An ESR and HPLC-EC assay for the detection of alkyl radicals

Article abstract of DOI:10.1021/tx015507h

The correlation of lipid peroxidation with release of alkanes (RH) is considered a noninvasive method for the in vivo evaluation of oxidative stress. The formation of RH is believed to reflect a lipid hydroperoxide (LOOH)-dependent generation of alkoxyl radicals (LO^·) that undergo β-scission with release of alkyl radicals (R^·). Alternatively, R^· could be spin-trapped with a nitrone before the formation of RH and analyzed by ESR. Extracts from the liver and lung of CCl₄- and asbestos-treated rats that were previously loaded with nitrones exhibited ESR spectra suggesting the formation of iso-propyl, n-butyl, ethyl, and pentyl radical-derived nitroxides. In biological systems, various nitroxides with indistinguishable ESR spectra could be formed. Hence, experiments with N-tert-butyl-α-phenylnitrone (PBN) for spin trapping of R^· were carried out in which the nitroxides formed were separated and analyzed by HPLC with electrochemical detection (EC). The C_1-5 homologous series of PBN nitroxides and hydroxylamines were synthesized, characterized by ESR, GC-MS, and HPLC-EC, and used as HPLC standards. For in vivo generation and spin trapping of R^·, rats were loaded with CCl₄ and PBN. The HPLC-EC chromatograms of liver extracts from CCl₄-treated rats demonstrated the formation of both the nitroxide and hydroxylamine forms of PBN/^·CCl₃, as well as the formation of a series of unidentified PBN nitroxides and hydroxylamines. However, formation of PBN adducts with retention times similar to these of the PBN/C_2-5 derivatives was not observed. In conclusion, we could not correlate the production of PBN-detectable alkyl radicals with the reported CCl₄-dependent production of C_1-5 alkanes. We speculate that the major reason for this is the low steady-state concentrations of R^· produced because only a small fraction of LO^· undergo β-scission to release R^·.

Full text of DOI:10.1021/tx015507h

Chem. Res. Toxicol. 2001, 14, 1239-1246
1239
An ESR a n d HP LC-EC Assa y for th e Detection of Alk yl
Ra d ica ls
Christo P. Novakov, Dennis Feierman, Arthur I. Cederbaum, and
Detcho A. Stoyanovsky*
Department of Biochemistry and Molecular Biology, Mount Sinai School of Medicine,
New York, New York 10029
Received April 3, 2001
The correlation of lipid peroxidation with release of alkanes (RH) is considered a noninvasive
method for the in vivo evaluation of oxidative stress. The formation of RH is believed to reflect
a lipid hydroperoxide (LOOH)-dependent generation of alkoxyl radicals (LO•) that undergo
â-scission with release of alkyl radicals (R•). Alternatively, R• could be spin-trapped with a
nitrone before the formation of RH and analyzed by ESR. Extracts from the liver and lung of
CCl4- and asbestos-treated rats that were previously loaded with nitrones exhibited ESR spectra
suggesting the formation of iso-propyl, n-butyl, ethyl, and pentyl radical-derived nitroxides.
In biological systems, various nitroxides with indistinguishable ESR spectra could be formed.
Hence, experiments with N-tert-butyl-R-phenylnitrone (PBN) for spin trapping of R• were carried
out in which the nitroxides formed were separated and analyzed by HPLC with electrochemical
detection (EC). The C1-5 homologous series of PBN nitroxides and hydroxylamines were
synthesized, characterized by ESR, GC-MS, and HPLC-EC, and used as HPLC standards. For
in vivo generation and spin trapping of R•, rats were loaded with CCl4 and PBN. The HPLC-
EC chromatograms of liver extracts from CCl4-treated rats demonstrated the formation of both
the nitroxide and hydroxylamine forms of PBN/•CCl3, as well as the formation of a series of
unidentified PBN nitroxides and hydroxylamines. However, formation of PBN adducts with
retention times similar to these of the PBN/C2-5 derivatives was not observed. In conclusion,
we could not correlate the production of PBN-detectable alkyl radicals with the reported CCl4-
dependent production of C1-5 alkanes. We speculate that the major reason for this is the low
steady-state concentrations of R• produced because only a small fraction of LO• undergo
â-scission to release R•.
nitroxide may be extracted from tissues subjected to
oxidative stress and analyzed by ESR. It has been
reported that extracts from the liver and lung of CCl4-
and asbestos-treated rats that were previously loaded
with nitrones exhibited ESR spectra with hyperfine
structure which allows the assignment of the adducts as
those formed by the addition of iso-propyl and n-butyl,
and ethyl and pentyl radicals, respectively (6, 7).
In tr od u ction
The correlation of lipid peroxidation with release of
alkanes (RH)1 has attracted considerable interest as a
noninvasive method for the in vivo evaluation of oxidative
stress. It has been reported that patients with cystic
fibrosis, alcohol liver disease, rheumatoid arthritis,
asthma, and Alzheimer’s disease have an increased, GC-
detectable exhalation of ethane, propane, n-butane,
isobutane, pentane, and isopentane (reviewed in ref 1).
Under conditions of oxidative stress, the formation of RH
is believed to reflect a lipid hydroperoxide (LOOH)-
dependent generation of alkoxyl radicals (LO•) that
undergo â-scission with release of alkyl radicals [R•;
Scheme 1, 4 f 5 f 7 (2-5)]. The latter are a highly
reactive species that can trigger chain reactions within
the cellular membranes, react with oxygen to form
peroxyl radicals (ROO•; Scheme 1, 7 f 9) and/or to react
with H-donors to form RH (Scheme 1, 7 f 8).
The detection and quantitation of alkyl radicals has
proven to be difficult as they are species with half-lives
within the microsecond scale; furthermore, R• reacts in
a diffusion-controlled manner with oxygen to form sec-
ondary, oxygen-centered radicals (ROO•). Hence, when
experiments are carried out for spin trapping of R• in
biological systems, a particular complication in the
interpretation of the ESR data may be the formation of
multiple nitroxides with similar ESR spectra. Recently,
Iwahashi et al. have reported that pentyl radical could
be spin-trapped with R-(4-pyridyl-1-oxide)-N-tert-butylni-
trone (POBN) in a complex reaction system; the resulting
POBN/C5H11 nitroxide was analyzed by HPLC with ESR
and MS detection (8). Alternatively, characterization of
alkyl-derived nitroxides could be carried out by HPLC
analysis with electrochemical detection [EC (9-12)]. In
this report, we have extended the HPLC-EC methodology
for detection of alkyl radicals within the homologous
series of methyl to pentyl radicals (C1-5). PBN/C1-5
The in vivo observation of RH, however, provides
limited information about the sites of radical formation.
Alternatively, R• could be spin-trapped with a nitrone
before the formation of RH; the resulting alkyl-derived
* To whom correspondence should be addressed. Phone: (212) 241-
6810, Fax: 212-996-7214, E-mail: stoyanovsky@hotmail.com.
Abbreviations: ESR, electron spin resonance; PBN, N-tert-butyl-
R-phenylnitrone; Rt , retention time; EC, electrochemical; LPO, lipid
1
peroxidation.
10.1021/tx015507h CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/08/2001

1240 Chem. Res. Toxicol., Vol. 14, No. 9, 2001
Novakov et al.
Sch em e 1
nitroxides were prepared under the conditions of the
Wurtz reaction (13) and/or via direct coupling of PBN
with Grignard reagents; ESR, GC-MS, and HPLC-EC
analysis were used for characterization of the corre-
sponding adducts. Liver extracts from PBN plus CCl4-
treated rats were analyzed by HPLC-EC with the ulti-
mate goal to evaluate whether the formation of PBN/C2-5
nitroxides could be used as an in vivo index of radical-
dependent phospholipid alteration.
Ta ble 1. ESR Hyp er fin e Cou p lin g Con sta n ts (in G) for
P BN/C1-5 Nitr oxid es in 80% Meth a n ol a t 20 °Ca
Ma ter ia ls a n d Meth od s
Rea gen ts. All reagents used were purchased from Sigma
Chem. Co. (St. Louis, MO). The solutions used in the experi-
ments were prepared in deionized and Chelex-100 treated (batch
method) water or 0.1 M potassium phosphate buffer (pH 7.4).
ESR Mea su r em en ts. ESR measurements were performed
on a Bruker ECS106 spectrometer with 50 kHz magnetic field
modulation at room temperature (25 °C). ESR spectrometer
settings were: modulation amplitude 0.7 G, scan time 41 s, time
constant 0.164 s, microwave power 20 mW and receiver gain 1
× 105-1 × 106. ESR spectra simulations were made using a
program created by Philip D. Morse II and Richard Reiter (EPR
Simulation System 2.01, Scientific Software Services, IL). The
hyperfine splitting constants used for ESR spectra simulation
were taken from Table 1 and/or (14).
P r ep a r a tion of Gr ign a r d Rea gen ts. Alkylmagnesium
bromides (RMgBr) were prepared by refluxing (1-1.5 h) the
corresponding alkylbromide (3 M) with an equimolar amount
of Mg in dry diethyl ether (30 mL). The reactions were carried
out by a dropwise addition of the bromide at a rate that secured
moderate boiling of the solvent; in some experiments, initiation
of the reaction required an addition of a crystal of iodine. In
dry atmosphere (20 °C), the solutions of alkylmagnesium
bromides were stable for several months.
P r ep a r a tion of P BN/C1-5 Ad d u cts. Alkyl-derivatization of
PBN was carried out via direct coupling of RMgBr (0.3 M) with
PBN (0.08 M) in helium-deaerated, dry diethyl ether (10 mL).
After an incubation of 10 min (20 °C), the excess of the Grignard
reagent was carefully decomposed with 50% ethanol (v/v; 2 mL),
the organic layer was separated, dried over Na2SO4 and rotor-
evaporated (25 °C). The dry residue containing both the nitrox-
ide and hydroxylamine forms of the corresponding PBN deriva-
tive was redissolved in ethanol and used as an HPLC-EC
standard. When stored at -80 °C, the PBN/C1-5 adducts were
stable for several months. Preparation of PBN/C1-5 adducts was
also carried out under the condition of the Wurtz reaction
(Supporting Information and ref 13).
GC/MS An a lysis. GC/MS analysis was performed with a
Hewlett-Packard (HP) 5890 Series II gas chromatograph,
equipped with a HP 7673 automatic injector, a DB-17 capillary
column (15 m × 0.25 mm; 0.25 µm film thickness; J & W
Scientific, Folson, CA). A HP 5972 mass selective detector (MSD)
was employed in the scan mode. The underivatized samples (2
µL) were injected with the purge valve closed for the first minute
(splitless mode). Helium was the carrier gas at a pressure of 72
kPa. Initial temperature settings for the oven, injector, and MSD
were 100, 225, and 280 °C, respectively. The oven temperature
a
Preparation and ESR analysis of PBN/C1-5 nitroxides was
carried out as described in the Materials and Methods. (*)
Nitroxides with ESR spectra that differ from the ESR spectra of
PBN/C2H5.
was increased at a rate of 10 °C per min to 280 °C and was
then held constant for the remaining time of the analysis.
HP LC Mea su r em en ts. HPLC was performed with an ESA
model 5102 liquid chromatograph (Milford, MA). Separation was
achieved with either a 4.6 mm × 15 or 25 cm C-18 column (5
µm, 100 A; Microsorb, Rainin Instrument Company, Inc.,
Emeryville, CA). The mobile phase was saturated with helium
and contained 20 mM lithium perchlorate and either 70 or 80%
(v/v) methanol. All HPLC analyses were conducted at a flow
rate of 1 mL/min. Electrochemical detection of PBN/C1-5 adducts
was carried out at + 0.7 V with an ESA coulochem detector. A
0.020 mL injection loop was used for all experiments.
An im a ls a n d An im a l Tr ea tm en ts. For induction of P450
2E1, Sprague-Dawley rats (n ) 9) weighing 120-150 g were
injected i.p. with pyrazole (200 mg/kg body wt/day, 3 days). The
rats were starved overnight prior to being killed and experi-
ments with CCl4 were conducted as described previously (6, 15).
Briefly, the animals were loaded with PBN (5-200 mg/kg body
weight) dissolved in olive oil (2 mL); 15 min later, CCl4 was
injected i.p. as a 20% solution in olive oil at a dose of 5 mL
solution per kg body weight (1 mL of pure CCl4/kg). In a typical
experiment, the animal was sacrificed 60 min after the admin-
istration of CCl4, its liver was homogenized in 20 mL of PBS

ESR and HPLC-EC Detection of Alkyl Radicals
Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1241
Sch em e 2
and the homogenate subjected to extraction with CHCl3 (40 mL)
and methanol (20 mL). The resulting emulsion was centrifuged
(15 min × 10000g), the chloroform phase was collected and the
solvent rotor-evaporated at room temperature. The residue was
re-suspended in 5 mL of ethanol (80%) and centrifuged (30 min
× 10000g). The supernatant was used for ESR and HPLC-EC
analysis, while the bottom layer (0.1-0.5 mL of fats) was
discarded.
peak 2 did not exhibit any ESR activity (Figure 1A, ESR
spectrum 2). An addition of potassium ferricyanide (0.02
mM) to this fraction, however, resulted in the appearance
of an ESR spectrum that was identical to ESR spectrum
1 (Figure 1A, ESR spectrum 3; aN ) 15.59 G; aH ) 3.56
G), suggesting that peak 2 reflects the elution of the
hydroxylamine form of the adduct. A brief pretreatment
of the PBN/C5H11 solution with either potassium ferri-
cyanide or ascorbate in 0.1 M phosphate buffer (pH 7.4)
resulted in an interconversion of peak 2 into peak 1
(oxidation of the hydroxylamine form) and vice versa
(reduction of the nitroxide form; data not shown). The
PBN/C5H11 structure was further supported by GS-MS
analysis (Figure 1B). The GC profile of the reaction
solution exhibited one major peak (Inset to Figure 1B,
peak 1). The same GC peak was obtained after injection
of HPLC fraction 2 (Figure 1A, peak 2) that supposedly
reflected the elution of the hydroxylamine form of PBN/
C5H11 (data not shown). The mass profile of the latter
included a species with a mass-to-charge ratio (m/z) of
249 that could be assigned to the protonated PBN/C5H11
nitroxide and molecular fragments with m/z values of 161
and 192 that most likely reflect the fragmentation of
PBN/C5H11 to PhC5H11+ and PhC5H11NHO+, respec-
tively. A more detailed assignment of the molecular
fragments shown in Figure 1B is proposed in Table 2.
Resu lts
P r ep a r a tion of P BN/C1-5 Ad d u cts. Nucleophilic
reagents in which the nucleophilic center is located on a
carbon atom, such as alkylmagnesium bromides (Grig-
nard reagents), CH3NO2, RNa, and CN- react with PBN
via a 1,3-addition mechanism to form the corresponding
PBN-derived hydroxylamines (Scheme 2; reviewed in ref
16). The latter readily undergo oxidation to nitroxides
that could be analyzed by direct ESR spectrometry. Both
nitroxide and hydroxylamine forms of PBN can be
quantified by HPLC-EC (9, 11, 12).
According to the Scheme 2, alkyl-derived PBN nitrox-
ides were prepared for the C1-5 homologous series (Table
1). When analyzed by HPLC-EC, the PBN/C1-5 adducts
exhibited increased retention times on a C18 matrix with
increasing length of the carbon chain of the alkyl group.
In Figure 1, panel A depicts the HPLC-EC profile of a
reaction solution consisting of PBN plus pentylmagne-
sium bromide. The fraction defined by peak 1 exhibited
an ESR spectra that could be assigned to the PBN/C5H11
nitroxide (Figure 1A, ESR spectrum 1; in gauss (Table
1) aN ) 15.59 and aH ) 3.56). The fraction collected under
The PBN/C1-5 hydroxylamines autoxidize in phosphate
buffer back to their nitroxide forms. The oxidation of the
hydroxylamines can also be carried out with potassium
ferricyanide (Figure 2A, first arrow); however, excess of
F igu r e 1. HPLC-EC and GC-MS analysis of the reaction products of pentylmagnesium bromide and PBN. C5H11MgBr-dependent
derivatization of PBN and subsequent HPLC-EC and GC-MC measurements were carried out as described in the Materials and
Methods. HPLC separation of the electroactive analytes was carried out with a C18 Microsorb column (4.6 mm × 15 cm; 5 µm) after
a 10-fold dilution of the reaction solution in ethanol. ESR spectra 1 and 2 refer to the spectrum of peaks 1 and 2, respectively, while
spectrum 3 refers to peak 2 treated with K3[Fe(CN)6]. HPLC profile: solid lines, minus K3[Fe(CN)6]; dashed lines, plus K3[Fe(CN)6].

1242 Chem. Res. Toxicol., Vol. 14, No. 9, 2001
Novakov et al.
Ta ble 2. Ma ss Sp ectr a l Da ta of P BN/C5H11
Hyd r oxyla m in ea
a
Preparation and GC-MS analysis of PBN/C5H11 hydroxylamine
was carried out as described in the Materials and Methods.
F igu r e 3. HPLC-EC chromatograms of PBN/C2-5 nitroxides
and hydroxylamines. (A) HPLC-EC profile of PBN/C2-5 nitrox-
ides (C2-5*) and hydroxylamines (C2-5). (B) HPLC-EC profile of
PBN/C2-5 hydroxylamines after reduction of their nitroxide
forms with ascorbic acid (15 mM) for 15 min in 0.1 M phosphate
buffer (pH 7.4; at 25 °C). PBN/C2-5 nitroxides were prepared
as described in the Materials and Methods and diluted 10 times
in 0.1 M phosphate buffer (pH 7.4). The HPLC separation was
achieved with a C18 Microsorb column (4.6 mm × 25 cm; 5 µ)
and a mobile phase consisting of either 70% (A) or 80% (B)
methanol, and 20 mM LiClO4.
F igu r e 2. ESR-monitored kinetics of redox interconversion of
PBN/C5H11 and PBN/CH3 adducts. All experiments were carried
out in 0.1 M phosphate buffer (pH ) 7.4; 25 °C). The nitroxide
and hydroxylamine derivatives of PBN were prepared as de-
scribed in the Materials and Methods. (A) Autoxidation of PBN/
CH3 (4) and PBN/C5H11 (O) hydroxylamines in buffered aqueous
solutions, and PBN/C5H11 hydroxylamine (b) in the presence
of K3[Fe(CN)6] (1 mM). The addition of K3[Fe(CN)6] is indicated
with an arrow. (B) ESR signal intensity of PBN/C5H11 (b) and
PBN/CH3 (4) nitroxides in the presence of ascorbic acid (5 mM).
their nitroxides. Since the separation of the hydroxyl-
amine derivatives could be carried out with a mobile
phase with lower polarity, the time of the analysis could
be considerably decreased (Figure 3B). Moreover, the
nitroxide and hydroxylamine forms of the alkyl adducts
can be further distinguished from each other by I/V
curves of their voltage dependence as shown in Figure 4
for the ethyl and pentyl PBN nitroxides and hydroxyl-
amines. The HPLC peaks obtained in analyses of complex
systems could be positively identified via their retention
time, co-injected of preformed standards (Figure 3), as
well as via redox interconversion (Figures 1 and 2), ESR
spectra (Figure 1A and Table 1), and I/V curves (Figure
4).
Alternatively, PBN/C1-5 nitroxides and hydroxylamines
could be prepared under the conditions of the reaction of
Wurtz [RBr + Na f R• (13); Supporting Information].
Figure 5 A, curve 1, depicts the ESR spectrum of a
solution of C3H7Br and C5H11Br in hexane containing
PBN and Na. The PBN/C3H7 and PBN/C5H11 nitroxides
have similar hyperfine splitting constants (Table 1). The
computer simulation of their ESR spectra presented in
Figure 5A, curve 2, illustrates that they are ESR indis-
tinguishable. The HPLC-EC analysis of the reaction
K3[Fe(CN)6] also reacts with the nitroxide form of the
adducts and causes loss of the ESR signal (Figure 2A,
second arrow). Reduction of radical-derived ESR active
nitroxides in spin trapping experiments is an unavoidable
occurrence in most in vivo systems (17). Therefore, when
biological samples are analyzed, any procedure that leads
to dilution of the reaction mixture could lower the
reductive potential of the system and may trigger the
autooxidation of the reduced nitroxide(s) (12). Although
both the nitroxide and hydroxylamine derivatives can
readily be detected by HPLC-EC (Figure 3A), to avoid
fluctuations in the ratio of these forms, the HPLC-EC
analysis can be preceded by incubation of the analyzed
solutions for 10-15 min with ascorbate. Figure 2B shows
the effectiveness of 5 mM ascorbate in reducing the PBN/
C5H11 and PBN/CH3 nitroxides. The latter procedure
would result in the detection of the total amount of the
C1-5 spin-trapped radicals as the hydroxylamine form of

ESR and HPLC-EC Detection of Alkyl Radicals
Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1243
alkanes, presumably formed via the intermediate gen-
eration of alkyl radicals (19). Hence, we have carried out
spin-trapping experiments with HPLC-EC detection to
attempt to identify alkyl radicals formed in the liver of
CCl4-treated rats.
Since CYP2E1 is known to be especially reactive in
catalyzing the metabolism of CCl4 (reviewed in ref 20),
Sprague-Dawley rats were treated for several days with
pyrazole, which is known to cause a 2-4-fold increase of
the hepatic CYP2E1 (21). Subsequently, the animals
were treated with CCl4 and PBN, and liver extracts were
analyzed by ESR and HPLC-EC for detection of hepatic
radical adducts. Figure 6 (ESR spectrum 1) depicts the
ESR spectrum of an ethanolic extract from the liver of a
CCl4 plus PBN-treated rat. The hyperfine structure (in
gauss) of the ESR spectrum presented in trace 1 suggests
the presence of the PBN/•CCl3 nitroxide (in 80% ethanol,
aN ) 14 G; aH ) 1.9 G). However, similar ESR spectra
may be exhibited by lipid and alkyl radical-derived PBN
nitroxides. The intensity of the ESR spectrum of the liver
extract was constant for a time period of 60 min. After
an addition of K3[Fe(CN)6] (1 mM), the intensity of the
observed ESR spectrum increased with a shift in the
F igu r e 4. Voltage dependence of HPLC-EC signals of PBN/
C2H5 and PBN/C5H11. Both hydroxylamine (A) and nitroxide (B)
forms of the adducts were subjected to EC-HPLC analysis at
different holding potential. The nitroxide and their hydroxyl-
amine derivatives were prepared as described in the Materials
and Methods and diluted 10 times in ethanol prior to HPLC
separation.
solution, however, revealed that both the nitroxide and
hydroxylamine forms of PBN/C3H7 and PBN/C5H11 were
produced (Figure 5B, spectrum I, peaks 1, 2 and 3, 4,
respectively). The chromatographic resolution was also
satisfactory in the analysis of the iso-pentyl- and pentyl-
derived PBN nitroxides that were obtained from 1-bromo-
2-methylbutane and 1-bromopentane, Na and PBN (Fig-
ure 5B, spectrum II, peaks 1 and 2, respectively). The
kinetics of formation of the isopentyl and the pentyl PBN
adducts is shown in Figure 5C.
hyperfine structure of the initial ESR spectrum to aN
)
14.3 G and aH ) 2.7, suggesting the presence of multiple
PBN-derived nitroxides (Figure 6, ESR spectrum 2). In
a detailed analysis of the liver extract from CCl4-treated
rats, J anzen et al. have identified PBN-derived nitroxides
•
of CCl3 and carbon- and oxygen-centered radicals (22,
23).
Within the time interval of PBN/C2-5 elution (Figure
7A, dashed rectangle, retention times of 7-18.5 min), the
HPLC-EC profile of the liver extracts exhibited three
major peaks. Peaks 1 and 2 had close but distinct
retention times as compared to those of the nitroxide and
hydroxylamine forms of PBN/C2H5 (Rt ) 8.2 and 10.6 min
compare to 8.4 and 11 min, respectively); their I/V curves
also differed from those of the authentic PBN/C2H5
ESR a n d HP LC-EC An a lysis of th e in Vivo Me-
ta bolism of CCl4. The hepatotoxicity of CCl4 is believed
to result largely from the metabolic activation of the
•
halocarbon molecule by P450 to •CCl3 radical. The CCl3
radical has been shown to form adducts with hepatic
molecules, to oxidize glutathione and protein thiols, and
to induce lipid peroxidation (LPO) (reviewed in ref 18).
The metabolism of CCl4 is also paralleled by a release of
F igu r e 5. ESR spectra and HPLC-EC chromatograms of alkyl derivatives of PBN obtained under the condition of the Wurtz reaction.
All reactions were carried out in dry hexane for 1 h at room temperature. The HPLC separations were carried out with a C18
Microsorb column (4.6 mm × 25 cm; 5 µm). (A) Trace 1: ESR spectrum of a solution of C3H7Br (0.08 M) and C5H11Br (0.08 M) in
hexane containing Na (0.01 g). Trace 2: overlapped, PC simulated ESR spectra of PBN/C3H7 and PBN/C5H11 nitroxides. (B) HPLC-
EC profiles of solutions of C3H7Br and C5H11Br (trace I) and C5H11Br and iso-C5H11Br (trace II) in hexane containing Na. In
chromatogram I, Peaks 1, 2 and 3, 4 reflect the elution of nitroxide and hydroxylamine forms of PBN/C3H7 and PBN/C5H11, respectively.
In chromatogram II, peaks 1 and 2 reflect the elution of the nitroxide forms of PBN/iso-C5H11 and PBN/C5H11, respectively. (C)
HPLC-EC-monitored kinetics of formation of PBN/iso-C5H11 and PBN/C5H11 nitroxides in a solution of hexane containing C5H11Br
(0.08 M), iso-C5H11Br (0.08 M), and Na (0.01 g).

1244 Chem. Res. Toxicol., Vol. 14, No. 9, 2001
Novakov et al.
well-resolved ESR spectrum (aN ) 14.39 G and aH ) 2.22
G; data not shown) that differed considerably from the
ESR spectra of the PBN/C2-5 nitroxides (Table 1). Peaks
3 and 4 reflected the elution of the nitroxide and hy-
droxylamine forms of PBN/•CCl3 (chromatograms I-III
compared to chromatogram IV). Chromatogram IV rep-
resents the HPLC-EC profile of a standard solution of
PBN/•CCl3 nitroxide and hydroxylamine, respectively
(15). At lower doses of PBN, peaks 3 and 4 considerably
decreased (Figure 7A, traces I-III), suggesting that an
•
increased CCl3-dependent formation of R• could have
occurred. However, our initial expectations to detect the
formation of ethyl, isopropyl, n-butyl, pentyl and iso-
pentyl derivatives of PBN were not confirmed as HPLC-
EC evidence for the presence of these adducts could not
be observed. When the HPLC separation was carried out
with a mobile phase with lower polarity, two additional
nitroxides and their hydroxylamine forms that previously
eluted with the front of the higher polarity eluent were
observed (Figure 7B, peaks 1-4). Upon treatment of the
liver extract with ascorbic acid, a decrease of peaks 1 and
2 was observed that was paralleled by an increase of
peaks 3 and 4 (Figure 7B, dashed lines), suggesting the
formation of the corresponding hydroxylamines. In sup-
port of the latter assumption, the fractions defined by
peaks 1 and 2 exhibited well resolved ESR spectra which
were missing in peaks 3 and 4 (Figure 7C, ESR spectra
1-4); the appearance of the same ESR spectra in frac-
tions 3 and 4 required an addition of potassium ferri-
cyanide (Figure 7C, spectra 5 and 6). The structure of
F igu r e 6. Effects of potassium ferricyanide on the ESR
spectrum of a liver extract from
a CCl4-treated rat. The
experiments were carried out in ethanol at 25 °C. Liver extract
was prepared as described in the Materials and Methods. ESR
spectra of liver extract from a CCl4-treated rat in the absence
(trace 1) or presence (trace 2) of K3[Fe(CN)6] (1 mM) were
recorded after an incubation of 20 min.
adducts (data not shown). The fraction defined by peak
1 did not exhibit any ESR activity in the absence or
presence of K3[Fe(CN)6], suggesting that the analyzed
solution did not contain either a nitroxide or hydroxy-
lamine form of PBN. The fraction under peak 2 had a
F igu r e 7. HPLC-EC analysis of liver extracts from CCl4-treated rats. Liver extracts were prepared as described in the Materials
and Methods. The HPLC separations were achieved with a C18 Microsorb column (4.6 mm × 25 cm; 5 µm). The HPLC-EC profiles
of the liver extracts were obtained via an elution with either 80% (panel A) or 30% (panel B) methanol. Chromatograms I-III were
obtained from experiments with rats loaded with 200, 100, and 5 mg of PBN/kg of body weight, respectively. Chromatogram IV
reflects the elution of a standard solution of the nitroxide (peak 3) and hydroxylamine (peak 4) form of PBN/CCl3 (prepared as
described in ref 15). Panel B depicts HPLC-EC profiles of a liver extract from a rat loaded with 200 mg of PBN/kg body weight. A
treatment of the extract for 10 min with ascorbic acid (0.1 M phosphate buffer, pH 7.4) resulted in a decrease of peaks 1 and 2 that
was paralleled by an increase of peaks 3 and 4 (dashed lines), respectively. In panel C are presented the ESR spectra of the fractions
defined by peaks 1-4 (panel B). The numeration of the ESR spectra corresponds to the numeration of the HPLC peaks in panel B;
ESR spectra 5 and 6 were obtained via treatment of fractions 3 and 4 for 15 min with K3[Fe(CN)6] (1 mM).

ESR and HPLC-EC Detection of Alkyl Radicals
Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1245
these adducts, however, remains unknown, but they are
not reflective of the C1-5 homologous series of alkyl PBN
adducts.
nitroxide and hydroxylamine forms of PBN/CCl3 (Figure
7A, peaks 3 and 4, respectively) were observed. When
the liver samples were eluted with a mobile phase with
lower polarity, the nitroxide and hydroxylamine forms
of two additional PBN adducts were observed (Figure 7,
Discu ssion
•
panels B and C). Since CCl3 and R• are equally reactive
(28), the detection of PBN/•CCl3 adducts attests that the
spin trapping efficiency of PBN should be sufficient for
intercepting alkyl radicals. Figure 7A illustrates that the
HPLC-EC peaks reflecting the elution of the PBN/•CCl3
adducts considerably decreased when lower doses of PBN
were injected. It could be speculated that the decreased
PBN/•CCl3 concentrations reflect less effective trapping
of •CCl3 by the lower concentrations of PBN which could
result in an increased amount of “free” •CCl3 to interact
with the hepatic phospholipids; this should ultimately
lead to an increased production of R•. Under these
experimental conditions, however, we could not detect the
formation of C1-5 alkyl radicals. We could not also detect
PBN/C1-5 adducts in model systems that are routinely
used for induction of lipid peroxidation, such as liver
microsomes, NADPH, and CCl4 (independently of the
presence or absence of O2), or liver microsomes plus iron
and ascorbate (data not shown), despite the evidence for
lipid peroxidation by assays for TBARS.
In conclusion, HPLC-EC protocols for the analysis of
PBN-spin trapped radical intermediates were estab-
lished. The HPLC-EC peaks reflecting the elution of the
PBN derivatives, in contrast to their ESR spectra, were
well resolved. The use of PBN as a spin-trapping agent
allowed a simultaneous detection of multiple radical
intermediates. The sensitivity of the electrochemical
detection of nitroxides was within the picomolar range.
Although an unsuccessful attempt was made for the in
vivo spin trapping of alkyl radicals, it is expected that
the HPLC-EC analysis of alkyl-derived PBN adducts may
be helpful for identification of spin-trapped radical spe-
cies in biological systems. The lack of PBN-detectable
alkyl radicals is likely due to the low steady-state
concentrations of R•, as only a small fraction of LO• go to
produce R• (29). Thus, the unstable nature of the lipid-
derived radicals and the multiple reaction pathways of
their decomposition reduces the efficient spin trapping
of R•, which remains a severe limitation of such indirect
detection techniques. However, the presented data do not
exclude the possibility that other nitrones may prove
more efficient scavengers of alkyl radicals.
In biological systems, the identification of spin trapped
radical species is difficult as usually various radical-
derived nitroxides with similar ESR spectra are formed.
Furthermore, the redox potential of the intracellular
milieu is in the range of -0.26 to -0.28 V (24). The
reducing environment in the cellular cytosol is main-
tained by reductants such as the thioredoxin and glu-
taredoxin systems, ascorbate and glutathione, and pro-
tein thiols. Recently, the redox potentials of a series of
nitroxides have been reported to be in the range of 0.5-
0.9V (25), suggesting that in spin trapping experiments
in biological systems a major reaction pathway can be
the reduction of the nitroxides to their ESR “silent”
hydroxylamine derivatives. The latter process would
considerably decrease the sensitivity of the ESR spin
trapping technique. To overcome these experimental
difficulties, we have conducted experiments for the in vivo
spin trapping of alkyl radicals in which the identification
of the PBN adducts was carried out by HPLC-EC. For
positive identification of alkyl metabolites, PBN/C1-5
nitroxides were prepared and used as HPLC-EC stan-
dards. HPLC separation with EC detection proved to be
reliable in distinguishing both the nitroxide and hydrox-
ylamine forms of the various PBN/C1-5 adducts.
In vivo generation of carbon-centered radicals was
carried out via administration of CCl4 to rats. The
hepatotoxicity of CCl4 is believed to result largely from
its metabolic activation by P450, especially P450 2E1, to
•CCl3 radical. The metabolism of CCl4 was also correlated
with the production of alkanes (18, 26, 27), presumably
formed via the intermediate generation of alkyl radicals.
According to Scheme 1, however, two experimental
problems for spin trapping of R• could be envisaged: (i)
at high concentrations of PBN, species 2, 3, and 5
•
(including CCl3) may be spin trapped, therefore, the
production of R• (7) could be inhibited, because of the
trapping of its radical precursors, and ii) at relatively low
concentrations of PBN, the reduction of R• to RH or/and
its recombination with O2 (Scheme 1, 7 f 8 and 7 f 9,
respectively) may be faster than its interaction with PBN.
Hence, we have treated rats with CCl4 and doses of PBN
ranging from those indicated in (6) (10 mg/kg of body
weight) to up to 200 mg/kg of body weight to attempt to
have an effective concentration of PBN which would allow
some formation of R• to occur and allow trapping of the
formed R• . One hour after the treatment, the animals
were sacrificed, and liver extracts were prepared and
analyzed by ESR and HPLC-EC. The ethanolic extracts
from the liver of CCl4 plus PBN-treated rats exhibited
ESR spectra with hyperfine structure suggesting the
formation of the PBN/CCl3 nitroxide. However, upon
addition of K3[Fe(CN)6] the intensity of the ESR spectra
increased with a small but noticeable shift of the hyper-
fine structure, suggesting the presence of multiple PBN-
derived hydroxylamines in the liver extract. The latter
observation should be taken into consideration when
attempts for reoxidation of the hydroxylamine derivatives
of spin trapped adducts are made.
Ack n ow led gm en t. This work was supported by
Grant ES-09648 from the National Institute of Environ-
mental Health Sciences (D.A.S.) and USPHS Grant AA-
09460 from the National Institute on Alcohol Abuse and
Alcoholism (A.I.C.).
Su p p or tin g In for m a tion Ava ila ble: Preparation of PBN/
1-5 nitroxides under the conditions of the Wurtz reaction. This
C
Material is available free of charge via the Internet at http://
pubs.acs.org.
Refer en ces
(1) de Zwart, L. L., Meerman, J . H., Commandeur, J . N., and
Vermeulen, N. P. (1999) Biomarkers of free radical damage
applications in experimental animals and in humans. Free
Radical Biol. Med. 26, 202-226.
Within the time interval of PBN/C1-5 elution, an
unidentified nitroxide (Figure 7A, peak 2) and the
(2) Gardner, H. W. (1989) Oxygen radical chemistry of polyunsatu-
rated fatty acids. Free Radical Biol. Med. 7, 65-86.

1246 Chem. Res. Toxicol., Vol. 14, No. 9, 2001
Novakov et al.
(3) Chamulitrat, W., and Mason, R. P. (1990) Alkyl free radicals from
the beta-scission of fatty acid alkoxyl radicals as detected by spin
trapping in a lipoxygenase system. Arch. Biochem. Biophys. 282,
65-69.
(4) Labeque, R., and Marnett, L. J . (1988) Reaction of hematin with
allylic fatty acid hydroperoxides: identification of products and
implications for pathways of hydroperoxide-dependent epoxida-
tion of 7,8-dihydroxy-7,8-dihydrobenzo[a]pyrene. Biochemistry 27,
7060-7070.
(16) Thennant, J . (1979) Imines, Nitrones, Nitriles and Isocyanamides.
In Comprehensive Organic Chemistry. The synthesis and Reac-
tions of Organic Compounds (Sir Derek Barton, W. D. O., I. O.
Sutherland, Eds.) Vol. 2, pp 715-721, Pergamon Press Ltd.,
Oxford.
(17) DeGray, J . A., and Mason, R. P. (1991) Biological spin trapping.
In Electron Spin Resonance (Atherton, N. M., Davies, M. J ., and
Gilbert, B. C., Eds.) Vol. 14, pp 246-301, Cambridge.
(18) Plaa, G. L. (2000) Chlorinated methanes and liver injury: high-
lights of the past 50 years. Annu. Rev. Pharmacol. Toxicol. 40,
42-65.
(19) Cohen, G. (1978) Lipid peroxidation: detection in vivo and in vitro
through the formation of saturated hydrocarbon gases. Ciba
Found. Symp. 177-185.
(20) Koop, D. R. (1992) Oxidative and reductive metabolism by
cytochrome P450 2E1. FASEB J . 6, 724-730.
(5) North, J . A., Spector, A. A., and Buettner, G. R. (1994) Cell fatty
acid composition affects free radical formation during lipid
peroxidation. Am. J . Physiol. 267, C177-C188.
(6) McCay, P. B., Lai, E. K., Poyer, J . L., DuBose, C. M., and J anzen,
E. G. (1984) Oxygen- and carbon-centered free radical formation
during carbon tetrachloride metabolism. Observation of lipid
radicals in vivo and in vitro. J . Biol. Chem. 259, 2135-2143.
(7) Ghio, A. J ., Kadiiska, M. B., Xiang, Q. H., and Mason, R. P. (1998)
In vivo evidence of free radical formation after asbestos instilla-
tion: an ESR spin trapping investigation. Free Radical Biol. Med.
24, 11-17.
(21) Winters, D. K., and Cederbaum, A. I. (1992) Time course
characterization of the induction of cytochrome P-450 2E1 by
pyrazole and 4-methylpyrazole. Biochim. Biophys. Acta 1117, 15-
24.
(8) Iwahashi, H., Albro, P. W., McGown, S. R., Tomer, K. B., and
Mason, R. P. (1991) Isolation and identification of alpha-(4-
pyridyl-1-oxide)-N-tert-butylnitrone radical adducts formed by the
decomposition of the hydroperoxides of linoleic acid, linolenic acid,
and arachidonic acid by soybean lipoxygenase. Arch. Biochem.
Biophys. 285, 172-180.
(9) Stronks, H. J ., J anzen, E. G., and Weber, J . R. (1984) Electro-
chemical detection of PBN spin adduct amynoxyls (nitroxides)
separated by high performance liquid chromatography. Anal. Lett.
17, 321-328.
(10) Pritsos, C. A., Constantinides, P. P., Tritton, T. R., Heimbrook,
D. C., and Sartorelli, A. C. (1985) Use of high-performance liquid
chromatography to detect hydroxyl and superoxide radicals
generated from mitomycin C. Anal. Biochem. 150, 294-299.
(11) Towell, J ., and Kalyanaraman, B. (1991) Detection of radical
adducts of 5,5-dimethyl-1-pyrroline n-oxide by the combined use
of high-performance liquid chromatography with electrochemical
detection and electron spin resonance. Anal. Biochem. 196, 111-
119.
(12) Stoyanovsky, D. A., Melnikov, Z., and Cederbaum, A. I. (1999)
ESR and HPLC-EC analysis of the interaction of hydroxyl radical
with DMSO: rapid reduction and quantification of POBN and
PBN nitroxides. Anal. Chem. 71, 715-721.
(22) J anzen, E. G., Towner, R. A., and Haire, D. L. (1987) Detection
of free radicals generated from the in vitro metabolism of carbon
tetrachloride using improved ESR spin trapping techniques. Free
Radical Res. Commun. 3, 357-364.
(23) J anzen, E. G., Towner, R. A., Krygsman, P. H., Lai, E. K., Poyer,
J . L., Brueggemann, G., and McCay, P. B. (1990) Mass spectros-
copy and chromatography of the trichloromethyl radical adduct
of phenyl tert-butyl nitrone. Free Radical Res. Commun. 9, 353-
360.
(24) Gilbert, H. F. (1990) Molecular and cellular aspects of thiol-
disulfide exchange. Adv. Enzymol. Relat. Areas Mol. Biol. 63, 69-
172.
(25) Morris, S., Sosnovsky, G., Hui, B., Huber, C. O., Rao, N. U., and
Swartz, H. M. (1991) Chemical and electrochemical reduction
rates of cyclic nitroxides (nitroxyls). J . Pharm. Sci. 80, 149-152.
(26) Ekstrom, T., Stahl, A., Sigvardsson, K., and Hogberg, J . (1986)
Lipid peroxidation in vivo monitored as ethane exhalation and
malondialdehyde excretion in urine after oral administration of
chloroform. Acta Pharmacol. Toxicol. 58, 289-296.
(27) Kieczka, H., and Kappus, H. (1980) Oxygen dependence of CCl4-
induced lipid peroxidation in vitro and in vivo. Toxicol Lett. 5,
191-196.
(28) Asmus, K. D., Lai, M., Monig, J ., and Schoneich, C. (1988) Radical-
induced degradation of organic halogen and sulfur compounds in
oxygenated aqueous solutions. In Oxygen radicals in biology and
medicine (Simic, M. G., Taylor, K. A., Ward, J . F., and von
Sonntag, C., Eds.) Vol. 71, pp 715-721, Plenum Press, New York.
(29) Qian, S. Y., Wang, H. P., Schafer, F. Q., and Buettner, G. R. (2000)
EPR detection of lipid-derived free radicals from PUFA, LDL, and
cell oxidations. Free Radical Biol. Med. 29, 568-579.
(13) LeGoff, E., Ulrich, S. E., and Denney, D. B. (1958) Studies on the
mechanism of the Wurtz reaction. The configuration of 2-bromo-
octane, 3-methylnonane and 7,8-dimethyltetradecane. J . Am.
Chem. Soc. 80, 622-625.
(14) Buettner, G. R. (1987) Spin trapping: ESR parameters of spin
adducts. Free Radical Biol. Med. 3, 259-303.
(15) Stoyanovsky, D. A., and Cederbaum, A. I. (1999) Metabolism of
carbon tetrachloride to trichloromethyl radical: An ESR and
HPLC-EC study. Chem. Res. Toxicol. 12, 730-736.
TX015507H