Introduction

Sepsis, characterized by a system-wide dysregulation of inflammation responses following primary bacterial infection, is one of the leading causes of death in intensive care unit worldwide1. Extravagant inflammation reactions are autodestructive and result in distant organ damage or failure, whereas an insufficient response due to immune anergy or tolerance can propagate further infection2. Some septic patients die at the early stage of sepsis characterized by an uncontrolled activation of the immune responses3, but most patients die at the late stage with significant immunosuppression, marked by an impaired activation of the immune responses and hypoinflammation4,5,6. Therefore, understanding the immune pathophysiology of the late stage will be mandatory for discovering novel therapeutic options.

Reactive oxygen species (ROS) play essential roles in regulating immune responses against pathogens, and are critical to macrophages' bactericidal activity7,8. In addition, ROS could function as signaling molecules to activate multiple signal transduction cascades and in turn modulate inflammatory responses9,10,11,12. Thus, mediators regulating oxidative stress may effectively modulate inflammation responses and improve survival in sepsis.

Parkinson disease 7 (Park7), also known as DJ-1, is highly conserved in all organisms and has diverse biochemical and cellular activities. Loss-of-function mutations of PARK7 are considered as causal factors for the early onset of Parkinson's disease (PD). In addition, decreased Park7 levels were detected in cerebrospinal fluid of late-onset PD patients (age > 50) compared with health controls13. Oxidative stress has been implicated in the pathogenesis of PD, because neuronal cells with either Park7 deletion or Park7 downregulation are sensitive to oxidative stress14,15. Given the antioxidant feature of Park7 and that oxidative stress is tightly related to inflammatory responses and survival in experimental sepsis12,16, we hypothesized that Park7 protects against sepsis by modulating the cellular redox state.

In this study, we demonstrate that Park7 depletion leads to higher susceptibility to sepsis. Park7−/− mice present increased bacterial burdens, reduced local and systemic inflammation, macrophage paralysis and impaired induction of proinflammatory cytokines under the condition of sepsis. These immunosuppression phenomena are largely attributable to the inactivation of NADPH oxidase caused by Park7 deficiency. Through its C-terminus, Park7 interacts with p47phox to facilitate NADPH oxidase-dependent ROS production. More importantly, we demonstrate that in vivo administration of Park7-restored macrophages rescues animals from septic death induced by LPS. Our findings for the first time demonstrate that Park7 could direct NADPH oxidase activation and indicate the putative therapeutic potential of Park7 in sepsis.

Results

Disruption of Park7 significantly increases lethality but drastically decreases local inflammation during CLP- and LPS-induced sepsis

The absence of Park7 was confirmed in Park7−/− mice (Supplementary information, Figure S1A). We first examined the role of Park7 in survival in a septic model induced by cecal ligation and puncture (CLP). 60 h after CLP, 85% of Park7−/− mice died while only 40% of WT mice died. After 5 days, 25% of WT mice still survived, whereas 15% of Park7−/− mice were alive (Figure 1A). Meanwhile, the higher susceptibility of Park7−/− mice was also verified in another clinically relevant septic model induced by Pseudomonas aeruginosa infection (Supplementary information, Figure S1B). We then investigated the role of Park7 in survival in LPS-induced sepsis. WT and Park7−/− mice were treated i.p. with a lethal dose of LPS (15 mg/kg BW) and survival was monitored for 5 days. 100% of Park7−/− mice died within 84 h whereas 30% of WT mice survived by day 5 (Figure 1B). As Park7 is critical for septic survival, the role of Park7 in regulating inflammation was investigated in nonlethal septic models. Lung injury was assessed by histopathology analysis and lung wet/dry ratio 24 h after CLP or LPS treatment (i.p. 5 mg/kg BW). Lung injuries were greatly decreased as indicated by less infiltration of inflammatory cells and lower wet/dry ratio in Park7−/− mice compared with WT ones (Figure 1C and 1D). Consistent with these results, much fewer inflammatory cells were observed in bronchoalveolar lavage (BAL) fluid of Park7−/− mice compared with WT mice (Figure 1E and 1F). Meanwhile, upon CLP and LPS treatment, significantly reduced recruitment of macrophages to peritoneal cavity was observed in Park7−/− mice compared with WT mice (Figure 1G and 1H). In response to impaired recruitment of inflammatory cells, significantly decreased levels of CXCL1, CXCL2 and CCL2 were detected in both peritoneal fluid and BAL fluid in Park7−/− mice compared with WT mice after CLP or LPS (Figure 1I-1K and Supplementary information, Figure S1C-S1E). Taken together, these results show that Park7−/− mice have greater susceptibility but reduced inflammatory responses to sepsis induced either by endotoxin or by polymicrobial infections.

Figure 1
figure 1

Park7−/− mice were more sensitive to CLP- and LPS-induced septic shock. (A, B) Survival of WT and Park7−/− mice subjected to CLP (n = 20/group) (A) and LPS (n = 10/group) (B). Data were analyzed using log-rank test. *P < 0.05. (C) Representative sections of HE staining for lung infiltration of inflammatory cells 24 h after CLP and LPS (n = 5/group). Magnification, 40×. (D) Lung wet/dry ratio 24 h after CLP and LPS. (E, F) Quantification of inflammatory cells in collected BAL fluid 24 h after CLP (E) and LPS treatment (F). (G) Quantification of inflammatory cells in peritoneal lavage fluid 24 h after CLP (n = 5/group). (H) Quantification of inflammatory cells in peritoneal lavage fluid after LPS treatment at the indicated time points. (I-K) CXCL1 (I), CXCL2 (J) and CCL2 (K) concentrations in peritoneal lavage fluid of WT and Park7−/− mice 24 h after CLP or LPS treatment (n = 5/group). Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

Park7−/− mice had greatly reduced secretion of IL-6 and TNF-α after CLP and LPS treatment

As IL-6 and TNF-α are proinflammatory cytokines elevated during the early septic hyperinflammation and are negatively correlated with the septic survival17,18, their concentrations in serum and peritoneal fluid were measured by ELISA. 24 h after CLP or LPS, levels of IL-6 and TNF-α in serum or peritoneal fluid were significantly lower in Park7−/− mice compared with WT mice (Figure 2A-2D). Furthermore, IL-6 concentrations in BAL fluid were lower in Park7−/− mice compared with WT mice (Figure 2E and 2F). Given that macrophages are the prime source of inflammatory cytokines19, levels of IL-6 and TNF-α in peritoneal macrophages were examined by qPCR and ELISA. LPS induced fewer transcripts and less secretion of IL-6 and TNF-α in Park7−/− macrophages compared with WT ones (Figure 2G and 2H). Because the secretion of TNF-α usually declines in 24 h, we did not observe significant difference in TNF-α levels 24 h after LPS. Given that IL-10 is a negative regulator of IL-6 and TNF-α upon inflammatory stimulation20, we then examined whether Park7−/− mice has increased IL-10. Interestingly, IL-10 levels in serum and peritoneal fluid were also significantly decreased in Park7−/− mice compared with WT mice after CLP or LPS treatment (Figure 2I and 2J). Similarly, less IL-10 induction was detected in Park7−/− macrophages after LPS (Figure 2K). We further examined whether Park7 deficiency could affect the IL-6 and TNF-α levels in RAW264.7, a macrophage cell line. Stable transfectants with expression of a Park7-shRNA or a shRNA not targeting any genes from mouse were established. Park7 knockdown was confirmed by qPCR and immunoblotting (Supplementary information, Figure S2A). We refer to NT and KD hereafter as non-target and knockdown groups, respectively. LPS induced significantly fewer IL-6 and TNF-α in KD compared with NT (Supplementary information, Figure S2B-S2E). These results show that, in sepsis models, the significantly attenuated production of IL-6 and TNF-α in Park7−/− mice is IL-10-independent and is at least partially because of the impaired cytokine production in Park7−/− macrophages after endotoxin treatment.

Figure 2
figure 2

Disruption of Park7 reduces proinflammatory cytokine induction during CLP- and LPS-induced septic shock. (A, B) IL-6 and TNF-α serum concentrations of WT and Park7−/− mice 24 h after CLP (A) or LPS treatment (B) (n = 5/group). (C, D) IL-6 and TNF-α levels in peritoneal lavage fluid of WT and Park7−/− mice 24 h after CLP (C) or LPS treatment (D) (n = 5/group). (E, F) IL-6 levels in BAL fluid of WT and Park7−/− mice 24 h after CLP (E) or LPS treatment (F) (n = 5/group). (G, H) Quantification of mRNA (G) and protein levels (H) of IL-6 and TNF-α, at the indicated time points, in peritoneal macrophages after LPS treatment. (I, J) IL-10 concentrations in serum or peritoneal lavage fluid 24 h after CLP (I) or LPS treatment (J). (K) Quantification of IL-10 protein levels, at the indicated time points, in peritoneal macrophages after LPS treatment. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001. RFC, relative fold change.

Toll-like receptor signaling was blunted in Park7-deficient macrophages

Toll-like receptors (TLRs) recognize pathogen-derived macromolecules, and play an important role in macrophage activation21,22. As Park7 deficiency impaired macrophage cytokine production, we then examined the role of Park7 in TLR signaling. Compared with WT, Park7−/− peritoneal macrophages produced less IL-6 and TNF-α after LPS, PolyI:C, LTA or CpG ODN treatment (Figure 3A). Next, we assessed the activation of MAPKs, NF-κB and TBK-IRF3 signaling pathways downstream of TLRs in primary macrophages and RAW264.723. Thirty min after LPS stimulation, activation of MAPKs was significantly lower in KO and KD macrophages than that in WT and NT, respectively (Figure 3B). In addition, phosphorylation of ΙκB-α and IRF3 was remarkably decreased in KD compared with NT (Supplementary information, Figure S3). Because AP-1, NF-AT and NF-κB are known transcription factors downstream of MAPKs and NF-κB pathways24,25,26, we examined their transactivation by luciferase reporter assays. Compared with NT, KD had markedly decreased luciferase activity after LPS treatment (Figure 3C-3E). Given that TLR signaling augments macrophage bactericidal activity8, we assessed the bactericidal activity in peritoneal macrophages of WT and Park7−/− 1 h after incubation with E. coli. Impaired phagocytosis and bacterial killing abilities were observed in Park7−/− macrophages compared with WT ones (Figure 3F). Consequently, the bacterial burdens in blood or peritoneal lavage fluid were greater in Park7−/− mice compared with WT mice 24 h after CLP (Figure 3G and 3H).

Figure 3
figure 3

Park7 depletion affects macrophage activation after LPS treatment. (A) Quantification of mRNA expression of IL-6 and TNF-α in macrophages upon treatment with various pathogen-derived macromolecules, including LPS (100 ng/ml), PolyI:C (20 μg/ml), LTA (10 μg/ml) and CpG.(1 μM; n = 4/group). (B) Phosphorylated p38, JNK and ERK shown by immunoblotting in WT and Park7−/− primary macrophages or NT and KD, at the indicated time points, upon LPS. (C-E) Transcriptional activities of NF-AT (C), NF-κB (D) and AP-1 (E) by luciferase reporter assay in NT and KD, at the indicated time points, after LPS treatment. (F) Phagocytosis and killing of bacteria by peritoneal macrophages isolated from WT and Park7−/− mice (n = 3/group). (G, H) Quantification of bacterial burden in blood (G) and peritoneal lavage fluid (H) 24 h after CLP (n = 3/group). (I) Quantification of bacterial burden in blood of mice facing imminent death after a lethal dose of LPS challenge. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

We then determined whether the increased bacterial burden also contributes to the higher mortality of Park7−/− mice after LPS treatment. We employed a novel strategy to detect bacterial burdens in mice facing imminent death after a lethal LPS challenge. Body temperature of mice was monitored every 6 h in the first 24 h and hereafter every 12 h after LPS. We artificially defined the mice with body temperature below 32 degrees as dying ones27, and sacrificed them for bacterial burden measurement. At the end of this experiment (5 days), 12 out of 20 WT mice and 16 out of 20 KO mice were sacrificed. No bacteria were detected in peritoneal fluid (data not shown). However, significantly higher bacterial burden was observed in blood of Park7−/− mice compared with WT mice (Figure 3I). These results show that Park7 deficiency severely blunts the TLR signaling and consequently impairs the bactericidal ability of macrophages, and strongly suggest that the higher mortality of Park7−/− mice from sepsis is caused by increased bacteria burden.

Park7 deficiency impairs NADPH oxidase-dependent ROS production in macrophages

As ROS produced by NADPH oxidase have an established role in regulating TLR signaling12,28, we assessed whether Park7 could modulate the accumulation of NADPH oxidase-derived ROS. Intracellular ROS were quantified through dichlorofluorescin diacetate staining followed by flow cytometry or lucigenin-derived chemilumiscence in RAW264.7 and peritoneal macrophages. After 4 h LPS treatment, greatly reduced induction of intracellular ROS was observed in Park7−/− and KD macrophages compared with WT and NT ones, respectively (Figure 4A and 4B). The decline of ROS in Park7-deficient macrophages apparently did not result from increase of ROS scavengers, because the expression of Nfe2l2 and its downstream targets, Gclm and Trx129,30,31, was either comparable between Park7−/− and WT macrophages or decreased in Park7−/− macrophages compared with WT ones (Supplementary information, Figure S4). To specifically target NADPH oxidase, we adopted phorbol-12-myristate-13-acetate (PMA) to exclusively stimulate for NOX-dependent respiratory burst. PMA caused dramatic ROS induction in WT and NT macrophages but not in Park7-deficient ones (Figure 4C-4F). Given that the main ROS produced by NADPH oxidase is superoxide anion, we adopted another assessment to measure the superoxide by combining a widely used sensitive superoxide (O2.−) probe, Dihydroethidium (DHE), with HPLC method. Because DHE oxidation yields two fluorescent products, 2-hydroxyethidium (EOH) and ethidium, and EOH is known to be more specific for O2.− than ethidium, we used HPLC analysis to measure EOH levels to assess O2.− production or NADPH oxidase activity32. As shown in Figure 4G, under the PMA treatment, WT macrophages generated more superoxide anion than Park7−/− macrophages.

Figure 4
figure 4

Depletion of Park7 results in impaired LPS- and PMA-induced ROS production in macrophages. (A, B) Quantification of cellular ROS levels by oxidized DCFDA and flow cytometry in peritoneal macrophages isolated from WT and Park7−/− mice (A) or in NT and KD RAW264.7 cells (B) 4 h after LPS. (C, D) Quantification of ROS, at the indicated time points, by DCFDA staining in WT and Park7−/− macrophages (C) or in NT and KD RAW264.7 cells (D) after PMA stimulation. (E, F) Quantification of superoxide, at the indicated time points, by lucigenin-derived chemiluminescence in WT and Park7−/− macrophages (E) or in NT and KD RAW264.7 cells (F). (G) HPLC analysis of DHE-derived products in WT and Park7−/− peritoneal macrophages after PMA (100 ng/ml) stimulation for 4 h. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

Restoration of Park7 expression rescues ROS production and inflammatory responses in Park7−/− macrophages

To test the specificity of Park7 on NADPH oxidase-dependent ROS production, we assessed whether re-expression of Park7 can rescue the ROS production by NADPH oxidase. Stable transfectants with doxycycline-induced Park7 expression were established in NT and KD and named NT-Park7 and KD-Park7, respectively. Doxycycline-induced Park7 expression was detected by qPCR and immunoblotting (Figure 5A and 5B). ROS production was restored in doxycycline-treated KD-Park7 after PMA stimulation (Figure 5C). In response to ROS restoration, IL-6 and TNF-α production was also reinstated (Figure 5D and 5E). Similar restoration of IL-6 and TNF-α production was detected in KD-Park7 cells treated with LPS (Figure 5F and 5G). These results indicate that restoration of Park7 expression rescues NADPH oxidase-dependent ROS production and in turn reinstates the production of proinflammatory cytokines.

Figure 5
figure 5

Restored expression of Park7 rescues the ROS and proinflammatory cytokine production in Park7-deficient macrophages. (A, B) Quantification of Park7 mRNA and protein levels by qPCR (A) and western blotting (B), respectively, in NT-Park7 and KD-Park7 RAW264.7 cells after doxycycline (1 μg/ml) treatment. (C) ROS levels, at the indicated time points, by DCFDA staining after PMA stimulation. (D, E) IL-6 (D) and TNF-α (E), at the indicated time points, in culture media by ELISA after PMA treatment. (F, G) IL-6 (F) and TNF-α (G), at the indicated time points, in culture media by ELISA after LPS treatment. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

ROS modulate LPS-induced proinflammatory cytokine production in macrophages

As, upon LPS stimulation, the induction of ROS and proinflammatory cytokines was decreased in Park7−/− macrophages, we investigated whether administration of an exogenous oxidant could augment proinflammatory cascade. IL-6 and TNF-α levels were measured in Park7−/− macrophages 6 h after treatment with LPS (1 ng) or LPS along with H2O2 (10 μM). Exposure of H2O2 significantly increased the levels of LPS-induced IL-6 and TNF-α expression in Park7−/− macrophages (Figure 6A and 6B). To answer whether ROS elimination could inhibit proinflammatory cascade, we treated WT macrophages with LPS (10 ng) or LPS along with antioxidant NAC (0.5 mM). After 6-h treatment, NAC significantly reduced IL-6 and TNF-α levels (Figure 6C and 6D). These results indicate that ROS play a critical role in modulating proinflammatory cytokine production in macrophages.

Figure 6
figure 6

ROS enhance LPS-induced proinflammatory cytokine production in macrophages. (A, B) IL-6 (A) and TNF-α (B) levels by ELISA in Pakr7−/− macrophages 6 h after H2O2 (10 μM) and LPS (1 ng/ml) treatment. (C, D) IL-6 (C) and TNF-α (D) levels by ELISA in WT macrophages 6 h after NAC (0.5 mM) and LPS (10 ng/ml) treatment. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

Park7 interacts with p47phox to activate NADPH oxidase

NADPH oxidase is composed of six subunits including gp91phox, p22phox, p40phox, p67phox, p47phox and rac2. Given that the function of Park7 is partially mediated by physical interactions with other proteins like TP5333 and PTEN34, we then examined putative interactions between Park7 and the NADPH oxidase subunits. We employed DOMINE to search for putative domains binding to Park735,36. This search identified several putative interacting domains including catalase, DJ-1/PfpI family, PX domain and so on. Interestingly, both p47phox and p40phox contain PX domains, suggesting their putative interactions with Park7. Immunoprecipitation indicated no interaction between Park7 and p40phox (data not shown). Interaction of endogenous Park7 and p47phox was observed in NT as well as in KD but with a much decreased level (Figure 7A). Remarkably, this interaction was enhanced in NT after LPS challenge in a time-dependent manner (Figure 7A). As Cys106 oxidation is important for Park7 interaction with other proteins37, we then examined whether Park7 was oxidized in our experimental system. NT cells showed significantly increased Park7 oxidation after PMA treatment (Supplementary information, Figure S8), suggesting that the enhanced Park7-p47phox interaction may be attributed to the increased Park7 oxidation level. Given that phosphorylation and plasma membrane translocation of p47phox are key events for NADPH oxidase assembly38, we investigated whether Park7 is required for activation of NADPH oxidase via modulating p47phox phosphorylation and membrane translocation. LPS and PMA treatment induced significant p47phox phosphorylation and membrane accumulation in NT but not in KD, respectively (Figure 7A and 7B). In agreement, fluorescent staining demonstrated membrane translocation of p47phox in NT but not in KD upon PMA stimulation (Figure 7C). Given that Rac2 plays a critical role in the assembly and activation of NADPH oxidase39, we then investigated whether Rac2 could affect the Park7-p47phox interaction. Immunoprecipitation indicated that Rac2 knockdown had no effect on this interaction (Supplementary information, Figure S5). Taken together, these results suggest that, by interacting with p47phox and modulating phosphorylation and membrane translocation of p47phox, Park7 affects NADPH oxidase activation.

Figure 7
figure 7

Park7 interacts with p47phox to activate NADPH oxidase. (A) Interaction between endogenous p47phox and Park7, at the indicated time points, in both NT and KD RAW264.7 cells by immunoprecipitation and western blotting after LPS. (B) p47phox levels in both membrane and cytoplasm fractions, at the indicated time points, in both NT and KD RAW264.7 cells by western blotting after LPS. (C) Fluorescent staining of p47phox and membrane of NT and KD RAW264.7 cells 16 h after PMA stimulation. CTB-FITC, Cholera Toxin B Subunit-FITC. Magnification, 40×.

The C-terminus of Park7 is required for the physical association with p47phox

To characterize the functional region(s) of Park7 contributing to p47phox interaction, we constructed a series of doxycycline-inducible Myc-tagged expression vectors encoding truncated Park7 (Figure 8A). ND, CD and MD stand for N-terminal, C-terminal and middle-domain deletion, respectively. Inducible expression of truncated Park7 in stable transfectants was detected by immunoblotting (Supplementary information, Figure S6). CD80 was undetectable. Immunoprecipitation indicated that Park7-CD40 was not able to interact with p47phox (Figure 8B), suggesting that the C-terminus of Park7 is indispensible for this interaction. Like full-length Park7, enhanced interaction between truncated Park7 and p47phox was also observed upon LPS treatment (Figure 8B). To answer whether this interaction is important for NADPH oxidase activation, we overexpressed ND40 and CD40 in Park7-deficient macrophages. ND40, but not CD40, rescued PMA-induced ROS production by NADPH oxidase (Figure 8C). In response to this ROS change, LPS-induced IL-6 and TNF-α production was also rescued by ND40 but not CD40 (Figure 8D and 8E). These data suggest that the C-terminus of Park7 is required for Park7-p47phox interaction which is critical for NADPH oxidase activation and proinflammatory cytokine production in macrophages.

Figure 8
figure 8

Characterization of the interaction between p47phox and Park7. (A) Diagram of five truncated Park7 fragments cloned into a doxycycline-inducible expression vector followed by a Myc-tag. (B) Interactions between p47phox and truncated Park7, at the indicated time points, by immunoprecipitation and western blotting after LPS. (C) NADPH oxidase-dependent ROS production, at the indicated time points, in RAW264.7 cells with ND40 and CD40 expression after PMA stimulation. (D, E) IL-6 (D) and TNF-α (E) levels by ELISA, at the indicated time points, in RAW264.7 cells with ND40 and CD40 expression after LPS treatment. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

Macrophages with restored Park7 expression confer strong protection against severe sepsis

To assess the therapeutic potential of Park7 on sepsis, we adopted a RAW264.7-based macrophage reconstituted mouse model, which was successfully used to reconstitute macrophage-depleted allogeneic BALB/c mice40. Given BALB/c origin of RAW264.7 cells, we first examined whether RAW264.7 cells will be eliminated by immune repelling in B6 host mice. We used a fluorescent vital dye SP-DiI (D-7777, Molecular Probes) to stain the RAW264.7 cells, and then transferred these labeled RAW264.7 cells into C57BL/6 or BALB/C mice with or without GdCl3 pretreatment. 24 h or 3 days after transferring, the mice were sacrificed and the lungs, livers and spleens were collected. The fluorescent signals were examined under fluorescent microscope. In the group without GdCl3 pretreatment, compared with BALB/c mice, great immune repelling was observed in B6 mice with RAW264.7 transferring (Supplementary information, Figure S7A and S7B). However, in the GdCl3-treated group, the intensity and pattern of fluorescent signals on RAW264.7 cells were comparable in both allogeneic and syngeneic transplantation. These results clearly demonstrated that RAW264.7 cells were not eliminated by either immune repelling or GdCl3-mediated killing in macrophage-depleted C57BL/6 mice in at least 3 days (Supplementary information, Figure S7A and S7B). Then, KD-Park7 RAW264.7 cells were treated with or without doxycycline and injected i.v. into mice depleted of macrophages (Figure 9A). Lethal dose of LPS treatment was carried out 24 h after transferring. Survival was monitored every 12 h. 90% of mice with KD-Park7 (−Dox) transferring died while only 50% of mice transferred with KD-Park7 (+Dox) died by the end of the experiment, whereas mice transferred with NT pretreated with or without doxycycline showed no survival difference (Figure 9B). In response to the improved survival rate, significantly decreased bacterial burden was detected in the blood of mice transferred with KD-Park7 (+Dox) compared with the ones with KD-Park7 (−Dox) transferring (Figure 9C). In addition, elevated IL-6 and TNF-α levels in both serum and peritoneal lavage and increased lung infiltration of inflammatory cells were also observed in mice transferred with KD-Park7 (+Dox) 24 h after a nonlethal dose of LPS challenge (Figure 9D-9F). These results strongly suggest that forced Park7 expression acts as a putative therapeutic option against septic shock.

Figure 9
figure 9

Park7 restoration strongly protects against LPS-induced sepsis. (A) Doxycycline-induced Park7 expression detected by western blotting. (B) Survival of macrophage-depleted mice after transferring with NT (Dox−), NT (Dox+), KD-Park7 (Dox−) or KD-Park7 (Dox+) macrophages followed by LPS challenge (n = 10/group). Data were analyzed using log-rank test. *P < 0.05. (C) Quantification of bacterial burden in blood of dying mice subjected to macrophage transferring followed by a lethal dose of LPS challenge. (D, E) IL-6 and TNF-α levels in serum (D) and peritoneal lavage fluid (E) of mice subjected to macrophage transferring 24 h after LPS (n = 5/group). (F) Representative sections of HE staining for lung infiltration of inflammatory cells of mice subjected to macrophage transferring 24 h after LPS (n = 5/group). Magnification, 40×. Data were analyzed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.

Discussion

We have shown that Park7−/− mice have greater susceptibility to CLP- and LPS-induced sepsis, which is associated with impaired bactericidal ability of macrophages in Park7−/− mice. Furthermore, we demonstrate that forced expression of Park7 restored macrophage activation characterized by recovered production of ROS and proinflammatory cytokines in Park7−/− macrophages. More importantly, mice transferred with Park7-restored macrophages showed significantly improved survival along with decreased bacterial burdens and reinforced proinflammatory responses in LPS-induced sepsis. Through its C-terminus, Park7 could interact with p47phox and facilitate its phosphorylation and membrane translocation, which in turn result in the NADPH oxidase activation.

LPS-induced septic death is largely dependent on excessive inflammatory response. Compared with WT mice, however, we demonstrated that Park7−/− mice are more susceptible to LPS-induced sepsis accompanied by impaired cytokine and chemokine production. Interestingly, similar phenomenon was also observed by Foster's group in Stat2−/− mice41. These findings strongly suggest the existence of an alternative, cytokine storm-independent mechanism of LPS-induced septic death. In fact, we demonstrated that the significant bacteremia may contribute to the higher susceptibility of Park7−/− mice to LPS-induced septic death.

In p53-null but not in p53 WT MEF cells, Vasseur et al.42 have recently demonstrated that Park7 knockdown could lead to reduced ROS accumulation. However, we showed that Park7 is essential for ROS induction upon LPS treatment in p53 WT macrophages. This discrepancy could be explained by Hassan's43 observation that LPS inhibits p53 activation in macrophages. Therefore, these results indicate that Park7 could direct ROS induction in p53-null or -inactive cells. Given 57% of sepsis patients infected with gram-negative bacteria, Park7 could have broad therapeutic potential for sepsis. On the other side, as inactivation or mutation of p53 are frequent in tumors, whether Park7 controls ROS production in tumors will be an intriguing question to answer.

As far as we know, we for the first time demonstrate that Park7 is involved in ROS production in a NADPH oxidase-dependent manner. This function of Park7 relies on its C-terminal interaction with p47phox. Upon LPS challenge, we observed an enhanced interaction between Park7 and p47phox concomitant with elevated phosphorylation and membrane translocation of p47phox. Similar to the Park7−/− mice, leukocytes from p47phox knockout mice also showed impaired capabilities to produce superoxide and kill staphylococci44. The exact mechanism responsible for the enhancement of the Park7-p47phox interaction after LPS remains to be determined. However, recent studies demonstrated that both p47phox and Park7 could be recruited into lipid rafts under the stimulation of LPS45,46, suggesting that the enhanced interaction between Park7 and p47phox may be due to their sub-cellular spatial changes.

ROS are not only cytotoxic molecules, but also signaling molecules regulating a wide variety of physiological processes. For example, by regulating TLRs-initiated signaling pathways, ROS appear to modulate the production of proinflammatory cytokines11,47. In the present study, addition of H2O2 enhanced LPS-induced proinflammatory cytokine production in Park7−/− macrophages, whereas NAC reduced proinflammatory cytokine levels in WT macrophages. These results further confirm the strong link between ROS and proinflammatory cytokines.

An uncontrolled release of proinflammatory cytokines, thought to lead to multiple organ failure and even death, is an important characteristic in early stage of sepsis48,49. Many agents with anti-inflammatory potency were tested for sepsis therapy but with limited or no success50. Most recently, accumulating evidence indicates that sepsis has a late immunosuppression stage characterized by lymphopenia and loss of immune function4, and this stage seems responsible for the majority of death of septic patients5,6. In our sepsis models, Park7−/− mice present immunosuppression phenotypes similar to the late stage of sepsis, suggesting that Park7−/− mice could serve as an ideal animal model for studying the immunopathophysiology of the late stage of sepsis. Actually, we demonstrate that restoration of Park7 expression in macrophages could improve survival in LPS-induced sepsis. This provides a possibility that agents that can induce Park7 expression may have clinical value to treat sepsis in late stage.

Materials and Methods

Mouse strains

Mice were bred and maintained under SPF conditions. Park7−/− mice (stock#: 006577) were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). Littermate controls of matched age and sex were used in all experiments.

Measurement of bacterial burden in mice facing imminent death after LPS

Mice (20/group) were i.p. injected with LPS (15 mg/kg BW) and the rectal temperature was measured every 6 h in the first 24 h and hereafter every 12 h after LPS injection. Mice with temperature below 32 °C were artificially defined as dying ones and sacrificed for collection of blood and peritoneal lavage. 10 μl blood or peritoneal fluid was immediately diluted in 990 μl PBS and 100 μl of the dilution was plated onto Agar-plate for 24 h incubation at 37 °C. Results were presented as CFU/ml.

Cell stimulation and extraction of HPLC

WT and Park7−/− peritoneal macrophages were grown in six-well plates at 3 × 106 cells/well in 1640 medium supplemented with 10% fetal bovine serum. Cells were starved for 1 h before stimulation with PMA (100 ng/ml) for 4 h. After stimulation, cells were washed twice with cold PBS and incubated in PBS/DTPA (500 μl) at final DHE concentration of 50 μM for additional 30 min. After incubation, cells were washed twice with cold PBS, harvested in acetonitrile (500 μl/well), sonicated (10 s, 1 cycle at 8 W), and centrifuged at 12 000× g for 10 min at 4 °C. Supernatants were dried under vacuum and pellets were stored at −20 °C in the dark until analysis. Samples were resuspended in 120 μl PBS/DTPA and 100 μl was injected into HPLC system.

HPLC conditions of analysis

The conditions of HPLC experiment were as described previously32. The optimal emission wavelength for both EOH and ethidium detection is 595 nm. Chromatographic separation was carried out with the use of a NovaPark C18 column (4.6 × 250 mm, 5 μm particle size) in a HPLC system equipped with a rheodyne injector and ultraviolet detector (SPD-M20A) and fluorescence detectors (RF-10A). Solutions A (pure acetonitrile) and B (water/10% acetonitrile/0.1% trifluoroacetic acid) were used as a mobile phase at a flow rate of 0.4 ml/min. Runs were started with 0% solution A, increased linearly to 40% solution A during the initial 10 min, kept at the proportion of 100% solution A for additional 5 min, and to 0% solution A for the final 10 min. DHE was monitored by ultraviolet absorption at 245 nm. EOH and ethidium were monitored by fluorescence detection with excitation at 510 nm and emission at 595 nm. Quantification was performed by comparison of integrated peak areas between the WT and Park7−/− cell supernatants under identical chromatographic conditions.

Standard solution: 100 μm DHE was incubated with xanthine/xanthine oxidase (0.5 mM/0.05 U/ml) in PBS buffer composed of (in mM) 7.78 Na2HPO4, 2.20 KH2PO4, 140 NaCl and 2.73 KCl, pH 7.4, containing 100 μM DTPA at 37 °C for 30 min. EOH was separated by HPLC, collected and lyophilized to dryness. The purple-pink solid was further resuspended in DMSO and used as standard.

Adoptive transfer of macrophages

Macrophage transferring was performed according to a published method with slight modification40. Briefly, C57BL/6 mice were injected i.v. with GdCl3 (Sigma-Aldrich) to eliminate macrophages. 24 h after GdCl3 injection, KD-Park7 macrophages (1 × 106) were injected i.v. into mice. Another 24 h later, mice were injected i.p. with LPS. Survival was monitored for 5 days.

Statistical analysis

Except survival studies, all other results were presented as mean ± SEM from at least 3 independent biologically replicated experiments. Data were analysed using Student's t-test. *P < 0.05; **P < 0.01; ***P < 0.001.