Formation of Biologically Active Oxysterols During Ozonolysis of Cholesterol Present in Lung Surfactant

Exposure of the lung to concentrations of ozone found in ambient air is known to cause toxicity to the epithelial cells of the lung. Because of the chemical reactivity of ozone, it likely reacts with target molecules in pulmonary surfactant, a lipid rich material which lines the epithelial cells in the airways. Phospholipids containing unsaturated fatty acyl groups and cholesterol would be susceptible to attack by ozone, which may lead to the formation of cytotoxic products. While free radical derived oxidized cholesterol products have been frequently studied for their cytotoxic effects, ozonized cholesterol products have not been studied, though they could reasonably play a role in the toxicity of ozone. The reaction of ozone with cholesterol yielded a complex series of products including 3 (cid:36) -hydroxy-5-oxo-5,6-secocholestan-6-al, 5-hydroperoxy-B-homo-6-oxa-cholestan-3 (cid:36) ,7a-diol and 5 (cid:36) ,6 (cid:36) -epoxycholesterol. Mass spectrometry and radioactive monitoring were used to identify the major cholesterol derived product during the reaction of 2 ppm ozone in surfactant as 5 (cid:36) ,6 (cid:36) -epoxycholesterol, which is only a minor product during ozonolysis of cholesterol in solution. A dose dependent formation of 5 (cid:36) ,6 (cid:36) -epoxycholesterol was also seen during direct exposure of intact cultured human bronchial epithelial cells (16-HBE) to ozone. Studies of the metabolism of this epoxide in lung epithelial cells yielded small amounts of the expected metabolite, cholestan-3 (cid:36) ,5 (cid:34) ,6 (cid:36) -triol, and more abundant levels of an unexpected metabolite, cholestan-6-oxo-3 (cid:36) ,5 (cid:34) -diol. Both 5 (cid:36) ,6 (cid:36) -epoxycholesterol and cholestan-6-oxo-3


INTRODUCTION
Human exposure to 0.2 ppm levels of ozone in ambient air has been shown to cause numerous pulmonary effects such as increased airway inflammation and decreased pulmonary function (1,2). Studies of ozone in animals using up to 3 ppm ozone have been shown to cause increased airway hyperresponsivenesss and epithelial cell death. It has been hypothesized that the very high chemical reactivity of ozone limits the distribution of this gas in the pulmonary system, preventing direct exposure to the cellular components of the lung. In part ozone may react with the various components of the epithelial cell lining fluid in the lung, also known as pulmonary surfactant, which includes proteins, lipids and single electron antioxidant agents such as ascorbic acid (3)(4)(5). Due to the very high reactivity of ozone with lipids containing double bonds, considerable emphasis has been placed on the reaction of ozone with lipid compounds in the lungs and the possibility that the adverse effects of ozone are mediated by lipid ozonized products (6).
Evidence in support of this theory has been accumulating with the identification of biologically active phospholipids (7) such as 1-hexadecanoyl-2-(9-oxo-nonanoyl)-glycerophosphocholine, found following ozone exposure to lung surfactant (8). This oxidized phospholipid which eluted as a somewhat polar product on normal phase HPLC was found to initiate apoptotic death in monocytes and macrophages. However, a relatively non-polar component was also found to elute from this normal phase HPLC separation that was also cytotoxic, and preliminary data suggested that several oxidized neutral lipid products were present in this fraction.
Cholesterol is the most abundant neutral lipid present in pulmonary surfactant and this molecule has a double bond that would be susceptible to attack by ozone (9,10). While there has been some controversy about the exact chemical structure of the major ozonolysis product when cholesterol is ozonized in solution at high ozone concentrations (>0.1%) (11)(12)(13)(14), electrospray 5 tandem mass spectrometry was recently used to characterize this chemically reactive cholesterol ozonolysis product as 5-hydroperoxy-B-homo-6-oxa-cholestane-3$,7a-diol (Scheme 1) (15).
It is important to consider that the interaction between ozone and cholesterol has primarily been studied in organic solvents with high levels of ozone, where 5-hydroperoxy-B-homo-6-oxacholestane-3$,7a-diol is the major product (11,14,15). However, environmentally relevant concentrations of ozone acting on lipid cellular membranes or in lipid rich pulmonary surfactant could involve different chemistry, because of the ordered nature of the lipid bilayer, yielding alternate products. Isolated bronchoalveolar lavage fluid was exposed in vitro to precise levels of ozone in a carefully controlled ozone chamber to study the formation of cholesterol derived ozonolysis products. This revealed the formation of 5$,6$-epoxycholesterol ($-epoxide) (Scheme1) as a more abundant product than 5-hydroperoxy-B-homo-6-oxa-cholestan-3$,7a-diol in this system.
The ability of this compound and its cellular metabolites to cause cytotoxicity and to inhibit cholesterol synthesis in cultured human bronchial epithelial cells was subsequently studied.
Bis(trimethylsilyl)fluoroacetamide (BSTFA) and trypan blue dye (0.4%) were purchase from Sigma. 9 Bis(trimethyl-d -silyl)acetamide (99%) was purchased from Isotech (Miamisburg, OH). Rat lung lavage fluid was provided by Dennis Voelker (National Jewish Medical and Research Center, Denver, CO). After ozonolysis, the samples in each well were diluted with 2 ml of water, transferred to glass tubes, and wells were washed with 2 ml of methanol. To the transferred samples, 3ml methylene chloride was added and lipids extracted essentially as described by Bligh and Dyer (19).

Identification of Cholesterol Ozonolysis Products in
After drying the extract with a stream of dry nitrogen, the products were dissolved in 100 :l ethanol, then injected onto a C18 (250 x 4.6 mm) reversed phase column (Phenomenex, Torrance, CA) at a flow rate of 1 mL/min. Solvent A was methanol:water:acetonitrile (v:v:v;60:20:20) with 1 mM ammonium acetate; solvent B was methanol with 1 mM ammonium acetate. The gradient ran from 50% to 100% B in 20 min and stayed at 100% B for 20 min. Radioactive monitoring coupled with mass spectrometry was used to detect the products of cholesterol ozonolysis as previously described (15).

Quantitation of Epoxycholesterol, Cholesterol and Phospholipids.
Both the alpha and beta isomers of 5,6-epoxycholesterol were synthesized based on the method of Sevanian et al. (20) with slight modifications. Briefly, 19.1 mg chloroperoxybenzoic acid was added to 38.6 mg 6 cholesterol in 12 mL methylene chloride and stirred overnight at 4<C. Deuterated (2,2,3,4,4,6-H ) 2 cholesterol was used for synthesis of deuterated 5,6-epoxycholesterol. The solution was washed three times with water, and then with a saturated salt solution. Finally, the solvent was evaporated using a roto-evaporator and the product was resuspended in ethanol. Separation of the alpha and beta isomers was achieved using RP-HPLC on an Alltech column (250 x 10.0 mm, C18;Deerfield, IL) at a flow rate of 4 mL/min with the gradient described above. Fractions were collected (1 min) and dried under vacuum; after weighing, the fractions were derivatized with BSTFA and analyzed by GC/MS to determine purity of the reference standards (>95%). The dwell time for each transition was 800ms. The standard curve was linear for the range tested, from 0.625 ng to 320 ng 5$,6$-epoxycholesterol and from 100 ng to 6.4 :g cholesterol.
Before ozonolysis the media was removed and cells were washed with PBS. A small volume (300 :l) of HBSS was added to each well to keep the cells from drying out during ozone exposure. After exposure cells were trypsinized and the lipids were extracted as described above.
Trypan blue exclusion was used to measure cytotoxicity (22). 16 been previously studied since it was observed to form during a very different type of oxidative stress, namely free radical mediated cholesterol peroxidation (26). Synthesis of both epoxide steroisomers had been previously described (20), which provided a facile means to synthesize deuterated 5",6"-epoxycholesterol ("-epoxide) as well as deuterated $-epoxide. Each isomer was added in separate experiments to pulmonary surfactant after exposure to relatively low concentrations of ozone and it was found that the radioactive peak eluting at 21 min was in fact only one of the epimers, namely 5$,6$-epoxycholesterol based on co-elution of the deuterated $-epoxide with peak B (data not shown).
Deuterated $-epoxide was subsequently used as a mass spectrometry internal standard to facilitate quantitation of both isomers of epoxycholesterol in rat BAL treated with ozone. When BAL was exposed for 4 hr to increasing levels of 0.2, 0.5, and 1.0 ppm ozone, a dose dependent formation of $-epoxide was observed ( Figure 2). Up to 200 ng of $-epoxide was observed to form under these conditions, while approximately 3ug of unreacted cholesterol remained in the surfactant.
The beta isomer was formed in preference to the alpha isomer at all ozone concentrations studied with a ratio of approximately 5:1. Samples of rat BAL exposed to filtered air had low but detectable levels of $-epoxide, however, "-epoxide was not detected.
Phospholipids that contain a double bond in a fatty acyl chain are also an abundant component of pulmonary surfactant, and therefore it was of interest to determine the abundance of cholesterol derived ozonolysis products relative to phospholipid derived ozonolysis products.  Figure 4B). This unknown metabolite was found to substantially increase with further cell incubation (data not shown). Unmetabolized $-epoxide was still present in the media as indicated by Figure 4C. The elution of an internal standard (deuterated analog of $-epoxide) added during sample work-up was used to precisely mark the elution of $epoxide ( Figure 4D).
In order to obtain sufficient quantities of the unknown metabolite for further structural studies, lipids were extracted from 16-HBE cells (10 )  being derived from the carbon-3 TMS ether substituent (29). Therefore, the hydroxyl group at carbon-3 was likely intact, suggesting that the substitution of oxygen as an oxo group was in fact at carbon-6. All data was consistent with the unknown metabolite as cholestan-6-oxo-3,5-diol. This compound was synthesized (25) and was found to have the same retention time and electron ionization mass spectrum as the unknown metabolite ( Figure 5b).

Biological activity of 5$,6$-epoxycholesterol and cholestan-6-oxo-3,5-diol. Previous
investigators studied the cytotoxic effects of cholesterol epoxides in Chinese hamster lung fibroblasts, rabbit aortic endothelial cells and in the human monocytic cell line, U937. In these systems $-epoxide was more cytotoxic than the "-epoxide (20,30,31). Additionally, the $-epoxide was shown to be more mutagenic than the "-epoxide in mouse embryo cells (32). The values obtained from a representative experiment are shown in Table 1. Cholesterol synthesis was found to be inhibited even with nanomolar concentrations of the epoxide, and was almost completely inhibited at concentrations that were found to cause significant cytotoxicity (Figure 7).
Both $-epoxide and 6-oxo-3,5-diol were able to inhibit cholesterol synthesis using this assay with 50 50 an IC of 350 nM. The effect was more potent in the cultured lung cell line A549, with and IC of 150 nM (data not shown). The role of cholesterol depletion in the observed cytotoxicity was examined by attempting to rescue the cells by supplementation of the media with cholesterol. Cells were protected from the effects of the oxysterol 6-oxo-3,5-diol by the addition of cholesterol in a dose dependent manner (Table 1), as exemplified when cells treated with 15 :M 3,5-diol-6-one that would have resulted in approximately 50% cell death had less than 20% cell death when 30 :M cholesterol was coincubated with the oxysterol.

DISCUSSION
Formation of 5$,6$-epoxycholesterol has been previously observed as a result of cholesterol peroxidation and autooxidation (34). This oxysterol was shown to initiate cellular apoptosis in some systems (20,31) and to play a role in lipid loading of macrophages (35), but had not previously been described as a major product of cholesterol ozonolysis in lipid membranes. Ozonolysis of olefins is thought to characteristically proceed via formation of a Criegee ozonide which breaks the carbon carbon double bond leaving a keto or aldehyde moiety and a hydroperoxy hydroxyl acetal substituent at these carbons atoms. The reaction of ozone with cholesterol by this mechanism would be expected to yield 3$-hydroxy-5-oxo-5,6-secocholestan-6-al. An alternative, less appreciated mechanism of ozone involves addition of one oxygen atom of the trioxygen molecule to a double bond, followed by loss of diatomic oxygen, which results in epoxidation but not scission of the carbon-carbon bond (36). The yield of such epoxides during ozonolysis can be dependent on the solvent used and the degree of substitution at the target double bond. For example, lanosterol derivatives have been shown to primarily form epoxides at the very hindered 8,9 double bond during ozonolysis in methylene chloride (37). Alternatively, $-epoxide could form via lipid peroxidation since it has been proposed that ozone can initiate the formation of radical species (38). However, other studies of cholesterol ozonolysis in solution suggest that $-epoxide is a unique cholesterol ozonolysis product rather than a secondary result of peroxide formation (11) The identification of 6-oxo-3,5-diol as the major metabolite of $-epoxide that accumulated in 16-HBE cells was not anticipated, since other studies had identified 3,5,6-triol as the major metabolite of both "-epoxide and $-epoxide (20,28). A metabolism study of orally administered 3,5,6-triol in rat yielded 6-oxo-3,5-diol as a major metabolite (39). In fact, treatment of 16-HBE cells with 3,5,6-triol led to the production of 6-oxo-3,5-diol suggesting that 3,5,6-triol was an 22 intermediate formed during the metabolism of $-epoxide to 6-oxo-3,5-diol (data not shown). Very little is known about the biological activity of 6-oxo-3,5-diol; however, there have been a few studies suggesting it may be an endogenous ligand for cytosolic-nuclear tumor promoter binding protein, with which PMA had been shown to bind with high affinity (40,41). Furthermore, 6-oxo-3,5-diol has been used for photoaffinity labeling of this protein (42). The observation of 6-oxo-3,5diol as a major metabolite of $-epoxide is significant aside from its potential biological activity, in that this compound is isobaric with other cholesterol products suggested as biomarkers for ozone exposure, such as 5,6-secosterol and its corresponding aldol condensation product (molecular weight of 418 daltons). Careful analysis of the cholesterol derived products formed during ozone exposure, beyond just molecular weight determination, would be required to distinguish between these compounds.
There is reasonable debate about the claims that oxysterols are cytotoxic since levels used for in vitro studies are extraordinarily high, most often in the low micromolar range. Our study also found these levels necessary for cytotoxicity; however, it is reasonable that these levels could be leading to changes in cell signaling and suggesting that this pathway is involved in the observed cytotoxicity of statins and phytosterols (45,46). The oxysterols formed during ozone exposure in the lung may also cause changes in the prenylation of proteins due to their effects on isoprenoid synthesis. Some studies suggest that blocking protein prenylation can cause changes in inflammatory signaling (47,48). Treatment of macrophages with lovastatin was shown to stimulate low levels of TNF-" and to enhance LPS stimulated TNF-" production (47). Also, mevalostatin and lovastatin were shown to upregulate cyclooxygenase-2 expression in human aortic smooth muscle cells (48). Oxysterols may similarly act to enhance the inflammatory response observed during ozone exposure in the lung.
In conclusion, the observation of biologically active ozonized cholesterol products formed during exposure of pulmonary surfactant and lung epithelial cells to ozone supports the theory that ozonized lipids and particularly oxysterols may mediate the toxicity of ozone. The two compounds seen in this study, $-epoxide and 6-oxo-3,5-diol, have not been studied extensively, but similar 24 oxysterols, such as 7$-hydroxycholesterol, 25-hydroxycholesterol, and 7-ketocholesterol, have been implicated in the initiation of inflammatory signaling associated with atherosclerosis (49)(50)(51). The formation of $-epoxide and 6-oxo-3,5-diol in the lung may also play a role in the increased risk of cardiopulmonary diseases observed with high ozone exposure in epidemiological studies (52,53).
Finally, with regard to the recent interest in ozone as a biochemical product during inflammation (54), the biological role of $-epoxide and its metabolites needs to be examined as they may be formed in cellular membranes by reaction with ozone formed in vivo, and these compounds may play a significant role in the formation of lipid loaded macrophages and the development of atherosclerosis. Neutral lipids were extracted and quantitated using stable isotope dilution mass spectrometry. Low to undetectable levels of both of $-epoxide (black bars) and "-epoxide (striped bars) were observed in the filtered air controls. Nanogram levels of both epoxide isomers were observed in the ozone treated samples (n=3).
Approximately 1 x 10 cells (16-HBE) were exposed to indicated levels of ozone for 1 hour.

Table 1
Representative data of the effect of 5$,6$-epoxycholesterol and cholestan-6-oxo-3,5diol on cholesterol biosynthesis and cell viability in 16-HBE cells. Cells (approximately 1 x 10 in 6 35 mm wells) were treated with the indicated concentrations of $-epoxide. After 24 hr, C-acetate 14 (7 µCi per dish) was added and incubation carried out for an additional 3 hr. Suppression of cholesterol biosynthesis by $-epoxide at the indicated concentrations was measured by comparing the ratio of C-acetate incorporation into cholesterol (cpm) with that of triacylglycerol (cpm). 14 Comparison of the observed ratio from controls (untreated) to that observed for $-epoxide exposed cells was used to calculate the percent inhibition of cholesterol synthesis. Similar data for 6-oxo-3,5-diol were obtained (Figure 7). Cells could be rescued from the cytotoxic effects of the 6-oxo-