Endothelial ENaC-α Restrains Oxidative Stress in Lung Capillaries in Murine Pneumococcal Pneumonia–associated Acute Lung Injury

Abstract

Infection of lung endothelial cells with pneumococci activates the superoxide-generating enzyme NOX2 (nicotinamide adenine dinucleotide phosphate hydrogen [NADPH] oxidase 2), involving the pneumococcal virulence factor PLY (pneumolysin). Excessive NOX2 activity disturbs capillary barriers, but its global inhibition can impair bactericidal phagocyte activity during pneumococcal pneumonia. Depletion of the α subunit of ENaC (epithelial sodium channel) in pulmonary endothelial cells increases expression and PMA-induced activity of NOX2. Direct ENaC activation by TIP peptide improves capillary barrier function—measured by electrical cell substrate impedance sensing in endothelial monolayers and by Evans blue dye incorporation in mouse lungs—after infection with pneumococci. PLY-induced hyperpermeability in human lung microvascular endothelial cell monolayers is abrogated by both NOX2 inhibitor gp91dstat and TIP peptide. Endothelial NOX2 expression is assessed by increased surface membrane presence of phosphorylated p47^phox subunit (Western blotting) in vitro and by colocalization of CD31 and gp91^phox in mouse lung slices using DuoLink, whereas NOX2-generated superoxide is measured by chemiluminescence. TIP peptide blunts PMA-induced NOX2 activity in cells expressing ENaC-α, but not in neutrophils, which lack ENaC. Conditional endothelial ENaC-α knockout (enENaC-α knockout) mice develop increased capillary leak upon intratracheal instillation with PLY or pneumococci, compared with wild-type animals. TIP peptide diminishes capillary leak in Streptococcus pneumoniae–infected wild-type mice, without significantly increasing lung bacterial load. Lung slices from S. pneumoniae–infected enENaC-α knockout mice have significantly increased endothelial NOX2 expression, compared with infected cyclization recombination mice. In conclusion, enENaC may represent a novel therapeutic target to reduce NOX2-mediated oxidative stress and capillary leak in acute respiratory distress syndrome, without impairing host defense.

Clinical Relevance

The research presented in this manuscript demonstrates a novel mechanism by which the epithelial sodium channel can repress oxidative stress in mouse lung capillaries during bacterial pneumonia. Because the same mechanism seems to apply to human microvascular endothelial cells, these data can foster the development of novel therapeutic agents towards pneumonia-associated acute respiratory distress syndrome.

Pneumococcal pneumonia is the main cause of severe pneumonia in infants and one of the primary causes of community-acquired pneumonia in the elderly (1, 2). Severe pneumococcal pneumonia can cause acute respiratory distress syndrome (ARDS), with mortality as high as 40%. Despite recent advances in ARDS treatment, severe ARDS is refractory to conventional therapy, with no successful pharmacologic treatments to date (3, 4). A potentially lethal complication of ARDS is pulmonary permeability edema, characterized by hyperpermeability of the capillary barrier, which is at least partially caused by oxidative stress. Streptococcus pneumoniae (Sp) significantly increases O₂⁻ generation in infected mammalian cells and, depending on the cell type, both the pneumococcal pore-forming toxin PLY (pneumolysin) (5) and the autolysin LytA (6) are important mediators. A major component of the oxidative stress response is the superoxide- (O₂⁻) generating enzyme NOX2 (nicotinamide adenine dinucleotide phosphate hydrogen [NADPH] oxidase 2) (7, 8). Increased levels of NOX2-derived peptide (sNOX2-dp), a marker of NOX2 activation, were found in plasma of patients with pneumococcal pneumonia (9). Enhanced NOX2 activity is deleterious to lung capillary endothelium and alveolar epithelium, because the activity can impair alveolar–capillary barrier function (10). In contrast to NOX2’s deleterious role in endothelium and alveolar epithelium upon excessive activation, NOX2 expression in polymorphonuclear neutrophils (PMNs) is a hallmark of their bactericidal activity in pneumococcal pneumonia (5, 11). Thus, NOX2 plays a dual role in pneumococcal pneumonia, exhibiting a deleterious effect on the alveolar–capillary compartments when hyperactivated, but with a beneficial role in bactericidal activity in phagocytes. This dual nature of NOX2 may provide an explanation why global inhibitors of the enzyme have not been evaluated in clinical trials, because of concerns of heightened susceptibility to harmful infections (12). As such, a strategy that reduces deleterious NOX2 activity in ARDS target cells, without significantly interfering with the enzyme’s activity in phagocytes, could be therapeutically significant in pneumonia, because it could overcome lung barrier dysfunction while preserving host defense.

The amiloride-sensitive ENaC (epithelial sodium channel) in its conventional native conformation consists of the three subunits α, β, and γ, although a fourth δ subunit exists, which can functionally replace α to form an alternative channel (13–15). The activity of the channel is the product of its surface membrane expression (N) and its open probability (Po). The α and γ subunits of the channel have to be proteolytically cleaved by furin for maturation of the channel (16). ENaC has a conductance of ∼5 pS and is highly selective for Na⁺ cations, facilitating their transport across the apical cell membrane from the lumen into the cytoplasm (16, 17). Absorbed Na⁺ is subsequently transported out of the cell into the interstitial space by basolateral Na⁺/K⁺ ATPase (18). The importance of the α subunit of ENaC is demonstrated by the inability of global ENaC-α knockout (ENaC-α KO) mice to remove lung fluid at birth, leading to respiratory failure and death (19). We have developed a 17–amino-acid peptide (AA sequence CGQRETPEGAEAKPWYC)—the TIP peptide (a.k.a. solnatide, AP301)—which is a mimic of the lectin-like domain of human TNF (20), that increases both the open probability of ENaC and surface expression of its α subunit through a direct binding interaction with its carboxyterminal domain (21–24).

Recently, we and others have shown a novel role for ENaC-α expressed in lung capillaries as a regulator of barrier function in the presence of bacterial toxins, including PLY (25) and LPS (26). In murine models of PLY-induced lung injury, intratracheal TIP peptide instillation improved both alveolar fluid clearance and alveolar–capillary barrier function (25, 27). This alveolar fluid clearance-activating activity of the TIP peptide seems to work across species; in a phase 2 clinical trial, when inhaled through the ventilator, it significantly reduced extravascular lung water in patients with ARDS with a sequential organ failure assessment score ⩾11, accompanied by a reduction in peak ventilation pressure, which in turn has the potential to reduce or prevent ventilator-induced lung injury (28).

In this study, we show that depletion of endothelial ENaC-α (enENaC-α) increases PMA-induced NOX2 activity in lung capillary endothelial cells in vitro and augments endothelial gp91^phox expression in mice infected with Sp. In view of the deleterious role of NOX2 oxidase in capillary leak, we investigated whether the direct activation of ENaC-α by TIP peptide can reduce NOX2 activity and preserve barrier function in capillary endothelial cells in murine pneumococcal pneumonia without impairing the enzyme’s activity in PMNs.

Methods

Mice

We used 8- to 10-week-old male wild-type (WT) C57BL6 mice (Jackson Labs) in the high-dose Sp and PLY instillation experiments. To selectively knock out ENaC-α in vascular endothelial cells, we cross-bred conditional Scnn1a^lox/lox mice (29) (breeding pairs of which were obtained from the European Mouse Mutant Archive) with tamoxifen-inducible vascular endothelial (VE)-cadherin–Cre/ert2 mice on a C57BL6 background (30). Mouse genotyping was done using the Kapa mouse genotyping kit, using the appropriate primers, with the floxed allele at 280 bp and the WT allele at 220 bp (see Figure E1 in the data supplement). Mice were injected intraperitoneally for 5 consecutive days with tamoxifen dissolved in corn oil and were used 2 weeks after the last injection. VE-cadherin–Cre/ert2 mice were injected for 5 consecutive days with vehicle (corn oil) and were used 2 weeks after the last injection as controls. Male 8- to 12-week-old tamoxifen-inducible enENaC-α KO mice (enENaC-α KO) were used in studies assessing low-dose Sp infection and low-dose PLY instillation. VE-cadherin–cre/ert2 cyclization recombination (CRE) driver mice were used as controls in the studies evaluating the enENaC-α KO mice. For the co-immunoprecipitation (co-IP) studies in lung homogenates, we used 8- to 10-week-old female C57BL6 mice (Jackson Labs). All animal studies conformed to National Institutes of Health (NIH) guidelines. The experimental procedures were approved by Augusta University (AU) Institutional Animal Care and Use Committee.

Experimental details are provided in the data supplement.

Results

Human Lung Microvascular Endothelial Cells Express Functional ENaC Channels, Which Are Activated by TIP Peptide and Inhibited by PLY

Human lung microvascular endothelial cells (HL-MVECs) have cation channels with a conductance of 5 pS, indicative of functional ENaC channels (Figures 1A and 1B). The TNF-derived TIP peptide (50 μg/ml) in the cell bath significantly increases ENaC Po within minutes. The increase in ENaC Po is observed each time TIP peptide is applied (Figures 1C and 1D). In contrast, the pneumococcal pore-forming toxin PLY (30 ng/ml) reduces ENaC activity within minutes in HL-MVECs (Figures 1E and 1F). We also investigated whether PLY can interfere with the surface expression of ENaC in HL-MVECs, the latter of which is reduced following ubiquitination and degradation of the subunits. Because HDAC7-mediated deacetylation of lysine residues within ENaC-α was shown to increase the subunit’s ubiquitination and degradation (31), we investigated whether PLY can reduce ENaC-α acetylation and, if so, whether TIP peptide can prevent this. Our results from a qualitative co-IP study in Figures 1G and 1H show that PLY (30 ng/ml) within 30 minutes reduces acetylation of ENaC-α, which is mitigated by TIP peptide (50 μg/ml). As such, our results indicate that PLY is able to negatively affect both expression and activity of ENaC in HL-MVEC.

Figure 1.

Human lung microvascular endothelial cell (HL-MVEC) ENaC (epithelial sodium channel) expression and open probability (Po) is activated by TIP peptide and inhibited by PLY (pneumolysin). (A) Typical record showing many low conductance (5 pS) channels (ENaC, a.k.a. highly selective channel) in HL-MVECs. (B) Current voltage (I-V) plot of ENaC/highly selective channel in HL-MVECs fitted to the Goldman equation from SigmaPlot 14. (C) ENaC response in HL-MVECs to two consecutive applications of TIP peptide (5 pS currents). (D) TIP peptide increases ENaC Po in HL-MVECs (P < 0.001 vs. untreated; n = 30). (E) ENaC current trace in HL-MVECs demonstrating that the application of PLY (30 ng/ml) potently inhibits ENaC activity within minutes. (F) PLY inhibits ENaC Po within 10 minutes in HL-MVECs (n = 5; P < 0.01 vs. untreated). (G) TIP peptide partially preserves ENaC-α acetylation in PLY-treated HL-MVECs. Confluent monolayers of HL-MVECs (one 100-mm dish per condition) were treated for 30 minutes with vehicle (CTRL), PLY (30 ng/ml), or PLY in the presence of TIP peptide (50 μg/ml). After 30 minutes, acetylated lysine residues within ENaC-α were assessed upon co-IP with anti–ENaC-α antibody (Ab) (Novus) and anti-acetylated lysine Ab (Cell Signaling). (H) Relative expression of acetylated ENaC-α in vehicle-, PLY-, or (TIP + PLY)-treated HL-MVECs. Ctrl = control; co-IP = co-immunoprecipitation (co-IP).

NOX2 Mediates PLY-induced Barrier Dysfunction in HL-MVEC Monolayers

Sp infection was proposed to increase reactive oxygen species (ROS) generation in pulmonary microvascular endothelium (32, 33). The pneumococcal pore-forming toxin PLY, a virulence factor that induces lung injury (22, 27, 34–36), has been proposed as an activator of NADPH oxidases (NOX), mainly NOX2 (5). As shown in Figure 2A, hyperpermeability induced by PLY (15 ng/ml) in HL-MVEC monolayers is completely inhibited by a specific NOX2 inhibitor, gp91dstat (10 μM), when given 30 minutes before PLY. The same gp91dstat pretreatment partially inhibits the effect of a higher PLY dose (60 ng/ml; Figure 2B). A 30-minute post-treatment with gp91dstat had no effect at this concentration (Figure 2B). We have previously shown that TIP peptide, which binds to ENaC-α, protects HL-MVEC monolayers from PLY-induced barrier dysfunction (27), and a specific siRNA-mediated depletion of ENaC-α in HL-MVECs makes these cells more susceptible to PLY (25). We have also shown that pharmacological inhibition of PKC-α reduces PLY-induced hyperpermeability in HL-MVEC monolayers (27). PKC-α is activated by PLY-induced Ca²⁺ influx and can phosphorylate cytosolic NOX2 subunits, leading to their recruitment to the surface membrane and NOX2 oxidase activation (7, 32). Although these findings do not exclude a role for other ROS-generating enzymes expressed in endothelium, like NOX1 or NOX4, they nevertheless show the relevance of NOX2 as an important mediator of PLY-induced barrier dysfunction, as well as the potential of enENaC activation to counteract PLY-induced barrier dysfunction. Recently, NOX activation was shown to subsequently activate mitochondrial ROS generation in endothelium (37). In agreement with a previous report from our group (33), we demonstrate in Figures 3A and 3B that PLY (30 ng/ml) within 15 minutes significantly increases mitochondrial ROS generation in HL-MVECs, assessed with MitoSox Red, and that a 30-minute pretreatment with TIP peptide (50 μg/ml) partially, but significantly, blunts this.

Figure 2.

NOX2 (nicotinamide adenine dinucleotide phosphate hydrogen [NADPH] oxidase 2) inhibition blunts PLY-induced hyperpermeability in HL-MVEC monolayers. HL-MVEC grown to confluence on ECIS arrays (ECIS 1600R, Applied Biophysics; initial resistance between 1,800 and 2,000 Ohms) were treated with (A) 15 ng/ml PLY (low concentration), or (B) 60 ng/ml PLY (high concentration) for 8 hours, in the presence or absence of vehicle or the NOX2 inhibitor peptide gp91dstat (10 μM, applied 30 min before, or, in the case of the higher concentration, 30 min after PLY) (n = 4 per group; mean ± SD; *P < 0.05 between PLY +  gp91dstat vs. PLY for the time period indicated by the bold line).

Figure 3.

Assessment of PLY-induced mitochondrial reactive oxygen species (ROS) production. (A) Representative confocal images of MitoSox Red staining in HL-MVECs treated for 15 minutes with PLY (30 ng/ml), upon a 30-minute pretreatment or not with TIP peptide (50 μg/ml). Scale bars, 100 μm. (B) Data points in the graph indicate the intensity of red fluorescence staining with MitoSOX normalized to the number of DAPI-positive nuclei (n = 3 wells for each treatment). ImageJ was used for image analysis and data graphed using GraphPad Prism. Mean and error bars indicating the SD are included in the dot blot image. Statistical analysis was performed using t test. *P < 0.0004 versus Ctrl. **P < 0.02 versus PLY.

enENaC-α Is a Novel Repressor of NOX2 Expression

Activated NOX2 consists of two membrane proteins (p22^phox and gp91^phox) and of four cytosolic proteins (p47^phox, p67^phox, and p40^phox, together with the small GTPase Rac2 [7]). NOX2 activation occurs upon phosphorylation of its cytosolic subunits and their subsequent migration to the plasma membrane; PKC is an important mediator of this process. Although a basal level of NOX2 oxidase activity has been proposed to be necessary for proper ENaC activity (38), it is unknown whether enENaC activation can modulate excessive NOX2 activity. With that in mind, we generated tamoxifen-inducible enENaC-α KO mice by cross-breeding conditional Scnna^lox/lox mice (29) (European Mouse Mutant Archive) with tamoxifen-inducible VE-cadherin–CRE/ert2 driver mice (30). As shown in Figures 4A and 4B, lung endothelial cells isolated from enENaC-α KO mice express significantly higher protein levels of gp91^phox (NOX2) than cells from control CRE driver mice (***P < 0.01, compared with CRE control group; n = 4–5 in each group). This finding indicates that ENaC-α can regulate NOX2 expression in lung endothelial cells. To investigate whether this also occurs in HL-MVECs, we transfected these cells with ENaC-α–specific siRNA and compared them to cells transfected with scrambled siRNA (Figure 4C). As shown in Figure 4D, ENaC-α depletion significantly increases gp91^phox protein expression at 72 hours in HL-MVECs. NOX2 has been shown to coimmunoprecipitate with ENaC-α in cell lysates (38), suggesting the possibility of a direct NOX2–ENaC-α interaction at the plasma membrane. Such an interaction could potentially interfere with subunit assembly and successful formation of the functional NOX2 complex. Our data in Figure 5A show that ENaC-α can associate with gp91^phox (NOX2) in lung homogenates from 8- to 10-week-old female C57BL6 mice. The co-IP efficacy was analyzed in Figure 5B. Taken together, these results indicate that enENaC-α represents a novel repressor of NOX2 expression in both mouse and human lung endothelial cells. Whether, apart from expression, NOX2 activity is also affected by ENaC-α is unknown.

Figure 4.

Absence of ENaC-α increases NOX2 expression in mouse and human lung endothelial cells. (A) Left: Representative Western blot (10% agarose gel) of basal expression of ENaC-α in mouse lung endothelial cells isolated from control cyclization recombination (CRE) mice (CRE) or EC-specific ENaC-α knockout (EC ENaC-α KO) mice. Right: ENaC-α to β-actin expression ratio is blunted in endothelial cells from ENaC-α KO mouse, compared with endothelial cells from CRE mice (***P < 0.01, compared with CRE group; n = 4–5 in each group). (B) Left: Representative Western blot (10% agarose gel) of basal expression of NOX2 in isolated lung endothelial cells from control CRE mice (CRE) or EC-specific ENaC-α knockout (EC ENaC-α KO) mice. Right: Ratio of NOX2 over β -actin expression is markedly increased in endothelial cells from the ENaC-α KO group, compared with those from the CRE group (***P < 0.01, compared with the CRE control group; n = 4–5 in each group). (C) Left: Representative Western blot of basal expression of ENaC-α in HL-MVECs. Right: ENaC-α depletion using siRNA ENaC-α (72 h, StressMarq rabbit polyclonal antibody) (***P < 0.001, compared with CTL group; n = 4 in each group). (D) Left: Representative western blotting (10% agarose gel) of basal expression of NOX2 in HL-MVECs. Cells were transfected at 70–80% confluence with transfection reagent (Lipofectamine RNAiMAX, Invitrogen) with siRNA control (CTL) or siRNA ENaC-α. Right: Depletion of ENaC-α markedly increases NOX2 expression on HL-MVECs at 72 hours after transfection (**P < 0.01, compared with CTL group; n = 4 in each group).

Figure 5.

Co-IP of ENaC-α and NOX2 in lung lysate (3 h and overnight precipitation with Protein A beads). (A) Total lung tissue (8- to 10-week-old female C57Bl6 mice). ENaC-α pAb-treated lysate (10 μg; Stressmarq SPC-403) was precipitated with Invitrogen Ag Dynabeads. Lane 1: 30 μg lung lysate incubated with antibody overnight before precipitation with Ag beads; lane 2: 30 μg lung lysate incubated with antibody overnight before precipitation with Protein A beads; lane 3: supernatant from #1 after binding to beads (unbound target protein); lane 4: supernatant from #2 after binding to beads (unbound target protein); lane 5: whole lysate used for #1; lane 6: whole lysate used for #2. Upper panel: blot was detected with ENaC-α rabbit polyclonal Ab (Stressmarq) to determine the efficiency of the immunoprecipitation. Lower panel: blot with anti-NOX2 (gp91^phox Ab) to determine ENaC–α-NOX2 coprecipitation. (B) Efficacy of immunoprecipitation. The density of the ENaC-α and NOX2 bands in the immunoprecipitate (lanes 1 and 2) and the supernatant (lanes 3 and 4) was quantified for four blots like those shown in A, using the Fiji variant of ImageJ. The ratios of density of the immunoprecipitate to supernatant are shown in the graph (mean ± SD; n = 4). The efficacy of the ENaC-α antibody is good, but, as expected, the coimmunoprecipitation of NOX2 is less efficient but still recovers as much from the lysate as is left in the supernatant. This could also be interpreted to mean that not every ENaC-α has a NOX2 associated with it.

TIP Peptide Reduces NOX2 Complex Activity in Cells Expressing ENaC-α

Increased gp91^phox subunit expression in lung endothelium due to the absence of ENaC-α does not automatically imply higher NOX2 activity, because that requires phosphorylation and recruitment of cytosolic subunits to the surface membrane (7). As demonstrated in Figure 6A, using L-012 chemiluminescence (39), mouse lung endothelial cells isolated from male 8-week-old enENaC-α KO mice have a significantly higher gp91^phox expression than control lung endothelial cells isolated from CRE driver mice but do not generate more ROS under basal conditions. However, upon treatment with the gold standard NOX2 activator PMA, ENaC-α KO lung endothelial cells make significantly more ROS than control cells. This increased ROS generation is completely inhibited by superoxide dismutase (SOD), indicating it is mainly composed of O₂⁻ (n = 8 per group; P < 0.01 for all time points between 20 and 70 min after start of experiment). As shown in Figure 6B of membrane and organelle fractions of HL-MVECs (Cell Signaling fractionation kit) (40), TIP peptide, which activates ENaC, blunts PMA-induced (1 μM, 15 min) phosphorylation (at residue S359) and promotes surface membrane expression of the p47^phox subunit, normalized to the membrane and organelle marker protein AIF (Apoptosis Inducing Factor) (Cell Signaling fractionation kit) (40). As shown in Figure 6C, direct ENaC activation with TIP peptide (50 μg/ml) also blunts PMA-induced ROS generation in PMA-stimulated ENaC-α–expressing Cos p22^phox cells, (inset, Figure 6), which express NOX2 and its organizer and activator subunits, but no other NOX isoforms (41). As shown in Figure 6D, TIP peptide does not reduce, but rather activates PMA-induced ROS generation in primary bone marrow–derived mouse PMNs, which do not express ENaC-α (inset, Figure 6). PMA-induced ROS generation in THP-1 macrophages, which is blocked by SOD, is not significantly affected by TIP peptide (50 μg/ml) or by the indirect ENaC activator aldosterone (1 μM) (Figure 6E). Together, these data show that after its direct activation by TIP peptide, ENaC serves as a potent repressor of NOX2 activity in capillary endothelial cells. By contrast, TIP peptide does not significantly blunt ROS generation in PMA-stimulated PMNs but may increase it in an ENaC-independent manner.

Figure 6.

TIP peptide represses NOX2-mediated ROS generation in cells expressing ENaC-α. (A) Increased ROS generation in PMA-treated lung endothelial cells from endothelial ENaC-α KO mice (KO), compared with endothelial cells from wild-type (WT) control mice. ROS generation was blunted by addition of superoxide dismutase (SOD) (n = 8 per group; P < 0.01 for all time points between 20 and 70 min after start of experiment). (B) Upper panel: Representative western blotting of HL-MVEC membrane/organelle fractions demonstrating that PMA (1 μM) within 15 minutes induces p47^phox phosphorylation at Ser359 in HL-MVECs, and this leads to recruitment of the subunit from the cytosol to the surface membrane. Lower panel: Densitometric analysis calibrating protein levels to AIF (Apoptosis Inducing Factor), a protein marker expressed exclusively in membranes/organelle fractions, demonstrating that TIP peptide significantly blunts p47 phosphorylation in the presence of PMA (n = 3; ***P < 0.00001 vs. vehicle ctrl; **P < 0.0001 vs. PMA; PMA + TIP: not significant [NS] vs. vehicle). (C) PMA significantly stimulates ROS generation in Cos p22^phox cells, which express only NOX2 subunits, and the positive control DPI (an inhibitor of NADPH oxidase) blocks this completely. ENaC-α activation with TIP peptide (−30 min) significantly blunts PMA-induced ROS generation at 60 minutes in these cells (*P < 0.05 between PMA and PMA + TIP; n = 6 per group; PMA: 1 μM; DPI: 5 μM; TIP peptide 50 μg/ml). Cos p22^phox cells express ENaC-α (inset). (D) TIP peptide does not inhibit PMA-induced ROS generation in mouse PMNs, which do not express ENaC-α (inset). PMNs are pretreated for 20 minutes with TIP peptide (50 μg/ml) or with vehicle, before addition of PMA (50 nM). PMA significantly increases ROS generation, and TIP peptide does not inhibit this (n = 8 per group). Although TIP peptide did not decrease PMA-induced ROS generation in PMNs, it significantly increased it after PMA stimulation for the time points between 14 and 80 minutes (P < 0.05). (E) Neither TIP peptide (50 μg/ml) nor the indirect ENaC activator aldosterone (1 μM) is able to inhibit PMA (1 μM)-induced ROS generation in THP-1 macrophages at 180 minutes, whereas SOD (150 U/ml) does (n = 6 per group; ****P < 0.001 vs. vehicle or, in case of SOD, vs. PMA).

ENaC-α Protects from PLY- and Sp-induced Capillary Barrier Dysfunction in Mice

As shown in Figure 7A, tamoxifen-inducible enENaC-α KO mice develop significantly more capillary leak than control CRE driver mice 4 hours after intratracheal instillation of a moderate dose of PLY (1.5 μg/kg) evaluated by Evans blue dye incorporation in lung tissue (Evans blue dye injected in jugular vein 1 h before killing mice). Intratracheal instillation with 10⁷ CFU D39 Sp in 8- to 10-week-old male C57BL6 WT mice causes a significant increase in capillary leak 24 hours after infection, but coinstillation of 2.5 mg/kg of TIP peptide significantly blunts hyperpermeability (Figure 7B). As shown in Table 1, intratracheal instillation of TIP peptide (2.5 mg/kg) together with 10⁷ CFU Sp D39/mouse does not significantly increase lung bacterial load compared with infected vehicle-treated mice 24 hours after infection, as assessed after plating several dilutions of lung homogenates for 2 days on blood agar plates, although a trend toward an increase was observed in the peptide-treated group. Tamoxifen-inducible enENaC-α KO mice develop significantly more capillary leak than control CRE mice upon 24 hours of infection with 2 × 10⁶ CFU Sp per mouse (Figure 7C). TIP peptide, however, does not decrease the capillary leak in mice lacking enENaC-α (Figure 7D). TIP peptide moreover partially, but significantly, decreased concentrations of the proinflammatory cytokine IL-6 in lung homogenates from Sp-infected mice at 24 hours, whereas it did not significantly affect increases in expression of the proinflammatory TNF cytokine and the PMN-attracting KC chemokine (a.k.a. CXCL1) or of the antiinflammatory cytokine IL-10 (Figure 7E).

Figure 7.

Endothelial ENaC-α (enENaC-α) expression strengthens barrier function in pneumococcal pneumonia. (A) 8- to 10-week-old male enENaC-α KO mice develop significantly more capillary hyperpermeability 24 later upon intratracheal instillation of a moderate PLY dose (1.5 µg/kg) than age-matched control CRE mice (n = 3–5 per group; **P < 0.02 vs. vehicle Ctrl; ***P < 0.05 vs. CRE + PLY). (B) TIP peptide (2.5 mg/kg) significantly protects from capillary leak induced by intratracheal instillation of 10⁷ CFU of Streptococcus pneumoniae (Sp) in 8- to 10-week-old male WT C57BL6 mice (n = 4 per group; *P < 0.01 vs. vehicle and **P < 0.04 vs. Sp). (C) Infection with 2 × 10⁶ CFU Sp induces higher capillary leak in enENaC-α KO mice than in control CRE mice (n = 3–6 per group; *P < 0.03 vs. vehicle control and **P < 0.05 vs. CRE + Sp). (D) TIP peptide does not protect from Sp-induced capillary leak in enENaC-α KO mice. enENaC-α KO mice develop significant capillary leak 24 hours later intratracheal instillation of 2 × 10⁶ CFU Sp compared with enENaC-α KO control mice instilled with saline. Treatment with TIP peptide (2.5 mg/kg) did not significantly differ from vehicle-treated infected mice (n = 3–5 per group; *P < 0.003 vs. ctrl). (E) Measurement of cytokine/chemokine expression in lung homogenates (mean ± SD) (MILLIPLEX MAP Mouse Cytokine/Chemokine Magnetic Bead Panel–Immunology Multiplex Assay, EMD Millipore). Compared with vehicle control, 8- to 10-week-old male C57BL6 mice (n = 4 per group), mice instilled with 10⁷ CFU Sp have a significantly higher expression of TNF (**P < 0.004 vs. ctrl), keratinocyte-derived chemokine (KC) (**P < 0.006 vs. ctrl), and IL-6 (**P < 0.002 vs. ctrl) at 24 hours in lung homogenates, whereas levels of IL-10 were not significantly increased (n = 4 per group). Coinstillation of TIP peptide does not significantly reduce expression of TNF (**P < 0.003 vs. ctrl; ^#NS vs. Sp) or KC (*P < 0.02 vs. ctrl; ^#NS vs. Sp), but it does partially, but significantly, inhibit IL-6 expression (**P < 0.03 vs. ctrl; *P < 0.05 vs. ctrl; ^#P < 0.01 vs. Sp; n = 5 per group). CFU = colony-forming unit; NS = not significant.

Table 1.

TIP Peptide Treatment Does Not Significantly Affect Bacterial Load in Lungs 24 Hours after Infection

Treatment	Sp + Vehicle	Sp + TIP Peptide
Bacterial load in lung homogenates (in 107 CFU/ml, mean ± SD)	5.6 ± 1.8	8.8 ± 2.1

Definition of abbreviation: CFU = colony-forming unit; Sp = Streptococcus pneumoniae.

Bacterial load was assessed in lung homogenates from four vehicle-treated Sp D39-infected (10⁷ CFU/mouse) versus four Sp D39-infected TIP peptide-treated mice, 24 hours after infection. Samples were plated on blood agar petri dishes for 2 days at dilutions 1:10, 1:1,000, 1:10,000, and 1:1,000.000. P nonsignificant between Sp + TIP and Sp + vehicle.

Taken together, these results demonstrate an important barrier-protective role for enENaC-α in pneumococcal pneumonia–associated ARDS.

Depletion of ENaC-α Increases gp91^phox Expression in Mouse Lung Endothelial Cells In Situ after Infection with Sp

To substantiate our results from the cell studies, indicating a repressive role of enENaC-α on gp91^phox expression, we performed colocalization studies of the endothelial marker CD31 with the NOX2 complex subunit gp91^phox in lung slices from CRE driver control mice versus enENaC-α KO mice infected with Sp using the DuoLink method, as described (42). As shown in Figures 8A–8F, lung slices from infected mice (both Cre controls and enENaC-α KO mice) display a significantly higher colocalization of gp91^phox with CD31 than noninfected Cre control animals, as quantified in Figure 8G. Moreover, Sp-infected enENaC-α KO mice have a significantly higher endothelial gp91^phox expression than Sp-infected Cre driver mice. These observations substantiate our main hypothesis that ENaC-α represents a novel suppressor of NOX2 expression in lung endothelial cells, both in vitro and in situ.

Figure 8.

(A–F) Representative images of lung sections from enENaC-α KO and Cre-driver mice infected with Sp via intratracheal instillation. Vehicle saline was used instead of bacteria in Cre-driver control mice (Saline Ctrl, A and B). Red immunofluorescence staining demonstrates successful binding of both gp91^phox and CD31 antibodies in close proximity (⩽40 nm) in the lung sections. Nuclei are stained with DAPI (blue). Scale bars, 2 μm. (G) Number of areas of positive proximity ligation assay signal (red) per field in different regions of mouse lung tissue. Mean and error bars indicating the SD are included in the dot blot image. Statistical analysis was performed using t test. *P < 0.000001 vs. vehicle control. **P < 0.00000003 vs. saline control. ^#P < 0.0000001 vs. Sp CRE.

Discussion

ARDS represents a potentially lethal complication of severe pneumococcal pneumonia (1–4). The National Institutes of Health–sponsored ARDS Clinical Trials Network has recently organized several large, well-controlled trials of novel ARDS therapies (3, 4). Thus far, however, the only treatment found to consistently improve survival in ARDS is nonpharmacological: mechanical ventilation using low tidal volumes.

Neutrophils (PMNs) represent the first line of natural defense in patients infected with pneumococci. Sp engulfed by PMNs are enclosed in phagosomes, and O₂⁻ is released by activated NOX2 on the PMN membrane. The pneumococcal virulence factor PLY, the main inducer of lung capillary leak in murine pneumococcal pneumonia (27, 34–36), was suggested to be an important activator of NOX2 in PMNs (5). After natural dismutation or SOD activity, the highly unstable O₂⁻ rapidly converts to H₂O₂, which represents the precursor of several bactericidal secondary ROS, including hypochlorous acid generated by myeloperoxidase (43). The importance of this pathway for host defense is clearly documented in patients with chronic granulomatous disease (44) or with decompensated cirrhosis (45). Both of these patient groups are significantly more susceptible to bacterial infections, and this is accompanied by defective NOX2 activation in PMNs, as can be assessed by impaired phosphorylation of the p47^phox subunit (45).

NOX2 is activated in patients with pneumococcal pneumonia (9), where it is involved not only in beneficial host defense but also in deleterious capillary leak, a hallmark of ARDS. This dual nature of NOX2 may at least partially explain why global NOX2 inhibitors have not been evaluated in clinical trials, because of concerns for increased susceptibility to harmful infections (12). Our data from lung endothelial cells isolated from enENaC-α KO mice and from HL-MVECs with siRNA-depleted ENaC-α expression (Figure 4) demonstrate a previously unrecognized role for ENaC-α as a potent repressor of NOX2 expression. These findings are substantiated by the increased expression of endothelial gp91^phox after Sp infection in mice lacking enENaC-α, compared with infected control animals (Figure 8). We propose at least two potential mechanisms by which ENaC-α represses gp91^phox expression in HL-MVECs. First, as suggested by our co-IP results in mouse lungs, ENaC-α can directly associate with gp91^phox and, as such, impair subunit assembly of the active NOX2 oxidase enzyme (Figures 5A and 5B). Future studies will have to demonstrate that this occurs in lung endothelial cells. Second, as suggested by Figures 1G and 1H, we propose that ENaC-α activation can inhibit HDAC activity in endothelial cells. Not only were HDACs shown to increase ENaC-α degradation (31) but also they can promote transcription of gp91^phox (46). We do not expect that ENaC will affect expression of other NOX2 subunits, but we rather propose that their phosphorylation and recruitment to the surface membrane is inhibited upon ENaC activation, as indicated by the inhibition of p47^phox phosphorylation in PMA-treated HL-MVECs.

NOX2 plays a crucial role in PLY-induced hyperpermeability, because its specific inhibitor gp91dstat abrogated this activity in HL-MVEC monolayers (Figures 2A and 2B). Although the physical presence of ENaC-α seems to be sufficient to repress gp91^phox expression in lung endothelial cells, basal superoxide generation in cells lacking the subunit is not increased compared with control endothelial cells (Figure 6A), indicating that under these conditions there is no increased recruitment of cytosolic NOX2 subunits, which would be necessary for full activation of the enzyme. Once NOX2 is activated by PMA, however, cells isolated from enENaC-α KO mice start generating significantly higher superoxide levels than cells isolated from CRE control mice. TIP peptide reduces phosphorylation and surface expression of the p47^phox NOX2 subunit in PMA-stimulated HL-MVECs. Both PMA and PLY can induce Ca²⁺ influx in endothelial cells, which in turn activates PKC-mediated phosphorylation of cytosolic NOX2 subunits. Because TIP peptide blunts PLY-induced activation of the PLC/PKC pathway in lung endothelial cells (23, 27), this may provide a plausible explanation for its potent protective activity in NOX2-dependent PLY-induced hyperpermeability (27). The absence of enENaC-α makes mice significantly more susceptible to capillary leak induced by a low dose of PLY or a low inoculum of Sp (Figures 7A and 7C), and this effect was not due to the tamoxifen treatment, which in studies by us and others (47) did not significantly increase capillary leak in infected CRE mice, compared with infected CRE mice treated with a comparable regimen of corn oil (data not shown). ENaC activation by TIP peptide, moreover, strengthens lung capillary barrier function in mice instilled with pore-forming toxin PLY (27) or infected with a high inoculum of Sp (Figure 7B). However, in mice lacking enENaC-α, TIP peptide no longer protected from Sp infection–induced capillary leak (Figure 7D), thus indicating the peptide’s specificity for ENaC-α in its barrier-protective effect. The TIP peptide also exerted some antiinflammatory actions, because it partially blunted the increase in IL-6 expression in Sp-infected animals (Figure 7E). ENaC-α is absent in neutrophils (PMNs) but is expressed in resident target cells in the lungs during ARDS (i.e., AT1/2 and MVECs). We found no inhibitory effect of TIP peptide on ROS generation in PMA-stimulated PMNs, although we recorded that the peptide rather enhanced superoxide generation in these cells. Although a trend toward an increased bacterial load was recorded in Sp-infected mice treated with TIP peptide, this was not significant (Table 1), indicating there was no major impairment of host defense under these conditions. Our results therefore indicate that direct ENaC stimulation may rather specifically blunt ROS generation in ARDS target cells during pneumococcal pneumonia without affecting neutrophil-mediated antibacterial defense.

The TIP peptide (a.k.a. AP301, Solnatide) has been shown to be protective in preclinical acute lung injury studies in several species, including mouse (21, 22, 27), rat (48), rabbit (49), and pig (50), as well as in a rat model of ischemia-reperfusion–induced lung injury after isotransplantation (51), a model characterized by NOX2-dependent oxidative stress in lung endothelium (52). All these models are characterized by lung capillary endothelial dysfunction, hyperinflammation, and increased oxidative stress. TIP peptide has also been tested in two double-blind, placebo-controlled phase 2a clinical trials in patients with ARDS (28) and after lung transplantation (53). When given in the ventilator twice daily over 7 days, TIP peptide significantly reduced extravascular lung water in patients with ARDS with a sequential organ failure assessment score ⩾11 and, moreover, reduced the ventilation pressure (20 patients per group) (28). Patients who underwent lung transplantation who inhaled the peptide twice daily over 7 days were released from the ventilator significantly faster than patients in the placebo group (10 patients per group) (53). A third phase 2 dose-escalation study, designed according to U.S. Food and Drug Administration recommendations, is currently being conducted in several university hospitals in Germany and Austria (clinicaltrials.gov identifier: NCT03567577) (54).

Our study has several limitations. Although TIP peptide did not reduce superoxide generation in PMA-stimulated PMNs, it did increase it in an ENaC-α–independent manner. Because the TIP peptide mimics the lectin-like domain of TNF, which binds to specific oligosaccharides, such as branched trimannoses and N,N’-diacetylchitobiose (20), one way this may occur is through interaction with glycosylated molecules at the surface of the PMNs. As a concern, an overactivation of PMNs by TIP peptide might induce tissue damage during pneumonia. However, our in vivo data demonstrate that TIP peptide does not significantly affect bactericidal activity in mice with pneumococcal pneumonia, which mainly rely on PMNs for bacterial killing. As such, although we observe an increase in oxidative stress in PMA-treated PMNs in the presence of TIP peptide in vitro, this may not significantly affect their bactericidal or tissue damaging activity in pneumococcal pneumonia in vivo. Besides PMNs, alveolar macrophages can contribute to pneumococcal infection host defense, especially the repair process (55), which our mouse studies did not address. Although we did not observe an inhibitory effect of TIP peptide on ROS generation in PMA-stimulated human THP-1 macrophages, it should be noted that mouse macrophages can express ENaC-like amiloride-sensitive channels conducting Na⁺ (56, 57). Further studies are therefore necessary to elucidate the effects of TIP peptide on the macrophage-mediated repair process at later time points in the infection. Another limitation is that patients with severe ARDS require 100% fraction of inspired oxygen during ventilation, which can induce oxygen toxicity. Rather than NOX2, NOX1 was proposed as the main contributor to hyperoxia-induced lung injury (58). We can therefore not exclude that ENaC activation may not protect from oxidative stress in the alveolar–capillary barrier under these conditions. Another limitation is that enENaC-α KO mice still express ENaC-α in the type 1 and 2 alveolar epithelial cells. As such, because TIP peptide can also improve barrier function in other models of lung injury (e.g., in a rat model of high-altitude pulmonary edema, where the peptide increases occludin expression [48]), we cannot exclude the possibility that the protective effects of the peptide in the Sp infection model are mainly due to barrier-protective and antiinflammatory effects in the alveolar compartment, which could indirectly affect capillary barrier function. Another limitation is that we have mainly used PLY in our experiments assessing endothelial barrier function, whereas pneumococcal autolysin LytA has also been implicated in NOX2 activation in Sp-infected epithelial cells (6).

In conclusion, our results suggest that enENaC presents a differential therapeutic target, whereby modulating its activity could on the one hand prevent barrier impairment by excessive oxidative stress in pulmonary endothelial cells, without significantly impairing the crucial bactericidal activity expressed by phagocytes during pneumococcal pneumonia. These combined observations suggest a novel strategy to address capillary leak–inducing oxidative stress in ARDS, without impairing host defense to infection.