Journal of Neuroscience Methods 144 (2005) 127–135

Lipid peroxidation measurement by thiobarbituric

acid assay in rat cerebellar slices
Yngo J. Garcia a, b, c, d, ∗ , Antonio J. Rodrı́guez-Malaver a , Nancy Peñaloza a
a Laboratorio de Bioquı́mica Adaptativa, Departamento de Bioquı́mica, Facultad de Medicina,
Universidad de Los Andes (ULA), Mérida 5101, Venezuela
b Hospital Clinico de Mérida, Servicio de Neurocirugı́a, Mérida, Venezuela
c Unidad de Neurocirugı́a, Hospital Vargas, Universidad Central de Venezuela, Caracas, Venezuela
d Instituto de Medicina Experimental, Laboratorio de Fisiopatologı́a, Universidad Central de Venezuela, Caracas, Venezuela

Received 12 August 2004; received in revised form 22 October 2004; accepted 22 October 2004

Abstract

Lipid peroxidation by reactive oxygen species (ROS) is known to be involved in the damaging mechanism of several acute and chronic brain
disorders. The most prominent and currently used assay as an index for lipid peroxidation products is the thiobarbituric acid assay (TBA test).
It is based on the reactivity of an end product of lipid peroxidation, malondialdehyde (MDA) with TBA to produce a red adduct. However, it
is known that the MDA levels are frequently overestimated, that the reaction lacks specificity and mainly reflects the susceptibility of brain
tissue to the generation and degradation of newly formed lipid hydroperoxides under the TBA test conditions. The present paper shows that
artifactual lipid peroxidation by TBA test conditions can be prevented and that the MDA level overestimation can be minimized in cerebellar
slices. This can be done by incubating the slices in a continuous tissue perfusion system, by adding butylated hydroxytoluene (BHT) to the
homogenization solutions and by carrying out the assay anaerobically on deproteinizated supernatants of cerebellar slice homogenates. The
present research also showed that lipid peroxidation products generated during incubation of the slices by hydrogen peroxide (H2 O2 ) could
be measured without artifactual interference by the TBA test conditions.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Lipid peroxidation; Thiobarbituric acid assay; Thiobarbituric acid reactive substances; Malondialdehyde; Cerebellar slices; Butylated hydroxytoluene

1. Introduction with several acute conditions such as hyperbaric oxygena-

Brain tissues are rich in phospholipids, which can be at-
tacked by the highly reactive oxygen species (ROS) for the
initiation of lipid peroxidation. The most prominent assay
currently being used as an index for lipid peroxidation in bi-
ological systems is the measurement of the thiobarbituric acid
reactive substances (TBARS) through the thiobarbituric acid
(TBA) test (Gutteridge and Halliwell, 1990). In the brain, this
assay has been extensively used in the last three decades in
several experimental models such as homogenates, synap-
tosomes, slices and in vivo, in an effort to elucidate the
possible contribution of ROS to brain damage associated
∗ Corresponding author. Tel.: +58 274 2403095; fax: +58 274 2403045.
E-mail address: garciayngo@intercable.net.ve (Y.J. Garcia).
tion, brain ischemia, focal hemorrhage, traumatic brain injury
and chronic disorders such as Huntington’s chorea, Parkin-
son’s and Alzheimer’s diseases (Bawari et al., 1995; Boehme
et al., 1977; Bralet et al., 1991; Chakraborty et al., 2001;
Cini et al., 1994; Dirks and Faiman, 1982; Kücükkaya et al.,
1996; Santamarı́a et al., 1999). This assay is based upon the
formation of a red adduct (absorption maximum 532 nm) be-
tween TBA and malondialdehyde (MDA), a colorless end
product of lipid peroxide decomposition (Janero, 1990). Yet,
the TBA test has been widely criticized because of the low
efficiency of fatty-acid hydroperoxide breakdown to MDA
(Janero and Burghardt, 1988). It, moreover, lacks the requi-
site chemical specificity in order to be used as a reliable mea-
sure or even as an indicator of lipid peroxidation. Indeed, the
end products of lipid peroxide breakdown, aldehydes other

0165-0270/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jneumeth.2004.10.018
128 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
than MDA are TBA-reactive and produce orange or yellow
and also red MDA and non-MDA pigments (Janero, 1990;
Kikugawa et al., 1992; Kosugi et al., 1987). TBA-positive
products with important absorbance at 532 nm are generated
through oxidative damage to non-lipid molecules (Janero and
Burghardt, 1988; Ceconi et al., 1991). What is more, MDA
from non-lipids by peroxidation of amino acids, carbohy-
drates or nucleic acids have also been reported (Gutteridge
and Halliwell, 1990; Janero, 1990). Last but not least, arti-
factual TBARS during tissue homogenization or during the
acid-heating stage of the test are also produced (Draper et al.,
1993; Götz et al., 1993; Verbunt and Egas, 1996). In order to
obtain an in vitro experimental protocol to study the effect of
ischemia/reperfusion and l-glutamate on lipid peroxidation
using the TBA assay in cerebellar slices, it is necessary to
take into account these considerations and standardize con-
ditions to measure the preformed TBARS, not the artifactual
ones produced during the several stages of the TBA test. The
advantage of using tissue slices is that they retain the greatest
part of the tissue organization. Consequently, if lipid per-
oxidation measurement on this preparation is optimized, it
could represent a model better than homogenates or cultures
for morphological, metabolic, electrophysiological or neuro-
chemical studies (Bralet et al., 1991; Garthwaite et al., 1992;
Garcia et al., 1995; Pellmar et al., 1989).
2. Materials and methods
2.1. Chemicals
The composition of the Krebs bicarbonate solution was
as follows: 5 mM KCl, 1.2 mM KH2 PO4 , 1.3 mM MgSO4 ,
125 mM NaCl, 2 mM CaCl2 , 10 mM glucose and 25 mM
NaHCO3 equilibrated with O2 /CO2 to pH 7.4. Buty-
lated hydroxytoluene (BHT), hydrogen peroxide (H2 O2 ), 2-
thiobarbituric acid, Lowry assay kit, bovine serum albumin
(BSA), chloroform, methanol and benzene were obtained
from Sigma.
2.2. Animals, preparation of cerebellar slices and
incubation protocols
Male Sprague–Dawley rats, weighing 180–220 g, were
decapitated. Twenty parasagittal slices (0.4 mm thick) of
cerebellum were obtained by using a McIlwain tissue chop-
per. Slices were immediately placed in a Petri dish contain-
ing Krebs bicarbonate solution under continuous oxygena-
tion with 95% O2 /5% CO2 . All these steps were carried out
at 4 ◦ C. Slices were transferred to a 30 ml tissue chamber and
superfused at a rate of 2 ml/min with Krebs bicarbonate so-
lution at 37 ◦ C under continuous oxygenation from another
30 ml chamber. Solution from tissue chamber was drained
back to the oxygenation chamber. A polystaltic pump (Buch-
ler), silicon tubes and water bath (Fisher Isotemp, Model 110)
were used. Slices were incubated for 30 min under control and
ischemic conditions with Krebs solutions gassed with 95%
O2 /5% CO2 and a glucose-free medium gassed with 95%
N2 /5% CO2 , respectively. Slices were also incubated with
hydrogen peroxide at different concentrations and lengths of
time in order to study its effect on lipid peroxidation.
2.3. Extraction and quantification of lipids and proteins
from homogenate, cerebellar membranes and
deproteinizated cerebellar membrane fractions
Lipids were extracted and purified by a modified
Bligh–Dyer procedure (1959). Four cerebella were homoge-
nized in 4480 ␮l of ice-cold deionized water (obtained from
Barnstead Mega-Pure MP-1 and Bantan Cartridge holder,
Model D0800) with a teflon pestle (10–12 strokes). An aliquot
of the tissue homogenate, i.e., 1600 ␮l, was distributed in
four glass tubes so that each one contained 400 ␮l of the final
volume. To liberate any covalently bound lipid, homogenates
were hydrolyzed in acid (250 ␮l TCA at 40%) at pH 3, 23 ◦ C,
for 10 min. The remaining homogenate was centrifuged at
3000 × g for 15 min. Aliquots of 400 ␮l from supernatants
(cerebellar membranes) were transferred to four glass tubes
and 250 ␮l of TCA at 40% were added. From the remaining
supernatant, 800 ␮l were transferred to four centrifuge tubes
and 500 ␮l of TCA at 40% were added and then centrifuged
at 3000 × g for 15 min to obtain cerebellar deproteinizated
membranes. From each fraction, 650 ␮l (400 ␮l of the cere-
bellar fraction and 250 ␮l of TCA at 40%) was used to purify
lipids. One thousand eight hundred sixty seven microliters
of chloroform and 934 ␮l of methanol (2:1, v/v) were then
added. After thorough mixing, the samples were centrifuged
at 2000 × g for 15 min. The supernatant was retrieved and
mixed with 1866 ␮l of water and 1866 ␮l of chloroform (1:1,
v/v). After centrifugation at 2000 × g for 15 min, the final
chloroform phase containing the purified lipids was with-
drawn. Two hundred fifty microliters of benzene was added
to aid removing water traces. The solvent was evaporated un-
til dryness under nitrogen at room temperature (23 ◦ C), and
the lipids were weighed with an analytical balance.
For protein quantification, a sample of 20 ␮l was taken
from each cerebellar fraction for determinations by Lowry’s
assay (1951) using BSA as a standard.
2.4. Assay of lipid peroxides and protein estimation on
cerebellar slices
Lipid peroxidation was evaluated by measuring the
TBARS according to the TBA test described by Kovachich
and Mishra (1980) with the following modifications. After
incubations in the perfusion system, five slices were added to
ice-cold deionized water containing 5 ␮l of BHT dissolved
in methanol for HPLC grade (10%, w/v). Water was previ-
ously gassed with 100% N2 for 60 min to remove oxygen.
Tissue homogenates were prepared on ice using a teflon pes-
tle (10–12 strokes) and centrifuged at 3000 × g for 15 min.
The supernatant that contained microsomes, mitochondria
 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
and synaptosomes (Subbarao and Richardson, 1990) was re-
trieved. From this supernatant, 20 ␮l was taken to determine
the amount of proteins using Lowry’s assay (1951). Three
hundred and fifty microliters was then transferred to a cen-
trifuge tube. To precipitate soluble and membrane proteins
and obtain deproteinizated membranes, 250 ␮l of TCA at
40% were added and centrifuged at 3000 × g for 15 min. All
these operations were carried out at 4 ◦ C or on ice. Five
hundred microliters of thiobarbituric acid reagent (0.67%
TBA and 0.05 N NaOH in 50 ml of deionized water previ-
ously gassed for 60 min with 100% N2 ) was added to 500 ␮l
of the resulting supernatant. This solution was placed un-
der a stream of nitrogen at 4 ◦ C for 1 min and was after-
wards heated at 100 ◦ C for 15 min. This stage was carried
out on a N2 gas phase obtained by displacing the air gas
phase with 100% N2 before closing the tubes. After heat-
ing, the tubes were cooled in a water bath at room tempera-
ture. Clear solutions were obtained with this procedure suit-
able for direct spectrophotometric measurement at 532 nm. A
Perkin-Elmer Spectrophotometer Lambda 3B was used. No
evidence for interfering substances was obtained by spectral
analysis of the pink pigment. For quantitation, calibration
curves were constructed using 1,1,3,3-tetramethoxypropane
(TMP, malondialdehyde-bis-(dimethylacetal)) as a standard,
ranging from 0.0 to 6070 pmol for absorption measurements.
The formation of TBARS was expressed as malondialdehyde
equivalents per mg protein.
For measuring TBARS in homogenate or membrane frac-
tions from slices, aliquots of 350 ␮l were taken and mixed
immediately with 250 ␮l of TCA at 40%. From these ho-
mogenate and membrane solutions, aliquots of 500 ␮l were
retrieved, and 500 ␮l of TBA reagent was added. The TBA
test was then performed as described and the samples were
centrifuged at 2000 × g, 4 ◦ C for 15 min. The absorbance of
the upper phase at 532 nm was read. In these fractions, pro-
teins were also measured in supernatants of cerebellar slice
homogenates.
2.5. Assay and estimation of lipid peroxides on lipids
from homogenates, membranes and deproteinizated
membrane fractions of cerebellar slices
Cerebellar slices were obtained and homogenized as de-
scribed. Lipids were purified from aliquots of 350 ␮l of ho-
mogenates and membrane samples. To obtain deproteinizated
Table 1
Biochemical properties of rat cerebellar membranes
Fraction Protein
% ␮g
Homogenate 100
Cerebellar membranes 55.02 ± 0.62
Deproteinizated cerebellar membranes 5.83 ± 1.07
Cerebellar fractions were isolated from rat brains as described in the Section 2. Results are expressed as means ± S.E.M. from at least four experiments. As
far as lipids are concerned, there was a significant difference between the fractions (overall one-way ANOVA F(2, 11) = 25.200 and p < 0.01). The post-hoc
analysis was not significant when comparing cerebellar membranes with deproteinizated cerebellar membranes.
129
membranes, 450 ␮l was taken from membranes samples and
deproteinizated with 350 ␮l of TCA at 40%. A total volume
of 600 ␮l (350 ␮l of the cerebellar fraction and 250 ␮l of TCA
at 40%) was taken from all cerebellar fractions to purify lipids
as described above. The lipid film was resuspended in 250 ␮l
of sodium dodecyl sulfate (SDS) at 5%. Two hundred fifty
microliters of TCA at 40% and 500 ␮l of TBA reagent was
added. The TBA assay was then performed in the N2 phase as
described. Because of the sample opalescence, blanks with
purified lipids were used in these experiments for the spec-
trophotometric lecture at 532 nm. For all cerebellar fractions,
proteins were measured in supernatants of cerebellar slice ho-
mogenates.
2.6. Assessment of “bound” MDA
To liberate any bound MDA to proteins in supernatants
before the TBA test analysis in deproteinizated membranes,
350 ␮l of supernatants of homogenate were hydrolyzed us-
ing 250 ␮l of 40% TCA, which gives a solution pH of 3. For
incubation, a 60 ◦ C temperature during 20 min before cen-
trifugation at 4 ◦ C was chosen (Ichinose et al., 1989). In an
effort to prevent lipid auto-oxidation during acid hydrolysis
incubations, experiments were carried out on the N2 phase.
2.7. Statistical analysis
All data are presented as means ± S.E.M. Data were an-
alyzed using one-way ANOVA for between-groups compar-
isons followed by a Tukey’s post-hoc test for multiple com-
parisons. A standard significance level of p < 0.05 was used.
3. Results
The biochemical properties of rat cerebellar fractions are
summarized in Table 1. The efficiency of cerebellar pro-
tein extraction by centrifugation and acidification with TCA
was readily apparent from the lower protein content of the
membranes and of previously acidified and centrifuged mem-
branes with respect to starting homogenate. Yet, the removal
of protein by TCA treatment was incomplete. By contrast, the
membranes and deproteinizated membranes represented ap-
proximately 73% of the cerebellar homogenate lipid. The im-
portant lipid recovery indicates that the isolated membranes
were representative of the cerebellar membranes in situ.
Lipid
% mg
493.64 ± 10.88 [4] 100 45.25 ± 1.38 [4]
271.60 ± 3.07 [4] 76.24 ± 4.05 34.50 ± 1.83 [4]NS
28.79 ± 3.04 [4] 70.60 ± 3.60 31.95 ± 1.63 [4]
130 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
Table 2
Effect of BHT on the TBA signal in supernatants of cerebellar slice ho-
mogenates at different stages of the TBA test
Stage Malondialdehyde (pmol/mg)
Without BHT 366.80 ± 47.60 [4]*
Pre-homogenization Not detected [4]
Post-homogenization 64.33 ± 12.66 [4]
Before heating Interferencea
Slices were obtained immediately after cerebellum removal and the TBA
assay in which BHT (5 ␮l; 10%, w/v) was added at different stages was
performed. Values are expressed as means ± S.E.M. of the MDA equivalent
concentrations from at least four experiments. There was a significant dif-
ference between BHT-free experiments and those in which BHT was added
after the homogenization (overall one-way ANOVA F(1, 7) = 50.287).
a Adding BHT before heating produces interference for spectrophotomet-
ric measurement by turbidity.
∗ p < 0.01.
The effect of the BHT antioxidant on tissue processing
as well as the acid-heating stage of TBA assay was as-
sessed in deproteinizated supernatants of cerebellar slices
homogenates obtained immediately after removal of the cere-
bellum. Table 2 shows that if BHT is omitted from the homog-
enization solution there is an important TBARS production.
By contrast, the addition of BHT to the homogenization solu-
tion yielded TBARS levels below the detection limit. When
BHT was added to the homogenate, it inhibited the TBARS
generation in 82%. The homogenization step would thus ac-
count for approximately only 18% of TBARS generation dur-
ing the assay. The BHT added to the deproteinizated super-
natants before heating produced interference due to turbidity.
This explains why spectrophotometric measurements could
not be carried out (Table 2).
It has been reported that in the liver, brain homogenates
and cardiac membranes, TBA test overestimates MDA lev-
els because some aldehydic compounds and other non-lipid-
related, non-MDA substances are capable of reacting with
TBA to produce absorbance spectra similar to those of the
MDA-complex (Janero and Burghardt, 1988; Kikugawa et
al., 1992). Moreover, it has been found that there is a close
agreement between the production of TBARS and the gen-
eration of MDA, only if the TBA test is performed in lipids
purified from membranes (Janero and Burghardt, 1988). This
is why we studied TBARS production in the different cere-
bellar fractions and the effect of adding BHT to their ho-
mogenization solutions. Table 3 shows that in the absence of
BHT, the highest values were found in membranes, followed
by homogenate and deproteinizated membranes. There was a
significant difference between these fractions, but there was
no difference between deproteinizated membranes and lipids
purified from them. BHT addition inhibited TBARS produc-
tion to the point that they became undetectable, in membrane
and deproteinizated membranes, but it decreases significantly
the TBA signal in homogenate and purified lipids from de-
proteinizated membranes.
We also performed a comparative study between the dif-
ferent cerebellar fractions and the lipids purified from them.
There was a significant difference between homogenate and
Table 3
Effect of BHT on the TBA signal in different fractions of cerebellar slices
Fraction Malondialdehyde (pmol/mg)
Without BHT With BHT
Homogenate 567.76 ± 40.41 [4]* 88.47 ± 48.32 [4]
Membranes 927.28 ± 63.97 [4] Not detected [4]
Deproteinizated membranes 366.80 ± 47.60 [4]** Not detected [4]
Purified lipids 246.22 ± 38.41[4] NS 140.06 ± 10.62 [4]
Slices were obtained immediately after cerebellum removal and the TBA
assay was performed with and without BHT (5 ␮l; 10%, w/v). Values are
expressed as means ± S.E.M. of the MDA equivalent concentrations from
at least four experiments. There was a significant difference between the
fractions without BHT (overall one-way ANOVA F(3, 15) = 46.055 and
p < 0.01). The post-hoc analysis was found to be not significant when com-
paring purified lipids with deproteinizated membranes but it was significant
when comparing homogenates with membranes and deproteinizated mem-
branes with homogenate.
∗ p < 0.01.
∗∗ p < 0.05.
membranes and the lipids purified from them, but not between
deproteinizated membranes and their lipids. The TBA signal
values of lipids purified from the different cerebellar fractions
were equivalent (Fig. 1). Table 3 and Fig. 1 also show that
the measurement of TBARS in homogenate and membrane
cerebellar fractions indeed significantly overestimates their
values in comparison with the lipids obtained from them.
TBA signal was 55% and 21% in lipids compared with start-
ing homogenate and membranes, respectively. However, this
is not so when comparing values from deproteinizated mem-
branes and their lipids.
In an attempt to explain the TBA signal results, obtained
in the comparative study carried out between the cerebellar
fractions and the lipids extracted from them, an absorption
spectra of these fractions was performed. As can be seen from
Fig. 1. TBA signal associated with tissue processing in the absence of BHT
in homogenate (H), lipids isolated from homogenate (LH), membranes (M),
lipids isolated from membranes (LM), deproteinizated membranes (DM)
and lipids isolated from deproteinizated membranes (LDM). The different
fractions and the lipids purified from them were obtained from cerebellum
slices as described in the Section 2 and analyzed spectrophotometrically for
MDA equivalents content. Values are expressed as means ± S.E.M. from at
least four experiments. A significant between-group difference was found
(overall one-way ANOVA F(5,23) = 52.630 and p < 0.01). The post-hoc anal-
ysis was found to be not significant when comparing DM with LH, LM and
LDM, but it was significant when comparing H with LH (*p < 0.01) and M
with LM (**p < 0.01).
 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
Fig. 2. Absorption spectra of homogenate, membranes and deproteinizated
membranes. The different fractions were obtained from cerebellum slices
as described in the Section 2 and analyzed by spectrophotometry for TBA
signals.
Fig. 2, the homogenate spectrum presents three peaks con-
sisting in a yellow 455 nm, orange 495 nm and red 532 nm
absorbing pigments. In the membrane spectrum, the yellow
pigment disappeared and there was a marked attenuation of
the orange pigment. By contrast, the red pigment exhibited
the maximum absorbance. The deproteinizated membrane
fraction and the lipids extracted from it (not shown) had sim-
ilar spectra with a unique peak at 532 nm. However, tiny
peaks at 455 and 495 nm were observed in lipids purified
from deproteinizated membranes 1 h after the acid-heating
stage. These peaks appear immediately in deproteinizated
membranes and their lipids if the acid-heating stage of the
TBA test was performed under an air or oxygen phase (data
not shown).
It is possible that contributions of bona fide MDA to TBA
signal in the cerebellar membrane fractions could be orig-
inated from bound MDA liberated from soluble and mem-
brane proteins during the heat-acid stage of the TBA test
(Ichinose et al., 1989). These could account for the higher
values recorded in the membranes and for the lower val-
ues obtained in deproteinizated membranes because of the
loss of bound MDA when soluble and membrane proteins
were discarded after centrifugation. To test this hypothe-
sis, an acid hydrolysis of homogenate supernatants in N2
phase to prevent lipid peroxidation during incubation was per-
formed at 60 ◦ C for 20 min (Ichinose et al., 1989). When com-
paring hydrolyzed membrane samples incubated with TCA
(386.46 ± 22.28 pm/mg protein; mean ± S.E.M. and n = 4)
with non-incubated membranes (366.62 ± 47.63 pmol MDA
equivalents/mg protein; mean ± S.E.M. and n = 4), only a
small increase of approximately 5% of TBA signal was ob-
served in the hydrolyzed membranes. This level of bound
MDA was too low to account for the 60% difference between
deproteinizated membrane TBA-reactivity and the TBA sig-
nal from membranes.
In order to test the behavior of the TBA signal in cere-
bellar slices previously incubated in the superfusion system,
the measurement of TBARS in the presence and absence of
BHT was carried out in deproteinizated supernatant fractions
131
Table 4
Effect of BHT on the TBA signal in supernatants of cerebellar slice
homogenates
Condition Malondialdehyde (pmol/mg)
Without BHT With BHT
Control 380.51 ± 54.30 [4] Not detected [4]
Ischemia 376.90 ± 70.45 [4] Not detected [4]
Immediate 366.80 ± 47.60 [4] Not detected [4]
Slices were incubated and superfused during 30 min either with a Krebs
bicarbonate solution gassed with 95% O2 /5% CO2 (control) or with a glu-
cose free-medium Krebs bicarbonate solution gassed with 95% N2 /CO2
(ischemia) or were taken immediately from the cerebellum. When BHT was
used (5 ␮l; 10%, w/v), it was added to the homogenization solutions. Val-
ues are expressed as means ± S.E.M. of the MDA equivalent concentrations
from at least four experiments. No significant difference was found between
the BHT-free groups (overall ANOVA F(2, 11) = 0.020 and p = 0.980).
under the following conditions: control, ischemic and imme-
diately after removal of the cerebellum. Under control and
ischemic conditions, the slices were incubated during 30 min
with Krebs solutions. Table 4 shows that in the absence of
BHT, there was an important production of TBARS. No sig-
nificant differences were observed between these groups. If
BHT is added to the homogenization solution, TBARS are
not detected in any of the conditions studied.
H2 O2 has been previously used to generate injury by
ROS on several systems (Ortega-Gutiérrrez et al., 2002;
Pellmar et al., 1989; Sewerynek et al., 1995). This is why
it was used to study its effect on lipid peroxidation under the
above-mentioned experimental conditions. Slices were incu-
bated with H2 O2 at a concentration of either 1, 10, 100 or
1000 mM for 30 min or of 200 mM H2 O2 for either 2, 15,
30, 45 or 60 min. After homogenization in the presence of
BHT, a TBA assay was performed on deproteinizated su-
pernatants of cerebellar slice homogenates. H2 O2 increased
TBARS levels in cerebellar slices incubated for 30 min in a
concentration-dependent manner (Fig. 3). The H2 O2 concen-
Fig. 3. Effect of different concentrations of H2 O2 on the level of lipid perox-
idation products (TBARS) in cerebellar slices. Incubation time was 30 min.
Values are indicated as means ± S.E.M. of the MDA equivalent concentra-
tions from at least four experiments. A significant between-group difference
was recorded (overall one-way ANOVA F(3,15) = 128.462 and p < 0.01).
The post-hoc analysis was significant when comparing 10 mM of H2 O2
with 1 mM of H2 O2 (*p < 0.05) and 100 mM of H2 O2 with 10 and 1000 mM
of H2 O2 (**p < 0.01).
132 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
Fig. 4. Accumulation of lipid peroxidation products (TBARS) in cerebellar
slices following incubation for either 2, 15, 30, 45 or 60 min in the presence
of 200 mM H2 O2 . Values are indicated as means ± S.E.M. of the MDA
equivalents content from at least four experiments. A significant between-
group difference was found (overall one-way ANOVA F(4,19) = 160.602 and
p < 0.01). The post-hoc analysis was significant when comparing the values
recorded at 15 min with those recorded at 2 and 30 min (*p < 0.05) and those
observed at 45 min with those recorded at 30 and 60 min (**p < 0.01).
tration used (200 mM) also increased lipid peroxidation in a
time-dependent manner (Fig. 4).
4. Discussion
The TBA test has been criticized for its lack of speci-
ficity and its risk of lipid auto-oxidation during the assay
procedure. In the present work, a modified method for deter-
mining lipid peroxidation on cerebellar slices is presented.
The objective of these modifications was to attempt to over-
come these problems while at the same time maintaining the
convenience, ease and rapidity of the original procedure.
Most of the TBA assay was performed on homogenates
or membranes. In these fractions, the risk of overestimation
of bona fide MDA levels and the susceptibility to oxygen-
induced lipid peroxidation in vitro is well documented (Cini
et al., 1994; Götz et al., 1993; Janero, 1990; Verbunt and
Egas, 1996). Our findings (see Table 1) indicate that lipids in
deproteinizated cerebellar membranes are also representative
of the cerebellar membranes in situ.
Table 2 shows that if the TBA assay on deproteinizated
supernatants of cerebellar slice homogenates is performed
with no antioxidants, then important amounts of TBARS are
produced, even if the homogenization and the acid-heating
stage are performed anaerobically. It is likely that these re-
sults mainly represent the artifactual basal values of TBARS
generated during the assay previously reported by several re-
searchers (Gutteridge and Halliwell, 1990; Götz et al., 1993;
Li and Chow, 1994). However, the concentrations of these
basal values were significantly lower than those obtained in a
previous study in supernatant cerebellum fractions (Subbarao
and Richardson, 1990). The reason for such a discrepancy ap-
pears to be methodological. Indeed, in our study there was
no incubation of homogenates before the TBA test was per-
formed, and homogenates spontaneously formed appreciable
amounts of MDA (Cini et al., 1994). Furthermore, the anaer-
obic conditions during the boiling step of the TBA test could
contribute to partially prevent lipid auto-oxidation. Finally,
the supernatants were deproteinizated before performing the
TBA test. Although the peroxidation of myocardial mem-
brane proteins did not yield appreciable amounts of TBA-
reactive substances, it is likely that under free-radical attack,
the integrity of phospholipid–protein-surface carbohydrate
chains of membrane milieu could be necessary to stimulate
the formation of non-lipid-related TBA reactive substances
more efficiently than MDA formation (Janero and Burghardt,
1988). Our findings (see Table 2) suggest also that homog-
enization could produce artifactual TBARS. This has been
reported before for the TBA–MDA complex, free MDA and
TBARS, and it has been attributed to the release of catalytic
iron from the cells injured by the procedure (Chakraborty
et al., 2001; Draper et al., 1993; Verbunt and Egas, 1996).
However, like in previous reports in homogenates, most of
the artifactual TBARS obtained were produced during the
acid-heating stage of the assay by the auto-oxidation of co-
existing unoxidized unsaturated fatty acids (Gutteridge and
Halliwell, 1990; Götz et al., 1993). This suggests that trace
amounts of oxygen from the starting homogenate still persist
in the supernatants. It can also be seen that the addition of
BHT to deionized water before but not after the homogeni-
sation, as is usually the case, allowed the suppression of the
artifactual TBA signal until it became undetectable.
When BHT was added, artifactual TBARS production as-
sociated with tissue homogenization and acid-heating stage
was inhibited in membranes and deproteinizated membranes.
The inhibition was lower in the homogenate, probably be-
cause this fraction is richer in TBARS and traces of oxygen.
In a study realized on rat brains, greater TBARS formation
was seen in cerebellum homogenates and pellets. This was
accounted for by the nucleic acid content of this fraction
(Subbarao and Richardson, 1990). The inhibition on puri-
fied lipids by BHT was lower. In these samples, the BHT
added to the homogenization solution could be lost during
the rest of the procedure of lipid extraction with organic sol-
vents and evaporation. The results summarized in Table 3
and Fig. 1 suggest that in anaerobic conditions lipid perox-
idation assessment, based on MDA measurement by a TBA
test in deproteinizated supernatants, could produce more re-
liable results than procedures performed in homogenate or
membranes.
The oxidation of polyunsaturated fatty acids is accom-
panied by the formation of a complex mixture of products
including aldehydes, such as alkanals, alk-2-enals, alka-2,4-
dienals and malondialdehyde (Esterbauer et al., 1982; Kiku-
gawa et al., 1992). Several reports have also dealt with the
reaction of TBA with these saturated and unsaturated mono-
functional aldehydes. Saturated aldehydes can produce a red
pigment via the formation of substituted alk-2-enals and pyra-
nopyrimidine adducts, but they mainly produce a yellow pig-
ment, particularly a reaction of hexanal with TBA (Draper
et al., 1993; Kosugi et al., 1987). Alk-2-enals can produce
yellow, orange and non-MDA red pigments (Kosugi et al.,
 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
1987). Alka-2,4-dienals produce true 1:2 adduct of MDA
red pigment in an amount 1/10–1/20 of that from MDA
(Kosugi et al., 1988). As shown in Fig. 2, the reaction of
homogenate cerebellar fractions with TBA produces a yel-
low pigment with a maximum absorption at 455 nm, an or-
ange pigment at 495 nm and a red pigment at 532 nm. It is
interesting to note that the absorbance of the yellow pigment
disappeared, and the orange pigment was greatly attenuated
with the first centrifugation. If alkanals and alk-2-enals are
less polar aldehydes than alka-2,4-dienals and malondialde-
hyde, these less soluble molecules could be excluded from
supernatants and retained mainly in the microsomal pellets.
Supernatants could be enriched by aldehydic products partly
soluble in water. Alkenals and mainly alka-2,4-dienals could
thus account for the high absorbance at 532 nm observed in
membrane cerebellar fractions. Furthermore, unlike alkanals
and alk-2-enals, the reaction of the alka-2,4-dienals with a
large amount of TBA (Kosugi et al., 1988), now available by
the relative absence of the less polar aldehydes, could also
contribute to the greater absorbance of the red pigment in
microsomal supernatants. Previous works have reported that
unsolubilized lipid could be removed by centrifugation be-
fore spectrophotometric measurement and that the proportion
of alkanals contained in liver microsomal pellets were signif-
icantly higher than in microsomal supernatants (Esterbauer
et al., 1982; Pikul et al., 1983). It is moreover well known that
the formation of pigments from aldehydes (except MDA) is
completely suppressed when the oxygen in the reaction mix-
ture is substituted with nitrogen before the reaction (Kosugi
et al., 1988; Kikugawa et al., 1992). It is possible that under
the N2 phase traces of oxygen are higher in homogenate and
cerebellar membranes than in the deproteinizated cerebellar
fraction. This could explain why the lowest levels of TBA
signal were found in deproteinizated membranes. The fact
that the TBA assay performed under O2 or air phase pro-
duced yellow and orange pigments and an increase of the red
pigment on deproteinizated membranes and purified lipids fa-
vors this hypothesis. Furthermore, it is possible that because
membrane acidification and centrifugation were performed
before the TBA assay, a non-lipid-related molecule associ-
ated with the membrane milieu (Ceconi et al., 1991; Janero
and Burghardt, 1988) can be excluded or inhibited to form a
non-MDA but red pigment. This could also explain the lower
values obtained in deproteinizated membranes. Whatever the
answer, the fact that the TBA signal measured in homogenate
and membranes overestimates the values of its lipids and the
equivalent results between deproteinizated membranes and
its lipids suggest that this important overestimation of MDA
by the TBA test could be greatly eliminated. Moreover, the
concentrations of TBARS were in order of magnitude lower
than the concentrations usually reported in the brain and my-
ocardium. This suggests a closer correlation between TBARS
and true MDA in these experimental conditions (Ceconi et al.,
1991). This procedure of precipitating any sample protein
and analyzing only the clarified supernatant for TBA reac-
tivity has been recommended before (see review by Janero,
133
1990). From the results reported here, it seems that in the
cerebellum this procedure could eliminate not only non-lipid
substances which could influence the TBA signal, but it could
also partially remove aldehydes other than MDA and mini-
mize their effects on the TBA reactivity of the remaining
aldehydes if oxygen levels are absent or low. These mod-
ifications could increase the specificity of the TBA assay
(see Janero, 1990).
The aim of the present work was also to minimize the sus-
ceptibility to oxidative stress during the different stages of the
TBA assay (Janero, 1990; Götz et al., 1993) searching for a
real basal value which would allow us to detect small amounts
of preformed lipid hydroperoxides and/or TBARS generated
under controlled conditions in vitro. To prevent artifactual
lipid peroxidation during incubation of the slices by oxygen
high-pressure or mechanical stress (Dirks and Faiman, 1982;
Kovachich and Mishra, 1980), a continual tissue perfusion
system was chosen. Table 4 shows that the stay of the slices
in the tissue chamber under aerobic conditions does not pro-
mote lipid peroxidation since their levels of TBARS are not
significantly different when compared with the values from
slices with no incubation and slices incubated under ischemic
conditions. It can be also seen that the use of BHT prevents
TBARS artifactual generation in the three experimental con-
ditions. These results could contribute to distinguish between
pigment formation resulting from degradation of lipophilic
peroxides having previously existed and those having been
newly formed during homogenization or by auto-oxidation of
unoxidized lipids during the acid-heating stage of the TBA
assay (Götz et al., 1993; Verbunt and Egas, 1996). To test
this hypothesis cerebellar slices were incubated with H2 O2
in presence of BHT.
The incubation of the slices with H2 O2 induced a sig-
nificant increase on TBARS levels in a concentration and
time-dependent manner. This has been previously reported
for homogenates (Ortega-Gutiérrrez et al., 2002; Sewerynek
et al., 1995) and cortical and hippocampus slices (Oubidar
et al., 1996; Pellmar et al., 1989). The principal mechanism
of H2 O2 toxicity is thought to involve the generation of the
highly reactive hydroxyl (OH) radical through its interaction
with Fe2+ by the Fenton reaction (Gutteridge and Halliwell,
1990; Oubidar et al., 1996; Sewerynek et al., 1995).
The use of BHT in the solution of homogenization pre-
vents artifactual lipid peroxidation by tissue processing and
lipid oxidation in the TBA test. Measurements of the TBA
signal on deproteinizated membranes in anaerobic conditions
were equivalent to the signal values detected in lipids pu-
rified from membranes. The use of the tissue chamber in
the superfusion system described here seems to prevent ar-
tifactual lipid peroxidation during incubation of the slices.
Altogether these modifications and procedures seem to in-
crease the fidelity of the lipid peroxidation measurement by
the TBA test on rat cerebellar slices. The concentration, time-
dependency and the amounts of the TBARS detected on cere-
bellar slices incubated with H2 O2 suggest that in this in vitro
system, it was possible to prevent artifactual TBARS gen-
134 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135
eration because of tissue processing and TBA test condi-
tions. Furthermore, the spectrophotometric lecture on depro-
teinizated supernatants allowed us to minimize the detection
of the TBA signal from peroxidation of non-lipid substances
by this oxidative stress during slice incubation. The amounts
obtained probably represent mainly the TBARS produced by
radical attack of fatty acids by hydroxyl radicals that can
be formed via Fenton reaction during slice incubation with
H2 O2 .
In conclusion, this procedure describes the reaction con-
ditions necessary to perform lipid peroxidation studies by
means of a TBA test following incubations of cerebellar
slice cultures under controlled gas conditions. The method
enables the detection of pmol amounts of MDA equivalents
per mg protein, which may be a reasonable sensitivity for
lipid peroxidation studies in slices. Unlike what happens in
homogenates, synaptosomes or cultures, the cellular integrity
or synaptic organization still persists in cerebellar slices.
This preparation could contribute to document morpholog-
ical, metabolic, electrophysiological, pharmacological and
neurochemical events associated with lipid peroxidation and
its relations with human disease.
Acknowledgements
This study was supported by Grants S1-2001001176
from F.O.N.A.C.I.T and M-583-96-C-07 from C.D.C.H.T,
U.L.A.Venezuela. We thank Dr. Francoise Meyer for review-
ing the manuscript and Mr. Milano Rivas for his technical
assistance.
References
Bawari M, Babu GN, Ali MM, Misra UK. Effect of neonatal
monosodium glutamate on lipid peroxidation on adult brain. Neu-
roReport 1995;6:650–2.
Bligh EG, Dyer WJ. A rapid method of total lipid extraction and purifi-
cation. Can J Med Sci 1959;37:911–7.
Boehme DH, Kosecki R, Carson F, Stern F, Marks N. Lipoperoxidation
in human and rat brain tissue: developmental and regional studies.
Brain Res 1977;136:11–21.
Bralet J, Bouvier C, Schreiber L, Boquillon M. Effect of acidosis on lipid
peroxidation in brain slices. Brain Res 1991;539:175–7.
Chakraborty H, Sibendra NR, Chakrabarti S. Lipid peroxidation associ-
ated protein damage in rat brain crude synaptosomal fraction mediated
by iron and ascorbate. Neurochem Int 2001;39:311–7.
Cini M, Fariello RG, Bianchetti A, Moretti A. Studies on lipid peroxida-
tion in the rat brain. Neurochem Res 1994;3:283–8.
Ceconi C, Cargnoni A, Pasini E, Condorelli E, Currello S, Ferrari R.
Evaluation of phospholipid peroxidation as malondialdehyde dur-
ing myocardial ischemia and reperfusion injury. Am J Physiol
1991;260:H1057–61.
Dirks RC, Faiman MD. Free radical formation and lipid peroxidation in
rat and mouse cerebral cortex slices exposed to high oxygen pressure.
Brain Res 1982;248:355–60.
Draper HH, Squires EJ, Mahmoodi H, Wu J, Agarwal S, Hadley M. A
comparative evaluation of thiobarbituric acid methods for the deter-
mination of malondialdehyde in biological materials. J Free Rad Biol
Med 1993;15:353–63.
Esterbauer H, Cheeseman KH, Dianzani MU, Poli G, Slater TF. Sepa-
ration and characterization of the aldehydic products of lipid perox-
idation stimulated by ADP.Fe2+ in rat liver microsomes. Biochem J
1982;208:129–40.
Garcia Y, Ibarra C, Jaffé EH. NMDA and non-NMDA receptor-mediated
release of [3 H] GABA from granule cell dendrites of rat olfactory
bulb. J Neurochem 1995;64:662–9.
Garthwaite G, Williams GD, Garthwaite J. Glutamate Toxicity: An
Experimental and Theoretical Analysis. Eur J Neurosci 1992;4:
353–60.
Götz ME, Dirr A, Freyberger A, Burger R, Riederer P. The thiobar-
bituric acid assay reflects susceptibility to oxygen induced lipid
peroxidation in vitro rather than levels of lipid hydroperoxides
in vivo: a methodological approach. Neurochem Int 1993;22:255–
62.
Gutteridge JM, Halliwell B. The measurement and mechanism of
lipid peroxidation in biological systems. Trends Biochem Sci
1990;15:129–34.
Ichinose T, Miller MG, Shibamoto T. Gas chromatographic analysis
of free and bound malonaldehyde in rat liver homogenates. Lipids
1989;24:895–8.
Janero DR, Burghardt B. Analysis of cardiac membrane phospholipid per-
oxidation kinetics as malondialdehyde: nonspecificity of thiobarbituric
acid-reactivity. Lipids 1988;23:452–8.
Janero DR. Malondialdehyde and thiobarbituric acid-reactivity as diagnos-
tic indices of lipid peroxidation and peroxidative tissue injury. Free
Rad Biol Med 1990;9:515–40.
Kikugawa K, Kojima T, Yamaki S, Kosugi H. Interpretation of the thiobar-
bituric acid reactivity of rat liver and brain homogenates in the pres-
ence of ferric ion and ethylenediaminetetraacetic acid. Anal Biochem
1992;202:249–55.
Kovachich GB, Mishra OP. Lipid peroxidation in rat brain cortical
slices as measured by the thiobarbituric acid test. J Neurochem
1980;35:1449–52.
Kosugi H, Kato T, Kikugawa K. Formation of yellow, orange, and red
pigments in the reaction of alk-2-enals with 2-thiobarbituric acid. Anal
Biochem 1987;165:456–64.
Kosugi H, Kato T, Kikugawa K. Formation of red pigment by a two-step
2-thiobarbituric acid reaction of alka-2 4-dienals. Potential products
of lipid oxidation. Lipids 1988;23:1024–31.
Kücükkaya B, Haklar G, Yalcin AS. NMDA excitotoxicity free radical
generation in rat brain homogenates: application of a chemilumines-
cence assay. Neurochem Res 1996;21:1535–8.
Li XY, Chow CK. An improved method for the measurement of malon-
dialdehyde in biological samples. Lipids 1994;29:73–5.
Lowry OH, Rosenbrough NG, Farr AL, Randall RJ. Protein measure-
ment with the folin phenol reagent. J Biol Chem 1951;93:265–
75.
Ortega-Gutiérrrez S, Garcia JJ, Martinez-Balları́n E, Reiter RJ, Millán-
Plano S, Robinson M, Acuña-Castroviejo D. Melatonin improves de-
feroxamine antioxidant activity in protecting against lipid peroxidation
caused by hydrogen peroxide in rat brain homogenates. Neurosci Lett
2002;323:55–99.
Oubidar M, Boquillon M, Marie C, Bouvier C, Beley A, Bralet J. Effect
of intracellular iron loading on lipid peroxidation of brain slices. Free
Rad Biol Med 1996;21:763–9.
Pellmar TC, Neel KL, Lee KH. Free radicals mediate peroxidative damage
in guinea pig hippocampus in vitro. J Neurosci Res 1989;24:437–
44.
Pikul J, Leszczynski DE, Kummerow FA. Elimination of sample auto-
oxidation by butylated hydroxytoluene additions before thiobarbituric
acid assay for malondialdehyde in fat from chicken meat. J Agric
Food Chem 1983;31:1336–42.
Santamarı́a D, Espinoza-Gonzáles V, Rı́os C, Santamarı́a A. N␻-Nitro-l-
arginine a nitric oxide synthase inhibitor antagonizes quinolinic acid-
 Y.J. Garcia et al. / Journal of Neuroscience Methods 144 (2005) 127–135 135

induced neurotoxicity and oxidative stress in rat striatal slices. Neu-
rochem Res 1999;24:843–84.
Sewerynek E, Poeggeler B, Melchiorri D, Reiter RJ. H2 O2 -induced lipid
peroxidation in rat brain homogenates is greatly reduced by melatonin.
Neurosci Lett 1995;195:203–5.
Subbarao KV, Richardson JS. Iron-dependent peroxidation of rat brain: a
regional study. J Neurosci Res 1990;26:224–32.
Verbunt RJ, Egas JM, Van der Laarse A. Risk of overestimation of free
malondialdehyde in perfused rat hearts due to homogenization arti-
facts. Cardiovasc Res 1996;31:603–6.