Opposing roles of LTB4 and PGE2 in regulating the inflammasome-dependent scorpion venom-induced mortality

Tityus serrulatus sting causes thousands of deaths annually worldwide. T. serrulatus-envenomed victims exhibit local or systemic reaction that culminates in pulmonary oedema, potentially leading to death. However, the molecular mechanisms underlying T. serrulatus venom (TsV) activity remain unknown. Here we show that TsV triggers NLRP3 inflammasome activation via K+ efflux. Mechanistically, TsV triggers lung-resident cells to release PGE2, which induces IL-1β production via E prostanoid receptor 2/4-cAMP-PKA-NFκB-dependent mechanisms. IL-1β/IL-1R actions account for oedema and neutrophil recruitment to the lungs, leading to TsV-induced mortality. Inflammasome activation triggers LTB4 production and further PGE2 via IL-1β/IL-1R signalling. Activation of LTB4-BLT1/2 pathway decreases cAMP generation, controlling TsV-induced inflammation. Exogenous administration confirms LTB4 anti-inflammatory activity and abrogates TsV-induced mortality. These results suggest that the balance between LTB4 and PGE2 determines the amount of IL-1β inflammasome-dependent release and the outcome of envenomation. We suggest COX1/2 inhibition as an effective therapeutic intervention for scorpion envenomation.

I ncidences of scorpion stings have surpassed 1.2 million a year and are a serious health problem worldwide, resulting in severe wounds and more than 3,000 deaths annually 1 . Clinical manifestations of severe scorpion envenomation are local pain and multi-organ failure, including cardiogenic shock and pulmonary oedema [1][2][3][4][5] . Scorpion antiserum is the only available treatment, but it is not always effective and might induce adverse effects 1,4 . To develop a specific and effective treatment, it is necessary to understand the pathogenesis of the reaction induced by a scorpion sting.
Human envenomation by Tityus serrulatus venom (TsV) induces a massive and fast release of cytokines such as IL-1b, IL-6, IL-8, IL-10 and TNF-a factor 6 . Animal models of envenomation have been instrumental for the characterization of the inflammatory response and cell activation induced by TsV (refs 5,7,8). TsV or venom-associated molecular patterns (VAMPs) act in a way analogous to pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns 9,10 and are also recognized by pattern recognition receptors (PRRs) in leukocytes 11 . The recognition of TsV by PRRS induces NFkB and AP-1 or PPAR-g activation, leading to the production of proinflammatory cytokines and eicosanoids 11,12 .
Inflammasome activation induces the production of lipid mediators 13 and contributes to IL-1b secretion induced by bee venom 14,15 . Lipid mediators such as PGs and LTs are arachidonate (AA)-derived eicosanoids produced by cyclooxygenases 1/2 (COX1/2) and 5-lipoxygenase (5-LO), respectively, and are produced in abundance by resident and inflammatory cells 16 . PGE 2 has important functions in innate and adaptive immune responses through the binding to one of its four different G-protein-coupled E-prostanoid (EP) receptors, known as EP-1-4 (ref. 17). The EP2 and EP4 are Gas-coupled receptors that activate adenylate cyclase and cAMP production, while EP1 receptor increases intracellular calcium, and EP3 is a Gai-coupled receptor that decreases cAMP formation 18,19 . PGE 2 is a potent vasodilator and vascular permeability factor 20,21 and induces gene expression of the inflammatory and edematogenic cytokine IL-1b via cAMP-PKA signalling pathway 22 . On the other hand, LTB 4 does not induce oedema formation 23 ; however, via GPCRs, BLT1 (high-affinity receptor) rather than BLT2 (low-affinity receptor) 24 induces recruitment and activation of leukocytes 25 and increases phagocytosis and killing of microorganisms by macrophages 26 . Interestingly, low concentrations of LTB 4 , but not LTD 4 , reduce intracellular cAMP production in macrophages 26,27 .
We found that intraperitoneal inoculation of mice with TsV induces the activation of the NLRP3 inflammasome, resulting in the production of IL-1b, LTB 4 and PGE 2 , lung oedema and neutrophil recruitment and animal mortality. In summary, we identified that, in scorpion envenomation, a complex and intricate cAMP-mediated mechanism is responsible for lung oedema and mortality and unveiled a heretofore unknown cross-talk between IL-1b and eicosanoids, where the balance between 5-LO and COX1/2 products appears to determine the severity of envenomation.

Results
TsV induces the activation of the inflammasome. Although bee venom activates the inflammasome to produce mature IL-1b (refs 14,15), whether scorpion venom activates inflammasome remains unknown. We initially tested whether TsV induces inflammasome activation and IL-1b release. Bone marrowderived macrophages (BMDMs) from C57BL/6 wild-type (WT) mice were either pre-treated or not with lipopolysaccharide (LPS) and incubated with a TsV concentration capable of activating macrophages in vitro 11,12 . TsV alone induces the significant production of IL-1b as early as 1 h after stimulation, demonstrating that the venom, per se, delivery signals are necessary for the activation of the inflammasome in BMDMs (Fig. 1a). Next, we determined which component of the NLRP3 inflammasome accounts for caspase-1/11 activation in response to TsV. Experiments with BMDMs deficient in various inflammasome constituents demonstrated that Nlrp3 À / À , Asc À / À and Casp1/11 À / À cells did not produce IL-1b (Fig. 1b) and did not activate caspase-1 in response to TsV (Fig. 1c). During the TsV challenge, macrophages from Nlrc4 À / À mice produced higher amounts of IL-1b than did macrophages from Nlrp3 À / À mice and lower amounts than those produced by macrophages from WT mice. These data suggest that NLRP3 and the NLRC4 inflammasome, to some extent, contribute to venom-induced IL-1b production (Fig. 1b). Western blotting analysis of the p20 subunit of caspase-1 (ref. 28) confirmed that TsV activated the NLRP3 inflammasome (Fig. 1c) as did nigericin, the positive control 28 . As a control, we determined the production of IL-6, an inflammasome-unrelated cytokine. As expected, BMDMs produced IL-6 in response to TsV regardless to the inflammasome components NLRP3, ASC and caspase-1/11 (Fig. 1b). To assess the mechanisms by which inflammasome triggers NLRP3 activation, we investigated the induction of pores in the macrophage membranes and the efflux of potassium, an essential process to enable the activation of the NLRP3 inflammasome 29 . We found that elevated concentrations of extracellular K þ impaired the inflammasome activation in response to TsV (Fig. 1d), suggesting that K þ efflux is involved in the activation of the inflammasome mediated by TsV. Notably, a higher concentration of NaCl potentiates TsV-induced IL-1b release (Fig. 1d), suggesting that the venom activates the inflammasome by a mechanism that is distinct from that induced by hyperosmotic stress, as reported previously 30 . To investigate whether TsV directly triggers cell death, we performed a pore-forming assay using WT-BMDMs treated with TsV in presence or absence of LPS. We verified that, even at 17 h after treatment, the cells did not present significant membrane damage, while cells treated with Triton-X, the positive control, did ( Fig. 1e), confirming that TsV-induced cell activation without significant cell death.
TsV induces lung inflammation and mortality via IL-1R/IL-1b. Because TsV activates the inflammasome in vitro and since in humans and animals, scorpion envenomation induces IL-1b (refs 5,6), lung oedema 1-3 and mortality 1-5 , we performed an in vivo dose-response experiment to identify the inflammatory, lethal and excessive doses of TsV for C57BL/6 and 129sv mice, which are the WT control mice for the inflammasome, and for Alox5 À / À mice, respectively (Supplementary Fig. 1). We used a sublethal dose of TsV (120 mg kg À 1 intraperitoneally) to study inflammasome activation, IL-1b release and IL-1R signallinginduced lung inflammation in vivo. We collected bronchoalveolar lavage fluid (BAL) or whole lungs 4 h after envenomation from C57BL/6 (WT), IL-1r À / À and Casp1/11 À / À mice. TsV activated inflammasome in lungs, inducing a significant increase in IL-1b release and protein extravasation (Fig. 2a), PGE 2 and LTB 4 production (Fig. 2b), and neutrophil recruitment to lung tissue and bronchoalveolar space in C57BL/6 mice (Fig. 2c). To evaluate if TsV induces the expression of enzymes involved in eicosanoid metabolism, peritoneal macrophages from C57BL/6 mice were incubated in vitro with TsV for 2 and 6 h and Ptgs2, Ptges2, Alox5 and Alox5ap mRNA expression was determined. Interestingly, we observed that the mRNA expression of enzymes involved in the PGE 2 biosynthetic pathway were increased earlier than mRNA for enzymes involved in LTs production ( Supplementary Fig. 2). Activation of the IL-1b-IL-1R axis during the TsV-induced mediator release and lung inflammation was confirmed in IL-1r À / À mice. After TsV inoculation, we detected lower levels of IL-1b and protein extravasation in IL-1r À / À mice (Fig. 2a). Moreover, PGE 2 and LTB 4 production and neutrophil accumulation in lungs were lower in the absence of IL-1R than in WT mice (Fig. 2b,c). These results indicated that IL-1b/IL-1R signalling is essential for TsV-induced lung inflammation. Next, we determined whether the absence of IL-1b/IL-1R signalling protects mice from a lethal BMDMs from C57Bl/6 mice were pre-treated or not (a) with LPS (1,000 ng ml À 1 ) for 4 h and then stimulated with TsV (50 mg ml À 1 ) for 0.5, 1, 3, 6, 12 and 24 h to induce IL-1b release, which was measured by ELISA. (n ¼ 4) (b) IL-1b and IL-6 production measured by ELISA in the culture supernatant of BMDMs from C57Bl/6 WT, Nlrc4 À / À , Nlrp3 À / À , Asc À / À and Casp1/11 À / À mice; the cells were incubated only with TsV (50 mg ml À 1 ) for 24 h. (n ¼ 4) (c) C57Bl/6 WT, Nlrp3 À / À , Asc À / À and Casp1/11 À / À BMDMs were pre-treated for 4 h with LPS (1,000 ng ml À 1 ) and later stimulated with TsV (50 mg ml À 1 ) for 24 h or nigericin as a positive control (20 mM) for 1 h. Immunoblotting for caspase 1 (p20) in the culture supernatant was performed and the percentage of p20 abundance was calculated by densitometric analysis and is shown next to the panels. The data are representative of two independent experiments. (d) BMDMs from C57BL/6 mice were stimulated with TsV (50 mg ml À 1 ) plus NaCl (50 mM) or KCl (50 mM) for 24 h and IL-1b was detected in the cell supernatant by ELISA. All data are representative of three independent experiments (n ¼ 3). (e) BMDMs from C57Bl/6 mice were pre-treated or not with LPS (1 mg ml À 1 ) for 4 h and then stimulated with TsV (50 mg ml À 1 ) or Triton-X (9%), which was used as the positive control for pore formation. Fluorometric plots show propidium iodide uptake (RFUs) over time to demonstrate the kinetics of pore formation in cell membrane. The data are representative of two independent experiments, performed in triplicate. Error bars (a,b,d), s.d. *Po0.05 (one-way ANOVA with Bonferroni's post-test compared with non-TsV-stimulated), # Po0.05 (one-way ANOVA with Bonferroni's post-test compared with TsV-stimulated). Images have been cropped for presentation. Full size images are presented in Supplementary Fig. 6. RFU, relative fluorescence units.
dose of TsV (180 mg kg À 1 ). We found that 100% of IL-1r À / À mice survived the lethal dose, while only 35% of the C57BL/6 mice survived (Fig. 2d). Oedema and cell accumulation induced by the lethal dose of TsV were also assessed macroscopically using Evans blue leakage assay and microscopically by haematoxylin and eosin (H&E) staining of lung tissue. Whole-lung staining in C57BL/6 mice showed intense Evans blue colouring and cell infiltration ( Supplementary Fig. 3a,c), whereas the lungs from IL-1r À / À mice did not stain positive for Evans blue and showed milder oedema and weaker infiltration by leukocytes ( Supplementary Fig. 3d,f). As expected, TsV intraperitoneal inoculation induced neutrophil recruitment to the peritoneal cavity and increased the number of circulating neutrophils. However, these were lower in IL-1r À / À mice ( Supplementary Fig. 3g,h). To confirm the role of caspase-1/11 in the TsV-induced IL-1b secretion in vivo, we performed experiments using the Casp1/11 À / À mice. We found that caspase-1/11-deficient animals showed a reduction in protein extravasation (Fig. 2a), in PGE 2 and LTB 4 release (Fig. 2b), and in neutrophil recruitment to the lung and BAL (Fig. 2c) 4 h after administration of 120 mg kg À 1 TsV. The impact of caspase-1/11 deficiency on mortality among the mice after lethal dose of TsV (180 mg kg À 1 ) was examined and, up to 24 h, 100% of Casp1/11 À / À survived, while only 25% of WT mice survived (Fig. 2d). These data confirmed that caspase-1/11 and IL-1b are essential participants in lung inflammation and mediated animal death. Because we observed that IL-1r À / À and Casp1/11 À / À mice still produced significant amounts of PGE 2 after TsV administration, we speculated that PGE 2 might be necessary for the release and the amplification loop for IL-1b production.

5-LO metabolites control lung inflammation and mortality.
Our data show that the earlier release of LTB 4 was dependent on inflammasome, while PGE 2 production was only partially dependent on the inflammasome (Fig. 2b). Therefore, we hypothesized that Alox5 À / À mice would develop a weaker lung inflammation with reduction in mortality after envenomation. Unexpectedly, when we performed a TsV dose-response curve (60, 120, 180 or 360 mg kg À 1 ) and evaluated the survival up to 24 h, we observed that Alox5 À / À mice were more susceptible than 129sv and C57BL/6 mice ( Supplementary Figs 1,4). We next investigated lung inflammation in Alox5 À / À and in WT 129sv mice, 4 h after injection of 120 mg kg À 1 TsV. According to the increased susceptibility of these mice to TsV challenge, inflammation was stronger in the lung parenchyma of Alox5 À / À mice than that in the lung parenchyma of 129sv mice, as evidenced by an increase in IL-1b and protein concentration (Fig. 3a), PGE 2 release (Fig. 3b) and leukocyte accumulation in the lung tissues (Fig. 3c). However, the number of neutrophils in BAL of Alox5 À / À mice was lower than that of BAL from 129sv WT mice (Fig. 3c). As expected, LTB 4 amount in the lungs of (a-c) C57BL/6 WT, IL-1r À / À or Casp1/11 À / À mice were inoculated intraperitoneally with a sublethal dose of TsV (120 mg kg À 1 ) or PBS and the lungs were excised 4 h later for analysis of (a) IL-1b production and total protein, (b) PGE 2 and LTB 4 concentrations, (c) MPO activity and in a separate group of mice, BAL was performed to count neutrophils within the same period. (d) C57BL/6 WT, IL-1r À / À or Casp1/11 À / À mice were inoculated intraperitoneally with a lethal dose of TsV (180 mg kg À 1 ) for assessment of survival; mice were monitored for 24 h. (a-c) Experiments with a sublethal dose of TsV were performed once with six mice and error bars denote s.d. (d) Data on the lethal dose of TsV are shown as a percentage of surviving animals (n ¼ 6). *PBS versus TsV; # TsV in WT mice versus TsV in IL-1r À / À mice or in Casp1/11 À / À mice. These differences were considered significant with Po0.05 according to one-way ANOVA with Bonferroni's post-test (a-c) or the log-rank test (d).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10760 129sv mice increased after TsV injection (Fig. 3b). Because we observed a higher production of IL-1b in the lungs of Alox5 À / À mice (Fig. 3a), we next assessed the effects of exogenous IL-1 receptor antagonist (IL-1Ra) on TsV-induced lung inflammation and mortality. We found that, regardless of the mouse strain, IL-1Ra treatment decreased TsV-induced lung inflammation ( Fig. 3a,b) and BAL neutrophil infiltration (Fig. 3c), confirming the crucial role of IL-1b for oedema and cell recruitment. Furthermore, treatment with exogenous IL-1Ra increased the rate of TsV-induced mortality among Alox5 À / À mice closer to that of 129sv IL-1Ra-treated mice. About 60% of the mice in both strains survived up to 24 h (Fig. 3d). To confirm that the high mortality of Alox5 À / À mice was not related to a nonspecific compensatory mechanism, we inoculated C57BL/6 mice with a lethal dose of TsV and either treated them or not with MK886, a leukotriene synthesis inhibitor. As expected, MK886 increased mortality, decreased LTB 4 and IL-1b and increases PGE 2 (Supplementary Fig. 5a-d). Taken together, these results indicate that mediators downstream of 5-LO activation protected the host from excessive inflammasome activation and consequently mortality after scorpion envenomation.
As described above, 100% of Alox5 À / mice died after a lethal dose of TsV and produced 100% more PGE 2 than 129sv mice did. Given that PGE 2 induces IL-1b gene expression 22,31 and oedema 20,21 , we next tested whether treatment of 129sv and 129sv WT or Alox5 À / À mice were inoculated intraperitoneally with PBS or sublethal (120 mg kg À 1 ) dose of TsV and lung tissues were recovered 4 h later for analysis of (a) IL-1b production and protein content; (b) PGE 2 and LTB 4 release; and (c) histological features of infiltration by inflammatory cells, as determined by staining with H&E, and the number of neutrophils was determined in BAL within the same period in another group of mice. (d) For survival analysis, 129sv WT or Alox5 À / À mice that received PBS or a lethal (180 mg kg À 1 ) dose of TsV were monitored for 24 h. (a-c) A group (n ¼ 6) of 129sv WT or Alox5 À / À mice was treated with vehicle or IL-1Ra (10 mg kg À 1 intraperitoneally) 1 h before and again 1 h after TsV injection and was subjected to the above analysis. (a-c) The experiment with a sublethal dose of TsV was performed once on six mice and the error bars denote s.d. (d) Data on the lethal dose of TsV are shown as a percentage of surviving animals among six mice. *PBS versus vehicle þ TsV, & vehicle þ TsV in 129sv versus vehicle þ TsV in Alox5 À / À , # 129sv or Alox5 À / À inoculated with vehicle þ TsV versus 129sv WT or Alox5 À / À inoculated with IL-1Ra þ TsV; $ PBS in 129sv WT versus PBS in Alox5 À / À . These differences were considered significant with Po0.05 according to one-way ANOVA with Bonferroni's post-test (a-c) or the log-rank test (d). The histological analyses were single-blinded. Scale bars, 100 mm. Alox5 À / À mice with inhibitors that block both COX1 and COX2 (indomethacin) and COX2 alone (celecoxib) abrogates mortality induced by the lethal dose of TsV. Indomethacin abrogated TsVinduced mortality in both 129sv and Alox5 À / À mice and significantly inhibited protein extravasation, PGE 2 and IL-1b production, and myeloperoxidase (MPO) expression in the lungs (Fig. 4a). Similar results were observed in C57BL/6 mice injected with a lethal dose of TsV and treated with indomethacin, celecoxib or an EP2 antagonist, AH6809. Although B50-70% of C57BL/6 mice died after administration of 180 mg kg À 1 TsV, indomethacin, celecoxib and EP2 antagonist treatment decreased mortality, PGE 2 and IL-1b production and neutrophil infiltration in the lungs (Fig. 4b-d). As expected, COX1/2 inhibition in C57BL/6 mice also resulted in increased LTB 4 release (Fig. 4b,c). Inhibition of COX1 by SC-560 partially prevented TsV-induced mortality ( Supplementary Fig. 5e). These results indicate that TsV-COX1/2-PGE 2 -EP2 signalling mediates excessive production of IL-1b, accounted for lung inflammation and is deleterious for the host.

PGE 2 increases IL-1b production via cAMP production.
Previous studies showed that the PGE 2 -EP2 axis drives cAMP production 32 and mediates IL-1b production 22,31 . First, we determined the relative basal expression of all EP receptors in resting macrophages and examined whether TsV upregulates the expression of EP1-EP4 receptors. We observed that (a) 129sv WT mice and Alox5 À / À mice or (b) C57BL/6 mice were treated with indomethacin (Indo; 2 mg kg À 1 intraperitoneally), 4 h and again 30 min before inoculation with a lethal (180 mg kg À 1 ) dose of TsV. Additional administration of vehicle or indomethacin was performed 4 and 8 h later and the survival was monitored for 8-12 h. (c) C57BL/6 mice were treated with celecoxib (5 mg kg À 1 intraperitoneally), or with (d) EP2 antagonist (AH6809) (5 mg kg À 1 intraperitoneally), 24 h and again 1 h before inoculation of 180 mg kg À 1 of TsV. In all experiments, TsV-inoculated mice treated with vehicle were used as controls. The lungs were excised immediately after an animal died or from mice that survived for 8-12 h; then protein content, PGE 2 release, IL-1b production, LTB 4 release and MPO activity were quantified in the lung parenchyma. The experiment was conducted once with six mice and the error bars denote s.d. *PBS versus vehicle þ TsV; # vehicle þ TsV versus Treatments þ TsV; $ PBS in 129sv WT mice versus PBS in Alox5 À / À mice; & Vehicle þ TsV in 129sv WT mice versus Vehicle þ TsV in Alox5 À / À mice. These differences were considered significant with Po0.05, according to one-way ANOVA with Bonferroni's post-test (for soluble mediators and MPO) or the log-rank test (for survival).
non-stimulated macrophages exhibit a higher expression of EP2 and EP4, low expression of EP3 and no expression of EP1. Interesting, TsV increased the expression levels of Ptger2-4 alone (Fig. 5a). Next, we investigated the effects of PGE 2 on TsV-induced IL-1b and whether EP2/EP4 mediates cAMP increases, as previously described 33 . We observed that TsV-induced IL-1b expression in C57BL/6-peritoneal or immortalized macrophages (J774.1) was higher in PGE 2pre-treated cells and that indomethacin, the EP2 antagonist AH6809 and the EP4 antagonist AH23848 abrogated IL-1b production (Fig. 5b,c). To further evaluate the participation of PGE 2 -cAMP in TsV-induced IL-1b expression, we measured cAMP in the supernatant of J774.1 cells pre-treated as indicated.
While forskolin, the adenylate cyclase agonist, potentiated TsV-induced IL-1b production, H89, the protein kinase A (PKA) inhibitor, reduced it (Fig. 5e). Moreover, PGE 2 increased IL-1b production, while the EP2 antagonist AH6809 and indomethacin inhibited TsV-induced cAMP production (Fig. 5d). Next, we determined whether cAMP and its downstream effector PKA, leading to NFkB activation 34,35 , are involved in TsV-induced IL-1b. To do so, J774.1 cells were pre-treated with forskolin or with H89, and then challenge with TsV. We discovered that, in the presence of TsV, PGE 2, similar to forskolin, amplified the phosphorylation of NFkB p65, while the PKA inhibitor, H89, decreased it (Fig. 5f). The role of NFkB for TsV-induced IL-1b was confirmed by treatment with an NFkB inhibitor (Fig. 5g). Our results suggest that TsV via the PGE 2 -cAMP-PKA-NFkB pathway increases IL-1b production by inflammasome activation (Fig. 5h).
LTB 4 decreases cAMP generation and rescues mice from death.
Exogenously administered LTB 4 rescued Alox5 À / À mice from mortality (Fig. 7a) and decreased lung inflammation as noticed by the production of IL-1b and PGE 2 and protein extravasation, besides increasing neutrophils recruitment to the lungs (Fig. 7b).

COX1/2 inhibition rescues animals from lethal envenomation.
Overproduction of PGE 2 -cAMP-IL-1b accounts for uncontrolled lung oedema and mortality in lethal envenomation. Thus, we next investigated whether COX1/2 inhibition can be used as a therapeutic treatment of scorpion stings. For this purpose, we injected C57BL/6 mice with a lethal dose (180 mg kg À 1 ) or superdose (360 mg kg À 1 ) of TsV and then therapeutically treated the animals with indomethacin as indicated. Under both regimens, indomethacin rescued the mice from the mortality induced by the lethal dose. Moreover, indomethacin strongly inhibited the envenomation by the superdose: 67% and 50% of the mice survived under the regimens '15' and '30 min,' respectively ( Fig. 8a). As our data suggested that the balance between PGE 2 and LTB 4 regulates the amount of IL-1b produced and lung oedema (Supplementary Table 1), we next calculated the ratio of PGE 2 to LTB 4 in mice that received TsV at 180 mg kg À 1 (Fig. 8b). Our data showed that the PGE 2 /LTB 4 ratio affects IL-1b production in vivo and that when LTB 4 is low, higher levels of PGE 2 and IL-1b are produced and that a positive correlation exists between PGE 2 and IL-1b. As illustrated in Fig. 8c, our data show for the first time that, during scorpion envenomation, an intricate and complex network controls lung oedema, cell recruitment and the stung animal's fate. Altogether, these results indicate that COX1/2 inhibition might be an effective treatment for patients stung by a scorpion in emergency rooms.

Discussion
Bee venom activates the inflammasome 14,15 , resulting in eicosanoid production 13,14 . However, whether scorpion venom-induced pulmonary oedema and mortality in humans 1-5 and animals 5,7 are determined by inflammasome-dependent production of bioactive lipids remains unknown. We hypothesized that TsV activates the inflammasome and induces IL-1b, LTB 4 and PGE 2 production leading to lung oedema and mortality. In this study, macrophage-derived NLRP3 inflammasome was identified as a platform essential for a systemic response to TsV, resulting in PGE 2 , IL-1b and LTB 4 production, lung inflammation and mortality. Treatment with IL-1Ra or experiments with Casp1/11 À / À mice confirmed that the inflammasome is also activated by TsV in vivo, resulting in the production of IL-1b and PGE 2 , which are key determinants of the envenomation. The ability of IL-1Ra to control inflammation was reported previously 36 and blocking IL-1b along with IL-1Ra has been proposed as an effective treatment of human inflammatory diseases 37 . Moreover, PGE 2 upregulates IL-1b (refs 22,31), and both of these mediators promote vascular permeability and oedema 20,21,38 , whereas IL-1b recruits neutrophils 39 . To the best of our knowledge, the present report is the first to demonstrate that IL-1Ra protects mice from envenomation, due to the diminished production of PGE 2 and IL-1b and the consequent reduction of lung oedema.
The results demonstrating that the NLRP3 inflammasome is necessary for the LTB 4 release led us to hypothesize that animals deficient in mediators downstream of 5-LO would be unaffected by scorpion envenomation, because it was previously described that LTs are necessary for the recruitment and activation of polymorphonuclear cells and for pulmonary oedema 40,41 . However, Alox5 À / À animals show overproduced PGE 2 and IL-1b in the lungs leading to strong oedema and leukocyte infiltration in the tissue. Data from the literature demonstrated that PGE 2 causes an array of beneficial and deleterious effects during inflammation 42 . Among these effects, PGE 2 presents lung protective effects during chronic inflammation 43,44 . However documented. The finding that Alox5 À / À mice are more sensitive to TsV than 129sv and C57BL/6 mice are underscores the protective role of LTB 4 during scorpion envenomation. The recovery of envenomed Alox5 À / À mice by exogenously administered LTB 4 confirmed its antiedematogenic role, although we cannot disregard a potential role of others eicosanoids such as lipoxins 45 in antiedematogenic effects. In fact, we previously suggested that LTs could present antiedematogenic effects 46 . This was the first demonstration that LTB 4 acts in an autocrine and paracrine manner as an antiedematogenic mediator by controlling PGE 2 and IL-1b production, via cAMP-PKA-NFkB inhibition. Interestingly, our data also indicate that, in scorpion envenomation, IL-1b accounts for neutrophil recruitment to the lungs, but not to the bronchoalveolar space. In fact, decreased IL-1b concentration in IL-1r À / À , Casp1/11 À / À and IL-1Ra-, indomethacin-, celecoxib-and EP2-treated mice correlated with diminished MPO activity in the lung. Although our data indicate that LTB 4 also regulates the intensity of neutrophil recruitment to (a) C57BL/6 mice were inoculated with a lethal (180 mg kg À 1 ) or excessive (superdose; 360 mg kg À 1 ) dose of TsV and were treated with vehicle or indomethacin (Indo, 2 mg kg À 1 ) 15 or 30 min after the venom injection, followed by an additional dose of the treatment 4 and 8 h later. Survival was monitored for 12 h. Significant differences (Po0.05) are marked with an asterisk. The experiment was performed once on six mice and the log-rank test was used to compare TsV þ vehicle versus TsV þ Indo. (b) Using the eicosanoid data obtained from the animals (129sv WT and Alox5 À / À ) inoculated with TsV at 180 mg kg À 1 for 24 h, we demonstrated the correlation between the PGE 2 /LTB 4 balance and the IL-1b production level, in each mouse. Data are representative of one experiment on six mice. (c) Mechanism scheme showing the production of eicosanoids after scorpion envenomation, inflammasome activation and IL-1b release, and indicating that the balance between metabolites of PGE 2 and LTB 4 determines the outcome of inflammasome-mediated envenomation, the severity of lung inflammation and the outcome of envenomation mortality. the lung parenchyma by controlling IL-1b production, methodologically it is difficult to prove, since as observed in Fig. 7b and as previously described 25 , LTB 4 , per se, induces neutrophil recruitment. Altogether, our results show that, during scorpion envenomation, a cross-regulation mechanism exists between PGE 2 , IL-1b, and LTB 4 . According to our data and to one report showing that a high dose of indomethacin decreases the severity of lung oedema induced in rats by one purified scorpion toxin (tityustoxin TsTX) 7 , we hypothesized that inhibition of COX1/2 or COX2 or blocking EP2/EP4-PGE 2 interaction would result in the downregulation of PGE 2 and IL-1b and would abrogate the lethality of TsV envenomation. These treatments increased LTB 4 production and downregulated IL-1b and PGE 2 in the lung and abrogated the mortality induced by a lethal dose of scorpion venom. Moreover, treatment with low dose of indomethacin rescued mice from the death induced by a lethal dose of TsV and significantly increased the survival rate after a high dose of TsV. In this context, we found that the ratio of 5-LO to COX products determined the amount of IL-1b released and the outcome of the envenomation. Furthermore, exacerbated lung inflammation and greater synthesis of IL-1b (ref. 47) and PGE 2 (ref. 48) were previously observed in the presence of reduced level or in the absence of 5-LO products, but the effects of the augmented production of both mediators were not elucidated.
In summary, we can suggest that, in scorpion envenomation, PGE 2 is produced in two waves. The first wave comes from resident macrophages and is inflammasome-independent, but involves the TLR4/MyD88-dependent activation of NFkB and TLR4-dependent activation of AP-1 (ref. 11). Then PGE 2 -EP2/EP4 actions activate cAMP-PKA-NFkB and triggers inflammasome activation and IL-1b release. The inflammasome-mediated IL-1b maturation signals via IL-1R in macrophages and induces the second and more prominent wave of PGE 2 and discrete amounts of LTB 4 . Inflammasome-dependent PGE 2 production in turn acts on EP2/EP4 receptors on macrophages (and perhaps neutrophils) and drives IL-1b production via the COX1/2-cAMP-PKA-NFkB axis. Meanwhile, TsV, sensed by TLR2, TLR4 and CD14 on macrophages induces neutrophil chemoattractants such as TNF-a, IL-6 and LTB 4 (refs 11,39). Therefore, LTB 4 , supplied by resident macrophages and newly recruited neutrophils, has autocrine and paracrine effects and, via BLT1/2, downregulates cAMP, restrains PGE 2 and IL-1b production, consequently controlling lung oedema and possibly neutrophil recruitment. However, in contrast to our results, the regulation of NLRP3 inflammation by cAMP was previously demonstrated 49 , suggesting that more studies are needed. We propose that, in non-severe envenomation, the production of mediators is self-controlled and the patient presents discrete lung oedema, but survives. However, in severe envenomation, the production of PGE 2 and IL-1b is highly induced and LTB 4 delayed production is not sufficient to control the oedema and the patient succumbs.
Our findings may open new avenues of research and we speculate that patients with low levels of LTB 4 or high levels of PGE 2 , which is the case for patients with immunodeficiencies 50,51 , are more likely to die after scorpion envenomation. Given the fact that many non-steroidal drugs are safe to use in humans, we anticipate that therapeutic strategies aiming to block PGE 2 synthesis during envenomation would be effective, safe, and lowcost attractive therapies.

Methods
Animals. Female or male mice (6-8 weeks old) were used for in vivo and in vitro experiments. The mice were matched by sex and age in all procedures. Sample size was determined based on previous studies from our laboratory and literature and considering an alpha and beta errors of 0.05 and 0.20, respectively. Nlrp3 À / À T. serrulatus venom. The venom was extracted from several T. serrulatus scorpions by electric stimulation, dried at room temperature in a vacuum desiccator, and stored at À 20°C. Before the experiments, TsV was diluted in PBS and filtered through a 0.22-mm sterilizing membrane (Millipore, Tullagreen, CO, EUA). Limulus Amebocyte Lysate Test (LAL; QCL-1000, Bio Whittaker, Cambrex Company, Walkersville, MD, USA) was performed to detect LPS in the TsV samples, according to the manufacturer's instructions. No LPS was detected in any TsV samples.
Activation of BMDMs and peritoneal macrophages by TsV. BMDMs were isolated from C57BL/6 (WT), Nlrc4 À / À , Nlrp3 À / À , Asc À / À and Casp1/11 À / À , mice as described previously 56 . After differentiation, the culture medium was replaced with RPMI 1640 containing 10% (v/v) fetal bovine serum and 5% (v/v) of L929-cell conditioned medium. BMDMs were seeded (2 Â 10 5 per well in 1 ml) in a 24-well plate and either pre-treated or not-pre-treated with Escherichia coli LPS (1,000 ng ml À 1 ; Sigma-Aldrich, St Louis, MO, USA) for 4 h and then incubated with TsV (50 mg ml À 1 ) for 24 h at 37°C in a humidified atmosphere containing 5% of CO 2 . Next, the supernatants were harvested for IL-1b and IL-6 quantification using an ELISA kit (R&D Systems, Minneapolis, MN, USA). In another set of experiments, resident peritoneal macrophages from naive C57BL/6 mice, harvested by peritoneal washes with RPMI-1640 and the J774.1 cell lineage of peritoneal macrophages, obtained from cell culture were plated at the density of 2 Â 10 5 cells per well in 200 ml of incomplete RPMI-1640 supplemented with antibiotics. The cells were then cultured at 37°C (5% CO 2 ) for 2 h. Next, the supernatants were removed and the cells were pre-treated or not with either LTB 4 for 10 min (0.01, 0.1, 1, 10 or 100 nM; Cayman Chemical, Ann Arbor, MI, USA) 57,58 or PGE 2 (0.01, 0.1, 1 or 10 mM; Cayman Chemical) 59 or forskolin 60 (50 mM; Sigma-Aldrich) in 200 ml of incomplete RPMI. LTB 4 and PGE 2 from ethanol stock solution were diluted in the cell culture medium and the same concentration of ethanol (maximum 0.1%) was added to the medium only (control). Forskolin from DMSO stock solution was diluted in the cell culture medium and the same concentration of DMSO (maximum 0.1%) was added to medium only (control). Next, the cells were stimulated with TsV (50 mg ml À 1 ) under the same experimental conditions, and after 24 h at 37°C in a humidified atmosphere containing 5% of CO 2 , supernatants were collected for IL-1b quantification.
Pore formation assay and inhibition of the K þ efflux. BMDMs (1 Â 10 5 ) were plated in 96-well tissue culture dishes for 16 h at 37°C, 5% CO 2 . Next, cells were stimulated with LPS (1 mg ml À 1 ) during 4 h and then treated with 50 mg ml À 1 of TsV venom for 17 h. Pore-forming activity was measured by the quantification of propidium iodide uptake 61 . As a positive control, the cells were treated with Triton-X (9%). Throughout the treatment, plates were incubated at 37°C in a SpectraMax fluorimeter plate reader and the propidium iodide fluorescence was measured every 10 min. For the inhibition of the K þ efflux, BMDMs from C57BL/6 mice were stimulated with TsV (50 mg ml À 1 ) plus NaCl (50 mM) or KCl (50 mM) for 24 h (ref. 62). The supernatant was used to detect IL-1b by ELISA.
In vitro pharmacological treatments. J774.1 macrophages were plated at the density of 2 Â 10 5 cells per well in 200 ml of serum-free RPMI supplemented with antibiotics. The cells were then cultured at 37°C in 5% CO 2 for 2 h. Next, the supernatants were removed, and the cells were treated or not with specific inhibitors/antagonists for 30 min: indomethacin (10 mM; Cayman Chemical); AH6809 (1 mM; Cayman Chemical); AH23848 (1 mM; Cayman Chemical); U-75302 (0.1 and 1 mM; Cayman Chemical); and NFkB Activation Inhibitor (20 nM; Calbiochem, Darmstadt, Germany). H89 dihydrochloride hydrate (25 mM; Sigma-Aldrich) was added for 2 h in the cell culture medium before stimulation. AH6809 and U-75302 from ethanol stock solutions were diluted in cell culture medium and the same concentration of ethanol (maximum 0.1%) was added to the medium only (control). The AH23848 and NFkB inhibitor from DMSO stock solutions were diluted in the cell culture medium and the same concentration of DMSO (maximum 0.1%) was added to the medium only (control). All compounds were diluted in 200 ml of serum-free DMEM, and the same solution with solvent diluents was used as control. After treatment, the cells were stimulated with TsV (50 mg ml À 1 ) under the same experimental conditions and after 24 h at 37°C in a humidified atmosphere 5% of CO 2 , the supernatants were collected for IL-1b quantification.
Quantitative PCR with reverse transcription. RNA was extracted using a guanidine-based column method, according to the manufacturer protocol (Purelink, Ambion) and the quantity of RNA was determined by means of a fluorometric assay (Qbit, Invitrogen, Carlsbad, CA, USA). Complimentary DNA (cDNA) was synthesized from 1 mg of total RNA (High Quality cDNA Reverse Transcriptase Kits, Applied Biosystems, Foster City, CA, USA). Aliquots (2 ml) of the total cDNA were amplified by quantitative reverse transcriptase-polymerase chain reaction (quantitative PCR with reverse transcription (qRT-PCR)) using TaqMan primers for Ptgs2, Ptges2, Ptger1, Ptger2, Ptger3, Ptger4, Alox5, Alox5ap, Ltb4r1 and Ltb4r2 in a StepOne Plus machine (Applied Biosystems). Gapdh and Actb were used as reference genes. Reactions were performed in duplicate and amplification was performed under the following conditions: denaturation at 95°C for 2 min; followed by 40 cycles of 95°C for 2 s and 60°C for 20 s. The results were normalized to the expression levels of the endogenous internal controls, Actb and Gapdh. The DDCt method was used for the analysis, and the data were expressed as n-fold difference relative to the control.
Measurement of intracellular cAMP. For intracellular cAMP measurement, 1 Â 10 6 macrophages (peritoneal or J774.1) per well (48-well plate) were incubated with 1 ml of DMEM-serum free with or without PGE 2 (1 mM) for 3 min, followed or not by TsV (50 mg ml À 1 ) for 5 min at 37°C in 5% CO 2 . In addition, cells were treated or not with indomethacin (10 mM), EP2 antagonist-AH6809 (1 mM) for 30 min before PGE 2 (1 mM), or TsV (50 mg ml À 1 ). In other sets of experiments, cells were pre-treated or not with LTB 4 (10 nM) for 1 min, and then stimulated with PGE 2 (1 mM) or TsV (50 mg ml À 1 ). LTB 4 , PGE 2 , AH6809 from ethanol stock solution were diluted in the cell culture medium and the same concentration of ethanol (maximum 0.1%) was added to medium only (control). All stimuli (or ethanol) were diluted in serum-free DMEM medium. Culture supernatants were aspirated and the cells were lysed by incubation for 10 min with 0.1 M HCl at room temperature, followed by disruption using a cell scraper 27 . Intracellular cAMP was quantified by ELISA, using an acetylation protocol, according to the manufacturer (Enzo Life Sciences, Farmingdale, NY, USA).
Immunoassay of phosphoproteins. To assess the levels of transcription factors (phosphoproteins) in J774.1 macrophages, a sandwich ELISA kit was used (PathScan Phospho-NFkB p65 (Ser536) and Total NFkB p65 ELISA Kit-Cell Signaling, Danvers, MA, USA). J774.1 macrophages were plated in 24-well micro-culture plates at a density of 2 Â 10 6 cells per well in serum-free DMEM medium and cultured at 37°C in a humidified 5% CO 2 atmosphere for 2 h. The cells were then stimulated with TsV (50 mg ml À 1 ) for 2 h at 37°C in a 5% CO 2 atmosphere. In another set of experiments, PGE 2 (10 mM) or forskolin (100 mM) were added in the culture 10 min before TsV stimulation. In addition, PKA-inhibitor H89 (25 mM) was added 2 h before TsV stimulation. After treatments, the adherent macrophages were washed with ice-cold PBS. Then, cell lysis buffer (Cell Signaling), containing protease and phosphatase inhibitors, was added to the cell culture. The lysates were centrifuged at 10,000g for 10 min at 4°C in tubes and the supernatant was directly used or stored at À 80°C. The total protein content of the lysates was quantified, and normalized for used in ELISA kits according to the manufacturer's instructions (Cell Signaling). Briefly, the lysates were diluted with sample diluent and 100 ml of the lysate was added to wells that had been pre-coated with the primary antibody. The plate was incubated overnight at 4°C and then washed. Next, the plate was incubated with the detection antibody and horseradish peroxidase-conjugated secondary antibody, with subsequent washes. The substrate 3,3 0 ,5,5 0 -tetramethylbenzidine was added to the wells, the reaction was stopped with acid solution, and the absorbance of the samples was read at 450 nm. The results are expressed as relative percentages of the levels of phosphorylated NFkB to total NFkB, normalized by 100 mg of total protein.

Dose-response experiments with TsV in vivo.
In all experiments under all conditions, the mice were weighed before an i.p. injection of TsV. To determine the sublethal, lethal and excessive doses in WT mice (C57BL/6 and 129sv) or Alox5 À / À , we tested TsV at 60, 120, 180 and 360 mg kg À 1 in 200 ml of PBS and the survival was monitored for 24 h. Death of each animal was recorded immediately and used to calculate the rate of survival (in percentage points). On the basis of these results ( Supplementary Fig. 1), the TsV dose of 120 mg kg À 1 was considered sublethal and used to assess lung inflammation 4 h later; 180 mg kg À 1 was designated as lethal, and 360 mg kg À 1 as a superdose. These doses were used in survival experiments (with or without treatment) as indicated below. The animals were monitored for 8, 12 or 24 h. The animals that were inoculated with 200 ml of PBS (i.p.) served as negative controls. The mice were killed with an overdose of a solution of 20% ketamine and 10% xylazine.
In vivo experiments and drug treatment. IL-1r À / À , Casp1/11 À / À and C57Bl/6 (WT) mice without treatment were inoculated with a sublethal or lethal dose of TsV (or PBS) as described above. Alox5 À / À mice and 129sv mice were pre-treated or not with IL-1 receptor antagonist (IL-1Ra) (ref. 63) at 10 mg kg À 1 , i.p., 1 h before and again 1 h after the sublethal or lethal TsV injection. IL-1Ra was kindly provided by Dr Stephen Poole, from the National Institute for Biological Standards and Control (South Mimms, Hertfordshire, UK). In a specific experiment, the mice were either treated or not treated with MK886 (5-LO inhibitor, 5 mg kg À 1 i.p., in 200 ml of 1% alcohol in water) 64 , indomethacin (COX1/2 inhibitor, 2 mg kg À 1 i.p. in 200 ml of Tris[hydroxymethyl]aminomethane-HCl; TRIS-HCl, pH 8.2) 65 , SC-560 (selective COX1 inhibitor, 3 mg kg À 1 i.p., in 200 ml of PBS, Sigma-Aldrich) 66 , celecoxib ( COX2 inhibitor, 5 mg kg À 1 i.p., in 200 ml of water) 64 or EP2 antagonist (AH6809, 5 mg kg À 1 i.p., in 200 ml of PBS, Cayman Chemical) 67 . The drugs (MK886 or indomethacin) or vehicles were administered four times, at 4 h and 0.5 h before and again 4 and 8 h after the lethal dose of TsV. The others drugs (SC-560, celecoxib, and EP2 antagonist) or vehicles were administered 1 day and again 1 h before the i.p. injection of lethal dose of TsV (180 mg kg À 1 ). In other experiments, the Alox5 À / À mice were treated or not with LTB 4 (50 ng per mice, intranasal (i.n.) administration, in 20 ml of PBS, Cayman Chemical) 68 . The LTB 4 or vehicle (PBS þ 0.05% of ethanol) were administered 2 h and 0.5 h before the dose lethal of TsV (180 mg kg À 1 ). The lungs were excised immediately after death or from mice survivors that were killed 8-12 h after the injection of TsV or vehicle. In some sets of experiments, two groups of mice were inoculated with PBS or a sublethal (120 mg kg À 1 ) dose of TsV and, in only one, BAL fluids were collected 4 h later, to count the total cell number and neutrophils, as described previously 47 . In the other group, without BAL, the lungs were excised and weighed and 2 mg of tissue was homogenized in 2 ml of incomplete RPMI. After centrifugation (400g for 10 min), the supernatants were transferred to new tubes, split into two samples of 1 ml and stored at À 80°C until use. One sample was used for IL-1b and protein quantification analysis and the other for PGE 2 and LTB 4 measurement. For analysis of MPO activity, one lobule of a lung was cut out, immediately frozen in liquid nitrogen, and stored at À 80°C until use. In the therapeutic protocol, the lethal dose (180 mg kg À 1 ) and superdose (360 mg kg À 1 ) of TsV were injected and indomethacin (2 mg kg À 1 i.p.) or vehicle were administered either 15 or 30 min after and again 4 and 8 h later. Mice survivors were killed 12 h after the envenomation.
Quantification of inflammatory markers. IL-1b present in the supernatant of a lung homogenate or cell culture and IL-6 in cell culture were quantified using IL-1b (BD Biosciences, Franklin Lakes, NJ, USA or R&D Systems) and IL-6 (R&D Systems) ELISA kits. Lipids were purified from 1 ml of filtered supernatants from the lung homogenates using Sep-Pak C 18 cartridges (Waters Corp). Measurement of LTB 4 and PGE 2 levels was performed by means of an EIA assay (Enzo Life Sciences). The supernatants of the lung homogenates were used for the measurement of MPO activity as described previously 69 and the quantification of total protein was conducted using the Coomassie Protein Assay Reagent (Pierce Chemical). In one set of experiments, peritoneal cavity fluid and 1 ml of peripheral blood were also collected to count total cells and neutrophils, using a Neubauer chamber and cytospin preparation. Immediately after an animal died or in mice survivors, the lungs were always weighed before processing and these data were used for the calculation of the lung/body weight index to evaluate oedema. When indicated in figures, the samples of lung parenchyma were stained with H&E for histological analysis and blind observers performed the analysis. In one set of experiments, three mice received 200 ml of a 1% Evan blue solution (Sigma-Aldrich) intravenously 60 min before euthanasia and the lungs were excised, weighed and photographed.
Statistical analysis. For comparison of multiple groups, we performed one-way analysis of variance (ANOVA) followed by Bonferroni's post-test. The differences between any two groups were evaluated using two-tailed Student's t-test. All calculations were performed in the GraphPad Prism 5.0 software (GraphPad, San Diego, CA, USA). Differences in survival were analysed using the log-rank test. Differences with Po0.05 were considered statistically significant.