspended in DMEM and quantified using a Neubauer chamber. Cells (1×10⁶ cells/100 μL) were incubated with anti-FcγR antibodies CD16/CD32 (clone 2.4G2; BD Biosciences Pharmingen) to prevent nonspecific antibody binding, followed by incubation with phycoerythrin-conjugated CD11b mAb (BD Biosciences) and allophycocyanin-conjugated anti-Ly6G mAb (BD Biosciences). After incubation, the cells were washed, fixed, and analyzed by flow cytometry, using a BD LSR II flow cytometer.

Peritoneal macrophage isolation

Peritoneal cells from naïve mice or septic mice were harvested 12 h after CLP were harvested as shown before (32). Ly6G⁺ cells were isolated by Neutrophil isolation Kit mouse and F4/80⁺ cells by Anti-F4/80 Microbeads Ultrapure mouse kit followed the manufacturer’s instructions (MACS, Miltenyi Biotec).

BMDM cell culture

Bone marrow cells were flushed from femurs and tibias of mice with 10 mL of DMEM supplemented with 10% FBS (Gibco), antibiotic/antimycotic solution (100 X, HyClone), GlutaMAX (100X, Gibco) and HEPES (25 mM, Sigma). Cells were pelleted by centrifugation and resuspended in ACK lysis buffer for 5 min and resuspended in DMEM. 10⁷ cells were resuspended in DMEM supplemented with 20 ng/mL of M-CSF (Peprotech) and 5 ng/mL of GM-CSF (Peprotech) and plated in 10 cm tissue culture-treated petri dishes. Three days later, we added 10 mL of BMDM medium and on day 6, the medium was replaced with 10mL of fresh BMDM medium. 24 hours later, cells were treated with different doses of LPS from E. coli serotype 0111: B4 (Sigma-Aldrich) for different time points. In different experiments, cells were treated with anti-IL10R (10 μg, clone 1B1.3A BioXcel), cyclooxygenase inhibitors (Aspirin 100 μM, and Indomethacin 10 μM, Sigma Aldrich) and mPGES1 inhibitor CAY10526 (Cayman Chem.).

Cytokine and chemokine quantification

Cytokine and chemokine production from the serum and peritoneal exudate were measured by ELISA according to the manufacturer’s instructions (Biolegend or R&D Systems).

Measurement of eicosanoids by mass spectrometry

Supernatant of BMDM from miR-21^fl/fl or miR-21^Δmyel with or without LPS challenge were placed in a microcentrifuge tube containing 1 mL of 25% methanol in water and an internal standard mix (1ng each deuterated eicosanoid). The Vanderbilt University Eicosanoid Core Laboratory processed the samples and injected for liquid chromatography-MS (LC-MS) as previously described (34). Eicosanoids were identified and quantified based on the mass and amount of known standards as previously described (34). The levels of PGE₂ were also measured in the serum or the peritoneal lavage fluid of septic mice by EIA according to the manufacturer’s instruction (Cayman Chem.).

Lactate activity assay

Lactate in peritoneal exudate was quantified using a Lactate Kit (Quibasa Quimica Basica - Brazil), according to the manufacturer’s instructions.

Creatine kinase-MB (CK-MB) and glutamic-oxaloacetic transaminase (TGO) activity

The levels of CK-MB and TGO were measured in the serum of septic mice, 6 h after surgery as a biochemical indicator of heart and liver injury, respectively. The determinations were made using a commercial colorimetric kit (Labtest, Brazil).

Gene expression analysis

RNAs from BMDMs and peritoneal macrophages were isolated using the miRNeasy Mini kit (Qiagen) according to the manufacturer’s instructions and as previously described (35).. Primers for Rnu6 and miR-21 were purchased from Sigma Aldrich. Primers for β-actin, Tnfa, Il1b, Il6, Il10, Nos2, Hif1a, Hk1, Hk2, Scl2a1, Pkm, Ldha, Slc16a3, Pla2g4a, Pla2g2a, Pla2g6, Ptgs1, Ptgs2, Ptges1, Ptges2, Slco2a1 and Hpgd were purchase from Integrated DNA Technologies (table I). Relative gene expression was calculated using the comparative threshold cycle (C_t) and expressed relative to control or WT groups (ΔΔCt method)

Table I:

Primers list

Primer	Assay ID	Ref. Seq.
Il1b	Mm.PT.58.41616450	NM_008361(1)
Tnf	Mm.PT.58.12575861	NM_013693(1)
IL6	Mm.PT.58.10005566	NM_031168(1)
IL6ra	Mm.PT.58.31166746	NM_010559(1)
IL10	Mm.PT.58.13531087	NM_010548(1)
Hif1a	Mm.PT.58.11211292	NM_010431(1)
Hk1	Mm.PT.58.9947184	NM_001146100(2)
Hk2	Mm.PT.58.32698746	NM_013820(1)
Slc2a1	Mm.PT.58.7590689	NM_011400(1)
Pkm	Mm.PT.58.6642152	NM_001253883(2)
Ldha	Mm.PT.58.29860774	NM_001136069(2)
Slc16a3	Mm.PT.58.13307682.gs	NM_001038653(3)
Nos2	Mm.PT.58.43705194	NM_010927(1)
Slco2a1	Mm.PT.58.13147950	NM_033314(1)
Hpgd	Mm.PT.56a.9684089	NM_008278(1)
Pla2g4a	Mm.PT.58.42486653	NM_008869(1)
Pla2g2a	Mm.PT.58.13649588	NM_001082531(2)
Pla2g6	Mm.PT.58.15922201	NM_001199024(1)
Ptgs1	Mm.PT.58.13984299	NM_008969(1)
Ptgs2	Mm.PT.58.9154407	NM_011198(1)
Ptges	Mm.PT.58.28527240	NM_022415(1)
Ptges2	Mm.PT.7480753	NM_133783(1)
Actb	Mm.PT. 39a.22214843.g	NM_007393(1)
miR21	M_Mir21_1	NR_029738
RNU6	MIRCP00001

List of the primers, assay ID and reference number of the primer pairs

RNA sequencing and data Analysis

RNA was isolated from BMDMs as described above and sent to VANderbilt Technologies for Advanced GEnomics (VANTAGE) core at Vanderbilt University. cDNA synthesis, end-repair, A-base addition, and ligation of the Illumina indexed adapters were performed according to the TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA). The concentration and size distribution of the completed libraries was determined using an Agilent Bioanalyzer DNA 1000 chip (Santa Clara, CA) and Qubit fluorometry (Invitrogen, Carlsbad, CA). Paired-end libraries were sequenced on an Illumina HiSeq 4000 following Illumina’s standard protocol using the Illumina cBot and HiSeq 3000/4000 PE Cluster Kit. Base-calling was performed using Illumina’s RTA software (version 2.5.2). Paired-end RNA-seq reads were aligned to the mouse reference genome (GRCm38/mm10) using RNA-seq spliced read mapper Tophat2 (v2.1.1). Each sample was analyzed in triplicate.

Raw read quality was assessed using FastQC (v0.11.5). Salmon (v0.14.0) was used to quantify transcript expression using annotation gencode v21 mouse transcriptome (https://www.gencodegenes.org/mouse/release_M21.html) under quant mode with default parameters (transcript index was generated with k = 31 using transcripts fa file from gencode v21). Transcript expression was summarized at the gene level and imported into R using txImport (36, 37). R package EnsDb.Mmusculus.v79 (https://bioconductor.org/packages/release/data/annotation/html/EnsDb.Mmusculus.v79.html) was used to annotate the genes. Then the differentially expressed genes were called using edgeR (v2.26.5) with Benjamini-Hochberg adjusted p-value < 0.05 and log2FoldChange > 2. R package clusterProfiler (v3.12.0) was used for the gene set over-representation analysis with KEGG database. K-means cluster was applied to genes expression values (normalized by count per million) using R function kmeans (from R package stats), the number of clusters was determined by total within sum of square. R package ggplot2 was used to make the volcano plots. Heatmap was generated using pheatmap (v1.0.12).

Immunoblotting

Western blots were performed as previously described (35). Protein samples were resolved by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with commercial available antibodies - glycolysis antibody Sampler (Kit 8337, Cell Signaling) and β-actin Dylight 800 (1:1000, Invitrogen). Membranes were then washed and incubated with appropriate fluorophore-conjugated secondary antibodies (1:10,000, anti-rabbit IgG, IRDye 800CW antibody, Licor). Relative band intensities were quantified using ImageJ software (NIH), as previously described.

Elicited macrophages

Peritoneal cavity from miR-21^fl/fl or miR-21^Δmyel mice were elicited with an injection of 1 ml 3% thioglycolate for four days, and collected as previously described (26). Phagocytes were stimulated with 10 ng/ml LPS (Sigma-Aldrich), and for indicated time points, the supernatant was used to determine cytokines production.

Extracellular Flux Analyses (Seahorse)

Glycolytic stress assay (Extracellular acidification rate - ECAR) was performed using the Seahorse Bioscience XFe96 Extracellular Flux Analyzer. 50,000 BMDMs from miR21^fl/fl or miR21^∆myel were seeded in a Seahorse 96-well plate in DMEM and incubate overnight at 37C an atmosphere of 5% CO₂. On the next day, BMDMs challenged with vehicle (US) or stimulated with LPS (1,10 and 100 ng/mL). In different experiments, cells were pretreated with anti-IL10R (clone 1B1.3A, BioXcel) and CAY10526 (Cayman CC) for 12 and 24 h, before the incubation with LPS. After different time points of LPS stimulation, the culture medium was replaced for XF Base Media (DMEM) supplemented with 2mM glutamine. All experiments were done under sterile conditions and the freshly prepared media adjusted to pH of 7.4 before each assay. The glycolysis test compounds were injected in the following order: Glucose (10 μM), Oligomycin (1.0 μM), and 2-deoxyglucose (2-DG, 50 μM). Glycolysis stress was measured by recording extracellular acidification rates (ECAR, milli-pH units per minute).

Statistics analysis

Data analysis was performed in GraphPad Prism software. Statistical tests used are listed for each experiment in the corresponding figure legends. Briefly, Student’s 2- tailed t tests were used to compare 2 experimental groups. One-way ANOVA followed by Tukey’s multiple comparison corrections was used to compare 3 or more groups. Two-way ANOVA with repeated measured followed by Tukey’s multiple comparison corrections was used to compare infection areas over time between 2 or more mouse groups. Survival curves were expressed as percent survival, and Log-rank (Mantel-Cox) test was used to determine differences between survival curves. p < 0.05 was considered statistically significant.

Results

miR-21 expression during sepsis.

Initially, we evaluated the expression of miR-21 in total peritoneal cells of mice undergoing CLP-induced polymicrobial sepsis. Our data show that miR-21 abundance was increased after 12 h and remained enhanced up to day 7 post-infection in total peritoneal cells (Figure 1A). Since macrophages and neutrophils play key roles in both microbial control and cytokine storm, we assessed whether these phagocytes express different levels of miR-21 in both sham and septic mice. We observed that miR-21 expression is higher in bone marrow neutrophils than macrophages in naïve mice, but 12 h after sepsis; miR-21 abundance was similar in both cells (Figure 1B). These data show that miR-21 is greatly increased in both macrophages and neutrophils after sepsis.

Figure 1: Myeloid-based expression of miR-21 drives poor sepsis outcomes.

(A) miR-21 expression levels (relative to sham) in the peritoneal cells from septic mice at days 1, 7, and 10 after CLP determined by qPCR. (B) miR-21 expression (relative to naive cells) in bead-selected bone marrow Ly6G⁺ and F4/80⁺ peritoneal cells of septic or naive mice determined by qPCR. Data are from n = 8 mice per group and were analyzed by t-test. (C) The survival rate of miR-21^fl/fl and miR-21^Δmyel septic mice monitored for 10 days after CLP. Data are from n = 10 mice per group and were analyzed by the log-rank (Mantel-Cox) test. (D) Bacterial burden in the blood and peritoneal lavage from miR-21^fl/fl and miR-21^Δmyel determined by CFU counts 18h after sepsis. (E, F) AST (liver) and CKMB (heart) damage markers measured in the serum of miR-21^fl/fl and miR-21^Δmyel 18h after sepsis, as described in the methods. (D-F) Data are from n = 7 mice per group and were analyzed by t-test. Violin plots show the frequency distribution of the data. For the appropriate panels, *p < 0.05 compared to sham mice or miR-21^fl/fl CLP (control) in all experiments.

Genetic disruption of miR-21 in macrophages and neutrophils improve survival in septic mice

The role of miR-21 during LPS-induced endotoxemia or sepsis is inconclusive (29). The cellular and molecular targets by which miR-21^−/− improves animal survival, and the role of miR-21 in aberrant and damaging metabolism during sepsis remains unresolved. To define a role for myeloid-based expression of miR-21 in sepsis, we performed CLP in both miR-21^fl/fl and miR-21^∆myel. We found that while miR-21^fl/fl (WT) mice are susceptible to sepsis (80% mortality), miR-21^∆myel are protected (25% mortality) (Figure 1C). These findings correlate with a reduced bacterial burden in blood and the peritoneal cavity of miR-21^∆myel septic mice compared with septic WT control (Figure 1D). Next, we determined whether decreased bacterial burden correlated with lower organ damage. We observed that miR-21^∆myel showed reduced production of AST (aspartate transaminase) and CK-MB (creatine kinase-MB), a hepatic and cardiac marker of tissue damage, respectively (Figure 1E, F). Together these data show that myeloid miR-21 expression is detrimental during sepsis and negatively impacts animal mortality and bacterial load.

miR-21 regulation of cytokine production drives mortality during sepsis.

Mice mortality during sepsis is controlled by early and exaggerated production of inflammatory cytokines, followed by impaired microbial clearance at later time points (9). Since we did not detect any differences in antimicrobial effector functions in vitro (data not shown), we focused our study on the role of miR-21 in inflammation-induced mortality in sepsis. We challenged miR-21^fl/fl and miR-21^∆myel mice with either lethal or sublethal doses of LPS and our data show that miR-21 deficiency also protected mice from LPS-induced septic shock (Figure 2A). Next, we determined the production of local and systemic cytokines in LPS-challenged and CLP mice. We observed that myeloid-specific disruption of miR-21 genetic locus led to a decreased production of IL-6, IL-1β, and TNF-α in the serum of both LPS and CLP septic mice (Figure 2B, D). However, we observed increased IL-10 in the serum and peritoneal exudate of miR-21^∆myel septic mice (Figure 2B, C and D). Therefore, these data show that miR-21 is an important regulator of the systemic inflammatory milieu during sepsis.

Figure 2: Myeloid-based expression of miR-21 balances pro- and anti-inflammatory cytokines levels in septic shock.

(A) Survival rate of miR-21^fl/fl and miR-21^Δmyel mice challenged i.p. with lethal (10mg/kg) and sub-lethal (5mg/kg) doses of LPS. The survival curve was determined for 7 days (n=5/7 mice per group). *p < 0.05 versus LPS-treated miR-21^fl/fl analyzed by log-rank (Mantel-Cox) test. (B) Concentrations of IL-1β, TNF-α, IL10, and IL-6 in the serum of LPS-challenged miR-21^fl/fl and miR-21^Δmyel mice by ELISA. (C) IL-10 levels in the peritoneal exudate of LPS-challenged miR-21^fl/fl and miR-21^Δmyel was determined by ELISA. Data are from n= 4–7 mice per group and analyzed by t-test. D) Concentrations of IL-1β, TNF-α, IL10, and IL-6 in the serum of miR-21^fl/fl and miR-21^Δmyel septic mice determined 18h after CLP by ELISA. Data are from n = 4–7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. Violin plots show the frequency distribution of the data. For the appropriate panels, *p < 0.05 compared to miR-21^fl/fl septic mice in all experiments.

miR-21 induces the production of pro-inflammatory cytokines in BMDM and peritoneal macrophages.

Here, we began investigating the mechanisms by which miR21 regulates systemic inflammatory response during sepsis. Our data show that miR-21 expression is required for the Il1b, Il6, and Tnfa mRNA expression, but not for Nos2 mRNA expression in total peritoneal cells from septic mice (Figure 3A). To start investigating the mechanisms underlying miR-21 actions, initially, we compared the time-dependent miR-21 effects in LPS response in macrophages derived from the bone marrow or peritoneal cavity. We found that LPS-challenged BMDM deficient miR-21 showed decreased Il1b, Il6, and Tnfa mRNA expression (Figure 3B, C, D) and protein (Figure 3E, F, G) at different time points when compared to LPS-challenged WT BMDMs. Likewise, miR21^−/− resident PMs also showed decreased LPS-enhanced IL-6 production (Supplementary Figure 1). However, no differences were noted in TNF-α production (data not shown). Since miR-21 is required for cytokine production during LPS challenge of both BMDMs and PMs, we have decided to proceed the next experiments with BMDM for two reasons: 1) during sepsis, the majority of peritoneal macrophages are derived from monocytes and therefore the bone marrow; 2) the fact that we observed a similar dependency of miR-21 in both LPS-challenged BMDMs and peritoneal cells from septic mice production of inflammatory mediators, along with the higher cell numbers and low cell variability of BMDMs, we decided to keep using BMDMs for our next studies. Together these data suggest that miR-21 is a global regulator of inflammatory cytokine production in different population of macrophages.

Figure 3: miR-21 locus controls LPS-induced cytokine production in phagocytes.

(A) Fold change gene expression of Tnfα, Il1b, Il6, and Nos2 in peritoneal cells 12h after CLP or sham control was determined by qPCR. Data are from n = 4–7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. (B-D) Fold change gene expression of Tnfα, Il1β, and Il6 mRNA in vehicle or LPS-challenged miR-21^fl/fl and miR-21^Δmyel BMDM overtime (6, 12, and 24h) determined by qPCR. (E-G) The concentration of TNF-α, IL-1β, and IL-6 in the supernatant in vehicle or LPS-challenged miR-21^fl/fl and miR-21^Δmyel BMDMs for 24h determined by ELISA. For B-G, data are from 2 mice per group/experiment from at least 3 independent experiments and were analyzed by one-way ANOVA, followed by Bonferroni correction. Violin plots show the frequency distribution of the data *p < 0.05 compared in vivo by miR-21^fl/fl mice or in vitro in LPS-challenged miR-21^fl/fl BMDM (control) or unchallenged.

miR-21 is required for the global expression of genes involved in the inflammatory response.

Given the substantial and conflicting body of literature showing the effects of miR-21 in macrophages, we sought to determine what are the potential candidate genes involved in improved survival and decreased inflammation in miR-21^∆myel using RNAseq analysis in both naïve and LPS-challenged BMDM from miR-21^fl/fl and miR-21^∆myel. Our data showed that more genes were downregulated than upregulated in macrophages from miR-21^∆myel in both basal levels and 6 h after LPS challenge (Figure 4A and Figure 4B). However, we observed an increase in gene clusters involved in cytokine production and responses, hypoxic responses, and cell metabolism. We also detected differential regulation of gene clusters in vehicle and LPS-challenged miR-21^−/− BMDM. In naïve miR-21 deficient BMDMs, we observed a differential expression of genes involved in cell homeostasis, such as the cell cycle, DNA replication, and oxidative phosphorylation. When cells were challenged with LPS, we found that genes that belong to cluster 6 are increased in both naïve, and LPS challenged miR-21^−/− BMDMs. The relevant genes in this group are involved in PI3K, HIF-1α, and TNFα signaling, as well as antigen presentation, and phagosome formation (Figure. 4B). Furthermore, pathway analysis of LPS-challenged BMDMs showed regulation of genes involved in metabolic programs and inflammatory cytokine generation (Figure 4C and D). When we performed a gene enrichment analysis of the top 50 regulated genes, we detected extensively decreased mRNA expression in macrophages from miR-21^∆myel of genes involved in cell homeostasis, tissue homeostasis, repair, and metabolisms, as well as inflammatory response, indicating that miR-21 is required for overall cell homeostasis (Figure 4D). We also observed a decreased expression of inflammatory and repair genes in miR-21 deficient macrophages. Overall, these data suggest that miR-21 is a crucial mediator of macrophage homeostasis and a unique component of LPS responses and mortality during sepsis.

Figure 4: miR-21 regulates gene clusters involved in cell metabolism and cytokine responses in macrophages.

BMDMs from miR-21^fl/fl and miR-21^Δmyel were challenged with LPS or vehicle control for 6 h, and RNAseq performed as described in the Methods. A) Heat-map of genes up-or down regulated. Genes were clustered using a K-cluster tool (green bars on the right). B) Expression of the top 50 up or downregulated genes in BMDMs as in A. C) KEGG analysis of gene clusters as in A. The X-axis shows the number of clusters and genes/clusters. D) Pathway analysis of genes from BMDMs as in A. Data are from n = 4–7 mice per group and analysis was performed as described in Methods.

Genetic disruption of miR-21 decreases aerobic glycolysis in macrophages

Exaggerated glycolysis drives the expression of inflammatory cytokines, enhances tissue injury and mortality during sepsis (17, 38). Using RNAseq analysis, we observed a reduction of HIF-1α expression (a master regulator of glycolysis) in BMDMs from miR-21^Δmyel at homeostasis and in LPS-challenged macrophages, when compared to the BMDMs from miR-21^fl/fl mice. We therefore speculated that genetic miR-21 deletion would reduce glycolytic capability in macrophages challenged with LPS. Using the Seahorse assay, we detected lower basal glycolysis and glycolytic capacity (as measure by the ECAR) in LPS-challenged miR-21^−/− versus LPS-challenged WT macrophages, in both BMDMs and peritoneal macrophages (Figure 5A, B, Supplementary Figure 3A, B). These data led us to speculate that miR-21 regulates the mRNA expression of enzymes involved in glycolysis that could culminate in decreased glycolytic capacity of macrophages. Therefore, we aimed to determine if changes in glycolysis were due to alterations in the expression of key glycolytic enzymes in macrophages from miR-21^Δmyel mice. Peritoneal cells from miR-21^Δmyel septic mice showed decreased mRNA expression for many glycolytic enzymes (Hk1, Pkm, and Ldha), glucose, and lactate transporter (Slc2a1 and Mct4) compared to cells from WT septic mice (Figure 5C). Moreover, we also observed decreased expression of glycolysis-related genes in miR-21^Δmyel BMDM challenged with LPS (Supplementary Figure 3C). We also detected decreased HK1, PDHK1, GLUT1 protein expression, and in both LPS-challenged miR-21^−/− BMDMs (Figure 5D) and CD11b⁺ peritoneal cells from miR-21^Δmyel septic mice (Figure 5E and Supplementary Figure 3D). Since HIF-1α is a key transcription factor involved in the expression of glycolytic enzymes, we sought to determine whether miR21 regulates the expression of HIF-1α in macrophages. Our data show that HIF-1α expression is decreased in peritoneal macrophages from miR-21^Δmyel septic mice and in miR-21^−/− BMDM in vitro (Figure 5C, D, and E).

Figure 5: Myeloid-miR-21 locus deletion decreases aerobic glycolysis in vivo and in vitro.

(A) Real-time ECAR evaluation of glycolysis in vehicle or LPS-challenged miR-21^fl/fl and miR-21^Δmyel using Seahorse XF glycolysis stress test. (B) ECAR of glycolytic capacity (ECAR) after LPS in miR-21^fl/fl and miR-21^Δmyel BMDMs. Data are from n = 1–2 mice per group/experiment from at least 3 independent experiments and were analyzed by one-way ANOVA, followed by Bonferroni correction. (C) Hif1a, Scl2a1, Hk1, Hk2, Pkm, Lhda, and Slc16a3 mRNA expression (fold change relative to sham) determined in peritoneal cells from miR-21^fl/fl and miR-21^Δmye septic and sham mice 18h after CLP by qPCR. Data are from n = 4–7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. (D) Expression of HIF-1α, GLUT-1, HK1, HK2, PKM, PDHK1, LDHA, and βActin in LPS-challenged miR-21^fl/fl and miR-21^Δmyel BMDMs min the indicated time points by immunoblotting. Numbers underline represents the densitometric quantification of two independent experiments. (E) Expression of HIF-1α, HK1, and GLUT1 in CD11b⁺ peritoneal cells from septic mice by flow cytometer. (F-G) Pyruvate and lactate measured in cells lysate and supernatant of the peritoneal cavity by colorimetric assay, respectively. Data are from n = 4–7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. Violin plots show the frequency distribution of the data. *p < 0.05 compared ex vivo with miR-21^fl/fl mice or in vitro with miR-21^fl/fl BMDM LPS (control). ^#p < 0.05 compared in vitro with miR-21^fl/fl BMDM unchallenged.

Since the ratio of pyruvate/lactate is decreased in cells undergoing aerobic glycolysis, leading to pyruvate-dependent lactate production (39), we evaluated the abundance of pyruvate and lactate in peritoneal cells from septic mice 18h after CLP. We found lower levels of lactate in the peritoneal exudate and higher amounts of pyruvate in peritoneal cells from miR-21^Δmyel septic mice when compared to WT septic animals (Figure 5F, G). Together, our findings show that myeloid-miR-21 genetic locus disruption prevents aerobic glycolysis and lactate generation during sepsis.

miR-21 regulation of prostaglandin E₂ production dictates the production of inflammatory mediators in vivo and in vitro.

We and others have shown that PGE₂ inhibits the production of pro-inflammatory mediators, while enhances the generation of anti-inflammatory actions of macrophages (40, 41). Here, we speculated that PGE₂ is a central mediator of the miR-21 effects during sepsis/endotoxemia. when we determined the abundance of PGE₂ in the serum of miR-21^flfl and miR-21^Δmyel mice during endotoxemia, we observed increased circulating PGE₂ in miR-21^Δmyel septic mice (Figure 6A). Next, we sought to determine whether miR-21 regulates macrophage eicosanoid production in LPS-challenged BMDMs. We confirmed that PGE₂ production is increased in miR-21 deficient macrophages; however, the production of other eicosanoids, such as prostaglandins D₂ and F2α (PGD₂, PGF_2α) and thromboxane A2 (TxA2) and the oxylipids 12,13-DiHOME and 15 HETE were unaffected (Figure 6B). Increased PGE₂ levels also correlate with greater amounts of arachidonic acid (AA) in miR-21 deficient LPS-challenged BMDMs (Figure 6C). Next, we aimed to determine whether miR-21 regulates the expression of biosynthetic enzymes responsible for AA release and PGE₂ production. Interestingly, we detected upregulation of a single gene involved in AA release from the membranes (Pla2g2a - encodes Phospholipase A2 group IIA [PLA2]) and three genes involved specifically in PGE₂ synthesis in miR-21^−/− BMDMs: Ptgs1 and Ptgs2, encoding cyclooxygenases 1 and 2 (COX-1 and 2), respectively; and Slco2a1, encoding the prostaglandin transporter (PGT) (Figure 6D). Together these data suggest that increased PGE₂ expression in miR-21deficient macrophages is due to a pleiotropic regulation of enzymes involved in multiple steps of the PGE₂ biosynthetic pathway.

Figure 6: miR-21 regulates PGE₂/IL10 axis in macrophages.

(A) PGE₂ levels measured in the serum of mice challenged with LPS miR-21^fl/fl and miR-21^Δmyel mice for 12h. Data are from n = 5 mice per group and were analyzed by t-test and Mann-Whitney U test. (B and C) Arachidonic acid (B) and eicosanoid (C) levels measured in the supernatant of miR-21^fl/fl and miR-21^∆myel BMDMs challenged or not with LPS (10ng/mL) for 6h by LC-MS. Data are from n = 3 mice per group and were analyzed by t-test and Mann-Whitney U test. (D) Pla2g4a, Pla2g2a, Pla2g6, Ptgs1, Ptgs2, Ptges1, Ptges2, Slco2a1 and Hpgd mRNA expression (fold change relative to unchallenged cells) of miR-21^fl/fl and miR-21^∆myel BMDMs challenged or not with LPS (10 ng/mL) for 6h by qPCR. (E) The abundance of IL-10 was measured in the supernatant of miR-21^fl/fl and miR-21^Δmyel BMDM treated with aspirin (ASA, 100 μM), Indomethacin (IDO, 10 μM) and CAY10526 (CAY, 1 μM) one hour before challenge with LPS (10 ng/mL) for 6h by ELISA. (F, G) The concentration of IL-6 and TNF-α was determined in the supernatant of miR-21^fl/fl and miR-21^Δmyel BMDM treated with CAY10526 (1 μM) or DMSO (0.1%), followed by challenge with LPS for 6h by ELISA. B, D-E data are from at least 3 independent experiments and were analyzed by one-way ANOVA, followed by Bonferroni correction. Violin plots show the frequency distribution of the data *p < 0.05 compared in vivo by miR-21^fl/fl mice or in vitro in LPS-challenged miR-21^fl/fl BMDM (control). ^#p < 0.05 compared in vitro by miR-21^fl/fl BMDM untreated (UT) group.

To determine whether increased basal PGE₂ is involved in differential production of pro- and anti-inflammatory cytokines in miR21−/− cells, we incubated BMDMs from miR-21^fl/fl and miR-21^Δmyel with the specific microsomal prostaglandin E synthase-1 (mPGES-1) inhibitors CAY10526, as well as different COX inhibitors (aspirin and Indomethacin) followed by LPS challenge. Our results showed that PGE₂ inhibition decreased IL-10 abundance in miR-21^Δmyel macrophages (Figure 6E), and that this effect was accompanied by increased pro-inflammatory cytokines production miR-21^Δmyel BMDMs (Figure 6F and G). Thus, these data suggest that miR-21 drives a PGE₂/IL-10 axis to potentially regulate the abundance of inflammatory cytokines in LPS-treated cells.

Enhanced IL-10 production drives protective effects in LPS-challenged miR-21^Δmyel mice

IL-10 is classically considered an anti-inflammatory cytokine (42), and more recently, Ip et al. have shown that the inhibitory effect of IL-10 is dependent on the inhibition of LPS-induced glucose uptake and glycolysis and increased oxidative phosphorylation (43). Here, we are investigating whether miR-21-dependent IL-10 production is the primary driver of endotoxin resistance in miR-21^Δmyel. We treated miR-21^Δmyel BMDMs with an IL-10R blocking antibody, as we have previously shown (32), followed by LPS challenge for 24 h,. Blocking IL-10 actions rescued the production/secretion of TNF-α and IL-6 in LPS-challenged miR-21^Δmyel BMDM. These data indicate that increased IL-10 is responsible for low production of inflammatory cytokines in miR-21^Δmyel macrophages (Figure 7A). Next, we tested if blocking IL-10R would render miR-21^Δmyel mice susceptible to endotoxin. anti-IL-10R treatment increased mortality to endotoxin in LPS-challenged miR-21^Δmyel mice (Figure 7B), as well as restored production of IL-6, TNF-α, and IL-1β in the serum of miR-21 deficient mice (Figure 7C, D, and E). Importantly, blocking IL-10R in vitro also restored glycolysis in BMDM from LPS-challenged miR-21^Δmyel mice (Figure 7F). Together, these data show a heretofore-unknown link between miR-21, PGE₂, IL-10 and glycolysis in macrophage inflammatory response and resistance to sepsis/endotoxemia in mice.

Figure 7: IL-10 receptor blockage prevents the anti-inflammatory effects of myeloid-specific miR-21 deletion.

(A) The concentration of TNF-α and IL-6 levels in the supernatant of miR-21^Δmyel BMDM treated with anti-10R (10 μg/mL) or PBS (Vehicle) 1-hour prior to LPS challenge for 12 hours by ELISA. Data are from n = 2 mice per group/experiment from at least 3 independent experiments and were analyzed by one-way ANOVA. (B) The survival rate of miR-21^Δmyel mice challenged with a lethal dose of LPS (10 mg/kg) and treated with anti-10R (10 mg/kg) or PBS (Vehicle) 1 hour before and 24h after LPS i.p injection. The survival curve was analyzed for 7 days (n=5/7 mice per group) *p < 0.05 versus LPS/PBS treated miR-21^Δmyel analyzed by log-rank (Mantel-Cox) test. (C-E) The concentration of IL-1β, IL-6, and TNF-α in the serum of miR-21^fl/fl and miR-21^Δmyel mice as in A. Cytokines were measured 24h after challenge with LPS by ELISA. Data are from n = 5 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. (F) Real-time ECAR evaluation of glycolysis in miR-21^fl/fl and miR-21^Δmyel BMDMs treated with anti-10R (10 μg/mL) or PBS (Vehicle) 1-hour prior, followed by challenge with LPS for 12 hours by Seahorse XF glycolysis stress test. Data are from n = 2 mice per group/experiment from at least 3 independent experiments and were analyzed by one-way ANOVA. Violin plots show the frequency distribution of the data *p < 0.05 compared by miR21^fl/fl BMDM LPS (control) or vehicle.

Discussion.

The outcome of sepsis is dictated by the impact of macrophage-induced production of inflammatory mediators along with the loss of structural cells and the release of danger-associated molecular patterns (44). Macrophages are exquisite cells in their capacity to adapt to environmental clues; however, intracellular regulatory mechanisms must be in place to prevent macrophage hyperactivation and cause tissue injury (45). Therefore, macrophage pro-inflammatory programs need to be adequately controlled at multiple levels, including transcriptional and posttranscriptional gene expression, post-receptor signaling events, and generation of metabolites that allow cells to adapt and respond to environmental cues (45, 46). Herein we are employing complementary approaches to unveil a unique requirement of myeloid expression of miR-21 in controlling a cytokine-dependent metabolic shift resulting in the overwhelming inflammatory response, organ damage, and death in septic mice.

In summary, our findings show: 1) miR-21 is highly expressed in macrophages and neutrophils after sepsis; 2) disruption of the miR-21 genetic locus in myeloid cells enhance animal survival, decreases bacterial load, cytokine storm and organ injury; 3) deletion of miR-21 is accompanied by low glycolysis and decreased expression of glycolytic enzymes; 4) miR-21 inhibits the release of AA and PGE₂ production that balances the production of pro-inflammatory cytokines and IL-10 abundance. 5) In vivo and in vitro IL10R blockage restores cytokine production and glycolysis and enhanced mortality during endotoxin shock in miR-21^Δmyel.

miR-21 is one of the most abundant microRNAs expressed during LPS challenge, cytokine stimulation (TNF-α, IL-1β, IL-6, TGF-β, GM-CSF), and lipid responses (PGE₂ and resolvin A1) (23). miR-21 is upregulated in septic patients and mice (19). The role of myeloid-miR-21 in LPS-induced endotoxemia or sepsis is inconclusive. The miR-21 global knockout mice are more resistant to sepsis by controlling IL-1β dependent inflammasome activation than septic WT mice (30). These findings also corroborated with McClure et al. that showed that treatment of BALB/c mice with a miR-21 antagomiR enhances animal survival and decreases bacterial load after CLP (29). Delivery of a miR-21 antagomiR was cardioprotective in septic mice by increasing the number of myeloid suppressor cells (28). On the other hand, Barnett et al. showed that miR-21 KO mice are more susceptible to LPS-induced septic shock. Still, they did not identify any differences in animal mortality after CLP-induced sepsis (31). The reason for the discrepant data mentioned above might stem from different factors, 1) animal strains used, 2) the severity of sepsis, 3) source and quantity of LPS tested, 4) degree of gene silencing/inhibition using miR-21 antagomir, as well as genetic and pharmacological approaches and global vs. cell-specific miR-21 knockdown. Here, we are not only confirming the protective effect of miR-21 locus disruption during sepsis and endotoxemia but also moving the field forward by dissecting the importance of the miR-21 expression in macrophages (the critical cells involved in cytokine storm and organ damage during trauma) in the outcome of sepsis. Further studies are needed to determine whether the timing of treatment of mice with miR-21 antagomir or mimic after the induction of CLP will also play a role in the outcome of sepsis.

There is no common scientific consensus about the exact mechanism of miR-21 in the induction or repression of the inflammatory program. Several manuscripts suggest miR-21 as an anti-inflammatory factor (31, 47, 48), while others suggested this miRNA as a crucial pro-inflammatory factor (26, 30). We observed an overall decrease in IL-6 and IL-1β and an increase in IL-10 abundance in septic miR-21^∆myel when compared to septic WT mice. Changes in cytokine production could be explained by the overall changes in metabolic changes in HIF-1α-dependent expression of glycolytic enzymes in macrophages of both people and mice with sepsis (17, 38). Increased expression of individual glycolytic enzymes is also associated with enhanced mortality in sepsis (7, 38). Here, we observed decreased HIF-1α, GLUT-1, and HK-1, along with lower ECAR response and cytokine production in miR-21^∆myel mice. If decreased HIF-1α expression is the primary gene responsible for overall lower glycolysis in miR-21^∆myel mice remains to be determined. Although lower glycolysis in miR-21^−/− macrophages is not in agreement with the data from Hackett et al.,(49) that shows increased expression of PFKM, ECAR, and production of inflammatory cytokines, other reports have demonstrated that miR-21 deficiency leads to decreased HIF-1α-dependent glycolysis in different cancer cells (50, 51). The reason for the discrepancy between our findings and Hackett, et al., remains to be thoroughly dissected, but possible differences include the method used to differentiate BMDM (M-CSF and GM-CSF for 6 days) vs. 5% L929 supernatant throughout the experiments, and therefore, these cells are potentially different.

It has been shown that PGE₂ inhibits IL-1β and TNF-α, while enhances IL-10 production in macrophages (52). Also, PGE₂ decreases glycolysis in different cell types (53, 54). Therefore, we speculated that PGE₂ is a central hub for miR-21 actions in macrophage glycolysis and cytokine. We observed increased PGE₂ levels in vivo and in vitro resulting of the genetic disruption of miR-21 in myeloid cells. Herein, we have shown the expression of miR-21 in BMDMs reduces the mRNA expression for crucial enzymes involved in PGE₂ biosynthesis, as Pla2g2a (PLA₂), Ptgs (COX1) and Ptgs2 (COX-2), as well as Slco2a1 (PGT), but not Ptges and Ptges2. These results suggest that miR-21 inhibits PGE₂ production by downregulation AA levels and PGT expression are sufficient to impair PGE₂ production and secretion in WT macrophages. The inhibition of PGE₂ synthesizing enzymes decrease IL-10 and increase IL-6 and TNF-α, suggesting that miR-21 regulates the balance between PGE₂ and IL-10, controlling macrophage proinflammatory activation. None of the mRNAs involved in PGE₂ biosynthesis and secretion that are inhibited by miR-21 in BMDMs were previously predicted as targets of miR-21. That miR-21 can target the mRNA for 15-PGHDS (Hpgd, gene), enzyme that degrades PGE₂, has been suggested in cancer models (55, 56). Here, we did not see any difference in Hpgd mRNA and 15-PGHDS expression in macrophages from WT and miR-21^∆myel mice (not shown). Whether miR-21 mimics enhances 15-PGHDS expression in macrophages remain to be determined. Therefore, it is possible that miR-21 might be directly targeting mRNAs that controls the expression of PLA₂, COX-1/2 and PGT.

In the present study we are unfolding the new events that are controlled by endogenous miR-21 expression that culminates in PGE₂-mediated increased IL-10 production and inhibition of proinflammatory cytokine production in macrophages, including IL-6. We speculated that low IL-6 abundance dictates the overall decreased LPS responses observed in miR-21 deficient macrophages. Therefore, we performed “add back” experiments to determine if exogenous IL-6 rescues LPS response in miR-21^−/− macrophages. Intriguingly, the challenge of miR-21^−/− BMDM with exogenous recombinant IL-6 did not rescue LPS responses (Supplementary Figure 2 B, C) what suggest the IL-6R signaling is not involved in miR-21 actions. We then hypothesized that IL-10 is the final mediator of miR-21 regulation in macrophages. Sheedy et al. have shown that miR-21 enhances IL-10 abundance by directly controlling the expression of the IL-10 inhibitor PDC4 (25). IL-10 expression is regulated by different transcription factors, including STAT3 and CREB (57). We and others have shown that PGE₂ enhances IL-10 production via cyclic AMP-dependent protein kinase A (PKA)-mediated CREB phosphorylation in macrophages (40, 41, 52). IL-10 is a pleiotropic inhibitor of the inflammatory response. It is well established that IL-10 inhibits NF-kB, STAT1, and AP1 activation by mechanisms that are dependent or independent on the STAT3/SOCS3 axis (58, 59). Here, we showed that deficiency of miR-21 enhances IL-10 that in turn decreases glycolysis-mediated proinflammatory cytokine production in LPS-challenged BMDMs. There are four hypothetical models to explain how IL-10/STAT3 can inhibit inflammatory gene transcription: 1) via induction of transcription repressor of NF-kB; 2) gene-specific silencing at the chromatin level; 3) posttranslational modification of proinflammatory transcriptional factor; 4) via sequestration of key pro-inflammatory transcriptional mediators (58, 59). Interestingly, STAT3 induces the expression of miR-21, which, in turn, downregulates the translation of important proinflammatory factors like NF-kB and PDCD4 (60). In addition, our group showed miR-21 deficient macrophages have a higher expression of STAT3 (31). Moreover, IL-10/STAT3 impairs mTORC1 activation inhibiting LPS-induced glucose uptake, aerobic glycolysis and IL-1β production while eliminating damaged mitochondria in macrophages (61).

All these finding strongly support our hypothesis that miR-21/PGE₂/IL-10 axis controls aerobic glycolysis and inflammatory profile of macrophages, which is critical for sepsis outcome. In summary, we are unveiling a heretofore-unknown program by which a single microRNA regulates the actions of different classes of molecules, including lipids, protein, and metabolites that influence the outcome of an exaggerated inflammatory response triggered during sepsis.

Key points:

• Myeloid-based miR-21 expression drives the hyperinflammatory state in sepsis.

• miR21 inhibits macrophage glycolysis in a manner dependent on PGE₂/IL10 axis

• Enhanced IL-10 production leads to protective effects in miR-21^Δmyel septic mice.