Pressure-dependent NOS activation contributes to endothelial hyperpermeability in a model of acute heart failure

Acute increases in left ventricular end diastolic pressure (LVEDP) and pulmonary capillary pressure (P_pc) can cause rapid and severe pulmonary edema (PE) [1]. Clinically, however, the severity of PE is often out of proportion to the increase in P_pc suggesting that other mechanisms may be operational. To understand this clinical disparity, we explored the relationship between sudden changes in hydrostatic pressure and acute changes in endothelial permeability. We have previously demonstrated that small changes in hydrostatic pressure produced large, rapid, and sustained increases in endothelial permeability, an effect mediated by nitric oxide (NO) [2] and reactive oxygen species (ROS) [3]. In fact, NO-mediated increases in endothelial permeability during increased hydrostatic pressure are common throughout the vascular system [4,5]. Thus, increases in both pressure and shear stress can activate nitric oxide synthase (NOS) and increase endothelial NO. Nitrosylation of adherens junction proteins is a trigger for junctional disassembly and results in acute increases in endothelial permeability [6]. The mechanisms whereby ROS contributes to endothelial hyperpermeability, however, are less understood. Reactive oxygen species are generated in a controlled and compartmentalized manner in the vasculature and although NADPH oxidases (Nox) are considered to be the main source of ROS in the vascular system, there is no evidence that Nox regulates barrier function; thus, other sources for ROS production, such as uncoupled endothelial NOS (eNOS) [7], may play a role in endothelial hyperpermeability and PE.

Interestingly, the majority of patients with chronic heart failure (HF) do not develop frank PE, even when P_pc is elevated [8,9] suggesting that alterations in endothelial signaling pathways, structural alteration of the vascular wall, or both, contribute to compensatory mechanism(s) that protect against sustained high pressure. We recognize that while NO causes acute changes in endothelial permeability, organic nitrites are used to treat pulmonary congestion and mitigate the symptoms of chronic HF. This paradox suggests, perhaps, that the physiological response of the pulmonary vasculature to NO and/or oxidative stress changes during the progression from acute HF (AHF) to chronic HF and, therefore, warrants further mechanistic characterization. Understanding the early mechanisms that regulate disease progression maybe an important step to better understand this paradox and improve therapeutic approaches to HF.

Herein, we describe a series of studies demonstrating the contribution of NOS activation to acute endothelial barrier failure in a model of AHF. The results demonstrate that eNOS has a dual role in AHF-related PE: it acts by increasing NO concentrations and second as a source of superoxide anion production leading to endothelial hyperpermeability and pulmonary vascular failure. Inhibition of NOS attenuates the physiological and histological pulmonary changes associated with AHF. The novel findings demonstrate that pressure-dependent NOS-activation results in enhancement of endothelial permeability and rapid PE. This represents a major paradigm shift in our understanding of hydrostatic PE and may lead to new therapeutic strategies to re-establish barrier integrity.

NADPH, Norepinephrine (NE), Dihydroethidium (DHE), protease inhibitor cocktail, and Krebs Ringer Buffer were bought from from Sigma Co, Ltd (St. Louis, MO); BSA was purchased from Proliant Biologicals (Boone, IA); antibodies were bought from EDM Millipore (Billerica, MA), Aviva Systems Biology (San Diego, CA) and Cell Signaling (Danvers, MA); Lucigenin and L-NAME were bought from Cayman Biochemicals; apocynin was bought from Calbiochem.

Animal studies were approved by the University of Illinois Institutional Animal Care and Use Committee. The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health. Male Sprague–Dawley rats were divided in the following groups: Sham, AHF; L-NAME + AHF (L-NAME + AHF); L-NAME; Apocynin (Apo); Apocynin + AHF (Apo + AHF); Ethanol (EtOH); Folic Acid (FA) and FA + AHF (FA + AHF).

Rats received a NE infusion starting at 7 μg/Kg/min and titrated to a mean arterial pressure (MAP) of 150 mm Hg for 2 h. Sham rats received a lactated ringers infusion (1.5 ml/h). For specific experiments, L-NAME or Apo (200 µmol/l/kg) was administered as a bolus over a 10-min period immediately before the induction of AHF. Additional control groups included solo administration of L-NAME and Apo to determine if it had any effects on the measured parameters independent of NE infusion. In order to assess the role of mechanotransduction in the development of PE during AHF, a non-hypertensive dose of L-NAME was chosen to be used in this model. With this approach, although increased mechanical forces and hemodynamics of HF are still present, we can specifically assess the contribution of NOS signaling to PE development.

Arterial blood gases, hematocrit (HCT) and pH were measured in 250 µl of whole blood using a GEM Premier 3000 machine (Instrumentation Laboratory, Orangeburg, NY), according to manufacturer’s instructions.

Left ventricular end systolic pressure (LVESP), LVEDP, and pulmonary artery (PA) pressures were measured using a saline-filled PE tubing (PE 50) connected to an arterial pressure transducer interfaced to a TAM-A amplifier (Harvard Apparatus, Holliston, MA).

Wet-to-dry (W/D) ratios were determined after drying lungs for 24 h at 60°C.

Lung injury was assessed by five blinded investigators and scored based on perivascular cuffing (PVC) and intra-alveolar hemorrhage (IAH). Total lung injury score (LIS) was determined by the weighted average of PVC and IAH.

The rat isolated perfused lung preparation was used as previously described [2,10]. Briefly, rats were anesthetized and mechanically ventilated. The PA and left atria were cannulated and lungs were perfused with Krebs–Ringer bicarbonate solution containing 3% BSA. Lungs were exposed to low pressure (6 cm H₂O) or high pressure (12 cm H₂O) for 60 min.

To assess if NE could induce lung hyperpermeability, 10⁻⁶ mol/L of NE was added to the perfusate reservoir and circulated for 1 h at low pressure. Lung W/D ratios were determined.

To assess if NE had a direct effect on lung vasculature, a concentration–response curve for NE (10⁻⁸ to 10⁻³ mol/L) was performed and pulmonary artery pressure (P_pa) was recorded.

BAL albumin content was assessed using a fluorimetric detection kit (Active Motif, Carlsbad, CA) according to manufacturer’s instructions.

MPO activity was measured in lung tissue and in BAL as previously described [11].

Total eNOS, iNOS, Nox1, and Nox2 expression was assessed in whole lung lysates. eNOS activity was determined by p-Ser¹¹⁷⁷/eNOS ratio; NO production was assessed for nitrotyrosine content as a marker for mechanotransduction activation. Immunoblots were performed as previously described [12]. Signal was detected by chemiluminescence using Li-COR system and band intensities were measured using Image Studio software (Li-COR, Lincoln, NE).

To assess whether eNOS is uncoupled, we performed a low-temperature gradient gel following standard Western blot techniques as previously described [13]. Signal was detected by chemiluminescence using Li-COR system. Band intensities were measured using Image Studio software (Li-COR).

ROS production was assessed by lucigenin enhanced chemiluminescence (ECL) as previously described [12]. To determine NOS-dependent superoxide generation, rats were treated with L-NAME and lucigenin ECL assay was performed. To confirm uncoupled eNOS as a source of ROS production in lung homogenates, we assessed ROS levels in lung lysates with exogenous tetrahydrobiopterin (BH4) supplementation (in vitro). Addition of PEG-SOD in vitro to lung samples was done as an additional control to show superoxide anion is being detected in this assay.

Nox activity was assessed in lung membrane fractions as previously described [12]. Results were normalized by protein content in the samples.

Data are presented as mean ± S.D. Groups were compared using one-way ANOVA or Student’s t test as appropriate. Tukey’s post-test was used to compensate for multiple test procedures. P<0.05 was considered statistically significant.

During general anesthesia, baseline MAP was 109.8 ± 10 mm Hg and heart rate (HR) averaged 348/min (Table 1). All hemodynamic variables were stable over the 2-h time period. Acute hypertension was induced by a NE infusion resulting in a rapid increase in MAP to 150 mm Hg and a significant increase in LVESP and LVEDP without changes in HR (see Table 1).

Acute hypertension caused a significant increase in pulmonary artery systolic pressure (PASP, 26 ± 8 compared with 39 ± 16 mm Hg) and a doubling of mean P_pa (mPAP; Sham = 13.14 ± 1.0 compared with AHF = 25.1 ± 6.3 mm Hg; Table 1). To characterize the direct effects NE on the pulmonary pressures, we used the in situ isolated perfused rat lung preparation and tested if NE infusion (10⁻⁸ to 10⁻³ mol/L) altered P_pa. NE had no direct effect on pulmonary vascular pressure (Supplementary Figure S1A), indicating that NE-induced pulmonary hypertension requires an intact circulation.

To clarify if NE per se could induce lung hyperpermeability, we used the isolated perfused lung preparation to test the effects of high pressure and NE on lung edema. While high pressure induced lung edema when compared with control group, NE did not alter lung W/D ratio when compared with control group (Supplementary Figure S1B). These results demonstrate that pressure, and not NE, is the main stimulus of endothelial hyperpermeability during AHF.

Elevations in LVEDP and PAP during AHF were associated with the development of PE. P_aO₂ dropped on an average to 50% (Figure 1A) and lung W/D ratio increased significantly (5.3 ± 0.52 compared with 6.44 ± 0.75; Figure 1B). Histological evidence of lung injury was manifested as significantly increased PVC and increased IAH. Higher LIS were observed in the dependent lung regions (dorsal) compared with non-dependent regions (ventral). In the dorsal regions, total LIS for Sham were 0.17 ± 0.05 and for AHF rats = 0.38 ± 0.05 (Supplementary Figures S2 and S3).

To determine if AHF and associated increase in P_cp caused an increase in leukocytes recruitment to the lungs that contributed to barrier failure, CD45 expression in lung sections, MPO activity in BAL and lung tissue were measured. As shown in Supplementary Figure S4A,B, lung MPO activity was not increased in the AHF BAL and lungs. In the same manner, no differences in CD45 staining was observed in lung sections from AHF lungs when compared with Sham lungs (Supplementary Figure S4D), ruling out increased white blood cell accumulation or activation as a cause for hyperpermeability. We also assessed the levels of interleukin 1-β (IL1-β), macrophage inflammatory protein 2 (MIP2), and tumor necrosis factor-α (TNF-α) by real-time PCR in lung lysates from sham and AHF rats. No differences in IL1-β, TNF-α, and MIP2 were found corroborating our previous data that this is not an inflammatory model of PE during AHF (Supplementary Figure S5).

To determine if epithelial barrier disruption contributed to the reduction in P_aO₂ and increase in lung W/D, albumin content of BAL fluid was measured. Acute increases in pulmonary vascular pressures were not associated with an increase in BAL albumin concentration (Supplementary Figure S4C) when compared with Sham BAL, indicating that the alveolar epithelial barrier remained intact.

To determine if acute vascular pressure increased lung NO production, we indirectly assessed NO generation by quantitating nitration of tyrosine residues in lung lysates. As shown in Figure 1C, acute hypertension and associated increase in mPAP resulted in a 40% increase in nitro-tyrosine content, consistent with our previous findings [2] and indicates activation of endothelial mechanotransduction.

To determine if pressure-dependent PE was caused by NO-mediated barrier failure, rats were treated with the NOS inhibitor, l-NAME, prior to the induction of AHF. l-NAME attenuated PE, returning P_aO₂ to Sham values (Figure 1A). In addition, l-NAME abolished the AHF-induced increase in lung W/D ratio (l-NAME + AHF wet/dry = 4.8 ± 0.41 compared with AHF = 6.4 ± 0.75) (Figure 1B). Pretreatment with l-NAME reduced histological evidence of lung injury during AHF. Histological LIS in the dorsal region were reduced in l-NAME + AHF to 0.381 ± 0.05, compared with LIS = 0.287 ± 0.06 (in AHF); P=0.0098 (Supplementary Figure S3).

In order to evaluate if ROS contribute to PE during pressure-induced AHF, we first measured ROS concentrations in lung homogenates from Sham and AHF lungs. AHF lungs showed increased ROS concentration when compared with Sham lungs (299.7 ± 140.3 in AHF compared with 50.91 ± 35.47 in Sham; Figure 2A). The increase in ROS concentration was inhibited when rats were pretreated with L-NAME (L-NAME + AHF: 52.94 ± 45.32) (Figure 2A) suggesting eNOS uncoupling. L-NAME, alone, did not affect ROS concentration (111.7 ± 51.10 in L-NAME compared with 50.91 ± 35.47 in Sham). To investigate if ROS contribute to PE, we assessed lung W/D ratio in rats treated with apocynin prior to induction of AHF. Apo inhibited PE development in AHF (W/D = 4.80 ± 0.21 in Apo + AHF compared with 6.44 ± 0.75 in AHF). Rats that received either apocynin only or EtOH (vehicle used to dissolve Apo) did not show differences in lung W/D ratio (Apo = 5.02 ± 0.7; EtOH = 4.8 ± 0.36) when compared with sham rats (5.25 ± 0.59) (Figure 2B).

To evaluate the participation of Nox in ROS production during AHF we assessed Nox activity in isolated membrane fraction of lungs collected from Sham and AHF rats. Nox activity was not altered in AHF rats relative to Shams (Figure 2C). To confirm that Nox was not involved in ROS production in this model, we assessed the expression of Nox isoforms 1 and 2 in lung lysates from Sham and AHF rats. No changes were observed in NOX1 and NOX2 content in lung lysates collected from AHF rats when compared with Sham rats (Figure 2D,E).

As L-NAME decreased ROS levels during AHF, we then considered eNOS as a source for oxidative stress in AHF and assessed eNOS uncoupling by Western blot. AHF lungs showed eNOS uncoupling which was not observed in Sham lungs. Pretreatment of AHF rats with L-NAME attenuated eNOS uncoupling during AHF (Figure 3A) thus supporting the findings that L-NAME inhibited ROS production. Collectively, these results suggest that eNOS uncoupling occurs due to excessive eNOS stimulation and the lack of enzyme substrate.

We evaluated the hypothesis that depletion of BH4 was contributing to eNOS uncoupling. Exogenous BH4 was added to lung lysates in vitro and ROS production was assessed by lucigenin ECL. As shown in Figure 3B, supplementation of BH4 in lung lysates decreased ROS levels in AHF lungs (from 290.5 ± 55.41 in AHF to 125 ± 71.17 in AHF + BH4). The same effect was observed when PEG-SOD was added to the lysates (51.88 ± 24.13 in AHF + PEG SOD), indicating that superoxide anion was the ROS being measured. These results support the hypothesis that uncoupled eNOS was a source of ROS. To further investigate if uncoupled eNOS contributes to PE development in vivo, we treated rats with FA, known to recouple eNOS by promoting BH4 recycling, before inducing AHF and assessed lung W/D ratio. In the presence of FA (FA + AHF group), rats that underwent AHF did not develop PE as evidenced by similar lung W/D ratio when compared with sham rats (4.43 ± 0.78 in FA + AHF compared with 5.2 ± 0.59 in Sham), Figure 3C.

We then investigated if caveolae degradation was part of the mechanism(s) leading to eNOS uncoupling in AHF lungs. Western blot analysis of lung homogenates did not demonstrate caveolae degradation (Figure 4A), ruling out caveola deterioration as a pathway to eNOS uncoupling in pressure-dependent AHF. Additionally, we investigated eNOS activity in lung lysates from sham and AHF rats at the final time point of the experimental model (2 h). No significant changes in eNOS activity were observed in AHF lungs when compared with sham lungs (Figure 4B).

In order to determine if differences in NOS isoforms expression occurred during AHF, we assessed eNOS and iNOS expression in lung lysates. No differences in NOS isoform expression was seen between groups (Figures 4C,D), thus ruling out iNOS contribution to increase NO production during AHF.

We used a hemodynamic model of HF that was induced by acute afterload mismatch, resulting in increased LVEDP, increased P_pa and subsequently, rapid PE [14–16]. This clinically relevant model allowed us to investigate the role of acute pressure-dependent endothelial mechanotransduction on vascular barrier failure as a mechanism of AHF-associated PE. Acute hypertension and the associated afterload mismatch resulted in: (i) a significant decrease in P_aO₂ that occurred within 30 min, (ii) an increase in lung W/D ratio and, (iii) an increase in histological LIS. Inhibition of NOS during AHF prevented the reduction in P_aO₂, maintained normal lung W/D ratio, and reduced histological LIS. The major novel finding of the present study is that in the absence of NOS-dependent enhancement in endothelial permeability, acute increases in pulmonary vascular pressure alone are insufficient to cause significant PE. This represents a major paradigm shift in our understanding the pathophysiology of hydrostatic PE.

These findings extend previous work on pressure-dependent endothelial mechanotransduction to a clinically relevant model of AHF. Kuebler et al. (2003) [17] were the first to demonstrate that acute increases in P_pc resulted in endothelial NO production. Previous work from our laboratory showed that pressure-dependent NOS activation, increased whole lung filtration coefficient (K_f) and endothelial hydraulic conductivity (L_p) [2,18].

The increase in pulmonary vascular pressure during NE-infusion resulted from a combination of increased LV afterload and a translocation of blood volume from the peripheral circulation to the lungs [19]. The combined effect of increased venous return and increased resistance to pulmonary outflow (increase in LVEDP) resulted in a significant increase in P_pc [20]. Because this is a model that combines two vasoactive effectors (pressure and NE) we also sought to identify the primary stimulus for endothelial hyperpermeability. In order to do so, we assessed the direct effects of NE on the pulmonary hemodynamics and edema development in the isolated perfused lung preparation. No changes in P_pa or in lung W/D ratio were found, corroborating previous findings by Krishnamoorthy et al. [21]. Given that HR did not increase during NE infusion, the increase in hydrostatic pressure appears the major stimulus for activation of NOS. Collectively, these results indicate that pressure, and not NE, is the main trigger for endothelial hyperpermeability

Inflammation is a controversial process in HF progression [22]. Although many animal studies suggest that inflammation is a definitive component of AHF, clinical evidence indicates that is not always the case [23]. Particularly, in conditions of elevated pulmonary vascular pressure, localized inflammation may arise from increased leukocyte margination in the lungs [24,25]. In order to evaluate if leukocyte margination contributed to lung endothelial hyperpermeability in the current model of AHF, we assessed myeloperoxidase activation. Myeloperoxidase activity remained unaltered in AHF lungs and BAL. In the context of inflammation, iNOS could be contributing to hyperpermeability in AHF as it has been previously reported to participate in barrier dysfunction [26]. We validate that iNOS expression was not increased in AHF lungs relative to Sham ruling iNOS as a mediator in this model of AHF. Finally, immunohistochemistry for leukocytes failed to demonstrate an increase in immune cells in the lungs from AHF rats and measurement of IL1-β, TNF-α, and MIP2 indicated that production of these cytokines are not altered during AHF.

The protective effects of L-NAME on reducing histological indices of lung injury during acutely increased vascular pressure were most notable for the reduction in IAH. This was a novel and an unexpected finding and contributes to the hypothesis that capillary stress failure may be a regulated process [27]. In the dorsal sections of the lung, where hydrostatic pressure would be highest in a supine rat, L-NAME attenuated hypertension-induced IAH. Collectively, these results suggest that pressure-dependent increases in endothelial permeability, including capillary stress failure, occur as a part of NO-mediated process. Studies are underway to further characterize the role of pressure-dependent NO production in capillary stress failure.

Activation of NOS by mechanical forces is a hallmark of endothelial mechanotransduction [2–4,28,29] and the signaling pathways that lead to NOS activation include glycocalyx-dependent signaling [3,30], neutral sphingomyelinase (NSMase) [31], IK_ATP [32], and TRP channels [33,34]. We have recently demonstrated that inhibition of NSMase prevents pressure-dependent increase in the whole lung filtration coefficient [35]. It is likely therefore, that pressure-dependent activation of NSMase and subsequent release of ceramide is part of the pathway leading to NOS activation.

Little is known about the mechanisms that lead to pulmonary oxidative stress during AHF. We observed that ROS levels were increased in AHF lungs when compared with Sham lungs and we expected that Nox would play a role in ROS production as they are considered the main source for ROS in the vasculature [36]. Our results however indicated that this was not the case: we found no differences in Nox activity or isoforms expression (Nox1 and Nox2) in AHF lungs when compared with Sham lungs. The fact that we did not see increased leukocyte recruitment in BAL or MPO activity in lung tissue and BAL fluid corroborate our findings that Nox 2 is not involved in ROS production during AHF and support the rationale that macrophages are not a source for ROS during AHF.

The evidence clearly suggested that NOS is activated in this model, so we explored eNOS as a potential source for oxidative stress. ROS levels were decreased with pretreatment of AHF rats with L-NAME suggesting eNOS was uncoupled in AHF rat lungs. Immunoblots of lung lysates confirmed a higher eNOS monomer/dimer ratio in this model supporting the idea that uncoupled eNOS was the key element in pulmonary oxidative stress. Treatment of AHF rats with apocynin mitigated lung edema, confirming a role for ROS in barrier failure. We explored the mechanisms that result in eNOS uncoupling, first by evaluating substrate availability (e.g. BH4) and second by assessing caveolae degradation. Supplementation of BH4 to lung lysates reduced ROS formation and confirmed that substrate depletion lead to eNOS uncoupling, thus supporting previous reports in the literature [37].

As eNOS is localized to the caveolae and caveolae dynamics is related to hyperpermeability in lung injury models [38,39], we investigated if caveolae degradation could be part of the mechanism leading to eNOS uncoupling [40]. No changes in caveolin-1 monomer/dimer ratio were found indicating that caveolae degradation is not a determinant for barrier failure in AHF. Our data support previous findings reporting uncoupled eNOS as a mediator of hyperpermeability [7] and as a mechanism in HF [37,41].

This is the first report to demonstrate a role for pressure-dependent endothelial hyperpermeability in the pathogenesis of AHF using an intact animal model. The results indicate that pressure-dependent NOS activation contributes to PE development during AHF by increasing NO and ROS production. The observation that L-NAME attenuates PE without altering pulmonary hemodynamics indicate that the increase in pressure per se is insufficient to cause lung edema challenging the long-held notion that Starling forces alone can explain rapid and severe PE during AHF. To the contrary, our results suggest that rapid endothelial hyperpermeability, the end result of endothelial mechanotransduction is required in order for the prevailing hydrostatic pressure to cause clinically significant PE (Figure 5). This is a translational study that reveals NOS as a potential target for treating PE during AHF.

• The endothelial cell is an active participant in AHF.

• Endothelial mechanotransduction contributes to pathogenesis of AHF.

• Targetting endothelial hyperpermeability could be a novel approach for treating AHF.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the Department of Anesthesiology at University of Illinois at Chicago .

A.Z.C. wrote the manuscript and contributed to study conception, research design, data acquisition, analysis, and interpretation. A.I., R.R., J.S., and J.S. contributed to data acquisition. M.P. contributed to data acquisition and analysis. B.A.B. participated in writing the manuscript. R.O.D. wrote the manuscript and was responsible for study conception, research design, and supervised data acquisition and interpretation.

AHF

acute heart failure

Apo

apocynin

BAL

bronchoalveolar lavage

BH4

tetrahydrobiopterin

eNOS

endothelial nitric oxide synthase

EtOH

ethanol

FA

folic acid

HF

heart failure

HR

heart rate

IAH

intra-alveolar hemorrhage

iNOS

inducible nitric oxide synthase

LIS

lung injury score

LVEDP

left ventricular end diastolic pressure

LVESP

left ventricular end systolic pressure

MAP

mean arterial pressure

MIP2

macrophage inflammatory protein 2

mPAP

mean pulmonary artery pressure

MPO

myeloperoxidase

NE

norepinephrine

NOS

nitric oxide synthase

Nox

NADPH oxidase

NSMase

neutral sphingomyelinase

PE

pulmonary edema

P_pa

pulmonary artery pressure

Ppc

pulmonary capillary pressure

PVC

perivascular cuffing

ROS

reactive oxygen species

TNF-α

tumor necrosis factor-α

W/D

wet to dry