Protective and Detrimental Effects of Sodium Sulfide and Hydrogen Sulfide in Murine Ventilator-induced Lung Injury

• Sulfides such as sodium sulfide and hydrogen sulfide have both protective and detrimental effects on cells

• High tidal volume ventilation can produce acute lung injury

• Inhalation of hydrogen sulfide has no beneficial effects on ventilator-induced lung injury in mice

• Intravascular administration of sodium sulfide attenuates the development of ventilator-induced lung injury through antioxidative signaling pathways

MECHANICAL ventilation can be life-saving in acute respiratory failure but may contribute to lung injury because of its side effects.1,2Cyclic lung stretch during mechanical ventilation induces tissue disruption and activates proinflammatory pathways, promoting the formation of pulmonary edema and neutrophil infiltration.3In addition, cyclic lung stretch is associated with generation of reactive oxygen species and production of redox imbalance in the lung.4,–,6Together, these events may initiate and perpetuate oxidative stress and local and systemic inflammatory responses, enhance lung injury, and lead to multiorgan failure and mortality.7,–,9Developing novel therapies targeted at oxidative stress and inflammation in ventilator-induced lung injury (VILI) could be useful clinically.

Hydrogen sulfide (H²S) is an endogenous gaseous transmitter.10The antiinflammatory and antiapoptotic effects of H²S and other sulfides, such as sodium hydrogen sulfide or disodium sulfide (Na²S), have been studied in various animal models of inflammation (reviewed in Baumgart et al.  11) and oxidative stress.12,13In particular, Calvert et al.  reported that Na²S protects against cardiac ischemia or reperfusion injury by up-regulating Nrf2-dependent antioxidant and detoxification proteins.14Nrf2 (i.e. , nuclear factor E2-related factor 2) is a key transcription factor that regulates antioxidant genes as an adaptive response to oxidative stress or pharmacologic stimuli. Downstream targets of Nrf2 include direct antioxidant proteins (such as glutathione peroxidase, NAD(P)H oxidoreductase, etc .), thiol-metabolism associated detoxifying enzymes (such as glutathione-S-transferase, glutamate–cysteine ligase, thioredoxin, etc .), stress-response genes (heme oxygenase, heat-shock proteins, ferritin, etc .), and others. In a murine model of VILI, it has been shown that Nrf2-dependent antioxidant genes play a key protective role in reducing oxidative stress.5Whether or not H²S can protect against VILI by modulating the expression of Nrf2-dependent antioxidant genes has not been reported.

Inhalation of H²S gas is known to cause airway mucosa irritation and cytotoxicity,10,15has been reported to exert proinflammatory effects in various models (reviewed in Szabó10and Baumgart et al.  11), and traditionally has been considered a health hazard. On the other hand, Faller et al.  recently showed that inhaled H²S at 80 ppm prevents lung injury in mice ventilated with a tidal volume of 12 ml/kg (plateau pressure 10–13 cm H²O, i.e. , moderate stretch).16Nonetheless, molecular mechanisms responsible for the protective effects of H²S inhalation against VILI are not completely understood. In addition, it remains uncertain whether inhaled H²S is protective or deleterious during mechanical ventilation that produces increased lung stretch, such as in patients with acute respiratory distress syndrome.

To elucidate the role of H²S in VILI, we examined the effect of inhaled H²S in a model of lung injury induced by high tidal volume (HV^T) ventilation. We found that H²S gas promotes VILI and enhances the pulmonary expression of leukocyte adhesion and chemoattractant molecules. Subsequently, we hypothesized that intravascular administration of Na²S, avoiding direct exposure of the lung to H²S gas, could provide better protection against VILI than could inhaled H²S. We report that intravascular Na²S both attenuates the pulmonary expression of chemoattractant and leukocyte adhesion molecules and enhances Nrf2-dependent expression of antioxidant genes, thereby attenuating VILI from HV^Tventilation.

This study was approved by the Institutional Animal Care and Use Committee (Subcommittee on Research Animal Care, Massachusetts General Hospital, Boston, Massachusetts), and conforms to the revised Guide for the Care and Use of Laboratory Animals.

Male C57BL/6 mice (23.3 ± 1.0 g, mean ± SD) were subjected to HV^Tventilation as described previously6with minor modifications of inspired oxygen fraction (F^IO2) positive end-expiratory pressure, and alveolar recruitment. Briefly, after anesthesia was induced and tracheostomy performed, mice were ventilated in a volume-controlled mode (Inspira; Harvard Apparatus, Boston, MA) at a tidal volume (V^T) of 10 ml/kg, respiratory rate of 90 breaths/min, positive end-expiratory pressure of 2 cm H²O, and FIO²0.4 for 1 h (baseline ventilation). Immediately after the tracheostomy and after the initiation of ventilation, a carotid catheter was inserted for blood pressure monitoring, continuous infusion of anesthetics, and blood sampling. After 1 h of baseline ventilation, the ventilator settings were switched to an HV^Tof 40 ml/kg, positive end-expiratory pressure of 1 cm H²O, and respiratory rate of 60 breaths/min, and mice were ventilated for a maximum duration of 240 min of HV^Tventilation. The FIO²(0.4) was not changed. Because a plateau pressure exceeding 30 cm H²O is associated with increased mortality1and commonly is used as a trigger for initiating rescue therapies in patients with acute respiratory distress syndrome, we aimed for plateau pressures of more than 30 cm H²O to induce VILI in mice. In this model using normal lungs, a tidal volume of 40 ml/kg was necessary to increase plateau pressure above 30 cm H²O, resulting in a peak pressure of ∼35 cm H²O.

Alveolar recruitment maneuvers were performed every 30 min during the baseline period and every 60 min during HV^Tventilation. Body temperature was maintained at 37°C with a heating pad. See Materials and Methods in Supplemental Digital Content 1, http://links.lww.com/ALN/A775, which provides the detailed methodology of the mouse model of VILI.

After 30 min of baseline ventilation (i.e. , 30 min before  the onset of injurious HV^Tventilation), mice were treated with either inhaled H²S or intravascular Na²S as follows: H²S gas (hydrogen sulfide 100 ppm, balance nitrogen, MedTechGases, Medford, MA) was diluted and added continuously to the inhaled gas mixture (FIO²= 0.4) at a concentration of 0, 1, 5, or 60 ppm (n = 6 each) until the end of the study. Alternatively, mice received a single intraarterial bolus injection of Na²S (sodium sulfide nonahydrate, Sigma–Aldrich, St. Louis, MO; 0.55 mg/kg in 5.5 ml vehicle/kg body weight, n = 8) or vehicle alone (Dulbecco's phosphate buffered saline, Sigma–Aldrich; 5.5 ml/kg body weight, n = 8).

The animals were killed after 240 min of HV^Tventilation or when their mean arterial pressure decreased to less than 60 mmHg for more than 5 min. Immediately after the animals were killed, respiratory mechanics were assessed and arterial blood was collected. Bronchoalveolar lavage (BAL) and the collection of lung tissues were performed as described in Supplemental Digital Content 1, http://links.lww.com/ALN/A775.

Control mice not subjected to HV^Tventilation (n = 4) were killed after 5 min of baseline ventilation.

We measured V^T, peak inspiratory airway pressure (PIP), inspiratory capacity, compliance of the respiratory system, pressure-volume curves, and arterial blood gas tensions, as described in Supplemental Digital Content 1, http://links.lww.com/ALN/A775.

Protein concentration in BAL fluid was measured (Bradford assay; Sigma–Aldrich) to assess pulmonary edema formation. Lung inflammation was determined by measuring interleukin-6 (IL-6 enzyme-linked immunosorbent assay; R&D Systems, Minneapolis, MN) and leukocyte concentrations in BAL fluid.

Paraffin-embedded 6-μm lung sections were stained with hematoxylin and eosin or reacted with an antimouse neutrophil antibody (Cedarlane Laboratories, Burlington, Ontario, Canada) to evaluate histopathologic changes and neutrophil infiltration (see Materials and Methods in Supplemental Digital Content 1, http://links.lww.com/ALN/A775, which provides the detailed methodology of staining tissue sections).

Total, reduced (GSH) and oxidized (GSSG), glutathione concentrations in lung tissue were measured using an enzymatic assay (Cayman Chemical, Ann Arbor, MI).

Quantitative real-time polymerase chain reaction was used to determine the pulmonary expression of the neutrophil chemoattractant cytokine CXCL-2 (i.e. , macrophage inflammatory protein 2), the leukocyte adhesion molecules CD11b (i.e. , integrin α M), and L-selectin, as well as the following Nrf2-dependent antioxidant genes: NAD(P)H:quinone oxidoreductase (NQO1), glutathione-S-transferase A4 (GST-A4), and glutathione peroxidase 2 (GPX2) (see Materials and Methods in Supplemental Digital Content 1, http://links.lww.com/ALN/A775, which provides the detailed methodology of polymerase chain reaction).

Animals subjected to HV^Tventilation without  H²S (0 ppm) were compared with H²S-treated animals (60 ppm) subjected to HV^Tventilation and with animals not subjected to HV^Tventilation (controls) using ANOVA and Bonferroni multiple comparison tests (two comparisons: controls vs.  0 ppm H²S; 0 vs.  60 ppm H²S). Na²S-treated animals subjected to HV^Tventilation were compared with vehicle-injected animals using Student unpaired t  test (between-group comparison). Survival curves were compared with the log-rank Mantel–Cox test. Compliance of the respiratory system and PIP values at the beginning (1 h of HV^T) and end of HV^Tventilation were compared with the Student paired t  test (within-group comparison). All tests were two-tailed and were performed with GraphPad Prism® version 5.01 for Windows (GraphPad Software, San Diego, CA). Data are expressed as mean ± SEM, unless indicated otherwise.

Mice breathing 40% oxygen without added H²S died of acute lung injury or fulfilled the criteria for being killed after 224 min (median) of HV^Tventilation (fig. 1). That is similar to the median survival time of mice ventilated with 1 ppm (236 min) or 5 ppm (215 min) of inhaled H²S. In contrast, the median survival time during HV^Tventilation was shorter in mice breathing 60 ppm H²S (166 min; P = 0.0016 vs.  0 ppm H²S). The median survival time in vehicle-treated animals was 231 min. Remarkably, all Na²S-treated animals were alive at the end of the experiment (240 min of HV^Tventilation) and were then killed.

Ventilation with a tidal volume of 40 ml/kg resulted in a PIP of 33–35 cm H²O in all groups during the first hour of HV^Tventilation. In mice breathing 40% oxygen without added H²S, HV^Tventilation induced a gradual increase of PIP to 43 cm H²O by the end of the experiment. A comparable increase in PIP was observed in mice ventilated with 1, 5, or 60 ppm H²S (table 1). This increased PIP was associated with a marked reduction of respiratory system compliance (table 1). Reduced respiratory system compliance was also reflected by a significant downward shift of the pressure-volume curves (fig. 2) measured in mice after HV^Tventilation with and without H²S compared with those curves measured in mice not subjected to HV^Tventilation.

In control mice, killed after 5 min of baseline ventilation (FIO²0.4), the PaCO²was 40 ± 4 mmHg, and the PaO²was 239 ± 16 mmHg (fig. 3). After HV^Tventilation (0 ppm H²S), PaCO²did not change, but PaO²decreased to 59 ± 17 mmHg at the end of the experiment. Similarly, in mice breathing 1 or 5 ppm H²S, PaCO²did not change but PaO²decreased to 60 ± 6 and 56 ± 8 mmHg, respectively. In contrast, in mice breathing 60 ppm H²S, PaCO²increased to 88 ± 13 mmHg (P = 0.0001 vs.  0 ppm) and PaO²decreased to 77 ± 10 mmHg (P = not significant vs.  0 ppm) in response to HV^Tventilation.

To investigate whether alveolar-capillary barrier disruption with pulmonary edema formation contributes to the observed impairment of arterial oxygenation, the protein concentration in BAL fluid was measured (fig. 4). BAL fluid protein concentrations were consistently higher in all mice subjected to HV^Tventilation with and without inhaled H²S than in controls.

To evaluate pulmonary inflammation, we measured the concentrations of the proinflammatory cytokine IL-6 and the concentration of leukocytes in BAL fluid (fig. 5). BAL fluid IL-6 concentrations were greater in mice subjected to HV^Tventilation than in controls. In contrast, BAL fluid IL-6 concentrations were decreased in mice ventilated with 60 ppm H²S than without H²S (fig. 5A). BAL fluid leukocyte concentrations (fig. 5B) invariably were reduced by HV^Tventilation in mice compared with controls, independent of whether the mice were ventilated with or without H²S.

The pulmonary messenger RNA (mRNA) concentrations of CXCL-2, CD11b, and L-selectin in mice subjected to HV^Tventilation were approximately 7-, 3-, and 1.5-fold greater than in controls (fig. 6). Inhalation of 60 ppm H²S during HV^Tventilation augmented the increase in CXCL-2, CD11b, and L-selectin mRNA concentrations (15-, 6-, and 3-increase vs.  controls).

The concentration of total glutathione and the ratio of reduced to oxidized glutathione (GSH/GSSG) were significantly lower in lungs subjected to HV^Tventilation than in controls (fig. 7). This decrease was independent of whether mice were ventilated with or without H²S.

The pulmonary mRNA concentrations of NQO1 and GPX2 in mice subjected to HV^Tventilation were approximately 1.6- (not significant) and 2-fold higher than in controls (fig. 8). Inhalation of 60 ppm H²S during HV^Tventilation attenuated the increase in NQO1 and GPX2 mRNA concentrations (1.1- and 1.3-fold increase vs.  controls). HV^Tventilation with 0 or 60 ppm H²S reduced the pulmonary mRNA concentrations of GST-A4 to 66% and 54%, respectively, of the concentrations measured in control mice.

In vehicle-treated animals, PIP and respiratory system compliance deteriorated to the same extent as in mice receiving HV^Tventilation alone (table 1). In contrast, treatment with Na²S attenuated the PIP increase, preserved respiratory system compliance, and reduced the downward shift of the pressure-volume curve (fig. 2).

In vehicle-treated animals, PaCO²was stable and PaO²decreased after HV^Tventilation (fig. 3). Compared with vehicle-treatment, Na²S treatment attenuated the decrease in PaO²in mice subjected to HV^Tventilation. At the end of the experiment, PaO²was higher in Na²S-treated animals than in vehicle-treated animals (157 ± 45 vs.  71 ± 14 mmHg, P = 0.0001).

BAL fluid protein concentrations were greater in vehicle-treated mice subjected to HV^Tventilation than in controls (fig. 4). In contrast, BAL fluid protein concentrations were lower in Na²S-treated mice than in vehicle-treated mice.

BAL fluid IL-6 concentrations were greater in vehicle-treated animals subjected to HV^Tventilation than in controls. Na²S treatment reduced BAL fluid IL-6 concentrations in mice subjected to HV^Tventilation (fig. 5A).

High tidal volume ventilation reduced the leukocyte concentration in BAL fluid obtained from vehicle-treated mice compared with that from controls. In contrast, Na²S treatment restored the concentration of leukocytes in BAL fluid of mice subjected to HV^Tventilation (fig. 5B).

The CXCL-2 and CD11b mRNA concentrations measured in the lungs of vehicle-treated animals subjected to HV^Tventilation (7- and 3-fold increase vs.  controls) were significantly attenuated by Na²S treatment (3- and 0.8-fold increase vs.  controls, fig. 6, A and B). In contrast, the pulmonary mRNA concentrations of L-selectin measured in vehicle-treated animals (1.5-fold increase vs.  controls) did not differ from those measured in Na²S-treated animals (1.3-fold increase vs.  controls, fig. 6C).

Total glutathione concentrations and the GSH/GSSG in lung tissues obtained from mice subjected to HV^Tventilation were significantly higher in Na²S-treated animals than in vehicle-treated animals (fig. 7). The increase of antioxidant NQO1 and GPX2 mRNA concentrations measured in the lungs of vehicle-treated animals subjected to HV^Tventilation (1.7- and 1.9-fold increase vs.  controls) was significantly enhanced by Na²S treatment (2.5- and 2.9-fold increase vs.  controls, fig. 8). Moreover, Na²S treatment prevented the reduction of GST-A4 mRNA concentrations measured in vehicle-treated animals subjected to HV^Tventilation.

Histologic analysis of lung sections from mice treated with either 60 ppm inhaled H²S or with injection of Na²S revealed that HV^Tventilation induced signs of edema formation and airway epithelial disruption (fig. 9). These pathologic signs were similar in mice ventilated with or without 60 ppm H²S. Fewer signs of edema and epithelial disruption were observed after Na²S pretreatment. The number of neutrophils in lung tissue sections taken from three mice subjected to HV^Tventilation was found to be greater than the number of lung tissue neutrophils determined in two control animals not subjected to HV^Tventilation (fig. 10). In addition, the numbers of lung tissue neutrophils found after HV^Tventilation in three mice with and three mice without inhalation of 60 ppm H²S were similar. In contrast, the number of neutrophils found in lung tissue sections taken from three mice subjected to HV^Tventilation after Na²S pretreatment was smaller. Despite the small sample size of tissue sections used for neutrophil staining, the variance of lung tissue neutrophil concentrations is small. The interpretation of these data must be made with great caution.

Additional figures describing the effect of various concentrations of inhaled H²S on BAL fluid leukocyte and IL-6 concentrations, pulmonary expression of leukocyte chemoattractant and adhesion molecules, and oxidative stress parameters and antioxidant gene expression are provided in Supplemental Digital Content 2, http://links.lww.com/ALN/A776, figures 1–4.

This study was performed to evaluate whether inhalation of H²S gas protects against lung injury from HV^Tventilation in mice. We found that inhalation of 1 or 5 ppm H²S produces neither beneficial nor deleterious effects, whereas inhalation of 60 ppm H²S accelerates the development of lung injury and death from HV^Tventilation. In contrast, treatment with a bolus infusion of Na²S attenuates pulmonary edema formation, thereby limiting the deterioration of respiratory system compliance and arterial oxygenation. Our results suggest that these protective effects of Na²S in VILI are linked to attenuated expression of chemoattractant and leukocyte adhesion molecules, enhanced expression of Nrf2-dependent antioxidant genes, and improved availability of reduced glutathione in the lung.

Our findings underscore the potential toxicity of H²S gas inhalation in a rodent model of VILI. Lopez et al.  reported cytotoxic effects and edema formation in respiratory tract tissues of rats breathing 10–400 ppm H²S.17The primary biochemical mechanism responsible for the toxicity of H²S gas is believed to involve the inhibition of cytochrome c oxidase.18Inhibition of cytochrome c oxidase blocks oxidative phosphorylation, a major source of cellular adenosine triphosphate synthesis. In turn, the abolishment of oxidative phosphorylation implies functional hypoxia (inability to use available oxygen for oxidative metabolism).18Both hypoxia and a lack of adenosine triphosphate are associated with pulmonary vasoconstriction, impaired alveolar fluid clearance, and pulmonary edema (reviewed in Bärtsch et al.  19). Based on these considerations, the inhalation of 60 ppm H²S is likely to have contributed to the formation of pulmonary edema in our murine model.

Faller et al.  reported that inhalation of 80 ppm H²S limited cytokine release in a model of lung injury in C57BL/6 mice caused by ventilation with a moderate tidal volume (12 ml/kg). Similarly, we found that BAL fluid IL-6 concentrations were lower when mice developed lung injury in the presence of 60 ppm H²S (fig. 5). However, the current study demonstrates that under conditions of higher mechanical lung stretch (peak airway pressure 35 cm H²O, as opposed to ∼12 cm H²O in the Faller et al.  study), the potential antiinflammatory effects of inhaled H²S are outweighed by the deleterious effects of inhaling 60 ppm H²S, including enhanced pulmonary expression of chemoattractant and leukocyte adhesion molecules and earlier deaths (figs. 1and 6). In addition, H²S inhalation prevented the up-regulation of antioxidant gene expression in mice subjected to HV^Tventilation (fig. 8) and caused a decrease in both the GSH/GSSG and total glutathione concentrations in the lung (fig. 7).

Mechanical stretching of the lung may initiate intravascular pulmonary leukocyte sequestration in mice.20Various chemoattractant and adhesion molecules that are derived from epithelial and endothelial cells, as well as alveolar macrophages and fibroblasts, are responsible for inflammatory recruitment and migration of leukocytes from the circulation into injured lung tissue.21In particular, the neutrophil chemoattractant CXCL-2 and the adhesion molecules CD11b and L-selectin are reported to play a critical role in the pathogenesis of VILI and for control of leukocyte trafficking into the lung.22,23Ultimately, pulmonary neutrophil activation and accumulation promotes microvascular permeability and edema formation.3,22Our results demonstrate that HV^Tventilation increases the pulmonary expression of CXCL-2, CD11b, and L-selectin. Moreover, the HV^T-induced increase in CXCL-2, CD11b, and L-selectin was enhanced markedly by concomitant inhalation of 60 ppm H²S (fig. 7). These findings suggest that the pathophysiologic changes leading to an earlier death in mice ventilated with 60 ppm H²S appear to be linked to enhanced expression of chemoattractant and leukocyte adhesion molecules in the lung.

As an alternative to the inhalation of potentially toxic H²S gas, we examined whether intravascular administration of Na²S can attenuate the development of VILI during HV^Tventilation. We found that systemic treatment with Na²S protects against VILI and reduces GSH/GSSG imbalances (fig. 7) in mice subjected to HV^Tventilation. Our results also suggest that Na²S attenuates the pulmonary expression of CXCL-2 and CD11b during VILI (fig. 6). This finding is supported by other reports demonstrating that both Na²S and sodium hydrogen sulfide can impair leukocyte adherence to endothelium and subsequent diapedesis during inflammation in vivo  by reducing the expression of endothelial and leukocyte adhesion molecules.24,25Increased pulmonary leukocyte adhesion in mice subjected to HV^Tventilation may also account for the low concentrations of BAL leukocytes reported here (fig. 5B) and elsewhere.26The protective effect of Na²S in this model of lung injury produced by HV^Tventilation is consistent with the beneficial effect of sodium hydrogen sulfide in other rodent models of lung injury induced by oleic acid27or a skin burn and smoke inhalation.28 

The current study demonstrates that Na²S pretreatment, but not H²S inhalation during HV^Tventilation, enhances the pulmonary expression of Nrf2-dependent antioxidant genes and stabilizes the concentration of reduced glutathione in the lung (fig. 6). A transcription factor for several antioxidant genes, Nrf2 is a master regulator of antioxidant response mechanisms (reviewed in Maher and Yamamoto29). Activation of Nrf2 results in the induction of many cytoprotective proteins. In the lung, Nrf2-dependent genes regulate the cellular redox status and have been reported to modulate pulmonary cellular responses and oxidative stress induced by mechanical ventilation.5The reasons inhaled H²S and intravenous Na²S have distinct effects in VILI in the current study are likely to be multifactorial. It is possible that pretreatment with Na²S before injurious ventilation is required to induce the Nrf2-dependent antioxidant defense mechanisms that contribute to protecting against VILI. It is also conceivable that intravascular administration of Na²S may preferentially target activation of Nrf2-dependent signaling in different cell types compared with inhaled H²S (e.g. , activating vascular endothelial vs.  airway epithelial cells). Analyzing the candidate cellular and molecular targets of H²S and Na²S in future studies may provide additional insights into their differential biologic effects.

In patients, ventilator-associated lung injury usually occurs in lungs with reduced compliance caused by preexisting conditions (two-hit or multifactorial lung injury), including septic inflammation. The current study uses a one-hit model of VILI in healthy mouse lungs. Thus, the current study has limited clinical relevance. Although the current study demonstrates that Na²S pretreatment protects against VILI, this study does not elucidate the potential dose-dependency of this protective effect and does not investigate the potential toxicity of Na²S to central or peripheral organs. Lung tissue H²S concentrations were not measured in this study. Comparing the lung tissue concentrations of sulfide species in future studies may help explain the differential biologic effects of inhaled H²S and intravascular Na²S and their dose-dependency.

Our study demonstrates that continuous inhalation of H²S gas enhances the expression of leukocyte adhesion and chemoattractant molecules (CXCL-2, CD11b, and L-selectin), accelerates pulmonary edema formation, and promotes lung injury when it is inhaled at high concentrations (60 ppm) during mechanical ventilation with a high tidal volume. The deleterious effects of H²S gas inhalation reemphasize that H²S can cause pulmonary toxicity and do not suggest a substantial role for H²S in the prevention and treatment of VILI in patients. In contrast, our data suggest that systemic intravascular treatment with Na²S may represent a novel therapeutic strategy to prevent both VILI and GSH/GSSG imbalance by activating Nrf2-dependent antioxidant gene transcription.

The authors thank their collaborators from the Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts: Rosemary C. Jones, Ph.D. (Associate Professor of Pathology), and Diane E. Capen and Yuko Beppu (Research Assistants) for their skillful assistance and advice in the preparation and interpretation of histologic lung sections, and Andrea U. Steinbicker, M.D., M.P.H., Matthias Derwall, M.D., and Kentaro Tokuda, M.D. (Research Fellows), for advice and technical assistance in quantitative real-time PCR techniques. The authors also thank Yasuko Nagasaka, M.D., Ph.D. (Instructor), for multifaceted advice and assistance with the study.