Genetic inactivation of the phospholipase A₂ activity of peroxiredoxin 6 in mice protects against LPS-induced acute lung injury

Abstract

Peroxiredoxin 6 (Prdx6) is a multifunctional enzyme that serves important antioxidant roles by scavenging hydroperoxides and reducing peroxidized cell membranes. Prdx6 also plays a key role in cell signaling by activating the NADPH oxidase, type 2 (Nox2) through its acidic Ca²⁺-independent phospholipase A2 (aiPLA2) activity. Nox2 generation of O₂^·−, in addition to signaling, can contribute to oxidative stress and inflammation such as during sepsis-induced acute lung injury (ALI). To evaluate a possible role of Prdx6-aiPLA₂ activity in the pathophysiology of ALI associated with a systemic insult, wild-type (WT) and Prdx6-D140A mice, which lack aiPLA₂ but retain peroxidase activity were administered intraperitoneal LPS. LPS-treated mutant mice had increased survival compared with WT mice while cytokines in lung lavage fluid and lung VCAM-1 expression, nitrotyrosine levels, PMN infiltration, and permeability increased in WT but not in mutant mice. Exposure of mouse pulmonary microvascular endothelial cells in primary culture to LPS promoted phosphorylation of Prdx6 and its translocation to the plasma membrane and increased aiPLA₂ activity as well as increased H₂O₂ generation, nitrotyrosine levels, lipid peroxidation, NF-κB nuclear localization, and nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3 (NLRP3) inflammasome assembly; these effects were not seen in Nox2 null cells, Prdx6-D140A cells, or WT cells pretreated with MJ33, an inhibitor of aiPLA₂ activity. Thus aiPLA₂ activity is needed for Nox2-derived oxidant stress associated with LPS exposure. Since inactivation of aiPLA₂ reduced mortality and prevented lung inflammation and oxidative stress in this animal model, the aiPLA₂ activity of Prdx6 could be a novel target for prevention or treatment of sepsis-induced ALI.

INTRODUCTION

The acute lung injury (ALI) syndrome can result from both direct lung as well as systemic insults (57). Sepsis, a life-threatening inflammatory response to infection, is an important underlying systemic cause (7, 57). The exposure to bacterial endotoxin is a significant contributor to sepsis-induced ALI (21) through the activation of endothelial and innate immune cells resulting in an increased production of inflammatory cytokines, the recruitment of inflammatory cells to the lung, and the generation of reactive oxygen species (ROS) (23). Excessive ROS generation can directly cause tissue damage through oxidation of cellular components including lipids and proteins and indirectly can lead to increased lung permeability through its effects on Ca²⁺-mediated cell signaling (19). Furthermore, ROS-mediated signaling is crucial for the activation of the transcription factor nuclear factor-κ-light-chain-enhancer of activated B cells (NF-κB) and of the inflammasome nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3 (NLRP3), both associated with lung inflammation (30, 38, 50). To extend our previous studies using intratracheal LPS as a model for ALI (34), the present study utilized systemic (intraperitoneal) administration of LPS; this route of administration induces systemic symptoms and represents a model for sepsis, albeit in the absence of active bacterial infection.

From previous studies using various lung disease models, it is known that NADPH oxidase type 2 (Nox2) is the major source of endothelial ROS in experimental ischemia/reperfusion, exposure to hyperoxia, and bacterial lipopolysaccharide (LPS) instillation (2, 19, 39, 44, 46, 48). We have shown that the widely expressed protein peroxiredoxin 6 (Prdx6) plays a crucial role in endothelial ROS production through its role in the production of mediators that regulate Nox2 activation (6, 53). Prdx6 is a multifunctional enzyme with peroxidase, phospholipase A₂ (PLA₂), and lysophosphatidylcholine acyltransferase activities (10, 12). The PLA₂ activity of Prdx6 has been called phospholipase A₂ (aiPLA₂) since initial studies showed that the native protein functions at acidic pH and is Ca²⁺ independent. The three enzymatic activities of Prdx6 participate in the scavenging of lipid hydroperoxides and in the repair of peroxidized cell membranes (10, 16, 36). Prdx6 null mice show increased mortality and susceptibility to ALI when exposed to stressors associated with increased oxidant generation such as hyperoxia, paraquat ingestion, or intratracheal instillation of LPS (55, 56, 60). In contrast, pharmacological inhibition of aiPLA₂ activity blunts ROS generation and ameliorates lung inflammation associated with hyperoxia, ischemia/reperfusion, and intratracheal LPS instillation (3, 34, 35). Thus the peroxidase and aiPLA₂ activities of Prdx6 may play counterbalancing roles in mediating the onset and progression of ALI.

In the present study, we used a genetic mouse model that lacks aiPLA₂ activity while maintaining Prdx6 peroxidase activity to study the role of aiPLA₂ in LPS-induced ALI. We found that genetic inactivation of aiPLA₂ reduces mortality, lung inflammation, and Nox2-driven oxidative stress after intraperitoneal LPS. These results suggest that aiPLA₂ could be an important therapeutic target to limit the degree of sepsis-induced lung injury.

METHODS

Reagents.

Amplex red and horseradish peroxidase (HRP) were purchased from Invitrogen (Carlsbad, CA). Liperfluo was from Dojindo Molecular Technologies (Rockville, MD). Lithium 1-(hexadecyloxy)-3-(2,2,2-trifluoroethoxy)propan-2-yl methyl phosphate (MJ33), LPS derived from Escherichia coli O111:B4 cell membranes and purified by gel-filtration chromatography (LPS), protease and phosphatase inhibitors, and the Duolink assay kit were obtained from Sigma (St. Louis, MO). 1-Palmitoyl,2-[³H] \palmitoyl,sn-glycero-3-phosphocholine (³H-DPPC) and 1-palmitoyl,2-linoleoyl,sn-glycero-3-phosphocholine (PLPC) were from American Radiolabeled Chemicals (St. Louis, MO); PLPC was treated as described previously to generate a lipid hydroperoxide as substrate for the peroxidase assay (15).The anti-CD144-Alexa Fluor 488 antibody was obtained from eBioscience (cat no. 53-4301-80; San Diego, CA). The myeloperoxidase (MPO) activity kit was from Cayman Chemical (Ann Arbor, MI). IL-1β, CCL2/monocyte chemoattractant protein-1 (MCP-1), and tumor necrosis factor-α (TNF-α) ELISA kits and the anti-MPO (cat no. MAB3174) antibody were purchased from R&D Systems (Minneapolis, MN). The wheat germ agglutinin-Alexa Fluor 594 conjugate (W11262) was from Molecular Probes (Eugene, OR). The anti-3-nitrotyrosine antibody (17) was a kind gift of Dr. Harry Ischiropoulos (Children’s Hospital of Philadelphia). Anti-NIMP-R14 (cat no. sc-59338) and anti-NF-кB p50 H-119 (cat no. sc-7178) antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-ICAM-1 (cat no. ab119871) antibody was purchased from Abcam (Cambridge, MA). Anti-GAPDH (cat no. 2118), anti-NLRP3 (cat no. 15101), and the monoclonal anti-VCAM-1 (cat no. 32653) were from Cell Signaling Technology (Danvers, MA). The anti-CD31 antibody (cat no. 553370) was purchased from BD Biosciences (Billerica, MA). The anti-ASC (cat no. 653902) antibody was from BioLegend. The antibody against Prdx6 was made to a peptide (aa 196-211 in human Prdx6) by Strategic Solutions (Newark, DE), and its specificity has been demonstrated previously (40). The antibody to phosphorylated Prdx6 was generated by Proteintech (Chicago, IL) to a 15 amino acid peptide from Prdx6 surrounding a phosphorylated Tyr and its specificity also has been validated previously (6, 33).

Animals.

All animal care and experimental use was approved by the University of Pennsylvania Animal Care and Use Committee. Animals were bred and maintained in University of Pennsylvania animal care facilities. C57Bl/6J wild-type (WT) and NOX2 (gp91^phox) null breeder pairs were obtained from The Jackson Laboratory (Bar Harbor, ME). The generation of Prdx6 null and Prdx6-D140A “knockin” mice has been described previously (36, 40). These mice have been fully backcrossed to the C57Bl/6J background. The Prdx6-D140A knockin mouse carries a single amino acid mutation (D140A) in one of the three amino acids comprising the PLA₂ catalytic triad; this mutation inactivates aiPLA₂ activity in the knockin mice (12). Animals were genotyped to confirm the presence of the Prdx6-D140A mutation using previously published primer sequences (36) (Fig. 1A). However, these mice still retained the neomycin resistance cassette (NeoR) that was used in clonal selection. For the present study, the NeoR cassette was excised from the genome of Prdx6-D140A knockin mice through use of flanking flippase recombinase target (FRT) sites (Fig. 1B). We used the following primers and conditions to detect the presence of the insert between FRT sites: forward primer, 5′-GCAGCTTTACCGTCATGTACTAACG-3′; and reverse primer, 5′-GGAGCTGGGATTTTGGGGTTACG-3′. PCR was 35 cycles; within each cycle, denaturation was for 30 s at 95º, annealing was 15 s at 62º, and elongation was 1 min at 72º. The expected product if the insert was present would be 2,350 nucleotides in length; if the insert between the FRT sites was deleted, the expected product would be 522 nucleotides in length. To confirm the absence of the neomycin resistance cassette, the same conditions and the same forward primer was used as in the above PCR but a different oligonucleotide, complementary to a sequence inside the neomycin resistance cassette, was substituted for the reverse primer: 5′-GTCTGTTGTGCCCAGTCATAGCCG-3′. If the neomycin cassette was present, the expected PCR product would be 1,107 nucleotides in length; if it had been deleted, no PCR product would be detected. The FLP gene itself was removed by breeding and confirmed using primers and protocols designed by The Jackson Laboratories: https://www2.jax.org/protocolsdb/f?p=116:5:0::NO:5:P5_MASTER_PROTOCOL_ID,P5_JRS_CODE:26666,003946.

Fig. 1.

Peroxiredoxin 6 (Prdx6)-D140A mutation abolishes acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) activity without affecting Prdx6 expression or peroxidase activity. A: genotyping results of wild-type (WT) and Prdx6-D140A knockin mice. B: PCR showing excision of the neomycin resistance cassette NeoR. C: WT and Prdx6-D140A knockin lungs stained with hematoxylin and eosin; the scale bar = 100 µm. D: representative Western blot for Prdx6 expression in WT, Prdx6 null (KO), and Prdx6-D140A knockin (KI) lungs. Prdx6 antibodies were generated by Covance Research Products (Denver, CO) and have been previously characterized (6, 58). E–G: enzymatic activities of Prdx6 in WT, Prdx6 null, and Prdx6-D140A knockin lungs: aiPLA₂ activity using radiolabeled phospholipid substrate (E) and peroxidase activity with H₂O₂ (F) and phospholipid hydroperoxide (PLPCOOH) substrates (G). Error bars indicate the means ± SE for n = 3. Statistical significance was tested using ANOVA with Bonferroni post hoc test. *P < 0.05 vs WT.

Animal model of sepsis-induced ALI.

Mice were used at ~8 wk of age with a body weight of 20–23 g. WT and Prdx6-D140 knockin mice were injected intraperitoneally with either LPS (10 mg/kg body wt) or an equal volume of saline. After 24 h, animals were euthanized, the lungs were cleared of blood and then either were lavaged by instilling 1 ml of PBS through a tracheal catheter three times, were frozen by immersion in liquid nitrogen, or were fixed with 4% paraformaldehyde. Bronchoalveolar lavage fluid (BALF) was centrifuged for 8 min at 375 g. Supernatants were aliquoted, and pellets were resuspended in 200 µl of PBS. Nucleated cells were counted using a hemocytometer. Cell pellets and supernatants were snap-frozen and stored at −80°C for further analysis. Fixed lungs were embedded in paraffin and sectioned at the Children’s Hospital of Philadelphia Pathology Core Facility. For survival studies, animals were closely monitored and euthanized 4 days after initial LPS administration.

BALF analysis.

TNF-α, interleukin-1β (IL-1β), and MCP-1 expression were measured in BALF using commercial ELISA kits. Total protein in the BALF was measured using the Quick Start Bradford Dye reagent (Bio-Rad, Richmond, CA) with bovine γ-globulin as standard. Frozen BALF cell pellets were sonicated and MPO activity was measured using a commercial kit. Concentration measurements in the BALF were multiplied by BALF volume to determine total yield per lung; this latter value was normalized to the mouse weight.

Isolated lung perfusion.

Intravascular oxidant generation was measured using Amplex red and HRP in isolated perfused lungs as previously described (14). WT and Prdx6-D140A mice were injected intraperitoneally with either LPS or saline. After 24 h, mice were anesthetized and the lungs were cleared of blood, placed in a temperature-controlled perfusion chamber, and continuously ventilated through a tracheal cannula with 5% CO₂ in air. Lungs were perfused in a recirculating system with Krebs-Ringer bicarbonate buffer containing 10 mM glucose, 3% fatty-acid free bovine serum albumin, 50 μM Amplex red, and 25 μg/ml HRP. Aliquots of the perfusate were removed at 15-min intervals and its fluorescence intensity was measured (excitation/emission: 545/610 nm) using a spectrofluorimeter (Photon Technology International, Birmingham, NJ).

Isolation of primary endothelial cells.

Pulmonary microvascular endothelial cells (PMVECs) were isolated from lungs of WT, Nox2 null, and Prdx6-D140A knockin mice using collagenase, anti-CD31 antibodies, and magnetic Dyna beads (Dynal, Oslo, Norway) as reported previously (6). A second selection was conducted by sorting cells labeled with CD144-Alexa Fluor 488 antibodies using flow cytometry. The endothelial phenotype of the preparation was confirmed by evaluating cellular uptake of DiI-acetylated low-density lipoprotein and immunostaining for five endothelial markers as described previously (6).

Measurement of oxidant generation in intact endothelial cells.

Extracellular hydrogen peroxide generation was measured using Amplex red and HRP as described previously (53) in WT, Nox2 null, and Prdx6-D140A knockin PMVECs treated with 1 µg/mL LPS for 8 h. Intracellular H₂O₂ generation was measured after LPS treatment in WT PMVECs that, as shown previously, stably express the genetically encoded pHyPer-Cyto intracellular hydrogen peroxide sensor (19, 37). The HyPer-expressing cells, a gift of Dr. Madesh Muniswamy (Temple University), were maintained in culture as described previously (19, 37). In some experiments, cells were pretreated for 30 min with MJ33 (10 µM), a pharmacological inhibitor of aiPLA₂ activity. Fluorescence was measured (excitation: 488 nm; emmission: 520 nm) using an epifluorescence microscope (Nikon Diaphot TMD) with Metamorph software. Fluorescence intensity was quantified using ImageJ software.

Immunofluorescence.

Paraffin-embedded tissue sections were deparaffinized and rehydrated using xylene and ethanol, permeabilized, blocked, and incubated overnight at 4°C with primary antibodies as previously described (53). Cells were fixed for 20 min at −20°C with an ice-cold 1:1 solution of acetone/methanol and treated as described above for tissue sections. For phosphorylated Prdx6 (P-Prdx6) membrane localization studies, cells were fixed with 4% paraformaldehyde for 30 min at 4°C and incubated with wheat germ agglutinin-Alexa 594 conjugates (5 μg/ml) for 10 min in the dark before permeabilization; this antibody, while indicating Prdx6 phosphorylation, also shows nonspecific nuclear staining (6). The following dilutions were used for primary antibodies: 1:50 (NIMP-R14), 1:80 (MPO), 1:250 (P-Prdx6), 1:100 (ICAM-1), 1:50 (NF-κBp50), and 1:200 (VCAM-1) and 7 μg/ml 3-nitrotyrosine, Alexa Fluor 488 or 594 specific secondary antibodies were used at a 1:200 dilution. Samples were imaged using a Zeiss Meta 510 or a Zeiss 780 laser-scanning confocal microscope with Zen software. Nuclei were counterstained with either propidium iodide, Sytox green or DAPI (Invitrogen). Fluorescence was quantified using ImageJ software.

In situ proximity ligation assays.

NLRP3 inflammasome assembly was detected by in situ proximity ligation (Duolink) assay using a mouse antibody against the adaptor protein ASC (apoptosis-associated speck-like protein containing a COOH-terminal caspase recruitment domain) and a rabbit antibody against the NLRP3 subunit. PMVECs were treated with LPS for 8 h, fixed with acetone/methanol, permeabilized, blocked, and incubated overnight with primary antibodies diluted 1:200. Proximity (<40 nm) of ASC and NLRP3 was detected using a Duolink kit containing rabbit and mouse secondary antibodies. Nuclei were counterstained with Sytox green. Cells were imaged using a confocal microscope. Rabbit and mouse IgGs were used as negative controls. Quantitative analysis was performed using the Fiji program of ImageJ software; 10 sections were evaluated and counted using a “plugin” created in our laboratory for equal number of cells (43). NLRP3:ASC interactions were normalized to the number of nuclei.

Western blot analysis.

To study protein expression, lungs were homogenized and cells were sonicated in RIPA buffer containing protease and phosphatase inhibitors. Proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes, blocked with Odyssey Blocking Buffer, and incubated overnight with primary antibodies diluted 1:250 (phospho-Prdx6), 1:1,000 (NLRP3), or 1:1,500 (Prdx6). Membranes were washed and incubated with IRDye 800CW secondary antibodies and imaged using a two-color Odyssey (Li-Cor, Omaha, NE) or an Azure c500 ECL system (Azure Biosystems, Dublin, CA). Protein loading was evaluated by stripping the blots and re-probing them for GAPDH. Individual bands were quantified using ImageJ software (https://imagej.nih.gov).

Lipid peroxidation.

Lipid peroxidation was visualized in WT and Prdx6-D140A PMVECs using Liperfluo. Cells were treated with 1 µg/ml LPS for 16 h. Liperfluo was dissolved in DMSO to make a stock solution and subsequently diluted in PBS. Liperfluo in DMSO was added for the last 30 min of the incubation at 10 µM final concentration. Cells were rinsed with phenol red-free medium and observed using a Nikon epifluorescence microscope with Metamorph software (https://www.moleculardevices.com). Fluorescence intensity was quantified using ImageJ. Data are reported as change from corresponding PBS control.

Enzyme activities.

aiPLA₂ activity was measured in cells and lung homogenates from the liberation of [³H]palmitate from ³H-DPPC radiolabeled liposomes in Ca²⁺-free buffer at pH 4 as previously described (14, 53).The peroxidase activity in the presence of GSH with both H₂O₂ and PLPCOOH substrates was measured using the previously described protocol (15) with minor modifications (36). Activities were normalized to total protein content and are expressed in nanomoles per hour per milligrams of protein.

Statistics.

Results are presented as means ± SD or SE along with n = the number of biological replicates. Statistical significance was assessed using GraphPad Prism 7 software (https://www.graphpad.com). Group differences were evaluated by t-test or one-way ANOVA followed by the Bonferroni post hoc test. Differences between mean values were considered statistically significant at P < 0.05.

RESULTS

Comparison of mice with Prdx6-D140A mutation with WT.

The Prdx6-D140A mice that were maintained in our animal facility developed and reproduced normally and remained physically active, similar to WT, during ~12 mo of observation. Aside from minor differences related to lung perfusion and fixation, the lung histology of WT and Prdx6-D140A mice was similar (Fig. 1C). Although there were some borderline differences, the measured parameters of BALF and lung function under control conditions also were similar between the WT and mutant mice. One borderline difference was the protein content of the BALF from D140A mice, which appeared to be higher compared with WT (Fig. 2); however, this result was not statistically significant (P = 0.0722). Furthermore, control lung wet-to-dry weight ratios for WT vs DI40A mice also were not different (Table 1). Likewise, TNF-α levels in BALF appeared elevated in the control mutant lungs compared with WT(Fig. 3), but this also was not statistically significant (P = 0.1389). Thus the WT and D140A mutant lungs under basal conditions appear to have been similar .

Fig. 2.

Genetic inactivation of acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) reduces mortality and prevents sepsis-induced lung inflammation. Wild-type (WT) and peroxiredoxin 6 (Prdx6)-D140A knockin (KI) mice were treated with 10 mg/kg LPS intraperitoneally. A: survival plots constructed using the Kaplan-Meier estimator. B: protein content of bronchoalveolar lavage fluid (BALF). C: total cells in BALF. D: myeloperoxidase (MPO) activity of cells in the BALF. E: lung staining for the presence of PMN as indicated by MPO (green, n = 4) using an antibody from R&D Systems (cat no. MAB3174; Minneapolis, MN). F: staining of lungs with anti-PMN antibody (NIMP-R14, green; cat. no. sc-59338; Santa Cruz Biotechnology) (n = 4). E and F: nuclei are counterstained with propidium iodide (red) and the scale bars = 20 µm. Blue circles and red squares represent individual animals per experiment. Error bars indicate the means ± SD. Group differences (saline vs. LPS) were evaluated by t-test, and the P value for comparison of the 2 groups is indicated; no P value is shown for those comparisons where P > 0.05.

Table 1.

Lung wet-to-dry weight ratio

	Wild Type	Prdx6-D140A
Control	5.73 ± 0.16	5.80 ± 0.24
+LPS	7.79 ± 0.16*	5.85 ± 0.38

Values are means ± SD for n = 4 for each condition.

^*P < 0.05 vs. control.

Fig. 3.

Genetic inactivation of acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) ameliorates sepsis-induced lung VCAM-1 expression, cytokine release, and oxidative stress. Wild-type (WT) and peroxiredoxin 6 (Prdx6)-D140A knockin (KI) mice were treated with 10 mg/kg LPS intraperitoneally. A: VCAM-1 expression measured by immunofluorescence of control (saline) and LPS-treated WT and Prdx6-D140A mice (n = 4). The monoclonal VCAM-1 antibody (catalog no. 32653) was purchased from Cell Signaling Technology (Danvers, MA). B: VCAM-1 expression measured by Western blot in whole lung homogenates of control (saline) and LPS-treated WT and Prdx6-D140A mice; the gels for WT (control and LPS) and mutant mice (control and LPS) were done at different times so that gel densities for the 2 types of mice should not be compared. The rabbit polyclonal VCAM-1 antibody (catalog no. sc-1504-R) was purchased from Santa Cruz Biotechnology. C–E: cytokine/chemokine expression in bronchoalveolar lavage fluid (BALF): TNF-α (C); IL-1 (D); and monocyte chemoattractant protein-1 (MCP-1; E); this cytokine was detected only in WT animals stimulated with LPS. F: staining of lung for 3-nitrotyrosine (green, n = 3) using an antibody generated by Dr. Harry Ischiropoulos (Children’s Hospital of Philadelphia) and previously described (17). In A and F, nuclei are counterstained in red and the scale bars = 20 µm. Blue circles and red squares represent individual animals per experiment. Error bars indicate the means ± SD. Group differences (saline vs. LPS) were evaluated by t-test, and the P value for comparison of the 2 groups is indicated; no P value is shown for those comparisons where P > 0.05.

D140A mutation abolishes aiPLA₂ activity without affecting Prdx6 expression or its peroxidase activity.

Measurements of Prdx6 expression and enzymatic activities in homogenates of whole lung confirmed the inactivation of aiPLA₂ activity by the D140A mutation of Prdx6 and demonstrated that Prdx6 expression levels and peroxidase activities (both H₂O₂ and a lipid hydroperoxide as substrate) were unaffected (Fig. 1, D–G). These results are consistent with our previous results that were obtained before removal of the neomycin-resistance cassette from the mutant mouse genome (36). Thus the mutation to Prdx6-D140A appears to have the absence of aiPLA₂ activity as its sole effect.

Genetic inactivation of aiPLA₂ prevents sepsis-induced lung inflammation and oxidative stress.

We used an animal model of sterile sepsis to study the role of aiPLA₂ activity in ALI associated with systemic inflammation. Control mice, both WT and mutant, gained ~3% of their initial body weight during the 24-h experimental period. WT mice lost 5.8 ± 0.2% of their body weight while the mutant mice lost 6.1 ± 0.1% (not shown); the differences between the two mouse groups is not statistically significant. On the other hand, there was a significant difference in survival between WT and mutant mice (Fig. 2A); at 48 h after LPS injection, WT mice showed ~60% mortality while all of the Prdx6-D140A mice were still alive. A significant difference in mortality was still present at 96 h of observation after LPS (Fig. 2A). Lungs from WT mice showed the well-documented ALI reaction when examined at 24 h after systemic administration of LPS. LPS treatment of WT mice resulted in an increased lung vascular permeability, as indicated by increased BALF protein (Fig. 2B) and increased lung wet-to-dry weight ratio (Table 1). Lung inflammation was indicated by an increased number of neutrophils (PMN) in the BALF (Fig. 2, C and D) and by increased staining with antibodies to MPO (Fig. 2E) and NIMP (Fig. 2F) in the lung tissue. All of these alterations after LPS in lungs from WT mice were absent or markedly diminished in the lungs of Prdx6-D140A mice (Fig. 2, B–F).

The increase in the number of lung PMN that was seen in WT mice at 24 h after treatment with LPS was associated with increased lung VCAM-1 expression as indicated by both immunofluorescence and Western blot analysis (Fig. 3, A and B). BALF content of three cytokines (TNF-α, IL-1β, and MCP-1) also was increased in WT after LPS (Fig. 3, C–E). Finally, LPS-induced oxidative stress in the WT lung was indicated by increased nitrotyrosine content in lung tissue (Fig. 3F). As for the altered lung permeability and inflammation, the increased VCAM, cytokines, and nitrotyrosine after LPS were significantly reduced in the Prdx6-D140A as compared with WT mice (Fig. 3). We postulate that these manifestations of ALI reflect the role of Prdx6 in the activation of Nox2 and subsequent ROS production in WT but not in D140-Prdx6 mutant lungs.

LPS induces Prdx6 phosphorylation and translocation to the plasma membrane.

Evaluation of lung epithelium/endothelium has shown that Prdx6 is primarily localized to the cytosol and lysosomal type organelles, functioning as a peroxidase at cytosolic pH and as a PLA₂ at acidic pH (11). Phosphorylation of the enzyme, however, changes its conformation thereby increasing its lipid-binding affinity; the result is its translocation to the plasma membrane, its increased PLA₂ activity at cytosolic pH, and its participation in Nox2 activation (6, 45, 58). To evaluate further the role of Prdx6 in ALI, we tested the effect of LPS treatment in isolated PMVECs in primary culture from WT mice on Prdx6 phosphorylation, its cellular localization, and its PLA₂ activity. Exposure of cells to LPS resulted in Prdx6 phosphorylation as shown by Western blot analyses using an antibody to phosphorylated Prdx6 (Fig. 4A) and an increase in aiPLA₂ activity by 260% (Fig. 4B); cells from Prdx6-D140A mice were not evaluated for phosphorylation since this mutation does not show aiPLA₂ activity (Fig. 1E). Treatment of WT cells with MJ33, a pharmacological inhibitor of this enzymatic activity, resulted in a marked decrease (~90%) in aiPLA₂ activity after LPS (Fig. 4B); aside from aiPLA₂, the only mammalian enzyme with known sensitivity to inhibition by MJ33 is pancreatic (secreted type I) PLA₂, which is not expressed in the lung (13). Phosphorylation resulted in a shift in localization of Prdx6 to the plasma membrane (Fig. 4C), consistent with previous observations (6). The intense nuclear staining with this antibody has been observed previously (6) and likely represents an artefact (28).These results indicate that the PLA₂ activity of Prdx6 can be significantly increased during inflammation, allowing it to function in Nox2 activation (53).

Fig. 4.

LPS induces peroxiredoxin 6 (Prdx6) phosphorylation and promotes its translocation to the plasma membrane in lung pulmonary microvascular endothelial cells (PMVECs). Wild-type (WT) PMVECs were treated with 1 μg/ml LPS for 8 h. A: representative Western blots using Ab to phosphorylated Prdx6 (P-Prdx6; top), Prdx6 (middle), and GAPDH (bottom); Prdx6 is phosphorylated after treatment of cells with LPS. Prdx6 and phosphorylated Prdx6 antibodies were generated by Covance Research Products (Denver, CO) and Proteintech Group (Chicago, IL), respectively, and have been described previously (6); GAPDH antibodies (cat no. 2118) were purchased from Cell Signaling Technology. B: acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) activity in control cells (PBS), in cells after treatment with LPS, and in cells that were pretreated with 10 μM MJ33 for 30 min before the addition of LPS. C: immunofluorescence using an anti-phosphorylated Prdx6 antibody (green) (58) and the membrane marker WGA-Alexa Fluor 594 (red). The scale bar = 20 µm. Note colocalization of the red and green markers at the cell membrane following LPS. The basis for the green nuclear staining is unknown. Error bars indicate the means ± SD for n = 3. Statistical significance was tested using ANOVA with Bonferroni post hoc test. Group differences (LPS vs. control or MJ33) were evaluated by ANOVA with Bonferroni post hoc test, and the P value for comparison of the groups is indicated.

Genetic inactivation of aiPLA₂ prevents LPS-induced oxidative stress and inflammation.

We have previously demonstrated that Prdx6 is needed for Nox2-driven oxidant generation in response to phorbol ester and angiotensin II in both intact lungs and PMVECs (6, 34, 35, 53). In the present study, we evaluated the effect of LPS on Nox2-driven oxidant generation and the role of aiPLA₂ in LPS-induced Nox2 activation in lungs that were isolated and perfused in a recirculating system at 24 h after LPS administration. In contrast to lungs from WT mice, lungs from Prdx6-D140A mice failed to generate intravascular H₂O₂ at 24 h after stimulation with LPS (Fig. 5A). Similarly, treatment of PMVECs with LPS for 8 h increased H₂O₂ generation in WT but not in Nox2 null or Prdx6-D140A cells (Fig. 5B). Furthermore, intracellular H₂O₂ generation increased after LPS treatment in WT PMVEC stably expressing the HyPer-Cyto sensor and this effect was abolished by pretreatment with MJ33 (Fig. 5C). Both lipid peroxidation and nitrotyrosine levels increased after LPS treatment in WT but not in Prdx6-D140A cells (Fig. 5, D and E). Overall, these data show that aiPLA₂ activity is needed for LPS-mediated Nox2 activation in the lung endothelium that leads to oxidant generation and lipid peroxidation.

Fig. 5.

Genetic inactivation of acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) prevents NADPH oxidase, type 2 (Nox2)-mediated oxidant generation and oxidative stress in intact lungs and endothelial cells. A: intravascular oxidant generation was measured using Amplex red in the recirculating perfusate of isolated lungs from wild-type (WT) and peroxiredoxin 6 (Prdx6)-D140A mice at 24 h after intraperitoneal administration of LPS or saline (control). The time axis indicates the time of in vitro lung perfusion. Numbers in parenthesis represent the slope of the change in Amplex red fluorescence vs time calculated by least mean squares; n = 4 for Prdx6-D140A LPS; for all other groups, n = 3. B: extracellular H₂O₂ generation measured by the change in Amplex red fluorescence in WT, Nox2 null and Prdx6-D140A PMVECs at 8 h after administration of LPS. Data are the %change vs. control (cells treated with PBS for 8 h); n = 4 for WT; n = 3 for Nox2 and Prdx6-D140A; statistical significance was tested using ANOVA with Bonferroni post hoc test; the individual P values are for comparison to WT. C: intracellular H₂O₂ generation in PMVECs stably expressing pHyPer-Cyto. Cells were stimulated with LPS with or without MJ33 pretreatment for 8 h. An increase in cellular H₂O₂ is indicated by increased fluorescence. n = 3–4. The y-axis in A and C is in arbitrary fluorescence units (a.u.). D: lipid peroxidation (green fluorescence) was visualized using Liperfluo in WT and Prdx6-D140A PMVECs stimulated with LPS for 16 h; n = 3. Data are reported as change from corresponding PBS control for each cell type (WT and Prdx6-D140A). E: nitrotyrosine staining (green) in WT and Prdx6-D140A PMVECs stimulated with LPS for 16 h. Nuclei (red) are counterstained with propidium iodide. The antibody to 3-nitrotyrosine was a kind gift of Dr. Harry Ischiropoulos (Children’s Hospital of Philadelphia) and has been previously described (17). D and E: the scale bar for the fluorescence images indicates 20 µm and the y-axis on the bar graphs indicates fold-change from corresponding PBS control in cell fluorescence. WT, n = 4; Prdx6-D140A, n = 3. Error bars indicate the means ± SD. Group differences (PBS vs LPS) were evaluated by t-test; no P value is shown for those comparisons where P > 0.05.

We further studied the effect of aiPLA₂ deficiency on LPS-mediated oxidative stress and inflammation in PMVECs. LPS treatment promoted the nuclear translocation of NF-κB and led to increased expression of both ICAM-1 and NLRP3 in WT but not in Prdx6-D140A cells (Fig. 6, A–C). Moreover, NLRP3:ASC colocalization was observed in WT but not in Prdx6-D140A cells using in situ proximity ligation assays (Fig. 6D). These results suggest that, aside from promoting vascular oxidative stress by a mechanism that involves Nox2, aiPLA₂ also participates in the inflammatory response by regulating ROS-mediated activation of NF-κB and the NLRP3 inflammasome.

Fig. 6.

Genetic inactivation of acidic Ca²⁺-independent phospholipase A2 (aiPLA₂) prevents LPS-induced inflammation in lung endothelial cells. A: immunofluorescence staining for NF-κB (red) in wild-type (WT) and peroxiredoxin 6 (Prdx6)-D140A pulmonary microvascular endothelial cells (PMVECs) stimulated with LPS for different times. Nuclei are counterstained with SYTOX green; images are representative of three biological replicates. B: ICAM-1 staining (green) in WT and Prdx6-D140A PMVECs stimulated with LPS for 8 h. Nuclei are counterstained with propidium iodide (red); n = 3. ICAM-1 antibodies (cat no. ab119871) were obtained from Abcam (Cambridge, MA). C: representative Western blot for nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3 (NLRP3) expression in WT and Prdx6-D140A PMVECs stimulated with LPS. WT, n = 6; Prdx6-D140A, n = 3. NLRP3 expression was not detected in Prdx6-D140A PMVECs even after stimulation with LPS. NLRP3 antibodies were purchased from Cell Signaling Technologies (cat no. 15101). D: in situ proximity ligation assays (Duolink) to evaluate the colocalization of the NLRP3 and ASC (BioLegend antibody no. 653902) inflammasome subunits. KI, knockin. Red fluorescence indicates proximity (<40 nm) of the 2 proteins. Gray indicates nuclear staining. Images are representative of at least 3 biological replicates. Quantitative analysis was performed using Fiji software. For each biological sample, ten pictures were evaluated and counted. NLRP3:ASC interactions were normalized to nuclei. The scale bars in A, B, and D indicate 25 µm. Error bars indicate the means ± SD. Group differences (PBS vs. LPS) were evaluated by t-test; no p\P value is shown for those comparisons where P > 0.05.

DISCUSSION

Increased oxidative stress and inflammation are hallmarks of sepsis-induced ALI (23). Excessive generation of oxidants in the lung during sepsis causes dysregulated Ca²⁺ signaling, endothelial cell activation, increased expression of vascular adhesion molecules, increased release of inflammatory cytokines, and increased neutrophil infiltration; these effects of oxidative stress ultimately result in increased vascular permeability, lung edema, and endothelial cell death. Several recent studies have shown that the Nox2 enzyme is critical for the generation of oxidants and the progression of ALI in LPS as well as polymicrobial models of sepsis (5, 19, 20, 29, 39). Furthermore, Nox2-deficient lung endothelial cells as compared with WT cells show increased survival after exposure to LPS or cecal ligation puncture-induced sepsis (19, 20).

Our previous study using intra-tracheal LPS as a model of acute ALI indicated an important role for aiPLA₂ in the activation of Nox2 and subsequent oxidant stress; this latter study demonstrated that treatment of mice with the aiPLA₂ inhibitor MJ33 ameliorated LPS-induced lung injury (34). In the present study, we have expanded our studies to a model of sepsis (systemic LPS administration) and shown that, as with the model of ALI, aiPLA₂ regulates activation of Nox2 and that decreased aiPLA₂ activity improves the outcome. In the present study, the role of aiPLA₂ is confirmed using a specific mutation to ablate this activity. These mutant mice had the absence of aiPLA₂ activity as the sole observable difference from WT C57Bl/6 mice. Of course, like all studies of this type, subtle changes that are compensatory to the genetic alteration may have occurred although they have not been observed. Furthermore, these knockin mice would be expected to show increased susceptibility to infection due to the altered Nox2 response, but this has not been studied. The animals used in the present study showed no evidence of lung infection and were housed under conditions that are expected to minimize that possibility.

Using phorbol ester or angiotensin II as agonists, we have previously characterized the mechanism by which aiPLA₂ activity can modulate Nox2 activation (6, 53). While cytosolic Prdx6 is essentially inactive as a PLA₂, its phosphorylation by mitogen-activated protein kinases at amino acid T177 induces a conformational change in the protein that results in increased binding affinity for phospholipids and increased aiPLA₂ activity (31, 45, 58). Similarly to phorbol esters and angiotensin II, LPS can activate a mitogen-activated protein kinase (p38) (24, 25) that leads to Prdx6 phosphorylation and translocation to cell membranes, as confirmed in Fig. 4. Cell membrane phospholipids represent an aiPLA₂ substrate to generate lysoPC as required for Nox2 activation (6, 53). Since Prdx6-D140A lacks aiPLA₂ activity (Fig. 1E) (16, 36), we did not evaluate translocation of mutant Prdx6 to the plasma membrane. We have previously shown that translocation of Prdx6 occurs in cells treated with the aiPLA₂ inhibitor MJ33, but activation of Nox2 does not occur (6). Thus aiPLA₂ activity, but not its translocation per se, is required for Nox2 activation.

Based on our previous studies documenting the pathway from aiPLA₂ to activated Nox2 (53), we attribute the failure of Nox2 activation in the Prdx6-D140A mice to an absence of activated Rac, a key cytosolic component required for Nox2 activation. Although Rac participates in a variety of other cellular process (47), the lack of significant tissue abnormalities in the Prdx6-D140A knockin, as well as in Prdx6 null mice, suggests either that Rac can be activated through other non-Prdx6 mechanisms or that other pathways can compensate for the cellular effects of Rac inhibition.

In addition to a failure of Nox2 activation, the Prdx6-D140A knockin mice might be expected to show alterations in several other parameters. First, aiPLA₂ activity has a role in the degradation and remodeling of lung surfactant phospholipids; this process takes place within the acidic lysosomal-related organelles called lamellar bodies (13). However, this is a long-term effect that leads to altered surfactant lipid content in the lung but would likely have a limited role in ALI. A second function of aiPLA₂ is its participation, along with the peroxidase activity of Prdx6, in the repair of peroxidized membrane phospholipids; however, repair of membranes by Prdx6 is complete, but delayed, in the absence of aiPLA₂ activity (10, 16). Finally, Prdx6 may play a role in the scavenging of intracellular ROS, but this function requires the peroxidase activity that is independent of the PLA₂ activity. Thus the role of Prdx6 in NOX2 activation appears to be the major effect following LPS. Although other physiological roles for aiPLA₂ may yet be discovered, these would likely have lesser effects on the course of LPS-induced ALI as compared with the considerable role for Nox2 activation.

Other PLA₂ enzymes such as cytosolic PLA₂ and type II secretory PLA₂ have previously been implicated in the progression of sepsis-induced ALI (9, 41, 49, 51). Although their precise roles in ALI remain elusive, their effects derive mainly from the liberation of arachidonic acid for the generation of inflammatory mediators (4, 42, 54). Of note, aiPLA₂, in contrast to these other PLA₂ enzymes, does not show a substrate preference for arachidonic acid-containing phospholipids (1, 32).

The NLRP3 inflammasome is a critical component of innate immunity and a contributor to the pathology of several pulmonary diseases including ALI (8, 22). Assembly of NLRP3 inflammasomes stimulates the proteolytic conversion of procaspase-1 into active caspase-1, which promotes the maturation of pro-IL-1β into active IL-1β (26). With the use of intact lungs and PMVECs derived from Prdx6-D140A mutant mice, this study shows that aiPLA₂ is required for LPS-driven Nox2 activation and that LPS treatment increases NLRP3 expression and promotes inflammasome assembly in WT but not in Prdx6-D140A PMVECs. Although there is a consensus that oxidants can modulate NLRP3 inflammasome assembly, the role of Nox2 in NLRP3 activation is controversial, and it appears to be cell specific. While peripheral blood mononuclear cells and neutrophils isolated from chronic granulomatous disease patients and bone marrow-derived macrophages isolated from Nox2-deficient mice show normal NLRP3 activation (18, 27, 52), lung endothelial cells isolated from Nox2-deficient mice show impaired NLRP3 activation (59). Our results show that aiPLA₂ activity is needed for NLRP3 inflammasome assembly in endothelial cells and support the idea that activation of endothelial cells by bacterial endotoxin plays an important role in the progression of ALI.

Our results also show that LPS stimulation induces nuclear translocation of NF-κB in WT but not in Prdx6-D140A PMVECs. Those events are associated with increased ICAM-1 and NLRP3 expression. TLR4 signaling is known to “prime” the inflammasome by increasing NLRP3 and pro-IL-1β transcription by a mechanism that involves NF-κB (26). In human lung endothelial cells, inhibition or silencing of Nox2 but not Nox1 or Nox4 blunts LPS-driven NF-κB activation (38). Moreover, Nox2-deficient mice do not show increased lung ICAM-1 or IL-1β expression after intraperitoneal LPS administration (39). Thus it appears that the main role of aiPLA₂ in ALI during sepsis is through its effects on Nox2 activation.

In summary, here we show that genetic inactivation of aiPLA₂ activity (by mutation of the D140 amino acid of Prdx6) ameliorates ALI by preventing lung inflammation and oxidative stress in a sterile model of sepsis (summarized in Fig. 7). As shown in this scheme, LPS treatment results in the phosphorylation of cytosolic Prdx6, its binding to the plasma membrane, and aiPLA₂ activity that results in activation of endothelial Nox2 and oxidant generation. These oxidants modulate the expression of several proinflammatory molecules in lung endothelial cells following LPS (e.g., NF-κB and NLRP3 inflammasomes). These present results suggest that aiPLA₂ could be a potential therapeutic target for the prevention or treatment of sepsis-induced ALI.

Fig. 7.

Schematic for the effect of the D140 mutation of peroxiredoxin 6 (Prdx6-D140A) on the lung response to intraperitoneal LPS. LPS through interaction with Toll-like receptor 4 (TLR4) leads to MAP kinase activation with subsequent phosphorylation of Prdx6 and its translocation to the plasma membrane. An increased Prdx6-PLA₂ activity leads to NADPH oxidase, type 2 (Nox2) activation and generation of oxidants, resulting in tissue oxidative stress and lung injury. Oxidative stress also results in NF-κB translocation to the cell nucleus, which promotes the expression of various inflammatory agents, recruitment of PMNs to the lung, and amplification of oxidant-induced lung injury. Mutation of D140 to A140 in Prdx6 abolishes its PLA₂ activity and prevents the tissue-damaging cascade associated with Nox2 activation. aiPLA₂, acidic Ca²⁺-independent phospholipase A2.

GRANTS

This research was funded by National Heart, Lung, and Blood Institute (NHLBI) Grant HL-R01-105509 (to A. B. Fisher). J. P. Vázquez-Medina was partially supported by NHLBI Grant F32-HL-127972 and University of California (UC), Berkeley startup funds. Confocal imaging for VCAM-1 experiments was conducted at the UC Berkeley CRL Molecular Imaging Center, supported by National Science Foundation Grant DBI-1041078.

DISCLOSURES

S. I. Feinstein and A. B. Fisher have a patent application pending for a peptide inhibitor of peroxiredoxin 6 PLA₂ activity and have part ownership in a start-up company based on use of the peptide inhibitor. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

S.C. and A.B.F. conceived and designed research; J.P.V.-M., J.-Q.T., P.P., R.B.-F., C.D., E.M.S., and S.C. performed experiments; J.P.V.-M., R.B.-F., S.I.F., S.C., and A.B.F. analyzed data; J.P.V.-M., S.I.F., and A.B.F. interpreted results of experiments; J.P.V.-M., C.D., S.C., and A.B.F. prepared figures; J.P.V.-M. drafted manuscript; A.B.F. edited and revised manuscript; J.P.V.-M., J.-Q.T., P.P., R.B.-F., C.D., E.M.S., S.I.F., S.C., and A.B.F. approved final version of manuscript.