A Novel Mechanism of Mesenchymal Stromal Cell-Mediated Protection against Sepsis: Restricting Inflammasome Activation in Macrophages by Increasing Mitophagy and Decreasing Mitochondrial ROS

Sepsis, a systemic inflammatory response to infection, is the leading cause of death in the intensive care unit (ICU). Previous studies indicated that mesenchymal stromal cells (MSCs) might have therapeutic potential against sepsis. The current study was designed to investigate the effects of MSCs on sepsis and the underlying mechanisms focusing on inflammasome activation in macrophages. The results demonstrated that the bone marrow-derived mesenchymal stem cells (BMSCs) significantly increased the survival rate and organ function in cecal ligation and puncture (CLP) mice compared with the control-grouped mice. BMSCs significantly restricted NLRP3 inflammasome activation, suppressed the generation of mitochondrial ROS, and decreased caspase-1 and IL-1β activation when cocultured with bone marrow-derived macrophages (BMDMs), the effects of which could be abolished by Mito-TEMPO. Furthermore, the expression levels of caspase-1, IL-1β, and IL-18 in BMDMs were elevated after treatment with mitophagy inhibitor 3-MA. Thus, BMSCs exert beneficial effects on inhibiting NLRP3 inflammasome activation in macrophages primarily via both enhancing mitophagy and decreasing mitochondrial ROS. These findings suggest that restricting inflammasome activation in macrophages by increasing mitophagy and decreasing mitochondrial ROS might be a crucial mechanism for MSCs to combat sepsis.


Introduction
Sepsis syndrome, the inflammatory response to infection, is one of the leading causes for death in hospitalized patients [1]. Complications of sepsis, such as acute respiratory distress syndrome (ARDS) and multiple organ dysfunction syndrome (MODS), are major causes for morbidity and mortality in critical patients. Sepsis-associated mortality in the intensive care unit (ICU) is extremely high, with rates of 20% for sepsis, 35% to 45% for severe sepsis, and 60% for septic shock [2]. Therefore, it is essential to identify effective therapeutic strategies for sepsis.
The pathophysiologic process for sepsis is complicated, including invading microorganisms, proinflammatory response, anti-inflammation, and associated immunoparalysis [3,4]. Extensive data have demonstrated that mesenchymal stromal cells (MSCs), a type of adult stem-like cells, are capable of inhibiting inflammation and immunity responses [5]. Therefore, stem cells have drawn growing interest in the treatment of inflammatory diseases. Of note, it has been reported that MSCs can improve organ function and decrease mortality [6,7] as well as reduce sepsis-induced inflammation in an animal sepsis model induced by cecal ligation and puncture (CLP) [8]. However, the underlying mechanism is poorly understood, which vastly limits the therapeutic potential of cytotherapy. Previous studies have shown that mitophagy is a prosurvival mechanism associated with cellular exposure to various mitochondrial stressors [9]. Recent studies have indicated that excessive reactive oxygen species (ROS) production is involved in activating mitophagy [10,11]. It is also well established that the cytosolic E3 ubiquitin ligase, Parkin, and the outer mitochondrial membrane kinase, PTEN-induced putative kinase 1 (PINK1), are two main regulators of mitophagy in mammalian cells [9,12,13].
Based on previous reports, we hypothesized that bone marrow-derived mesenchymal stem cells (BMSCs) could reduce sepsis-associated inflammation and organ dysfunction in CLP-induced sepsis. We further investigated the immune-modulatory effects of BMSCs on bone marrowderived macrophages (BMDMs) to determine the mechanisms for the beneficial effects of BMSCs against sepsis.

Materials and Methods
2.1. Ethics. All animal procedures were conducted in conformity with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The experimental protocol was approved by the Chinese PLA General Hospital and Fourth Military Medical University Committee on Animal Care.

Animal Study Protocol and CLP Procedures.
Transgenic C57BL/6-Tg (CAG-EGFP)1Osb/J mice (C57BL/6, 8-10 weeks, 20-24 g), which continuously express eGFP in all tissues and organs, were commercially purchased from Jackson Laboratory (stock number 003291). Wild-type C57BL/6 mice (8-10 weeks, 20-24 g) were from Animal Laboratory of Fourth Military Medical University. All animals had free access to water and food during the experiment. C57BL/6 mice that underwent the CLP operation were randomly divided into three groups depending on the intravenously injected solution at 1 h postoperation (n = 20 in each group), the CLP + saline group (injected with 100 μl of saline); CLP + Fbs group (injected with 2.5 × 10 5 fibroblasts in 100 μl); and CLP + BMSC group (injected with 2.5 × 10 5 BMSCs in 100 μl). The fate of BMSCs and myocardial function were detected 24 h after the operation. Visualization of BMSCs in C57BL/6 mice liver tissue at 1-6 h after intravenous injection was shown in supplemental Figure S1C. The survival rate was analyzed every 6 h after the operation until 96 h.
CLP was performed according to the protocol proposed by Rittirsch et al. [14]. Briefly, C57BL/6 mice were anesthetized with persistently inhaled 2% isoflurane during the operation. The abdominal zone was shaved and prepared with 70% ethanol. A ventral midline incision (approximately 1 cm) was made, and the cecum was exteriorized. The cecum was ligated at the designated position with silk suture and penetrated through-and-through with a 22 G needle. Then, the abdominal incision was closed. In the sham group, the cecum was exteriorized, without being ligated or punctured. Immediately after surgery, animals were subcutaneously injected with 1 ml/30 g saline and imipenem-cilastatin for fluid resuscitation and infection prevention.
2.3. Cardiac Function Evaluation by Echocardiography. Echocardiography was performed using the Vevo 2100 ultrasound system (Visual-Sonics, Toronto, Canada) with a 30 MHz linear transducer. Anesthesia was conducted with persistently inhaled 1.0% isoflurane. Data of the left ventricular enddiastolic diameter (LVEDd), left ventricular end-systolic diameter (LVESd), and left ventricle (LV) internal dimension in diastole (LVID, d) and systole (LVID, s) were measured. The left ventricular ejection fraction (LVEF) and fractional shortening (FS) were calculated accordingly. The whole procedure was performed by two blinded investigators.
Mice were sacrificed, and femurs were removed and cleansed of tissue. Marrows were flushed from the femurs with PBS and collected by centrifugation (200g, 5 min). Cells were cultured at the density of 5 × 10 5 /ml in culture dishes with DMEM medium supplemented with 10% (vol/vol) FBS, 1% (vol/vol) penicillin and streptomycin. Following 6 h of incubation at 37°C in 5% CO 2 , nonadherent cells (primary bone marrow-derived macrophages) were decanted and seeded in plates and incubated in complete medium with 25% (vol/vol) conditioned medium from L929 mouse fibroblasts for 7 days (over 90% cells are positive for the cell type surface marker CD11b) to form proliferative nonactivated cells (M0 macrophages) [16,17].
2.6. Immunohistochemical Staining. Immunohistochemical staining was performed to detect inflammatory cell infiltration. Tissue sections were blocked with 5-10% normal goat serum for 30 min and then incubated with monoclonal antibodies of anti-Ly-6G and anti-Mac-3 (from Biolegend, San Diego, CA, USA) overnight at 4°C. Then, sections were washed and incubated with secondary horseradish peroxidase-conjugated goat anti-rabbit antibodies (from Zhongshan Biotechnology Co. Ltd., Beijing, China) at 37°C for 1 h. Sections were randomly selected, and images were visualized and photographed with inverted or confocal microscopy (Olympus, Japan).

Western Blot.
Western blot was performed as previously described [18]. Briefly, proteins were harvested from cells or tissues with RIPA Lysis Buffer (Beyotime Biotechnology, Beijing, China). Total proteins were loaded onto SDS-PAGE gels and transferred electrophoretically to PVDF membranes (Millipore, Billerica, MA). After blocking with 5% skim milk for 1 h, the membranes were incubated with primary antibody at 4°C overnight. Afterwards, membranes were washed and incubated with corresponding secondary antibody at 37°C for 1 h. The blots were developed with enhanced chemiluminescence (ECL) reagent (Millipore) and visualized using UVP Bio-Imaging Systems. Blot densities were analyzed with Quantity One System Software.

Transmission Electron Microscopy (TEM)
. TEM was performed to observe the mitochondrial morphology. Briefly, collected BMDMs underwent the procedures of fixation, stepwise alcohol dehydration, embedding, polymerization, sectioning, and staining. Images were observed with an electron microscope (JEM-2000EX TEM, JEOL Ltd., Tokyo, Japan). Random sections were visualized by a blinded technician. 2.10. Statistical Analysis. Data were expressed as the mean ± SD and were analyzed using ANOVA followed by a Bonferroni correction for the post hoc t-test. The survival rate was analyzed with the Kaplan-Meier test followed by the log-rank post hoc test. All statistical tests were performed using SPSS software version 17.0 (IBM, Armonk, NY, USA) and GraphPad Prism software version 5.0 (GraphPad Software, San Diego, CA). A value of p < 0 05 was considered statistically significant.

BMSC Treatment Improved Survival Rate and
Multiorgan Functions after CLP. Figure 1(a) shows the survival curve of C57BL/6 mice after CLP. Compared with the CLP + saline and CLP + Fbs groups, the survival rate was significantly increased in the CLP + BMSC group, with survival rates of 10%, 5%, and 40% in the three groups at the time point of 96 h (p < 0 05). As revealed by HE staining, BMSC observably improved morphological changes in vital organs, such as the heart, liver, spleen, lungs, and kidneys, which was manifested by interstitial edema, red blood cell, and inflammatory cell infiltration (Figure 1(b)). In addition, serum biomarkers, including cardiac troponin I (cTnI), creatinine kinase (CK), lactate dehydrogenase (LDH), alanine aminotransferase (ALT), aspartate aminotransferase (AST), amylase, serum creatinine values (SCr), and blood urea nitrogen (BUN), were significantly reduced in the CLP + BMSC group compared with the CLP + Saline and CLP + Fbs groups (P < 0 05), indicating that BMSC treatment improved organ functions after CLP (Figure 1(c)).

BMSCs Reduced the CLP-Induced Inflammation Level.
Inflammatory cell infiltration in the heart and lung tissue was determined by the immunohistochemistry for Ly-6G and MAC-3, which are well-recognized biomarkers for neutrophils and macrophages. It could be inferred that BMSC treatment inhibited inflammatory cell infiltration, as evidenced by decreased macrophage (Mac-3) (Figures 2(a) and 2(c)) and neutrophil (Ly6G) infiltration (Figures 2(b) and 2(c)) in the lung and liver tissues. Furthermore, BMSCs decreased the proinflammatory cytokine levels of IL-6, IL-1β, and TNF-α, while increasing anti-inflammatory cytokine IL-10 in the serum (Figure 2(d), p < 0 05, CLP + BMSC group versus CLP + saline group; p < 0 05, CLP + BMSC group versus CLP + Fbs group).

BMSCs Inhibited NLRP3 Inflammasome Activation in
BMDMs by Decreasing Mitochondrial ROS. Compared with the LPS + ATP-stimulated group, BMSC coculturing reduced the mitochondrial ROS (mtROS) level in BMDMs ( Figure 5(a)). The flow cytometry analysis of MitoTracker Deep Red-MitoTracker Green ( Figure 5(b)) indicated that BMSCs alleviated LPS + ATP stimulation-induced mitochondrial damage in BMDMs. Antioxidant Mito-TEMPO was used to inhibit mtROS generation. Figure 5(c) shows the colocalization of NLRP3 (green), mitochondria (red), and nucleus (blue) by immunofluorescence staining. In control BMDMs, very little NLRP3 cosedimented with mitochondria. However, much more NLRP3 was colocalized with or adjacent to mitochondria after LPS-ATP stimulation. Both BMSCs and Mito-TEMPO resulted in less colocalization of NLRP3 and mitochondria. After LPS and ATP stimulation, Mito-TEMPO treatment decreased the expression levels of caspase-1 p20 and IL-1β p17 (p < 0 01, p < 0 05), suggesting that NLRP3 activation was associated with mtROS level. Additionally, the expression levels of caspase-1 p20 and IL-1β p17 were also reduced in the BMSC-

BMSCs Inhibited NLRP3 Inflammasome Activation by
Increasing BMDM Mitophagy. Using electron microscopy (EM), we directly assessed the mitochondrial integrity and state. We observed more mitophagy (indicated by red arrows) and fewer swollen mitochondria in the BMSCtreated group than for treatment with LPS and ATP in BMDMs (Figure 6(a)). In addition, autophagy-associated LC3 puncta were found to accumulate around the mitochondria in BMDMs when coculturing with BMSCs after stimulation by LPS and ATP (Figure 6  indicating that the inhibitory effects of BMSCs on mtROS generation were mediated by mitophagy (Figure 7(c)).

Discussion
Several studies have indicated the beneficial effects of MSCs against sepsis, but the underlying mechanisms remain unclear. In the present study, we demonstrate that BMSCs exert beneficial effects on experimental sepsis via inhibiting NLRP3 inflammasome activation in macrophages. Furthermore, one important finding in the present study was that BMSCs restricted NLRP3 inflammasome activation through increasing mitophagy activation and decreasing mitochondrial ROS in macrophages.
The antisepsis efficacy of MSCs may be attributed to their ability to home injured tissue, secrete paracrine cytokines, decrease apoptosis in injured tissues, and modulate immune cells [19,20]. It was reported that MSCs attenuated septic injury by reducing the infiltration of inflammatory cells and cell death in various targeted organs [21]. However, Krasnodembskaya et al. determined that human BMSCs attenuated live bacteria-induced injury due to direct antimicrobial activity [22]. Moreover, the same group has published a study on therapeutic potential of human MSC in the model of sepsis, in which they found that MSC effect was associated with increased phagocytic activity of blood monocytes and M2 polarization of monocytes/macrophages in the spleen [23]. Although complex, the acquisition of a mechanistic knowledge is essential for developing BMSC cell therapy against sepsis injury. In the present study, our data showed that BMSCs increased the CLP murine survival rate and alleviated organ injury via improving organ function and alleviating proinflammatory cytokines.
Cardiac function is crucial to the clinical outcomes of sepsis. Therefore, myocardial dysfunction has been gaining increasing attention in recent years [24,25]. In the current study, we observed impaired cardiac function in septic mice by echocardiography and hemodynamic measurements.     [26]. However, the precise mechanism of MSCs' beneficial function in cardiac recovery is far from clear. Data from Rogers et al. revealed that MSCs exerted reparative effects on myocardium through molecular reprogramming of the cardiomyocytes themselves [27]. Notably, studies on sepsis indicate that mitochondria could mediate organ dysfunction, including heart function depression. Considering our data, mitochondrial dysfunction may be a fundamental contributor to cardiac depression in sepsis. However, further studies are still needed to clarify this.
According to previous studies, sepsis is initiated when well-conserved microbial structures (also known as pathogen-associated molecular patterns, PAMPs) bind to receptors embedded either on the cell membranes or inside the cell cytoplasm of cells of the innate immune system, namely, blood monocytes and tissue macrophages. The characteristic of the first phase of sepsis is the production of proinflammatory cytokines, including tumor necrosis factor-alpha (TNF-α), interleukin-(IL-) 1β, IL-6, and IL-8. These proinflammatory mediators orchestrate septic reaction of the host. Soon after this first phase of hyperproduction of proinflammatory mediators, a second phase ensues during which Th2 cells, monocytes, and macrophages stimulated by PAMPs secrete a large amount of anti-inflammatory mediators like IL-10. This phase is considered a state of immunosuppression or immunoparalysis of the host when multiple organ dysfunctions take place [28]. Consistent with previous reports, we found that the inflammatory biomarkers (IL-1β, TNF-α, IL-6, and IL-10), serum markers (cTnI, CK, LDH, ALT, AST, Amylase, Scr, and BUN), and inflammatory cell infiltration were increased in CLP mice compared with those in control mice [7,26,29]. In addition, mitochondrial ROS generation was also elevated in CLP-induced septic mice. These results were consistent with the findings by Chang et al., who proposed that CLP-induced sepsis triggered a rigorous inflammatory response with the generation of ROS [30]. Sepsis is characterized by overwhelming activation of inflammatory and immune responses. An interesting finding in the present study was that BMSCs restricted the inflammatory responses and ROS generation in CLP animals. This is comparable to previous reports. Yip and his colleagues reported that combined therapy with melatonin and apoptotic adipose-derived mesenchymal stem cells alleviated sepsis-induced organ injury through attenuating inflammatory reactions and oxidative stress [2,31]. Notably, the inflammatory cytokines were produced by macrophages rather than injected BMSCs, suggesting that the therapeutic effect of BMSCs against excessive inflammatory responses was mediated by an interaction between BMSCs and macrophages. The in vitro coculture experiments further confirmed this finding.
Inflammasomes are a variety of protein complexes that could recognize inflammation-inducing stimuli and control inflammatory cytokine production [32]. Among them, the NLRP3 (nucleotide-binding domain, leucine-rich-repeatcontaining family, and pyrin domain-containing 3) inflammasome is one of the most widely studied inflammasomes. The NLRP3 inflammasome, consisting of the regulatory subunit NLRP3, adaptor ASC, and effector caspase-1, is a molecular complex that could be activated to trigger immune defenses through secreting inflammatory cytokines such as IL-1β or IL-18 [33,34]. Previous researches showed that the NLRP3 inflammasome was upregulated and activated in the liver during sepsis [35,36]. Our data revealed that NLRP3 inflammasome-mediated and IL-1β activation in septic liver tissue, and BMSCs inhibited NLRP3 activation and resultant IL-1β activation. Although the level of NLRP3 or its adaptor ASC did not reveal a significant alteration, its cellular localization changed. As inferred by our immunofluorescence results, the resting NLRP3 was localized in the endoplasmic reticulum, while under the condition of inflammasome activation, NLRP3 colocalized with endoplasmic reticulum and mitochondrion structures. Mitochondrial ROS was capable of inducing NLRP3 inflammasome activation [37]. As revealed in our data, with the addition of Mito-TEMPO (a mitochondria-targeted antioxidant), the relocalization of NLRP3 inflammasome was further attenuated, in combination with subsequent decreased activation of caspase-1 and IL-1β. Similar to the previous results, we observed that Nacetylcysteine (NAC, a general reactive oxygen inhibitor) abrogated the apparent increase in caspase-1 activation and IL-1β expression in BMDMs in response to LPS and ATP. These results suggested that the protective effect of BMSCs on the NLRP3 inflammasome was mediated by regulating mitochondrial ROS generation.
Maintaining a healthy mitochondrion population is important for cells. Autophagic clearance is the major pathway in mitochondrial turnover, termed as mitophagy [38]. Recently, Mahrouf-Yorgov et al. found that MSCs that engrafted into infarcted hearts of mice could reduce damage via upregulating HO-1 and increasing mitochondrial biogenesis, and inhibition of mitophagy or HO-1 failed to protect against cardiac apoptosis [39]. Activation of the autophagy process has been reported to attenuate cardiac dysfunction and liver injury in septic mice [40][41][42]. Similarly, Carchman et al. demonstrated that mitophagy was necessary to prevent organ injury in sepsis, but the authors attributed this protective process to the activation of TLR9 signaling [43]. Parkin, an E3 ubiquitin ligase, promotes dysfunctional mitochondrion clearance by autophagy [44]. In our current study, the results indicated that BMSCs increased Parkin-mediated mitophagy and decreased mtROS generation of BMDMs. Furthermore, inhibition of mitophagy with 3-MA agent blocked the inhibitory role of BMSCs in IL-1β activation and mitochondrial ROS production, indicating that the effect of BMSCs on inflammatory regulation and mitochondrial ROS generation was mainly mediated by mitophagy.
Although the data from our study bear some clinical relevance, several limitations remain. First, cardiac function depression in sepsis may be detoxified by various factors, such as inflammatory factors, which may not represent all contributors. We only observed the effects of MSCs, but the underlying mechanisms are still unclear. Second, the data were based on a murine CLP model, and the dose of BMSCs may be different in a clinical setting. Besides, as the in vitro cell study cannot fully simulate the in vivo circumstances, the weak link between in vivo and in vitro studies should be taken into consideration. Although the mechanisms of inflammasome inhibition by MSCs were dissected in vitro, it was less well confirmed in vivo. Therefore, more studies are warranted to confirm if our current findings are true for human MSCs in clinical settings.

Conclusion
BMSCs improved the murine survival rate and alleviated organ injury in experimental sepsis induced by CLP. BMSCs' beneficial effects were mediated by inhibiting NLRP3 inflammasome activation in macrophages, which was primarily through increasing the mitophagy of macrophages and decreasing mitochondrial ROS generation (Figure 8). Taken together, MSCs can act as promising cellular therapy, as an immune-modulator, for fighting against sepsis.