Acute lung injury (ALI), which is a cause of high mortality and morbidity, associates with predisposing conditions such as sepsis, gastric acid aspiration, and trauma. As pathogenic mechanisms underlying ALI continue to be inadequately understood, there is considerable interest in developing animal models for mechanistic studies. Animal models of ALI are established by exposing lungs to pathogenic conditions either through the airway, or the vascular route. Several models are available (Matute-Bello et al., 2008; Matute-Bello et al., 2011). Examples are the pneumonitis model in which endotoxin is given intratracheally or intraperitoneally, or the acid-injury model in which concentrated hydrochloric acid (HCl) is given intratracheally.
Animal models are a good platform for addressing a well-known feature of ALI, namely, the rapidity with which edema develops across the lung. Although several mechanisms could account for the rapid onset, in a recent report, we showed that intercellular communication in lung microvessels plays a role (Parthasarathi et al., 2006). Uncaging-induced increases of Ca2+ spread through endothelial gap junctional channels, spreading proinflammatory effects to regions more extensive than that directly affected by the uncaging. Although these findings indicate that proinflammatory signaling spreads spatially in the lung microvascular bed, it is not clear whether similar mechanisms cause spatial spread of lung injury.
The difficulty with testing this hypothesis is that available ALI models induce lung injury in a global manner, precluding quantification of spatial spread. Here, we developed a novel lung injury model in which our objective was to deliver the injurious agent to a localized region of the lung parenchyma. As the lung parenchyma is eminently accessible to focal sampling, our plan was to cause focal lung injury, then sample lung tissue from the directly injured region as well as from adjoining regions that did not receive the injury agent. To this end, we delivered concentrated HCl through a catheter wedged in the left lower lobe of the rat lung. Our findings indicate that this approach provides a suitable model for inducing focal lung injury and that the injury spreads to the adjoining parenchyma by gap junctional mechanisms.
Animals were treated in accord with protocols approved by the Institutional Animal Care and Use Committee of the Columbia University Medical Center or that of the University of Tennessee Health Science Center, as appropriate. Animals were given ad libitum access to food and water and placed on a 12-hr light-dark cycle. We used adult male Sprague Dawley rats weighing 400–500 g. We prepared the isolated blood-perfused lungs as reported (Parthasarathi et al., 2002). Briefly, lungs were excised and continuously pump-perfused with autologous blood at 37°C. Final perfusate hematocrit was 20%. The perfusion rate was maintained at 14 mL/min. Perfusion pressures were 10 and 3 cmH2O in the pulmonary artery and left atrium, respectively. Lungs were continuously ventilated with room air at a tidal volume of 6 cc/kg, with positive end-expiratory pressure set to 5 cmH2O for the entire duration of the experiment. Lungs were allowed to stabilize for a minimum of 30 min after initiation of perfusion.
We used HCl in concentrations of 0.1, 0.05, and 0.025 N (Fisher Scientific, Pittsburgh, PA). For gap junction inhibition, we used 18α-glycyrrhetinic acid (GA; 5 μM) (Calbiochem, La Jolla, CA).
A PE-10 (Becton-Dickinson, Sparks, MD) catheter was advanced into the airways through the tracheal cannula until it met firm resistance. Subsequently, 0.1 mL of HCl in Hepes-buffered Ringers was delivered via the tracheal catheter over a 2-min duration using a syringe infusion pump (Braintree Scientific, Braintree, MA). To mark the site of instillation, we mixed evans blue (EB; 0.4 mg/mL) with HCl before instillation. All lungs were continuously ventilated and perfused during acid instillation. At the end of the instillation, the catheter was withdrawn. After 1 hr, the ventilation and perfusion connections were disconnected; the lungs were inflated to 5 cmH2O and flash frozen in liquid nitrogen. Lungs were stored in a −70°C freezer until further processing.
Subsequently, we extracted tissue samples at multiple sites from each frozen lung preparation using a sharp hollow cylinder. Lung samples were collected from: (a) the site of acid instillation identified by the blue stain, (b) 1 cm away from the site of acid instillation, (c) 2 cm away from the site of acid instillation, and (d) contralateral lung. The blood-free extravascular lung water (EVLW) was quantified for each sample as reported (Bhattacharya et al., 1989; Ying et al., 1994; Safdar et al., 2005; Eckle et al., 2008; Munoz et al., 2009). In brief, the tissue samples were separately homogenized. A portion of each homogenate was weighed using a microgram balance with accuracy in the 10-μg range, oven-dried at 70°C until the dry weight remained unchanged over a 24-hr period. The remaining homogenate was centrifuged at 15,000g for 30 min. The supernatant was extracted and used to correct for residual blood volume in the corresponding sample as in our previous reports (Ying et al., 1994; Safdar et al., 2005). As shown therein, the method is sensitive in detecting lung water changes of 25%.
For inhibiting endothelial gap junctions, GA was added to the perfusate 10 min before acid instillation. For inhibiting epithelial gap junctions, GA was mixed with HCl and instilled together into the airways.
In a separate set of lungs, we instilled 0.1 N HCl mixed with fluorescein isothiocyanate-dextran 70 kD (FDx: 1mg/mL) via the tracheal catheter and extracted lung samples as detailed above. We then determined the FDx level in each tissue sample, as in the next subsection.
The various experimental protocols of this study are listed below, with the number of repetitions in brackets. Tissue sample EVLW was quantified in the following protocols (a) instillation of buffer + EB (n = 4), (b) instillation of 0.1 N HCl + EB (n = 7), (c) instillation of 0.05 N HCl + EB (n = 5), (d) instillation of 0.025 N HCl + EB (n = 3), (e) vascular pretreatment with GA followed by instillation of buffer + EB (n = 4), and (f) vascular pretreatment with GA followed by instillation of 0.1 N HCl + EB (n = 7). EVLW was not quantified in lungs of the following protocols: (g) instillation of 0.1 N + EB to establish the spread of the instilled acid using EB absorbance (n = 6), and (i) instillation of 0.1 N + FDx to establish the spread of the instilled acid using FDx fluorescence (n = 6).
Quantifying EB and FDx in Lung Samples
To determine the levels of EB and FDx in the extracted tissue samples, we homogenized the samples individually, centrifuged the homogenate at 15,000g for 30 min, and then collected the supernatant. For EB, we quantified the absorbance of the supernatant at 620 nm using a spectrophotometer (Molecular Devices). For FDx, we filled capillary tubes individually with supernatant from the various tissue samples and captured the FITC fluorescence emission at 520 to 490 nm excitation using a standard epifluorescence microscope. Subsequently, we quantified the fluorescence using image analysis software. Separately, we established that EB absorbance and FDx fluorescence were not significantly changed by the addition of 0.1 N HCl to EB or FDx. Moreover, we also verified that pH of HCl did not change significantly by addition of either EB or FDx.
All data are reported as mean ± standard error of the mean. All measurements were made in triplicate. Lung water data for a tissue sample was grouped with data for other tissue samples from corresponding sites of all lungs treated similarly. Multiple groups were compared with one-way analysis of variance (ANOVA) or Kruskal–Wallis one way ANOVA on ranks, depending on whether the datasets passed the normality test. ANOVA was followed by pairwise comparisons by Holm-Sidak method. ANOVA on ranks was followed by pairwise multiple comparisons by Dunn's method (for groups of unequal sample sizes) or Tukey's test. The exact combination of tests used for each set of data is indicated in the respective figure legend.
Focal Acid Delivery
The intratracheally inserted catheter usually wedged in either the left or right lower lobe. Instillation of HCl alone through the catheter caused localized discoloration of lung tissue (Figs. 1A,B). To mark the acid-treated region more prominently, we instilled a mixture of EB and HCl. EB staining was restricted to a well-defined region of the lung, indicating that the catheter enabled focal delivery to regions comprising ∼20% of the surface area of the lobe (Fig. 1B).
To determine the dispersion of the injected acid, we instilled HCl+EB and extracted deep-tissue samples obtained directly from EB-stained regions, as well as from adjacent, unstained regions. Assays of supernatants derived from homogenates of the tissue samples indicated that EB-absorbance was present only in surface-stained regions (Fig. 2A). Similar results were obtained when we instilled HCl+FDx and quantified FDx fluorescence (Fig. 2A) in supernatants as above. Fluorescent images of glass capillaries filled with supernatant from tissue samples of HCl+FDx-instilled lungs show that FDx fluorescence was detectable only in the acid-instilled regions (Fig. 2B). These data indicate that catheter method was effective in delivering acid to a focal region of the lung.
To determine the extent of lung injury at the acid-instilled and adjoining sites, we quantified EVLW of tissue samples obtained at increasing distances from the acid-instilled site (Fig. 3, inset). At the site of instillation, as indicated by EB discoloration, acid induced a concentration-dependent increase in EVLW (Fig. 3). In nonstained sites, EVLW increases were detectable in deep-tissue samples taken at distances of up to 4 cm from the site of direct acid instillation (Fig. 3). However, no increase in EVLW was evident in the contralateral lung (Fig. 3). These data indicate that acid caused lung injury not only at sites that received the acid, but also at adjacent sites that did not directly receive the acid.
Gap Junction Inhibition
As our findings indicated that there was a spread of lung injury from the primary site of acid instillation to adjacent sites, we considered the possibility that gap junctions were involved. To test this hypothesis, we pretreated lung vessels with the gap junctional blocker, glycerrhetinic acid (GA). In GA-treated lungs, EVLW increase at the acid-instilled site was similar to untreated lungs (Fig. 4A). However, in contrast to the untreated lung, the EVLW increase was reduced in the adjacent acid-free sites (Fig. 4A). Comparison of the individual data indicated that vascular GA pretreatment significantly blunted the acid-induced EVLW increases in adjacent sites located 1 cm away (Fig. 4B). Further, inhibiting epithelial gap junction communication with GA did not block the response at sites located 1 cm away from the site of acid instillation (Fig. 4C). Hence, we conclude that inhibiting endothelial gap junction-dependent intercellular communication inhibited spread of lung injury.
Mechanisms underlying the spatial expansion of lung injury remain unclear. Here, we developed a novel, catheter-based method for localized acid-induced lung injury. Our study indicates that following focal acid instillation in a lung lobe, lung injury occurs not only directly in the acid instilled region but also in adjacent regions that did not receive the acid.
To establish the region that made direct contact with the instilled acid, we co-instilled the fluorescent marker, EB. The dye prominently demarcated the surface boundaries of the acid-treated region. Reports indicate that arterial oxygen content decreases and protein leakage increases 1-hr post-acid instillation (Folkesson and Matthay, 1997; Yamada et al., 2000; Matt et al., 2009) and endothelial permeability increases in the acid treated regions (Modelska et al., 1999). Consistent with these reports, our findings show that lung water increased at the site of acid instillation.
The novel finding was that in addition to lung injury at the focal instillation site, lung water increased in adjacent regions that did not directly receive acid. Both epithelial and endothelial mechanisms may contribute to acid injury. The absence of the markers, EB and FDx, in the adjacent regions clearly suggests that these regions did not receive any acid and hence, not directly injured by the acid. The rapid buffering of acid in airways further limits the possibility that the instilled acid spread far from the site of instillation and induced direct injury in the adjacent areas. Inhibiting endothelial gap junctional communication with intravascular GA blunted the increases in lung water at the adjacent sites. This blunting was not the result of intravascular GA leaking into airspaces before acid instillation and blocking epithelial gap junctions, as the isolated lung preparations were maintained at pressures that do not allow small solutes to cross the microvascular barrier (Parthasarathi et al., 2002). However, post-instillation, GA may have leaked into air spaces with the edema fluid. However, GA instilled into airways with acid did not inhibit the acid-induced responses at any measurement site. Hence, we interpret that inhibition of epithelial gap junctional communication by GA leaking with the edema fluid did not contribute to the reduction in lung water increases at the adjacent sites. This suggests that the acid-induced responses were independent of epithelial mechanisms. Thus, endothelial mechanisms may be predominant in increasing lung water at the acid-free regions.
At sites of direct acid injury proinflammatory second messengers might be generated in endothelial cells as a result of epithelial-endothelial crosstalk, as suggested previously for tumor necrosis factor (TNF)-treated lungs (Kuebler et al., 2000). Endothelial gap junctions might have mediated the communication of the proinflammatory mediators from the acid-treated site to adjacent uninjured sites, as previously reported (Parthasarathi et al., 2006). Thus, lung water increase at these uninjured sites might have occurred as a consequence of this communication. Thus, we interpret that interendothelial communication mediates the propagation of lung injury. While cytokines released into the circulation during injury are reported to expand injury (Tutor et al., 1994), the absence of lung water increases in the contralateral lung suggests that this mechanism was not relevant here.
To inhibit endothelial gap junctional communication, we pretreated vessels with GA. Although it is well known that GA blocks intercellular gap junctional communication (Braet et al., 2003; Earley et al., 2004; Yu et al., 2010), nonspecific effects need to be considered. GA might have induced anti-inflammatory effects (Feinstein and Schleimer, 1999; Kao et al., 2010). However, this appears to be unlikely here since at the site of direct acid injury, increases in lung water were similar for both GA-treated and -untreated lungs. Moreover, the anti-inflammatory effect of 18-alpha-GA isomer used here is small compared with that of the 18-beta-GA isomer (Antolini et al., 1992; Kroes et al., 1997), for which the anti-inflammatory effect requires hours of exposure (Kao et al., 2010). Here, GA exposure was for 10 min, which is very short for a substantial anti-inflammatory. Thus, the present lung water effect by GA was due to inhibition of endothelial gap junctional communication.
These observations extend our previous findings that interendothelial gap junctional communication is the primary mechanism for the spread of the proinflammatory responses in microvessels (Parthasarathi et al., 2006). While second messengers involved in this study remain unknown, it is possible that cytosolic Ca2+ increases mediate the spatial spread of acid injury since endothelial Ca2+ mediates permeability changes in response to thrombin, stretch, TNF, and pulmonary hypertension (Ichimura et al., 2005; Tiruppathi et al., 2006; Townsley et al., 2006; Kuebler et al., 2010). Moreover, we previously reported that alveolar TNF instillation increases vascular endothelial Ca2+ (Kuebler et al., 2000). It is possible that airway acid instillation initiates endothelial Ca2+-dependent spatial expansion of acid injury. Further studies are required to better understand the role of Ca2+ in acid injury.
In this study, we use focal instillation of acid to establish the role of endothelial gap junctions in acid injury. Other injury models such as intratracheal instillation of acid may initiate a more global injury. Even under these conditions, inhibiting gap junctional communication might protect the healthy uninjured parenchyma not directly subjected to acid contact. However, this issue remains to be resolved.
In conclusion, using a novel method of focal acid instillation, we show that focal acid injury induces lung water increases in regions within the same lung lobe but spatially distant from the site of acid instillation. Endothelial gap junctional communication mediates this spatial extension of injury. Inhibition of gap junctional communication may be therapeutic in the treatment of ALI.
This project was supported by NIH grants HL75503 to KP, and HL57764 and HL36024 to JB. The authors thank Dr. Christopher Waters for his helpful comments on the manuscript.