Products and mechanism of the reaction of ozone with phospholipids

Katherine Windsor , Thiago C. Genaro-Mattos , Sayuri Miyamoto , Donald F. Stec , Hye-Young H. Kim , Keri A. Tallman , and Ned A. Porter. Chemical Rese...
0 downloads 0 Views 1MB Size
Chem. Res. Toxicol. 1992,5, 134-141

134

Products and Mechanism of the Reaction of Ozone with Phospholipids in Unilamellar Phospholipid Vesicles Jeffrey Santrock,*if Robert A. Gorski,? and John F. O'Garat Biomedical Science and Analytical Chemistry Departments, General Motors Research Laboratories, Warren, Michigan 48090-9055 Received June 11, 1991

While considerable effort has been expended on determining the health effects of exposure to typical urban concentrations of 03,little is known about the chemical events responsible for toxicity. Phospholipids containing unsaturated fatty acids in the cell membranes of lung cells are likely reaction sites for inhaled ozone (03). In this study, we examined the reaction of O3 with l-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) in unilamellar phospholipid vesicles. Reaction of ozone with the carbon-carbon double bond of POPC yielded an aldehyde and a hydroxy hydroperoxide. The hydroxy hydroperoxide eliminated H202to yield a second aldehyde. Upon further ozonolysis, the aldehydes were oxidized to the corresponding carboxylic acids. A material balance showed that no other reaction consumed POPC and O3 or produced these products. As a mechanistic probe, we measured incorporation of oxygen-18 from I8O3into aldehyde, carboxylic acid, and H202. Approximately 50% of the aldehyde oxygen atoms were derived from 03.Oxygen in H202was derived solely from 03,where both oxygen atoms in a molecule of H202were from the same molecule of 03.One of the carboxylic acid oxygen atoms was derived from the precursor aldehyde, while the other was derived from 03.These results support the following mechanism. Cleavage of the carbon-carbon double bond of POPC by O3 yields a carbonyl oxide and an aldehyde. Reaction of H 2 0 with the carbonyl oxide yields a hydroxy hydroperoxide, preventing formation ozonide by reaction of the carbonyl oxide and aldehyde. Elimination of H202 from the hydroxy hydroperoxide yields a second aldehyde. Oxidation of the aldehydes by O3 yields carboxylic acids.

I ntroductlon Ozone is a toxic air pollutant formed in the atmosphere by a complex series of photochemical reactions (1, 2). Exposure to ozone causes a number of physiological responses in laboratory animals (3)and in humans (4),which include an influx of inflammatory cells into the lungs, decrements in pulmonary function, and changes in immune function. The chemical reactions of O3in the lungs which are responsible for these effects are unknown. Unsaturated fatty acids in membrane phospholipids are among possible reaction sites for ozone in biological systems. Traditionally, mechanistic studies have been concerned largely with the study of ozonolysis of olefins in pure, nonreactive solvents, where ozonides are the major products (5-9). These studies do not bear directly on questions of ozone toxicity. In one study, however, Tiege et al. examined the reaction of O3 with unsaturated lipids in biological-like membranes-liposomes and red blood cell ghost membranes (10). Ozonolysis resulted in cleavage of the carbon-carbon double bonds of unsaturated fatty acids, yielding HzOz and aldehydes or carboxylic acids as the terminal groups of the cleaved fatty acids. The authors postulated the mechanism shown in Schemes I and I1 to explain these results. Reaction of O3with the carboncarbon double bond of lipid 1 yielded aldehyde 2 and carbonyl oxide 3. Reaction of the carbonyl oxide with HzO yielded hydroxy hydroperoxide 5, preventing formation of ozonide 4 and leaving aldehyde 2 unreacted. Elimination of HzOz from the hydroxy hydroperoxide yielded a second aldehyde, aldehyde 6. Oxidation of the aldehydes *Address correspondence to this author at Research Biomedical Laboratories, 30500 Mound Rd., Box 9055, Warren, MI 48090-9055. +Biomedical Science Department. Analytical Chemistry Department.

Scheme I

5 -

4 HOO, ,C,H

,OH

A

R

5 -

0 I1

R,C.H

t HOOH

6 -

Scheme I1 0

1 1

R'c'H

t

0,

-

2 16-

0

II

R"'0

H

t

02

/a

7

by ozone yielded the corresponding carboxylic acids. A t least two questions arise concerning this reaction mechanism. First, hydroxy hydroperoxides were not observed directly. Tiege et al. based Scheme I solely on observation of aldehydes and HzOz (lo),which has been the usual criteria for inferring this reaction mechanism (11-15).Since then, others have observed aldehydes and HzOz as ozonolysis products of unsaturated lipids and inferred the mechanism shown in Scheme I (16). But identification of aldehydes and HzOZalone does not constitute unambiguous proof of Scheme I. Second, formation

0893-228x/92/2705-0134$03.00/00 1992 American Chemical Society

Chem. Res. Toxicol., Vol. 5, No. 1, 1992 135

Reaction of Ozone with Phospholipids in Vesicles

from the average decrease in absorbance at X = 600 nm, using of hydroxy hydroperoxide by Scheme I requires that the a value of 15500 for the molar absorptivity of indigotrisulfonate carbonyl oxide be accessible to H20. A lipid membrane (19). An aliquot ('/& of each solution was removed for meais a multiphase system composed of hydrated phosphosurement of peroxides. 1,2-Ditridecanoyl-sn-glycero-3-phoepholipids. The polar head groups of the lipids are located at choline (DTPC, 0.228 mg, 350 nmol) was added to the remainder the outer surfaces of the membrane, in contact with the as an internal standard, and lipids were extracted and concenaqueous medium. The fatty acid side chains are sequestrated for HPLC analysis. 1,2-Ditetradecanoyl-sn-glycero-3tered within the hydrophobic interior of the membrane, phosphocholine (DMPC, 0.257 mg, 380 nmol) was added to the where the concentration of H 2 0 is exceedingly low. The extract to determine recovery, which was 296% in all cases. low concentration of H20within the bilayer may have been Chromatographic Methods. Phosphocholines in the product mixture were separated by reverse-phase HPLC (20, 21). A insufficient for the reaction shown in Scheme I, and some Pecosphere ODS column (0.46 cm i.d. X 8.3 cm length, 3 pm other mechanism yielded these products. particle size; Perkin Elmer) fitted with a Pellicular ODS guard In this study, we elucidated the structures of organic ozonolysis products of l-palmitoyl-2-oleoyl-sn-glycero-3- column was used for analytical separations. The mobile phase was a gradient of water (A), hexane (B), methanol (C), and 200 phosphocholine (POPC)' in unilamellar vesicles. OzonomM ammoNum acetate in methanol (D): isocratic at 15% A, 0% lysis of POPC yielded the expected aldehyde, hydroxy B, 80% C, and 5% D for 5 min; to 9.5% A, 4.8% B, 80.7% C, hydroperoxide, and carboxylic acid: l-palmitoyl-2-(9'and 5% D in 10 min; and isocratic at 9.5% A, 4.8% B, 80.7% C, oxononanoyl)-sn-glycero-3-phosphocholine(PN'PC), 1and 5% D for 30 min. The column temperature was 40 "C,and palmitoyl- 2- (9'- hydroxy-9'-hydroperoxynananoyl)-snthe flow rate was 1.5 mL/min. Lipids were deteded using a mass glycero-3-phosphocholine(PN,PC), and l-palmitoyl-2detector (Model 750/14; Peris Industria, State College, PA). Mole response factors were determined using authentic DPPC, DTPC, (9'-carboxynonanoyl)-sn-glycero-3-phosphocholine(PN2PNIPC, PNzPC, and POPC. An Ultrasphere ODS column (0.46 PC), respectively.2 To address the question of reaction cm i.d. X 25 cm length, 5 pm particle size; Altex) fitted with a mechanism, we determined a material balance for ozonoPellicular ODS guard column was used for preparative separations. lysis of POPC in vesicles and measured incorporation of The mobile phase was isocratic 7.5% A, 90% B, and 2.5% C, with lSO from lSO3into the aldehyde, the carboxylic acid, and 10 mM ammonium acetate, and the flow rate was 1 mL/min. I42024

Experimental Section Materials. Oxygen (99.99% purity) was from Scott Specialty Gases (Troy, MI), and (99 atom % oxygen-18) was from Cambridge Isotope Laboratories (Woburn, MA). Deuterated chloroform (99.996 atom % D, CDC13)was from MSD Isotopes (St. Louis, MO). Potassium indigotrisulfonate was from Aldrich (Milwaukee, WI).All phospholipids were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Potassium phthalate, horseradish peroxidase, p-hydroxyphenylacetic acid, catalase, and tris(hydroxymethy1)aminomethane (Tris) were from Sigma (St. Louis, MO). All solvents were HPLC grade. Phosphate-buffered saline [PBS; 10 mM PO4", 0.8% (wt/v) NaC1, pH 6.51 was prepared from water obtained from a Milli-Q purification system (Millipore, Bedford, MA). The PBS was sparged with O3 (5% in 0,)for 2 h, allowed to stand for 48 h, and sparged with Ar for 2 h before use. Phospholipid Vesicles. Unilamellar phospholipid vesicles with an average diameter of 0.1 pm were prepared by extrusion through polycarbonate filters (17). 1,2-Dipalmitoyl-sn-glyceroSphosphocholine (DPPC, 17.6 mg,24 pmol) and POPC (18.3 mg, 24 pmol) were hydrated in 5 mL of PBS at 50 "C. The resulting mixture was extruded 10 times through a polycarbonate membrane filter (0.1-pm pores, Nucleopore, Pleasanton, CA). The vesicles were diluted to 100 mL with PBS and divided into ten 10-mL aliquots. Ozonolysis. Ozone was prepared from either lSO2or l8OZin a corona discharge (18). The product was purified by cryogenic distillation and dissolved in PBS at 5 "C. The ozone solution was added to vesicles at 45 "C. Equal volumes were added to potassium indigotrisulfonate immediately before and after each addition to the vesicles. The amount of O3 added was determined Abbreviations: CAD, collisionally activated dissociation; DMPC, 1,2-ditetradecanoyl-sn-glycero-3-phosphocholine; DPPC, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine;DTPC, 1,2-ditridecanoyl-snglycero-3-phosphocholine;FAB-MS, fast atom bombardment mass spectrometry; GC/MS, gas chromatography/maea spectrometry; HPLC, high-performance liquid chromatography; LAH, lithium aluminium hydride; NMR, nuclear magnetic resonance; PBS, phosphate-buffered saline; PC, phosphatidylcholine; PNIPC, l-palmitoyl-2-(9'-oxononanoyl)sn-glycero-3-phosphocholine; PN,PC, l-palmitoyl-2-(9'-carboxynonanoyl)-an-glycero-3-phosphocholine;PN,PC, l-palmitoyl-2-(9'hydroxy-Y-hydroperoxynonanoyl)-sn-glycero-3-phosphocho~ne;POPC, l-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; RD, relaxation delay; TMS, trimethylsilyl; Tris, tris(hydroxymethy1)aminomethane. * For ozonolysis products of POPC, the capitol N represents a nonanoyl ((39) side chain in the 2-position and the subscript represents the number of oxygen atoms on the terminal carbon of nonanoyl side chain.

Fractions corresponding to 80% of the peak areas were collected. Samples were stored in CDC13 at -70 "C until analyzed. Mass Spectrometry. Mass spectra were obtained using a JEOL HXllO (Peabody, MA) or a Kratos MS 50 (Ramsey, NJ) mass spectrometer. The instruments were operated with mass resolutions of 3000 and ion accelerating potentials of 10 kV. Spectra were recorded from 100 to 800 amu. Fast atom bombardment (FAB)with Xe atoms at a potential of 6-8 kV was used to desorb sample ions from the matrix. Glycerol was the FAB matrix for all analyses except the exact mass determination of PN'PC, where a mixture of glycerol-thioglycerol(1:l) was used. Daughter ion spectra of the protonated molecular ions of PNIPC and PN2PC were obtained by linked scan at constant B / E of positive ions generated by collisionally induced decomposition (CID) of [M + HI+. Xenon was introduced into a collision cell in the f i t field-free region until the signal from the protonated molecular ion was reduced by 75%. High-resolution mass measurements of the [M + H]+ were obtained by peak matching at a resolution of 10OOO and an ion-acceleratingpotential of 10 kV. The [M H]+ of DMPC and DTPC were used as reference massea for high-resolution measurements. For isotopic analysis, the detector slits were opened to obtain flat-topped peaks. The abundances of l80in the products were determined from the abundances of the isotope peaks in the molecular ion clusters. Additionally, trimethylsilyl (TMS) derivatives of the alcohols and diols produced by reduction of the products by lithium aluminum hydride were prepared by reaction with N(trimethylsily1)imidazole in acetonitrile and analyzed by gas chromatography/ mass spectrometry (GC/ MS). Nuclear Magnetic Resonance Spectrometry. Nuclear magnetic resonance (NMR) experiments were performed on an IBM NR270AF spectrometer (Danbury, CT) operating at 270 MHz for 'H, 67.9 MHz for 13C,and 109.4 MHz for 31P. Samples were dissolved in 0.5 mL of CDC13. One-dimensional 'H spectra were obtained with a 3-pa 40" pulse, 16K data points, 3789-Hz spectral width, 1.1-srelaxation delay (RD), and -250 transients. Two-dimensional 'H-correlated spectra were obtained with the COSY pulse sequence RD-90°-t-45"-detect, using a 7-ps 90" pulse, a 1.6-kHz spectral width, and 2K data points. A total of 400 spectra were used to provide the equivalent of a l.6-msweep width in the second frequency dimension. Free induction decays were enhanced with sine function multiplications and zero-fied in the second frequency dimension before Fourier transformation. The final COSY spectrum was symmetrized and displayed in the absolute mode. One-dimensional 13Cspectra were obtained with a 3-ps 70" pulse, 32K data points, 16667-Hz spectral width, power-gated 'H broad-band decoupling with a RD of 6 s, and -60000 transients. All chemical shifts were referenced to the CHC1, impurity in the solvent, which was assigned chemical shifta

+

Santrock et al.

136 Chem. Res. Toxicol., Vol. 5,No. 1, 1992

3.93 (m, 2 H, Hg3),4.14 (m, 1 H, H,J, 4.31 (m, 1H, H c ~ )4.42 , ppm (m, 1 H, Hg13,5.21 (m, 1H, He), 7.2 ppm (bs, 1 H, Ha); 13C N M R (CDC13) (24) 6 14.06 (CJ, 18.42-31.90 (C.,), 22.66 (Cf), 24.72 (Cp sn-2), 24.86 (Cp sn-1), 34.07 (C, sn-l), 34.13 (C, sn-2)) 54.62 (m, Cm),60.05 (d, Jp.c 6.0 Hz, Ccz), 62.43 (8, C J, 64.30 (d, Jp.c 4.9 Hz, C d , 66.15 (d,JPC5.5 Hz, Ccl), 69.94 (m, bd,173.13 (OCO sn-2), 173.48 (OCO sn-1), 176.12 ppm (COOH); (+) FAB MS (70

of 6 7.26 ppm for 'H and 6 77.0 ppm for 13C relative to tetramethylsilane. One-dimensional ,'P NMR spectra were obtained with a 12-ps 9 0 O pulse, 8K data points, 35.7-kHzspectral width, inverse-gated broad-band proton decoupling with a RD of 15 s, and -3800 transients. Vesicles in DzO (0.8% NaC1, 10 mM Tris, pH 6.8) were placed in a 10 mm 0.d. tube with a central capillary tube containing 100 mM Na3P04as an external standard, which was arbitrarily assigned a chemical shift of b 0.00 ppm. PrC1, dissolved in the deuterated buffer was added to the vesicles to obtain final concentrations of 5 and 10 mM Pr3+. Hydrogen Peroxide. For quantitation of HzOz,p-hydroxyphenylacetic acid (2 mL, 1.5 mM) buffered with potassium phthalate (50 mM, pH 6.35) was added to 1mL of the ozonized vesicles. Fluorescence was measured (Aex = 320 nm and A,, = 400 nm) 1 min after addition of horseradish peroxidase (583 purpurogallin units). Standard solutions of H20zwere calibrated by titration with aqueous KMnO4 For isotopic analysis, H202 was oxidized to O2 with KMn04 (22). The Ozwas adsorbed on 5-A molecular sieves at liquid nitrogen temperature and desorbed into the inlet of a Delta E isotope ratio mass spectrometer (FinniganMAT, San Jose, CA). The abundanm of '60'60, '60'80, and '80'80 were determined from the intensities of the ion currents at m / z 32, 34, and 36, respectively. l-Palmitoyl-2-(9'-oxononanoyl)-sn-glycero-3-phosphocholine (PN'PC). Ozone (0.6 mol equiv) was added to vesicles (50 mg of DPPC and 50 mg of POPC) in 10 mL of PBS at 40 OC. Peak 2 (seeResults and Discussion) was isolated from the product mixture by HPLC, yielding a waxy solid. 'H NMR (CDC13)(23)

H,C'

-

y

41 2

+

0

V

-o",'o

H,c/"\cH,

7

eV, glycerol) (25,261 m / z 666 (6, [M + HI), 496 (1, [M H C9H14031), 480 (1, [M + H - CgHi4041), 428 (1, [M + H C15H31011),412 (1, [M + H - C1BH30021);CAD of m/z 666 at constant EIE, 496,480,428, and 412; (+) FAB HRMS [70 eV, glycerol (1:1)] m / z 666.4310 [666.4346 calculated for C s w O l $ (231. 1-Palmitoyl-2-(9'- hydroxy-9'-hydroperoxynonanoy1)-snglycero-3-phosphocholine(PNSPC). Hydrogen peroxide, prepared by vacuum distillation of 30% HzO2,was added to dry CH2Cl2,and the mixture was stirred rapidly for 1 h. When the stirring was stopped, excess H202 (density 1.4422) settled to the bottom of the flask, allowingremoval of saturated CH2C12. PNIPC (10 mg,15.4 pmol) was added to 200 mL of the saturated CHZClz, and the solution was stirred for 120 min at room temperature. Solvent was removed by rotary evaporation. Traces of CHzC12 and H20zwere removed in high vacuum, yielding a waxy solid.

V H - . o

-0' "0

428

nu f

u

'CH,

a

81

II

g,g2g,

i

l

l

1

HOO - , o

I

V

m

I

6 0.88 (t, 3 H, HJ, 1.27-1.30 (m, 30 H, H.J, 1.57 (8, 6 H, Hp),2.36 (s,4 H, Ha), 2.42 (dt, 2 H, Hd), 3.36 (8, 9 H, Hm),3.80 (m, 1 H, Hcl), 3.93 (s, 2 H, Hg3),4.14 (m, 1 'H, H,J, 4.31 (m, 1H, H c ~ ) , 4.42

(8,

w m

'H NMR (CDCl,) (23) 6 0.88 (t,3 H, HJ, 1.27-1.30 (m, 32 H, 1.45 (m, 2 H, HUT),1.57 (m, 4 H, Hp), 2.36 (m, 4 H, Ha), 9 H, Hm),3.80 (m, 1H, HcJ, 3.93 (m, 2 H, H,,), 4.14 (m, 1 'H, H,,), 4.31 (m, 1H, Hcz), 4.42 (m, 1H, H 5.21 (m, 1H, H 5.1 ppm (m, 1 H, Ha); 13C NMR (CDb13) (24) b 14.06 (&), 18.42-31.90 (C.,), 22.66 (Cf), 24.72 (Cp Sn-2), 24.86 (Cp en-1), 34.07 (C, sn-I), 34.13 (C, sn-2), 43.62 (Cd), 54.62 (m, CJ, 60.05 (d, J p e 6.0 Hz, Ccz), 62.43 ( 8 , Cgl),64.30 (d, Jp-c 4.9 Hz, C 31, 66.15 (d, Jp.c5.5 Hz, Ccl), 69.94 (m, C 1,101.54 [CH(OH)(ObH)],173.13 (OCO sn-2), 173.48 ppm (OC8sn-1); (-) FAB MS (70 eV, glycerol)

1H, H ',), 5.21 (m, 1H, He), 9.80 ppm (t, 1 H, H*); 13C

NMR (CDCI,) $24)6 14.06 (CJ, 18.42-31.90 (cy),22.66 (Cf), 24.72 (Cp sn-2), 24.86 (Cp sn-1), 34.07 (C, sn-11, 34.13 (C, sn-2),43.62 (Cd), 54.62 (m, Cm),60.05 (d, Jp.c 6.0 Hz, CCJ, 62.43 (8, C J, 64.30 4.9 Hz, C ,), 66.15 (d, JP.c 5.5 Hz, Ccl), 69.94 fm, C 2), (d, Jp.c 173.13 (OCO sn-2!, 173.48 (OCO sn-1), 202.64 ppm (CHO); ?+)

/o

H,C\:A./O\ H,C'

-0' '0

'CH,

a,

7 : /I

396

V

MW 6 4 9

41 2

FAB MS (70 eV, glycerol) (25,26) 650 (6, [M + HI), 496 (1, [M + H - CgH1502]), 480 (1, [M + H - CgH1503]), 412 (1, [M + H - C16H310]),396 (1, [M + H - C16H3102]);CAD of m / z 650 at constant E / E , 496,480,412, and 396; (+) FAB HRMS [70 eV, glycerol-thioglycerol (1:1)] 650.4447 [650.4397 calculated for C33Hd09P (231. l-Palmitoyl-2-(9'-carboxynonanoyl)-sn -glycero-3phosphocholine (PNzPC). Ozone (2.0 mol equiv) was added to vesicles (50 mg of DPPC and 50 mg of POPC) in 10 mL of PBS at 40 "C. Peak 1 (see Results and Discussion) was isolated from the product mixture by HPLC, yielding a waxy solid. 'H NMR V n

v u

,0 HI C

ct

H,C'

'CH,

\

~ -0"b

u

H, C""CH,

gsg2g1

8

0

\ 0

p

~

8'

O

& I

7

m

(CDCl,) (23) 6 0.88 (t, 3 H, H,J, 1.27-1.30 (m, 30 H, HJ,1.57 (m, 6 H, H&, 2.36 (m, 6 H, Ha), 3.36 (8, 9 H, H,), 3.80 (m, 1 H, HcJ,

/

'd

MW 6 8 3

684 (0.6, [M + HI), 664 (60, [M - H - HZO]), 650 (22, [M - H Oz]),619 (4, [M - H - HzO - 3CH,]), 605 (15, [M - H - HzO (CHd3N1,480 (4, [M - H - ~ ~ ~ O 431 S I(4,) [M , - H - Ci&nOzI).

u

O

431

577l

OH,

~

/-o"'b

I

Results and Dlscussion Vesicles. To avoid potential anomalies resulting from ozonolysis of POPC in multilamellar vesicles, we used a procedure reported to yield unilamellar vesicles (17). We determined the percentage of unilamellar vesicles in our preparations by 31PNMR spectrometry. The 31PNMR spectrum O ~of vesicles prepared in D20contained a single resonance (Figure 1A). Lanthanide ions bind to the phosphate moieties of lipids in the outer leaflet of the vesicles only, inducing a broadening and downfield shift of these resonances in the NMR (28). Addition of PrC13 (5 and 10 mM) to the vesicle preparation produced two separate resonances for the phosphate moieties of lipids

Chem. Res. Toxicol., Vol. 5, No. 1, 1992 137

Reaction of Ozone with Phospholipids in Vesicles inner and outer

A

inner

B

O

-

W

R

0

II

/'\OH 0

II

HOO, /

7

,

,

1

1

,

1

1

1

1

1

,

,

40 30 20 10 0 -10 - 2 0

1

,

1

~

1

,

1

,

1

~

,

,

,

,OH 'H

I

40 30 20 10 0 -10 -20

Figure 1. 31PNMR spectra of phospholipid vesicles at 50 OC: 0 mM P?+ with external standard (A); 10 mM Pr3+without external standard

C

(B).

2

,

.

r I . . . , I . . . . , . . . . , , I . . I 200

150

100

50

0

ppm

Figure 3. 13C NMR spectrum of the product mixture from reaction of 1 mol equiv of O3 with POPC in unilamellar phoepholipid vesicles (1:l DPPC/POPC). The CHO group of PNIPC produced the resonance at 6 202.6 ppm, the COOH group of PN2PC produced the resonance at 6 176.1 ppm, and the CH(0H)(OOH) group of PNSPC produced the resonance at 6 101.5 ppm.

4

I

I

I

I

I

0

10

20

30

40

retention time, m h

Figure 2. Chromatogram of ozonolysis products of POPC (1mol equiv of 0,)in unilamellar phospholipid vesicles (1:l DPPC/ POPC): peak 1, PN2PC; peak 2,PN,PC; peak 3,PN,PC; peak 4,DTPC (internal standard); peak 5, DMPC (extemal standard); peak 6,DPPC; peak 7,POPC.

on the inner and outer surfaces of the vesicles (Figure 1B). The ratio of the areas of these signals was 4258 (outer: inner), with no difference between 5 or 10 mM P P . For vesicles with a diameter of 0.1 pm, this ratio indicated that 285% of the vesicles were unilamellar. Tris (10 mM) was substituted for phosphate as buffer for this measurement because P r P 0 4 has low solubility in water (29). Ozonolysis Products of POPC. A chromatogram of ozonolysis produds of POPC (1mol equiv of 0,)in vesicles at 45 "C is shown in Figure 2. No reaction occurred when the vesicles were at room temperature. Compounds 1-3 were ozonolysis products, compounds 4 and 5 were the internal (DTPC) and external (DMPC) standards, respectively, compound 6 was DPPC, and compound 7 was unreacted POPC. Fractions corresponding to 80% of the areas of compounds 1 and 2 were collected for analysis. Compounds 1 and 2 were identified from their positive FAB mass spectra, one-dimensional 'H NMR spectra,

two-dimensional 'H COSY spectra, and one-dimensional 13CNMR spectra: compound 1was PN2PC and compound 2 was PNIPC. Peak 3 appeared sporadically, and an amount sufficient for analysis could not be collected. However, authentic PN3PC, prepared by reaction of PNIPC with H202, coeluted with compound 3. Since elimination of H202 from hydroxy hydroperoxides is acid-catalyzed (30),it is likely that the buffer in the chromatographicsolvent system accelerated decomposition of PN3PC to PNIPC and H202by general-acid catalysis. Compound 3 therefore was tentatively identified as PN3PC. Direct evidence of hydroxy hydroperoxide was obtained from a 13CNMR spectrum of the product mixture. A 13C NMR spectrum of the lipid extract from reaction of 0.75 mol equiv of O3with POPC in vesicles is shown in Figure 3. The structures of PNIPC, PN2PC, PN3PC, and POPC are similar, differing only in the functional group at the CY of the side chains in the sn-2 positions. The 13CNMR spectra of these compounds were correspondingly similar. Carbon atoms in the aliphatic side chains and the choline and glycerol groups produced resonances between 6 14 and 71 ppm. The COO groups of the side chains (the ester bonds) produced resonances at 6 173.52 and 173.23 ppm, irrespective of the structure of the side chain in the sn-2 position. The structure of the side chain in the sn-2 position could not be deduced from the spectra in these regions. Resonances at 6 101.54, 173.05, and 202.71 ppm correspond to the CH(OH)OOH),COOH, and CHO group at the termini of the nonanoyl chain in the sn-2 positions of PN3PC, PN,PC, and PNIPC, respectively. The resonance at 6 176.12 ppm was produced by the CH-CH group of residual POPC. When O3 was added to POPC in dry CH2Cl2,the 13C NMR spectrum of the product mixture contained resonances at 6 103.99,104.04,104.09, and 104.15 ppm, which is characteristic of an unsymmetrically substituted ~ z o n i d e . Ozonide ~ was not detected

138 Chem. Res. Toxicol., Vol. 5, No. 1, 1992

Santrock et al.

Table I. Oxygen Isotope Abundances in Ozonolysis Products of 1-Palmitoyl-2-oleoyl-sn -al~cero-3-~hos~hocholine mole fraction

HzOzb

aldehyde" reagents 180 d

160:/1803g

carboxylic acid'

160

'80

160160

160180

180'80

160160

160180

180'80

0.94e/0.55f ndh

0.06e/0.4d nd

0.02 0.46

0.03

0.95 0.50

0.09 nd

0.42 nd

0.43 nd

0.04

Only PNIPC was measured. The isotopic species of HzOz were measured as O2 produced by reduction of HzOz by KMnO,. Only PNzPC was measured. dThe O3was 96% oxygen-18. eDetermined from the positive FAB mass spectrum of PNIPC. fDetermined by GC/MS (positive methane C1) of bis-TMS-nonanediol, produced by reduction of PNIPC with LAH and derivatization with N-(trimethylsily1)imidazole. gThe mixture was 52% I6O3and 48% h n d = not done. 1 .o 1

,

0.8 t

0.0

A

*

0.5

1.0

1.5

2.0

2.5

mole fraction ozone

Figure 4. Material balance for reaction of O3with POPC in uniiamellar phospholipid vesicles (1:l DPPC/POPC). All quantities were normalized to the initial amount of POPC: POPC ( 0 ) ;PN,PC (m); PNzPC ( 0 ) ; H,Oz (0). The curves represent least-squares approx'lmations of fhese data using polynomial equations of degree 2: POPC (A); PNIPC (B);HzOz (C); PNzPC (D).

in the 13CNMR spectrum of ozonolysis products of POPC in vesicles. Material Balance. Figure 4 shows a plot of the mole fraction of POPC, PNIPC, PN,PC, and HzOzas a function of the mole fraction of O3 added. All quantities were normalized to the initial amount of POPC. The ratio of PNIPC to PN3PC was highly variable. Since the buffer in the chromatographic mobile phase catalyzed decomposition of PN3PC to PNIPC, points corresponding to PNIPC in Figure 4 are the sum of PNIPC and PN3PC, which is a measure of the amount of POPC consumed by 03.The response factor determined for PNIPC was used for PN3PC. The curves in Figure 4 are polynomial fits to these data. Nonanal and nonanoic acid were also detected

In unpublished observations, the 'H NMR spectrum of the ozonide of methyl oleate in CD2Cl, contained triplets at 6 5.12 and 5.17 ppm, corresponding to the trans and cis isomers of the ozonide, respectively. The 13C NMR spectrum of the same compound contained resonances at 6 104.59, 104.65, and 104.73 ppm. Four signals were expected for an unsymmetrically substituted ozonide. A lH-I3C cross-correlation experiment contained triplets in the 'H frequency domain centered at (5.12, 104.65), (5.12,104.73),(5.17,104.59),and 617,104.65). The first number in each pair correspondsto the 'H chemical shift, and the second number corresponds to the 13Cchemical shift. Thus, the two ring carbons of the unsymmetrically substituted ozonide produced separate resonances in the 13CNMR spectrum. The upfield signal from the trans isomer overlapped the downfield signal of the cis isomer at 6 104.65 ppm, apparently producing only three signals. The ozonide of POPC was prepared by adding O3to POPC in dry CH2C12 The I3C NMR spectrum of the crude ozonide of POPC in CDC13contained four resolved resonances at 6 103.99,104.04, 104.09, and 104.15 ppm, corresponding to the Cg and Clo carbons of the cis and trans isomers of the unsymmetrically substituted ozonide. In the 'H NMR spectra, theg, proton produced a broad resonance at 6 5.21 ppm (see Experimental Section) which obscured signals from protons on the trioxolane ring. No purification or further characterization of the ozonide of POPC was attempted. We determined the absence of ozonide from lack of signals in the 13CNMR spectrum at 6 103.99,104.04,104.09,and 104.15 ppm.

in the product mixture, but not quantified because the response of the detector to these compounds was low. The sequence of disappearance of POPC and appearance of products indicated that two reactions occured (1) ozonolysis of POPC yielded PNIPC or PN3PC and (2) ozonolysis of PNIPC yielded PN2PC. With addition of 10.5 mol equiv of 03,ozone reacted with only POPC. However, with addition of >0.5 mol equiv of 03,ozone reacted with both POPC and PNIPC. The stoichiometries of these reactions were determined from the polynomial fits to these data. For ozonolysis of POPC, 1 mol of O3 reacted with 1 mol of POPC, producing 1 mol each of (PNIPC + PN3PC) and HzOz. For ozonolysis of PNIPC, 1mol of O3reacted with 1mol of PNIPC, producing 1mol of PN2PC. The products (PNIPC + PN3PC) and PNzPC accounted for 296% of POPC consumed by O3with addition of 11.5 mol equiv of Os, but accounted for only 285% of POPC consumed with addition of >1.5 mol equiv of 03. Incorporation of Oxygen-18 into Products. The sources of oxygen in the aldehyde moiety of PNIPC, the carboxylic acid moiety of PN,PC, and HzOzwere determined by measuring incorporation of l80from 1803 into these products (Table I). Isotopic abundances in PNIPC and PNzPC were determined from the positive FAB mass spectra of these compounds. The molecular ion cluster of PNIPC contained [M H]+ (loo%), [M - H]+ (20%), and traces of [Me]+. Isotopic abundances were deconvoluted from the overlapping isotopic envelopes of [M + H]+ and [M - H]+ using a model that allowed the isotope abundances in each ion to vary independently. To correct for the contribution of [(l80)M - H]+ to [(l60)M+ HI+, we assumed that the ratio [M H]+/[M + H]+ was the same for labeled and unlabeled with POPC yielded PNIPC material. Reaction of 1803 which contained 6% l80in the 1-oxygen position (Table I), compared to the natural abundance of 0.2% l80(27). However, when the 9'-oxononanoyl group of PNIPC was converted to 1,9-nonanediol by reduction of PNIPC with LAH immediately after ozonolysis, the tetramethylsilyl (TMS) derivative of l,9-nonanediol was 45% l80in one oxygen position. Variabilities in the spectral background and in the ratio [M - H]+/[M HI+ were the largest contributions to uncertainty in this calculation. The mass spectrum of PN2PCdisplayed a nonrandom distribution of l80in the two carboxylic acid oxygen positions: 9% C160160H, 42% Cl6OlsOH, and 43% C180180H(Table I). Isotopic abundances in PN2PC were estimated using a model that allowed l80in the two positions to vary independently. As before, we assumed that [M - H]+/[M + H]+ was the same for labeled and unlabeled material. These calculations showed that one position contained 56% l80, while the other position contained 86% l80.Unlike the aldehyde moiety of PNIPC, the rate of exchange of oxygen isotopes between water and carboxylic acids is slow (31-33), and loss of isotopic information with time was not a problem with PN2PC.

+

+

Reaction of Ozone with Phospholipids in Vesicles

The source of oxygen in H202was determined using 1803 or a 1:l mixture of 1603and l8OB. Excess KMn04 was added directly to the product mixture, and isotopic species of the O2evolved were quantitated from ita electron impact mass spectrum. Oxidation of H202by KMn04 yields O2 in which both oxygen atoms are derived from the same molecule of H202(34-36). Moreover, the rates of exchange of oxygen between H202,H20, and O2 are slow (34,361. Therefore, the isotopic composition of the O2was identical to that of ita precursor H202. When 1803 was a reactant, the H202was 96% l8O2(Table I). When a 1:l mixture of 1603 and '80, was a reactant, the H202was 46% HI6O160H, 4% H160180H,and 50% H180180H (Table I). Mechanism of Ozonolysis. The products identified here, the sequence of appearance of products, and the into aldehyde, pattern of incorporation of l80from 1803 carboxylic acid, and H202are consistent with the mechanism originally proposed by Tiege et al. (IO). Ozonolysis of POPC in a lipid bilayer yielded an aldehyde (PNIPC) and a hydroxy hydroperoxide (PN3PC). The hydroxy hydroperoxide eliminated H202to yield a second aldehyde (also PNIPC). Upon further ozonolysis, the aldehydes were oxidized to the corresponding carboxylic acids (PN2PC). Although PN3PC was not isolated, authentic PN3PC coeluted with peak 3 in the chromatogram of the product mixture. Moreover, hydroxy hydroperoxide was clearly present in the 13CNMR spectrum of the product mixture. PNIPC would have accumulated in the membrane as the hydroxy hydroperoxide decomposed and H202diffuses away from the membrane. As observed, PNIPC always was in excess of PN3PC. The carboxylic acid PN2PC was an ozonolysis product of the aldehyde PNIPC. The pattern of incorporation of l80from 1803 into these products is consistent with this mechanism. Consider Scheme I. Addition of O3 to the carbon-carbon double bond of POPC yields aldehyde 2 and carbonyl oxide 3. Oxygen in the CHO group of aldehyde 2 and in the COOgroup of the carbonyl oxide is derived from 03.Addition of H20to the carbonyl oxide yields hydroxy hydroperoxide 5. Oxygen in the OOH group is derived from 03,and oxygen in the OH group is derived from H20. Elimination of H202from hydroxy hydroperoxide appears to occur by proton transfer from the OH group of the hydroxy hydroperoxide to H 2 0 and from H 2 0 to the OOH group of the hydroxy hydroperoxide, possibly through a cyclic transition state involving one or more hydrogen-bound water molecules (30). According to this scheme, H202is

derived from the OOH group of hydroxy hydroperoxide, with both oxygen atoms in a molecule of H202 coming from the same molecule of 03.Oxygen in the CHO group of aldehyde 6 is derived from the OH group of the hydroxy hydroperoxide, and ultimately from H20. As shown in Scheme 11, ozonolysis of the aldehydes yields carboxylic acids (37-42). The reaction proceeds by insertion of O3into the C-H bond of the aldehyde, yielding a hydrotrioxide (39-41). The hydrotrioxide is probably stabilized by an intermolecular hydrogen bond between the hydrotrioxide hydrogen and the carbonyl oxygen, forming a six-membered ring which facilitates elimination of singlet O2 from the OOOH group (40-42). The carboxylic acid then retains one oxygen atom from the pre-

Chem. Res. Toricol., Vol. 5, No. 1, 1992 139 '0 ' 0 kH>0

-o'w

/a

2

*OH

7/& -

cursor aldehyde and derives the other oxygen atom from 03. Aldehyde 2 is the precursor to carboxylic acid 7, whereas aldehyde 6 is the precursor to carboxylic acid 8. The pattern of incorporation of l80from la03into PNIPC, PN2PC,and H202(Table I) was consistent with Schemes I and 11. When 1803 was a reactant, PNIPC was 45% [180]PN1PC (aldehyde 2) and 65% [160]PN1PC. Presumably, ["jO]PN,PC was aldehyde 6 which derived ita oxygen from H20, but verification using H,l80 as solvent was too costly and was not done. The concentration of 6 approaches that of 2 as hydroxy hydroperoxide decomposed. In the limit, PNIPC would be an equimolar mixture of aldehydes 2 and 6, where 2 contained l80from 1803.As our difficulty in collecting sufficient PN3PC demonstrated, any manipulation resulted in significant decomposition of the hydroxy hydroperoxide. Thus, we observed a nearly equimolar mixture of aldehydes 2 and 6 in Table I. Ozone was the only source of oxygen for H202;hydrogen peroxide was 95% H1a0180Hwhen 1803 was the reactant. Both oxygen atoms in a molecule of H202 were derived from the same molecule of 0,;hydrogen peroxide was 46% H21602and 50% H21802when a 1:l mixture of 1 6 0 3 and lag3was the reactant. When 1803 was a reactant, PN2PC was 42% C160180H(carboxylic acid 8) and 43% C1s0180H (carboxylic acid 7). This isotopic distribution indicated that one of the oxygen atoms was derived from ['sO]PN,PC (56% lag)and the other oxygen atom was derived from O3 (86% l80).Both oxygen atoms in the COOH group of 7 were derived from laOB. In contrast, one oxygen atom in the COOH group of 8 was derived from 03,while the other oxygen atom was derived from H20. The amount of l 6 0 observed in PNIPC was less than the value of 50% predicted from Scheme I. In water, aldehydes are in equilibrium with the corresponding geminal diols according to the following reaction: RCHO H20 + RCH(OH)2 (43-53). Successive hydration of the aldehyde and dehydration of the gem-diol distribute oxygen isotopes randomly between the populations of aldehyde and water (43, 46, 47). The abundance of l80in each species is controlled by the relative concentrations of each species and by the intrinsic equilibrium isotope effect for the reaction. Since H 2 0 is in great excess, l80is lost from the aldehyde position. The reaction is acid-catalyzed (&), and the transition state contains two or three auxiliary water molecules (44,52),which act to lower the activation energy for the reaction (52). Incorporating the aldehyde into a micelle shifts the equilibrium toward the aldehyde (53) and may slow the rate of exchange because of a decrease in the activity of water and changes in the activity of catalytic species at the micelle surface. But hydration of the aldehyde still occurs. In contrast, hydroxy groups do not undergo exchange of oxygen with H20 under these conditions. In the lipid bilayer, PNIPC lost l80to H20. Reduction of the aldehyde by LAH trapped l80as OH in 1,9-nonanediol. Since some exchange occurred prior to reduction, bis-TMS-nonanediol always contained less l8O than was incorporated into the aldehyde during ozonolysis. The abundance of l80in 1,9-nonanediolwas closer to that in PNIPC immediately after ozonolysis of POPC. Two scenarios can explain how carbonyl oxide, produced in the hydrophobic bilayer, reacted with H2O. First, the concentration of H 2 0within the hydrophobic bilayer may

+

Santrock et al.

140 Chem. Res. Toxicol., Vol. 5, No. 1, 1992

have been sufficient for the reaction to occur within the membrane, at the site of ozonolysis. Alternatively, polar-nonpolar interactions between the charged carbonyl oxide and the hydrocarbon side chains may have forced the carbonyl oxide to the interface region, where it was in contact with H20. To distinguish between these two scenarios, we would need to know the concentration of H20 in the lipid bilayer accurately, the second-order rate constants for reaction of carbonyl oxide with aldehyde and with H20, and how the lipid bilayer affects mobility and reactivity. Although a lipid bilayer is permeable to water, the steady-state concentration of water within the bilayer is low. Neutron diffraction studies of phospholipids hydrated in 2H20indicated that bulk water penetrated into the region of the phospholipid head group, but not into the hydrocarbon core (54-56). The detection limit in these studies was estimated to be