Journal of Neurochemistry
Lippincott—Raven Publishers, Philadelphia
© 1997 International Society for Neurochemistry
High-Density Lipoprotein Aggregated by Oxidation
Induces Degeneration of Neuronal Cells
Silvia C. Kivatinitz, Maria A. Peisman, Alejandra del C. Alonso, Luis Bagatolli,
and Santiago Quiroga’
Centro de Investigaciones en QuImica BiolOgica de Córdoba (CIQUIBIC), Facultad de Ciencias QuI,nicas,
Universidad Nacional de Córdoba/CONJCET, Córdoba, Argentina
Abstract: We have previously reported that high-density
lipoprotein (HDL) exhibits antineuritogenic effects on
chicken cerebral cells in culture. In the present study, we
show the effects of HDL5, oxidized by UV irradiation or
heating, on chicken cerebral neurons in culture. Both
treatments produced several physical and chemical
changes in the HDLs, i.e., formation of lipid peroxides,
enlargement of HDL diameters, an increased exposure of
the tryptophan groups of the apolipoprotein A-I to a more
hydrophilic environment, formation of bityrosines, and
cross-linking of apolipoprotein A-I. When these treat-
ments were performed in the absence of EDTA, most of
the modifications described above were more intense
and HDLs formed a macroaggregate that displays a ro-
sette-like structure. The aggregated HDLs produced neu-
rodegeneration and death when added to both undiffer-
entiated and differentiated cerebral neurons in culture.
This process was accompanied by the disorganization of
the cellular microtubular cytoskeleton and hyperphos-
phorylation of the microtubule-associated protein tau.
Native HDL or HDLs treated in the presence of EDTA
inhibited the neuritogenesis of undifferentiated neurons
but did not show any significant effect on the differenti-
ated neurons in culture. The effects on the cellular cy-
toskeleton and morphology of aggregated HDLs recall
those of the fibrillar /1-amyloid peptide. The present re-
sults suggest that aggregated HDLs could participate in
neurodegeneration associated with oxidative stress in the
CNS. Key Words: Neurodegeneration—Aggregated
high-density lipoproteins—Oxidized lipoproteins—Alz-
heimer disease— Laurdan fluorescence—Hyperphos-
phorylated tau.
J. Neurochem. 69, 2102—2114 (1997).
We have previously reported that high-density lipo-
protein (HDL) exhibits antineuritogenic effects on
chicken undifferentiated neuronal cells in culture and
that it inhibits the biosynthesis of gangliotetraosyl gan-
gliosides (Kivatinitz et a!., 1995). There are many
reports on the activity of HDL as an antiatherogenic
agent because of its ability to stimulate removal of
cholesterol from the peripheral cells to the liver for
2102
excretion, through reverse cholesterol transport (Roth-
blat et al., 1992; Ross, 1993; Rubin and Smith, 1994).
It has also been demonstrated that oxidized HDLs have
a reduced cholesterol effluxing capacity (Miyazaki et
al., 1994; Morel, 1994; Bonnefont-Rousselot et al.,
1995), but their effects on neuronal cells have not been
analyzed. Yet, apolipoprotein A-I, the main protein-
aceous component of avian HDL, comprising >90%
of the total HDL protein (Shackelford and Lebherz,
1983), has been found in nervous tissues of several
species, e.g., birds (Mao et al., 1980), fishes (Harel
et al., 1989), and mammals (Möckel et al., 1994).
Although most apolipoproteins are synthesized in the
liver and small intestine, apolipoprotein A-I is synthe-
sized in several avian tissues, including the brain (Blue
et al., 1982; Byrnes et al., 1987; Ferrari et al., 1987).
In human CSF there is a high concentration of lipopro-
tein-containing apolipoprotein A-I (Borghini et al.,
1995), and HDL particles containing apolipoprotein
A-I and E have been reported (Roheim et al., 1979).
Also, an active mechanism of transport of HDL across
the blood—brain barrier has been suggested (Devries
et al., 1995). There is no evidence indicating synthesis
of apolipoprotein E or apolipoprotein A-I! in birds
(Jackson et a!., 1976; Hermier and Dillon, 1992); but
as the tissue distribution between avian apolipoprotein
A-I and mammalian apolipoprotein E is very similar,
Received January 20, 1997; revised manuscript received May 6,
1997; accepted June 16, 1997.
Address correspondence and reprint requests to Dr. S. C. Kivati-
nitz at Departamento de QuImica Biologica-CIQUIBIC, Facultad de
Ciencias Qulmicas, Universidad Nacional de Córdoba, cc. 61, AP
4 5000 Córdoba, Argentina.
Abbreviations used: DMSO, dimethyl sulfoxide; HDL, high-den-
sity lipoprotein; HDLm+ and HDLm—, HDL modified by heating
in the presence and the absence of EDTA, respectively; HDLn and
HDLn—, HDL dialyzed against Tris-HCI with and without EDTA,
respectively; HDLuv+ and HDLuv—, HDL modified by irradiation
in the presence and the absence of EDTA, respectively; LDL, low-
density lipoprotein; mAb, monoclonal antibody; PAGE, polyacryl-
amide gel electrophoresis; P1, propidium iodide; SDS, sodium dode-
cyl sulfate.
 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL
it has been proposed that avian apolipoprotein A-I may
function analogously to mammalian apolipoprotein E,
participating in the process of local cholesterol trans-
port or during nerve degeneration and regeneration
(Dawson et al., 1986; Harel et al., 1989). Another
putative role for avian apolipoprotein A-I, analogous
to that of apolipoprotein E, is the modulation of neurite
outgrowth (Nathan et al., 1994; Kivatinitz et al.,
1995).
HDL has been oxidized in vitro under different con-
ditions, i.e., ultraviolet irradiation (Salmon et al.,
1992), incubation with Fe2~,Mn2~,or Cu2~(Marcel
et al., 1989; Morel, 1994; Shoukry et a!., 1994), or
prolonged storage at 4°Cwithout EDTA (Marcel et
a!., 1987). It is interesting that some common modifi-
cations of HDL were observed under all these condi-
tions, such as an increase in peroxidized lipids and the
appearance of apolipoprotein A-I oligomers.
The presenceof oxidized HDLs in vivo has not been
clearly established, but the occurrence and actions of
oxidized low-density lipoprotein (LDL) in different
pathological conditions have been thoroughly studied.
Atherosclerotic lesions from Watanabe Heritable Hy-
perlipidemic rabbits and LDL isolated from these le-
sions contain material reactive with antibodies raised
against oxidized LDLs (Boyd et al., 1989; Rosenfeld et
al., 1990), and hypochlorite-modified apolipoprotein
B in human atherosclerotic lesions has been reported
(Hazell et al., 1996). Also, the following facts suggest
the occurrence of oxidized HDLs in vivo: (1) HDL is
modified in an oxidative manner by human polymor-
phonuclear leukocytes, probably due to the actions of
superoxide anions (Cogny et al., 1994); (2) there are
deposits of fragments of apolipoprotein A-I, corre-
sponding to amino acids 1—69, in human senile aortic
amyloid ( Westermark et al., 1995); and (3) it has been
reported very recently that in brains of patients with
Alzheimer’s disease at least some of the amyloid
plaques contain apolipoprotein A-I (Harr et al., 1996).
Besides atherosclerosis, there are many oxidative
stress-mediated diseases, i.e., Alzheimer’s disease,
multiple sclerosis, Parkinson’s disease, and ischemic—
reperfusion injury. The disruption of the blood—brain
barrier observed in some of these pathological condi-
tions, like ischemia (Farooqui et al., 1994) or Alzhei-
mer’s disease (Mon et al., 1991), could produce
higher than normal HDL levels in brain, which could
be susceptible to oxidation by free radicals produced
during ischemia and subsequent reperfusion (Farooqui
et al., 1994), and in Alzheimer’s disease frontal cortex
(Subbarao et al., 1990). Another potential source of
oxidized HDLs could be HDL of the interstitial fluid
bathing the epidermis of UV-exposed skin (Salmon et
al., 1992).
These facts prompted us to study the effects of HDL
and of oxidized HDLs on neurons in culture and to
correlate these effects with the chemical and structural
changes in the HDL molecule. We conclude that HDL,
2103
UV-irradiated or heated in the absence of antioxidants,
forms a particular aggregate and that these modified
HDLs are able to cause neurodegeneration. The disor-
ganization of the microtubules, the accumulation of
hyperphosphorylated forms of tau, and the aggregation
of the cells and fragmentation of their neurites are
some of the characteristic features of the neurodegener-
ation produced by HDL.
MATERIALS AND METHODS
Materials
Trizma base, KBr, NaC1, EDTA, dimethyl sulfoxide
(DMSO), and all the reagents for electrophoresis and immu-
noblotting were from Sigma Chemical (St. Louis, MO,
U.S.A.). Monoclonal antibody (mAb) Tau-l was raised
against bovine tall (Binder et al., 1985). Laurdan was from
Molecular Probes (Eugene, OR, U.S.A.).
HDL preparation
HDL was prepared essentially as described (Kivatinitz et
al., 1995). In brief, whole blood was obtained from 10-day-
old chickens and allowed to clot at 37°C. After centrifuga-
tion, the serum was adjusted to 0.4946 g/ml KBr (density,
1.30 g/ml) by addition of the solid salt. Fractionation was
performed by using a single near-vertical spin-density gradi-
ent ultracentrifugation method (modified from Chung et a!.,
1986). Samples were centrifuged at 305,000 g for 330 mm
by using a Beckman TV 865 rotor. The fraction collected
as HDL had a density of 1.12 g/ml and was bright yellow-
orange, typical of chicken serum HDL (Kruski and Scanu,
1975). HDL was dialyzed against Tris-HC1 with EDTA (5
mM) (HDLn) or without EDTA (HDLn—); the modified
HDLs were prepared by heating the preparations at 85°Cfor
10 mm or irradiating at 290 nm X 385 nm in quartz
cuvettes (1-cm light path containing 1 ml of preparation)
for 30 mm (as described by Salmon et al., 1992), in the
absence (HDLm— and HDLuv—, respectively) or the pres-
ence (HDLm+ and HDLuv+, respectively) of EDTA (5
mM). EDTA was used as an antioxidant agent (Richter,
1987). Although the above procedures were used routinely
to produce HDL aggregation, this phenomenon occurs at any
temperature tested (20 or 4°C) but at a slower rate. The
aggregated HDL is a homogeneous dispersion that remains
in solution unless it is centrifuged at 7,000 g for 30 mm.
Electron microscopy
Aliquots of native and modified HDLs were negatively
stained with 1% sodium phosphotungstate. Samples were
applied to Formvar-carbonated grids and examined with a
Siemens Elmiskop 101 electron microscope and photo-
graphed. Particle size was determined from digitized images
by using the morphometric menu of the Metamorph 2.0 soft-
ware.
Cell cultures
The cell cultures were performed essentially as described
(Pettman and Sensenbrenner, 1979). Cerebral hemispheres
of 7-day-old chicken embryos were dissected in ice-cold
Ca2~-and Mg2~-free Tyrode’s buffer solution containing
25 mM glucose. After removal of the meninges, the cells
were dissociated using a low concentration of trypsin fol-
lowed by trituration. The dissociated cells were plated onto
.1. Neurochenr, Vol. 69. No. 5. 1997
2104 S. C. KIVATINITZ ET AL.
coverslips previously coated wmth po!ylysine (20 ,ug/m! in
2, and
0.1 M borate
incubated at buffer)
37°C in atDulbecco’s 1 X io~
a density ofmodified cells/cm
Eagle medium
containing 10% fetal calf serum in a 5% CO
2 atmosphere.
When mndicated, 50 ~tg/ml HDLn, HDLm+, HDLuv+,
HDLm—, or HDLuv— was added to the culture medium.
Neuronal counts
The assessment of cell viability was performed as de-
scribed (Busciglio et al., 1995). In brief, living cells were
incubated with 10 ~tg/mlpropidium iodide (P1) for 30 mm,
washed with phosphate-buffered saline, and fixed. Viable
neurons exclude P1.
Immunofluorescence
Cells were fixed after mild detergent extraction under mi-
crotubule-stabilizing conditions and processed for immuno-
fluorescence as described (Cáceres et a!., 1992). The pri-
mary antibody used was an mAb against tyrosinated cs-tu-
bulin (clone-TUB-1A2, mouse IgG, Sigma) diluted 1:1,000.
Incubations with primary antibody were performed for 2 h
at room temperature, whereas incubations with the secondary
antibody (fluorescein isothiocyanate-labeled goat anti-rabbit
IgG, Boehringer-Mannheim) were performed for 1 h at 37°C.
The cells were observed with an inverted microscope (Zeiss
Axiovert 35M) equipped with epifluorescence optics and
photographed using a bOX objective (Zeiss) and Tn X-Pan
(400 ASA) film (Eastman Kodak).
Measurement of lipid peroxidation indices
The lipid peroxidation indices were determined as de-
scribed by Klein (1970). The absorbency spectra measure-
ments were made in a Gilford Response II UV-VIS spectro-
photometer using quartz cuvettes of 0.25 cm.
Fluorescence analysis of HDLs
Steady-state excitation and emission spectra were mea-
sured on an SLM 4800C fluorometer (SLM/Aminco, Ur-
bana, IL, U.S.A.) equipped with a xenon-arc lamp using 8
nm of bandwidth. The laurdan excitation spectra were cor-
rected with a quantum counter (rhodamine B in ethylene
glycol 3 g/ml; Lakowicz, 1983) and the emission spectra
were corrected for instrument response. We used a 3 x 3-
mm quartz cuvette to minimize inner filter effects (Lako-
wicz, 1983) and made blank subtraction in all samples. The
actual temperature measured in the cuvette was 37°Cin all
fluorescence experiments and was controlled through a circu-
lating bath.
One microliter of a 0.77 mM solution of laurdan in DMSO
was added to a l.5-ml solution of HDL in 20 mM Tris-
HC!—50 mM NaC1, pH 7.4 (the HDL/laurdan ratio was
300:1; laurdan final concentration was 0.51iM). The solu-
tions were incubated in the dark at 37°Cfor 1 h and analyzed
immediately after preparation. The laurdan emission general-
ized polarization (GPem) values were calculated as follows:
GPe0, = ‘390 — ‘360
‘390 + ‘360
where 1390 and 1360 are the intensities at the emission wave-
length of 440 nm, obtained by using fixed excitation wave-
lengths of 390 and 360 nm. In the same manner, the excita-
tion GP values (GP~~) were calculated using:
GPe~,= ‘440 1490
1445 + 1495
J. Neurochem., Vol. 69, No. 5, 1997
where 1443) and 1490 are the intensities at an excitation wave-
length of 360 nm obtained by using a fixed emission wave-
length of 440 or490 nm (Parasassi eta!., 1990, 1991, 1993).
The properties of the GP parameter have been discussed
extensively (Parasassi et al., 1990, 1991; Parasassi and Grat-
ton, 1995). Some of the advantages of laurdan are (1) in
contrast to common radiometric measurements made by us-
ing probes that display spectra! sensitivity to the properties
of the environment, GP measurements do not require calibra-
tion; (2) characteristic values of GP parameters are related
to the extent of water motion included in a lipid matrix
around the.probes; and (3) the measurement is fast and easy.
To collect the HDL tryptophan intrinsic fluorescence
emission spectra, we prepared solutions of pure HDL in
buffer as described above and the samples were excited at
295 nm. Bityrosines, products of oxidized proteins, were
measured as fluorescence intensity at excitation and emission
wavelengths of 327 and 400 nm, respectively (Miller et al.,
1996).
Electrophoresis and immunoblotting
Cells were collected, lysed with 10% sodium dodecyl sul-
fate (SDS), and sonicated for 10 s at intensity 7 in a Branson
sonicator. Samples were prepared as previously described
(Grundke-Iqbal et al., 1984). To detect abnormally phos-
phorylated tau, the blots were pretreated with alkaline phos-
phatase (86 j.tg/ml in 0.1 MTris, pH 8.0/1 mMphenylmeth-
ylsulfonyl fluoride) for 15 h before immunostaining with
mAb Tau-1 as previously described (Alonso et al., 1994).
RESULTS
Chemical and physical modifications of HDL
induced by heat or irradiation
Purified chicken HDL preparations dialyzed against
buffer containing EDTA (HDLn) or against buffer
without EDTA (HDLn—) were subjected to heating at
85°C for 10 mm or UV-irradiated for 30 mm. The
respective heated and irradiated preparations were
termed HDLm+ and HDLuv+ (when heated or irradi-
ated with EDTA) and HDLm— and HDLuv— (when
heated or irradiated without EDTA). To ascertain the
peroxidation status of the lipid moiety and the struc-
tural changes in the apolipoprotein A-I of these HDL
fractions, we performed various spectroscopic anal-
yses.
We determined the UV spectrum of the lipids ex-
tracted from HDLn, HDLn—, HDLm+, HDLuv+,
HDLm—, and HDLuv— (Klein, 1970). The level of
lipid peroxidation was different between the HDL
preparations, being more noticeable as an increment
in the content of conjugated diene and triene chromo-
phores in the HDLm— and HDLuv— produced by the
oxidation of fatty acid side chains than the changes
observed in the other HDL preparations (Fig. 1A
and B).
To further study the modifications in the lipid matrix
of HDL, the amphiphilic fluorescence probe laurdan
was used. The reasons for using this particular probe
were as follows: (1) Fluorescence emission and excita-
tion spectra of laurdan are sensitive to the polarity of
 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL 2105

FIG. 1. Chemical characterization of the differ-

ent HDL preparations. Open columns, HDLs
treated in the presence of 5 mM EDTA; filled
columns, HDLs treated in the absence of EDTA.
HDLn, native HDL; HDLm, HDL heated for 10
mm at 85°C; HDLuv, HDL irradiated for 30 mm
at 290 nm m 385 nm at 20°C.Determinations
of peroxidation indexes (A and B) were per-
formed in the lipid extracts of the different HDL
preparations. Fluorescence analysis was per-
formed using the whole HDL particles (C—H).
The statistical significance of the differences be-
tween HDLm+ (open columns) and HDLm—
(filled columns) was p < 0.001 in A, B, E, and
F; p < 0.01 in C and 0; and not significantly
different (NS) in H. The statistical significance of
the differences between HDLuv+ (open col-
umns) and HDLuv— (filled columns) was p
<0.001 in A, E, and F; p < 0.01 in C, B, and 0;
and NS in H. There were no statistical differences
between HDLn+ (open columns) and HDLn—
(filled columns) in any experiment. The statistical
significance of the differences between HDLn
and HDLm+ wasp < 0.001 in D, E, F, G, and
H; and p < 0.05 in A and C. The statistical sig-
nificance of the differences between HDLn and
HDLuv+ was p < 0.001 in D, E, F, G, and H,
and not significantly different in the other panels.
The statistical significance of the differences be-
tween HDLn and HDLm— wasp < 0.001 for all
the conditions tested. The statistical significance
of the differences between HDLn and HDLuv—
was p < 0.01 in B and p < 0.001 in all other
panels.

the environment, the quantum yield in the membrane
being high but virtually zero in aqueous environment
(Parasassi et al., 1990); (2) the probe affinity for pro-
teins and the fluorescence intensity have been shown
to be very low (Parasassi et al., 1992, 1993); and (3)
the presence of dipolar relaxation due to the dynamics
of water molecules in the phospholipid liquid crystal-
line phase modify the spectral position of laurdan com-
pared with the gel phase (Parasassi eta!., 1991). The
results of our experiments using laurdan showed a re-
markable coincidence of values obtained with all four
procedures. We observed a slight blue shift in the laur-
dan excitation spectra of HDLm—, HDLuv—,
HDLm+, and HDLuv+ compared with HDLn (Fig.
1C). We observed an emission blue shift of 8 nm for
HDLm— and HDLuv— and of 4 nm for HDLm+ and
HDLuv+ compared with HDLn (Fig. 1E). We also
noticed an increase in the laurdan GPex and GPe,,, val-
ues of the modified HDLs, which was more noticeable
in the fractions prepared in the absence of EDTA (Fig.
1D and F).
It is known that ionizing radiation produces peroxida-
tion in HDL (Bonnefont-Rousselot et al., 1995) and
modifies laurdan fluorescence parameters when it is in-

J. Neuroche,n., Vol. 69, No. 5, 1997

2106 S. C. KJVATJNITZ ET AL.
serted into unsaturated phospholipid vesicles (Parasassi
et al., 1994). Taken together, these facts and our results
indicate that the laurdan molecules sense differences
in the peroxidation status of the HDL preparations;
i.e., HDLm— = HDLuv— > HDLm+ = HDLuv+
> HDLn = HDLn—, which correlates with the conju-
gated diene values (see Fig. IA, C, D, and F).
To evaluate the effects of heating or UV irradiation
in HDL apolipoproteins, we recorded the intrinsic flu-
orescence emission spectra of tryptophan residues from
HDL. Tryptophans emit an intrinsic UV fluorescence,
which is sensitive to solvent polarity in addition to
having a high quantum yield. These properties make
tryptophans a useful built-in probe with which to study
protein structure (Lakowicz, 1983). The intrinsic flu-
orescence emission spectra of tryptophan residues
showed a red shift compared with HDLn, of ~—~8nm
for HDLm— and HDLuv— and of ~-~5nm for HDLm+
and HDLuv+ (Fig. IG). A red shift in the protein
fluorescence emission spectra indicates a major expo-
sure of the tryptophan residues to the aqueous medium
(Lakowicz, 1983). This finding indicates that the
protein conformational changes exposing the trypto-
phan residues are in the following order: HDLm—
= HDLuv- > HDLm+ = HDLuv+ > HDLn
= HDLn-.
Although oxidative cross-linking of apolipoprotein
A-I of HDL has been observed by us (Kivatinitz et
al., 1995) and others (Marcel et al., 1989; Salmon et
al., 1992; Morel, 1994; Shoukry et al., 1994) em-
ploying different conditions or due to prolonged stor-
age at 4°Cwithout EDTA (Marcel et a!., 1987), the
nature of the mechanisms leading to protein cross-
linking has not been identified. Recently, it has been
shown that oxidative cross-linking of apolipoprotein B
of LDL also occurs during LDL oxidation (Miller et
al., 1996) and it has been suggested that intermolecular
bityrosines result in apolipoprotein B cross-linking
(Miller et al., 1996). As seen from the results shown
in Fig. 1H, the levels of bityrosines in all four modified
HDL preparations were increased to the same extent
compared with HDLn. It is interesting that the amounts
of apolipoprotein A-I dimers detected in SDS—poly-
acrylamide gel electrophoresis (SDS-PAGE) immu-
noblots by using an antibody against apolipoprotein
A-I were also the same for the four modified HDLs
(~—12%of the total immunostained protein; results not
shown). In contrast, no dimers were detectable in
HDLn (Kivatinitz et al., 1995).
As HDLm— and HDLuv— were turbid, we investi-
gated the structure of the aggregate by electron micros-
copy. HDL stored in the absence of EDTA became
turbid at any temperature but at slower rate (see Mate-
rials and Methods). HDLn and HDLn— (HDL that
was dialyzed against Tris buffer without EDTA) had
the usual appearance of HDL fractions (Fig. 2A and
B), whereas heating produced larger particles in the
HDLm preparations (Fig. 2C and D). The average
.1. Neurochem., Vol. 69, No. 5, 1997
radii of the four different HDL preparations are as
follows: HDLn, 5 ± 1 nm; HDLn—, 5.5 ± 1.5 nm;
HDLm+, 9 ± 3 nm; and HDLm—, 9 ± 3 nm (p
<0.0001 for HDLm+ versus HDLn and for HDLm—
versus HDLn—; no significant difference for HDLn
versus HDLn— and for HDLm+ versus HDLm—).
From these results it is clear that in the heated prepara-
tions, the average particle diameter is 1.70 ± 0.05
times greater than in the nonheated preparations; but
what is most impressive is that these larger HDLs form
rosette-lik’e macroaggregates if the heating procedure
is performed in the absence of EDTA (Fig. 2, compare
C with D). The appearances of HDLuv + and HDLuv —
were identical to those of their heated counterparts
(data not shown).
Effects of HDL preparations on the morphology
of immature and differentiated chicken cerebral
cells in culture
We have previously shown that HDLn inhibits neu-
ritogenesis of undifferentiated chicken cerebral cells
in culture (Kivatinitz et al., 1995). To determine the
effects of the different HDL preparations,cerebral cells
were cultured as described in Materials and Methods
and the HDL preparations added to the culture medium
immediately after plating. After 2 days in culture, the
control cells showed the usual polygonal shape of neu-
ronal cells, with abundant and long processes and
clumping of several cells (Fig. 3A). In contrast, cells
cultured in HDLn or HDLm+ preparations displayed
basically a round shape and only a few cells exhibited
scant processes (Fig. 3C and E). After 4 days in cul-
ture, the morphology of control neurons was similar
to that after 2 days in culture but the processes were
longer and there were more cells forming the clumps
(Fig. 3B). The morphology of the neural cells cultured
in the presence of HDLn was similar to that of control
cells, i.e., the neurons were more clumped and rounded
but exhibited very long processes (Fig. 3D). Con-
versely, the cells cultured in the presence of HDLm—
showed a very similar morphology to that after 2 days
in culture; the only noticeable change was clumping
of a few cells and elongation of the few short neurites,
and a significant number of the cells cultured in this
condition appeared to be nonviable (Fig. 3F). To ver-
ify this last observation, we used the P1 method. P1 is
selectively taken up into the nuclei of nonviable cells,
due to loss of membrane integrity, and becomes
brightly fluorescent after binding to nuclei acids. The
results indicated that in the control cultures (i.e., Fig.
3B) 92 ± 3% of the cells were viable, whereas in
HDLm— treated cultures (i.e., Fig. 3F) only 45 ±2%
of the cells were negative to the staining (significance
of the difference, p < 0.001, by ANOVA). Thus, al-
though the effects of HDLn on the differentiation pro-
cess were relatively reversible, those of HDLm— were
persistent. The observations from experiments per-
formed with HDLm+ were identical to those for HDLn
 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL 2107

FIG. 2. Electron microscopy of the different HDLs. A: Native HDL without EDTA (HDLn~-).B: Native HDL plus EDTA (HDLn). C: HDL
heated for 10 mm at 85°Cin the absence of EDTA (HDLm—). D: HDL heated for 10 mm at 85°Cin the presence of EDTA (HDLm+).

after 2 and 4 days in culture (data not shown). These
results indicated that, in contrast to the actions of
HDLn and of HDLm+, the effects of aggregated
HDLm— were irreversible and many of the neurons
cultured in its presence died.
We also studied the effects of the addition of the
HDL preparations to neuronal cells after 2 days of
differentiation in culture, so at this time neurons were
already differentiated, showing abundant and long neu-
rites. The different HDL preparations were added to
the culture medium and the incubation continued for
another 48 h. Results are shown in Fig. 4. Control cells
and cells cultured in the presence of HDLm+ or HDLn
looked alike; i.e., they displayed several long and
branched processes (Fig. 4A, B, and D). In contrast,
cells cultured in the presence of HDLm— were
rounded, some of them were edematized (see arrow,
Fig. 4C), and their processes were fragmented, form-
ing numerous vesicles that suggested rosary beads (see
arrowhead, Fig. 4C).
The neurotoxicity of aggregated HDL on the neu-
rons in culture was assessed by quantification of neu-
ron viability of control cultures and of neurons cultured
in the presence of HDLm—, using the PT fluorescence
method. The results indicate that HDLm— caused pro-
gressive neuronal degeneration in the cultures of
chicken brain cortical neurons, resulting in ~-~50%loss
of neuron viability after 5 days (Fig. 5).
Throughout all the above experiments, the effects
of HDLuv+ and HDLuv— on neurona! cells were in-
distinguishable from those produced by HDLm+ and
HDLm— (data not shown).
Effects of HDL preparations on the dynamic
cytoskeleton of chicken cerebral cells in culture
The effects of aggregated HDL on the cytoskeleton
of chicken cerebral cells were investigated by adding
HDLm— to cerebral cells differentiated in culture for 2
days. At 48 h after the addition, the cells were detergent
extracted and fixed in conditions that preserved the
cytoskeleton as described (Cáceres et al., 1992; Black
et al., 1994) and immunostained with a mAb against
tyrosinated tubulin. Neurons showing several small
processes (Fig. 6A and B) and others showing some

J. Neurochem., Vol. 69, No. 5, 1997

2108 S. C. KIVATINJTZ ET AL.

FIG. 3. Differential effects of native HDL (HDLn) and aggregated HDL (HDLm—) on differentiating cerebral cells in culture. Micrographs
of chicken cerebral cells after 48 h in culture (A, C, and E) and after 96 h in culture (B, D, and F). A and B: Control cells. C and D:
Cells cultured in the presence of native HDL (HDLn). E and F: Cells cultured in the presence of HDL heated for 10 mm at 85°Cin the
absence of EDTA. The results shown are representative of six different experiments performed using different fresh preparations of
HDL in which the morphology of the neurons was essentially identical. Calibration bar: 20 tim.

small processes and an elongated one, most probably
an axon (Fig. 6C and D), were analyzed. The microtu-
bular cytoskeleton of cells treated with HDLm— was
disorganized at the neuronal body level (large arrow-
heads, Fig. 6B and D), the staining was more unevenly
distributed than in control cells, and the neurites’ mi-
crotubule bundles became thinner, fragmented, and
twisted (small arrowheads, Fig. 6B and D).
As the effects of HDLm— on the morphology and
on the cytoskeleton of neuronal cells are similar to
those produced by the /3-amyloid peptide, and as it is
known that /3-amyloid peptide induces the appearance
of the hyperphosphorylated species of tau protein
(Busciglio et al., 1995), we investigated the state of
phosphorylation of tau protein in control cells and in
cells cultured in the presence of HDLm+ or HDLm—
for 48 h. Cells were cultured as described in Materials
and Methods, their total proteins resolved by SDS-
PAGE, and the immunoreactivity of western blots to
Tau-I antibody analyzed before and after treatment

J. Neurochem., Vol. 69, No. 5, 1997

 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL 2109

FIG. 4. Differential effects of the HDL fractions on differentiated chicken cerebral cells in culture. Micrographs of chicken cerebral
cells cultured for 96 h; after 2 days in culture the different HDLs were added to the culture medium. A: Control cells. B: Cells cultured
in the presence of native HDL (HDLn). C: Cells cultured in the presence of HDL heated for 10 mm at 85°C(HDLm—). D: Cells cultured
in the presence of HDL heated for 10 mm at 85°Cin the presence of EDTA (HDLm+). The arrow in C points to an edematized cell
with vacuoles; the arrowheads point to fragmented neurites. The results shown are representative of six different experiments performed
using different fresh preparations of HDL in which the morphology of the neurons was essentially identical. Several hundred neurons
were observed in each experiment and no significant differences in either the number of processes per cell or the average length (p
> 0.1) were found between the control neurons and the neurons cultured in the presence of either HDLn or HDLm+, as measured
on digitized pictures by using the integrated morphometric menu of Metamorph 2.5 software. Calibration bar: 15 /.tm.

with alkaline phosphatase. Tau-1 is an mAb that recog-
nizes the unphosphorylated form of Ser’99 to Ser202 of
tau. These residues are not largely phosphorylated un-
der normal conditions. As shown in Fig. 7, treatment
with alkaline phosphatase revealed a faint additional
band of higher M. (as detected by the Tau- 1 antibody)
in control cells and in cells cultured in the presence of
HDLm+. In contrast, in the cells cultured with
HDLm— for the last 48 h there was a high accumula-
tion of hyperphosphorylated species of tau as shown
by the high immunoreactivity to Tau-l and the shift
to a higher Mr of the major band detected in the strip
treated with alkaline phosphatase (Fig. 7B, lane 3).
To quantify the extent of this accumulation, similar
experiments were performed in which the regions of
the blots showing immunoreactivity to Tau-1 were ex-
cised and the amount of bound ‘251-protein A measured
in a y counter. The results are shown in Table 1. Tn
the control cells and in the cells treated with HDLm+,
only .—~30%of the total tau protein was found to be
phosphorylated at the Ser199 to Ser202 region, whereas
in the HDLm— treated cells this proportion was in-
verted and 74% of the tau protein was phosphorylated
at these residues.
DISCUSSION
The aim of the present study was to ascertain the
chemical and physical changes that HDL fractions un-
dergo when they are subjected to different levels of
oxidative stress and to evaluate the effects of oxidized
HDLs on neuronal cells in culture. We found that HDL
heated or UV-irradiated in the absence of EDTA forms
a supramolecular aggregate and under these conditions
is neurodegenerative. The mechanism underlying these
effects occurs with disorganization of cytoskeleton and
Tau hyperphosphorylation. Finally, as it has been pro-
posed that /3-amyloid is neurotoxic only when it is
aggregated, we compared our results of the neurode-
generative effects of aggregated HDL with those re-
ported for /3-amyboid peptide.
Chemical and physical characterization of
aggregated HDL
Our experiments show that heating or UV-irradiat-
ing HDL produced the formation of larger HDL parti-
cles and that these larger particles can form a macroag-
gregate when the HDL was heated or UV-irradiated in
the absence of EDTA. These results are probably re-

.1. Neurochem., Vol. 69. No. 5, 1997

2110 S. C. KIVATINITZ ET AL.

lated to the changes in the oxidation status of HDL

FIG. 5. Neurotoxicity in cerebral cells treated with aggregated
HDL. Chicken cerebral cells were cultured for 48 h and for the
further indicated times in the absence (hatched columns) or in
the presence of HDLm— (filled columns). Neuronal viability was
evaluated by the P1 fluorescence method. Data are mean ± SD
values of four independent experiments; a total of 400 cells were
scored for each condition at the indicated times. Significance of
the differences, as calculated by ANOVA: ns, not significant; *p
< 0.02; °°p < 0.002; ~ < 0.001.
lipids because the appearance of larger lipid vesicles
(Gast et al., 1982) and of larger HDL particles has
been described under oxidative conditions both in vitro
(Shoukry et a!., 1994) and in vivo (Cogny et al.,
1996). Our data indicate that for the formation of these
macroaggregates it is necessary to have both a confor-
mational change of the apolipoprotein A-I (increase
in tryptophan maximum emission wavelength) and an
altered lipid environment, as seen both by the increase
in the lipid peroxidation index and by changes in laur-
dan fluorescence. In fact, the HDLm— and HDLuv—
preparations have more peroxidized lipids and the tryp-
tophan residues are more exposed to hydrophilic envi-
ronment than HDLm+ or HDLuv+. Our findings are
in line with those of Shoukry et al. (1994) who pro-
posed that the shift in maximum peak wavelength of
tryptophan in HDL after oxidation appears highly ap-
plicable to the investigation of HDL oxidation. Peroxi-
dative phenomena induce physicochemical modifica-
tions in the lipid structure that have been already de-
scribed for HDL (Bonnefont-Rousselot et a!., 1995),
model membrane systems (Parasassi et a!., 1994), and
cell membranes (Richter, 1987). These modifications
are mainly identified as a progressive rigidification of
the lipid moiety. It has been suggested that this is partly

FIG. 6. Microtubular cytoskeleton disorganization in cerebral cells cultured in the presence of aggregated HDL. Micrographs showing
immunolabeling of chicken cerebral cells cultured for 96 h, fixed after mild detergent extraction under microtubule-stabilizing conditions,
and stained using anti-tyrosinated tubulin. A and C: Control cells. B and D: Cells cultured for the last 48 h in the presence of HDL
heated for S mm at 85°Cin the absence of EDTA. A and B: Neurons in stage 2 of development. C and D: Neurons in stage 3 of
development. Large arrowheads in B and D point to the disorganized microtubular cytoskeleton at the cell body level; small arrowheads
point to fragmented neurites. Calibration bar: 5 tim.

J. Neurochem., Vol. 69, No. 5, 1997

 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL
FIG. 7. Accumulation of hyperphosphorylated tau in cerebral
cells cultured in the presence of aggregated HDL. A: Immu-
noblots of total protein of control cells (lane 1) and cells cultured
in the presence of HDLm+ (lane 2) or HDLm— (lane 3) reacted
with Tau-1 without any treatment. B: Immunoblots of total pro-
tein of control cells (lane 1) and cells cultured in the presence
of HDLm+ (lane 2) or HDLm— (lane 3) reacted with Tau-1 anti-
body after treatment with alkaline phosphatase (see Materials
and Methods). Arrow points to the accumulation of tau-immuno-
reactive species at a higher Mr in the cells cultured with HDLm—
after treatment with alkaline phosphatase.
due to double bond breakages and, after a delay, to
formation of linkages between phospholipid and pro-
teins (Richter, 1987). The extent of apolipoprotein
A-I cross-linking and bityrosine formation does not
correlate with macromolecular aggregate formation be-
cause there are no significant differences between their
levels in HDLs modified in the presence or absence of
EDTA, so we speculate that the formation of apolipo-
protein A-I oligomers occurs within the larger HDL
particles of all four modified HDLs.
We were not able to clarify whether lipid peroxida-
tion per se could be the only cause of the macroaggre-
gate formation, because we always observed that an
increase in lipid peroxidation status is concomitant
with the conformational changes in the protein moiety.
In fact, experiments performed by other investigators
in which HDL preparations were oxidized under differ-
ent conditions have shown protein modifications (i.e.,
inter- and intramolecular cross-linking of apolipopro-
tein A and changes in tryptophan fluorescence) (Mar-
cel et a!., 1989; Morel, 1994; Shoukry et al., 1994).
However, the appearance of protein modifications ap-
pears to require harder oxidative conditions than the
appearance of lipid modifications (Morel, 1994). Fur-
ther studies are required to investigate a causal rela-
tionship between modifications of the protein moiety
and the lipid peroxidation status, and the formation of
aggregated HDL species.
Neurodegenerative effects of aggregated HDL on
neurons in culture
The effects of native and aggregated HDLs on neu-
ronal cells in culture are clearly different. Native HDL
produces a relatively reversible inhibition of neurite
formation and outgrowth of undifferentiated chicken
brain neurons but does not have any significant effect
on differentiated neurons of the same origin. These
results could be explained by the presence of HDL
binding sites in the undifferentiated neuronal cells,
which were not detected in differentiated neurons by
2111
using ligand blotting techniques (C. Alfonso, S. C. Ki-
vatinitz, and S. Quiroga, unpublished observations),
and because native HDL and oxidized HDL exert their
actions through different mechanisms, as oxidized
HDL is no longer recognized by the HDL receptors
(Brinton et a!., 1986; Rothblat et a!., 1992; Laville et
al., 1994). A putative candidate receptor through
which HDLs could exert their effects on differentiated
neurons is the LDL receptor homologue found recently
in the avian brain (Novak et a!., 1996). Aggregated
HDL, in contrast, produces an irreversible inhibition
of neurite formation and outgrowth of undifferentiated
chicken brain neurons, and neurodegeneration of dif-
ferentiated neuronal cells in culture. It is noteworthy
that the HDL preparations heated or UV-irradiated in
the presence of EDTA did not produce any significant
effect on differentiated neurons in culture, although
some of the chemical modifications produced by heat-
ing or UV irradiation are similar in both preparations
(Fig. 1). These results suggest that aggregation itself
is the condition required for HDL to exert its neurode-
generative effects on cerebral neurons in culture. Be-
cause changes in both the lipid peroxidation status and
the apolipoprotein are concomitant with the aggrega-
tion process, we cannot assert that one of these features
alone would be able to cause the observed effects. The
mechanisms by which the HDL aggregates interact
with the neuronal cells remain to be clarified. Some
characteristics of the cells cultured in the presence of
aggregated HDL (neurite fragmentation and edemati-
zation) closely resemble those previously shown for
human brain cells cultured in the presence of aggre-
3-amy-
gated /3-amyloid (Busciglio et a!., 1992). The /
bid peptide is neurotoxic to neuronal cells both in
culture and in vivo (Yankner et al., 1990). This neuro-
toxicity is dependent on the formation of a fibrillar
aggregate, which requires specific conditions of incu-
bation of the soluble form of the peptide such as time,
temperature, and pH (Pike et al., 1991; Busciglio et
al., 1992; Mattson et al., 1993; Lorenzo and Yankner,
1994). Moreover, HDL is just the second endogenous
compound reported (after /3-amyloid peptide) to in-
duce degeneration of neurons in culture depending on
its aggregation state. The effects of aggregated HDL
on the microtubular cytoskeleton and the state of phos-
TABLE 1. Effect of aggregated HDL on the
phosphorylation status of tau from cultured cells
Unphosphorylated Phosphorylated
HDLn 70 ±3 30± 2
HDLm+ 74 ±3 26±4
HDLm— 26 ±4 74± 5
Data are expressed as percentages of total tau protein and are
mean ± SEM values of four different determinations. Values of
phosphorylated tau for HDLm—, compared with HDLm+ and
HDLn, are significantly different at p < 0.02.
J. Neurochem., Vol. 69. No. 5, /997
2112 S. C. KIVATINITZ ET AL.
phorylation of the microtubule-associated protein tau
are also similar to those described for the aggregated
/3-amyloid peptide (Busciglio et al., 1995). The hyper-
phosphorylation of the tau species could be one of
the earliest factors initiating the disorganization of the
microtubules displayed by the neurons cultured in the
presence of aggregated HDL, because it has been dem-
onstrated that the hyperphosphorylated tau does not
show affinity for microtubules and it inhibits tubulin
polymerization (Alonso et a!., 1994).
It is particularly tempting to speculate about the
presence of oxidized aggregated HDL in the brains of
patients with neurodegenerative diseases that occur
with an increase in oxidative stress, such as Alzhei-
mer’ s disease, Parkinson’s disease, amyotrophic lateral
sclerosis, and Huntington’s disease (Bowling and Beal,
1995). In Alzheimer’s disease the necessary condi-
tions for the existence of aggregated HDL by oxidation
have been reported to occur, i.e., (I) disruption of the
blood—brain barrier (Mon et a!., 1991) to allow the
accumulation of HDL in brain, and (2) an increase in
lipid peroxidation (Chia et a!., 1984; Subbarao et a!.,
1990). In fact, it was reported very recently that, in
brains of patients with A!zheimer’ s disease, some of
the amyloid plaques contain deposits of apolipoprotein
A-I (Harr eta!., 1996). It is interesting that in Down’s
syndrome patients who develop early-onset Alzhei-
mer’ s disease, there is an increment of superoxide dis-
mutase activity concomitant with a higher lipid peroxi-
dation (Ceballos et al., 1988; Hyman, 1992; Busciglio
and Yankner, 1995).
Besides, several apolipoproteins other than apolipo-
protein A-I are known to be closely associated with
amyloid fibrillogenesis, mainly apolipoprotein E (Wis-
niewski and Frangione, 1992). One isotype of this
apolipoprotein, apolipoprotein E4, is associated with
microtubule depolymerization and inhibition of neurite
outgrowth (Nathan et a!., 1994).
Additional experimentation is needed to ascertain
the potential in vivo significance of our findings. As a
first approach, we are currently developing in vivo
models using in situ injection of aggregated HDL or
dietary induction of higher peroxidability of HDL to-
gether with conditions that favor oxidative stress.
Acknowledgment: This study was supported by grants
from Fundación Antorchas, CONICET, CONICOR, and
SeCyT-Universidad Nacional de Córdoba to S.C.K. and S.Q.
S.Q. is a member of the National Career of Investigators,
CONICET; LB. is a recipient of a CONICET fellowship.
We thank Rodolfo Artola, Susana Deza, and Gabriela
Schahner for skillful technical assistance. We are indebted
to Dr. A. Cáceres for his collaboration in the determination
of HDL diameters. We warmly thank Dr. G. Fidelio for his
collaboration in the fluorescence studies and Drs. H. S.
Barra, G. Fidelio, and A. Cáceres for critical reading of the
manuscript.
J. Neurochem., Vol. 69. No. 5, 1997
REFERENCES
Alonso A. del C., Zaidi T., Grundke-Iqbal I., and Iqbal K. (1994)
Role of abnormally phosphorylated tall in the breakdown of
microtubules in Alzheimer disease. Proc. Nan. Acad. Sci. USA
91, 5562—5566.
Binder L. I., Frankfurter A., and Rebhum L. I. (1985)The distribu-
tion of tau in the mammalian central nervous system. J. Biol.
Chem. 101, 1371—1378.
Black M., Slaughter T., and Fischer I. (1994) Microtubule-associ-
ated protein lb (MAP Ib) is concentrate in the distal region of
growing axons. J. Neurosci. 14, 857—870.
Blue M. L., Ostapchuk P., Gordon J. S., and William D. L. (1982)
Synthesis of apolipoprotein A-I by peripheral tissues of the
rooster. .1. Biol. Chem. 257, 11151—11159.
Bonnefont-Rousselot D., Motta C., Khalil A. 0., Sola R., Laville
A. E., Delattre J., and Gardesalbert M. (1995) Physicochemical
changes in human high-density-lipoproteins (HDL) oxidized
by gamma radiolysis-generated oxyradicals. Effect on their cho-
lesterol effluxing capacity. Biochim. Biophys. Acta 1255, 23—
30.
Borghini I., Barja F., Pometta D., and James R. W. (1995) Charac-
terization of subpopulations of lipoprotein particles isolated
from human cerebrospinal-fluid. Biochim. Biophys. Acta 1255,
192—200.
Bowling A. C. and Beal M. F. (1995) Bioenergetic and oxidative
stress in neurodegenerative diseases. Life Sci. 56, 1151 — 1171.
Boyd H. C., Gown A. M., Wolfbauer G., and Chait A. (1989) Direct
evidence for a protein recognized by a monoclonal antibody
against oxidatively modified LDL in atherosclerotic lesions
from Watanabe Heritable Hyperlipidemic rabbit. Am. J. Pathol.
135, 815—825.
Brinton E. A., Oram J. F., Chen C-H., Albers J., and Bierman E. L.
(1986) Binding of high density lipoprotein to cultured fibro-
blasts after chemical alteration of apoprotein aminoacid resi-
dues. J. Biol. Chem. 261, 495—503.
Busciglio J. and Yankner B. A. (1995) Apoptosis and increased
generation ofreactive oxygen species in Down’s syndrome neu-
rons in vitro. Nature 378, 776—778.
Busciglio J., Lorenzo A., and Yankner B. A. (1992) Methodological
variables in the assessment of /3-amyloid neurotoxicity. Neuro-
biol. Aging 13, 609—612.
Busciglio J., Lorenzo A., Yeh J., and Yankner B. A. (1995) /3-
Amyloid fibrils induce tau phosphorylation and loss of microtu-
bule binding. Neuron 14, 879—888.
Byrnes L., Luo C-C., Li W.-H., Yang C.-Y., and Chan L. (1987)
Chicken apolipoprotein A-I: eDNA sequence, tissue expression
andevolution. Biochem. Biophys. Res. Commun. 148,485—492.
Cáceres A., Mautino J., and Kosik K. (1992) Suppression of MAP-
2 in cultured cerebellar macroneurons inhibits minor neurite
formation. Neuron 9, 607—618.
Ceballos I., Delabar J.-M., Nicole A., Lynch R. E., Hallewell R. A.,
Kamoun P., and Sinet P.M. (1988) Expression of transfected
human CuZn superoxide dismutase gene in mouse L cells and
NS2OY neuroblastoma cells induces enhancement of glutathi-
one peroxidase activity. Biochim. Biophys. Acta 949, 58—64.
Chia L. S., Thompson J. E., and Moscarello M. A. (1984) X-Ray
diffraction evidence for myelin disorder in brain from humans
with Alzheimer’s disease. Biochim. Biophys. Ada 775, 308—
312.
Chiu H. C., Jeng J. R., and Shieh S. M. (1994) Increased oxidizabil-
ity plasma low density lipoprotein from patients with coronary
artery disease. Biochim. Biophys. Acta 1225, 200—208.
Chung B. H., Segrest J. P., Ray M. J., Brunzell J. D., Hokanson J. E.,
Kraus R. M., Beaudrie K., and Cone J. T. (1986) Single vertical
spin density gradient ultracentrifugation, in Methods in Enzy-
mology, Vol. 27, Plasma Lipoproteins (Segrest J. P. and Albers
J. J., eds), pp. 170—181. Academic Press, New York.
Cogny A., Paul J. L., Atger V., Soni T., and Moatti N. (1994)
Structural-changes of high-density-lipoprotein apolipoproteins
 NEURODEGENERATIVE EFFECTS OF AGGREGATED HDL
following incubation with human polymorphonuclear cells. Eur.
I. Biochem. 222, 965—973.
Cogny A., Atger V., Paul J. L., Soni T., and Moatti N. (1996) High-
density lipoprotein 3 physicochemical modifications induced by
interaction with human polymorphonuclear leukocytes affect
their ability to remove cholesterol from cells. Biochem. J. 314,
285 —292.
Dawson P. A., Schech N., and Williams D. L. (1986) Induction of
rat E and chicken A-I apolipoproteins mRNAs during optic
nerve degeneration. J. Biol. Chem. 261, 568 1—5684.
Devries H. E., Breedveld B., Kuiper J., Deboer A. G., Vanberkel
T. J. C., and Breimer D. D. (1995) High-density-lipoprotein
and cerebral endothelial-cells in-vitro. Interaction and transport.
Biochem. Pharmacol. 50, 271 —273.
Farooqui A. A., Haun S. E., and Horrocks L. A. (1994) Ischemia
and hypoxia, in Basic Neurochemistry: Molecular, Cellular,
and Medical Aspects (Siegel G. J., Agranoff B. W., Albers
R. W., and Molinoff P. B., eds), pp. 867—883. Raven Press,
New York.
Ferrari S., Tarugi P., Drusiani E., Calandra S., and Fregni M. (1987)
The complete sequence of chick apolipoprotein Al mRNA and
its expression in the developing chick. Gene 60, 39—46.
Gast K., Zirwer D., Ladhoff A., Schreiber J., Koelsch R., Kretschmer
K., and Lasch J. (1982) Auto-oxidation-induced fusion of lip-
ids. Biochim. Biophys. Acta 686, 99—109.
Grundke-Iqbal I., Iqbal K., Tung Y.-C., Wang G. P., and Wisniewski
H. M. (1984) Alzheimer paired helical filaments: immuno-
chemical identification of polypeptides. Acta Neuropathol.
(Berl.) 62, 259—267.
Harel A., Fainaru M., Shafer Z., Hernandez M., Cohen A., and
Schwartz M. (1989) Optic nerve regeneration in adult fish and
apolipoprotein A-I. J. Neurochem. 52, 1218—1228.
Han S. P., Vint L., Hollister R., Hyman B. T., and Mendez A. J.
(1996) Brain expression of apolipoprotein E, apolipoprotein J
and apolipoprotein A-I in Alzheimer’s disease. J. Neurochem.
66, 2429—2435.
Hazell L. J., Arnold L., Flowers D., Waeg G., Malle E., and Stocker
R. (1996) Presence of hypochiorite-modified proteins in human
atherosclerotic lesions. J. Clin. Invest. 97, 1535—1544.
Hermier D. and Dillon J.-C. (1992) Characterization of dietary-
induced hypercholesterolemia in the chicken. Biochim. Biophys.
Acta 1124, 178—184.
Hyman B. T. (1992) Down syndrome and Alzheimer disease. Prog.
Clin. Biol. Res. 379, 123—142.
Jackson R. L., Morrisett J. D., and Gotto A. M. Jr. (1976) Lipopro-
tein structure and metabolism. Physiol. Rev. 56, 259—316.
Kivatinitz S. C., Grabois V. R., and Quiroga S. (1995) High-density
lipoprotein inhibits UDP-N-acetylgalactosamine:GM3,N-ace-
tylgalactosaminyltransferase and differentiation of cultured ce-
rebral cells: comparison with a formerly described inhibitor of
this enzyme. J. Neurochem. 65, 775—781.
Klein R. A. (1970) The detection of oxidation in liposome prepara-
tions. Biochim. Biophys. Acta 210, 486—489.
Kruski A. W. and Scanu A. M. (1975) Properties of rooster serum
high density lipoproteins. Biochim. Biophy,s. Acta 409, 26—38.
Lakowicz J. R. (1983) Principles of Fluorescence Spectroscopy.
Plenum Press, New York.
Laville A. E., Sola R., Balanya J., Turner P. R., and Masana L.
(1994) In vitro oxidized HDL is recognized by the scavenger
receptor macrophages. Implications for its protective role in-
viva. Atherosclerosis 105, 179—189.
Lorenzo A. and Yankner B. A. (1994) /3-Amyloid neurotoxicity
requires fibril formation and is inhibited by Congo red. Proc.
Natl. Acad. Sci. USA 91, 12243—12247.
Mao S. J. T., Downing M. R., and Kottke B. A. (1980) Physical,
chemical and immunochemical studies of apolipoprotein A-I
from pigeon plasma high density lipoproteins. Biochim. Bio-
phys. Acta 620, 100—110.
Marcel Y. L., Jewer D., Vézina C., Miltrop P., and Weech P. K.
(1987) Expression of human apolipoprotein A-I epitopes in
2113
high density lipoprotein and in serum. J. Lipid Res. 28, 768—
777.
Marcel Y. L., Jewer D., Leblond L., Weech P. K., and Milne R. W.
(1989) Lipid peroxidation changes the expression of specific
epitopes of apolipoprotein A-I. J. Biol. Chem. 264, 19942—
19950.
Mattson M., Tomaselli K. J., and Rydel R. E. (1993) Calcium desta-
bilizing and neurodegenerative effects of aggregated /3-amyloid
peptide are attenuated by basic FGF. Brain Res. 621, 35—49.
Miller I. Y., Felikman Y., and Shaklai N. (1996) Hemoglobin in-
duced apolipoprotein B crosslinking in low-density lipoprotein
peroxidation. Arch. Biochem. Biophys. 326, 252—260.
Miyazaki A., Sakai. M., Suginohara Y., Hakamata H., Sakamoto
Y. I., Morikawa W., and Horiuchi 5. (1994) Acetylated low-
density-lipoprotein reduces its ligand activity for the scavenger
receptor after interaction with reconstituted high-density-lipo-
protein. J. Biol. Chem. 269, 5264—5269.
Mockel B., Zinke H., Flach R., WeiB B., Weiler-Giittler H., and
Gassen H. G. (1994) Expression of apolipoprotein A-I in por-
cine brain endothelium in vitro. J. Neurochem. 62, 788—798.
Morel D. W. (1994) Reduced cholesterol efflux to mildly oxidized
high-density-lipoprotein. Biochem. Biophvs. Res. Commun. 200,
408—4 16.
Mori S., Sternberger N. H., Herman M. M., and Sternberger L. A.
(1991) Leakage and neuronal uptake of serum protein in aged
and Alzheimer brains. Lab. Invest. 64, 345—351.
Nathan B. P., Bellosta S., Sanan D. A., Weisgraber K. H., Mahley
R. W., and Pitas R. E. (1994) Differential effects of apolipopro-
teins E3 and E4 on neuronal growth in-vitro. Science 264, 850—
852.
Novak S., Hiesberger T., Schneider W. J., and Nimpf J. (1996) A
new low density lipoprotein with 8 ligand binding repeats in
brain of chicken and mouse. J. Biol. Chem. 271, 11732—11736.
Parasassi T. and Gratton E. (1995) Membrane lipid domains and
dynamics as detected by laurdan fluorescence. J. Fluoresc. 5,
59—69.
Parasassi T., De StasioG., d’Ubaldo A., and Gratton E. (1990) Phase
fluctuations in phospholipid membranes revealed by laurdan
fluorescence. Biophys. J. 57, 1179—1186.
Parasassi T., De Stasio G., Ravagnan G., Rusch R. M., and Gratton
E. (1991) Quantitation of lipid phases in phospholipid vesicles
by the generalized polarization of laurdan fluorescence. Bio-
phys. J. 60, 179—189.
Parasassi T., Di Stefano M., Ravagnan G., Sapora 0., and Gratton
E. (1992) Membrane aging during cell growth ascertained by
laurdan generalized polarization. Exp. Cell Res. 202, 432—439.
Parasassi T., Loiero M., Raimondi M., Ravagnan G., and Gratton
E. (1993) Absence of lipid gel-phase domains in seven mamma-
lian cell lines and in four primary cell types. Biochim. Biophys.
Acta 1153, 143—154.
Parasassi T., Giusti A. M., Gratton E., Monaco E., Raimondi M.,
Ravagnan G., and Sapora 0. (1994) Evidence for an increase
in water concentration in bilayers after oxidative damage of
phospholipids induced by ionizing radiation. mt. Radiat. Biol.
65, 329—334.
Pettman J. C. L. and Sensenbrenner M. (1979) Morphological and
biochemical maturation of neurons cultured in the absence of
glial cells. Nature 281, 378—380.
Pike C. J., Walencewicz A. J., Glabe C. G., and Cotman C. W.
(1991) Aggregation-related toxicity of synthetic /3-amyloid
protein in hippocampal cultures. Fur. J. Pharmacol. 207, 367—
368.
Richter C. (1987) Biophysical consequences of lipid peroxidation
in membranes. Chem. Phys. Lipids 44, 175—189.
Roheim P.S., Carey M., Forte T., and Vega G. L. (1979) Apolipo-
proteins in human cerebrospinal fluid. Proc. Nati. Acad. Sci.
USA 76, 4646—4649.
Rosenfeld M. E., Palinki W., Ylä-Herttuala S., Butler S., and Witz-
tum J. L. (1990) Distribution of oxidation specific lipid-protein
J. Neurochem., Vol. 69, No. 5, /997
2114 S. C. KIVATINITZ ET AL.
adducts and apolipoprotein B in atherosclerotic lesions of vary-
ing severity from WHHL rabbits. Arteriosclerosis 10, 336—349.
Ross R. (1993) The pathogenesis of atherosclerosis: a perspective
of the 1990s. Nature 362, 801—808.
Rothblat G. H., Mahlberg F. H., Johnson W., and Phillips M. C.
(1992) Apolipoproteins, membrane cholesterol domains, and
the regulation of cholesterol efflux. J. Lipid Res. 33, 1091—
1097.
Rubin E. M. and Smith D. J. (1994) Atherogenesis in mice: getting
to the heart of a polymeric disorder. Trends Genet. 10, 199—
203.
Salmon S., Santus R., Maziere J. C., Aubilly M., and Haigle J.
(1992) Modified apolipoprotein pattern after inadiation of hu-
man high-density lipoproteins by ultraviolet B. Biochim. Bio-
phys. Ada 1128, 167—173.
Shackelford J. E. and Lebherz H. G. (1983) Synthesis and secretion
of apolipoprotein Al by chick breast muscle. J. Biol. Chem.
258, 7175—7180.
J. Neurochem., Vol. 69, No. 5, 1997
Shoukry M. I., Gong E. L., and Nichols A. V. (1994) Apolipopro-
tein-lipid association in oxidatively modified HDL and LDL.
Biochim. Biophys. Acta 1210, 355—360.
Subbarao K. V., Richardson S. J., and Ang L. C. (1990) Autopsy
samples of Alzheimer’s cortex show increased peroxidation in
vitro. J. Neurochem. 55, 342—345.
Thomas T., Thomas G., McLendon C., Sutton T., and Mullan M.
(1996) /3-Amyloid mediated vasoactivity and vascular endothe-
hal damage. Nature 380, 168—17 1.
Westermark P., Mucchiano G., Marthin T., Johnson K. H., and Slet-
ten K. (1995) Apohipoprotein A-1 derived amyloid in human
atherosclerotic plaques. Am. J. Pathol. 147, 238—244.
Wisniewski T and Frangione B. (1992) Apohipoprotein-E. A patho-
logical chaperone protein in patients with cerebral and systemic
amyloid. Neurosci. Lett. 135, 235—238.
Yankner B. A., Duffy L. K., and Kirshner D. A. (1990) Neuro-
trophic and neurotoxic effects of amyloid /3 protein: reversal
by tachykinin neuropeptides. Science 250, 279—282.