Formation of isoprostane bicyclic endoperoxides from the autoxidation of cholesteryl arachidonate

Article abstract of DOI:10.1021/ja0201092

Autoxidation of polyunsaturated fatty acids and esters leads to a complex mixture containing hydroperoxides and cyclic peroxides. Prostaglandin bicyclic endoperoxides have been detected from the autoxidation of cholesteryl arachidonate by LC-MS and GC-MS

Full text of DOI:10.1021/ja0201092

Published on Web 06/11/2002
Formation of Isoprostane Bicyclic Endoperoxides from the
Autoxidation of Cholesteryl Arachidonate
Huiyong Yin,† Christine M. Havrilla,† Jason D. Morrow,‡ and Ned A. Porter*,†
Contribution from the Department of Chemistry, Medicine and Pharmacology, and the Center in
Molecular Toxicology, Vanderbilt UniVersity, NashVille, Tennessee 37235
Received January 22, 2002
Abstract: Autoxidation of polyunsaturated fatty acids and esters leads to a complex mixture containing
hydroperoxides and cyclic peroxides. Prostaglandin bicyclic endoperoxides have been detected from the
autoxidation of cholesteryl arachidonate by LC-MS and GC-MS techniques. All four possible types (I-IV)
of bicyclic endoperoxides have been found starting from different regioisomeric hydroperoxides of cholesteryl
arachidonate. Furthermore, the stereochemistry of Type IV bicyclic endoperoxides has been determined
by conversion to pentafluorobenzyl (PFB) ester, trimethylsilyl (TMS) derivatives, and comparison with
synthetic standards by the use of GC-MS. All eight possible diastereomers of the derivatized isoprostanes
were observed and were separated by gas chromatography. The bicyclic endoperoxides with the two alkyl
chains syn on the cyclopentane ring were formed preferentially to those with anti configuration, a result
anticipated from earlier work. Substantial amounts of the anti-substituted isoprostanes, including PGF2R
,
were, however, observed in the product mixture.
Scheme 1
Introduction
Free radical-initiated lipid peroxidation has been linked to
many human diseases such as atherosclerosis, cancer, and
neurodegenerative disorders.1 In particular, oxidative modifica-
tion of the lipid components of low-density lipoprotein (LDL),
the major carrier of cholesteryl esters in human blood, has been
attributed to the pathogenesis and progression of atherosclerosis.2
Deposition of the oxidized cholesteryl esters in cells and vascular
tissues is a pathological hallmark of atherosclerotic lesions.3
However, the relevant lipid oxidation products and the mech-
anism of the LDL oxidation process have not been well-defined.
It has been shown that oxidized lipids extracted from the
atherosclerotic lesions are rich in cholesterol and cholesterol
esters.4 Therefore, characterization of the oxidation products of
cholesterol esters may have relevance to this disease. Cholesteryl
linoleate (Ch 18:2) and cholesteryl arachidonate (Ch 20:4)
comprise a substantial fraction of the neutral lipids present in
LDL.
The oxidation of cholesteryl linoleate leads to several primary
hydroperoxides. However, the mixture of oxidation products
* Corresponding author. E-mail: n.porter@vanderbilt.edu. Phone: 615-
343-2693. Fax: 615-343-5478.
of cholesteryl arachidonate is extremely complicated because
insertion of molecular oxygen may occur at C5, C8, C9, C11,
C12, or C15 of the 20 carbon fatty acid chain, Scheme 1.5
Furthermore, the resultant peroxyl radicals can undergo reactions
to give a complex mixture of primary hydroperoxides and cyclic
peroxides with dozens of regioisomers and diastereomers
possible. Conventional techniques of product identification
† Department of Chemistry.
‡ Medicine and Pharmacology.
(1) (a) Chisolm, G. M.; Steinberg, D. Free Radical Biol. Med. 2000, 28, 1815-
1826. (b) Nicholson, A. C.; Hajjar, D. P. Am. Sci. 1995, 83, 460-470.
(2) (a) Patel, R. P.; Moellering, D.; Murphy-Ullrich, J.; Jo, H.; Bechman, J.
S.; Darley-Usmar, V. M. Free Radical Biol. Med. 2000, 28, 1780-1794.
(b) Dhaliwal, B. S.; Steinbrecher, U. P. J. Lipid Res. 2000, 41, 1658-
1665.
(3) (a) Heineke, J. W. Atherosclerosis 1998, 141, 1-15. (b) Berliner, J.;
Leitinger, N.; Watson, A.; Huber, J.; Fogelman, A.; Navab, M. Thromb.
Haemostasis 1997, 78, 195-199.
(4) (a) Carpenter, K. L. H.; Tarlor, S. E.; van der Veen, C.; Williamson, B.
K.; Balloantine, J. A.; Mitchinson, M. J. Biochim. Biophys. Acta 1995,
1256, 141-150. (b) Suarna, C.; Dean, R. T.; May, J.; Stocker, R.
Arterioscler. Thromb. Vasc. Biol. 1995, 15, 1616-1624.
(5) (a) Porter, N. A. Free Radicals Biol. 1980, 4, 261-294. (b) Porter, N. A.;
Funk, M. O. J. Org. Chem. 1975, 40, 3614-3615. (c) Porter, N. A.; Funk,
M. O.; Gilmore, D.; Isaac, R.; Nixon, J. J. Am. Chem. Soc. 1976, 98, 6000.
(d) Marnett, L. J. Curr. Opin. Chem. Biol. 2000, 4, 545-552.
9
10.1021/ja0201092 CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 7745-7754
7745

A R T I C L E S
Yin et al.
Scheme 2
(purification by chromatography, spectroscopic characterization,
and conversion to or comparison with known compounds) are
not possible for the analysis of the oxidation products of
cholesteryl arachidonate.6 The challenges of chromatography
of such complex product mixtures alone makes analysis by
classical methods unrealistic. In response to this analytical
problem, we have developed methods for the analysis of
complex mixtures of peroxide compounds based upon Ag+
coordination ionspray mass spectrometry. This technique relies
on coordination of the soft Lewis acid, Ag+ to soft Lewis base
sites such as carbon-carbon double bonds in unsaturated
molecules and it provides characteristic fragmentation patterns
for peroxidic compounds.
The bicyclic endoperoxide 1b (prostaglandin G2, PGG2) is
formed in the oxidation of arachidonic acid promoted by
cyclooxygenases (COX) enzymes.7 Formation of analogous
endoperoxides has been implied in the nonenzymatic free radical
oxidation of arachidonic acid or its esters and these endoper-
oxides are presumed intermediates in the formation of isopros-
tanes, compounds that have been used as a reliable index for
oxidative injury in vivo.5,7,8
Given the fact that PGG2 and its closely related analogue
PGH2, 1c, are involved in important physiological events such
as smooth muscle contraction and the inflammatory process, it
seems likely that the many other regioisomeric and stereoiso-
meric endoperoxides formed in the free radical oxidation of
arachidonates will have their own biological activities. Com-
pounds derived from the endoperoxide mixtures that are
analogous to compounds such as 1d (PGF2R) and other PG’s
and thromboxanes may have interesting physiological properties
as well. Furthermore, the mixture of regioisomers and stereo-
isomers having structures analogous to PGF2R make up the class
of compounds known as isoprostanes.
LDL is esterified primarily to phospholipids and cholesterol
esters and the formation of PGG2-like bicyclic endoperoxides
from cholesteryl esters (or phospholipids) has not been studied
due to the complexity of the product mixture and the lack of
appropriate analytical techniques.9 Both Ag+ coordination
ionspray mass spectrometry and GC-MS of suitable isoprostane
derivatives made from the isolated endoperoxides were used to
provide information about the distribution of products formed
in lipid peroxidation of cholesteryl arachidonate.
Results
Autoxidation of cholesteryl arachidonate in the presence of
good hydrogen atom donors such as 1,4-cyclohexadiene or
methyl Trolox leads to six hydroperoxides of cholesteryl
arachidonate: Ch-5-HPETE, Ch-8-HPETE, Ch-9-HPETE, Ch-
11-HPETE, Ch-12-HPETE, and Ch-15-HPETE.6a These regio-
isomeric hydroperoxides of cholesteryl arachidonate can be
separated by semipreparative HPLC except for the 8 and 9 pair
of regioisomers. The separated regioisomeric hydroperoxides
can, themselves, serve as a source of specific peroxyl radicals
when reacted under free radical autoxidation conditions. In this
way, the complexity of the arachidonate autoxidation problem
can be simplified, the product mixture that results from reaction
of an individual arachidonyl peroxyl radical being simpler than
the mixture that results from reaction of the six peroxyl radicals
formed from arachidonate.
Characterization of Isoprostane Bicyclic Endoperoxides
of Cholesteryl Esters by LC-MS Techniques (Ag+ Coordina-
tion Ionspray Mass Spectrometry). (a) Isoprostane (Pros-
taglandin) Bicyclic Endoperoxides from Ch-15-HPETE
(Type IV). The technique of Ag+ coordination ionspray MS
has proved to be a powerful tool for analyzing the oxidation
mixture derived from Ch-15-HPETE. Coupled with normal
phase HPLC, this CIS-MS technique provides information about
different classes of peroxides that result from peroxidation.6a,10
The possible oxidation products from Ch-15-HPETE are il-
lustrated in Scheme 2. According to the mechanism shown in
Scheme 2, a peroxyl radical forms by H-abstraction from the
hydroperoxide of Ch-15-HPETE. This peroxyl radical can
undergo â-fragmentation and O2 addition to give a peroxyl
radical at C11 that can undergo double cyclization to give the
bicyclic endoperoxide, 2. Other cyclic peroxide products can
We report here on the bicyclic endoperoxides formed in the
oxidation of cholesterol arachidonate. Arachidonate present in
(6) (a) Havrilla, C. M.; Hachey, D. L.; Porter, N. A. J. Am. Chem. Soc. 2000,
122, 8042-8055. (b) Khan J. A.; Porter, N. A. Angew. Chem. 1982, 94,
220-221.
(7) (a) Needleman, P.; Kulkarni, P. S.; Raz, A. Science 1977, 195, 409. (b)
Kuehl, F. A.; Humes, J. L.; Egan, R. W.; Ham, E. A.; Beveridge, G. C.;
Van Arman, C. G. Nature 1977, 265, 171. (c) Needleman, P.; Moncada,
S.; Bunting, S.; Vane, J.; Hamberg, M.; Samuelsson, B. Nature 1976, 261,
558. (d) Funk, C. D. Science 2001, 294, 1871-1875.
(8) (a) Morrow, J. D.; Hill, K. E.; Burk, R. F.; Nammor, T.; Badr, K. F.;
Roberts, L. J. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 9383-9387. (b)
Pratico, D.; Lawson, J. A.; Rokach, J.; FitzGerald, G. A. TRENDS Endocrin
Met. 2001, 12 (6), 243-247. (c) Lawson, J. A.; Rokach, J.; FitzGerald, G.
A. J. Biol. Chem. 1999, 274, 24441-24444. (d) Rokach, J.; Khanapure, S.
P.; Hwang, S. W.; Adiyaman, M.; Lawson, J. A.; FitzGerald, G. A.
Prostaglandins 1997, 54, 823-851.
(9) Van Heek, M.; Schmitt, D.; Toren, P.; Cathcart, M. K.; DiCorleto, P. E. J.
Biol. Chem. 1998, 273, 19405-19410.
(10) (a) Bayer, E.; Gfrorer, P.; Rentel, C. Angew. Chem., Int Ed. 1999, 38, 992-
995. (b) Yin, H.; Hachey, D. L.; Porter, N. A. J. Am. Soc. Mass. Spectrom.
2001, 12, 449-455.
9
7746 J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002

Formation of Isoprostane Bicyclic Endoperoxides
A R T I C L E S
Figure 3. Selective reaction monitoring of 2 from m/z 843 to 375.
ment are summarized in Scheme 3. An apparent fragment with
m/z 771 is obtained by the loss of malondialdehyde (MDA), a
well-known process for the endoperoxides.12 The Hock rear-
rangement has also been observed in Ag+ coordination experi-
ments and in this experiment, aldehydes with m/z 743 (R )
Ch) or 375 (R ) H) are obtained.13 The R-cleavage of
hydroperoxides can also occur to give aldehydes with m/z 755
(R ) Ch) or 387 (R ) H). The fragments with m/z 753 and
385 can be rationalized as resulting from dehydration of the
fragment at m/z 771 and subsequent loss of cholesterol. The
fragment with m/z 475.4 corresponds to the loss of cholesterol
from the parent ion and its subsequent dehydration gives m/z
457.3. Overall, the CID experiment supports the notion that
bicyclic endoperoxide compounds are present in the reaction
mixture.
Figure 1. The oxidation products of Ch-15-HPETE detected by direct liquid
infusion with Ag+ CIS-MS.
Selective reaction monitoring (SRM) was carried out by
coupling HPLC to Ag+ CIS-MS, see Figure 3. The SRM
chromatogram from the parent ion m/z 843 to the characteristic
endoperoxide fragment, m/z 375, is shown in this chromatogram.
SRM chromatograms monitoring fragmentation to other char-
acteristic ions from Figure 2 are essentially identical with the
one shown in Figure 3. We conclude that these chromatograms
indicate the formation of several stereoisomeric bicyclic endo-
peroxides in the free radical reactions of Ch-15-HPETE.
Chromatography of the same product mixture with detection
by UV at 234 or 207 nm gave the chromatograms shown in
Figure 4. The endoperoxide structures do not have a conjugated
diene and they are therefore not detected at 234 nm. Other
products formed from the reaction of arachidonate hydroper-
oxides include serial cyclic and monocyclic peroxide compounds
that contain a conjugated diene substructure. These compounds
are detected at 234 nm and peaks in the HPLC-UV (234 nm)
chromatogram shown in Figure 4a have previously been
identified as having the serial cyclic structure.6a
The HPLC-UV (207 nm) chromatogram shown in Figure 4b
does give evidence of compounds having no conjugated diene.
These compounds elute in regions of the HPLC-MS chromato-
gram that correspond to isoprostane endoperoxides, as identified
by SRM. The isoprostane endoperoxides have no UV chromo-
phore above 210 nm and they are therefore detected at 207 nm
but not at 234 nm. Comparison of the chromatograms shown
in Figure 4a,b and the SRM chromatogram shown in Figure 3
Figure 2. CID of Type IV bicyclic endoperoxides, 2.
form.11 Thus, addition of oxygen to the intermediate carbon-
centered radical could give a peroxyl radical that would lead to
the monocyclic peroxides, 3, or the serial cyclic peroxides, 4.
Figure 1 shows the results of a Ag+ CIS-MS direct liquid
infusion (DLI) experiment on the product mixture derived from
exposure of Ch-15-HPETE to a free radical initiator in benzene
under air at 37 °C. The doublet peaks observed are characteristic
of the Ag+ adducts of peroxidic compounds formed in the
reaction. Silver ion adducts appear as doublets due to the fact
that there are two isotopes of Ag 107 and 109, present in a 1:1
ratio. The m/z of 811.7/813.5 corresponds to the Ag+ adduct of
the starting hydroperoxide, Ch-15-HPETE. The m/z 843.5/845.6
corresponds to the Ag+ adducts of either prostaglandin bicyclic
endoperoxide 2 or monocyclic peroxide 3 since they have the
same mass, resulting from addition of one molecule of oxygen
to the starting material. The Ag+ adduct of the serial cyclic
stereoisomeric peroxides 4 appears at m/z 875.5/877.5.
Structural information was obtained by collision-induced
dissociation (CID), using a triple quadrupole mass spectrometer.
The CID experiment was carried out by selecting the parent
ion (m/z 843.4) in the first quadrupole and fragmentation in the
second quadrupole by collisional activation with argon. The third
quadrupole was set to scan all of the resulting fragments. A
CID spectrum of the parent ion m/z 843.4 is shown in Figure
2. The fragmentation patterns consistent with the CID experi-
(12) Chaudhary, A. K.; Nokubo, M.; Reddy, G. R.; Yeola, S. N.; Morrow, J.
D.; Blair, I. A.; Marnett, L. J. Science 1994, 265, 1580-1582.
(13) (a) Hock, H. Angew. Chem. 1936, 49, 595. (b) For a review see: Frimer,
A. A. Chem. ReV. 1979, 79, 359-387. (c) Tanigawa, S.; Kajiware, T.;
Hatanaka, A. Phytochemistry 1984, 23, 2349-2344. (d) Gardner, H. W.;
Planter, R. D. Lipids 1984, 19, 294-298.
(11) (a) Porter, N. A.; Caldwell, S. E.; Mills, K. A. Lipids 1995, 30, 277-290.
(b) Porter, N. A. Acc. Chem. Res. 1986, 19, 262-268.
9
J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002 7747

A R T I C L E S
Yin et al.
Scheme 3
Scheme 4
gives independent support for the formation of isoprostane
endoperoxide structures in the reaction.
(b) Isoprostane Bicyclic Endoperoxides from Regioiso-
meric Ch-HPETEs (Types I-III). As summarized in Scheme
4, H-abstraction from cholesteryl arachidonate can occur at C7,
C10, or C13 (bisallylic positions) and this results in different
regioisomeric hydroperoxides of cholesteryl arachidonate.14 By
analogy to the chemistry observed for Ch-15-HPETE, Type I
isoprostane bicyclic endoperoxides could form from Ch-5-
HPETE and Ch-9-HPETE. Type II and Type III endoperoxides
could form either from Ch-8-HPETE or Ch-12-HPETE. To
provide evidence for formation of these endoperoxides, the 5-
and 12-Ch-HPETEs and the mixture of the 8- and 9-Ch-HPETEs
were purified by semipreparative HPLC and reacted as described
for the 15- and 11-Ch-HPETEs.
The mechanism for formation of the Type IV endoperoxides
from Ch-15-HPETE requires fragmentation of the 15-peroxyl
radical, which ultimately forms the 11-substituted peroxyl
radical, see Scheme 2. This approach was dictated by the fact
that it was relatively easy to obtain significant quantities of Ch-
15-HPETE by a chemo-enzymatic pathway while the acquisition
of the regioisomeric hydroperoxides required tedious preparative
chromatography. Ch-11-HPETE should also give access to Type
IV endoperoxides and indeed, when Ch-11-HPETE was reacted
under conditions similar to those described above for the 15-
isomer, an LC-MS chromatogram identical with the one
presented in Figure 3 was obtained. This provides additional
evidence in support of the mechanism for formation of the
endoperoxides.
(14) (a) Morrow, J. D.; Harris, T. M.; Roberts, L. J., II Anal. Biochem. 1990,
184, 1-10. (b) Roberts, L. J., II; Morrow, J. D. AdV. Prostaglandin,
Thromboxane, Leukotriene Res. 1995, 23, 219. For an alternate nomen-
clature for the isoprostanes, see refs 8c and 8d.
9
7748 J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002

Formation of Isoprostane Bicyclic Endoperoxides
A R T I C L E S
Figure 4. HPLC chromatogram of the oxidation mixture from Ch-15-HPETE
monitored at UV 234 and 207 nm.
Figure 6. GC-NICI MS SIM (m/z 569) chromatograms of isoprostane
derivatives from (a) Ch-15-HPETE and (b) Ch-11-HPETE. (c) Internal
standard of PGF2a-d4 (m/z 573).
acteristic fragmentation pathways gave poor signal-to-noise and
conclusive identification of isoprostane endoperoxide fractions
in these chromatograms was not possible. Evidence for the
formation of Type II and Type III isoprostane endoperoxides
was obtained, however, by GC-MS as described in the following
section.
Figure 5. Selective reaction monitoring of Type I bicyclic endoperoxides
from Ch-5-HPETE: m/z 843.6 to 359 (UV 234 nm).
Derivatization of Isoprostane Endoperoxides Formed from
Ch-HPETEs: Characterization by Gas Chromatography-
Negative Ion Chemical Ionization Mass Spectrometry (NICI-
MS). The GC-NICIMS technique has been extremely useful in
the study of the isoprostanes due to its high sensitivity.
Moreover, the isoprostanes have been employed as a reliable
index for oxidative injury.15 Oxidized lipid mixtures, either from
in vitro oxidation of arachidonate or from biological sources,
are typically converted to pentafluorobenzyl (PFB) esters and
trimethylsilyl ether (TMS) derivatives in preparation for analysis
by GC-NICIMS. The derivatization procedures of the bicyclic
endoperoxides of cholesteryl arachidonate used in this study
are illustrated in Scheme 5. The bicyclic endoperoxides were
reduced by triphenyl phosphine to give the trihydroxyl choles-
terol ester. We found that PPh3 gave better results than other
reductants such as SnCl2, NaBH4, and SmI2.16 After hydrolysis
of the ester under aqueous basic condition, esterification to the
pentafluorobenzyl ester, this product was converted to the TMS
ether derivative 5 by reaction with BSTFA.17
Each of the purified hydroperoxides (and the mixture of the
8 and 9 isomers) was reacted with a free radical initiator in
benzene under air at 37 °C. Mass spectral fragmentation patterns
(DLI) and CID spectra obtained from the various Ch-HPETE’s
are presented in Supporting Information. In short, mass spectral
fragmentation of endoperoxides of Type I-III occurs by the
same pathways described for the Type IV compounds. Loss of
malondialdehyde, Hock fragmentation of allylic hydroperoxides,
and R-cleavage reactions of hydroperoxides were observed for
all of these compounds.
SRM experiments were carried out in an attempt to analyze
the isoprostane bicyclic endoperoxides of Type I, Type II, and
Type III. Shown in Figure 5 are the SRM and UV (234 nm)
chromatograms of the reaction mixture. The SRM chromatogram
gives strong evidence for the formation of Type I endoperoxides,
based upon fragmentation patterns typical of the structures.
Experiments (SRM) that monitored several characteristic frag-
mentation pathways gave similar chromatograms, supporting
the assignment of structure to compounds.
Attempts to find endoperoxides of Type II and Type III by
SRM HPLC-MS were not successful. These isoprostanes derive
from the 8 and 12 peroxyl radicals and isolation of the requisite
Ch-HPETE precursors in substantial quantities in pure form is
difficult. HPLC-MS SRM chromatograms that monitor char-
GC-MS chromatograms of the isoprostane derivatives, 5, from
oxidation mixtures derived from Ch-15-HPETE and Ch-11-
HPETE are presented in Figure 6. Panel a is the selective ion
monitoring (SIM) chromatogram from reaction of the 15 isomer
and panel b presents a similar chromatogram from the reaction
9
J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002 7749

A R T I C L E S
Yin et al.
Scheme 5
of Ch-11-HPETE. The chromatograms are remarkably similar
with the exception that the relative intensities of some peaks
are reversed. Thus 5c and 5d as well as 5e and 5f have reversed
intensities depending upon the particular precursor. Panel c
shows the internal standard PGF2R-d4.
The isoprostanes from the other types of bicyclic endoper-
oxides have also been studied by this technique and the SIM
chromatograms are presented in Figure 7. The Type II and Type
III derivatives from Ch-12-HPETE are shown in panel a. The
chromatograms of isoprostane derivatives from the mixture of
8-HPETE/9-HPETE (b) and Ch-5-HPETE (c) are also presented
in Figure 7.
The similarity of the chromatograms shown in panels b and
c of Figure 7 is noteworthy. Ch-5-HPETE can give only Type
I isoprostanes and the chromatogram shown in Figure 7c must
therefore represent a stereoisomeric mixture of Type I isopros-
tanes. On the other hand, the mixture of Ch-8- and -9-HPETE’s
that give rise to the chromatogram shown in Figure 7b can give
rise, at least in theory, to Type I, Type II, and Type III
isoprostanes. We conclude from these studies that the formation
of Type II or Type III isoprostanes from 8-HPETE occurs to a
lesser extent than the formation of Type I hydroperoxides from
Ch-5 or 9-HPETE. This suggestion is consistent with the results
from autoxidation of arachidonic acid and in vivo studies of
the rat, i.e., Type I and Type IV isoprostanes are the major
regioisomers.18 The preference for formation of Type I and Type
IV isoprostanes is not well understood and experiments are
ongoing to provide further information on this point.
Figure 7. GC-NICI MS SIM (m/z 569) chromatograms of isoprostane
derivatives from (a) Ch-12-HPETE, (b) Ch-8/9-HPETE, and (c) Ch-5-
HPETE. (d) Internal standard of PGF2a-d4 (m/z 573).
The GC-MS method has been widely used to study lipid
oxidation products from biological sources. Compared to the
LC-MS techniques reported in the previous section, the GC-
MS method gives much better sensitivity. Furthermore, the LC-
MS method provides excellent qualitative information about
peroxide structures but its utility in quantitative analysis is
limited. Silver ion complexes of stereosomeric endoperoxides
give significantly different response factors in the mass spec-
trometer.
Separation of the Stereoisomers of Type IV Isoprostanes.
The bicyclic endoperoxide 2 has five stereogenic centers that
can form 16 racemic diastereomers. There are, however, only
eight possible racemic diastereomers observed due to the
required cis substitution of the peroxide bond on the cyclopen-
tane ring, which leads only to the cis-1,3-dihydroxycyclopentane
substructures. The eight possible derivatives are shown in
Scheme 6. Isoprostanes 5a-f have been synthesized by Taber
et al.,19 and by comparing these known isoprostanes with the
products formed from Ch-15 and 11-HPETE, assignment of
structures to most of the products, both isoprostanes and
cholesteryl ester endoperoxide precursors are possible. The six
major peaks in the chromatogram shown in Figure 6a were
(15) (a) Morrow, J. D.; Zackert, W. E.; Yang, J. P.; Kurkhts, E. H.; Callewaert;
Dworski, R.; Kanai, K.; Taber, D.; Moore, K.; Oates, J. A.; Roberts, J. L.,
II Anal. Biochem. 1999, 269, 326-331. (b) Mori, T. A.; Croft, K. D.;
Puddey, I. B.; Beilin, L. J. Anal. Biochem. 1999, 268, 117-125.
(16) (a) Hamberg, M.; Svensson, J.; Wakabayashi, T.; Samuelsson, B. Proc.
Natl. Acad. Sci. U.S.A. 1974, 71, 345-349. (b) Suzuki, M.; Noyori, R.;
Hamanaka, N. J. Am. Chem. Soc. 1981, 103, 5606-5607. (c) Feldman, K.
S.; Simpson, R. E. Tetrahedron Lett. 1989, 30, 6985-6988.
(17) (a) Pouzar, V.; Slavikowa, T.; Cerny, I. Collet. Czech. Chem. Commun.
1998, 63, 1623-1634. (b) Linnet, K. Clin. Biochem. 1994, 27, 177-182.
(18) (a) Waugh, R. J.; Morrow, J. D.; Roberts, L. J., II; Murphy, R. C. Free
Radical Biol. Med. 1997, 23, 943-954. (b) Waugh R. J.; Murphy, R. C. J.
Am. Soc. Mass. Spectrom. 1996, 7, 490-499.
(19) (a) Taber, D. F.; Herr, R. J.; Gleave, D.; Mark. J. Org. Chem. 1997, 62,
194-198. (b) Taber, D. F.; Kanai, K. Tetrahedron 1998, 54, 11767-11782.
(c) Taber, D. F.; Jiang, Q. J. Org. Chem. 2001, 66, 1876-1884. (d) Taber,
D. F.; Kanai, K.; Pina, R. J. Am. Chem. Soc. 1999, 121, 7773-7777. (e)
Taber, D. F.; Hoerrner, R. S. J. Org. Chem. 1992, 57, 441-447.
9
7750 J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002

Formation of Isoprostane Bicyclic Endoperoxides
A R T I C L E S
Scheme 6
soft ionization of analytes. Addition of a Lewis acid such as
silver ion to the analyte in conjunction with electrospray,
Coordination Ionspray MS, has proved to be extremely useful
in providing information about the structure of organic perox-
ides. Silver ion is a soft Lewis acid that can coordinate to soft
Lewis basic sites, such as carbon-carbon double bonds.
Moreover, fragmentation and rearrangement reactions of the
silver ion-molecule adduct usually give valuable structural
insight. Information about the location of the peroxide func-
tionality in an analyte molecule can be obtained in this way.
The technique of Ag+ CIS/MS has opened the possibility of
analyzing intact peroxides formed from free radical chain
oxidation of cholesteryl arachidonate.6a,10b
The bicyclic endoperoxides formed in the peroxidation of
cholesteryl arachidonate have not been studied due to the lack
of appropriate analytical techniques. Four types of bicyclic
endoperoxides are expected based on the position of H-
abstraction on the arachidonate chain. As shown in Schemes 2
and 4, the Type IV bicyclic endoperoxides can be formed from
Ch-15-HPETE or 11-HPETE. In this mechanism, generation
of the peroxyl radical from the hydroperoxide of Ch-15-HPETE
is followed by â fragmentation and re-addition of oxygen at
the 11-position. Subsequent bicyclization of the 11-peroxyl
provides the Type IV compounds. In a similar way, Type I
bicyclic endoperoxides may form from Ch-5-HPETE or 9-H-
PETE.
assigned structure by adding the known isoprostane derivatives
to the oxidation mixtures followed by GC-MS. The results were
unequivocal. The predominant stereoisomers are the C-15
epimeric pair, 5f and 5e, as well as the 5d and 5c pair. Both of
these epimeric pairs have cis substitution of the alkyl chains
on the cyclopentane ring. As to the relative geometry of the
hydroxyl at C15, the diastereomers with the hydroxyl group on
the side chain at C15 having R configuration (based upon
prostaglandin stereochemistry) elute first on the chromatogram
for each of the known pairs. The first pair of peaks at retention
times around 10 min is tentatively assigned to 5h and 5g,
respectively, although no standards for comparison are available.15b
To provide evidence of stereochemical configuration for the
isoprostane endoperoxides, the oxidation mixture of Ch-15-
HPETE was separated into fractions based upon the SRM
chromatograms shown in Figure 3. The separated fractions were
reduced with triphenylphosphine, converted to PFB-TMS
derivatives, and analyzed by SIM GC-MS. The stereochemistry
of the known isoprostanes was used to prove the stereochemistry
of the original endoperoxide, 2. The six major diastereomers
of the endoperoxides, 2, were found and their stereochemical
configuration was assigned. The assignment of endoperoxide
stereochemistry is shown in Figure 3, in which the configuration
of the endoperoxides, 2a-h, is shown, 2a having the same
configuration as 5a and so on.
The situation with Ch-8-HPETE and Ch-12-HPETE is
distinctly different from that of the other regioisomers. For these
compounds, both Type II and Type III bicyclic endoperoxides
can be formed from a single hydroperoxide, either the 8- or
12-substituted precursor. The mechanism for forming both Type
II and Type III bicyclic endoperoxides from Ch-12-HPETE is
shown in Scheme 7. The peroxyl radical 12-t,c-OO• can undergo
either 5-exo cyclization or â-fragmentation. The cyclization
leads to the Type III endoperoxides while fragmentation,
followed by re-addition of O2 at C8 gives the radical 8-t,t-OO•.
(The nomenclature for these intermediate peroxyl radicals
indicates the position of oxygen substitution on the C-20 chain
and the configuration of the conjugated diene double bonds.)
Subsequent bicyclization of 8-t,t-OO• provides entry to the Type
II endoperoxides.
The rate constant for 5-exo cyclization of peroxyl radicals
has been estimated to be 800 s-1, while the rate constant for
â-fragmentation is approximately 400 s-1 11
. These two competi-
tive reactions for 12-t,c-OO• account for the final formation of
both Type II and Type III bicyclic endoperoxides with the Type
III compounds being the major products formed from Ch-12-
HPETE. The same mechanism applied to Ch-8-HPETE would
give both Type II and Type III, the former being the major
endoperoxides in this instance.
The stereochemistry of the Type IV bicyclic endoperoxides
was determined in this study by conversion of the oxidation
mixtures to pentafluorobenzyl (PFB) ester-trimethylsilyl (TMS)
derivatives and by comparison of the GC-MS chromatogram
of the reaction mixture with those of the synthetic standards.21
Discussion
Although mass spectrometry plays an indispensable role in
structural elucidation of organic and biological compounds,
conventional MS techniques cause difficulties in the analysis
of peroxidation product mixtures because most of the ionization
techniques result in the fragmentation of peroxide molecular
ions.20 Electrospray MS is becoming more and more important
in the analysis of biological compounds because it involves the
(20) (a) Murphy, R. C. Chem. Res. Toxicol. 2001, 14, 1-10. (b) Murphy, R.
C.; Fieldler, J.; Helvko, J. Chem. ReV. 2001, 101, 479-526. (c) Yin, H.;
Hachey, D. L.; Porter, N. A. Rapid Commun. Mass. Spectrom. 2000, 14,
1248-1254.
(21) Thomas, M. J.; Chen, Q.; Sorci-Thomas, M. G.; Rudel, L. L. Free. Radical
Biol. Med. 2001, 30, 1337-1346.
9
J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002 7751

A R T I C L E S
Yin et al.
Scheme 7
Scheme 8
Eight possible racemic diastereomers were identified in the
oxidation mixture of Ch-15-HPETE, a reaction that provides
Type IV endoperoxides. Our results are consistent with those
of O’Connor and Mihelich in which the endoperoxides with
syn configuration of the cyclopentane side chains are favored
compared to products having anti side chain arrangement.22,23
Our results also suggest that the endo-endo diastereomers are
more abundant than exo-exo products. The stereochemical
course of the free radical process is in sharp contrast to that of
the cyclooxygenase (COX) promoted conversion in which the
anti product, 2a, is formed exclusively.7
According to the proposed mechanism, the bicyclic endo-
peroxides formed from Ch-15-HPETE should be identical with
those formed from Ch-11-HPETE. As shown in Figure 6, similar
chromatograms are indeed obtained from reactions of both of
these compounds. We note, however, that the relative intensity
of the C-15 epimeric products corresponding to 5c and 5d is
dependent upon the precursor hydroperoxide. Ch-15-HPETE
gives more 5c while the 11-regioisomer gives more 5d product.
Precursor-dependent product stereochemistry is also observed
for the C-15 epimers 5e and 5f while this dependence is not
observed for products 5a and 5b.
The only substantive difference between the mechanisms of
reaction of Ch-11- and 15-HPETE is the geometry of the
conjugated diene present in the first-formed peroxyl radicals,
6a and 6b. The conjugated diene of Ch-11-HPETE and the 11-
peroxyl derived from this precursor, 6a, have ∆12-13 trans,
stereochemistry translates directly from allyl radical geometry
to the C-15 epimers 5c and 5d. The radical 7a, having trans,cis
substitution, adds oxygen preferentially to give 5d while 7b,
having trans,trans allyl geometry, adds oxygen to give 5c as
the major product. Scheme 8 shows this reaction course for the
oxygen addition. The geometry of 7a and 7b is chosen such
that it is consistent with the products observed, if oxygen adds
to the face of the radical opposite to the vicinal side chain on
the endoperoxide.
A similar analysis of the epimeric products 5e and 5f, formed
from radicals 8a and 8b, suggests that the preferred geometry
for these radicals is as shown in Scheme 9. Again, preferential
addition of oxygen to the radical face opposite to the vicinal
chain gives the preferred product. We note that the formation
of 15-R isoprostane epimers from 11-HPETE (5d and 5f) is
consistent with the observations reported earlier for the simpler
linolenate system.22
∆14-15 cis geometry while the corresponding diene peroxyl 6b,
formed from Ch-15-HPETE, has ∆12-13 trans, ∆14-15 trans
configuration. These intermediate peroxyls are stereoisomeric.
We speculate that this stereochemical difference, translated
through the reaction course, results in stereoisomeric allyl
radicals 7a and 7b. Furthermore, we suggest that product
(22) (a) O’Connor, D. E.; Mihelich, E. D.; Coleman, M. C. J. Am. Chem. Soc.
1984, 106, 3577-3584. (b) Mihelich, E. D. J. Am. Chem. Soc. 1980, 102,
7141-7143. (c) O’Connor, D. E.; Mihelich, E. D.; Coleman, M. C. J. Am.
Chem. Soc. 1981, 103, 223-224.
(23) (a) Corey, E. J.; Shih, C.; Shih, N.-Y.; Shimoji, K. Tetrahedron Lett. 1984,
25, 5013-5016. (b) Corey, E. J.; Wang, Z. Tetrahedron Lett. 1994, 35,
539-544.
It is also noteworthy that product distribution is apparently
independent of precursor diene geometry for the other isopros-
tane C-15 epimeric pairs, 5a/b and 5h/g, which have side chains
substituted trans on the endoperoxide ring. A cis vicinal side
9
7752 J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002

Formation of Isoprostane Bicyclic Endoperoxides
A R T I C L E S
Scheme 9
purification from Aldrich Chemical Co. (Milwaukee, WI). N,O-Bis-
(trimethylsilyl)trifluoroacetamide (BSTFA) was obtained from Supelco
Inc. (Bellefonte, PA). Organic solvents such as dichloromethane,
hexane, benzene, and ethyl acetate were HPLC quality purchased from
Allied Signal (Burdick and Jackson Brand, Muskegon, MI). Preparative
TLC was performed on silica gel 60ALK6D plates (Whatman
International Ltd., Maidstone, UK). The free radical initiator di-tert-
butyl hyponitrite (DTBN) was synthesized according to the literature
procedures.24
Instruments. Analytical HPLC was carried out with a Waters Model
600E pump and a Waters 996 Photodiode array detector. Millenium32
chromatography software (Waters Corp., Milford, MA) was used to
control the 996 and to collect and process data. Cyclic peroxide analysis
by analytical HPLC utilized a single Beckman Ultrasphere 5 µm (4.6
mm × 25 cm) silica column. A flow rate of 1 mL/min was used for
analytical NP HPLC. Preparative NP HPLC was performed with a
Dynamax-60 Å (83-121-C) silica column (21.4 mm × 25 cm × 8 µm
particles) with a flow rate of 10 mL/min. Narrowbore HPLC for MS
analysis used a single Beckman Ultrasphere 5 µm (2.0 mm × 25 cm)
silica column for analysis of cyclic peroxides and two Beckman
Ultrasphere 5 µm (2.0 mm × 25 cm) silica columns for acyclic
hydroperoxide analysis. The flow rate for the LC-MS separations and
analysis was 150 µL/min.
Mass Spectrometry. CIS-MS was accomplished with a Finnigan
TSQ-7000 (San Jose, CA) triple quadrupole mass spectrometer operat-
ing in positive ion mode equipped with a standard API-1 electrospray
ionization source. The source was outfitted with a 100 µm deactivated
fused Si capillary. Data acquisition and evaluation were conducted on
ICIS EXECUTIVE INST, version 8.3.2, and TSQ 7000 software INST,
version 8.3. Data collected for Selected Reaction Monitoring (SRM)
experiments was also processed with Xcalibur, version 1 (Finnigan,
San Jose, CA).
chain on the endoperoxide is an important factor for the modest
stereoselection observed. Steric shielding arguments have been
invoked to explain stereoselectivity in radical additions to
carbon-carbon double bonds or atom transfer reactions. One
expects that steric shielding would have a marginal effect in
the addition of oxygen to carbon radicals, a reaction to proceed
at or near diffusion control, and this is confirmed by our
observations.
Our experiments are relevant to a novel alternate mechanism
for isoprostane formation described in Scheme 10a.23b,8c,d In this
mechanism, the Type IV isoprostanes are formed from 15-
HPETE by a 4-exo-peroxyl radical cyclization followed by
oxygen addition and double 5-exo cyclization. This mechanism
does not account for the formation of monocyclic and serial
cyclic peroxides, 3 and 4, which are major products in the
reaction and which are consistent with the mechanism described
in Scheme 2. The mechanism described in Scheme 10 also does
not account for the formation of Type IV isoprostanes from
Ch-11-HPETE, see Scheme 10b. In fact, we see very similar
product distributions starting from either the 15 or the 11
regioisomers, Figure 6, again consistent with the mechanism
outlined in Schemes 2 and 4. Finally, our LC-MS analysis of
reactions of Ch-15-HPETE shows evidence for the formation
of Ch-11-HPETE during the reaction. This observation, which
is consistent with reactions of other lipid hydroperoxides,
supports the notion that fragmentation of the 15-peroxyl radical
and re-addition of oxygen to give the 11-peroxyl radical
accompanies the formation of isoprostane products. This
observation is consistent with the mechanism shown in Scheme
2. The weight of the experimental evidence would appear to
support the mechanism we propose.
Nitrogen gas served both as a sheath gas and an auxiliary gas; argon
served as the collision gas. The electrospray needle was maintained at
4.6 kV and the heated capillary temperature was 200 °C. The tube
lens potential and capillary voltage were optimized to maximize ion
current for electrospray, with the optimal determined to be 80 and 20
V, respectively, for cholesteryl ester analysis. Positive ions were
detected scanning from 100 to 1000 amu with a scan duration of 2 s.
Profile data was recorded for 1 min and averaged for analysis. For
CID experiments the collision gas pressure was set from 2.3 to 2.56
mT. To obtain fragmentation information of each compound, the
dependence of offset-voltage and reconstructed ion current (RIC) was
studied. The collision energy offset was varied from 10 to 40 eV
depending on the compound being analyzed.
Samples were introduced either by direct liquid infusion or by HPLC.
For direct liquid injection, stock solutions of the lipids (100 ng/µL in
1% IPA in hexane) were prepared and mixed 1:1 with silver tetrafluoro-
borate (51.4 ng/µL in IPA). Samples were introduced to the ESI source
by syringe pump at a rate of 10 µL/min. For HPLC sample introduction
a Hewlett-Packard 1090 HPLC system was used. The auxiliary gas
flow rate to the ESI interface was increased to between 5 and 10 units
to assist in desolvation of the samples. For cholesteryl arachidonate
hydroperoxide analysis, normal phase HPLC sample introduction was
carried out by using two tandem Beckman Ultrasphere narrowbore 5µ
silica columns (2.0 mm × 25 cm) operated in isocratic mode with 0.5%
2-propanol in hexanes. For analysis of cyclic peroxide mixtures, sample
introduction was carried out with a single Beckman Ultrasphere
narrowbore 5 µm silica column (2.0 mm × 25 cm) operated in isocratic
mode with 1.0% 2-propanol in hexanes. The flow rate for both modes
of chromatography was 150 µL/min. Column effluent was passed through
an Applied Biosystems 785A programmable absorbance UV detector
with detection at 234 nm. An Upchurch high-pressure mixing tee was
Experimental Section
General Methods. Reactions involving hydroperoxides were moni-
tored by TLC by using a stain of 1.5 g of N,N′-dimethyl-p-phenylene-
diamine dihydrochloride/25 mL of H2O/125 mL of MeOH/1 mL of
acetic acid. Hydroperoxides yield an immediate pink color. TLC was
carried out by using a 0.2 mm layer thickness of Si coated on aluminum
(EM Scientific) and was visualized by UV 254, phosphomolybdic acid
char, or the peroxide stain. In general, peroxides were stored as dilute
solutions in either hexanes or benzene at -78 °C and were never
exposed to temperatures above 40 °C. All HPLC solvents were filtered
through Whatman Nylon membrane filters (0.45 µm pore size) prior
to use.
Chemicals. All lipids were purchased from Nu Chek Prep (Elysian,
MI) and were of the highest purity (>99+%). Pentafluorobenzyl
bromide and diisopropylethylamine were purchased without further
(24) (a) Mendenhall, G. D. Tetrahedron Lett. 1983, 24, 451-454. (b) Traylor,
T. G.; Kiefer, H. Tetrahedron Lett. 1966, 49, 6163-6168.
9
J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002 7753

A R T I C L E S
Yin et al.
Scheme 10
connected next in series for the postcolumn addition of the silver salts.
The silver tetrafluoroborate (AgBF4) solution (0.25 mM in 2-propanol)
was added via a Harvard Apparatus (Cambridge, MA) syringe pump
at a flow rate of 75 µL/min. A long section of PEEK tubing (1.04 m,
0.25 mm i.d.) allowed time for the complexation of the silver to the
lipid while delivering effluent to the mass spectrometer. A Rheodyne
7725 injector was fitted with a 100 µL PEEK loop for 20-50 µL sample
injections.
GC-NICI MS was performed with a Hewlett-Packard HP5989A GC/
MS instrument interfaced with an IBM Pentium II computer system.
GC was performed with a 30 m, 0.25 mm diameter, 0.25 µm film
thickness, DB-Wax column (J & W Scientific, Folsom, CA). The
column temperature was programmed from 190 to 260 °C at 10 deg/
min. Methane was used as the carrier gas at a flow rate of 1 mL/min.
The ion source temperature was 250 °C, the electron energy was 70
eV, and the filament current was 0.25 mA.
Oxidation of Cholesteryl Arachidonate to Primary Hydroper-
oxides. In a round-bottomed flask, cholesteryl arachidonate (400 mg,
0.584 mmol) was dissolved in benzene to make a 0.15 M solution. To
the mixture were added 0.5 equiv of methyl trolox (methyl 6-hydroxy-
2,5,7,8-tetramethylchroman-2-carboxylic acid ester)25 and 0.1 equiv of
DTBN. The reaction was stirred under oxygen at 37 °C for 24 h. The
reaction was quenched by adding 2 mg of BHT. Analytical HPLC
(tandem Si column, 0.5% 2-propanol in hexane, λmax ) 234 nm)
indicated seven major fractions: A, tR ) 13.80 min, 15-HPETE; B,
to the mixture. After evaporation of the solvent, the residue was
dissolved in 200 µL of ethanol and to the solution was added 200 µL
of 8 M aqueous KOH solution. After being stirred at 40 °C for an
hour, the mixture was cooled to room temperature and the pH of the
solution was adjusted to 3 by adding dilute HCl. The aqueous solution
was extracted by ethyl acetate three times. The combined organic phases
were dried over Na2SO4. After evaporation of the solvent, the residue
was dissolved in 20 µL of CH3CN. To the resulting solution was added
20 µL of 10% pentafluorobenzyl bromide in acetonitrile and 10 µL of
10% N,N-diisopropylethylamine in acetonitrile and the mixture was
kept at room temperature for 30 min. The reagent was dried under
nitrogen and the residue was subjected to TLC separation by using the
solvent ethyl acetate/methanol (98:2, v/v). Approximately 5 µg of the
PFB ester of PGF2R was applied on a separate lane and visualized by
spraying with a 10% solution of phosphomolybdic acid in ethanol
followed by heating. Compounds migrating in the region of the PFB
ester of PGF2R (Rf ) 0.4) and the adjacent area 2 cm above and under
were scrapped and extracted from the silica gel by ethyl acetate.
After evaporation of the ethyl acetate, 20 µL of BSTFA and 10 µL
of dimethylformamide were added to the residue and the mixture was
incubated at 40 °C for 20 min. The reagents were dried under nitrogen
and the derivatives were dissolved in 10 µL of dry undecane for analysis
of GC-MS.
Acknowledgment. We thank Lisa M. Manier and Dr. David
L Hachey at the Vanderbilt University Mass Spectrometry
Research Center for their assistance with MS measurements and
Dr. Douglass F. Taber of the Department of Chemistry and
Biochemistry at the University of Delaware for providing
isoprostane samples. Financial support from NIH HL17921,
GM15431, DK 48831, CA 77839, P30 ES00267, and NSF CHE
0107697 is gratefully acknowledged. H.Y. acknowledges the
Warren Graduate Fellowship of the Chemistry Department at
Vanderbilt University. J.D.M. is the recipient of a Burroughs
Wellcome Fund Clinical Scientist Award in Translational
Research.
tR ) 14.54 min, 12-HPETE; C, tR ) 16.11 min, 11-HPETE; D, tR
)
20.91 min, 9-HPETE; E, tR ) 21.29 min, 8-HPETE; F, tR ) 26.11
min, 5-HPETE; G, tR ) 51.30 min, methyl trolox. The excess of
hydrogen atom donor methyl trolox was separated by flash column
chromatography on silica with use of a solvent mixture of hexane:
ethyl acetate (9:1). All the peroxide stain positive fractions were
combined and separated by semipreparative HPLC (0.5% 2-propanol
in hexane, 234 nm). Components D and E were collected as a mixture.
Autoxidation of the Hydroperoxides of Cholesteryl Arachidonate.
The primary hydroperoxides separated by HPLC were converted to
more highly oxidized peroxides according to the literature.6a
Derivatization of the Oxidation Products of Cholesteryl Ara-
chidonate. To 100 µL of the oxidation mixture was added 10 ng of
PGF2R-d4 as internal standard. An excess amount of PPh3 was added
Supporting Information Available: DLI and CID spectra
obtained from the various Ch-HPETE (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
(25) (a) Cohen, N.; Lopresti, R. J.; Neukom, C. J. Org. Chem. 1981, 46, 2445-
2450. (b) Yoda, H.; Takabe, K. Chem. Lett. 1989, 465-468. (c) Suarna,
C.; Dean, R. T.; Southwellkeely, P. T. Aust. J. Chem. 1997, 50, 1129.
JA0201092
9
7754 J. AM. CHEM. SOC. VOL. 124, NO. 26, 2002