Genetic deletion of apolipoprotein A-I increases airway hyperresponsiveness, inflammation, and collagen deposition in the lung.

The relationship between high-density lipoprotein and pulmonary function is unclear. To determine mechanistic relationships we investigated the effects of genetic deletion of apolipoprotein A-I (apoA-I) on plasma lipids, paraoxonase (PON1), pro-inflammatory HDL (p-HDL), vasodilatation, airway hyperresponsiveness and pulmonary oxidative stress, and inflammation. ApoA-I null (apoA-I−/−) mice had reduced total and HDL cholesterol but increased pro-inflammatory HDL compared with C57BL/6J mice. Although PON1 protein was increased in apoA-I−/− mice, PON1 activity was decreased. ApoA-I deficiency did not alter vasodilatation of facialis arteries, but it did alter relaxation responses of pulmonary arteries. Central airway resistance was unaltered. However, airway resistance mediated by tissue dampening and elastance were increased in apoA-I−/− mice, a finding also confirmed by positive end-expiratory pressure (PEEP) studies. Inflammatory cells, collagen deposition, 3-nitrotyrosine, and 4-hydroxy-2-nonenal were increased in apoA-I−/− lungs but not oxidized phospholipids. Colocalization of 4-hydroxy-2-nonenal with transforming growth factor β-1 (TGFβ-1 was increased in apoA-I−/− lungs. Xanthine oxidase, myeloperoxidase and endothelial nitric oxide synthase were increased in apoA-I−/− lungs. Dichlorodihydrofluorescein-detectable oxidants were increased in bronchoalveolar lavage fluid (BALF) in apoA-I−/− mice. In contrast, BALF nitrite+nitrate levels were decreased in apoA-I−/− mice. These data demonstrate that apoA-I plays important roles in limiting pulmonary inflammation and oxidative stress, which if not prevented, will decrease pulmonary artery vasodilatation and increase airway hyperresponsiveness.

ture in a total volume of 50 µl. Rates of fl uorescence (Ex, 485 nm; Em, 530 nm) were determined over the next 2 h at 30 min intervals using a Spectra Max Gemini EM fl uorescence plate reader (Molecular Devices, Sunnyvale, CA).

Plasma PON1 arylesterase activity
Arylesterase activity of paraoxonase (PON1) was performed on whole plasma using phenyl acetate as the substrate as described ( 15 ). Initial rates of hydrolysis were determined spectrophotometrically at 270 nm on DU ® 640 spectrophotometer (Beckman Coulter ® Instruments, Brea, CA). An aliquot of 20 µl of 30× diluted mouse plasma was added to a fi nal reaction volume of 500 µl (phenyl acetate [100 mmol/l] and CaCl 2 [100 mmol/l] in Tris-HCl [40 mmol/l] buffer, pH 8.0) for 6 min at 25°C. Rates of spontaneous hydrolysis of phenyl acetate over the same time period were subtracted as blank. The extinction coeffi cient at 270 nm for phenyl acetate is 1310 mol•l One unit of arylesterase activity equals 1 µmole of phenyl acetate hydrolyzed per ml per min.

Quantifi cation of plasma nitrite+nitrate
Nitrite+nitrate concentrations were determined by ozone chemiluminescence using the NO Analyzer (Model 280i, GE Analytical-Sievers, Boulder, CO) as described ( 16,17 ). An aliquot of 30 µl was injected (plasma was diluted by 1:30) into a sealed glass reaction chamber at 95°C containing VCl 3 ( 17 ). Nitric oxide chemiluminescence signals were quantifi ed and peak areas compared with the areas of external nitrate standards. Results are expressed in M.

Estimates of plasma 3-nitrotyrosine
An aliquot of plasma from C57BL/6J and apoA-I Ϫ / Ϫ mice was pipetted onto nitrocellulose membranes and allowed to bind. Membranes were blocked with 5% nonfat dry milk dissolved in fresh PBS-Tween (0.1%) and then incubated overnight at 4°C with antibodies for 3-nitrotyrosine (3-NT, 1:5000; Millipore, Billerica, MA). The next day, the membranes were washed and incubated with the appropriate HRP-conjugated secondary antibody for 1 h. Bands of identity were visualized with ECL chemiluminescence (GE Healthcare, Piscataway, NJ) following the manufacturer's recommendations. Autoradiograms were scanned with a laser densitometer or a UMax scanner. Dot blot densities were quantifi ed using UN-SCAN-IT Gel 6.1 Software (Silk Scientifi c, Orem, UT).

Effects of genetic deletion of apoA-I on vasodilatation
Facialis and pulmonary arteries were isolated from C57BL/6J and apoA-I Ϫ / Ϫ mice by microdissection as previously described ( 14 ). Vasodilatation of pressurized (60 cm of H 2 O) facialis arteries was examined in the absence and presence of L-nitroargininemethylester (L-NAME; 200 µM, fi nal concentration) as previously described ( 14 ). Changes in pulmonary artery tension in response to acetylcholine (ACh) were recorded on a DMT wire-myograph using protocols similar to that previously described ( 18 ).

Measurements of lung mechanics
Mice (mean age = 15-16 weeks) were anesthetized with sodium pentobarbital (90 mg/kg) and placed on a heated surgical pad set at a constant temperature of 37°C. Mice were tracheostomized with an 18-gauge cannula and mechanically ventilated in a quasi-sinusoidal fashion with a small animal ventilator (fl exi-Vent, SCIREQ, Montreal, PQ, Canada) at 150 breaths per minute ( 19 ) and a tidal volume ( V T ) of 10 ml × kg Ϫ 1 body weight. To eliminate involuntary smooth muscle cell contraction during sickle cell anemia patients with pulmonary arterial hypertension (PAH) consistently had lower apoA-I levels than sickle patients without PAH ( 2 ). Interestingly, genetic deletion of endothelial lipase resulted in a nearly 2-fold increase in HDL, which was credited with decreasing airway hyperresponsiveness and pulmonary infl ammation in ovalbumin (OVA)-sensitized BALB/c mice ( 3 ). Although these reports provide some support for the idea that HDL helps maintain healthy lungs, no studies have directly determined the effects of apoA-I, or lack thereof, on pulmonary infl ammation, vasodilatation, and airway hyperresponsiveness. Increasing evidence suggests that elevated levels of HDL are not always atheroprotective (4)(5)(6)(7). Indeed, chronic states of infl ammation and oxidative stress have been shown to convert HDL from an anti-infl ammatory and anti-atherogenic lipoprotein into a pro-infl ammatory and pro-atherogenic lipoprotein, making it useless for protecting the vessel wall against the effects of atherogenic concentrations of LDL ( 8 ). Asthma also increases infl ammation and oxidative stress (9)(10)(11)(12). Such infl ammatory changes may explain why adult-onset asthma is associated with signifi cant increases in carotid artery intimal-medial thickness in women ( 13 ). In this article we examine the effects of genetic deletion of apoA-I on pulmonary infl ammation, vasodilatation, collagen deposition, and airway hyperresponsiveness. Our fi ndings suggest that apoA-I plays a critical role in protecting pulmonary artery and airway function as well as preventing infl ammation and collagen deposition.

Mice
Male apoA-I Ϫ / Ϫ (B6.129P2-Apoa1 tm1Unc /J; SN 002055, which is on a C57BL/6J genetic background) and C57BL/6J (SN 000664) mice were purchased from Jackson Laboratory (Bar Harbor, ME) at 8 weeks of age. Mice were housed in sterile autoclavable microisolation cages with 12 h dark and light cycles, free access to water, and standard chow diet. At 16 weeks of age, mice were either anesthetized for airway hyperresponsiveness or euthanized by intraperitoneal injection of Nembutal prior to collection of plasma, bronchial lavage fl uid (BALF), or tissues. The Institutional Animal Care and Use Committee (IACUC) of Medical College of Wisconsin approved all animal protocols used in the study.

Plasma total cholesterol, HDL cholesterol, and pro-infl ammatory HDL
Blood was collected in sodium heparin via left cardiac puncture and then centrifuged to harvest plasma, which was aliquoted and frozen at Ϫ 80°C until analysis. Total cholesterol (TC) was quantifi ed using a cholesterol oxidase/esterase kit from Wako Chemical, Inc. (Richmond, VA). HDL was isolated from whole plasma with a solution of dextran-sulfate-MgCl 2 (10 g/l, 0.5 M) (Berkeley HeartLab Inc., Alameda, CA), which precipitates apoBcontaining lipoproteins. HDL cholesterol was quantifi ed using a HDL Cholesterol E kit from Wako Diagnostics. Pro-infl ammatory HDL (p-HDL) was determined using a modifi ed method of a previously published cell-free assay ( 14 ). Briefl y, HDL was incubated with CuCl 2 (5 µmol/l, fi nal concentration) for 1 h at 37°C in a 384-well microtiter plate (MJ Research Inc., Waltham, MA). After incubation, 10 µl of 2 ′ ,7 ′ -dichlorodihdrofl uorescein (H 2 DCF) solution (0.2 mg/ml) was added to the HDL-Cu 2+ mix-eosin (H and E) for histology or with McLetchie's trichrome to assess collagen deposition.

Isolation of BALF and analysis
After exsanguination, the tracheas of C57BL/6J and apoA-I/ mice were cannulated with polyethylene tubing. BALF was obtained by fl ushing the lung with PBS, fi rst 1 ml followed by 0.5 ml. The rinses were combined, and the BALF was centrifuged at 2000 rpm for 10 min at 4°C. The supernatant was removed and stored at Ϫ 80°C. The cell pellet was gently resuspended in 1 ml PBS. An aliquot of 500 µl was used to prepare slides using a Cytospin (Model2; Shandon Scientifi c Co., Pittsburg, PA). Differential cell counts were made from slides stained with Diff-Quick. Five hundred cells were counted and identifi ed (magnifi cation 40×) for each mouse by a pathologist who had no prior knowledge of slide identities. The remainder of the cell suspension was centrifuged again, and the pellet was stored at Ϫ 80°C until further analysis.

Immunofl uorescence
Three C57BL/6J and apoA-I Ϫ / Ϫ mice were anesthetized and euthanized by exsanguination. The lungs were harvested, fi xed in zinc-formalin, and embedded in paraffi n. Sections were chosen at random by a technician who had no prior knowledge concerning slide identity. Immunofl uorescence was performed on 5 µm sections of paraffi n-embedded, PBS-zinc-formalin-fi xed lungs. Two sections were present on each slide. Sections were deparaffi nized with xylene and rehydrated in a descending alcohol row (from 95% to 50%). The sections were incubated separately with antibodies against 3-NT ( 24 ), anti-T15 autoantibodies (anti-T15 antibodies were prepared from hybridoma cells, which were kindly provided by Dr. J. F. Kearney, University of Alabama, Burmingham, AL) as previously described ( 25 ), anti-4-hydroxynonenal (4-HNE) Michael's adducts antibody (Calbiochem-EMD, LaJolla, CA), or with transforming growth factor ␤ -1 (TGF ␤ -1) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) for colocalization studies. Slides were washed with PBS (3×) and then incubated with the appropriate secondary antibodies. The slides were washed and sealed under cover slips. Images were captured using a krypton argon laser Nikon Eclipse TE2000U confocal microscope (Melville, NY) with 10×/0.17 aperture objective. Total magnifi cation was 100 with Ex/Em at 488/580 nm for Alexa 488 and 633/661 nm for TO-PRO-3. Images were captured (10/lung section; 2 sections per slide for a total of 60 images per test group) and analyzed using IBM EZC1 software (Armonk, NY). Controls for staining were slides incubated in the absence of primary antibodies.

Western blot analysis
Frozen lungs were pulverized in a stainless steel mortar and pestle that were prechilled with liquid nitrogen. The powder was quickly transferred into a 50 ml conical tube, homogenized in MOPS buffer with Polytron R (PT 1200E, Kinematica, Switzerland), and the homogenate was centrifuged at 13000 rpm (at 4°C for 10 min). Protein concentrations were quantifi ed using BCA reagent.
After one 1 min of regular ventilation with positive end-expiratory pressure (PEEP) set at 3 cm H 2 O, the lung volume history was standardized by two deep inspirations, delivered over 6 s at constant fl ow with a pressure limit of 30 cm H 2 O, followed by 5 min of regular ventilation. Basal respiratory system impedance ( Zrs ) was then assessed using the forced oscillation technique (FOT) applied over 2 s (a Prime-2 perturbation) using the fl exi-Vent system. Then, following a 60 s period of default ventilation, a series of challenges were performed at different levels of PEEP: 0, 3, 6, and 9 cm H 2 O. Each level of PEEP was held for 80 s with the Prime-2 perturbation repeated twice, once at 35 s and the other at 70 s Measurements at each PEEP were repeated twice, and the parameter estimates from each of the signal applications averaged and logged.

Methacholine challenge
Airway hyperresponsiveness was then assessed by administering series of progressively increasing concentrations of methacholine (MCh) to induce successive increases in bronchoconstriction. PEEP was set at 3 cm H 2 O for the MCh studies. Mice were challenged with 70 l of MCh (Sigma-Aldrich) of increasing concentrations (3.125-100 mg/ml). MCh aerosols were generated using an ultrasonic nebulizer (DeVilbiss LtraNeb 2000, Somerset, PA, USA) and delivered to the inspiratory line of the ventilator using room air. Each aerosol was delivered for 30 s, during which time ventilation was maintained mechanically by the fl exiVent. Immediately following MCh delivery, the Aeroneb nebulizer was quickly removed from the inspiratory arm of the fl exiVent system and default ventilation was reinstated. Then for the next 3 min, each aerosol was delivered for 30 s, during which time regular ventilation was maintained. A Prime 2-perturbation was applied every 20 s (by the fl exiVent) and changes in respiratory impedance ( Zrs ) in response to each MCh dose was determined using force oscillation techniques (FOT) generated by the fl exiVent.
Briefl y, the pressure and fl ow data obtained during Prime-2 perturbations were used by the fl exiVent to calculate Zrs , which was then fi t to a model consisting of a single airway serving a constant-phase viscoelastic tissue unit ( 19,20 ).
In this model equation, R N is Newtonian resistance composed mostly of the fl ow resistance of the conducting pulmonary airways, Iaw refl ects the inertance of the gas in the central airways (this can be ignored in mice below a breathing frequency of 20 Hz ( 21 ), G refl ects tissue resistance, viscous dissipation of energy in the parenchymal respiratory tissues ( 20 ), H refl ects elastic energy storage in the tissues (tissue stiffness), f is frequency i = -1 , and ␣ couples G and H in the following equation ( 20 ).
This model has been shown to accurately describe respiratory impedance both under control conditions and during mild bronchoconstriction (21)(22)(23). The advantage of this model is that it allows for a better distinction between central and peripheral airway changes in the lung.

Histology
A 0.5 ml aliquot of zinc-formalin was used to infl ate the lung prior to removal. The lung was fi xed in zinc-formalin, embedded in paraffi n, sectioned, and then stained with hematoxylin and

Effects of apoA-I defi ciency on vasodilatation
Studies on ACh-dependent vasodilatation of isolated and pressurized facialis arteries revealed that genetic loss of apoA-I had no effect on relaxation responses in these vessels ( Fig. 4A versus B). These observations are consistent with previous studies showing that apoA-I Ϫ / Ϫ mice are not more susceptible to atherosclerosis ( 27,28 ). However, relaxation responses of pulmonary artery (PA) rings isolated from apoA-I Ϫ / Ϫ mice were impaired at the highest and incubated with the appropriate HRP-conjugated secondary antibody for 1 h. Bands of identity were visualized with ECL chemiluminescence (GE Healthcare, Piscataway, NJ) following the manufacturer's recommendations. Autoradiograms were scanned with a laser densitometer or an UMax scanner. Band densities were quantifi ed using UN-SCAN-IT Gel 6.1 Software (Silk Scientifi c, Orem, UT).

BALF DCF-detectable oxidants
Oxidation of H 2 DCF was used to obtain an index of the levels of oxidants in BALF. An aliquot of BALF (75 µl) was mixed with H 2 DCF (100 µl PBS + 25 µl H 2 DCF [a 1 × 10 dilution of 2 mg/ml] in a total volume of 200 µl) and this mixture incubated for 2 h at 37°C. Absolute changes in fl uorescence (Ex 485 nm; Em 530 nm) were determined at the end of the 2 h incubation period using Spectra Max Gemini EM fl uorescence plate reader (Molecular Devices, Sunnyvale, CA).

Statistical analysis
Data are presented as mean ± SEM. Results were analyzed by Student's t -test, Mann-Whitney test, or Fisher's exact test where appropriate. Airway and pulmonary artery ring data were analyzed by 2-way ANOVA to determine signifi cance between curves and a Bonferroni posttest to determine signifi cance of points between the curves. All statistical analysis was performed using Graph Pad Prism Software (version 5.0).

Effects of apoA-I defi ciency on lipids and oxidative stress
ApoA-I Ϫ / Ϫ mice had reduced levels of plasma total cholesterol and HDL cholesterol compared with control mice ( Fig. 1A , B ). Although apoA-I Ϫ / Ϫ mice contained less HDL than control mice, their HDL oxidized at a faster rate than HDL from control mice, indicating that HDL in apoA-I Ϫ / Ϫ mice was pro-infl ammatory ( 26 ) ( Fig.  1C ). PON1 plasma protein in a poA-I Ϫ / Ϫ mice was increased compared with protein levels in C57BL/6J mice ( Fig. 2A , B ). However, PON1 activity was decreased ( ≈ 34%, P < 0.001) compared with controls ( Fig. 2C ).
Although the concentration of plasma nitrite+nitrate was decreased in apoA-I Ϫ / Ϫ mice, plasma 3-NT levels were increased compared with the levels in control mice ( Fig.  3A and B , respectively).  physiological responses of pulmonary arteries in apoA-I Ϫ / Ϫ mice are consistent with the idea that chronic exposure to infl ammation and oxidative stress impair vasodilatation.

Effects of apoA-I defi ciency on airway hyperresponsiveness
The fl exiVent provides quantitative information regarding the mechanical properties of the entire airway tree. In the large airways it assesses Newtonian resistance (R N ), while in the smaller airways it is able to determine changes in tissue damping (G) and tissue elastance (H). In both apoA-I Ϫ / Ϫ and C57BL/6J mice, methacholine (MCh) induced dose-dependent increases in airway resistance associated with tissue dampening (G) ( P < 0.01; Fig. 5B ) and tissue elastance (H) ( P < 0.001; Fig. 5C ), but not R N ( Fig.  5A ). Effects of increasing PEEP on airway mechanical parameters, R N , G and H, are shown in Figs. 5D, 4E, and 4F , respectively. G was increased in apoA-I Ϫ / Ϫ mice compared with control mice at baseline PEEP = 0 ( P < 0.01) as well as throughout the entire PEEP curve ( P < 0.001). H was increased in apoA-I Ϫ / Ϫ mice compared with controls throughout the entire PEEP range ( P < 0.001).

Effects of apoA-I defi ciency on histology
H and E sections of lungs harvested from a poA-I Ϫ / Ϫ mice ( Fig. 6A , two lower-left images) contained twice the number of leukocytes than lungs from C57BL/6J mice ( Fig. 6A, B , one upper-left image, P < 0.025, 2-tailed Student t -test). The cell counts in sections of lung isolated from apoA-I Ϫ / Ϫ mice were composed predominantly of neutrophils, leukocytes, and eosinophils. In contrast, the cell counts in sections of lungs isolated from C57BL/6J mice contained an occasional neutrophil. Although infl ammatory cells in the lungs are increased by the absence of apoA-I, the number of pro-infl ammatory cells per high powered fi eld in these mice is much lower than that in OVA-sensitized mice. OVA sensitization often increases leukocyte infi ltration in the lung as high as 25-50 cells per HPF ( 29 ). Sections of lungs harvested from ACh concentration ( Fig. 4C ). Two-way ANOVA (ANOVA) was performed on these data to determine whether mouse strain had any infl uence. Although on a concentration basis, the major difference in the curves was at the highest ACh concentration, two-way ANOVA indicated that that the strain of mouse (C57BL/6J versus apoA-I Ϫ / Ϫ mice) infl uenced the shape of the curves and that the curves were "extremely" signifi cantly different ( P < 0.0004). The importance of this analysis is that there is less than a 0.044% chance of these curves occurring by random events. The sudden decrease in vasodilatation by the apoA-I Ϫ / Ϫ PA rings means that the apoA-I Ϫ / Ϫ PA rings are actually contracting in response to the highest dose of ACh. In contrast, PA rings isolated from control mice continued to vasodilate. This change in vasodilatory response is an important phenotypic change and, to our knowledge, the fi rst demonstration that genetic loss of apoA-I specifi cally alters pulmonary vascular physiology. Such changes in the  Michael's adducts than sections from controls ( Fig. 7A , B ). In contrast to these two biomarkers of oxidative stress, no differences were observed for immunofl uorescent staining for T15-type autoantibodies between apoA-I Ϫ / Ϫ and control mice ( Fig. 7A ). Genetic loss of apoA-I increases colocalization of 4-HNE-derived Michael's adducts with active TGF ␤ -1 in pulmonary airways ( Fig. 7B ). Lungs from apoA-I Ϫ / Ϫ mice express higher levels of XO, MPO, and eNOS than lungs from C57BL/6 mice ( Fig. 8 ). These data are in contrast to nitrite+ nitrate data showing that BALF isolated from apoA-I Ϫ / Ϫ mice contains less nitrite+nitrate than BALF isolated from C57BL/6J mice ( Fig. 9A ). Finally, BALF from apoA-I Ϫ / Ϫ mice increased DCF fl uorescence to a greater extent than BALF from C57BL/6J mice ( Fig. 9B ). These data indicate that BALF from apoA-I Ϫ / Ϫ mice contains more pro-oxidant compounds than BALF from C57BL/6J mice.

Effects of apoA-I defi ciency on biomarkers of oxidative stress and infl ammation
Immunofl uorescence studies revealed that lung sections from apoA-I Ϫ / Ϫ mice stained stronger for 3-NT and 4-HNE Deletion of apoA-I altered plasma lipoprotein profi les by decreasing total cholesterol and HDL cholesterol ( Fig.  1A, B ), confi rming previous reports ( 27,28 ). Although it has long been known that HDL is decreased in apoA-I Ϫ / Ϫ mice, we report here that the levels of pro-infl ammatory HDL are also increased ( Fig. 1C ). This observation is consistent with our previous report examining the effects of a genetic loss of apoA-I on pro-infl ammatory HDL in Ldlr Ϫ / Ϫ /apoA-I Ϫ / Ϫ mice relative to levels in Ldlr Ϫ / Ϫ mice ( 14 ). The cell-free assay for quantifying pro-infl ammatory HDL is a convenient means for determining whether HDL isolated from a group of patients, different strains of mice or experimental test groups is more easily oxidized than HDL isolated from controls. We and others have shown that pro-infl ammatory HDL levels are increased in diseases characterized by chronic states of oxidative stress ( 25,30,31 ). This increase in pro-infl ammatory HDL in the apoA-I Ϫ / Ϫ mice coincides with a decrease in PON1 activity even though PON1 protein levels were actually increased ( Fig.  2 ). The lower PON1 activity levels in the apoA-I Ϫ / Ϫ mice are consistent with the fact that PON1 can be inactivated by chronic increases in oxidative stress ( 32 ). These data provide additional evidence that apoA-I Ϫ / Ϫ mice experience more oxidative stress than C57BL/6J mice. Although plasma from the apoA-I Ϫ / Ϫ mice contained less nitrite+nitrate, than plasma from C57BL/6J mice, it also contained a modest, but statistically signifi cant increase in 3-nitrotyrosine (3-NT) ( Fig. 3 ). This signifi cant increase in 3-NT, a biomarker of oxidative stress, provides a third level of support for the idea that apoA-I Ϫ / Ϫ mice experience greater oxidative stress than C57BL/6J mice. Interestingly, although apoA-I Ϫ / Ϫ mice had lower levels of HDL and increased plasma 3-NT, peripheral vascular function in apoA-I Ϫ / Ϫ mice was unaffected compared with controls ( Fig. 4 ). We have shown before that ACh-dependent vasodilatation of facialis arteries is endothelial nitric oxide synthase-dependent ( 14 ). The observations here are consistent with previous studies showing that apoA-I Ϫ / Ϫ mice are not more susceptible to atherosclerosis ( 27,28 ). Although genetic loss of apoA-I did not impair vasodilatation in facialis arteries ( Fig. 4A, B ), it did alter the relaxation responses of pulmonary artery rings ( Fig. 4C ). At 10 Ϫ 4 M ACh, C57BL/6J pulmonary rings continue to vasodilatate; however, rings from apoA-I Ϫ / Ϫ mice actually constrict when treated with 10 Ϫ 4 M ACh. These data are consistent with the idea that the oxidative stress and/or infl ammation associated with loss of apoA-I impairs pulmonary vascular function ( 25,33 ). Genetic loss of apoA-I also increased airway hyperresponsiveness, collagen deposition, and biomarkers of oxidative stress in the lungs. Methacholine is a bronchoconstrictor used clinically to assess airway responses in patients suspected of having asthma. Inspiration of nebulized methacholine increases bronchoconstriction which in turn increases airway resistance. Measurements of airway resistance after inspiration of methacholine provide a sensitive means of determining whether airway physiology is impaired. Using force oscillatory techniques, changes in airway resistance can be assessed not only in the central

DISCUSSION
The role of apoA-I in pulmonary infl ammation and airway hyperresponsiveness was investigated in mice that were genetically engineered to be defi cient in apoA-I. Genetic loss of apoA-I in mice resulted in changes in lung histology and airway physiology that are consistent with the idea that apoA-I plays an important anti-infl ammatory role in the lung. airways (R N ) but also in the lung parenchyma with respect to tissue dampening (G) or elastance (H). Our studies showed that genetic loss of apoA-I had little effect on central airway resistance (R N ), but it did increase resistance with respect to G and H ( Fig. 5 ). G refl ects changes in either the physical properties of pulmonary tissues or regional heterogeneity related to tissue damping ( 19 ). H is a parameter indicative of lung elastance or the ability of the lungs to recoil after expansion resulting from inspiration. Accordingly, an increase in G would most likely refl ect airway closure due to liquid bridges forming within the small airways and alveoli (contributing to increased G), whereas chronic changes in H would most likely refl ect changes in the intrinsic mechanical properties of the lung parenchyma ( 19,(34)(35)(36). Increases in G and H in response to a bronchoconstrictor are consistent with our histology studies showing that collagen deposition is greater in apoA-I Ϫ / Ϫ mice than C57BL/6J mice ( Fig. 6 ). Histological studies revealed that there were more infl ammatory cells in the lungs of apoA-I Ϫ / Ϫ mice than in C57BL/6J mice. The number of cells in the lungs of these mice is low in comparison to ovalbumin-sensitized mice in which very high numbers can be achieved (unpublished observations). The composition of the infl ammatory cells in the lungs of apoA-I Ϫ / Ϫ mice was predominantly neutrophils and leukocytes, whereas in C57BL/6J mice an occasion neutrophil was noted. Finally, although there was a tendency for leukocytes to be increased in BALF of apoA-I Ϫ / Ϫ mice, statistical analysis indicated that the difference between the two groups was not signifi cant ( Fig. 6 ). Performing forced oscillatory techniques at different PEEP levels to determine R N , G, and H has been shown to be a sensitive means of detecting differences in airway mechanics in response to increasing pressure between different strains of mice ( 19 ). We observed no differences in R N between control and apoA-I Ϫ / Ϫ mice with increasing PEEP; however, G and H were increased in the apoA-I Ϫ / Ϫ mice throughout the entire PEEP range compared with C57BL/ 6J mice ( Fig. 5 ). As R N corresponds to central airways resistance, a change in R N is expected only if the caliber of the conducting airways is signifi cantly reduced. Thus, our data suggest that the increases in G and H represent inherent changes in the biophysical properties of very small airways or lung parenchyma in the apoA-I Ϫ / Ϫ mice. Again, this fi nding is consistent with histological data showing that lungs from apoA-I Ϫ / Ϫ mice contain greater levels of collagen than C57BL/6J mice ( Fig. 6 ).
Immunofl uorescence studies of the lung clearly indicated that the lungs of apoA-I Ϫ / Ϫ mice were under greater oxidative and nitrosative stress and infl ammation than lungs of control mice. Lungs isolated from apoA-I Ϫ / Ϫ mice contained increased levels of 3-NT, an index of either peroxynitrite formation or nitrogen dioxide production ( 37-39 ) and 4-HNE Michael's adducts, an index of lipid peroxidation with respect to polyunsaturated fatty acid chain breakage ( 40 ), but not T15-autoantibodies, which detect the presence of oxidized phosphotidylcholine ( 41,42 ). The marked increase in 4-HNE Michael's adducts is important in that, for the most part, it colocalized activation of TGF ␤ -1, which plays important roles in mesenchymal transition to generate fi broblasts and increase collagen deposition. Accordingly, our data begin to explain why  sponsiveness. Loss of apoA-I and its associated anti-infl ammatory and anti-atherogenic properties has a profound and severe negative impact on the lung with respect to vascular function and airway physiology. Additional studies aimed at determining how apoA-I prevents infl ammation and oxidative stress in the lung should reveal new insight into the cellular mechanisms by which apoA-I protects the lung.
The authors wish to thank Maria Zaidi for technical assistance.
genetic loss of apoA-I, and possibly apoA-I dysfunction, increases collagen deposition and airway stiffening. The lungs of apoA-I Ϫ / Ϫ mice are under greater oxidative stress than the lungs of control mice. Western blot analysis revealed that homogenates of apoA-I Ϫ / Ϫ lungs contained greater levels of XO, MPO, and eNOS than homogenates of control lungs ( Fig. 8 ). Where cholesterol-rich diets ( 43 ) or ischemia/reperfusion ( 44,45 ) are well recognized for increasing injury and, therefore, XO release from the liver, we observed that apoA-I deficiency induces a nearly 2-fold increase in XO expression in lung homogenates without apparent injury to the liver (data not shown). Our data indicate that pulmonary injury and subsequent XO generation and release can and does occur throughout the lung. Others have shown that the endothelium and epithelium are important sources of XO activity ( 46 ). For example, exposure of endothelial cells to hypoxia/reoxygenation increases both the conversion of XDH to XO and XO activity nearly 2-fold ( 47 ). Further, rhinovirus infection, a common mechanism for aggravating viral-induced asthma, increases superoxide anion generation in primary bronchial respiratory epithelial cells by proteolytic conversion of XDH into XO ( 48 ). Thus, several mechanisms exist to increase XO in pulmonary tissues. Likewise, increased MPO expression in the lungs of apoA-I Ϫ / Ϫ mice, presumably a result of neutrophil or leukocyte recruitment, is well recognized for increasing oxidative damage, injury, and tyrosine nitration in infl amed lungs. Finally, although one might assume that an increase in eNOS expression is protective because of increased potential for generating •NO, it has been shown that overexpression of eNOS in atherogenic mice actually increases lesion formation ( 49 ). Thus, while an increase in eNOS should be protective, the fact that lungs of apoA-I Ϫ / Ϫ mice are subjected to increased states of oxidative stress and infl ammation suggests that pulmonary eNOS in apoA-I Ϫ / Ϫ mice might be uncoupled. Data supporting this notion are the decreased nitrite+nitrate concentrations in BALF in apoA-I Ϫ / Ϫ mice, the marked increases in 3-NT and 4-HNE Michael's adducts that colocalize with TGF ␤ -1 in the lungs of apoA-I Ϫ / Ϫ mice ( Fig.  7 ), and ultimately, the marked increase in DCF-detectable oxidants in the BALF isolated from apoA-I Ϫ / Ϫ mice ( Fig. 9 ). Quantifi cation of free and esterifi ed F2-isoprostanes in lung homogenates suggest that genetic loss of apoA-I increases the formation of F2-isoprostanes by 20-52% over and above the levels in C57BL/6J mice (data not shown). Numerous studies in other systems indicate that such conditions are a common prescription for uncoupled eNOS activity, a potential mechanism that will be examined in greater detail in future studies.

CONCLUSION
Our studies demonstrate that apoA-I plays an important protective role in preventing pulmonary infl ammation, impaired vasodilatation, and increased airway hyperre-