Glutathione-S-transferase P protects against endothelial dysfunction induced by exposure to tobacco smoke

Abstract

Exposure to tobacco smoke impairs endothelium-dependent arterial dilation. Reactive constituents of cigarette smoke are metabolized and detoxified by glutathione-S-transferases (GSTs). Although polymorphisms in GST genes are associated with the risk of cancer in smokers, the role of these enzymes in regulating the cardiovascular effects of smoking has not been studied. The P isoform of GST (GSTP), which catalyzes the conjugation of electrophilic molecules in cigarette smoke such as acrolein, was expressed in high abundance in the mouse lung and aorta. Exposure to tobacco smoke for 3 days (5 h/day) decreased total plasma protein. These changes were exaggerated in GSTP^−/− mice. Aortic rings isolated from tobacco smoke-exposed GSTP^−/− mice showed greater attenuation of ACh-evoked relaxation than those from GSTP^+/+ mice. The lung, plasma, and aorta of mice exposed to tobacco smoke or acrolein (for 5 h) accumulated more acrolein-adducted proteins than those tissues of mice exposed to air, indicating that exposure to tobacco smoke results in the systemic delivery of acrolein. Relative to GSTP^+/+ mice, modification of some proteins by acrolein was increased in the aorta of GSTP^−/− mice. Aortic rings prepared from GSTP^−/− mice that inhaled acrolein (1 ppm, 5 h/day for 3 days) or those exposed to acrolein in an organ bath showed diminished ACh-induced arterial relaxation more strongly than GSTP^+/+ mice. Acrolein-induced endothelial dysfunction was prevented by pretreatment of the aorta with N-acetylcysteine. These results indicate that GSTP protects against the endothelial dysfunction induced by tobacco smoke exposure and that this protection may be related to the detoxification of acrolein or other related cigarette smoke constituents.

extensive evidence has demonstrated that cigarette smoking (25) or exposure to secondhand tobacco smoke (4) increases the risk for cardiovascular disease (CVD). Smoking causes an estimated 1.69 million deaths per year world wide (17). Smoking is strongly and positively associated with an increase in the risk for myocardial infarction and sudden cardiac death and an increase in thrombosis and atherosclerosis (39). Smoking also induces endothelial dysfunction, which is a characteristic feature of CVD. Nevertheless, pathophysiological mechanisms that mediate and exacerbate cardiovascular injury due to exposure to tobacco smoke remain poorly understood. In particular, the role of individual constituents of tobacco smoke in promoting CVD is not known, and it is unclear whether metabolic pathways for the detoxification of chemicals in cigarette smoke (CS) modulate excessive CVD risk due to exposure to tobacco smoke.

CS contains more than 4,000 chemicals, but most of these are present in vanishingly small concentrations (39). In addition to nicotine, some of the most toxic CS constituents present in high abundance include carbon monoxide, saturated and unsaturated aldehydes, vinylpyridine, hydrogen cyanide, particulate matter, and polyaromatic hydrocarbons (PAHs). Of these, unsaturated aldehydes, such as acrolein, are likely to be particularly important because of their high reactivity and high cardiovascular toxicity (8). In most tissues, unsaturated aldehydes, such as acrolein and crotonaldehyde, are metabolized via conjugation by glutathione, a reaction that is catalyzed by glutathione-S-transferases (GSTs). The GSTs are a large family of enzymes [GST isoform A, GST isoform M (GSTM), and GST isoform P (GSTP), among others] that catalyze the conjugation of a variety of electrophilic xenobiotics and have been implicated in the development of tumor resistance to anticancer drugs (23) and cellular signaling.

Results of several epidemiological studies have suggested a link between polymorphism in GST genes and the risk of cancer due to smoking (38). The GSTM-null phenotype (16) and polymorphisms in the GSTP gene (11) have both been found to be associated with the elevated risk of bladder cancer in smokers. The association between the GST genotype and lung cancer is less clear, because both positive and negative data have been reported (34, 38). Similarly, in some studies, the GSTM-null phenotype was associated with an increase in coronary artery disease in smokers (33), whereas in others the null phenotype has been reported to be associated with a decrease in risk of myocardial infarction in smokers (46). Although GSTM participates in the metabolism of several PAHs present in CS, many of the small reactive carbonyls present in CS, such as acrolein and crotonaldehyde, are preferentially conjugated by GSTP (8). Nonetheless, the role of GSTP in the cardiovascular effects of CS has not been studied. Accordingly, the present study was designed to examine whether GSTP regulates smoking-induced endothelial dysfunction. In humans, passive smoking diminishes endothelium-mediated relaxation, and long-term smoking is associated with impaired endothelium-dependent relaxation of conduit and coronary arteries (13, 14). Tobacco smoke exposure also decreases endothelium-dependent relaxation in rats and rabbits (29, 30) and increases arterial stiffness in mice (20). The results of our study show that deletion of the GSTP gene increases tobacco smoke-induced endothelial dysfunction and exacerbates acrolein-induced vascular injury, suggesting that GSTP protects against smoking-induced endothelial dysfunction potentially by promoting the detoxification of acrolein and related electrophilic constituents of tobacco smoke.

MATERIALS AND METHODS

Mice.

GSTP1/P2-null mice were obtained from C. Henderson and R. Wolf (University of Dundee), in which both GSTP genes were knocked out using a single construct (24). These mice grow and breed normally. Mice were treated according the American Physiological Society's Guiding Principles in the Care and Use of Animals, and all protocols were approved by the University of Louisville Institutional Animal Care and Use Committee.

Exposure to tobacco smoke or acrolein.

Male wild-type (WT) and GSTP-null mice (12–16 wk old) were exposed to air (control) or to tobacco smoke for 1 or 3 days (5 h/day) and were killed either immediately after exposure (1 day) or at 16 h after the last treatment (3 days). Tobacco smoke was generated from Kentucky 2R4F reference cigarettes (Tobacco Research Institute, University of Kentucky, Lexington, KY), which have a declared content of 2.45 mg nicotine each, with a 23-mm butt remaining after smoking. These cigarettes have a higher total content of acrolein and crotonaldehyde than 1R4F cigarettes (15). Cigarettes were kept in standardized atmosphere humidified with 70% glycerol and 30% water for 48 h before use. Mice were exposed to tobacco smoke using a smoke chamber (model TE-10, Teague Enterprises, Woodland, CA). A mixture of sidestream (89%) and mainstream (11%) CS was used in tobacco smoke exposures. Each smoldering cigarette was puffed for 2 s once every minute for a total of 8 puffs at a flow rate of 1.05 l/min to provide a standard puff of 35 cm³. Ten 2R4F cigarettes were burned at one time for 5 h continuously per day. The total suspended particulate level was 88.1 ± 2.5 mg/m³ estimated from five separate 3-day exposures.

Acrolein atmospheres were generated from liquid acrolein (Sigma, >90%) diluted in distilled H₂O (1:10) in a custom vapor system (Teague Enterprises) using a primary chamber as a constant source. Acrolein vapors were diluted with high-efficiency particulate air-filtered room air in a secondary chamber. The acrolein exposure concentration was continuously monitored using an in-line photoionization detector (ppbRAE+, Rae Industries, Sunnyvale, CA) before delivery via a cage insert vapor delivery unit (Teague Enterprises) into a standard polycarbonate rat cage (16 × 8.75 × 13.5 in., ∼31 liters) for exposures. Air or acrolein was distributed through a fine mesh screen at 3 l/min by delivery units with a cyclone-type top that distribute air within 10% of the mean concentration at six locations in the cage. Exposure cages were placed partially over heating pads (∼71°F) to allow mice to select their preferable temperature. In the first protocol, mice were exposed to 5 ppm acrolein for 5 h (4,944 ± 44 ppb of 4 different exposures) and in the other protocols to 1 ppm acrolein for 3 days (1,053 ± 22 ppb of 9 different exposures).

After exposure, mice had free access to food (not during exposure) and water, after which they were killed with pentobarbital sodium (0.1 ml, 40 mg/ml ip). Blood was collected via cardiac puncture in Na₄·EDTA (0.2 M, 16 μl/ml blood)-containing tubes. Plasma protein and albumin levels were determined using Bradford and bromocresol green reagents, respectively (Wako). Plasma alanine transferase, aspartate transferase (Infiniti), creatine kinase, and lactate dehydrogenase (Promega) levels were measured using commercially available assay reagents in 96-well plates or with a Cobas Mira Plus 5600 Autoanalyzer (Roche). Organs (i.e., heart, kidney, liver, and lung) were weighed and then snap frozen in liquid nitrogen for Western blot or GST activity analyses and/or pieces of each were formalin fixed (10% neutral buffered formalin) and processed for histology and immunohistochemistry. Total plasma high-density lipoprotein (HDL), low-density lipoprotein (LDL), cholesterol, triglycerides, and phospholipids were determined using a Cholesterol CII Enzymatic Kit (Wako) and L-Type TG-H Kit (Wako) using calibrated standards and a Cobas Mira Plus 5600 Autoanalyzer.

Isolated aorta and vascular reactivity.

Mice were anesthetized with pentobarbital sodium (0.1 ml, 40 mg/ml ip) and given heparin (200 units, 0.2 ml ip), and the aorta was removed via a midventral thoracotomy. Thoracic aorta segment ends were trimmed, one 4-mm (tobacco smoke and acrolein experiments) or four 3-mm (naïve mice) rings were cut from each aorta, and rings were hung on stainless steel hooks in 15-ml water-jacketed organ baths in physiological salt solution (PSS) bubbled with 95% O₂ and 5% CO₂ at 37°C. The composition of PSS was (in mM) 130 NaCl, 4.7 KCl, 1.17 MgSO₄·7H₂O, 1.18 KH₂PO₄, 14.9 NaHCO₃, 2.0 CaCl₂, and 5.0 glucose (pH 7.4). One hook was connected to an isometric strain-gauge transducer (Kent Scientific, Litchfield, CT), and the other hook was attached to a fixed support glass rod. Transducer signals were fed into an eight-channel PowerLab analog-to-digital converter and recorded on a personal computer using Chart software (version 3.4.9, iWorx, Dover, NH). After 10 min without tension, rings were equilibrated to ∼1 g of loading tension over 30 min. Aortic rings were stimulated with 100 mM K⁺-PSS (100 mM K⁺) to test for viability, washed three times with PSS over 30 min, reequilibrated to ∼1 g of resting tension, and then restimulated with 100 mM K⁺ followed by three bath changes and reequilibration to ∼1 g. Rings from in vivo CS or acrolein exposures were contracted with cumulative concentrations of phenylephrine (PE; 0.1 nM–1 μM) and then relaxed with cumulative concentrations of ACh (0.1 nM–1 μM) to determine endothelium-dependent relaxation. After a tension plateau, sodium nitroprusside (SNP; 100 μM) was added to determine endothelium-independent relaxation. Vessel contractions were quantified as raw (mg tension) or normalized as a percentage of the maximum PE contraction (in vivo exposure) or of the pretreatment PE contraction [i.e., postacrolein PE tension/preacrolein PE tension (in%), in vitro exposure]. Relaxation was calculated as the percent reduction of PE-induced tension. The effective concentration producing a 50% response (EC₅₀) was determined by normalizing cumulative concentration responses to 100%, plotting the response versus log [M] of the agonist, and interpolating the EC₅₀ value.

Western blot analysis.

SDS-PAGE and Western blot analysis were performed with aorta and lung homogenates and plasma (100 μg) with mouse monoclonal antibody against human GSTP1 (Invitrogen, 1:2,500) or IgG-purified rabbit anti-Keyhole Limpet hemocyanin-acrolein polyclonal antibody (1:1,000). Western blot analysis was also performed using whole lung homogenates for aldehyde dehydrogenase (ALDH)3A1, FGF-regulated protein (FR)-1, heme oxygenase (HO)-1, and NAD(P)H:quinone oxidoreductase 1 (NQO1) using in house-raised polyclonal or commercially available mouse monoclonal primary antibodies (1:1,000, HO-1, Stressgen). Additionally, hepatic microsomes were probed for NQO1 and cytochrome P-450 isoforms (1A1/1A2, 2B1, and 2E1) using polyclonal in house-raised antibodies (1:1,000). Western blots were developed using appropriate secondary antibodies and ECL plus reagent (Amersham Biosciences), and band intensity was detected with a Typhoon 9400 variable mode imager (Amersham Biosciences). Quantification of band intensity was performed using Image Quant TL software (Amersham Biosciences), and bands were normalized to actin or amido black staining. Relative protein levels were calculated as fold increase of the appropriate air-exposed group.

GST activity and immunohistochemistry.

Total GST-conjugating activity toward a general substrate, 1-chloro,2,4-dinitrobenzene (CDNB; 1 mM), and a GSTP-selective substrate, ethacrynic acid (EA; 200 μM), was determined in lung and aorta homogenates according to Habig et al. (21). Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue sections (4 μm) stained with rabbit polyclonal primary antibody against human GSTP1 (1:1,500, Novo Castra) (IgG-purified preimmune rabbit serum served as negative control) (19). The secondary antibody was an anti-rabbit goat antibody with a Vector Elite staining kit using diaminobenzadine (Dako) as the chromagen.

Statistical analysis.

Data are reported as means ± SE. For comparing two groups, an unpaired Student's t-test or Mann-Whitney rank sum test was used. Comparisons among multiple groups or between two groups at multiple time points were performed by one-way ANOVA with Bonferroni's post test where appropriate. Statistical significance was accepted at P < 0.05.

RESULTS

GSTP localization and abundance.

The GST gene superfamily is composed of several members that show tissue-specific distribution (1, 47). To examine whether GSTP was expressed in the lung and aorta, immunohistochemistry and Western blot analysis were performed. High levels of GSTP protein were detected in the mouse aorta and lung (Fig. 1). In the aorta, positive GSTP staining was associated with areas also stained by anti-von Willebrand factor (data not shown), although intense staining was also associated with smooth muscle cells of the media (Fig. 1A). No staining was detected in vessels from GSP-null mice (Fig. 1A). Positive staining for GSTP protein was also observed in the WT lung, where it was highly localized to lung airway columnar epithelium (Fig. 1A). GSTP protein was variably expressed in several other organs as well. The highest levels of GSTP were expressed in the liver, aorta, stomach, and bladder (Fig. 1B).

Fig. 1.

Tissue abundance of glutathione-S-transferase (GST) isoform P (GSTP) protein. A: lung and aortic distribution of GSTP protein in male naïve wild-type (WT) and GSTP-null mice. Positive immunohistological staining was present in the WT lung (most intense in the pseudostratified columnar epithelium nuclei, arrows) and aorta but was absent in GSTP-null tissues. B: Western blots with anti-GSTP antibody using lysates obtained from the aorta, urinary bladder, stomach, small intestine, lung, liver, kidney, and heart from WT (+) and GSTP-null (−) mice. C: catalytic activity of GST in the mouse lung and aorta. Total GST activity was measured using 1-chloro-2,4-dinitrobenzene (CDNB) in the lung (n = 3 WT and 3 null) and whole aorta (n = 4 WT and 4 null) homogenates of WT and GSTP-null mice. *P < 0.05.

To determine the contribution of GSTP to total GST activity, GST activity was measured in lysates prepared from the mouse lung and aorta (pooled from 3 mice). Total GST activity, measured using CDNB, in the lung was 27% of that obtained with the liver. Deletion of GSTP led to a <15% decrease in CDNB activity, suggesting that GSTP accounts for only a fraction of the total GST activity in the lung. In contrast, WT aortic GST activity was half that present in the lung and was twice that in the aorta of GSTP-null mice (Fig. 1C). Collectively, these results show GSTP is expressed in the lung and the aorta, and, although GSTP contributes to a small, but significant, fraction of the total tissue GST activity in the lung, it comprises a majority of the GST activity in the murine aorta.

Tobacco smoke toxicity.

To understand the role of GSTP in tobacco smoke toxicity, mice were exposed to a mixture of main- and sidestream smoke for 3 days, and plasmatic changes were measured to assess systemic toxicity. As shown in Table 1, no increase in liver or muscle enzymes was observed, indicating that the exposure protocol did not induce overt toxicity. A decrease in total cholesterol was observed, which was due to a small decrease in HDL. There was also a slight decrease in blood glucose in tobacco smoke-exposed mice. GSTP-null mice showed no changes in glucose and cholesterol levels. Levels of HDL and LDL in tobacco smoke-exposed GSTP-null mice were similar to air-exposed mice, which indicates that deletion of the GSTP gene attenuates tobacco smoke-induced changes in blood glucose and cholesterol (Table 1). Even though no change in total protein was observed with WT mice, it was significantly decreased in GSTP-null mice.

Table 1.

Blood parameters in WT and GSTP-null mice exposed to air or CS

	WT Mice		GSTP-Null Mice
	Air	CS	Air	CS
Blood glucose, mg/dl	226±8	167±6*	199±9	186±15
Hct (%)	45.9±0.3	49.3±0.9*	47.9±0.6	47.8±1.4
Buffy Coat (%)	1.4±0.3	1.5±0.2	1.3±0.1	1.3±0.2
Cholesterol, mg/dl	117±4	101±6*	97±4	90±2
HDL, mg/dl	88±3	69±5*	73±3	68±2
LDL, mg/dl	23±2	29±3	18±1	15±1
Triglycerides, mg/dl	38±5	20±4*	28±3	27±3
TP, g/l	4.99±0.18	4.97±0.12	4.81±0.06	4.61±0.05*
Albumin, g/l	3.19±0.03	3.18±0.04	3.05±0.05	2.94±0.04
LDH, U/l	181±11	202±17	165±16	154±13
CK, U/l	166±22	196±22	164±11	202±34
ALT, U/l	20±2	20±2	19±2	19±2
AST, U/l	48±6	62±6	56±2	62±6

Values are means ± SE; n = 7-8 mice/group. Male 12- to 14-wk-old wild-type (WT) or glutathione-S-transferase isoform P (GSTP)-null mice were exposed to air or cigarette smoke (CS; 5 h/day) for 3 days. Mice were euthanized 24 h after exposure after an 8-h fast. Hct, hematocrit; HDL, high-density lipoprotein; LDL, low-density lipoprotein; TP, total protein; LDH, lactate dehydrogenase; CK, creatine kinase; ALT, alanine aminotransferase; AST, aspartate aminotransferase.

^*P < 0.05 vs. the air-exposed group.

Tobacco smoke-induced endothelial dysfunction.

To examine how GSTP affects the vascular toxicity of tobacco smoke, mice were exposed for 3 days to tobacco smoke. The aorta was removed, and its responses were studied ex vivo. As shown in Fig. 2, tobacco smoke exposure for 3 days did not significantly affect PE-induced tension development in WT mice. No significant change in ACh-induced relaxation was observed. In contrast, aortic rings prepared from similarly exposed GSTP-null mice showed a much smaller extent of ACh-induced relaxation than air-exposed mice (Fig. 2D). In the aorta isolated from mice exposed to tobacco smoke, maximal relaxation due to ACh was nearly 60% of that observed in the aorta of mice that inhaled air alone (Table 2). GSTP-null mice also showed higher levels of PE-mediated contractility compared with WT mice. Moreover, even though tobacco smoke exposure did not affect SNP-mediated relaxation in WT mice, it led to a minimal, but statistically significant, decrease in SNP-induced relaxation in the GSTP-null aorta (Table 2). Responses of the aorta from naïve WT and GSTP-null mice to ACh, nitroglycerin, and SNP were similar, indicating that there were no basal differences in aortic sensitivity between these two strains of mice (Table 2). Taken together, these data indicate that deletion of GSTP increases endothelial dysfunction induced by tobacco smoke.

Fig. 2.

GSTP deficiency increased the susceptibility to tobacco smoke-induced endothelial dysfunction. Aortas were removed from WT and GSTP-null mice exposed for 3 days to cigarette smoke (CS) as described in the text, and their contractile and relaxant responses to cumulative concentrations of phenylephrine (PE; A and C) and ACh (B and D) were measured (n = 4 WT and 4 null). Control mice (n = 4 WT and 3 null) were exposed to air alone. *P < 0.05.

Table 2.

Vascular effects of air or CS exposure in WT and GSTP-null mice

	WT Mice		GSTP-Null Mice
	Air	CS	Air	CS
PE
Tension, mg	1256±196	1112±119	875±203	1246±214
Tension, g/mg wet wt	1.16±0.19	0.88±0.13	0.57±0.13	1.07±0.21
Tension, PE/80 mM K+ buffer	106±3	93±10	107±4	102±12
EC50, nM	68±17	94±19	83±24	107±33
pD2	7.18±1.61	7.07±0.12	7.11±1.59	7.01±1.57
ACh
Relaxation, %PE	−61±8	−73±10	−80±9	−48±9†
EC50, nM	328±115	665±454	132±54	251±131
pD2	6.55±1.47	6.52±0.32	6.71±1.52	6.76±1.52
GTN
Relaxation, %PE*	−114±17	ND	−103±4	ND
EC50, nM*	27±7	ND	83±67	ND
pD2*	7.63±0.10	ND	7.60±0.32	ND
SNP
Relaxation, %PE	−110±8	−105±17	−158±20	−100±6†
EC50, nM*	38±21	ND	9±3	ND
pD2*	7.68±0.19	ND	8.12±0.13	ND

Values are means ± SE; n = 4 mice/group. Male 12- to 14-wk-old WT or GSTP-null mice were exposed to air or CS (5 h/day) for 3 days. Mice were euthanized 24 h after exposure after an 8-h fast. PE, phenylephrine; GTN, glyceryl trinitrate; SNP, sodium nitroprusside; ND, not determined.

^*Data from naïve mice only (n = 4 mice/group).

^†P< 0.05 vs. the matched air-exposed control group.

Tobacco smoke and antioxidant defenses.

Because tobacco smoke is known to induce oxidative stress, we tested whether the induction of antioxidant enzymes in the lung and liver were similar in WT and GSTP-null mice. After 3 days of tobacco smoke exposure, changes in lung and liver antioxidant enzymes were quantified by Western blot analysis (Table 3). No changes were observed in lung or hepatic expression of NQO1, a nuclear factor-E₂-related factor 2 (Nrf2)-dependent gene; however, a significant depression of hepatic NQO1 expression occurred in smoke-exposed GSTP-null mice compared with air-exposed GSTP-null controls (Fig. 3A). Pulmonary expression of FR-1, an aldo-keto reductase involved in aldehyde metabolism (38), was increased in WT mice after 3 days of tobacco smoke exposure (Fig. 3B), but this protein was downregulated in lungs of GSTP-null mice after 3 days of tobacco smoke exposure (Fig. 3B). Hepatic microsomal CYP1A1 expression was significantly increased in tobacco smoke-exposed WT mice (Fig. 3C); however, CYP1A1 expression was significantly greater in air-exposed GSTP-null mice compared with WT air-exposed mice, indicating a higher basal level of CYP1A1 expression in null mice that was neither enhanced nor suppressed by tobacco smoke exposure in GSTP-null mice (Fig. 3C). No changes in the protein expression of other aldehyde-metabolizing enzymes, including ALDH3A1, GSTP, CYP2B1, or CYP2E1, were observed with tobacco smoke exposure (Table 3). After 3 days of tobacco smoke exposure, there was an ∼10% increase in total GST and GSTP activities as well as increased immunohistochemical staining for GSTP localized in the airway epithelium (data not shown) in WT lungs, but there were no changes in total GST activity in GSTP-null lungs compared with air-exposed GSTP-null lungs (Fig. 3D). Moreover, GSTP activity measured with EA was ∼2% of the overall GST activity yet was completely absent in the lungs of air- and tobacco smoke-exposed GSTP-null mice (Fig. 3D). These data indicate that the tobacco smoke exposure used in the present study was sufficient to induce mild oxidative stress and that the responses elicited by tobacco smoke exposure in the lung and liver were dependent on the presence of GSTP.

Table 3.

Expression of antioxidant enzymes in the lung and liver after 3-day air or CS exposure in male WT and GSTP-null mice

	FR-1	ALDH3A1	GSTP	CYP1A1/1A2	CYP2E1	CYP2B1	NQO1
Lung
WT mice
Air	1.00±0.03	1.00±0.01	1.00±0.05	-	-	-	1.00±0.01
CS	1.22±0.03*	1.08±0.05	0.92±0.05	-	-	-	1.11±0.05
GSTP-null mice
Air	1.01±0.02	1.10±0.06	Null	-	-	-	1.00±0.02
CS	0.87±0.03*	1.04±0.02	Null	-	-	-	0.83±0.05
Liver
WT mice
Air	-	-	-	1.00±0.08	1.00±0.03	1.00±0.05	1.00±0.16
CS	-	-	-	1.38±0.09*	1.07±0.09	1.06±0.04	0.85±0.07
GSTP-null mice
Air	-	-	-	1.74±0.05*	1.28±0.14	1.19±0.08	1.14±0.07
CS	-	-	-	1.64±0.52	1.16±0.07	1.32±0.05	0.91±0.01*

Values are means ± SE; n = 3 mice/group. Western blot band intensity was normalized to total lane amido black staining. Relative protein expression was calculated as a percentage of the WT air-exposed value. FR-1, FGF-regulated protein 1; ALDH3A1, aldehyde dehydrogenase 3A1; CYP, cytochrome P-450; NQO1, NAD(P)H:quinone oxidoreductase 1.

^*P < 0.05 vs. the WT air-exposed group.

Fig. 3.

CS exposure (3 days) induced lung and liver antioxidant proteins. A: FGF-regulated protein (FR)-1 was induced in the lung of WT mice and was downregulated in GSTP-null mice. B: basal hepatic expression of NAD(P)H:quinone oxidoreductase 1 (NQO1) was greater in GSTP-null mice compared with WT mice. CS exposure downregulated hepatic NQO1 expression in GSTP-null mice. C: basal hepatic expression of cytochrome P-450 (CYP)1A1/1A2 was greater in GSTP-null mice compared with WT mice. CS induced hepatic CYP1A1 expression in WT mice. D: catalytic activity of GST in the mouse lung. Total GST and GSTP activities in lung homogenates of WT and GSTP-null mice exposed to air or CS were measured using CDNB or ethacrynic acid (EA), respectively. Whole lung homogenate or hepatic microsomal proteins (25–100 μg) were blotted with polyclonal antibodies, and band density was normalized to total amido black staining. Protein expression was calculated as a percentage of WT air-exposed samples. *P < 0.05, significant difference compared with WT air-exposed mice (n = 3 mice/group).

Protein-acrolein adducts of CS exposure.

Given that acrolein is one of the most toxic components of CS (7) and a high-affinity substrate of GSTP (8), we next tested whether exposure to tobacco smoke leads to systemic delivery of acrolein from the lung to the peripheral tissues. As shown in Fig. 4, exposure to tobacco smoke led to an increase in the abundance of protein adducts of acrolein in the lung, plasma, and aorta of tobacco smoke-treated WT mice compared with air-exposed controls immediately after 5 h of exposure to tobacco smoke as measured by Western blot analysis. Although low levels of protein-acrolein adducts, presumably due to basal oxidative stress, were observed in air-exposed lungs, significant increases were observed in the intensity of acrolein-adducted proteins of molecular weights of ∼22, ∼30, ∼75, and ∼250 kDa in the lungs of tobacco smoke-exposed mice. A similar increase in the 150-kDa protein band also was observed in the plasma and aorta (Fig. 4, B–D). Additional proteins of different molecular weights were also modified by acrolein in the aorta, and a few protein bands were selectively increased in WT or GSTP-null mice exposed to tobacco smoke. For example, the protein band at ∼150 kDa was significantly increased (P = 0.003) in the GSTP-null aorta but not the WT aorta, whereas the ∼250- and ∼22-kDa bands were significantly increased in the WT but not GSTP-null aorta exposed to tobacco smoke (Fig. 4, C and D). There was less immunohistochemical protein-acrolein adduct staining present in formalin-fixed tissues of 3-day tobacco smoke-exposed mice killed 24 h after the final exposure compared with tissues taken immediately after a 5-h tobacco smoke exposure, indicating that protein adducts of acrolein were rapidly removed or repaired (data not shown). Taken together, these data support the notion that tobacco smoke entering the lung delivers acrolein to the plasma and aorta (and liver) and that deletion of GSTP, while not grossly altering systemic delivery of acrolein, did affect the aorta-specific generation of protein adducts of acrolein.

Fig. 4.

Exposure to tobacco smoke results in the systemic delivery of acrolein. A–D: Western blots showing protein-acrolein adducts in the lung (A), plasma (B), and aorta (C and D) of mice exposed to air or CS. Adult male WT and GSTP-null mice were exposed for 5 h to tobacco smoke or air and euthanized immediately after exposure. Protein-acrolein adducts were detected in lysates of tissues frozen immediately after exposure using purified IgG rabbit polyclonal anti-keyhole limpet hemocyanin (KLH)-acrolein antibody. M, marker. Summary data of major protein-acrolein adducts in tissues were normalized to total amido black-stained protein. n = 3 mice/group in A, C, and D; n = 4 mice/group for B. *P < 0.05; †0.10 > P > 0.05.

Acrolein-induced endothelial dysfunction.

Given our results showing that acrolein was delivered to vascular sites in tobacco smoke-exposed mice, we tested whether GSTP regulates the endothelial and systemic toxicity of acrolein (Tables 4 and 5). For this, both WT and GSTP-null mice were exposed to inhaled acrolein. After exposure, aortas from acrolein- and air-exposed mice were removed, and their sensitivity to ACh was examined ex vivo. As shown in Fig. 5, the 3-day acrolein inhalation protocol (1 ppm, 5 h/day) had no effect on PE sensitivity of isolated aortas of WT or null mice (Fig. 5, A and B). However, acrolein decreased ACh-induced relaxation in GSTP-null but not WT mice (Fig. 5, C and D), indicating that deletion of the GSTP gene exacerbates acrolein-induced endothelial dysfunction (Table 4).

Table 4.

Vascular effects of air or acrolein in WT and GSTP-null mice

	WT Mice		GSTP-Null Mice
	Air	Acrolein	Air	Acrolein
PE
Tension, mg	792±140	1030±130	762±120	888±132
Tension, g/mg wet wt	ND	ND	0.4±0.07	0.7±0.1
Tension, PE/80 mM K+ buffer	116±17	139±28	130±14	118±7
EC50, nM	125±15	120±18	110±24	110±18
pD2	6.91±0.05	6.94±0.06	6.95±0.08	7.00±0.08
ACh
Relaxation, %PE	−66±7	−76±13	−87±14	−51±7*
EC50, nM	119±29	138±50	110±36	280±83
pD2	6.98±0.13	6.97±0.19	7.04±0.15	6.70±0.15
SNP
Relaxation, %PE	−118±28	−152±46	−136±51	−92±7

Values are means ± SE; n = 4-8 mice/group. Male 12- to 14-wk-old WT or GSTP-null mice were exposed to air or acrolein (1 ppm, 5 h/day) for 3 days. Mice were euthanized 16 h after the final exposure. ND, not determined.

^*P < 0.05 vs. the matched air-exposed control group.

Table 5.

Blood parameters in WT and GSTP-null mice exposed to air or acrolein

	WT Mice		GSTP-Null Mice
	Air	Acrolein	Air	Acrolein
Hct, %	45.0±0.5	46.0±1.0	41.3±0.8	44.7±0.5
Buffy coat, %	0.9±0.0	1.7±0.0*	1.5±0.1	1.4±0.1
Cholesterol, mg/dl	94±2	95±3	113±3	112±2
HDL, mg/dl	68±3	64±2	75±3	77±2
LDL, mg/dl	14±1	16±1	15±1	13±1
Triglycerides, mg/dl	76±8	72±5	76±7	56±4*
TP, g/l	4.78±0.08	4.83±0.11	4.73±0.08	4.70±0.13
Albumin, g/l	2.96±0.03	3.04±0.02	2.88±0.07	3.04±0.08
LDH, U/l	140±12	160±1	205±17	235±24
CK, U/l	209±28	275±26	157±12	156±13
ALT, U/l	28±1	28±1	34±2	31±2
AST, U/l	59±6	69±4	64±4	66±5

Values are means ± SE; n = 4 WT mice/group and 8 GSTP-null mice/group. Male 12- to 14-wk-old WT or GSTP-null mice were exposed to air or acrolein (1 ppm, 5 h/day) for 3 days. Mice were euthanized 16 h after exposure.

^*P < 0.05 vs. the matched air-exposed control group.

Fig. 5.

GSTP-dependent endothelial dysfunction of inhaled acrolein. WT and GSTP-null mice were exposed to air or acrolein (1 ppm) by inhalation for 3 days (5 h/day) and killed 16 h after the final exposure. Isolated thoracic aortic (∼5 mm) sensitivities to PE (A and B) and ACh (C and D) were measured to assess contractility and endothelium-dependent relaxation, respectively. n = 4 WT mice (4 mice/group) and 8 null mice (8 mice/group). *P < 0.05.

Protein-acrolein adducts of acrolein exposure.

Because protein-acrolein adducts were present in peripheral tissues after tobacco smoke exposure, we tested whether direct exposure to acrolein would also lead to a similar tissue distribution and accumulation of protein-acrolein adducts. As shown in Fig. 6, acrolein exposure led to an increase in the abundance of acrolein-protein adducts in the lung, plasma, and aorta of acrolein-treated mice compared with air-exposed controls immediately after 5 h of exposure as measured by Western blot analysis. Although low levels of protein-acrolein adducts were observed in air controls (as in Fig. 4), increases in the intensity of acrolein-adducted proteins of molecular weights of ∼75, ∼130, ∼150, and ∼250 kDa were observed in the lung. A similar increase in an ∼150-kDa protein band also was observed in the plasma and aorta (Fig. 6, B–D). Additional proteins of different molecular weights were also modified by acrolein in the aorta, with a few protein bands selectively increased in GSTP-null mice exposed to acrolein. For example, bands of ∼150 (P = 0.055), ∼75 (P = 0.029), and ∼70 (P = 0.002) kDa were significantly increased in GSTP-null aortas but not WT aortas, whereas the 250-kDa band was also significantly increased in GSTP-null aortas exposed to acrolein, but to a lesser degree in WT mice (Fig. 6, C and D). There was less immunohistochemical protein-acrolein adduct staining present in formalin-fixed tissues of 3-day acrolein-exposed mice killed 16 h after the final exposure compared with tissues taken immediately after a 5-h acrolein exposure, indicating that protein adducts of acrolein were rapidly removed or repaired (data not shown). Taken together, these data confirm that tobacco smoke entering the lung delivers acrolein to the plasma and aorta and that deletion of GSTP, while not altering the systemic delivery of acrolein, increases the formation of specific acrolein-protein adducts in the aorta.

Fig. 6.

Exposure to acrolein results in the systemic delivery of acrolein. A–D: Western blots showing protein-acrolein adducts in the lung (A), plasma (B), and aorta (C and D) of mice exposed to air or acrolein. Adult male WT and GSTP-null mice were exposed for 5 h to acrolein (5 ppm) or air and then euthanized immediately after exposure. Protein-acrolein adducts were detected in lysates of tissues frozen immediately after exposure using purified IgG rabbit polyclonal anti-KLH-acrolein antibody. Summary data of major protein-acrolein adducts in tissues were normalized to total amido black-stained protein. n = 3 mice/group for A, C, and D; n = 4 mice/group for B. *P < 0.05; †0.10 > P > 0.05.

Finally, to examine the role of GSTP in directly modulating acrolein toxicity, aortic rings prepared from naïve mice were treated with acrolein in the tissue bath. As shown in Fig. 7A, treatment with acrolein led to a concentration-dependent decrease in ACh-induced relaxation of PE-precontracted aortic rings prepared from C57BL/6 mice in vitro. Although ACh-induced relaxations in control (untreated) aortas of WT and GSTP-null mice were similar, the extent of acrolein-induced inhibition was much greater in GSTP-null mice than in WT mice (Fig. 7B). The exaggerated endothelial dysfunction in GSTP-null mice induced by acrolein was rescued by pretreatment with N-acetylcysteine (NAC; Fig. 7B). NAC treatment alone did not affect aortic reactivity. Collectively, these results suggest that GSTP protects against endothelial dysfunction induced by direct acrolein exposure. That the high sensitivity of GSTP-null mice to acrolein could be prevented by NAC pretreatment supports the notion that the toxicity of acrolein could be attenuated by thiol conjugation.

Fig. 7.

Acrolein toxicity in WT and GSTP-null mice. A: ACh-stimulated relaxation of PE-precontracted aortas of male naïve C57BL/6 mice treated with acrolein. Naïve aortic rings (n = 3–5) were exposed to normal buffer (control) or acrolein (10, 40, or 80 μM) for 90 min before responses to ACh were measured in PE-precontracted aortas. B: effects of N-acetylcysteine (NAC) on ACh-induced relaxation in acrolein-pretreated aortas of naïve WT and GSTP-null mice. n = 4 mice/group. *P < 0.05 vs. GSTP-null control; †P < 0.05 vs. acrolein treatment in WT mice.

DISCUSSION

The major findings of this study are that deletion of the GSTP gene exacerbates endothelial dysfunction induced by exposure to tobacco smoke or to inhaled acrolein. We hypothesized that smoking-induced endothelial dysfunction was mediated in part by electrophilic constituents of CS such as acrolein. Hence, detoxification of electrophilic CS constituents by GSTP may be a protective mechanism against the vascular effects of tobacco smoke and other acrolein-rich pollutants (e.g., coal smoke, wood smoke, and automobile exhausts). The results obtained support the hypothesis and significantly advance our understanding of the vascular toxicity of tobacco smoke and the processes that modulate the CVD risk of smoking.

The endothelium appears to be a highly vulnerable target of tobacco smoke. Smoking injures endothelial cells, and vessels isolated from chronic smokers show degenerative changes (10). In humans, smoking diminishes flow-mediated endothelium-dependent vasodilation (5), and serum from smokers decreases nitric oxide (NO) production and endothelial NO synthase activity in endothelial cells (6). Moreover, aortas isolated from tobacco smoke-exposed animals have shown impaired endothelium-dependent relaxation (36). In agreement with these results, we found that a short, 3-day exposure to tobacco smoke diminished ACh-induced relaxation in aortas isolated from exposed GSTP^−/− mice. This early and significant dysfunction underscores the high vulnerability of the endothelium to tobacco smoke exposure. Significantly, similar inhibition of ACh-mediated relaxation was observed when GSTP^−/− mice were exposed to acrolein, which is one of the most reactive and toxic chemicals in CS. Taken together, these observations raise the possibility that endothelial dysfunction induced by tobacco smoke could be in part attributable to acrolein and related electrophilic components of CS.

A significant role of acrolein and related electrophiles in tobacco smoke toxicity is also consistent with the observation that deletion of GSTP, which catalyzes the conjugation of acrolein with glutathione with 6- to 35-fold higher efficiency than other GSTs (8), increased both acrolein- and tobacco smoke-induced endothelial dysfunction. Endothelium-dependent effects of acrolein were abrogated by pretreatment with NAC, consistent with the ability of thiol conjugation to mitigate unsaturated aldehyde-induced injury. Thus, the observations that both chemical (NAC) and enzymatic (GSTP) quenching of acrolein, an electrophilic constituent of CS, were effective in preventing the vascular toxicity of tobacco smoke support the idea that thiol-reactive components of CS (of which acrolein is the most reactive) are significant mediators of CS-induced endothelial dysfunction.

Acrolein and related aldehydes are present in high concentrations in CS. Several reports have estimated that between 100 and 600 μg of acrolein are generated per cigarette (50–70 ppm) and that acrolein constitutes 50–60% of total vapor-phase electrophiles (15, 18, 39). Because most of acrolein is generated during smoldering, its concentration in sidestream smoke is 10- to 12-fold higher than in mainstream smoke (18). This could explain, in part, the finding that even though the dose of smoke delivered to active smokers is ∼100 times more than that delivered to passive smokers, the relative rate of CVD for smokers is 1.78 compared with 1.31 for passive smokers (4). Hence, decreasing acrolein emissions and exposures or increasing acrolein metabolism via GSTP may help in preventing the cardiovascular toxicity of tobacco smoke.

Our results show that exposure to tobacco smoke results in the appearance of protein-acrolein adducts in the lung, plasma, and aorta of exposed animals. These observations suggest that despite its high reactivity, acrolein is delivered from the lung into the systemic circulation and to vascular sites. Smokers’ urine contains high concentrations of acrolein metabolites (6–8 μM) demonstrating the presence of systemic acrolein and contribution of acrolein metabolism (12). However, it is not clear whether these metabolites are generated in the lung and then excreted in the urine or whether they are derived from nonpulmonary metabolism, possibly in the liver. Hence, our observation that tobacco smoke (and acrolein) exposure led to the formation of specific and shared protein-acrolein adducts in the plasma and aorta suggests that free acrolein is transported from the lung into the blood and, thus, could cause direct vascular toxicity. Although the present data cannot rule out the possibility that glutathione conjugates formed in the lung dissociate in the blood, the increased formation of adducts at nonpulmonary sites supports the concept that nonpulmonary metabolism of acrolein by enzymes such as GSTP may be an important determinant of vascular toxicity due to CS. This view is consistent with our observation that deletion of GSTP led to an increase in the abundance of specific acrolein protein adducts in the aorta. These observations indicate that the extent of acrolein reactivity in the aorta may be, in part, regulated by the metabolic capacity intrinsic to the aorta.

Exacerbation of tobacco smoke-induced protein-acrolein adduct accumulation and endothelial dysfunction in GSTP-null mice indicates that this enzyme may be an important modulator of CVD risk due to smoking in humans. Although complete GSTP deficiency is rare in humans, several known polymorphisms of the human GSTP gene have been identified. Of these, the human (h)GSTP1 (l104,A113) allele is the most frequent in human populations; however, the frequency of the hGSTP (V104, A113) and/or hGSTP (V104, V113) allele is higher in certain cancers (22, 45). A previous study (35) has shown that the catalytic efficiency of the GSTPVV form with acrolein is much less than that of other polymorphic variants of GSTP (130 vs. 90 mM⁻¹·s⁻¹). The GSTPVV form also differs in catalytic efficiency with benzo[a]pyrene diol epoxide (26, 27). Hence, our observation that GSTP deficiency in mice increases the vascular toxicity of tobacco smoke and acrolein in mice raises the possibility that humans with the GSTPVV isoform may be at a higher risk of tobacco smoke-induced CVD than those who possess other allelic forms of GSTP. Epidemiology studies have associated GSTP polymorphisms with a variety of respiratory outcomes that may reflect functional changes in airway responsiveness to endogenous and exogenous electrophiles (3, 31, 32, 40). Nevertheless, the effects of GSTP polymorphism on CS-induced endothelial dysfunction remain unknown. Although additional epidemiological studies are required to assess this risk, we have observed that human coronary artery bypass graft blood vessels (arteries and veins) express GSTP in high abundance, which indicates that this enzyme may be an important determinant of the response of human vessels to tobacco smoke as well (D. J. Conklin, unpublished observations). Furthermore, in addition to differences in polymorphic alleles, individual differences in GSTP activity may arise also from enzyme induction. GSTP is a highly inducible enzyme. Although GSTP was not induced by tobacco smoke in our model, the gene is induced by a variety of environmental and dietary factors such as garlic organosulfur compounds (44), chemopreventive selenocysteine conjugates (43a), and coffee (42). Hence, GSTP induction by such agents could modify the CVD risk of cigarette smoking.

It is currently believed that the induction of GSTP protects tumors from anticancer therapy via conjugation and removal of electrophilic products or by preventing apoptosis (28). In the present study, we did not observe significant induction of lung GSTP, but we did see tobacco smoke-induced upregulation of other antioxidant proteins in a GSTP-dependent manner. As shown in Fig. 3, tobacco smoke exposure significantly increased AKR1B8 (FR-1) protein in WT mice and tobacco smoke increased the pulmonary expression of FR-1 mRNA (37). FR-1 is an antioxidant protein involved primarily in the reduction of lipid peroxidation aldehydes, such as 4-hydroxy-trans-2-nonenol (HNE), and is a mouse homolog of aldose reductase (41). Thus, it is surprising that pulmonary FR-1 protein expression under the regulatory control of Nrf2 was significantly downregulated in GSTP-null mice after tobacco smoke exposure, because this would likely lead to a decrease in the overall aldehyde detoxification capacity. Because FR-1 reduces a wide array of aldehydes, a decrease in FR-1 levels, coupled with GSTP deficiency, could significantly elevate the levels of free acrolein and other secondary aldehydes generated by CS-induced oxidative stress in the lung and, thus, increase the overall toxicity of tobacco smoke. In addition, GSTP also seems to protect against the systemic toxicity of tobacco smoke. Our observations that deletion of the murine GSTP gene prevented tobacco smoke-induced decreases in HDL, triglycerides, and blood glucose (Table 1) and the induction of CYP1A1/1A2 (Table 3) suggest that adaptive changes to tobacco smoke are blunted in GSTP-null mice. This surprising effect of GSTP deficiency on the transcriptional regulation of other antioxidant enzymes and the induction of adaptive responses has not been described before and could provide an important clue for understanding how GSTP polymorphisms may contribute to increased cardiopulmonary sensitivity to xenobiotic exposures in humans.

In addition to tobacco smoke, acrolein and related aldehydes are also a component of automobile exhaust, smog, cotton, wood, and coal smoke as well as ambient air (9). The concentration of acrolein in ambient air ranges from 0.003 to 0.01 ppm, whereas 0.04–2.2 ppm has been detected near automobile exhaust (18). Acrolein is also endogenously generated during inflammation and lipid peroxidation (9). High (0.7–2 μM) concentrations of S-(3-hydroxypropyl)mercapturic acid, the major metabolite of acrolein, have been detected in the urine of healthy young adults, and the concentration of the acrolein metabolite exceeds that of another lipid peroxidation product (HNE) by a factor of 100 (43), suggesting that high levels of acrolein are generated endogenously. Hence, our observation that GSTP deficiency increases the vascular toxicity of inhaled acrolein and acrolein exposure ex vivo raises the possibility that GSTP may be a critical determinant of cardiovascular injury due to inflammation or exposure to several environmental pollutants other than tobacco smoke as well.

GRANTS

This work was supported by National Institutes of Health Grants ES-11860 (to A. Bhatnagar) and HL-89380 (to D. J. Conklin), the American Health Assistance Foundation/National Heart Foundation (to D. J. Conklin), the Environmental Protection Agency (to A. Bhatnagar and D. J. Conklin) and Philip Morris USA Incorporated and by Philip Morris International (to A. Bhatnagar).